Myeloid Lineage–Specific Deletion of Antioxidant System Enhances Tumor Metastasis

Cancer cell metastasis is one of the major health burdens, which often leads to cancer-related death in humans. It has been well known that cellular oxidative stress accelerates the pathogenesis of a number of chronic diseases, including cancer growth and its metastasis to multiple organs (1–3). Although the importance of oxidative stress in cancer metastasis is well recognized, it remains uncertain how oxidative stress provokes metastasis or how metastasis therapy is attained through targeting oxidative stresses.

Transcription factor NF-E2–related factor-2 (Nrf2) plays an important role in cellular defense against toxic electrophiles and oxidative stresses (4–6). Target genes of Nrf2 include NAD(P)H quinone oxidoreductase 1 (Nqo1), glutamate-cysteine ligase catalytic subunit (Gclc), and heme oxygenase-1 (Ho-1). Disruption of Nrf2 enhances susceptibility of mice to chemically induced carcinogenesis in stomach (7), bladder (8), colon (9), and upper aerodigestive tract (10). The cancer preventive activity of Nrf2 seems to be mainly enhanced by detoxification and/or excretion of carcinogens.

We recently demonstrate that host Nrf2 exerts antimetastatic activity against inoculated cancer cells (11). Germline deletion of Nrf2 increases susceptibility to experimental lung metastasis of mouse 3LL (Lewis lung carcinoma) cells (11). In the cancer-bearing Nrf2-deficient mice, accumulation of reactive oxygen species (ROS) was observed in the cells that coexpress Mac1 (or CD11b) and Gr1, which is known as myeloid-derived suppressor cells (MDSC; ref. 12). MDSCs have been known to be a heterogeneous population of myeloid precursors composed of macrophages, dendritic cells, and granulocytes (13) and suppress anticancer immune system through influencing T-cell activation (14). Thus, we speculate that Nrf2 in myeloid lineage cells may exert antimetastatic activity by maintaining the cellular redox balance.

For therapeutic use to prevent onsets of various diseases, substantial attention has been paid to identify and develop Nrf2-inducing chemicals. Because Nrf2 stability is tightly regulated by Kelch-like ECH-associated protein 1 (Keap1; refs. 15, 16), many types of chemicals that inactivate Keap1, thereby induce Nrf2, have been identified (reviewed in ref. 17). Of the chemicals, synthetic triterpenoid 1-[2-cyano-3,12-dioxoleana-1,9(11)-dien-28-oyl] (CDDO) derivatives show potent inducer activity of Nrf2 (18–20). Although CDDO derivatives are reported to suppress incidence of cancer metastasis (21, 22), there remains uncertainty as to how Nrf2 contributes to the suppression of cancer cell metastasis.

Selenoenzymes, another major antioxidant system, function cooperatively with the Keap1–Nrf2 system. Selenoenzymes, such as glutathione peroxidase (GPx) and thioredoxin reductases (TrxR), reduce hydrogen and lipids peroxide, and contribute to the control of cellular redox homeostasis (23–25). Selenoenzymes contain selenium in the form of selenocysteine, an amino acid that is indispensible for catalytic activity of selenoenzymes (26). The selenocysteine-tRNA acts as an adaptor for selenocysteine and for translational insertion of the amino acid into selenoenzymes. Selenoenzymes are important for maintaining cellular redox balance in various tissues, including myeloid lineage cells, erythrocytes, and hepatocytes (27, 28). Indeed, dietary supplementation with high selenium reduces incidence of metastasis in murine lung cancer models (29). Hence, these observations suggest that antioxidant systems, such as Nrf2-battery genes and selenoenzymes in myeloid lineage cells, may be important for the suppression of metastasis.

To address this hypothesis, we generated myeloid lineage cell-specific knockout mice of Nrf2 gene (N-MKO: Nrf2-Myeloid knockout) or selenocysteine-tRNA (Trsp) gene (T-MKO: Trsp-Myeloid knockout). These two types of antioxidant system–deficient mice were inoculated in the metastatic 3LL lung cancer cells. We herein demonstrate that the antioxidant systems directed by Nrf2 and selenoenzymes in myeloid lineage cells are crucial for maintenance of redox balance in MDSCs and eventual prevention of cancer cell metastasis.

