Targeting prooxidant MnSOD effect inhibits triple-negative breast cancer (TNBC) progression and M2 macrophage functions under the oncogenic stress

Triple-negative breast cancer (TNBC) has been shown with high mitochondrial oxidative phosphorylation and production of reactive oxygen species (ROS). MnSOD (SOD2) is a mitochondrial antioxidant defense that has been implicated in inhibition of human malignancies. However, the impact of MnSOD on immunosuppressive macrophage functions and TNBC aggressiveness has never been explored. We found here that SOD2 is primarily observed in the aggressive subtypes of HER2(+) breast cancers and TNBCs patients. Further analyses demonstrated that the oncoprotein multiple copies in T-cell malignancy-1 (MCT-1 or MCTS1) induces mitochondrial superoxide dismutase (MnSOD) in TNBC cells by stabilizing the transcription factor Nrf2. SOD2/MCTS1 expression correlates with a poor prognosis in breast cancer patients. MnSOD in TNBC cells functions as a prooxidant peroxidase that increases mitochondrial ROS (mROS) and adaptation to oxidative stress under the oncogenic effect. Interleukin-6 (IL-6) in the MCT-1 pathway elevates Nrf2/MnSOD and mROS levels. Knockdown of MnSOD inhibits TNBC cell invasion, breast cancer stem cells (BCSCs), mROS, and IL-6 excretion promoted by MCT-1. TNBC cells deficient in MnSOD prevent the polarization and chemotaxis of M2 macrophages but improve the ability of M1 macrophages to engulf cancer cells. Quenching mROS with MitoQ, a mitochondria-targeted non-metal-based antioxidant MnSOD mimics, effectively suppresses BCSCs and M2 macrophage invasion exacerbated by MnSOD and MCT-1. Consistently, silencing MnSOD impedes TNBC progression and intratumoral M2 macrophage infiltration. We revealed a novel stratagem for TNBC management involving targeting the MCT-1 oncogene-induced mitochondrial prooxidant MnSOD pathway, which prevents the development of an immunosuppressive tumor microenvironment.


INTRODUCTION
Triple-negative breast cancers (TNBCs) are characterized by a lack of hormone receptors for estrogen (ER), progesterone (PR), and human epidermal growth factor 2 (HER2) [1]. Patients with TNBC not only have a poor prognosis and high incidences of relapse and metastasis [2], but also frequently develop drug resistance [1,3]. Finding novel targetable biological features or therapeutic regimens for TNBCs thus become a high priority.
Cancer aggressiveness is reinforced by the interaction between tumor and host immune cells that secrete protumorigenic chemokines to support the development of an immunosuppressive tumor microenvironment (TME) [4]. Tumorassociated macrophages (TAMs) are the predominant immune cells infiltrating solid tumors [5,6]. TAMs are recruited to and differentiate in primary or metastatic sites, depending on the inflammatory cytokines or chemokines secreted by tumoral or stromal cells [6]. TAMs are mainly classified as inflammatory M1 and protumorigenic M2 macrophages. M1 macrophages act as innate host defense with tumoricidal functions via the production pro-inflammatory cytokines [4]. Conversely, M2 macrophages secrete anti-inflammatory cytokines and highly express immunosuppressive receptors, which enhance tumor growth and immune escape [4][5][6].
Oxidative stress creates intrinsic insults in carcinogenesis by inducing genetic instability, activating growth factor-dependent pathways, and disrupting aerobic metabolism [7]. To mitigate the impact of mitochondrial reactive oxygen species (mROS), mitochondrial superoxide dismutase (MnSOD or SOD2) controls superoxide radical anions (O 2 . − ) derived from electron transport chain byproduct and converts O 2 . − into the diffusible strong oxidant H 2 O 2 [8]. Since elevated ROS cause cancer [9], increased MnSOD are expected to suppress tumors. However, contrary to this expectation, recent studies have shown that MnSOD promotes signaling cascades that support malignant transformation [10], cell survival [11], and cancer stemness capacity [12], which overtake its oxidative defense mechanism. In multiple myeloma cells, MnSOD inhibits the binding of the transcription factor AP-1 to regulate proinflammatory cytokines interleukin 6 (IL-6 or IL6) [13]. The IL-6 promoter contains an antioxidant response element (ARE) [14], thus suggesting crosstalk between IL-6 and ROS scavenger activity, such as MnSOD. Nevertheless, the impact of MnSOD on tumor immunity remains unclear.
Further elucidating the role of the MCT-1/MnSOD axis in the oxidative TME, we now verify that MCT-1 shifts the antioxidant role of MnSOD toward a prooxidant effect via IL-6 signaling that elevates mROS in aggressive TNBC cells, which enhances M2 macrophage functions and TNBC expansion.

