Cancer-Associated Oxidoreductase ERO1-α Drives the Production of Tumor-Promoting Myeloid-Derived Suppressor Cells via Oxidative Protein Folding

The tumor microenvironment was shown to be an immunosuppressive microenvironment. Myeloid-derived suppressor cells (MDSCs) are a major component of the immune-suppressive network in cancer and many other pathological conditions. Experimental models using mice showed that MDSCs can facilitate tumor progression by promoting the suppression of antitumor immunity (1–4), promoting inflammation (4, 5), stimulating angiogenesis (6), and enhancing tumor cell migration and metastasis (7). Clinical studies further demonstrated that the presence of MDSCs correlates with adverse outcomes and shorter survival in various types of cancer, including breast cancer (8). MDSCs are a heterogeneous group of myeloid cells that are characterized by potent immunosuppressive activity. In mice, they are characterized as CD11b⁺ Gr-1⁺ cells. In recent years, two major groups of cells that make up MDSCs have been identified: cells with a morphology and phenotype (CD11b⁺ Ly6C^low Ly6G⁺) typical of polymorphonuclear (PMN)-MDSCs and cells with a morphology and phenotype (CD11b⁺ Ly6C^high Ly6G⁻) typical of monocytes (monocytic MDSCs). Monocytic MDSCs consist of immature myeloid cells with the ability to differentiate into macrophages and dendritic cells. PMN-MDSCs are the largest population of MDSCs in tumor-bearing mice, representing >75% of all MDSCs. They suppress Ag-specific T cell responses, primarily via the release of reactive oxygen species. PMN-MDSCs also have been found in cancer patients. Thus, PMN-MDSCs play a pivotal role in tumor progression. However, the underlying mechanism through which PMN-MDSCs proliferate and infiltrate into tumor sites has been unclear. G-CSF is a cytokine with potent neutrophil-proliferative activity. G-CSF is produced by macrophages, fibroblasts, and endothelial cells. Recently, it was shown that tumor cells are a source of G-CSF (9–12) and that production of G-CSF by tumors is responsible for the recruitment of immunosuppressive PMN-MDSCs, which promote tumor growth via inhibition of antitumor immune responses (13, 14). A vital role for tumor-derived G-CSF in tumor-bearing mice was demonstrated by Waight et al. (13) In addition, depletion of G-CSF resulted in reduced tumor growth in a murine mammary gland cancer 4T1 model. 4T1 cells were previously shown to express a G-CSF transcript; thus, G-CSF is a candidate molecular target for cancer treatment.

ERO1-α is an endoplasmic reticulum (ER)-resident oxidase. ERO1-α and protein disulfide isomerase (PDI) play central roles in disulfide bond formation of secreted and cell surface molecules (15–18). Disulfide bond formation (i.e., oxidative protein folding) is the most common posttranslational modification and is required for proper conformation and function of these molecules. Thus, these secreted and cell surface molecules need to be regulated at the gene expression level, as well as undergo proper posttranscriptional modification. We demonstrated recently that various types of tumor cells expressed high levels of ERO1-α and that ERO1-α is a marker of poor prognosis in breast cancer (19). In this article, we demonstrate that ERO1-α plays a pivotal role in PMN-MDSC induction via upregulation of G-CSF production from cancer cells in collaboration with PDI.

The murine breast cancer cell line 4T1 and human breast cancer lines MCF7, BT-474, UACC-893, SK-BR-3, MDA-MB-157, MDA-MB-231, and MDA-MB-468 were purchased from the American Type Culture Collection (Manassas, VA). 4T1, BT-474, and MDA-MB-157 cells were cultured in RPMI 1640 (Sigma-Aldrich, St. Louis, MO); MCF7, MDA-MB-231, and MDA-MB-468 cells were cultured in DMEM (Sigma-Aldrich); UACC-893 cells were cultured in Leibovitz’s L-15 (Life Technologies, Carlsbad, CA); and SK-BR-3 cells were cultured in McCoy’s 5A media (Life Technologies) supplemented with 10% FCS at 37°C in 5% CO₂. Short hairpin RNA (shRNA) for murine ERO1-α (TR502816) was purchased from OriGene (Rockville, MD) and transfected into 4T1 cells using Lipofectamine RNAiMAX (Life Technologies). Cells were stably propagated under puromycin selection (6 μg/ml). The murine ERO1-α gene fragment was isolated from pCMV6-Entry Vector/mERO1-α (OriGene), digested with BamHI and XhoI, and inserted into an appropriate site of the expression vector pcDNA6/myc-HisA (Invitrogen, Carlsbad, CA). The resulting pcDNA6/mERO1-α or an empty vector as a control was transfected into 4T1 cells using Lipofectamine 2000 (Life Technologies). Cells were stably propagated under blasticidin (5 μg/ml; Life Technologies) selection.