3LL cell line (30) was a kind gift from the Institute of Development, Aging and Cancer at Tohoku University (Sendai, Japan). Cells were regularly tested for Mycoplasma contamination using the e-Myco Mycoplasma PCR Detection Kit (iNtRON Biotechnology). No cell authentication was done by the authors. The cell culture medium was RPMI-1640 (Wako) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and penicillin–streptomycin (10 U and 0.1 mg/mL, respectively).

In the metastasis model experiments, 3LL cells (1 × 10⁶ in 100-μL PBS) were inoculated into the left thigh muscle. Weights of lungs and primary tumors were measured 21 days after inoculation. Numbers of lung surface nodules were counted after fixation with Tellyesniczky's solution (31).

Nrf2^F/F (32) and Trsp^F/F mice (27) on C57BL6/J genetic background were as described previously. Deletion of the floxed Nrf2 or Trsp gene in myeloid lineage cells was performed by crossing these floxed animals with heterozygous animals expressing Cre recombinase under regulation of the lysozyme M (LysM) locus (33). Nrf2^F/F::LysM-Cre (N-MKO) mice and Trsp^F/F::LysM-Cre (T-MKO) mice were used in this study. For CDDO-Im treatment study, animals were orally administrated CDDO-Im (30-μmol/kg body weight; CDDO-Im was a kind gift from Mochida Pharmaceuticals Co., Ltd.) or vehicle consisting of 10% dimethyl sulfoxide (DMSO), 10% Cremophor-EL, and 80% PBS by gavage. All animal experiments were carried out with the approval of the Tohoku University Animal Care Committee.

Lung tissues from the mice were fixed with Tellyesniczky's solution (31) and embedded into paraffin. Five-micrometer thick sections were stained with hematoxylin and eosin (H&E) using the standard techniques.

Single-cell suspensions from cancer-bearing mouse lungs were prepared as described previously (34). Analyses of lung cells, splenocytes, and bone marrow cells were performed using FACSCaliber (BD Biosciences). For the separation of MDSCs, each cell suspension was incubated with allophycocyanin (APC)-conjugated anti-Mac1 and phycoerythrin (PE)-conjugated anti-Gr1 monoclonal antibodies (eBioscience). For quantification of ROS levels, each cell suspension was incubated with 5-μmol/L 2′,7′-dichlorodihydrofluorececin diacetate (DCFDA; Invitrogen) at 37°C for 30 minutes (35). FACS analysis was performed using FlowJo (TOMY Digital Biology) software.

Macrophages were isolated by lavage from mice that had received a 2-mL intraperitoneal injection of 4% thioglycollate broth 4 days before. The cells were transferred onto a 10-cm dish at a density of 5 × 10⁵ cells/mL for immunoblot analyses or RNA extraction. The macrophages were maintained in RPMI-1640 medium containing 10% FBS and penicillin–streptomycin.

Whole spleen cells were dispersed and incubated on a 10-cm dish with RPMI-1640 medium containing 10% FBS and penicillin–streptomycin (10 U/0.1 mg/mL) for 2 hours. Nonadherent cells were removed and then the dish was washed twice with PBS. Adherent cells were subjected to the experiments as splenic myeloid cells. An anti-mouse Mac1 monoclonal antibody (clone M1/70; eBioscience) was used for the detection of splenic myeloid cells.

Peritoneal macrophages were collected and lysed in sodium dodecyl sulfate (SDS) sample buffer and stored at −80°C (36). The samples were subjected to immunoblot analysis using anti-Nrf2 (clone 103; ref. 37) and anti–α-tubulin (Sigma) antibodies.

Total RNA was extracted from peritoneal macrophages, plastic-adherent splenocytes, and liver tissues using ISOGEN (Nippon Gene). First-strand cDNA was synthesized from total RNA using ReverTra Ace qPCR RT Master Mix (TOYOBO). Real-time PCR was performed using an ABI PRISM 7300 sequence detector system (Applied Biosystems) and THUNDERBIRD Probe qPCR Mix (TOYOBO). The primer sequences for Nrf2, Nqo1, and Gclc were described previously (38).