Enhancement of MCT-1 and MnSOD is a poor prognostic marker in aggressive breast cancer
Using the Kaplan-Meier (KM) Plotter database [21] to inspect the clinical relevance of MCT-1 and MnSOD expression in breast cancer, we found that patients with high MCTS1 (Fig. 1A) or high SOD2 (Fig. 1B) expression showed a lower relapse-free survival (RFS) rate than those with low expression of MCTS1 or SOD2. We previously identified that high MCT-1 expression is largely observed in breast cancer patients at both the initial and the late stages, and found in over than 70% of ER(+)/HER2(+) subtype and TNBC [20]. From the Oncomine cancer microarray database [22], we found that MCTS1 and SOD2 expression was positively correlated in breast cancer patients (p < 0.0001) (Fig. 1C). Further characterization of SOD2 mRNA levels was conducted with breast cancer cDNA arrays (OriGene). SOD2 expression was enriched among stage II (3.57-fold) and advanced-stage (III/IV) (7.36-fold) tumor tissues (Fig. 1D) relative to normal breast tissues. High SOD2 expression was detected in 80% of breast cancer patients (Fig. 1E) and was positively correlated with MCTS1 (p < 0.001). C The correlation of MCTS1 and SOD2 gene expression in breast cancer patients (n = 159) was analyzed using the Pawitan dataset in the Oncomine cancer profiling database. D-F SOD2 expression was analyzed in TissueScan cDNA arrays with normal breast tissue (n = 11) and breast tumor biopsies (n = 124) across distinctive tumor stages (D, E), molecular subtypes and TNM classifications (F). The SOD2 mRNA levels in tumor samples were normalized to internal ACTB (β-actin) mRNA levels and then compared with the levels in normal breast samples. Data are presented as the mean ± s.e.m. G KM Plotter was used to estimate the survival of patients with MCTS1 high /SOD2 high (n = 37) expression compared with that of those with MCTS1 low /SOD2 low (n = 60) expression using the Pawitan dataset in the Oncomine database. H, I RFS was estimated for ER-negative (n = 434) (H) and basal-like (n = 309) (I) breast cancer patients with MCTS1 high /SOD2 high or MCTS1 low /SOD2 low expression using KM Plotter. Statistical analysis was performed using the log-rank Mantel-Cox test (A, B, and G-I), Pearson product-moment correlation coefficient (C), Kruskal-Wallis test followed by Dunn's multiple comparison test (D) and χ 2 test (E, F).

