TICB 1590 No.

of Pages 12

Trends in Cell Biology

Review

The Complex Interplay between Antioxidants

and ROS in Cancer
Isaac S. Harris1,* and Gina M. DeNicola2,*

Reactive oxygen species (ROS) play important roles in tissue homeostasis, cellu- Highlights
lar signaling, differentiation, and survival. In this review, we discuss the types New tools allow in vivo measurements of
of ROS, their impact on cellular processes, and their pro- and antitumorigenic ROS in tumors.
effects. Further, we discuss recent advances in our understanding of both
Mouse modeling and genetic screening
endogenous and exogenous antioxidants in tumorigenic processes. Finally, approaches have revealed novel com-
we discuss how aberrant activation of antioxidant programs by the transcription plexities and redundancies in endoge-
factor NFE2-related factor 2 (NRF2) influences tumorigenesis and metastasis, nous antioxidant systems.
and where the current gaps in our knowledge remain. Exogenous antioxidants may promote
cancer through complex mechanisms.

Introduction
Aberrant NRF2 activation has diverse,
ROS play important roles in tissue homeostasis, from the regulation of signaling and differentia- and sometimes contradictory, impacts
tion to the promotion of cellular damage and death. Unsurprisingly, their levels are tightly regu- on tumor growth and metastasis.
lated by cellular antioxidant defenses to prevent unwanted consequences of their actions.
Tissue of origin, tumor stage, and the mi-
While generally grouped together, ROS are a diverse class of molecules with distinct effects on croenvironment greatly influence the in-
cellular components. Consequently, their influence on cellular processes is complex and they fluence of ROS on cancer.
have both pro- and antitumorigenic effects. In this review, we discuss the role of both oxidants
and antioxidants in tumorigenic processes.

Types of ROS
ROS are defined as molecules that contain and engage in the transfer of electrons from reactive
oxygen. It is challenging to measure ROS directly, and consequently many tools have been devel-
oped for the indirect measure of ROS in cells and tissues (Box 1). Here we outline the different
forms of ROS, their sources, and their primary targets (Figure 1).

Superoxide
Superoxide (O2−) is primarily produced as a consequence of electron reaction with molecular oxygen
at complex I/III of the mitochondrial electron transport chain. O2− is moderately reactive but short lived.
It is easily dismutated to hydrogen peroxide (H2O2) by superoxide dismutases or nonenzymatically.
Its anionic charge prevents its diffusion through membranes. Substantial extracellular O2− is also pro-
duced by certain cell types (e.g., neutrophils) by NADPH oxidase enzymes. In the cell, it targets iron–
sulfur (Fe-S) clusters to release iron [1]. O2− can also form peroxynitrite (ONOO−) through a reaction 1
Department of Biomedical Genetics and
with nitric oxide (NO). ONOO− reacts with proteins to cause oxidation or nitration of amino acids, Wilmot Cancer Institute, University of
DNA to induce double-strand breaks, and lipids to induce lipid peroxidation. Rochester Medical Center, 601 Elmwood
Ave., Rochester, NY 14642, USA
2
Department of Cancer Physiology, H.
Hydrogen Peroxide Lee Moffitt Cancer Center and Research
H2O2 is formed from O2− and is moderately reactive but long lived. H2O2 is also generated by Ero1 Institute, Tampa, FL, USA
as a consequence of oxidative protein folding in the endoplasmic reticulum (ER) [2]. It can diffuse
through membranes and consequently can have effects distal from its site of production. It is the
*Correspondence:
primary ROS responsible for protein oxidation. While low levels (1–10 nM) play an important role in
[email protected]
signaling via redox signaling via oxidation [protein tyrosine phosphatases (PTPs), insulin signal- (I.S. Harris) and [email protected]
ing], higher levels (N100 nM) cause ‘oxidative stress’ [3]. (G.M. DeNicola).

Trends in Cell Biology, Month 2020, Vol. xx, No. xx https://doi.org/10.1016/j.tcb.2020.03.002 1

© 2020 Elsevier Ltd. All rights reserved.
Trends in Cell Biology

Box 1. Tools for Measuring ROS

Many methods exist for the measurement of ROS and they have distinct advantages and disadvantages [147,148]. Many
studies rely on indirect measurements of ROS. These include redox-active probes that yield fluorescent or luminescent
products on oxidation by ROS, but most can be oxidized by several ROS species so specificity is a problem. Spin trapping
with electron paramagnetic resonance (EPR) detection is the most unambiguous method for free radical detection, but cel-
lular antioxidants may scavenge ROS before they can react with spin traps. The use of multiple complementary methods,
including evaluation of redox ratios (e.g., GSH/GSSG), protein oxidation states, and ratiometric reporters (e.g., HyPER,
roGFP) is suggested. Further, ROS are generated in distinct subcellular compartments and consequently their location
should be considered when interpreting their effects. Importantly, distinct cell states including loss of ECM attachment
can alter cellular metabolism to support mitochondrial ROS metabolism [149]. While ratiometric reporters have been
targeted to subcellular compartments, recently tools to generate localized ROS have also been developed [150]. As the
oxidation states of the mitochondria and ER are quite different from the cytosol, these tools will greatly improve our under-
standing of how different compartments respond to changes in ROS.

Until recently, the evaluation of ROS levels in vivo has remained elusive. The development of redox-active positron emis-
sion tomography (PET) tracers for in vivo ROS imaging has the potential to greatly expand our toolkit. These include: [18F]
ROStrace [151], an analog of the O2− probe dihydroethidium; [18F]PC-FLT [152], which measures extracellular and intracel-
lular levels of H2O2; and [18F]ROS1 [153], which measures O−2 and OH• radicals.

Peroxyl Radical
Because O2− and H2O2 are only moderately reactive, most ROS-induced cellular damage is due
to their conversion to other species. The peroxyl radical (OH•) is formed from H2O2 and is the
most reactive of all ROS. OH• is formed when H2O2 reacts with iron (Fe2+) in the Fenton reaction
[4]. O2− also contributes to OH• formation by reducing Fe3+ to Fe2+.

Lipid Peroxides
Because of their carbon–carbon double bonds, polyunsaturated fatty acids (PUFAs) contain re-
active hydrogen atoms that are highly susceptible to lipid peroxidation, which compromises the
integrity of lipid bilayers in cells. Lipid peroxidation is initiated by OH•, leading to the formation
of lipid radicals and lipid peroxyl radicals, which react with PUFAs in a propagation reaction to
generate lipid peroxides. Excessive lipid peroxidation is associated with the iron-dependent
form of cell death known as ferroptosis [5].

Tumor-Initiating/Promoting Effects of ROS

ROS induce DNA damage through their oxidation of nucleobases including guanine. Repair of
these modified bases can result in errors leading to mutagenesis. Consistently, radiation is one
of the most well-known sources of ROS [6] and has long been associated with tumor-initiating
events [7]. ROS can also alter cellular processes through their effects on protein function
(Figure 2). The effects of ROS are related to the degree of protein oxidation. Mild oxidation pro-
motes cellular signaling and is typically reversible (disulfides, sulfenic acid, sulfinic acid), allowing
rapid changes in protein activity and signaling networks. By contrast, excessive oxidation leads
to terminal oxidation (sulfonic acid) and complete loss of protein function. While irreversible cys-
teine modifications can be detrimental to protein function, reversible modifications can be protec-
tive during stress. Protein modifications play a key role in adaptation to oxidative stress by
activating antioxidant (KEAP1) or metabolic (GAPDH, PKM2) programs to facilitate ROS metab-
olism. Other reversible modifications can occur endogenously, including CoAlation [8,9] and
glutathionylation [10], which can both protect proteins from terminal oxidation and alter their func-
tion to promote metabolic rewiring. The influence of ROS on tumor initiation and promotion is
complex and related to the amount, duration, location, and context.