All the data are presented as mean ± SD. Statistical differences were determined using the Student t test or the Mann–Whitney U test. P < 0.05 were considered statistically significant.

To ascertain whether the increased susceptibility of tumor cell metastasis in Nrf2^−/− mice is attributable to the Nrf2 deficiency in myeloid lineage cells, we crossbred Nrf2^F/F mice with LysM-Cre mice (Fig. 1A) to generate myeloid lineage–specific Nrf2-deficient mice (N-MKO). Through this crossbreeding, Nrf2 protein, which is normally accumulated by proteasome inhibitor MG132 in the control mice, was abrogated in peritoneal macrophages of N-MKO mice (Fig. 1B). We found that Nrf2^F/F mice showed slightly lower level of Nrf2 than wild-type (WT) mice did, and this may be due to residual Neo cassette. Therefore, to conduct reasonable comparison under this hypomorphic expression, we have exploited Nrf2^F/F mice as a control throughout this study. We also examined mRNA expression level of Nrf2 in peritoneal macrophages of N-MKO mice and found that the level also markedly decreased (Supplementary Fig. S1A). The upregulation of Nqo1, a typical Nrf2 target gene, in response to diethyl maleate (DEM), a well-known Nrf2 inducer, was observed in the peritoneal macrophages from WT and Nrf2^F/F mice, whereas the inducible Nqo1 expression was significantly diminished in those from N-MKO mice (Supplementary Fig. S1B). These results thus demonstrate that the transcript and protein levels of Nrf2 are significantly knocked out in N-MKO mice, and thereby the inducible Nqo1 expression is diminished in the macrophages.

We reported previously antimetastatic activity of Nrf2 through the analyses of systemic Nrf2 knockout mice. However, there remains uncertainty as to in which cell lineage Nrf2 deficiency actually contributes to the activity. To address this issue, we inoculated 3LL cells into left thigh muscle of Nrf2^F/F and N-MKO mice and developed primary tumors in the tissue. Mice were euthanized 3 weeks after the inoculation, and lung metastatic nodules were macroscopically and histologically examined (Fig. 1C). N-MKO mice exhibited a higher number of lung surface nodules (12 ± 9.1; n = 11) than did Nrf2^F/F mice (3 ± 5.7; n = 11; P = 0.0344; Fig. 1D; Table 1). In contrast, lung weight and primary tumor weight of thigh muscle were not significantly different between Nrf2^F/F and N-MKO mice (Supplementary Fig. S1C and Table 1). These results thus indicate that Nrf2 activity in myeloid lineage cells exerts the antimetastasis activity in the host lung tissue.

It has been shown that the MDSC population expands in many types of cancer patients and cancer model animals, and in mice. MDSCs are characterized by the coexpression of two myeloid lineage differentiation antigens, Mac1 and Gr1. Using fluorescence-conjugated antibodies and FACS, we examined the percentages of MDSCs (Mac1⁺Gr1⁺) in multiple tissues of cancer-bearing Nrf2^F/F and N-MKO mice. The lung cancer metastasis evoked expansion of MDSC fraction in the lung and spleen to the same extent in both Nrf2^F/F and N-MKO mice. Meanwhile, the MDSC population in the bone marrow was not significantly changed upon lung metastasis in both genotypes of mice (Fig. 2A and B).

We also analyzed intracellular ROS levels in MDSCs using a fluorescence marker H₂DCFDA. In this metastatic lung cancer model mice, ROS levels of MDSCs in the lung, spleen, and bone marrow of Nrf2^F/F and N-MKO mice were both elevated in comparison with tumor-free control mice of each genotype (Fig. 2C and D). Consistent with the increased metastasis susceptibility, the tumor-bearing N-MKO mice showed a higher ROS accumulation in the MDSCs of all the tissues in comparison with the tumor-bearing Nrf2^F/F control mice (Fig. 2C and D). These data thus indicate that Nrf2 is important for regulating intracellular ROS level in MDSCs and the suppression of metastasis.

