1 Introduction
Metastasis is the final step of a complex, multi-stage, and stepwise pathological process, known as cancer progression, which is the main cause of cancer-derived mortality [
1]. Through metastasis, tumour cells can disseminate, invade, and colonize distant secondary sites initiating a phase which is incurable from a clinical perspective. Cellular and environmental factors contribute to the process shaping the clinical evolution of cancer patients and highlight the extraordinary complexity of tumour evolution, which is at the base of the inefficacy of current clinical interventions and therapies against metastasis [
2].
The heterogeneity of tumour cells regarding their morphology, genetic background, and molecular signalling has been widely described. In addition and related to tumour cell heterogeneity, epigenetic modifications of DNA and/or chromatin in response to different environmental circumstances lead to changes in the selective regulation of gene transcription [
3]. The wide variety of cell-intrinsic and cell-extrinsic events, the microenvironmental circumstances faced by a growing tumour mass, together with individual genetic backgrounds, which can either promote or oppose metastasis development, lead to the generation of cell subpopulations characterized by a significant plasticity and increased metastatic potential [
3,
4]. However, because of the complexity of tumour growth and progression, it is improbable that isolated genetic alterations are the sole cause of metastasis. Rather, the combination of genetic and epigenetic modifications, together with metabolic adaptations, might be necessary to fuel tumour progression [
5].
Metastasis is considered an inefficient process based on the small proportion of cancer cells, known as disseminating tumour cells (DTCs), that colonize an organ which is distant from the anatomical site where the primary tumour originated [
6]. Tumour cell dissemination consists of a process known as the invasion-metastasis cascade, which involves different steps. Likewise, it has been proposed [
7] that DTCs able to generate a metastatic lesion are distinguished by four features known as the four hallmarks of metastasis: motility and invasion, plasticity, modulation of the local microenvironment of the secondary tissue, and colonization.
Specialized tumour cell subpopulations within the primary tumour mass have been shown to promote tumour cell dissemination based on the expression of a family of metastasis suppressor genes [
8]. Although apparently contradictory, the re-expression of such genes in metastasis-promoting tumour cells would block tumour cell dissemination and metastasis growth while stimulating angiogenesis and engaging inflammatory cells. The activation of these processes would promote the development and establishment of premetastatic niches through the communication of cells from the primary tumour with other sites of the body [
9]. Indeed, it is known that metastasis development is initiated long before any metastatic tumour mass is detectable, as initially described by Drs. Kaplan and Lyden [
9‐
11], due to the establishment of supportive metastatic environments induced by the primary tumour though the secretion of soluble factors and the recruitment of hematopoietic and mesenchymal stem cell populations that mobilize and condition the secondary site.
Likewise, tumour cell invasion requires alterations in the surrounding environment, as well as modifications in cell morphology and phenotype. During this step, there are three processes that have been considered as dynamically regulated: adhesion, reorganization of the extracellular matrix (ECM) and motility through the contractility of the cytoskeleton.
Adhesion is mediated by integrins and transmembrane glycoproteins. These surface molecules are considered the main cellular adhesion receptors at the interface with the ECM. Furthermore, integrins are associated with almost every step of the metastasis cascade through their diverse functions as signalling molecules, mechanotransducers, and crucial components of the cell migration machinery [
12]. These molecules are cell surface heterodimers that link the actin cytoskeleton to the cellular membrane and mediate cell-ECM interactions. The strength of the cellular adhesion to the ECM might be regulated through cell-intrinsic signalling pathways, whereas cellular phenotype can be modified through changes in cellular adhesion, highlighting the bidirectional nature of this interaction [
12,
13].
On the other hand, ECM remodelling occurs through the release of degradative enzymes by tumour cells or cells associated to the tumour. It has been described that these enzymes can act alone or together, through interactions that modulate their catalytic activity. Serine proteinases, cysteine proteinases, aspartyl proteinases, and matrix metalloproteinases (MMPs) are proteases well known to play a key role in tumour invasion [
14]. However, the function of some of these proteolytic enzymes, specifically MMPs, is not only the physically demolition of ECM barriers but also the modulation of several other cellular processes (such as cell growth, differentiation, apoptosis, angiogenesis, chemotaxis, and migration) through their substrates and cleavage products [
15,
16].
