Manganese superoxide dismutase (Sod2) and redox-control of signaling events that drive metastasis

Matrix Metalloproteinases and Invasion

Matrix metalloproteinases (MMPs) are Zn²⁺ and Ca²⁺ dependent endopeptidases involved in extracellular matrix (ECM) turnover. To date, 23 known human MMPs have been identified; many of these share a five domain architecture comprising of a signal peptide, prodomain region, catalytic domain, hinge region and the hemopexin domain. Based upon their substrate specificity and structural differences, the MMPs are broadly classified into five groups – collagenases, gelatinases, stromelysins, matrilysins and membrane-type MMPS (MT-MMPs). MMPs that differ from the general five-domain structure are: gelatinases (MMP-2 and 9), which have a fibronectin domain present within the catalytic domain; matrilysins (MMP-7 and 26), which lack the hinge and hemopexin domain; and MT-MMPs that have a transmembrane domain and cytoplasmic tail. All MMPs are secreted as zymogens into the extracellular space with the exception of MT-MMPs, which have a furin recognition sequence at the end of prodomain region that allows for MT-MMPs intracellular activation [75].

In tissues, MMP levels are extremely low and strictly regulated due to their ability to cleave a wide variety of substrates as they modulate many physiological processes including wound healing, cellular growth, embryogenesis, immune surveillance and bone remodeling [76]. Elevated MMP levels have been associated with a variety of pathological conditions such as rheumatoid and osteo-arthritis, lung emphysema, atherosclerosis, Alzheimer's, and cancer. Overexpression of MMPs by tumor cells and/or stromal cells infiltrating the tumor can lead to the remodeling and degradation of the extracellular matrix and basement membranes, subsequent tumor cell invasion and metastasis[77]. Similarly, Sod2 mediated increases in cellular oxidant/anti-oxidant ratios are directly correlated with tumor progression, angiogenesis and migration and invasion. Zhang et al. (2002) demonstrated that overexpression of Sod2 in estrogen-dependent human breast cancer MCF-7 cells resulted in activation of MMP-2 expression and a corresponding increase in ROS [78]. Co-expression of catalase or glutathione peroxidase in the Sod2 overexpressing cells reversed this effect. Conversely, in the highly metastatic estrogen-independent breast cancer MDA-MB231 and SKBR3 cells, basal Sod2 levels are significantly higher as compared to the MCF-7 cells, with catalase and peroxiredoxin 3 expression significantly lower [79]. This augmented expression of Sod2 is associated with increased MMP-9 activity and greater invasiveness both of which are reduced by expression of anti-sense Sod2 RNA or treatment of the cells with the H₂O₂ scavengers NAC or pyruvate. These studies support that Sod2-mediated H₂O₂ production result in MMP dysregulation and subsequent metastasis.

Our group has demonstrated that overexpression of Sod2 in HT-1080 fibrosarcoma cells significantly enhanced migration and invasion both in vitro and in vivo [72]. The severity of invasion was directly correlated to the Sod2 levels; HT1080 cells expressing high levels of Sod2-GFP exhibit greater invasiveness, a loss of focal adhesions and an increase in lung metastasis as compared to cells expressing low levels of Sod2-GFP. In tumor cells isolated from pulmonary metastases, those that expressed high levels of Sod2-GFP exhibited greater motility when examined by time-lapse microscopy. Concomitant increases in MMP-1 levels that also correlated with Sod2 levels were observed. The dependence of MMP-1 expression and the metastatic phenotypes associated with high Sod2 expression were abrogated upon co-expression with catalase targeted to the mitochondria.

The ability of Sod2 overexpression to modulate invasion is not restricted to HT-1080 fibrosarcoma cells and was also observed in 253J transitional bladder carcinoma cells [72]. Similar to the HT-1080 fibrosarcoma cells, co-expression of catalase in the 253J bladder carcinoma cells reversed the severity of the invasive phenotype. The highly metastatic bladder cancer cell line 253J-BV, which was derived from the 253J-P parental cell line, exhibited enhanced expression and activity of Sod2 as compared to the parental cells [36,80]. The increase in Sod2 expression was accompanied by a significant decrease in catalase activity, resulting in a net increase in H₂O₂ production in the 253J-BV cell line. Expression of MMP-9 was upregulated in the metastatic line and was attenuated by overexpression of catalase. Similarly, expression of catalase effectively reduced the clonogenic activity of 253J-BV cells.