Female BALB/c and BALB/c nu/nu mice (4 wk old) were obtained from The Jackson Laboratory (Bar Harbor, ME) and used at 5 wk of age. Mice were maintained in a specific pathogen–free mouse facility at Sapporo Medical University, according to institutional guidelines for animal use and care. For tumor-formation studies, mice were injected with 3 × 10⁴ 4T1 cells, ERO1-α–overexpressed (OE) cells, or ERO1-α–knockdown (KD) cells into the right fourth mammary glands. Tumor growth was measured two or three times/wk in two dimensions, and tumor volume was calculated using the equation 3.14 × (width² × length)/6. Tumor length and width were measured with a caliper.

Mice were challenged with 3 × 10⁴ 4T1 scrambled shRNA-transfected (SCR) cells. For the depletion of CD4⁺ and/or CD8⁺ T cells, mice were injected i.p. with a CD4-specific Ab and/or CD8-specific Ab at 200 μg/mouse on day 3 before and day 1 after tumor challenge. For depletion of Ly6G⁺ PMN-MDSCs, mice were injected i.p. with Ly6G-specific Ab clone 1A8 (Bio X Cell, West Lebanon, NH) or rat IgG (Sigma-Aldrich) at 100 μg/mouse every 2 d from day 15 after the tumor challenge.

Total RNA was isolated from cultured cells and normal breast tissues using Isogen reagent (Nippon Gene, Tokyo, Japan) and RNeasy Mini kits (QIAGEN, Valencia, CA) according to manufacturer's instructions. The cDNA mixture was synthesized from 1 μg total RNA by reverse transcription using Superscript III and oligo (dT) primer (Life Technologies) according to the manufacturer’s protocol. PCR amplification was performed in 20 μl PCR mixture containing 1 μl cDNA mixture, 0.1 μl Taq DNA polymerase (QIAGEN) and 6 pmol primers. Quantitative RT-PCR was performed with a QuantiTect SYBR Green PCR Kit (QIAGEN) to determine the expression levels of cxcl1, cxcl2, g-csf, and β-actin. Expression values for each sample were normalized to β-actin, and fold levels of the indicated genes represent the mean (± SEM) of replicate reactions. Primer sequences were as follows: β-actin (actb), QuantiTect Mm_Actb_1_SG Primer Assay; cxcl1, QuantiTect Mm_Cxcl1_1_SG Primer Assay; cxcl2, QuantiTect Mm_Cxcl2_1_SG; and g-csf (Csf3), QuantiTect Mm_Csf3_1_SG (QIAGEN). PCR cycles were performed using a StepOne Real-Time PCR System (Life Technologies) with the following cycle conditions: 10 min at 95°C, 45 cycles of 15 s at 95°C and 1 min at 60°C, followed by melting curve analysis. The δ-δ Ct method was used for data analysis.

Real-time relative PCR (real-time PCR) was performed to determine the expression levels of ERO1-α and β-actin. Expression values for each sample were normalized to β-actin, and fold levels of the indicated genes represent the mean (± SEM) of replicate reactions. Primer sequences were as follows: β-actin (ACTB), Hs0160665_g1; and ERO1-α (ERO1L), Hs00205880_m1 (Life Technologies). PCR cycles were performed using a StepOne Real-Time PCR System (Life Technologies) with the following cycle conditions: 2 min at 50°C, 10 min at 95°C, and 40 cycles of 15 s at 95°C and 1 min at 60°C. The ΔΔ Ct method was used for data analysis.