Finally, motility during tumour cell invasion has been distinctly described depending on whether it involves a single cell or a group of cells moving in concert through a mechanism known as collective migration [
17]. Cell migration is classified as mesenchymal or amoeboid, when referring to single cell migration, and collective when migration involves a coordinated group of cells. These three modes or types of cellular migration are interconvertible depending on the modulation of cytoskeletal structure [
18] and the relative levels of adhesion, cellular, and nuclear deformability [
19]. At the dynamical front edge of amoeboid migration, two types of cellular protrusions have been distinguished: amoeboid blebby and ameboid filopodial/pseudopodal. Amoeboid migration is also characterized by ability of the cells to deform their body, maintaining the tissue architecture and weakly adhering to its ECM [
20]. Conversely, mesenchymal migration showed a stronger adhesion and capability of tissue realignment and remodelling through deposition of ECM and cytokines [
21]. Cells undergoing mesenchymal migration are characterized by an elongated shape, decreased cell–cell interactions, and increased motility. Tumour cells are known to achieve this characteristic morphology through a process known as epithelial-to-mesenchymal transition (EMT). EMT is characterized by a decrease in the cellular expression of epithelial-specific Cadherin-1 (CDH1) and the increase of Cadherin-2 and/or -3 (CDH2 and CDH3, respectively) enabling more motility and individual migration in the cells [
22]. EMT will be further discussed in the context of redox signalling in metastasis, as part of specific sections in this review. Regarding to collective migration, three additional subtypes of collective movement have been described, which are characterized by an increasing degree of cell–cell adhesion: neuronal, epithelial (sheet/strand or ductal/glandular), and endothelial (vascular) collective migration. The morphological and molecular mechanisms regulating collective migration have been recently reviewed elsewhere and the readers are referred to these articles for more information [
23,
24].
Once local invasion is completed, cells must intravasate the circulatory system becoming circulating tumour cells (CTCs). To the succeed of this step, the ECM and the basement membrane must be partially degraded, so tumour cells can push between endothelial cells extending filopodia into the lumen, preserving the integrity of the endothelial barrier [
7,
25]. Although the metastatic process is highly inefficient, early steps in the metastatic cascade are achieved by tumour cells more frequently than later steps. For example, the number of tumour cells shed into the bloodstream is significantly higher than the number of tumour cells that eventually will give raise to an active metastatic lesion [
26,
27]. During their journey through the bloodstream, tumour cells are in contact with leukocytes, lymphocytes, and other immune components. Tumour cells have been observed to evade immune system by downregulating the expression of antigens, producing factors that help tumour cells to be recognized as “normal,” or directly killing immune cells. Despite the high degree of adaptability to a hard microenvironment, as it is the bloodstream, a significant proportion of the shed CTCs succumb to the exposure to hydrostatic pressures and hemodynamic shear forces [
28]. Even when CTCs are characterized for being deformable, cell origin and biophysical parameters may determine if the cell will survive or will be broken by shear. Interestingly, to protect cells from shear forces and immune attacks, cancer cells can be transported as an embolus or a CTC cluster [
29].
After surviving shear forces, tumour cells adhere in a selective manner to the endothelium. Endothelial cells found in each tissue express characteristic markers or a combination of them that can be recognized by tumour cells, promoting selective infiltration [
30‐
32]. Apparently, tumour cells adhere in a more efficient manner at anatomical locations where inflammation takes place. After recognition and adherence to the endothelium, cancer cells migrate through it and encounter the basement membrane. At the basement membrane, cancer cells produce proteinases that lead to its deformation and squeeze between endothelial cells and through holes in the matrix [
7,
33]. Extravasation was previously thought to be a limiting step of the metastatic cascade, but even its requirement for the completion of the metastatic process has been discussed, as it has been observed that tumour cells can form pulmonary metastases through the attachment to the lung endothelium and intravascular growth [
34]. Next, tumour cells start their interaction with the premetastatic niche, which promotes the proliferation and colonization of that secondary site. At this point, colonization depends on the ability of DTCs to interpret and adapt to the new tissue microenvironment, which will determine tumour cell fate. On one hand, tumour cells can remodel the microenvironment to allow continued growth, leading to the development of a clinically detectable metastatic lesion [
35,
36]. Interestingly, the requirements for colonization of secondary tissues are similar to those of the primary tumour, such as be provided of enough oxygen and nutrients [
37,
38], but in the secondary sites, additional players determine DTCs’ fate. In this regard, immune cells have been shown to play a role in the establishment of the premetastatic niche and tumour cell colonization. Neutrophils form the so-called neutrophil extracellular traps (NETs) by the extracellular deposition of DNA, which promote DTC colonization and growth [
39]. Conversely, DTCs require to evade immune surveillance and clearance to successfully colonize secondary sites and grow into a metastatic lesion. Malladi et al. [
40] showed that DTCs self-impose a slow-cycling state through the autocrine inhibition of Wnt signalling and downregulation of natural killer (NK) lymphocyte ligands, which impeded NK-mediated recognition and cytolysis and promoted immune evasion. The “slow-cycling state” described by the authors alludes to a stage of metastasis known as tumour dormancy.