Mechanistically, the regulation of MMP-1 by Sod2 seems to be mediated in part at least, at the level of transcription [81-85,85]. Based upon promoter conformation, MMPs have been categorized into three groups [84]. MMPs in group 1 have promoters that contain a TATA box in the −30 bp region with a proximal Activator Protein-1 (AP-1 site) at −70 bp (MMP-1, 3, 7, 9, 10, 12, 13, 19 and 26). Members in group 1 are transcriptionally activated following IL-1, TNF-α and phorbol ester treatment [86]. MMPs in group 2 have a TATA box without an AP-1 binding site (MMP-8, 11 and 21) whereas MMPs in group 3 lack all of the above elements (MMP-2, 14 and 28) [84]

Regulation of MMP-1 expression by Sod2 is influenced by a genetic variation in the MMP-1 promoter [73,82,83]. We have demonstrated that a single nucleotide polymorphism (SNP) located at −1607 bp, creates an Ets family transcription factor binding site of the sequence 5′-GGAT-3′ by the insertion of a guanine base (G) and confers Sod2-dependent MMP-1 promoter activity [73]. In HT-1080 fibrosarcoma cells, Sod2 overexpression also increases the mRNA levels of MMPs -2, -3, -7, -10, -9, -11, however, we have not yet identified promoter elements responsible [73]. Ets transcription factors normally do not bind DNA alone, but preferentially form coactivator complexes with transcription factors, such as AP-1. The Ets and AP-1 coactivating complex may also play a critical role in regulating the expression of various MMP family members, particulary that of MMP-1 [87]. The protooncoproteins Fos and Jun, which comprise the AP-1 complex, can homo or heterodimerize and bind its cognate consensus sequence (TGACTCA) in the regulatory domains of many genes including various MMP family members [88].

Numerous reports indicate that transcription factors important for MMP-1 expression are sensitive to redox-activation that occurs, in part, through activation of the MAPK family members ERK1/2 and JNK [89-91]. We have established that JNK confers redox-sensitivity to the MMP-1 promoter while both ERK and/or JNK are required for maximal basal promoter activity and the expression of AP-1 and Ets-1 [82]. In addition, both c-Jun and Ets-1 are recruited to the region of the MMP-1 SNP in response to alterations in the steady state production of H₂O₂.

Other lines of evidence suggest that H₂O₂-generated by the Sod2 is important to enhanced expression of MMP-1. Scharffetter-Kochanek and co-workers demonstrated that overexpression of Sod2 in dermal fibroblasts caused a profound increase in the expression of MMP-1 when superoxide was provided to the cells [92]. The investigators established that the induction of MMP-1 is due to the increase in the production of H₂O₂ as a result of the dismuting function of Sod2. This is further supported by work from our group that indicates binding of Ets-1 to the MMP-1 promoter and sustained JNK signaling requires the enzyme activity of Sod2. Furthermore, Sod2 is sufficient and required for the induction of MMP-1 and potentially other MMP family members via the H₂O₂-dependent activation of MAPK signaling [73,81,82].

Redox sensitive metastatic signaling networks: regulation of phosphatases

While we have extensively shown that enhanced expression of Sod2 leads to changes in MMP expression, our lab is interested in the mechanisms by which Sod2 expression can influence other pro-metastatic signaling event, potentially as a consequence of changing the redox environment of metastatic cancer cells. While Sod2 expression has been characterized to inhibit proliferation in some tumor cell types, an Sod2-dependent change in redox state appears to promote pro-migratory and -invasive signaling events.

ROS can act as second messengers via the oxidation and inactivation of Phosphatases. The active site cysteine residues responsible for phosphoryl transfer are particularly susceptible to oxidation at physiological pH, due to their low pKa (For review see [93]). This oxidation is reversible and dependent on the thiol status of cells and can result in significant changes in cellular signaling.