MCT-1 stabilizes Nrf2 to transcriptionally induce MnSOD
To examine how MCT-1 promotes MnSOD, V5-tagged MCT-1 was introduced into nontumorigenic human mammary epithelial cells (MCF-10A) and the TNBC cell lines HCC1395 (derived from early stage TNBC) and MDA-MB-231 (IV2-3) (a highly invasive subline derived from two rounds of in vivo selection of lung metastases) [23]. We found that MCT-1 overexpression elevated the MnSOD protein level ( Fig. 2A) and the level of nuclear factor-E2-related factor 2 (Nrf2 or NFE2L2), a transcriptional inducer of MnSOD. MCT-1 enrichment also led to increased SOD2 mRNA in MCF-10A cells and the TNBC cell lines (Fig. 2B). Conversely, Nrf2 and MnSOD were reduced while MCT-1 silencing in MCF-10A (shMCT-1 #1 and #2) and MDA-MB-231 (IV2-3) (shMCT-1 #2 and #3) cells (Fig. 2C, D). However, overexpressing or silencing MCT-1 did not affect catalase ( Fig. 2A, C, D), the H 2 O 2 scavenger. E Upon cycloheximide (CHX) treatment for the indicated times, Nrf2 protein stability and its amount relative to that of internal ACTB (right) were compared in MDA-MB-231 cells (control vs. MCT-1). Representative images (left) from four independent experiments are shown. F, G NFE2L2 mRNA levels were analyzed across distinctive breast tumor stages (F) and molecular subtypes or TNM classifications (G) using TissueScan breast cancer cDNA arrays. The NFE2L2 mRNA levels in tumor samples were normalized to the internal ACTB mRNA level and then compared with the levels in normal breast samples. H-I KM Plotter analysis estimated relapse-free survival in basal-like (H) breast cancer patients with MCTS1 high /NFE2L2 high (n = 155) expression versus those with MCTS1 low /NFE2L2 low (n = 154) expression and in breast cancer patients with lymph node metastasis (LN positive) (I) stratified by a SOD2 high /NFE2L2 high (n = 283) or SOD2 low / NFE2L2 low (n = 283) signature. J The correlation of SOD2 and NFE2L2 expression in TNBC patients was analyzed using the Stickeler dataset (n = 32) in the Oncomine database. Data (B, E) are presented as the mean ± s.e.m. Statistical analysis was performed using a two-tailed unpaired Student's t-test (B), linear regression followed by an analysis of variance (ANOVA) (E), the χ 2 test (F, G), the log-rank Mantel-Cox test (H, I), and the Pearson product-moment correlation coefficient (J). *p < 0.05; **p < 0.01; ***p < 0.001.
Constitutive Nrf2 activity occurs in cancer cells via either elevated Nrf2 transcription or disrupted Nrf2 turnover [24]. We found that NFE2L2 mRNA were unchanged in both the MCF-10A and MDA-MB-231 (IV2-3) cell lines with or without MCT-1 overexpression ( Supplementary Fig. S1A). However, upon inhibition of de novo protein synthesis with cycloheximide, MCT-1 overexpression extended the half-life (t 1/2 ) of Nrf2 to approximately two times longer than that observed with control cells (Fig. 2E). Thus, MCT-1 upregulates Nrf2 by enhancing its stability, which in turn activates MnSOD transcription.
Complex I is the key enzyme in the mitochondrial electron transport chain and is the major source of mROS [28]. To investigate the nexus between mROS and the IL-6/MnSOD circuit, MDA-MB-231 (IV2-3) cells were treated with diphenyleneiodonoium (DPI), an inhibitor of complex I that targets the flavin prosthetic group region [28], to block reverse electron transport-ROS formation. DPI reduced mROS production ( Supplementary  Fig. S4B) and decreased Nrf2 (Fig. 4G), MnSOD and IL-6 promoted by MCT-1. However, MDA-MB-231 (IV2-3) cells treated with a different inhibitor of complex I that targets the iron-sulfur cluster N2 [28], Rotenone (Rot), enhanced mROS production during forward electron transfer ( Supplementary Fig. S4C) and consequently induced Nrf2, MnSOD and IL-6 in a time-dependent fashion (Fig. 4H), especially in MCT-1-overexpressing cells. Thus, the enhanced mROS/IL-6/MnSOD signaling loop is existed in TNBC cells.