Endogenous Antioxidants
Antioxidant is a general term used to describe an enzyme or cofactor that participates in the
elimination of ROS (Figure 3). The most abundant endogenous antioxidant is the metabolic

2 Trends in Cell Biology, Month 2020, Vol. xx, No. xx

Trends in Cell Biology

Trends in Cell Biology

Figure 1. Types of Reactive Oxygen Species (ROS). Superoxide (O2−) is produced extracellularly by NADPH oxidase or
intracellularly by the mitochondrial electron transport chain (ETC). In the mitochondria, it targets iron–sulfur (Fe-S) clusters to
release iron (Fe2+) and reduces ferric iron (Fe3+) to ferrous iron (Fe2+). O2− is dismutated to hydrogen peroxide (H2O2) by
superoxide dismutases (SOD1, SOD2). H2O2 diffuses through membranes to react with proteins and DNA and is
detoxified to water by cellular peroxidases [catalase (Cat), glutathione peroxidase (GPX), peroxiredoxins (PRDX)]. O2−
produces peroxynitrite (ONOO−) through a reaction with nitric oxide (NO). The peroxyl radical (OH•) is formed from the
reaction of H2O2 with Fe2+ and the decomposition of ONOO− and initiates the lipid peroxidation cascade. First, OH•
reacts with lipids to form lipid radicals (L•), which react with oxygen to form lipid peroxide radicals (LOO•). LOO• reacts
with lipids to reform L• plus lipid peroxides (LOOH) and the cycle continues. Excessive lipid peroxidation leads to ferroptosis.

cofactor glutathione (GSH) [11]. GSH was first discovered over a century ago [12] and has long
been known to play a role in detoxifying reactions in cancer cells [13]. GSH is a tripeptide that
is synthesized in a two-step process. In the first step, the condensation of glutamate and cysteine
is catalyzed by glutamate–cysteine ligase catalytic subunit (GCLC). In the second, GSH synthe-
tase (GSS) incorporates glycine to form the tripeptide. Although cysteine is the rate-limiting me-
tabolite for this pathway in most contexts [14], both glutamate [15] and glycine [16,17] can also
be limiting for GSH synthesis. GSH is used as a cofactor by GSH S-transferases (GSTs) and
GSH peroxidases (GPXs) to eliminate ROS. GST and GPX enzymes comprise multiple families
and isoforms [18,19] and the exact targets of each are unclear. Besides GSH-dependent antiox-
idant systems, the sulfaredoxin (SRX) and thioredoxin (TXN) antioxidant networks regenerate
peroxiredoxins (PRDXs), a set of enzymes with high catalytic activity towards H2O2 [20,21]. Unlike
the highly abundant GSH metabolites, TXNs are small protein antioxidants and less abundant
[11]. While the TXN system can reduce PRDX disulfide bonds, SRX will reduce PRDXs that are
overoxidized to sulfinic acid. Distinct, but highly homologous, PRDX and TXN proteins localize
to either the mitochondria or the cytoplasm and the relative importance of each subcellular sys-
tem, as well as the crosstalk between them, is unclear [21]. Finally, the detoxification of ROS by

Trends in Cell Biology, Month 2020, Vol. xx, No. xx 3

Trends in Cell Biology

Trends in Cell Biology

Figure 2. Oxidative Protein Modifications. Oxidative protein modifications have important impacts on cellular signaling
and protein function. They include reversible modifications [CoAlation, disulfide (S–S) bond formation, nitrosylation,
glutathionylation, and persulfidation]. CoAlation and glutathionylation are the consequence of the reaction of the thiol of
coenzyme A or glutathione with the thiol (SH) of cysteine to form a disulfide bond. Nitrosylation is modification of the
cysteine thiol by nitric oxide (NO). Thiols can also be oxidized. Mild oxidation to sulfenic acid (SOH) and sulfinic acid
(SO2H) is reversible. By contrast, irreversible modifications [sulfonic acid (SO3H), tyrosine nitration (Tyr-NO2) by
peroxynitrate (ONOO−)] are terminal oxidation states that result in loss of protein function.

GSH and TXN generates oxidized forms of these antioxidants, which must be regenerated for
subsequent reactions. Oxidized GSH and TXN are both regenerated by reductases [GSH reduc-
tase (GR), TXN reductase 1 and 2 (TXNRD1/2)] using NADPH as an electron donor [22,23]. These
pathways are complementary and redundancy between GSH and TXN systems exists both in
normal and malignant tissue [24–27]. Furthermore, oxidative insults promote the expression of
enzymes in both the GSH and TXN systems, suggesting that they may work in unison to buffer
oxidative stress.

The importance of endogenous antioxidants in tumors depends heavily on the stage of tumori-
genesis. Antioxidants play an important role in preventing tumor initiation by preventing ROS-
induced oxidation of DNA and subsequent DNA damage. However, many studies rely heavily
on carcinogen-induced tumor model systems and because antioxidant systems participate in
carcinogen detoxification, their direct role in preventing ROS-induced tumor initiation is less
clear. By contrast, there are clear roles for antioxidant proteins in tumor progression. Below,
we provide an overview of these findings and highlight outstanding questions in the field.

Prevention of Tumor Initiation by Endogenous Antioxidants

Through the detoxification of ROS, antioxidants have the potential to prevent deleterious,
and sometimes oncogenic, outcomes. Multiple isoforms of GSTs can prevent skin, liver, and
colon tumor initiation in mice following exposure to carcinogens or loss of tumor suppressors
[28–31]. Similarly, GPXs can protect against carcinogen- and ROS-induced tumor initiation in

4 Trends in Cell Biology, Month 2020, Vol. xx, No. xx

Trends in Cell Biology

Trends in Cell Biology

Figure 3. Types of Antioxidants. Glutathione (GSH) is synthesized from cysteine, glutamate, and glycine in a two-step reaction by glutamate-cysteine ligase catalytic
(GCLC) and modifier (GCLM) subunits and GSH synthetase (GSS). Peroxidases and transferases (GPX and GST) use GSH as a cofactor to neutralize hydrogen peroxide
(H2O2). Thioredoxin (TXN) and sulfiredoxin (SRXN) promote peroxiredoxin (PRDX)-mediated H2O2 detoxification. GSH reductase (GSR) and TXN reductase (TXNRD1) use
NADPH to regenerate GSH and TXN as well as to reduce imported cystine to cysteine. NADPH is generated via multiple metabolic enzymes (IDH1/2, G6PD, ME1). Lipid
peroxidation is controlled by GSH-dependent GPX4 and GSH-independent ubiquinone (CoQ10) with ferroptosis suppressor protein 1 (FSP1). Exogenous supply of
vitamin E (α-tocopherol) buffers lipid peroxides. N-Acetyl cysteine promotes GSH production and protein persulfide-dependent ROS elimination.