Cultured cells were washed in ice-cold PBS, lysed by incubation on ice in a lysis buffer (50 mmol/l Tris-HCl [pH 7.5] 150 mmol/l NaCl, 5 mmol/l EDTA, 1% Nonidet P-40), and cleared by centrifugation at 21,880 × g for 30 min at 4°C. For blockade of free thiols, cells were pretreated for 5 min with 10 mM methyl methanethiosulfonate (Pierce, Rockford, IL) in PBS. Postnuclear supernatants were divided and heated for 5 min at 95°C in a nonreducing or reducing SDS sample buffer, resolved by SDS-PAGE, and electrophoretically transferred to polyvinylidene difluoride membranes (Immobilon-P; Millipore, Billerica, MA). The membranes were incubated with blocking buffer (5% nonfat dried milk in PBS) for 30 min at room temperature and then incubated overnight with anti–ERO1-α mAb (Abnova, Taipei, Taiwan), anti-PDI polyclonal Ab (Enzo Life Sciences, Farmingdale, NY), anti–mouse G-CSF (R&D Systems, Minneapolis, MN), or mouse anti–β-actin mAb AC-15 (Sigma-Aldrich). After washing three times with wash buffer (0.1% Tween-20 in TBS), the membranes were reacted with peroxidase-labeled goat anti-rabbit IgG Ab, peroxidase-labeled goat anti-mouse IgG Ab, or peroxidase-labeled rabbit anti-goat Ab (KPL, Gaithersburg, MD) for 3 h. Finally, the signal was visualized using an ECL detection system (Amersham Life Science, Arlington Heights, IL) or an IMMOBILON detection system (Millipore), according to the manufacturers’ protocols.

Mice were injected with 1 × 10⁵ 4T1 cells, OE cells, or KD cells into the right fourth mammary glands. Tumor tissues, peripheral blood, spleen, and bone marrow were collected on day 14 after tumor challenge. For analyses of leukocytic infiltrates in the tumor tissue, tumors were mechanically dissociated on a wire mesh by crushing with scissors and digested for 2 to 3 h at 37°C in RPMI 1640 medium supplemented with 5% FBS and Liberase (Roche, Tokyo, Japan; 1 mg/ml). Then, leukocytes were collected by density-gradient centrifugation with Lympholyte-M. Peripheral blood, spleen cells, and bone marrow were hemolyzed by a hemolytic agent. The cells were filtered through 70-μm nylon strainers (BD Biosciences, Bedford, MA), and cell numbers were counted. They were stained with specific markers and analyzed by flow cytometry. All preparations were preincubated with anti-CD16/32 mAb (BD Biosciences, San Diego, CA) to block FcR binding, followed by incubation with a directly conjugated primary Ab. Labeled cells were analyzed using a FACSCalibur flow cytometer (BD, San Jose, CA) and FlowJo software (TreeStar, Ashland, OR). Abs reactive against the following cell surface markers were used (including appropriate isotype controls): PE-labeled CD11b, PerCP/Cy5.5-labeled Ly6G and allophycocyanin-labeled Ly6C (BioLegend, San Diego, CA), and PerCP/Cy5.5-labeled Gr-1 (eBioscience, San Diego, CA).

Tissue was fixed in neutral 10% buffered formaldehyde, embedded in paraffin, and cut into 5-μm-thick slices for ERO1-α staining. Reactivity of the anti–ERO1-α mAb was determined by perinuclear staining within tumor cells, indicating ER localization. Tissues were frozen with OCT compound (Leica Biosystems, Nussloch, Germany) and cut into 7-μm-thick slices for CD11b and Gr-1 staining. Abs reactive against CD11b and Gr-1 were purchased from BioLegend and eBioscience, respectively. Secondary Abs were purchased from DAKO Japan (Tokyo, Japan). Tissue sections were developed using diaminobenzidine. Images were quantified by counting the number of positively stained cells in five randomly selected fields at ×200 magnification.

4T1 cells were plated at 1 × 10⁵ cells/well in six-well plates for 24 or 48 h. All samples were stored at −80°C until assayed. Supernatants were diluted, and mouse CXCL1 (GRO/KC; IBL, Gunma, Japan), mouse CXCL2 (MIP-2α), and mouse G-CSF (R&D Systems) levels were measured using a sandwich ELISA kit. Absorbance was determined at 450 nm.

The Student t test was used for analysis of two unpaired samples. Statistical differences with regard to depletion of CD4⁺ and/or CD8 T⁺ cells were analyzed by the Dunnett test. Differences in regard to depletion of Ly6G⁺ PMN-MDSCs were assessed by the Mann–Whitney U test. Overall survival rates were calculated by the Kaplan–Meier method, and differences in survival curves were assessed by the log-rank test. All analyses were carried out with StatMate version 3.19 (ATMS, Tokyo, Japan). A p value < 0.05 was regarded as statistically significant. All statistical tests were two sided.