DISCUSSION
MnSOD is typically known as a tumor suppressor due to its antioxidant role in protecting cells from oxidative damage [9]. However, the roles of MnSOD in TME modulation have been understudied. We now show that the oncogene MCT-1 induces Nrf2/MnSOD and mROS. MnSOD impacts TNBC cell aggressiveness, but MnSOD deficiency suppresses M2 macrophage phenotypes, the TNBC cell malignancies, TNBC progression, and M2 infiltration.
Prior investigation has shown that inducing oncogenic v-Src kinase in MCF-10A cells potentiates MnSOD expression [11,12]. High MnSOD expression in MCF-7 luminal breast cancer cells increases H 2 O 2 [11], which sustains the Warburg effect. Using a multistage skin carcinogenesis mouse model with human MnSOD promoter-enhancer elements [10], MnSOD transcripts upregulated at the advanced stage due to p53 loss. This also matches our findings that oncogenic MCT-1 increased MnSOD in p53-mutant aggressive MDA-MB-231 (IV2-3) cells, and MCT-1 reduces wild-type p53 in MCF-10A cells and invasive A549 cancer cells [19], leading to MnSOD induction. Besides, p53 binding sequences in the MnSOD promoter determine MnSOD expression in fibroblasts [35].
We found that MCT-1 overexpression drives the prooxidant function of MnSOD, in turn prompting TNBC cells to fight against and adapt to the oxidative environment. Similarly, increased A.T. Al Haq et al.
acetylation of MnSOD on the lysine 68 (K68) residue can enhance prooxidant MnSOD activity [32], thereby exacerbating stemness reprogramming [12]. This current study concomitantly identifies that the prooxidant peroxidase MnSOD expands BCSCs and invasiveness in high MCT-1 conditions by augmenting mROS, and these malignant phenotypes can be reversed by quenching mROS with MitoQ. Diminishing mROS or silencing MnSOD prooxidant effect could be a novel therapeutic regimen for aggressive breast cancer.
We uncovered that MnSOD induction in the oncogenic MCT-1 occurs via Nrf2 stabilization. Comparably, under oncogenic K-Ras stimulation, increased transcription or protein stability of Nrf2 enhances pancreatic and mammary cancer progression in an mROS-dependent or -independent manner [33,36]. Although Nrf2 binds to the ARE in the promoter region of multiple genes that control ROS detoxification [24], we found that MCT-1 overexpression does not alter catalase in TNBC cells. Likewise, increased MnSOD without catalase increment results in induced oxidative stress in fibroblasts and lymphoblasts [35].
Importantly, we demonstrated for the first time that the M2 macrophage migration and invasion can be diminished by shMnSOD, MitoQ, or shMCT-1. In chemically induced colitis, mROS production in colon tissues activates the NF-κB pathway [39], leading to the enhanced recruitment of M2 macrophages and regulatory T cells. IL-4-induced M2 macrophages are susceptible to MnTE-2-Pyp 5+ , a redox-active drug that mimics the activity of superoxide dismutases [40]. Thus, mROS inhibitor could boost the macrophage-based anti-tumor immunity.
Our study also first indicated that shMnSOD and shMCT-1 enhanced the phagocytosis of TNBC cells by M1 macrophages, signifying that phagocytosis can be controlled by redox homeostasis and oncogenic signaling. Consistently, M1 macrophage phagocytosis was increased in the MitoQ-treated TNBC cells, indicating that depleted mROS facilitate M1 macrophage tumoricidal activity. H 2 O 2 can travel from cell to cell and suppress the activation of bone marrow-derived macrophages [41]. Alternatively, H 2 O 2 induces TNF-α production in macrophages and triggers M1 polarization by activating the p38/JNK pathway [42]. Hence, balancing mROS and restricting mROS outflow in aggressive cancer cells could help immune surveillance and modify the TME.
In conclusion, our results provide new insights into the mechanism by which the oncogene MCT-1 deregulates redox homeostasis by changing the antioxidant MnSOD to a prooxidant that produces excess IL-6 and mROS and promotes an immunosuppressive TME. Targeting MnSOD or mROS with redox-active drugs could facilitate the engulfment of cancer cells by M1 macrophages and inhibit M2 macrophage functions, EMT progression, cancer stemness, and tumor advancement. , and HCC1395 cells as previously described [20], and the transfectants were maintained in complete medium containing 100 or 200 μg/ml neomycin (G418) (#345810, Millipore, Darmstadt, Germany). The macrophage-mediated phagocytosis. Data are presented as the mean ± s.e.m. Statistical analysis was performed using a two-tailed unpaired Student's t-test (A, D), the log-rank Mantel-Cox test (C), two-way analysis of variance (ANOVA) followed by Tukey-Kramer post hoc analysis (B, E), and the Spearman's rank correlation coefficient (Rho) (F). *p < 0.05; **p < 0.01; ***p < 0.001.