multiple models. GPX3 suppresses tumor initiation in mouse models of colon cancer [32]. Simi-
larly, mice with reduced expression of SOD2, either alone or in combination with loss of GPX1,
exhibit increased DNA damage and tumor incidence [33,34]. Furthermore, treatment with SOD
mimetics that localize to the mitochondria blocks cancer cell proliferation and tumor growth
[35,36]. Evidence for the tumor suppressive abilities of antioxidants also exists in the TXN system.
Loss of Prdx1 leads to the accumulation of DNA damage and increased tumor incidence in older
mice [37]. Further, PRDX1 inhibits cancer cell growth by acting as a reductant towards PTEN to
promote its phosphatase activity towards AKT [38]. In addition, loss of PRDX6 accelerates HPV8-

Trends in Cell Biology, Month 2020, Vol. xx, No. xx 5

Trends in Cell Biology

induced skin carcinogenesis. Importantly, GSH and TXN can act together towards tumor preven-
tion, as mice with combined loss of GR and liver-specific TXNRD1 have increased sensitivity to
carcinogen-induced liver malignancies [39]. Finally, the accumulation of oxidized DNA is sufficient
to promote tumor initiation in mice. Loss of OGG1, which repairs 8-oxo-deoxyguanine in DNA,
results in spontaneous mouse lung tumors in the absence of any carcinogenic treatment [40].
Therefore, multiple lines of evidence indicate that endogenous antioxidants play a role in
tumor prevention.

By contrast, in vivo genetic studies in mice have also demonstrated a role for endogenous anti-
oxidants in promoting tumor initiation. Gpx2−/- mice are protected against azoxymethane-
induced colorectal tumorigenesis [41], suggesting that Gpx2 may support survival during the
early stages of transformation. Further, Srx−/- mice had fewer and smaller urethane-induced
lung tumors [42] and DMBA/TPA-induced skin tumors [43]. In sum, while some studies provide
clear evidence for certain antioxidants in the prevention of cancer initiation, others promote
tumor initiation in similar contexts. Additional work is needed to dissect these seemingly contra-
dictory results and their direct relation to ROS and protein antioxidant function. Further, work
is needed to determine whether tumor suppressive antioxidant pathways can be selectively
upregulated therapeutically for cancer prevention without inducing cancer promotion.

Support of Tumor Progression by Endogenous Antioxidants

On transformation, cells upregulate processes including mitochondrial metabolism [44] and pro-
tein translation that lead to increased generation of ROS, necessitating an increased reliance on
antioxidants to maintain redox balance [45,46]. Recent studies have demonstrated a role for GSH
in limiting DNA damage and maintaining protein homeostasis in tumors [26,47–49]. Without
ample GSH synthesis, tumor cells reach a barrier in progression to more advanced and aggres-
sive malignancies. The GST and GPX enzymes involved in the downstream utilization of GSH
have also been implicated in tumor progression. GSTs metabolize chemotherapies, including cis-
platin [50–52], and activate oncogenic signaling proteins such as Akt [53]. GPXs are required to
buffer ROS generation during tumor progression [54], most notably GPX4 via its inhibition of lipid
peroxidation and ferroptosis [55,56]. Ferroptosis implicates multiple processes, including GSH-
independent pathways that use the antioxidant cofactor ubiquinone (CoQ10) [57–59]. Interest-
ingly, therapy-resistant cancer cells that have undergone epithelial–mesenchymal transition
(EMT) are more sensitive to ferroptosis [56,60,61]. Components of the TXN system, such as
TXN and TXNRD1, promote tumor growth [62,63]. Expression of PRDX1 and PRDX4 is elevated
in malignant tissues and supports tumor survival [64,65]. Further, PRDX6 overexpression accel-
erated malignant progression [66]. The metabolism of superoxide also plays a role in tumor pro-
gression. Inhibition of SOD1 through copper chelators blocked lung tumorigenesis [67], although
copper chelation can also have SOD1-independent antitumorigenic properties [68]. Further,
targeting the TXN or SOD1 antioxidant systems impaired lung cancer cell survival following O2−
exposure [69]. Thus, endogenous antioxidant programs have the potential to support tumor pro-
gression and viability. The significant redundancy between the antioxidant systems remains a
challenge for therapy [62] and more work is needed to understand compensatory mechanisms
between the different protein components in these systems.

NADPH, which regenerates endogenous antioxidants in GSH and TXN systems, must be regen-
erated from NADP+. Regeneration of NADPH is fueled by several metabolic processes, most no-
tably the pentose phosphate pathway (PPP) and one-carbon metabolism [70,71]. Interestingly,
glucose-6-phosphate dehydrogenase (G6PD), which generates NADPH in the first step of the
PPP, is the most common enzyme defect in humans [72]. G6PD-deficient patients have reduced
NADPH levels, resulting in resistance to malaria as well as susceptibility to hemolytic anemia [73].

6 Trends in Cell Biology, Month 2020, Vol. xx, No. xx

Trends in Cell Biology

Several studies have demonstrated the importance of NADPH generation and reductive capacity
for tumorigenesis [74,75]. In line with these reports, G6PD-deficient patients show a lower risk for
colorectal cancer [76]. It is difficult to ascertain the contribution of G6PD deficiency to cancer risk
as the cellular role of G6PD extends well beyond the regeneration of GSH and TXN systems. Re-
cent studies have demonstrated the importance of NADPH/NADP+ control for the maintenance
of folate metabolism [77]. In addition, local control of NADPH/NADP+ can influence the oxidation
of cellular components, such as Fe-S clusters, independent of the GSH and TXN system [78]. Ad-
ditional studies are required to better understand the subcellular interplay between G6PD,
NADPH, ROS, and GSH/TXN activity in tumorigenesis. Further, there is a lack of understanding
on the relative importance of these antioxidant programs across tumor types and between differ-
ing environments.

Exogenous Antioxidants
Exogenous antioxidants are commonly used to treat animal models of cancer and to interrogate
the causal role of intracellular ROS in various tumor processes. The most widely used tool for
these studies is N-acetyl cysteine (NAC). Treatment of mice with NAC impaired p53-null lym-
phoma and lung cancer growth by preventing oxidation of DNA and subsequent mutagenic
events [79]. NAC also blocks the stabilization of Hif1a and perturbs hepatocellular xenograft tu-
mors [80]. Preclinical evidence supported clinical trials, but ultimately these showed no benefit
for patients [81]. However, recent studies with NAC suggest it can promote tumorigenesis as
well. NAC supplementation promoted the initiation, progression, and metastasis of multiple ge-
netically engineered mouse models of cancer, including melanoma, and lung cancer [82–84].
Further, the exact mechanisms behind the effects of NAC supplementation on cellular redox sta-
tus are also unclear. While NAC can contribute to GSH synthesis in some contexts, its major an-
tioxidant function may be through the production of hydrogen sulfide and protein persulfidation
[85]. Importantly, protein persulfides can be reduced to regenerate unmodified thiols [86,87].

Similar to NAC, vitamin E (alpha-tocopherol) was largely regarded as having antitumor potential
as a supplement [88]. These beliefs were upheld, even when other exogenous antioxidants,
such as beta-carotene, were found to increase cancer incidence [89]. Ultimately, a large-scale,
multicenter clinical trial was initiated to investigate the ability of vitamin E supplementation to pre-
vent prostate cancer incidence [90]. This trial was stopped because the vitamin E supplementa-
tion arm was incurring higher rates of prostate cancer [91,92]. Similar to NAC, vitamin E promotes
lung tumor and melanoma growth and progression [82,83]. Further, vitamin E can directly prevent
lipid oxidation and ferroptosis [55,93].