PTEN (Phosphatase and Tensin Homolog) is a dual protein and lipid phosphatase that has been extensively studied as a tumor suppressor gene. Loss or functional mutants of PTEN are associated with a number of cancers and poor prognosis [94]. The lipid phosphatase activity of PTEN results in dephosporylation of PtdIns(3,4,5)P3 to PtdIns(4,5)P2, essentially regulating the phosphoinositide 3-kinase (PI3) Kinase signaling pathway by limiting PtdIns(3,4,5)P3 levels. PtdIns(3,4,5)P3 is an important regulatory lipid, functions of which range from involvement in migration to angiogenisis. Redox dependent changes of the PI3-kinase pathway have been attributed to the reversible oxidation and inactivation of PTEN [95]. We have found that HT-1080 fibrosarcoma cells that express Sod2 display more oxidation of this phosphatase than control transfected cells [96]. Further, oxidation of PTEN can be reversed by catalase expression, potentially implicating an increase in H₂O₂ as the primary ROS involved. PTEN oxidation is enhanced in the metastatic cell line, 253J-BV, which displays an increase in endogenous Sod2 expression, low catalase expression and an accompanying increase in H₂O₂ levels, when compared to it's non-metastatic parental line 253J-P (Hempel, Ye and Melendez, unpublished data). In the HT-1080 Sod2 over-expression model enhanced PTEN oxidation leads to a redistribution of PtdIns(3,4,5)P3 to the membrane, and leads to enhanced Akt signaling and increased angiogenesis, which were all Sod2 dependent [96].

We and others are actively investigating how increases in mitochondrial Sod2 and potential shifts in steady state [H₂O₂] lead to oxidation of phosphatases in membrand and cytosolic compartments.. While largely unknown, possibilities include the ability of oxidants to diffuse from the mitochondria to broad areas within the cell and/or the physical distribution of mitochondria. Mitochondria are often localized to leading edges of migrating cells, presumably to provide an energy source. Whether this also reflects site specific increases in mitochondrial ROS at the membrane remains to be elucidated. The importance of oxidants influencing cellular distribution of signaling players is highlighted by preliminary data from our laboratory. We have found that the oxidized pool of PTEN is recruited away from the membrane, it's primary site of action, and more abundantly found throughout the cytosol (Hempel and Melendez, unpublished data). This suggests that oxidation not only leads to a decrease in activity, but a change in cellular localization of this phosphatase.

Enhanced PtdIns(3,4,5)P3 distribution to the leading edge of cells has been implicated in regulating directionality and migration [97]. It is possible that cell surface accumulation of PtdIns(3,4,5)P3 in Sod2 over-expressing cells contributes to their enhanced migratory phenotype. Evidence from PTEN mutagenesis studies however suggest that some of the metastatic regulatory properties of PTEN may not be solely due to its lipid phosphatase activity. Leslie and colleagues have shown that the protein tyrosine phosphatase activity of PTEN negatively regulates epithelial to mesenchymal (EMT) transition [98]. Whether shifts in the Sod2-mediated oxidation of PTEN also drive EMT remains to be established. However, evidence from TGF-β induced EMT in human renal epithelial cells, suggests that Sod2 expression is down regulated during this transition [99]. Nonetheless, Sod2-mediated redox shift enhances Akt signaling. Sod2 expressing cells increase their angiogenic activity, a key contributor to metastatic spread and tumor growth [96]. Enhanced angiogenesis was accompanied by Sod2-dependent increases in VEGF expression [96]. The Sod2-dependent angiogenic switch was reversed by catalase expression, suggesting a role for H₂O₂ in this process. In 253J-BV cells, which display endogenously high Sod2 levels, we have similarly shown that enhanced VEGF levels and MMP-9 levels are dependent on a high intracellular H₂O₂ millieu [36]. Again, catalase expression could abrogate expression of these pro-metastatic genes.

Other phosphatases involved in pro-migratory signaling have similarly been shown to be redox regulated. We have investigated the role of Sod2 expression on changes in Focal Adhesion Kinase (FAK) signaling and found that the FAK-signaling cascade is activated in cells with high Sod2 expression. FAK signaling has also been shown to be H₂O₂-dependent, as catalase co-expression could reverse this effect, mediated via oxidation of phosphatase PTPN12 (Hempel and Melendez, unpublished data). Similar to Sod2, expression of Lysyl oxidase (LOX) has been shown to induce migration in a H₂O₂ dependent manner [100]. It was shown that FAK and Src activation were increased in LOX transfected breast cancer cells and catalase could abrogate the effects on migration and phosphorylation of the pro-migratory signaling kinases.