As described above, metastasis is a stepwise process; the kinetics of which are highly variable across the different tumour malignancies and within different subtypes of neoplasms originated at the same anatomical primary location [
7]. The highly heterogeneous rates of recurrence in breast cancer, depending on the estrogen receptor (ESR) status of the tumour, exemplify this phenomenon. Breast cancer patients diagnosed with ESR-negative tumours have a higher risk of recurrence during the first 5 years after diagnosis, as compared to patients diagnosed with ESR-positive tumours. On the contrary, patients diagnosed with ESR-positive tumours show a higher steady rate of recurrence that expands from 5 to 20 years from diagnosis [
41,
42]. This highly variable and potentially long period from tumour diagnosis to the detection of a metastatic lesion may result from DTCs. DTCs are thought to prevail in a dormant state until the appropriate stimuli trigger a switch that promotes colonization of the microenvironment and proliferation in the metastatic site, resuming tumour progression and the emergence of a clinically active metastatic lesion [
6]. Different concepts have been coined in relation to tumour dormancy. Tumour dormancy can be the result of a proliferative arrest at the G0 phase of the cell cycle (tumour cell dormancy), or the consequence of a balance between proliferation and apoptosis of cancer cells (tumour mass dormancy) [
43,
44]. On one side, data from different studies have associated apoptosis in tumour mass dormancy with the inexistence of a proper vascularity providing sufficient nutrients and oxygen, known as angiogenic dormancy [
45,
46]. On the other side, tumour mass dormancy has also been linked to the response of the immune system, probably due to the coordinated activity of cytotoxic CD8 + T cells, memory T cells, and humoral response, which is called immune system-induced dormancy [
43]. Despite the potential involvement of these different and presumably non-exclusive modes of tumour dormancy along the metastatic cascade, tumour cell dormancy has been best characterized, both at the molecular and clinical levels, in the context of metastasis.
As mentioned before, microenvironment has been shown to play a key role in the establishment and induction of tumour cell dormancy. Early seminal studies have demonstrated that DTCs undergo growth arrest and enter dormancy when they are not able to achieve appropriate interactions with the ECM. Supporting this idea, Aguirre-Ghiso et al. [
47] described that the
in vivo downregulation of urokinase plasminogen activator receptor (uPAR) induces tumour dormancy by inhibiting the interaction of the uPA/uPAR proteins with the α5β1 integrin. The downregulation of this interaction decreased tumour cell adhesion to fibronectin and reduced the activation of mitogen-activated protein kinase (MAPK)/extracellular signal-regulated kinase (ERK) pathway. An additional study reported that the uPA/uPAR interaction activate the generation of insoluble fibronectin fibrils inhibiting p38 MAPK activity. Furthermore, the authors also concluded that depending on p38 MAPK/ERK activity ratio Hep3 cells, a model of human head and neck squamous cell carcinoma (HNSCC) showed a distinct proliferative status. Hep3 cells showed growth arrest when the ERK/p38 ratio was high due to p38 phosphorylation and activation [
48]. Interestingly, downregulation of uPAR in Hep3 cells inhibited focal adhesion kinase (FADK 1) phosphorylation and downstream proto-oncogene tyrosine-protein kinase Src (Src) activation leading to cellular dormancy
in vivo [
49].