MATERIALS AND METHODS Cell culture and transfection
MCT-1 gene was stably knocked down in MCF-10A, 4T1, and MDA-MB-231 (IV2-3) cells as previously described [20], and the transfectants were maintained in complete medium containing 0.5 or 10 μg/ml puromycin (#A1113802, Millipore) or 8 μg/ml blasticidin (#ant-bl, InvivoGen, San Diego, CA, USA). MnSOD knockdown was established using a lentiviralbased shRNA vector (pGFP-C-shLenti) (#TR30021, OriGene, Rockville, MD, USA). To generate stable transfectants for MnSOD knockdown, pGFP-C-shLenti carrying scrambled shRNA or shMnSOD was transfected into HEK293T cells with Lipofectamine 3000 (#L3000001, Thermo Fisher Scientific, Waltham, MA, USA). After transfection for 24 h, the medium was replaced with BSA-enriched medium, and then the conditioned medium was collected at 24 and 48 h. The lentivirus-containing medium was centrifuged to remove cell debris and used to transduce MDA-MB-231 (IV2-3) cells with polybrene transfection reagent (Millipore); then, the cells were selected with 20 μg/ml puromycin for 2 weeks. To generate cells overexpressing MnSOD, True-ORF pCMV6-Entry-Flag-tagged SOD2 (#RC202330) and Sod2 (#MR202568) (all from OriGene) were transfected into MDA-MB-231 (IV2-3) and 4T1 cells with Lipofectamine 3000 (Thermo Fisher Scientific), and the transfectants were selected with 100 or 200 μg/ ml neomycin (G418) for 2 weeks. IL-6 knockdown in MDA-MB-231 (IV2-3) cells was established as previously described [20], and the cells were maintained in complete medium containing 0.5 μg/ml puromycin. Cell lines were not authenticated using STR profiling, but all cell lines were kept at low passages in order to maintain their identity. Generated stable cell lines were authenticated via Western blots to validate high and/or low expression of MCT-1/MnSOD and routine observation of cell morphology under the microscope.

Cycloheximide chase assay
Cells (8 × 10 5 ) were seeded in a 60 mm dish with complete medium and incubated for 24 h. The medium was then removed and replaced with a medium containing 200 µM cycloheximide (#C7698, Sigma-Aldrich). Cell lysates were collected at the indicated time intervals and subjected to Western blotting.

MnSOD dismutase activity
Superoxide dismutase activity in the sonicated cell homogenates was detected using the Superoxide Dismutase Assay Kit (#706002, Cayman Chemical) according to the manufacturer's protocol. MnSOD activity was measured after incubating the reaction mixture with 2 mM potassium cyanide (KCN) (gift from Institute of Biotechnology and Pharmaceutical Research at National Health Research Institute, Zhunan, Taiwan) for 30 min to inactivate CuZnSOD (SOD1) and extracellular SOD (SOD3) activity. The absorbance at 450 nm was monitored using a Tecan Infinite 200 PRO multimode microplate reader (Tecan).

Measurement of IL-6 secretion
To analyze IL-6 secretion, cancer cells (8 × 10 5 ) were seeded in a 60-mm culture dish with complete medium and incubated for 72 h. The complete medium was removed and replaced with 2 ml serum-free medium, and the cells were incubated for 48 h. The medium was collected by centrifugation at 2,000 rpm and 4°C for 10 min. The supernatants were immediately used for analysis of IL-6 secretion by the Human IL-6 ELISA MAX TM Deluxe Set (#430505, BioLegend, San Diego, CA, USA) according to the manufacturer's protocol. Reactions were stopped with 2 N H 2 SO 4 , and the absorbance was read at 450 nm and 570 nm using a Tecan Infinite 200 PRO multimode microplate reader (Tecan). The absorbance of samples at 570 nm was subtracted from the absorbance at 450 nm.

Conditioned medium production
To obtain MitoQ-treated conditioned medium (CM), MDA-MB-231 (IV2-3) cells (2 × 10 6 ) were seeded in a 100 mm dish, cultured until 80% confluent, treated with or without 0.5 μM MitoQ (Cayman Chemical) in complete RPMI medium for 24 h, rinsed twice with phosphate-buffered saline (PBS) to remove MitoQ, replaced and cultured with medium containing 1% FBS for another 24 h. The CM was harvested, followed by centrifugation at 200 × g for 10 min and filtration through 0.22-µm-pore filters to remove cell debris, and directly used for the experiment or stored at −80°C.

Macrophage differentiation and migration
Humane THP-1 monocytes and M0 murine RAW264.7 macrophages were differentiated into M1 and M2 macrophages as previously described [27]. Migratory abilities of macrophages were analyzed under a Leica AF6000LX microscope (Leica Microsystems, Wetzlar, Germany) using a 20x objective. Images were acquired for 18-24 h at 5 min intervals. Cell movements were tracked with Metamorph (Molecular Devices, San Jose, CA, USA) to quantify the total migration distance. Cell movement tracks emanating from the initial position were plotted using the DiPer macros [44]. For macrophage migration in response to MitoQ (Cayman Chemical), M2 THP-1 macrophages were cultured with CM and the movement was monitored as described above.