Other phosphatases that may be regulated due to a Sod2-dependent altered redox status likely include the Map Kinase phosphatases (MKP) that regulate MAPK signaling. MKPs are similarly sensitive to redox-dependent inactivation and this oxidation participates in the senescence-associated increase in MMP-1 expression [101]. MKPs are known targets of oxidation which leads to enhanced proteasomal degradation [102]. It is exciting to speculate that a similar signaling network participates in the Sod2-mediated increase in MMP expression in tumor cells.

Protection from Apoptosis

Many studies have shown that Sod2 is an important enzyme that protects cells from radiotherapy induced apoptosis. Here the action of Sod2 has primarily been attributed to it's antioxidant activity, rather than shifting the redox balance to a pro-H₂O₂ state. It has been shown that Sod2 stabilizes the mitochondrial membrane and leads to decreased caspase3 activation when cancer cells are treated with ionizing radiation, compared to cells that lack high Sod2 expression [103]. In addition Sod2 may inhibit tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-induced apoptosis by inhibiting cytochrome c and Smac/DIABLO release [104]. The authors hypothesize that within metastatic tumor cells the high levels of mitochondrial ROS, which normally drive release of Smac/DIABLO, are neutralized by Sod2, to prevent apoptosis. The high levels of Sod2 in tumor cells are therefore detrimental to chemo- and radiation therapy and investigators have proposed the use of Sod2 inhibitors as adjuvant therapy to radiation.

Wong and Goeddel were first to report that Sod2 was induced in response to tumor necrosis factor (TNF) and interleukin-1, as a protective mechanism from cytokine induced cell death [105,106]. Enforced Sod2 expression has been shown to confer resistance to numerous apoptotic stimuli [107-110]. The antiapoptotic effects of Sod2 are attributed to its dismuting activity. Recent work suggests that increases in steady state [H₂O₂] resulting from the overexpression of Sod2 are responsible for restricting TNF induced apoptotic cell death [111]. Regardless, of mechanism of protection afforded by Sod2 overexpression it does not precisely fit its role as a tumor suppressor as it would serve to prolong tumor survival and resistance to chemotherapeutic strategies. Recent work indicates radiation-resistance of cancer stem cells is attributed to their increased free radical scavenging capacity [112]. Chemoresistance of urothelial carcinoma cancer stem cells is associated with a high level expression of Sod2 [113,114]. Enhanced ROS scavenging confers genoprotection and allows cancer stem cells the unique ability to colonize new sites and survive after chemo- or radiation therapy. Thus, increased Sod2 expression may promote cancer stem cell survival and subsequent metastases.

Conclusions

It has been more than 30 years since Oberley and Buettner made the connection between low enzymatic Sod2 activity and cancer [1]. The idea that loss of Sod2 would potentiate tumorigenesis was well founded principally due to the wealth of information linking oxidants to tumor promotion [115]. This early observation has stood the test of time and it is now clear that in many tumor types suppression of Sod2 is regulated at both the genetic and epigenetic level [54,116]. While loss of Sod2 sensitizes tumor cells to free radicals associated with both radiation and chemotherapeutic it is equally not surprising that metastatic cancers that resist traditional ionizing and chemotherapies display an increase dismuting capacity. Why this increase is primarily restricted to Sod2 is not clear? Sod2 serves to preserve mitochondrial integrity but this is not likely the case as loss of mitochondrial genome integrity precedes progression to metastatic disease and is associated with increased ROS production [117]. Thus, increases in Sod2 levels may help cope with increased metabolic oxidant production. Alternatively, metastatic tumors cells are continuously exposed to inflammatory stimuli which can drive Sod2 expression. While possible, this does not explain the high level expression of Sod2 in isolated metastatic tumor cell types. Thus, many metastatic tumor cell types are programmed for high level Sod2 expression independent of inducing stimuli. There may be countless reasons for the increased Sod2 that is associated with a diverse array of malignant lesions and metastatic tumor cell lines. We feel it is the resulting consequences of increased Sod2 expression that further propagate the malignant phenotype as discussed in detail on the preceding pages and outlined in Figure 2. So far the pro-metastatic profile associated with increases in Sod2 expression has been attributed to increases in steady state [H₂O₂]. H₂O₂ can restrict the activity of many protein tyrosine phosphatase family members leading to an inherent increase in basal kinase activity. Our studies have established that Sod2-mediated increases in [H₂O₂] lead to oxidative inactivation of PTEN, Map kinase phosphatases and PTPN12 that play an important role in restricting invasion/migration, matrix turnover, cell proliferation, angiogenesis and cell survival. It is likely that elevated steady state [H₂O₂] resulting from Sod2 impact a broad array of signaling networks all of which exacerbate the malignant phenotype. The fact that many or all of these malignant properties are reversed by catalase co-expression makes them amenable to targeted and precise antioxidant intervention.