In line with these findings, MDA-MB-231 breast cancer cells infiltrated into the bone marrow have been shown to upregulate the expression of Src kinases and upregulate the phosphoinositide 3-kinase (PI3K)/protein kinase B (PKB, also named AKT) pathway, as a mechanism of survival in response to stromal cell-derived factor 1 (SDF-1)/C-X-C chemokine receptor type 4 (CXCR-4) and tumour necrosis factor ligand superfamily member 10 (TNFSF10, also named TRAIL), which are highly expressed in the bone microenvironment [
50]. Barkan et al. found that the dormant-to-proliferative switch was dependent on the engagement of integrin β1 (ITGB1) and downstream signalling via activation of FADK 1, Src, ERK, and myosin light-chain kinase (MLCK) in breast cancer D2.0R cells. They demonstrated that ITGB1-mediated phosphorylation of myosin light chain by MLCK is necessary for the development of actin stress fibers and proliferative growth. When either ITGB1 or MLCK were inhibited, the dormant-to-proliferative switch was prevented and a significant reduction of the metastatic burden was observed
in vivo [
51]. In addition, the authors showed that the dormant-to-proliferative switch was driven by cytoskeletal reorganization in otherwise dormant D2.0R cells due to the enrichment of the metastatic niche with type I collagen [
52]. Lastly, the team showed that combination treatment eliciting the simultaneous inhibition of Src and ERK1/2 could induce apoptosis in dormant breast cancer cells [
53]. In addition, a recent study revealed that D-HEp3 (dormant HNSCC) DTCs reorganize the ECM through the secretion of type III collagen via DDR1-mediated STAT1 signalling to maintain tumour cell dormancy. Conversely, the elimination of collagen III re-established the proliferative behavior of cancer cells and changes in the structure and quantity of collagen during the dormant-to-proliferative transition were observable [
54].
Other components of the microenvironment, such as the vascular system, have also been studied in the context of tumour cell dormancy. The endothelium has been shown to promote breast tumour cell dormancy via the expression and release of thrombospondin-1 (TSP-1). Conversely, the authors found that DTC proliferation was activated by endothelial tip cells in the neovasculature through secretion of transforming growth factor-beta (TGFβ) and periostin [
55,
56]. Finally, a recent study showed that the interaction of breast DTCs with resident epithelial cells of the secondary microenvironment significantly impacted tumour cell dormancy. Montagner et al. [
57] showed that dormant breast DTCs in the lung interacted with alveolar type 1 epithelial cells leading to the formation of fibronectin fibrils and driving integrin-mediated pro-survival signals. Mechanistically, the interaction was mediated by the secreted frizzled-related protein 2 (FRP-2), whose expression inhibition reduced the number of dormant breast DTCs.
While environmental factors have been studied and thoroughly discussed in the context of tumour dormancy induction and maintenance, only few studies have addressed the intrinsic molecular mechanisms which allow dormant DTCs to survive in the metastatic niche for extended periods. In this regard, autophagy, which is an evolutionary conserved mechanism of cell survival opposing metabolic stress, has been shown to promote dormant DTCs survival [
58]. Interestingly, this and other studies that will be discussed later in this review highlight the key role of redox signalling and the intracellular redox balance in tumour cell dormancy and metastasis. Herein, we will review the current experimental evidence showing the determining role of redox biology and redox signalling along the metastatic cascade. In addition, we will discuss the potential therapeutic implications of such knowledge in the design and implementation of new and improved antineoplastic therapies.
2 Redox signalling in tumour progression
ROS are produced intracellularly as a by-product, by mitochondria and other cellular elements, and exogenously by pollutants, tobacco smoke, drugs, xenobiotics, and radiation. Cancer cells exhibit persistently high levels of ROS because of genetic, metabolic, and microenvironment-associated instability [
59‐
61]. Therefore, cancer cells are chronically exposed to sublethal levels oxidative stress which are known to modulate cell signalling and fate. At this point, it is worth to review the current concept of oxidative stress, which has evolved since it was first introduced by the scientific literature. Oxidative stress is considered a cellular state in which living cells are exposed to highly reactive oxidizing molecules. Consequently, the pro-oxidant/antioxidant balance of the cell is disrupted in favor of pro-oxidant processes leading to a wide variety of cellular responses, ranging from the modulation of signalling networks to apoptosis due to persistent and/or severe oxidative damage [
62‐
64].