Although metastatic tumor nodules were increased on the surface of N-MKO mouse lungs, it remains unclear whether ROS accumulation in MDSCs underlies this phenomenon. To explore this issue, we used myeloid lineage–specific Trsp-deficient mice (T-MKO). Because selenocysteine-tRNA is indispensable for synthesizing selenoenzymes, such as GPx and TrxR, and maintaining cellular redox homeostasis, T-MKO mouse will serve as an important model of ROS-accumulated myeloid lineage cells.

T-MKO mice showed larger number of surface nodules (5 ± 9.36; n = 13) than Trsp^F/F mice did (3 ± 3.33; n = 13; P = 0.0365; Fig. 3A and B; Table 1). Lung weight and primary tumor weight on thigh were not significantly different between T-MKO and Trsp^F/F mice (Supplementary Fig. S2 and Table 1). In the metastatic lung cancer models, the MDSC population in lung and spleen were both increased in T-MKO and Trsp^F/F mice, whereas in contrast, bone marrows contained comparable number of MDSCs irrespective of the 3LL cell inoculation (Fig. 3C). Importantly, the cancer cell inoculation provoked remarkable ROS accumulation in MDSCs from lung, spleen, and bone marrow of T-MKO mice compared with those of control Trsp^F/F mice (Fig. 3D). Taken together, the increase in intracellular ROS levels in MDSCs upon deletion of either Nrf2 or Trsp is well correlated with the enhancement of metastatic progression.

We surmised that Nrf2 activation in the host microenvironment may exert preventive efficacy against cancer cell metastasis. To address this possibility, we used CDDO-Im, a potent Nrf2-activator. As shown in Fig. 4A, Nrf2^F/F and N-MKO mice were administrated CDDO-Im (30-μmol/kg body weight) or vehicle three times per week (Fig. 4A, black arrows) from 2 days before cancer cells inoculation (open arrow). CDDO-Im–treated Nrf2^F/F mice showed a significant decrease in the number of lung surface nodules (2.5 ± 3.0; n = 10) compared with vehicle-treated Nrf2^F/F mice (9 ± 5.6; n = 11; P = 0.0260; Fig. 4B; Table 1). Meanwhile, there was not significant difference in the number of surface nodules between vehicle-treated N-MKO (12 ± 13.1; n = 11) and CDDO-Im–treated N-MKO mice (10 ± 14.1; n = 13; P = 0.4996; Fig. 4B; Table 1). These observations clearly indicate that CDDO-Im treatment has efficacy for prevention of cancer cell metastasis and this efficacy depends on Nrf2 expression in myeloid lineage cells.

Both Nrf2^F/F and N-MKO mice did not display significant difference in the primary tumor size, lung weight, and primary tumor weight (Supplementary Fig. S3A–S3C and Table 1), regardless of CDDO-Im treatment, indicating that growth of the primary tumor in thigh tissue is independent of the Nrf2 activity in this model.

Next, we examined whether CDDO-Im treatment reduces the ROS accumulation in MDSCs from tumor-bearing mice. CDDO-Im treatment significantly reduced the ROS levels in MDSCs of Nrf2^F/F mice, whereas the MDSCs of N-MKO mice did not show apparent decrease in ROS levels even after CDDO-Im treatment (Fig. 4C). These data indicate that CDDO-Im–induced Nrf2 activity decreases the intracellular ROS levels in MDSCs. On the contrary, CDDO-Im treatment did not affect the populations of MDSCs in the lung, spleen, and bone marrow in each group, indicating that CDDO-Im treatment did not have any cytotoxic effect within the doses in our experiments (Supplementary Fig. S3D).