The interpretations of antioxidants are complicated by the ability of these molecules to be oxidized
themselves or to produce antioxidant-independent effects. Vitamin C (or ascorbate) is an
antioxidant that is absorbed through the diet and routinely supplemented exogenously in
experiments. Recent studies have shown that vitamin C is autoxidized to dehydroascorbate
(DHA) and can subsequently increase oxidative stress in cells [94,95]. Additionally, vitamin C
can negatively regulate hematopoietic stem cell (HSC) function by promoting the activity of
Tet2 [96]. As previously mentioned, NAC can produce H2S, which can influence metabolic and
signaling pathways [97,98]. In summary, caution must be taken when using exogenous antioxi-
dants to interrogate and interpret the impact of intracellular ROS on tumor biology.

Aberrant Activation of ROS Detoxification in Cancer

The tumor-promoting effects of cellular antioxidant programs are best evidenced by the aberrant
activation of the antioxidant transcription factor NRF2 in multiple cancer types. Under basal con-
ditions, NRF2 levels are constrained by its association with KEAP1, which targets NRF2 for

Trends in Cell Biology, Month 2020, Vol. xx, No. xx 7

Trends in Cell Biology

proteasomal degradation [99]. Following exposure of cells to oxidative or electrophilic stress,

cysteine residues on KEAP1 are modified, leading to impaired NRF2 ubiquitination and NRF2
accumulation. NRF2 promotes the transcription of many genes in the antioxidant system, includ-
ing those in the GSH and TXN antioxidant pathway [100]. Interestingly, NRF2 accumulation is
common in cancer, suggesting that increased antioxidant defense contributes to one or multiple
stages of the tumorigenic process. Various mechanisms exist for NRF2 accumulation across var-
ious cancer types. Mutations in NRF2 and KEAP1 that disrupt proper NRF2 degradation are
common in multiple cancers, including lung [100]. NRF2 exon skipping to delete the KEAP1 bind-
ing domain has been described in lung cancer [101] and oncogene-driven transcription can in-
crease the levels of NRF2 [102]. KEAP1 inactivation as a consequence of promoter methylation
[103], p62-mediated sequestration [104,105], and modification by the oncometabolites fumarate
and methylglyoxal [106–108] can also result in NRF2 accumulation.

Elegant animal studies have dissected the contribution of NRF2 to specific stages of tumorigen-
esis (Figure 4). Studies with NRF2 knockout mice have demonstrated that NRF2 contributes to
the incidence and growth of oncogene-driven lung tumors [102,109] and p62-driven pancreatic
tumorigenesis [110]. Further, KEAP1 deletion increases the tumor burden in lung tumor models
driven by KrasG12D/loss of p53 or loss of PTEN [111,112] and liver tumor burden driven by
Myc [113]. NRF2 activation in these models was associated with decreased levels of ROS and
oxidative DNA damage and the activation of metabolic processes. NRF2 regulates multiple met-
abolic pathways at the interface of antioxidant defense and proliferative processes, including the
PPP and serine biosynthesis [17,114,115], which may play a dominant role over antioxidant pro-
cesses at certain stages of tumor progression. Importantly, NRF2 activation causes widespread

Trends in Cell Biology

Figure 4. The Complex Role of NFE2-Related Factor 2 (NRF2) at Different Stages of Carcinogenesis. NRF2
plays dual roles in tumor initiation, progression, and metastasis. NRF2 protects against oxidation and carcinogen-induced
DNA damage via the antioxidant and detoxification programs. By preventing excessive oxidative damage, NRF2 promotes
the viability of transformed cells during early stages of tumorigenesis. Further, NRF2 promotes progression to higher-grade
tumors. The role of NRF2 in metastasis is complex and tumor and tissue specific. Loss of NRF2 promotes epithelial–
mesenchymal transition (EMT) via reactive oxygen species (ROS) to promote migration and invasion to support
intravasation/extravasation. By contrast, NRF2 can promote migration and invasion through the transcription factor
Bach1. ROS also promote the death of cells detached from the extracellular matrix (anoikis), which NRF2 may protect
against. Consequently, NRF2 and ROS play complex roles at different tumor stages.

8 Trends in Cell Biology, Month 2020, Vol. xx, No. xx

Trends in Cell Biology

changes in global cysteine reactivity [116,117], suggesting that the effects on NRF2 on metabo- Outstanding Questions
lism may not be limited to direct transcriptional targets. Do antioxidants have additional effects on
the tumor microenvironment compared
with the activation of tumor antioxidant
The influence of ROS on metastasis is complex and seemingly contradictory. ROS have
programs?
been shown to promote metastasis in multiple contexts [36,118–123]. By contrast, detachment
of cells from the extracellular matrix (ECM) induces oxidative stress that limits survival in circulation Do specific exogenous and endogenous
and antioxidants have been shown to be protective and metastasis promoting [82–84,124–127]. It antioxidants have unique effects on
tumor biology?
is therefore unsurprising that the influence of NRF2 on metastasis would be complex as well. NRF2
activation in KrasG12D; p53flox/flox lung tumors indirectly promoted the stability of the transcription Can we define the context-specific,
factor BACH1 via heme catabolism, which promoted metastasis via BACH1 transcriptional targets compartment-specific, and concentra-
[124], an effect that could be recapitulated by antioxidant treatment [125]. However, in a mouse tion-dependent roles of ROS during
different stages of tumor initiation, pro-
model of pancreatic cancer similarly driven by KrasG12D and p53 loss-of-function, ROS induced gression, and metastasis?
by deletion of TP53-induced glycolysis and apoptosis regulator (TIGAR) or NRF2 increased metas-
tasis to the lung [128]. Interestingly, BACH1 was not affected by ROS in the pancreatic cancer Are there roles for different types of
ROS (H2O2 vs O2- vs ONOO-)?
model; rather, DUSP6 expression was lost, thereby leading to increased ERK activity and EMT.
Notably, the difference between the lung and the pancreas is further exemplified by a recent Can we better define the functional
study examining the consequence of Keap1 deletion in the context of the KrasG12D and KrasG12D; consequences of oxidative protein
p53R172H pancreatic tumor models [129], which recapitulates the genetics of the lung tumor study. modifications?

However, Keap1 deletion in these pancreatic tumor models instead resulted in pancreatic atrophy Would targeting enzymes that use
[129]. Thus, the influence of NRF2 on metastasis is highly context dependent and may be influ- antioxidant cofactors, such as GSTs
enced by the tissue type, the NRF2 dosage, and ROS-dependent and -independent effects. and GPXs, instead of targeting enzymes
that synthesize or regenerate the
antioxidants cofactors themselves
The effects and degree of KEAP1 loss of function may also be context dependent within the provide a larger therapeutic window
same tissue and genetic context. In a competition model, KEAP1 deletion was not selected for for cancer treatment?
in KrasG12D, KrasG12D; p53flox/flox, or KrasG12D; LKB1flox/flox models compared with other tumor
suppressors [130]. Further, while a heterozygous KEAP1R554Q loss-of-function mutant modestly
increased tumor size, in agreement with deletion studies in the KrasG12D; p53flox/flox model, ho-
mozygous KEAP1R554Q expression actively antagonized tumor formation [131]. Additional work
is needed to understand whether specific KEAP1 mutations or degrees of NRF2 activation
have disparate effects on tumor growth and progression.