Accumulated evidence has suggested that highly metastatic cancer cells contain high levels of ROS and that intracellular redox state governs crucial steps for the metastatic process promoting cell invasion and metastatic spread. Early experiments showed that treatment of carcinoma cells with H
2O
2 prior to intravenous injection into mice enhanced metastasis [
65]. Additionally, subpopulations of the low- or non-motile breast cancer cell line MCF-7 that possess higher levels of endogenous ROS, as compared to the parental cell line, showed increased motility. In addition, orthotopic breast tumours generated with these “high endogenous ROS” cell subpopulations metastasized to the lung, liver, and spleen while the orthotopic tumours generated using the parental MCF-7 line did not [
66]. The role of ROS in metastasis is also supported by the fact that ROS attenuation by antioxidants suppressed hypoxia-induced metastasis of human pancreatic cancer cells in a xenograft nude mouse model [
67].
Mechanistically, ROS have been found to increase the expression and/or activate MMPs, adhesion molecules [
68], epidermal growth factor (EGF) [
69], epidermal growth factor receptor (EGFR) [
70], and vascular endothelial growth factor (VEGF) [
71] whose upregulation and activity is known to be crucial along the metastatic cascade. Several genes relevant to EMT, including CDH1, integrins, and MMPs, have shown to be directly or indirectly regulated by intracellular ROS levels [
71]. Interestingly, the dismutation of mitochondrially generated superoxide to H
2O
2 is considered an important step in oxidative stress-mediated expression of MMP genes [
72]. In this line, the treatment of SCp2 mouse mammary epithelial cells with the ROS scavenger N-acetyl-L-cysteine (NAC) abolished EMT though the inhibition of the expression of MMP-3, a stromal protease that is upregulated in mammary tumours [
73]. Additionally, to modulate expression of MMP genes, ROS can lead to the direct activation of MMPs through reaction with thiol groups in their catalytic domain [
74]. Interestingly, ROS regulate not only the expression and activity of MMPs, but also the inactivation of their inhibitors, such as the metalloproteinase inhibitor (TIMP) [
75,
76]. Increased MMP activity has also been associated with angiogenesis, increased tumour cell invasion, and blood vessel penetration [
77‐
80]. A role for ROS in angiogenesis though MMPs increase has been evidenced. ROS-induced secretion of MMP-1 from tumour cells promoted vessel growth within the tumour microenvironment [
81]. Moreover, a transient expression of MMP-1, MMP-2, and MMP-9 correlated with an increase in ROS during formation of capillary-like structures, implicating that MMP-mediated angiogenesis also occurs through upregulation of ROS [
82]. On the other hand, cancer cells have been shown to purposefully restrain pyruvate from entry into mitochondrial oxidative metabolism, given that the ROS produced as by-products of mitochondrial respiration exhibit anti-metastasis activity [
83].
Resistance to anoikis and independence from cell attachment signals promote tumour cell survival through increased generation of intracellular ROS. It has been suggested that an increase in oxidative stress mimics autocrine/adhesive signals, which in non-tumour cells is mediated by growth factor- and integrin-mediated signalling pathways [
70,
84,
85]. The activation of these signalling pathways would contribute to increase the threshold for anoikis induction in cancer cells, elevating their disseminating and metastatic potentials. Thus, metastatic cancer cells gain increased anoikis resistance and survival advantage through increased intracellular ROS generation and ROS-mediated signalling.
ROS have also been reported to participate in the regulation of mesenchymal-like tumour cell movement that likely involve, besides gene regulation, a direct modification of cytoskeleton dynamics through actin glutathionylation [
85,
86]. In general, ROS appear to promote an “explorative” behavior whereby membrane protrusions and fast-turnover focal contacts with ECM prevail over stable focal adhesions and cell contractility. These characteristics are typically observed in invadopodia, the actin cytoskeleton-based structures that tumour cells use to invade. NADPH oxidases (NOX) have been found, concentrated, and activated, at the invadopodia of several types of malignant cells [
87‐
89]. NOX1-mediated ROS generation has been shown to be necessary for the formation of invadopodia [
88], where also MMP activity is concentrated. Accordingly, invadopodia formation is impaired in the absence of NOX-derived ROS [
88]. ROS generated by NOX have been show to activate the cofilin pathway and thus contribute to increased cell migration [
90,
91].