We next examined whether CDDO-Im treatment induces Nrf2 activity in myeloid lineage cells from Nrf2^F/F and N-MKO mice. For this purpose, splenic myeloid lineage cells were collected as splenic adherent cells; majority of the adherent cells (89.87%) expressed Mac1 immunoreactivity (Supplementary Fig. S4), indicating that major cell type of the splenic adherent cells is myeloid lineage cell. The CDDO-Im treatment induced the expressions of both Nqo1 and Gclc genes in splenic myeloid cells from Nrf2^F/F mice, but not the cells from N-MKO mice (Fig. 4D). The inductions of these Nrf2 target genes in splenic myeloid cells from Nrf2^F/F mice were found to persist for at least 48 hours after treatment (Fig. 4D). These results indicate that Nrf2 activity is maintained at high level throughout the metastasis experiments with our CDDO-Im treatment protocol.

Collectively, these results demonstrate that CDDO-Im induces the expression of Nrf2 target genes in myeloid lineage cells, and thereby decreases the ROS accumulation in MDSCs. Thus, we conclude that Nrf2-mediated antioxidant system in myeloid lineage cells takes principal responsibility for the prevention of cancer cell metastasis.

This study demonstrates that the ROS levels in myeloid lineage cells are regulated by Nrf2 and selenoenzymes, and this regulation is critical for susceptibility of host animals for the lung cancer cell metastasis. The myeloid lineage–specific Nrf2-deficient mice provide a useful assay system to verify the participation of Nrf2 in myeloid lineage cells into the suppression of cancer cell metastasis. In addition to the Nrf2 pathway, myeloid lineage–specific selenoenzymes-deficient mice display increased ROS production in MDSCs and high-level susceptibility to metastasis of cancer cells compared with the control mice. In contrast, administration of a potent Nrf2 activator, CDDO-Im, reduced intracellular ROS levels in MDSCs and repressed lung metastasis of cancer cells in an Nrf2-dependent manner in myeloid lineage cells. These results thus indicate that the antioxidant systems regulated by both the Keap1–Nrf2 pathway and selenoenzymes play indispensable roles for maintaining redox balance in MDSCs and prevention of cancer cell metastasis (summarized in Fig. 5).

ROS accumulation has been suggested to participate in the MDSC-mediated immune suppression (reviewed in ref. 39). For instance, MDSCs with low level of ROS production (gp91-null mice) lack ability of inducing T-cell tolerance, suggesting that ROS production takes fundamental responsibility for MDSC-mediated CD8⁺ T-cell tolerance (40). The immunosuppressive activity of myeloid lineage cells is abrogated by the treatment with ROS scavenger, catalase (41). Showing very good agreement with these preceding observations, this study demonstrates that myeloid lineage–specific deficiency of antioxidant systems leads to aberrant ROS accumulation and resultant increase in lung metastasis. These observations further support the notion that ROS levels in myeloid cells are one of the critical determinants of the progression of cancer cell metastasis.

Although precise mechanisms underlying how ROS enhance metastasis remain unclear, several hypotheses have been proposed (reviewed in ref. 42). The present study showed that myeloid lineage–specific Nrf2- or Trsp-deficient mice did not show significant changes in the CD8 single-positive and CD4 single-positive T-cell populations despite significant increase in the lung metastasis (Supplementary Fig. S5), whereas the systemic Nrf2-null mice with cancer cell metastasis showed decreased in CD8⁺ T-cell population in our previous study (11). One plausible explanation for this discrepancy is that antioxidant system in T cells may be maintained to support the T-cell proliferation and/or survival in N-MKO and T-MKO mice even after cancer metastasis.

Alternatively, alterations in T-cell characters may give rise to the enhancement of cancer metastasis in the N-MKO and T-MKO mice. In this regard, one of the most likely mechanisms is the ROS-mediated modification of T-cell receptor (TCR)–CD8 complex. In combination with production of nitric oxide (NO) in cancer microenvironment, ROS accumulation in MDSCs seems to introduce peroxynitrite (ONOO⁻) modifications of the TCR–CD8 complex on the cell surface of CD8⁺ T cells. This aberrant modification disrupts the interaction between the TCR–CD8 complex and MHC class I molecules on antigen-presenting cells. Thereby, CD8⁺ T cell–mediated anticancer immunity is compromised eventually (40, 43).