Concluding Remarks
Stepping away from a ‘one size fits all’ view of the effects of ROS and antioxidants on tumor bi-
ology will help to reconcile many of the seemingly contradictory effects of these molecules across
studies (see Outstanding Questions). Recent studies support the idea that there are different
pools of ROS with differing functions. While NADPH oxidase-derived ROS was shown to promote
proliferation in the mouse intestine, ROS resulting from loss of TIGAR impaired proliferation in the
same cells [132,133]. It is important to note that NADPH oxidase generates extracellular O2−, while
TIGAR protects against ROS intracellularly by supporting the PPP [134]. In addition, the role of
antioxidant programs in the microenvironment needs to be considered when interpreting antioxidant
and whole-body gene knockout studies. These cell populations can rely on both antioxidant pro-
grams and ROS generation for function [135–139]. Further, the beneficial or detrimental roles of
ROS in the cell do not necessarily need to be mutually exclusive. Improvement in technologies, includ-
ing genetic screens that have brought increased clarity to enzymatic pathways [57,58,140–142], and
large-scale profiling efforts of cancer models using omic technologies [143–146], will undoubtedly ad-
vance our understanding of the complexities of ROS and antioxidant pathways.

References
1. Keyer, K. and Imlay, J.A. (1996) Superoxide accelerates DNA 2. Tu, B.P. and Weissman, J.S. (2002) The FAD- and O2-dependent
damage by elevating free-iron levels. Proc. Natl. Acad. Sci. reaction cycle of Ero1-mediated oxidative protein folding in the
U. S. A. 93, 13635–13640 endoplasmic reticulum. Mol. Cell 10, 983–994

Trends in Cell Biology, Month 2020, Vol. xx, No. xx 9

Trends in Cell Biology

3. Sies, H. (2017) Hydrogen peroxide as a central redox signaling 31. Li, J. et al. (2019) GSTZ1 deficiency promotes hepatocellular
molecule in physiological oxidative stress: oxidative eustress. carcinoma proliferation via activation of the KEAP1/NRF2
Redox Biol. 11, 613–619 pathway. J. Exp. Clin. Cancer Res. 38, 438
4. Winterbourn, C.C. (1995) Toxicity of iron and hydrogen 32. Barrett, C.W. et al. (2013) Tumor suppressor function of the
peroxide: the Fenton reaction. Toxicol. Lett. 82–83, 969–974 plasma glutathione peroxidase gpx3 in colitis-associated
5. Stockwell, B.R. et al. (2017) Ferroptosis: a regulated cell death carcinoma. Cancer Res. 73, 1245–1255
nexus linking metabolism, redox biology, and disease. Cell 33. Van Remmen, H. et al. (2003) Life-long reduction in MnSOD
171, 273–285 activity results in increased DNA damage and higher incidence
6. Riley, P.A. (1994) Free radicals in biology: oxidative stress and of cancer but does not accelerate aging. Physiol. Genomics
the effects of ionizing radiation. Int. J. Radiat. Biol. 65, 27–33 16, 29–37
7. Figge, F.H. (1947) Cosmic radiation and cancer. Science 105, 34. Zhang, Y. et al. (2009) Mice deficient in both Mn superoxide
323–325 dismutase and glutathione peroxidase-1 have increased oxida-
8. Tsuchiya, Y. et al. (2018) Protein CoAlation and antioxidant tive damage and a greater incidence of pathology but no re-
function of coenzyme A in prokaryotic cells. Biochem. J. 475, duction in longevity. J. Gerontol. A Biol. Sci. Med. Sci. 64,
1909–1937 1212–1220
9. Tsuchiya, Y. et al. (2017) Protein CoAlation: a redox-regulated 35. Weinberg, F. et al. (2010) Mitochondrial metabolism and ROS
protein modification by coenzyme A in mammalian cells. generation are essential for Kras-mediated tumorigenicity.
Biochem. J. 474, 2489–2508 Proc. Natl. Acad. Sci. U. S. A. 107, 8788–8793
10. Matsui, R. et al. (2020) Redox regulation via glutaredoxin-1 and 36. Porporato, P.E. et al. (2014) A mitochondrial switch promotes
protein S-glutathionylation. Antioxid. Redox Signal. 32, 677–700 tumor metastasis. Cell Rep. 8, 754–766
11. Winterbourn, C.C. and Hampton, M.B. (2008) Thiol chemistry 37. Neumann, C.A. et al. (2003) Essential role for the peroxiredoxin
and specificity in redox signaling. Free Radic. Biol. Med. 45, Prdx1 in erythrocyte antioxidant defence and tumour
549–561 suppression. Nature 424, 561–565
12. De Rey-Pailhade, J. (1885) Sur la Formation de l’Hydrogène 38. Cao, J. et al. (2009) Prdx1 inhibits tumorigenesis via regulating
Sulfuré dans l’Organisme à la Suite de l’Ingestion de Quelques PTEN/AKT activity. EMBO J. 28, 1505–1517
Médicaments, A. Delahaye et E. Lecrosnier (in French) 39. McLoughlin, M.R. et al. (2019) TrxR1, Gsr, and oxidative stress
13. Hirono, I. (1961) Mechanism of natural and acquired resistance determine hepatocellular carcinoma malignancy. Proc. Natl.
to methyl-bis-(beta-chlorethyl)-amine N-oxide in ascites Acad. Sci. U. S. A. 116, 11408–11417
tumors. Gan 52, 39–48 40. Sakumi, K. et al. (2003) Ogg1 knockout-associated lung