ROS may also promote tumour cell metastasis by increasing vascular permeability [
77] and triggering vasodilation through activation of the enzyme heme oxygenase 1 (HO-1), given that HO-1 is able to induce the formation of nitric oxide [
92]. The sources of ROS and their importance differ along the different steps of the metastatic process. Elevated ROS levels resulting from mutations in mitochondrial DNA have also been shown to promote metastasis [
70,
93]. Cell detachment during metastasis upregulates pyruvate dehydrogenase kinase 4 (PDK4) which inhibits pyruvate dehydrogenase complex (PDHc) and decreases the flux of glucose carbon into the tricarboxylic acid (TCA) cycle [
94]. On the other hand, cell adhesion and migration are dependent on integrin binding to ECM and these are able to elevate oxidant levels mainly by increasing prostaglandin G/H synthase 2 (PGHS-2, also named COX-2) [
68] but also through polyunsaturated fatty acid 5-lipoxygenase (5-LO) and even mitochondria [
68,
84]. In this context, an increase in mitochondrial ROS was linked to a first cellular contact with the ECM and increases in cytosolic ROS were shown to contribute to cytoskeleton remodelling and actin stress fiber formation during a later phase of the process [
84,
95]. In turn, these increases in ROS can trigger oxidative stress leading to oxidative damage to DNA and genomic instability. Despite NOXs have been clearly involved in invasion-related redox signalling, mitochondria may also contribute as sources of oxidant species in malignant growth [
96‐
99].
On the other hand, the effects of ROS are not specific to cancer cells and may result in the destruction of normal cells and tissues as well. Changes in the tumour microenvironment can bring about invasion and adhesion processes. Oxidative stress initiated in tumour cells is transferred to cancer-associated fibroblasts laterally and vectorially via H
2O
2. Excess of stromal ROS production has been shown to drive the onset of antioxidant defense in adjacent cancer cells, protecting them from apoptosis. Moreover, ROS can also act as players of immune regulation in cancer development. In particular, ROS are likely to participate as immunosuppressive agents [
100‐
102] in the tumour microenvironment facilitating tumour invasion, metastasis, and resistance. Most ROS-sensitive pathways transduce cytoplasmic signals to the nucleus, where they influence the activity of transcription factors that control the expression of a wide array of genes. In this regard, to prevent excessive intracellular ROS, cancer cells respond to oxidative stress by inducing the transcription of antioxidant enzymes, highlighting the relevance of an in-depth knowledge of these pathways for use in elaborating therapies that alter ROS levels.
Finally, antioxidant defense systems are expected to be also key signalling elements since they also modulate redox state. High levels of ROS are usually compensated by increased antioxidant capacity of the cancer cells. Due to the persistent high ROS microenvironment and increased intracellular ROS levels, cancer cells adopt efficient mechanisms of ROS detoxification. Consequently, they show high dependency on antioxidant systems for their survival. The cell’s defense against ROS includes antioxidant enzymes that detoxify ROS and prevent their intracellular accumulation at high concentrations [
103]. Evidence suggests that cancer progression involves numerous alterations in specific metabolic pathways involved in synthesis of proteins, lipids, and nucleotides. Besides, there is an increase in the generation of reductive equivalents, such as NADPH or GSH, and redox cofactors, such as NADH and flavin adenine dinucleotide (FADH2). There is a reciprocal crosstalk between metabolism and redox balance in cancer cells, with a particular emphasis on the role of glycolysis, glutaminolysis, fatty acid oxidation, one-carbon metabolism, and the pentose phosphate pathway [
104,
105]. Cell detachment has been shown to increase the expression of manganese superoxide dismutase (MnSOD), the main mitochondrial antioxidant enzyme, to detoxify mitochondrial ROS resulting from detachment [
106]. Moreover, it has been found that cells depleted of MnSOD were hypersensitive to matrix detachment [
106]. Therefore, through the activation of the antioxidant systems, cancer cells gain increased anoikis resistance and a survival advantage for the completion of subsequent steps of metastasis. However, metastatic breast cancer and highly invasive pancreatic cancer cells show lower levels and activity of the antioxidant enzyme MnSOD [
107‐
109]. Other redox proteins with redundant functions, such as thioredoxin and peroxiredoxin, may contribute to survive the raise in oxidative stress caused by anoikis. Indeed, it has been shown that reverse (basolateral-to-apical) transendothelial migration of human melanoma cells is induced by H
2O
2 and can be blocked by thioredoxin [
110].