We showed previously that ERO1-α expression was augmented in human breast cancer cell lines and breast cancer tissues (19). In this study, we showed that expression of ERO1-α on breast cancer cell lines was upregulated, regardless of the histological type, compared with the expression in normal breast tissues at the mRNA level (Fig. 1A). Immunohistochemical staining also showed that ERO1-α was preferentially expressed within tumor cells but not normal breast tissue (Fig. 1B, 1C). Our previous study demonstrated that none of the normal breast tissues (71 cases) were positive for ERO1-α staining (19). We observed that ERO1-α showed a patchy staining pattern within cancer nests. Because it was demonstrated that ERO1-α is induced under hypoxic conditions, we assumed that cancer cells residing within hypoxic areas show augmented expression of ERO1-α. Thus, heterogeneity of ERO1-α expression seems to be attributed to the oxygen and blood supply.

To examine the role of ERO1-α in tumor growth, we generated KD cells using shRNA against ERO1-α (Fig. 2A). KD cells did not show differences in a proliferation assay compared with wild-type 4T1 (WT) cells and SCR cells (Supplemental Fig. 1). When BALB/c nu/nu mice were challenged with SCR cells and KD cells, we observed that these two cell lines grew aggressively and at similar rates (Fig. 2B). In contrast, knockdown of ERO1-α retarded tumor growth compared with WT cells in BALB/c mice (Fig. 2C). KD cells also had a survival benefit compared with SCR cells (Fig. 2D). These results suggested that ERO1-α⁺ WT or SCR tumor cells inhibited antitumor immunity (i.e., depletion of ERO1-α within tumor cells might restore antitumor immunity). To clarify this, we performed a T cell–depletion assay in vivo. When CD4⁺ and/or CD8⁺ T cells were depleted during a tumor-growth assay using KD cells, KD cells showed tumor growth similar to that of SCR cells, indicating that both CD4⁺ and CD8⁺ T cells were responsible for the immunogenicity of KD tumor cells (Fig. 3A). Moreover, because some of the mice challenged with KD cells rejected the tumors, we rechallenged these mice with WT cells. All of the mice rechallenged with 4T1 tumor cells rejected tumor cells (Fig. 3B), indicating that KD tumor cells acted as immunogenic tumor cells due to ERO1-α knockdown. Based on these results, both CD4⁺ and CD8⁺ T cell–mediated antitumor immunity against 4T1 tumor was dampened by the expression of ERO1-α within 4T1 tumor cells.

To further confirm the effect of ERO1-α on tumor growth and antitumor immunity, we generated an OE 4T1 cell line by introducing murine cDNA of ERO1-α (Fig. 4A). When 4T1 cells transfected with a control vector (mock) and OE cells were inoculated into nude mice, there was no difference in tumor growth rates (Fig. 4B); however, when they were inoculated into BALB/c WT mice, OE tumors grew more aggressively (Fig. 4C, 4D). These results again suggested that expression of ERO1-α suppressed antitumor immunity against the 4T1 tumor.

Histopathological findings revealed that ERO1-α⁺ SCR 4T1 tumor tissues had a large amount of granulocytic infiltrates in the peritumor site, as well as within the tumor mass (Fig. 5A). In clear contrast, KD tumors showed fewer granulocytic infiltrates (Fig. 5B). We compared peritumoral- and intratumoral-infiltrating cells in SCR and KD tumors by immunohistochemical analysis using CD11b mAb and Gr-1 mAb. Although we observed that the major component was CD11b⁺ Gr-1⁺ MDSCs in ERO1-α⁺ SCR tumors and KD tumors, the total numbers of both CD11b⁺ cells and Gr-1⁺ cells were increased in SCR tumors (Fig. 5C, 5E, 5G) compared with KD tumors (Fig. 5D, 5F, 5H). Furthermore, we investigated the population of these infiltrates using a flow cytometer. Interestingly, Ly6G⁺ PMN-MDSCs were more common in the spleen (40.9 versus 17.2%), bone marrow (67.0 versus 57.7%), peripheral blood (57.6 versus 23.7%), and tumor (45.2 versus 27.9%) in SCR tumor–bearing mice compared with KD tumor–bearing mice (Fig. 6, Supplemental Table IA). In contrast, when we compared mice bearing mock tumors and OE tumors, we observed that PMN-MDSC infiltration was higher in OE tumor–bearing mice in the spleen (31.0 versus 18.1%), bone marrow (61.6 versus 60.4%), peripheral blood (50.3 versus 22.3%), and tumor (49.5 versus 37.6%) (Fig. 7, Supplemental Table IB). These results suggested that the expression of ERO1-α within the tumor resulted in the accumulation of PMN-MDSCs throughout the body, including the spleen, bone marrow, peripheral blood, as well as in the tumor, leading to the suppression of antitumor T cell–mediated immunity by PMN-MDSCs.