Cancer cell invasion and migration
Cell invasion assays were performed using Corning BioCoat Tumor Cell Invasion Systems (#354480, Corning) as previously described [19]. Cell migration was assayed using ibidi culture inserts (ibidi GmbH, Planegg, Germany). Cancer cells (3 × 10 4 ) were seeded in a well containing a culture insert for 24 h. The culture inserts were then removed to create a cell-free gap. The remaining cells were immediately washed with warm PBS and overlaid with complete medium. Migratory cells that repaired the cell-free gap were monitored with a Leica AF6000LX microscope (Leica Microsystems) using a 10x objective lens. Live-cell images were acquired at three different positions per sample over 24 h at one-hour intervals. Wound closure capacity was analyzed with Metamorph (Molecular Devices).

Soft agar colony formation assay
Anchorage-independent growth was characterized when MDA-MBA-231 cells (2 × 10 4 ) were seeded on 0.3% agarose (#50002, Lonza, Rockland, ME, USA) in RPMI medium over a bottom layer of 0.6% agarose in RPMI medium. The cells were fed every 5 days with RPMI medium containing 0.3% agarose. After 2 weeks of incubation, the colonies were imaged with a Leica AF6000LX microscope using a 20x objective.

Gelatin zymography study
MDA-MB-231 (IV2-3) cells (1 × 10 6 ) were seeded in a 60 mm culture dish with complete RPMI medium and incubated for 24 h. The complete medium was removed, and the cells were refed with 0.5 ml RPMI containing 1% FBS. After 24 h of incubation, the medium was collected and centrifuged at 2,000 rpm and 4°C for 10 min. The supernatants were resolved in an 8% SDS-PAGE gel containing 0.1% gelatin. Gels were subsequently washed in 2.5% (v/v) Triton X-100/PBS for 30 min and agitated in freshly prepared zymogram developing buffer (50 mM Tris-HCl, pH 7.5; 200 mM NaCl; 5 mM CaCl 2 ; 0.02% Brij 35 detergent (#203724, EMD Chemicals, San Diego, CA, USA)) at 37°C for 24 h. To visualize MMP activity, gels were stained with 0.5% (w/v) Coomassie Brilliant Blue R-250 (Sigma-Aldrich) for 30 min, immediately destained with the solution I (25% ethanol and 10% acetic acid) for 30 min, washed with solution II (5% ethanol and 7.5% acetic acid) for 4-5 h and agitated overnight in distilled water. Media containing 1% FBS or 10% FBS were used as negative and positive controls, respectively. Zymograms were scanned and analyzed using ImageJ (National Institutes of Health).

Xenograft tumor growth and immunohistochemistry analysis
For examination of orthotopic mammary tumor growth, 6-week-old female BALB/c nude mice (cAnN.Cg-Foxn1 nu /CrlNarl) (BioLASCO, Taipei, Taiwan) were randomly divided at the start of the experiment to be implanted with luciferase-expressing MDA-MB-231 (IV2-3) cells (1 × 10 6 ), which were stably transfected with MnSOD-specific shRNA or scrambled shRNA, on the bilateral side of the 4th mammary fat pads. These cell lines were tested with EZ-PCR Mycoplasma Detection Kit and showed negative results for mycoplasma contamination (#20-700-20, Biological Industries, Kibbutz Beit-Haemek, Israel). Tumor development was measured weekly and recorded as tumor width (W) and tumor length (L). Tumor volume was calculated using the formula V = 1/2 (W 2 × L). Observers were not blinded during tumor measurement because experimental identities were recorded on the cage cards. After implantation (9 weeks), the mice were euthanized by CO 2 inhalation, and the tumors were collected for weight and size analysis, under a protocol approved by the NHRI Institutional Animal Care and Use Committee (NHRI-IACUC-107053-A). Immunohistochemistry was performed on MDA-MB-231 (IV2-3) tumor sections as previously described [20] with staining for CD163 (1:600) (#GTX35247, GeneTex).