14. Lu, S.C. (2009) Regulation of glutathione synthesis. Mol. Asp. tumorigenesis and its suppression by Mth1 gene disruption.
Med. 30, 42–59 Cancer Res. 63, 902–905
15. McBrayer, S.K. et al. (2018) Transaminase inhibition by 2- 41. Muller, M.F. et al. (2013) Deletion of glutathione peroxidase-2
hydroxyglutarate impairs glutamate biosynthesis and redox inhibits azoxymethane-induced colon cancer development.
homeostasis in glioma. Cell 175, 101–116.e25 PLoS One 8, e72055
16. Maddocks, O.D. et al. (2012) Serine starvation induces stress 42. Mishra, M. et al. (2018) Nrf2-activated expression of
and p53-dependent metabolic remodelling in cancer cells. sulfiredoxin contributes to urethane-induced lung tumorigenesis.
Nature 493, 542–546 Cancer Lett. 432, 216–226
17. DeNicola, G.M. et al. (2015) NRF2 regulates serine biosynthe- 43. Wu, L. et al. (2014) Tumor promoter-induced sulfiredoxin is re-
sis in non-small cell lung cancer. Nat. Genet. 47, 1475–1481 quired for mouse skin tumorigenesis. Carcinogenesis 35,
18. Nebert, D.W. and Vasiliou, V. (2004) Analysis of the glutathione 1177–1184
S-transferase (GST) gene family. Hum. Genomics 1, 460–464 44. Hensley, C.T. et al. (2016) Metabolic heterogeneity in human
19. Margis, R. et al. (2008) Glutathione peroxidase family – an evo- lung tumors. Cell 164, 681–694
lutionary overview. FEBS J. 275, 3959–3970 45. Trachootham, D. et al. (2006) Selective killing of oncogenically
20. Trujillo, M. et al. (2007) Pre-steady state kinetic characterization transformed cells through a ROS-mediated mechanism by
of human peroxiredoxin 5: taking advantage of Trp84 fluores- beta-phenylethyl isothiocyanate. Cancer Cell 10, 241–252
cence increase upon oxidation. Arch. Biochem. Biophys. 46. Trachootham, D. et al. (2009) Targeting cancer cells by ROS-
467, 95–106 mediated mechanisms: a radical therapeutic approach? Nat.
21. Lu, J. and Holmgren, A. (2014) The thioredoxin antioxidant Rev. Drug Discov. 8, 579–591
system. Free Radic. Biol. Med. 66, 75–87 47. Diehn, M. et al. (2009) Association of reactive oxygen species
22. Arner, E.S. (2009) Focus on mammalian thioredoxin levels and radioresistance in cancer stem cells. Nature 458,
reductases – important selenoproteins with versatile functions. 780–783
Biochim. Biophys. Acta 1790, 495–526 48. Luo, M. et al. (2018) Targeting breast cancer stem cell state
23. Lu, S.C. (2013) Glutathione synthesis. Biochim. Biophys. Acta equilibrium through modulation of redox signaling. Cell
1830, 3143–3153 Metab. 28, 69–86.e6
24. Mandal, P.K. et al. (2010) Loss of thioredoxin reductase 1 ren- 49. Harris, I.S. et al. (2019) Deubiquitinases maintain protein ho-
ders tumors highly susceptible to pharmacologic glutathione meostasis and survival of cancer cells upon glutathione
deprivation. Cancer Res. 70, 9505–9514 depletion. Cell Metab. 29, 1166–1181.e6
25. Mandal, P.K. et al. (2010) System x-c and thioredoxin reductase 50. Oshimori, N. et al. (2015) TGF-β promotes heterogeneity and
1 cooperatively rescue glutathione deficiency. J. Biol. Chem. drug resistance in squamous cell carcinoma. Cell 160, 963–976
285, 22244–22253 51. Morrow, C.S. et al. (1998) Coordinated action of glutathione
26. Harris, I.S. et al. (2015) Glutathione and thioredoxin antioxidant S-transferases (GSTs) and multidrug resistance protein 1
pathways synergize to drive cancer initiation and progression. (MRP1) in antineoplastic drug detoxification. Mechanism of
Cancer Cell 27, 211–222 GST A1-1- and MRP1-associated resistance to chlorambucil in
27. Eriksson, S. et al. (2015) Dietary methionine can sustain MCF7 breast carcinoma cells. J. Biol. Chem. 273, 20114–20120
cytosolic redox homeostasis in the mouse liver. Nat. Commun. 52. Peklak-Scott, C. et al. (2008) Role of glutathione S-transferase
6, 6479 P1-1 in the cellular detoxification of cisplatin. Mol. Cancer Ther.
28. Abel, E.L. et al. (2010) Evidence that Gsta4 modifies suscepti- 7, 3247–3255
bility to skin tumor development in mice and humans. J. Natl. 53. Liu, C.J. et al. (2017) Glutathione-S-transferase A 4 (GSTA4)
Cancer Inst. 102, 1663–1675 suppresses tumor growth and metastasis of human hepato-
29. Henderson, C.J. et al. (2011) Increased skin papilloma forma- cellular carcinoma by targeting AKT pathway. Am. J. Transl.
tion in mice lacking glutathione transferase GSTP. Cancer Res. 9, 301–315
Res. 71, 7048–7060 54. Herault, O. et al. (2012) A role for GPx3 in activity of normal and
30. Ritchie, K.J. et al. (2009) Markedly enhanced colon tumorigen- leukemia stem cells. J. Exp. Med. 209, 895–901
esis in ApcMin mice lacking glutathione S-transferase Pi. Proc. 55. Yang, W.S. et al. (2014) Regulation of ferroptotic cancer cell
Natl. Acad. Sci. U. S. A. 106, 20859–20864 death by GPX4. Cell 156, 317–331