Search for novel Nrf2–inducing drugs has been dramatically accelerated in hope of identifying therapeutics against various types of diseases. Those Nrf2 inducers should also work for cancer chemoprevention, given that pharmacologic induction of Nrf2 is effective for the prevention of chemical carcinogenesis models in mice (reviewed in ref. 17). Of the Nrf2-inducing chemicals, the synthetic triterpenoid CDDO derivatives are the most potent Nrf2 activators (44). Indeed, CDDO-methyl (CDDO-Me; Bardoxolone Methyl) has been validated extensively for therapeutic efficacy against diabetic nephropathy in clinical trials (45).

Meanwhile, it has been found that NRF2 accumulation can be brought by a number of distinct mechanisms in human cancer cases (46, 47). We first discovered somatic mutations within the interfaces of KEAP1 and NRF2, which disturb interaction between KEAP1 and NRF2, leading to stabilization of Nrf2. Subsequently, epigenetic modifications of the KEAP1 gene promoter sequence and disruptor proteins of the KEAP1–NRF2 interaction were identified (reviewed in ref. 48). A metabolic modifier of the KEAP1 activity was also identified (48). Aberrant NRF2 accumulation brought by these diverged mechanisms in the cancer cells contributes to the cancer progression and results in poor prognosis for these NRF2-accumulated cancers (48, 49).

As for the molecular basis for the Nrf2 contribution to cancer cells, it has been postulated that Nrf2 facilitates activation of antioxidative enzymes, drug-metabolizing enzymes, and multidrug resistance–associated proteins/pumps (MRP), all of which renders drug resistance to cancer cells (50). In addition, we recently found that Nrf2 also confers malignancy to cancer cells by enhancing the proliferation ability of the cells (reviewed in ref. 48). There observations raise a concern that clinical application of the NRF2 activators may lead to cancer progression. However, it turns out that pharmacologic induction can attain only a weak level activation of NRF2 and may not be harmful for cancer cell development. In addition, our current study unequivocally provides emphatic answer that CDDO-Im administration reduces cancer metastasis. Thus, the therapeutic benefit by activating host Nrf2 exceeds the exacerbation risk by activating Nrf2 in cancer cells. We surmise that Nrf2-mediated reinforcement of cancer immunity generates robust cancer-resistant environment, which makes a countercharge against metastatic cancer cells.

In summary, we provide promising lines of evidence that the antioxidant activity in myeloid lineage cells contributes to the suppression of ROS accumulation and subsequent prevention of metastatic cancer progression. Furthermore, we have proved that Nrf2-activator drugs give rise to therapeutic benefits rather than adverse side effects. Taking all these observations into account, we argue that the Keap1–Nrf2 system in myeloid lineage cells could serve as an attractive therapeutic target for antimetastatic treatment.

No potential conflicts of interest were disclosed.

Conception and design: K. Hiramoto, H. Satoh, T. Suzuki, M. Yamamoto

Development of methodology: K. Hiramoto, H. Satoh, M. Yamamoto

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Hiramoto, H. Satoh, J. Pi, M. Yamamoto

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Hiramoto, H. Satoh, T. Moriguchi, M. Yamamoto

Writing, review, and/or revision of the manuscript: K. Hiramoto, H. Satoh, T. Suzuki, T. Moriguchi, T. Shimosegawa, M. Yamamoto

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Hiramoto, H. Satoh, T. Suzuki, T. Shimosegawa, M. Yamamoto

Study supervision: T. Suzuki, T. Shimosegawa, M. Yamamoto

The authors thank Ms. Eriko Naganuma for technical assistance, Mochida Pharmaceuticals Co. Ltd. for reagents, and the Biomedical Research Core of Tohoku University Graduate School of Medicine for technical supports.

This work was supported in part by Grants-in-Aids for Creative Scientific Research and Scientific Research from JSPS (Grant Number 24249015 to M. Yamamoto and 70508308 to T. Suzuki), JST-CREST, the NAITO foundation and the Takeda Science Foundation (to M. Yamamoto). H. Satoh was supported by JSPS Research Fellowships for Young Scientists.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.