Quantitative RT-PCR (qRT-PCR)
To measure mRNA expression levels, cells were washed with ice-cold PBS and subjected to total RNA extraction with RNAzol RT (#RN190, Molecular Research Center, Cincinnati, OH, USA) according to the manufacturer's protocol. Extracted RNA samples were immediately used for cDNA synthesis using the Maxima First Strand cDNA Synthesis Kit (#K1642, Thermo Fisher Scientific) according to the manufacturer's instructions. cDNA samples (100 ng) were mixed with PrecisionPLUS qPCR Master Mix (#PPLUS, Primerdesign, Southampton, UK) and subjected to quantitative PCR using a ViiA 7 Real-time PCR System (Thermo Fisher Scientific). All qRT-PCR assays were analyzed using comparative Ct methods (2 −ΔΔCt ). Results were normalized to ACTB mRNA results as an internal control. ΔΔCT = [(Ct target gene − Ct internal control) of treated cells − (Ct target gene − Ct internal control) of untreated cells].

Primer sequences
The primer sequences were designed according to the NCBI Probe Database and listed in Supplementary Table S1.

Clinical study
The SOD2 and NFE2L2 mRNA levels in normal breast and breast tumor specimens were quantified by qRT-PCR using TissueScan Breast Cancer cDNA Arrays I (BRCT101), III (BRCT103), and IV (BRCT104) (OriGene). The results were normalized to those for ACTB, which was used as an internal control, and then compared to those for normal breast tissue. Data were analyzed using comparative Ct (cycle threshold) methods (2 −ΔΔCt ). ΔΔCT = Ct of breast tumor − Ct of a normal breast. The expression levels of the MCTS1, SOD2, and NFE2L2 genes in breast cancer patients were also obtained from the Oncomine cancer profiling database (https://www. oncomine.org/resources/) [22]. Correlation plots between MCTS1 and SOD2, MCTS1 and NFE2L2, SOD2 and NFE2L2, and SOD2 and IL6 derived from the Oncomine dataset were then generated and analyzed using Minitab 16 software (Minitab Ltd., Coventry, UK), while KM plots were generated and analyzed using MedCalc (MedCalc, Ostend, Belgium). Alternatively, the probability of relapse-free survival in breast cancer patients was also analyzed using KM Plotter (https://kmplot.com/analysis/) [21]. The cohorts were divided into high and low expression groups according to the upper/lower quartile or median expression of the gene of interest. Additionally, the correlation between SOD2 expression and M2 macrophage infiltration in breast cancer patients in The Cancer Genome Atlas (TCGA) cohort was analyzed using TIMER 2.0. (http://timer.compgenomics.org) [34].

Statistics
Sample sizes was determined based on experience and preliminary study. For in vitro experiments, a minimum number of three biological replicates were used and shown in all figures. In vivo sample sizes were based on the previous studies in the laboratory that revealed that the number of animals to be sufficient to obtain significant differences. Cells were allocated into control and experimental groups based on its genetic manipulation (MCTS1, SOD2, or others). Investigators were not blinded during data collection and/or analysis for each in vitro experiment to allow correct identification of samples (control and other groups). Experiments were repeated at least twice on two independently grown cell cultures. The survival probability of breast cancer patients and tumor-free survival in mice were analyzed using the log-rank Mantel-Cox test. Correlations between MCTS1 and SOD2, MCTS1 and NFE2L2, SOD2 and NFE2L2, and SOD2 and IL6 were examined by Pearson productmoment correlation coefficients, whereas the correlation between MnSOD expression and M2 macrophage infiltration was evaluated using Spearman's rank correlation coefficient. The statistical significance of SOD2 and NFE2L2 mRNA in TissueScan Human Breast Cancer cDNA Arrays I, III, and IV (OriGene) was assessed by the chi-squared (χ 2 ) test. Experiments involving qRT-PCR, cancer migration and/or invasion, cycloheximide chase, mROS quantification, ELISA, the enzymatic activity, mammosphere formation, macrophage polarization/migration/invasion/phagocytosis, H 2 O 2 release, tumor burden, and tumor growth were analyzed with one-way or two-way ANOVA followed by the Tukey-Kramer or Newman-Keuls post hoc test or with a two-tailed unpaired t-test. These statistical tests were selected to be appropriate for the data properties (normality of distribution and homogeneity of variance). A pvalue less than 0.05 was considered significant. Statistical tests were performed with Minitab 16 software (Minitab Ltd.) and GraphPad Prism 8 (GraphPad Software, San Diego, CA, USA). Extreme outliers were removed using Grubbs' test provided by GraphPad (GraphPad Software) at p-value < 0.05. Survival estimation analysis and KM plot generation were performed using KM Plotter [21] or MedCalc (MedCalc).

DATA AVAILABILITY
Datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.