10 Trends in Cell Biology, Month 2020, Vol. xx, No. xx

Trends in Cell Biology

56. Wu, J. et al. (2019) Intercellular interaction dictates cancer cell 85. Ezerina, D. et al. (2018) N-Acetyl cysteine functions as a fast-
ferroptosis via NF2–YAP signalling. Nature 572, 402–406 acting antioxidant by triggering intracellular H2S and sulfane
57. Doll, S. et al. (2019) FSP1 is a glutathione-independent sulfur production. Cell Chem. Biol. 25, 447–459.e4
ferroptosis suppressor. Nature 575, 693–698 86. Millikin, R. et al. (2016) The chemical biology of protein
58. Bersuker, K. et al. (2019) The CoQ oxidoreductase FSP1 acts hydropersulfides: studies of a possible protective function of
parallel to GPX4 to inhibit ferroptosis. Nature 575, 688–692 biological hydropersulfide generation. Free Radic. Biol. Med.
59. Frei, B. et al. (1990) Ubiquinol-10 is an effective lipid-soluble 97, 136–147
antioxidant at physiological concentrations. Proc. Natl. Acad. 87. Ono, K. et al. (2014) Redox chemistry and chemical biology of
Sci. U. S. A. 87, 4879–4883 H2S, hydropersulfides, and derived species: implications of
60. Viswanathan, V.S. et al. (2017) Dependency of a therapy- their possible biological activity and utility. Free Radic. Biol.
resistant state of cancer cells on a lipid peroxidase pathway. Med. 77, 82–94
Nature 547, 453–457 88. Taylor, P.R. and Albanes, D. (1998) Selenium, vitamin E, and
61. Hangauer, M.J. et al. (2017) Drug-tolerant persister cancer prostate cancer – ready for prime time? J. Natl. Cancer Inst.
cells are vulnerable to GPX4 inhibition. Nature 551, 247–250 90, 1184–1185
62. Stafford, W.C. et al. (2018) Irreversible inhibition of cytosolic 89. Alpha-Tocopherol, Beta. Carotene. Cancer. Prevention. Study.
thioredoxin reductase 1 as a mechanistic basis for anticancer Group (1994) The effect of vitamin E and beta carotene on the
therapy. Sci. Transl. Med. 10, eaaf7444 incidence of lung cancer and other cancers in male smokers.
63. Gallegos, A. et al. (1996) Transfection with human thioredoxin N. Engl. J. Med. 330, 1029–1035
increases cell proliferation and a dominant-negative mutant 90. Lippman, S.M. et al. (2005) Designing the Selenium and
thioredoxin reverses the transformed phenotype of human Vitamin E Cancer Prevention Trial (SELECT). J. Natl. Cancer
breast cancer cells. Cancer Res. 56, 5765–5770 Inst. 97, 94–102
64. Cha, M.K. et al. (2009) Overexpression of peroxiredoxin I and 91. Klein, E.A. et al. (2011) Vitamin E and the risk of prostate
thioredoxin1 in human breast carcinoma. J. Exp. Clin. Cancer cancer: the Selenium and Vitamin E Cancer Prevention Trial
Res. 28, 93 (SELECT). JAMA 306, 1549–1556
65. Jiang, H. et al. (2014) Expression of peroxiredoxin 1 and 4 pro- 92. Lippman, S.M. et al. (2009) Effect of selenium and vitamin E on
motes human lung cancer malignancy. Am. J. Cancer Res. 4, risk of prostate cancer and other cancers: the Selenium and
445–460 Vitamin E Cancer Prevention Trial (SELECT). JAMA 301, 39–51
66. Rolfs, F. et al. (2013) Dual role of the antioxidant enzyme 93. Dixon, S.J. et al. (2012) Ferroptosis: an iron-dependent form of
peroxiredoxin 6 in skin carcinogenesis. Cancer Res. 73, nonapoptotic cell death. Cell 149, 1060–1072
3460–3469 94. Yun, J. et al. (2015) Vitamin C selectively kills KRAS and BRAF
67. Glasauer, A. et al. (2014) Targeting SOD1 reduces experimen- mutant colorectal cancer cells by targeting GAPDH. Science
tal non-small-cell lung cancer. J. Clin. Invest. 124, 117–128 350, 1391–1396
68. Brady, D.C. et al. (2014) Copper is required for oncogenic 95. Schoenfeld, J.D. et al. (2017) O·-2 and H2O2-mediated disrup-
BRAF signalling and tumorigenesis. Nature 509, 492–496 tion of Fe metabolism causes the differential susceptibility of
69. Torrente, L. et al. (2020) Inhibition of TXNRD or SOD1 over- NSCLC and GBM cancer cells to pharmacological ascorbate.
comes NRF2-mediated resistance to beta-lapachone. Redox Cancer Cell 32, 268
Biol. 30, 101440 96. Agathocleous, M. et al. (2017) Ascorbate regulates
70. Fan, J. et al. (2014) Quantitative flux analysis reveals folate- haematopoietic stem cell function and leukaemogenesis.
dependent NADPH production. Nature 510, 298–302 Nature 549, 476–481
71. Lewis, C.A. et al. (2014) Tracing compartmentalized NADPH 97. Vitvitsky, V. et al. (2018) Cytochrome c reduction by H2S po-
metabolism in the cytosol and mitochondria of mammalian tentiates sulfide signaling. ACS Chem. Biol. 13, 2300–2307
cells. Mol. Cell 55, 253–263 98. Mishanina, T.V. et al. (2015) Biogenesis of reactive sulfur spe-
72. Cappellini, M.D. and Fiorelli, G. (2008) Glucose-6-phosphate cies for signaling by hydrogen sulfide oxidation pathways.
dehydrogenase deficiency. Lancet 371, 64–74 Nat. Chem. Biol. 11, 457–464
73. Luzzatto, L. and Seneca, E. (2014) G6PD deficiency: a classic 99. Itoh, K. et al. (1999) Keap1 represses nuclear activation of an-
example of pharmacogenetics with on-going clinical tioxidant responsive elements by Nrf2 through binding to the
implications. Br. J. Haematol. 164, 469–480 amino-terminal Neh2 domain. Genes Dev. 13, 76–86
74. Wang, H. et al. (2017) The metabolic function of cyclin D3–CDK6 100. Hayes, J.D. and McMahon, M. (2009) NRF2 and KEAP1 muta-
kinase in cancer cell survival. Nature 90, 60 tions: permanent activation of an adaptive response in cancer.
75. Son, J. et al. (2013) Glutamine supports pancreatic cancer Trends Biochem. Sci. 34, 176–188
growth through a KRAS-regulated metabolic pathway. Nature 101. Goldstein, L.D. et al. (2016) Recurrent loss of NFE2L2 exon 2 is
496, 101–105 a mechanism for Nrf2 pathway activation in human cancers.
76. Dore, M.P. et al. (2016) Glucose-6-phosphate dehydrogenase Cell Rep. 16, 2605–2617
deficiency and risk of colorectal cancer in Northern Sardinia: a 102. DeNicola, G.M. et al. (2011) Oncogene-induced Nrf2 transcrip-
retrospective observational study. Medicine (Baltimore) 95, tion promotes ROS detoxification and tumorigenesis. Nature
e5254 475, 106–109
77. Chen, L. et al. (2019) NADPH production by the oxidative 103. Wang, R. et al. (2008) Hypermethylation of the Keap1 gene in
pentose-phosphate pathway supports folate metabolism. human lung cancer cell lines and lung cancer tissues. Biochem.
Nat. Metab. 1, 404–415 Biophys. Res. Commun. 373, 151–154
78. Ward, N.P. et al. (2020) Nicotinamide nucleotide 104. Komatsu, M. et al. (2010) The selective autophagy substrate
transhydrogenase regulates mitochondrial metabolism in p62 activates the stress responsive transcription factor Nrf2
NSCLC through maintenance of Fe-S protein function. through inactivation of Keap1. Nat. Cell Biol. 12, 213–223
J. Exp. Med. 217, e20191689 105. Lau, A. et al. (2010) A noncanonical mechanism of Nrf2 activa-
79. Sablina, A.A. et al. (2005) The antioxidant function of the p53 tion by autophagy deficiency: direct interaction between Keap1
tumor suppressor. Nat. Med. 11, 1306–1313 and p62. Mol. Cell. Biol. 30, 3275–3285
80. Gao, P. et al. (2007) HIF-dependent antitumorigenic effect of 106. Adam, J. et al. (2011) Renal cyst formation in Fh1-deficient
antioxidants in vivo. Cancer Cell 12, 230–238 mice is independent of the Hif/Phd pathway: roles for fumarate
81. Aitio, M.L. (2006) N-Acetylcysteine – passe-partout or much in KEAP1 succination and Nrf2 signaling. Cancer Cell 20,
ado about nothing? Br. J. Clin. Pharmacol. 61, 5–15 524–537
82. Sayin, V.I. et al. (2014) Antioxidants accelerate lung cancer pro- 107. Kinch, L. et al. (2011) Succination of Keap1 and activation of
gression in mice. Sci. Transl. Med. 6, 221ra15 Nrf2-dependent antioxidant pathways in FH-deficient papillary
83. Le Gal, K. et al. (2015) Antioxidants can increase melanoma renal cell carcinoma type 2. Cancer Cell 20, 418–420
metastasis in mice. Sci. Transl. Med. 7, 308re8 108. Bollong, M.J. et al. (2018) A metabolite-derived protein modifi-
84. Piskounova, E. et al. (2015) Oxidative stress inhibits distant cation integrates glycolysis with KEAP1–NRF2 signalling.
metastasis by human melanoma cells. Nature 527, 186–191 Nature 562, 600–604