Therefore, redox state has a profound impact on intracellular cell signalling [
111]. Not surprisingly, it is considered that ROS function as second messengers to regulate multiple metastasis-related signalling pathways by interacting with different proteins [
112]. From a biochemical standpoint, the oxidation of redox-sensitive cysteine and/or tyrosine residues located within or around the active site of many enzymes generates intra- and inter-protein bridges that affect their function [
113,
114]. These modifications generate a wide array of cellular responses [
115]. The possible mechanism involved in promoting targeted protein oxidation by H
2O
2 may involve the ability of ROS-scavenging enzymes, such as glutathione peroxidase (GPx), to sense and transduce the H
2O
2 signal, which is classified as a redox-relay mechanism. Another mechanism proposed is the so-called floodgate model in which oxidation causes inactivation of the ROS-scavenging enzymes by hyperoxidation or phosphorylation of key aminoacids, causing localized increases in H
2O
2 leading to protein oxidation and loss of function [
116].
ROS have been shown to regulate numerous signalling pathways (e.g., the MAPK and PI3K/AKT pathways) and activities of key transcriptional factors (e.g., hypoxia-inducible factor (HIF) and zinc finger protein SNAI1 (SNAI1)) to enhance cancer cell migration and invasion. Notwithstanding, ROS are also associated with epigenetic changes in genes. It is established that many transcription factors, including activator protein 1 (AP-1), hypoxia-inducible factor 1-alpha (HIF-1α), heat shock transcription factor 1 (HSF-1), nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB), nuclear factor erythroid 2-related factor 2 (Nrf-2), and tumour protein p53 (TP53) are activated by ROS and regulate the redox status of cells [
117]. The extent to which individual members of the abovementioned network of antioxidant transcription factors are differentially activated by oxidative stress is uncertain, although it is improbable that all are activated simultaneously. Rather, different transcription factors likely respond to distinct threshold levels of ROS/reactive nitrogen species (RNS), in a concentration- and/or time-dependent manner that is probably attuned to the coexistence of metabolic stress, proteotoxic stress, hypoxia, inflammation, or DNA damage. While these transcription factors have all been experimentally shown to be involved in carcinogenesis, more recent studies show that they also contribute to redox status and are implicated in tumour progression [
118].
From the extensive collection of experimental data summarized in this section, it is possible to conclude that the intracellular redox status of cancer cells has a profound impact on metastasis and tumour progression. In the coming sub-sections of this review, we will review and analyze the scientific evidence investigating the role of redox signalling along the different stages of metastasis.
Redox disbalance is a well-known feature of tumours. The involvement of free radicals in tumour cell signalling and cancer progression is evident from the data summarized and discussed in previous sections of this review, although most studies have focused on the presence/absence of free radicals and its impact on downstream signalling at defined steps of the metastatic cascade. To better understand the impact of redox signalling in tumour cell biology and disease progression, a wider picture is needed.
A higher but sub-lethal amount of free radicals has been observed in tumour cells actively proliferating, migrating, and invading the microenvironment since free radicals, mainly ROS, are known to act as second messengers that positively modulate the signalling pathways activating these biological processes. On the other hand, tumour cells have been shown to be highly dependent on efficient detoxification of free radicals through conserved mechanisms, such as autophagy, in other phases of the metastatic cascade, namely tumour cell dissemination and dormancy. Therefore, the question might not be whether free radicals do or do not promote tumour progression; rather, the molecular and cellular context along the metastatic cascade may be determinant in the matter. Indeed, the redox status of the cell is physiologically modulated through the interplay between the pro-oxidant and anti-oxidant systems to accommodate to the changing tissue microenvironment and different stages of cell development. The extensive, but not yet conclusive, scientific data on tumour cell redox biology suggest that cancer cells highjack the molecular processes conferring physiological cellular tolerance to disbalances in redox homeostasis as well as those that favor a motile phenotype. Metastasis competent tumour cells might benefit from the cellular plasticity endowed by the activation of redox signalling pathways throughout the metastatic cascade.