Interestingly, Oncomine data mining revealed that catalase levels are often decreased in a number of cancer types and expression appears to decrease further with progression to metastatic disease (Oncomine.org)[118]. Catalase levels are negatively regulated by a number of mechanisms, including oxidant dependent promoter hypermethylation [119], phosphorylation and consequential proteosomal degradation[120]. One could speculate that an increase in Sod2 activity and a concomitant decrease in catalase activity may further shift the steady state level of [H₂O₂] within tumor cells.

Recent studies have shown that Nox4 levels are increased in tumors of the breast, ovary and melanoma, and that this enhanced expression leads to an increased tumorigenic phenotype[121,122]. Further, Graham et al. reported that this enhanced Nox4 dependent tumorigenicity is dependent on Nox4 localization to the mitochondria and a concomitant increase in mitochondrial ROS production [121]. Whether an increase in mitochondrial Nox4 levels within metastatic tumor cells provides the source for enhanced O₂^.− production, which are consequentially rapidly dismuted by Sod2 within these cells, thereby resulting in a change in [H₂O₂] levels, remains to be elucidated.

While we have experimental evidence indicating that steady state [H₂O₂] levels increase in response to Sod2 expression, the exact biochemical rational for this shift is beyond the scope of this review and we refer the reader to reviews by Dr. Gary Buettner and Dr. Irwin Fridovich on this topic

The differences in pro- and anti-oxidant activity of Sod2 may also be related to dose dependent effects of Sod2 expression. For example, Wang et al. reported that Sod2 has a biphasic role on Hypoxia Inducible factor-1α (HIF-1α) regulation. Low level increases in Sod2 expression (2-6-fold) in MCF-7 breast cancer cells suppressed HIF-1α accumulation under hypoxic conditions[123]. The same authors later showed that this suppression was due to the dismuting activity of Sod2, as accumulation of O₂^.− following Sod2 knock-down was able to enhance HIF-1α levels[124]. On the contrary, when Sod2 was expressed at levels higher than 6-fold, HIF-1α levels increased in a H₂O₂ dependent manner, as peroxide removal from the mitochondria could reverse this effect[125]. We did not observe these biphasic effects of Sod2 expression on migration of HT-1080 cells. Low levels of Sod2 expression (2.5-fold increase in Sod2 activity) were able to drive migration of HT-1080 cells and additional increases in Sod2 expression lead to further enhancement of this migratory phenotype[126]. However, we have observed a biphasic role for Sod2 in the expression of the cytokine IL1α. At low Sod2 concentrations IL1α transcript levels are decreased [127], yet at higher Sod2 levels IL1α expression is H₂O₂-dependent.

There is a need to further our understanding of these complex biochemical mechanisms that govern the pro-oxidative and anti-oxidative nature of Sod2 during tumorigenesis. This knowledge will shed light on the roles of different oxidant species within tumor cells, as well as the consequences of altering antioxidant enzyme levels such as Sod2. It will also aid in the development of more targeted antioxidant and Sod2 mimetic therapies, considering that certain Sod2 mimetics may exert both pro- and anti-oxidative actions within tumor cells (review by Ines Batinic-Haberle in this issue). In addition, it is important to keep in mind that cancer stem cells bolster their free radical scavenging capacity for survival. Thus, it is not surprising that nutritional antioxidant supplementation has failed in cancer prevention studies as they have broad ranging impact and serve to scavenge all oxidants regardless of whether they are detrimental or beneficial to tumor cell survival. The advent of specific synthetic free radical scavengers which do not give rise to free radical propagating species may hold great promise in treatments targeted at metastatic disease.