To determine whether Ly6G⁺ MDSCs were the main immunosuppressor cells in the 4T1 tumor system, we depleted Ly6G⁺ cells during the tumor-growth assay using anti-Ly6G mAb (1A8). Although depletion of Ly6G⁺ cells seemed incomplete, ∼50% of Ly6G⁺ cells remained in the peripheral blood (Supplemental Fig. 2). Apparently, depletion of Ly6G⁺ cells retarded SCR tumor growth compared with the growth of isotype-matched control IgG-treated tumors, and SCR tumor growth was almost the same as tumor growth in mice challenged with KD cells (Fig. 8). These results indicated that (tumor-associated) Ly6G⁺ PMN-MDSCs were the main immunosuppressor cells for ERO1-α⁺ 4T1 tumor cells.

We examined the mechanism of induction of tumor-associated PMN-MDSCs. Because it was demonstrated that G-CSF and GM-CSF induced the proliferation of PMN-MDSCs (1, 13, 14), we measured the production of G-CSF and GM-CSF from 4T1 tumor cells using ELISA. We found that ERO1⁺ SCR cells produced a large amount of G-CSF compared with the amount produced by KD cells (Fig. 9A). In addition, the production of GM-CSF from SCR cells was greater than that from KD cells, although the difference was not as great as that seen for G-CSF (Supplemental Fig. 3A). Although G-CSF amplifies PMN-MDSCs in the spleen, bone marrow, and peripheral blood, PMN-MDSC recruitment from the circulation to tumor stroma is thought to occur mainly via interaction between the chemokines CXCL1 and CXCL2 (20, 21) and their receptor CXCR2 expressed on PMN-MDSCs (20). Therefore, we compared the concentrations of CXCL1 and CXCL2 in the culture supernatant of each cell line. SCR cells produced a larger amount of CXCL1 and CXCL2 for PMN-MDSC recruitment than did KD cells (Fig. 9B, 9C). In contrast, when we compared mock cells and OE cells, OE cells showed greater production of G-CSF, CXCL1, and CXCL2 (Fig. 9D–F). As expected, PMN-MDSCs in the spleen, bone marrow, and peripheral blood expressed CXCR2 (Supplemental Fig. 3B). These results suggested that tumor-derived G-CSF induced proliferation of PMN-MDSCs, followed by recruitment to the tumor sites by augmented production of CXCL1 and CXCL2, resulted in the inhibition of the antitumor T cell–mediated immune response.

We investigated the role of ERO1-α in the augmented production of G-CSF, CXCL1, and CXCL2. To determine whether the gene expression level was affected by ERO1-α, we compared g-csf mRNA expression levels by real-time RT-PCR. The mRNA expression levels were not different in mock cells and OE cells, indicating that the expression of ERO1-α did not influence the g-csf gene expression level (Fig. 10A). In addition, we observed that the gene expression levels of cxcl1 and cxcl2 were not altered (Fig. 10A). Next, we compared the protein levels using Western blot analysis. Because ERO1-α is known to act as an oxidoreductase, we further investigated the redox states of G-CSF in mock cells and OE cells by Western blot analysis under nonreducing conditions using methyl methanethiosulfonate. We found that the total amount of G-CSF protein in OE cells was greater than that in mock cells under reducing conditions (Fig. 10B). Moreover, we found that ratio of oxidized form (mature form)/reduced form (immature form) of G-CSF in OE cells was much higher than that in mock cells under nonreducing conditions (Fig. 10B). It was shown that G-CSF has two intramolecular disulfide bonds (22) and that disulfide bond formation is required for G-CSF to exert its biological activity. Taking this into account, these results indicated that tumor ERO1-α plays a pivotal role in the generation of disulfide bonds within G-CSF at the posttranscriptional level but not at the gene expression level. Thus, tumor ERO1-α plays a crucial role in the proper folding of G-CSF, leading to the augmented production of G-CSF. To further confirm our observations, we compared the gene expression levels of g-csf, cxcx1, and cxcl2 in SCR and KD cells. As expected, we did not observe any differences in the mRNA levels of these genes between the two cell lines (Fig. 10C). In addition, we compared the redox states of G-CSF in SCR and KD cells and found that the total amount of G-CSF protein was much greater in SCR cells under reducing conditions (Fig. 10D). Furthermore, both the reduced and oxidized forms of ERO1-α were clearly decreased in KD cells compared with SCR cells under nonreducing conditions. We speculated that the decrease in the total amount of G-CSF protein in KD tumor cells occurred via ER-associated degradation, because the immature reduced form of protein is shown to be destined to proteasomal degradation. This will be examined in the future. We concluded that tumor-associated ERO1-α facilitated the oxidative folding of factors including G-CSF, CXCL1, and CXCL2 for the induction and recruitment of PMN-MDSCs, resulting in inhibition of the antitumor T cell response.