These considerations may have a clinical impact when applying redox-based therapies against cancer. Indeed, current and novel chemotherapeutic approaches against cancer are already known to modulate oxidative stress. Most chemotherapeutic agents generate ROS and are known to alter redox balance in tumour cells. In this sense, anthracyclines, alkylating agents, platinum coordination complexes, and camptothecins are widely used chemotherapeutic drugs that rise ROS levels in tumour cells [
367,
368]. For example, cisplatin induces ROS through mitochondrial DNA damage, which impairs the synthesis of proteins involved in the electron transport chain [
369]. Other chemotherapeutic drugs like paclitaxel and doxorubicin have been shown to promote oxidative stress in cancer cells contributing, at least in part, to tumour shrinkage due to tumour cell death [
370,
371]. Paclitaxel treatment showed an increase in superoxide, H
2O
2, and nitric oxide, as well as oxidative DNA adducts, G2-M arrest, and cells with fragmented nuclei, suggesting the involvement of ROS and RNS in paclitaxel cytotoxicity. In breast cancer, a proton pump inhibitor known as lansoprazole has been observed to increase ROS generation and supress tumour invasion. Treatment with NAC and diphenyleneiodonium, a specific inhibitor of NOX, significantly reduced lansoprazole-induced ROS accumulation [
372]. In melanoma cells, the isoquinoline alkaloid berberine stimulates ROS production which in turn regulates AMPK phosphorylation and activation leading to the decrease of ERK activity and COX-2 expression, finally reducing metastatic capacity of the cells [
373]. Imexon is a small prooxidant molecule that bind to cellular thiols and depletes the cysteine and glutathione stores, therefore increasing intracellular ROS. Imexon has been shown to effectively increase non-Hodgkin lymphoma cells to oxidative stress and demonstrated therapeutic benefit in a clinical trial [
374]. In this sense, buthionine sulphoximine (BSO), an inhibitor of the glutamylcysteine synthetase, has been observed to also contribute to depletion of cysteine and glutathione achieving antitumour activity in several types of cancer cells [
375]. The therapeutic implications of ketogenic diets through the induction of oxidative stress and in the context of its possible role in the enhancement of radio-chemotherapy responses in lung cancer xenografts have also been studied [
376]. This type of diet increases dietary fatty acids, mainly PUFA levels, whose incorporation into membrane phospholipids increase cancer cells susceptibility to the accumulation of lipid peroxides and subsequent ferroptosis [
377]. Moreover, the inhibition of the activity of enhanced antioxidant system in cancer cells may induce intracellular oxidative stress, for example, with the employment of drugs that block the proteasome or cause ER stress [
378].
On the other side, tumour treatment through antioxidant-based therapies has also been investigated. In testicular cancer, an antioxidant cocktail of α-tocopherol, l-ascorbic acid, zinc, and selenium was used to modulate the expression of metastasis-associated protein 1 (MTA1), a gene involved in tumour growth and metastasis. The antioxidant cocktail effectively inhibited the expression of MTA1 and increased the susceptibility of tumour cells to apoptosis, suggesting that antioxidants may be helpful for metastasis prevention [
379]. Furthermore, many clinical trials have been developed in order to evaluate the potential of dietary supplementation with antioxidants as suppressors of tumour development [
380]. However, the context-dependent nature of redox signalling in cancer progression may contribute to the frequent contradictory results obtained in clinical trials evaluating the effect of adjuvant antioxidant therapies, in which the stage of the disease at which antioxidants are administered might influence the clinical outcome of the intervention. Therefore, much of the knowledge that we have acquired through research focused on redox signalling in cancer points to a stage-tailored strategy to develop redox-based therapies against cancer, conferring a “temporal dimension” to precision medicine.