Trends in Cell Biology, Month 2020, Vol. xx, No. xx 11

Trends in Cell Biology

109. Strohecker, A.M. et al. (2013) Autophagy sustains mitochon- 130. Rogers, Z.N. et al. (2017) A quantitative and multiplexed ap-
drial glutamine metabolism and growth of BrafV600E-driven proach to uncover the fitness landscape of tumor suppression
lung tumors. Cancer Discov. 3, 1272–1285 in vivo. Nat. Methods 14, 737–742
110. Todoric, J. et al. (2017) Stress-activated NRF2–MDM2 cas- 131. Kang, Y.P. et al. (2019) Cysteine dioxygenase 1 is a metabolic
cade controls neoplastic progression in pancreas. Cancer liability for non-small cell lung cancer. Elife 8, e45572
Cell 32, 824–839.e8 132. Cheung, E.C. et al. (2015) Opposing effects of TIGAR- and
111. Best, S.A. et al. (2018) Synergy between the KEAP1/NRF2 and RAC1-derived ROS on Wnt-driven proliferation in the mouse
PI3K pathways drives non-small-cell lung cancer with an al- intestine. Genes Dev. 30, 52–63
tered immune microenvironment. Cell Metab. 27, 935–943.e4 133. Myant, K.B. et al. (2013) ROS production and NF-κB activation trig-
112. Romero, R. et al. (2017) Keap1 loss promotes Kras-driven lung gered by RAC1 facilitate WNT-driven intestinal stem cell proliferation
cancer and results in dependence on glutaminolysis. Nat. Med. and colorectal cancer initiation. Cell Stem Cell 12, 761–773
23, 1362–1368 134. Bensaad, K. et al. (2006) TIGAR, a p53-inducible regulator of
113. Sanghvi, V.R. et al. (2019) The oncogenic action of NRF2 de- glycolysis and apoptosis. Cell 126, 107–120
pends on de-glycation by fructosamine-3-kinase. Cell 178, 135. Warnatsch, A. et al. (2017) Reactive oxygen species localiza-
807–819.e21 tion programs inflammation to clear microbes of different size.
114. Mitsuishi, Y. et al. (2012) Nrf2 redirects glucose and glutamine Immunity 46, 421–432
into anabolic pathways in metabolic reprogramming. Cancer 136. Ohl, K. and Tenbrock, K. (2018) Reactive oxygen species as
Cell 22, 66–79 regulators of MDSC-Mediated immune suppression. Front.
115. Singh, A. et al. (2013) Transcription factor NRF2 regulates miR-1 Immunol. 9, 2499
and miR-206 to drive tumorigenesis. J. Clin. Invest. 123, 137. Griess, B. et al. (2020) Scavenging reactive oxygen species se-
2921–2934 lectively inhibits M2 macrophage polarization and their pro-
116. Bar-Peled, L. et al. (2017) Chemical proteomics identifies tumorigenic function in part, via Stat3 suppression. Free
druggable vulnerabilities in a genetically defined cancer. Cell Radic. Biol. Med. 147, 48–60
171, 696–709.e23 138. Mak, T.W. et al. (2017) Glutathione primes T cell metabolism for
117. Chio, I.I.C. et al. (2016) NRF2 promotes tumor maintenance by inflammation. Immunity 46, 675–689
modulating mRNA translation in pancreatic cancer. Cell 166, 139. Rodriguez, A.E. et al. (2019) Serine metabolism supports mac-
963–976 rophage IL-1β production. Cell Metab. 29, 1003–1011.e4
118. Arnandis, T. et al. (2018) Oxidative stress in cells with extra cen- 140. Birsoy, K. et al. (2015) An essential role of the mitochondrial
trosomes drives non-cell-autonomous invasion. Dev. Cell 47, electron transport chain in cell proliferation is to enable aspar-
409–424.e9 tate synthesis. Cell 162, 540–551
119. Arora, S. et al. (2013) An undesired effect of chemotherapy: 141. Garcia-Bermudez, J. et al. (2019) Squalene accumulation in
gemcitabine promotes pancreatic cancer cell invasiveness cholesterol auxotrophic lymphomas prevents oxidative cell
through reactive oxygen species-dependent, nuclear factor death. Nature 567, 118–122
κB- and hypoxia-inducible factor 1α-mediated up-regulation 142. Alvarez, S.W. et al. (2017) NFS1 undergoes positive selection in
of CXCR4. J. Biol. Chem. 288, 21197–21207 lung tumours and protects cells from ferroptosis. Nature 551,
120. Goh, J. et al. (2011) Mitochondrial targeted catalase sup- 639–643
presses invasive breast cancer in mice. BMC Cancer 11, 191 143. Li, H. et al. (2019) The landscape of cancer cell line metabolism.
121. Ishikawa, K. et al. (2008) ROS-generating mitochondrial DNA Nat. Med. 25, 850–860
mutations can regulate tumor cell metastasis. Science 320, 144. Ghandi, M. et al. (2019) Next-generation characterization of the
661–664 Cancer Cell Line Encyclopedia. Nature 569, 503–508
122. O’Leary, B.R. et al. (2015) Loss of SOD3 (EcSOD) expression 145. Behan, F.M. et al. (2019) Prioritization of cancer therapeutic tar-
promotes an aggressive phenotype in human pancreatic ductal gets using CRISPR-Cas9 screens. Nature 568, 511–516
adenocarcinoma. Clin. Cancer Res. 21, 1741–1751 146. Nusinow, D.P. et al. (2020) Quantitative proteomics of the
123. Radisky, D.C. et al. (2005) Rac1b and reactive oxygen species Cancer Cell Line Encyclopedia. Cell 180, 387–402.e16
mediate MMP-3-induced EMT and genomic instability. Nature 147. Murphy, M.P. et al. (2011) Unraveling the biological roles of re-
436, 123–127 active oxygen species. Cell Metab. 13, 361–366
124. Lignitto, L. et al. (2019) Nrf2 activation promotes lung cancer 148. Dikalov, S.I. et al. (2018) Electron paramagnetic resonance
metastasis by inhibiting the degradation of Bach1. Cell 178, measurements of reactive oxygen species by cyclic hydroxyl-
316–329.e18 amine spin probes. Antioxid. Redox Signal. 28, 1433–1443
125. Wiel, C. et al. (2019) BACH1 stabilization by antioxidants stim- 149. Jiang, L. et al. (2016) Reductive carboxylation supports redox
ulates lung cancer metastasis. Cell 178, 330–345.e22 homeostasis during anchorage-independent growth. Nature
126. Schafer, Z.T. et al. (2009) Antioxidant and oncogene rescue of 532, 255–258
metabolic defects caused by loss of matrix attachment. Nature 150. Mishina, N.M. et al. (2019) Which antioxidant system shapes in-
461, 109–113 tracellular H2O2 gradients? Antioxid. Redox Signal. 31, 664–670
127. Labuschagne, C.F. et al. (2019) Cell clustering promotes a met- 151. Hou, C. et al. (2018) Development of a positron emission to-
abolic switch that supports metastatic colonization. Cell Metab. mography radiotracer for imaging elevated levels of superoxide
30, 720–734.e5 in neuroinflammation. ACS Chem. Neurosci. 9, 578–586
128. Cheung, E.C. et al. (2020) Dynamic ROS control by TIGAR 152. Carroll, V. et al. (2014) A boronate-caged [18F]FLT probe for
regulates the initiation and progression of pancreatic cancer. hydrogen peroxide detection using positron emission
Cancer Cell 37, 168–182.e4 tomography. J. Am. Chem. Soc. 136, 14742–14745
129. Hamada, S. et al. (2018) Simultaneous K-ras activation and 153. Al-Karmi, S. et al. (2017) Preparation of an 18F-labeled
Keap1 deletion cause atrophy of pancreatic parenchyma. hydrocyanine dye as a multimodal probe for reactive oxygen
Am. J. Physiol. Gastrointest. Liver Physiol. 314, G65–G74 species. Chemistry 23, 254–258

12 Trends in Cell Biology, Month 2020, Vol. xx, No. xx