Oxidative protein folding, characterized by intramolecular disulfide bond formation, is the most common posttranscriptional modification (18). A proper disulfide configuration provides the structural foundation for more nuanced intramolecular folding events that define protein activity (23). Most of the proteins that are secreted or expressed on the cell surface have intramolecular disulfide bonds. Thus, oxidative protein folding is critical for normal cell function and homeostasis. Among the oxidoreductases expressed in the ER, ERO1-α is central to oxidative protein folding, but its expression varies in a tissue-specific manner (19). Moreover, it was shown that ERO1-α is upregulated under hypoxic conditions (24, 25), which are observed frequently in a cancer microenvironment. Furthermore, we showed previously that ERO1-α is overexpressed in various types of cancer cell lines and cancer tissues (19). Because cancer cells may take advantage of factors that are induced within hypoxic environments or at different stages of carcinogenesis, these findings led us to hypothesize that overexpression of ERO1-α is functionally important for tumor cells and prompted further exploration of its role in cancer progression. In this study using a mouse mammary gland cancer 4T1 model, we found that suppression of ERO1-α protein resulted in a dramatic reduction in tumor growth. Because the expression of ERO1-α did not seem to play a major role in tumor cell growth in in vitro or in vivo tumor growth using nude mice, massive infiltration of PMN-MDSCs was considered to contribute to tumor growth in BALB/c mice. In fact, an in vivo depletion assay using anti-Ly6G, anti-CD4, or anti-CD8 mAbs and a tumor rechallenge assay clearly showed that PMN-MDSCs inhibited antitumor T cell responses. Because ERO1-α is a sulfhydryl oxidase (26), it is unlikely that it would directly induce proliferation of PMN-MDSCs. Therefore, we measured levels of the cytokines and chemokines responsible for the induction or recruitment of PMN-MDSCs in the culture supernatant. 4T1 tumor cells secreted a large amount of G-CSF. In contrast, KD cells showed decreased secretion of G-CSF, CXCL1, and CXCL2. Based on these results, we hypothesized that ERO1-α is involved in the proper oxidative protein folding of G-CSF before it is secreted and that the loss of ERO1-α leads to functionally inactive G-CSF (reduced form), as shown by Western blotting under nonreducing conditions (Fig. 10D). To further confirm that what we observed is a posttranscriptional event, we conducted real-time RT-PCR on SCR cells and KD cells. We did not observe any differences in the mRNA levels of g-csf, cxcl1, and cxcl2 between the two cell lines, suggesting that ERO1-α acted at the posttranslational level. Because ERO1-α–expressing 4T1 cells may affect the function of MDSCs through the production of some cytokines, this possibility should be investigated in the near future.

We reported that ERO1-α is overexpressed in breast cancer tissues and is a useful marker of poor prognosis (19). Many tumors, including breast cancer, have the capacity to promote immune tolerance and escape host immune surveillance (2). Tumors use numerous pathways to inhibit immune responses, including the elaboration of immune inhibitory cytokines, such as IL-10 and TGF-β, as well as inducing host cells to release immune inhibitors. In addition to these mechanisms, immune suppression through MDSCs has a crucial role in promoting tumor progression (1).

In this study using a mouse mammary gland carcinoma 4T1 model, we showed that ERO1-α contributes to suppression of antitumor immunity through upregulation of G-CSF and CXCL1/2 via facilitation of oxidative protein folding. We did not determine whether ERO1-α is solely responsible for the proper oxidative folding of these molecules or whether it cooperates with PDI while these molecules are folding in the ER and Golgi. These questions need to be clarified. Although further research is needed to understand the role of ERO1-α in vivo, the results presented in this article strongly suggest that targeted inhibition of ERO1-α may stall cancer progression.

The authors have no financial conflicts of interest.