The human NPC cell lines CNE-1 (highly differentiated), CNE-2 (poorly differentiated), 5–8F (high tumorigenesis and metastasis), 6–10B (low tumorigenesis and low metastasis) and the normal nasopharyngeal epithelial cell line NP69 were all gifts from Xiang-Ya and Sun Yat-sen University School of Medicine. NPC cells were cultured in RPMI 1640 (Biological Industries Israel Beit-Haemek, 01–100-1ACS) containing 10% FBS (Biological Industries Israel Beit-Haemek, 04–001-1ACS). NP69 cells were cultured in keratinocyte-SFM (Thermo Fisher Scientific, 17,005–042). THP-1 human monocytic leukemia cells were purchased from the American Type Culture Collection (ATCC) and cultured in RPMI 1640 medium containing 10% fetal bovine serum (FBS) and penicillin/streptomycin. Each cell line was maintained at 37°C in humidified air and 5% carbon dioxide according to the provider’s suggestions. All experiments were performed with mycoplasma-free cells.

Tissue and serum samples from pathologically confirmed cases of NPC were collected at the Affiliated Hospital of Nantong University following ethics committee approval (IRB number: 2018-L018). All included patients were informed and had not received any cancer therapies before biopsy. Detailed patient clinicopathological features are listed in Supplementary Table S1. The prognostic significance of MIF was assessed by Kaplan-Meier analysis. The X-tile Software (version 3.6.1; Rimm Lab; Yale School of Medicine) was used to define MIF low and high expression before analysis.

The study conformed to the standards set by the Declaration of Helsinki.

THP-1 cells were treated with 100 nM phorbol-12-myristate-13-acetate (PMA) for 1 day and then differentiated into M1 macrophages with LPS (50 ng/ml, Sigma) + rIFN-γ (20 ng/ml, PeproTech, Rocky Hill, United States). M2-polarized macrophages were induced by adding IL-4 (20 ng/ml, PeproTech) for 24 h.

CNE2 knockdown of MIF was obtained by infection with MIF shRNA lentiviral particles. The lentiviral vector and its corresponding control were obtained from Shanghai Genechem Co., Ltd. The specific vector used for cloning shRNA control and MIF shRNA: pRRLSIN-cPPT-U6-shRNA-SFFV-EGFP-SV40-puromycin. LV-MIF-RNAi (86638-12) CCC​GGA​CAG​GGT​CTA​CAT​CAA. The transfection efficiency was examined by fluorescence microscopy, qRT-PCR, and Western blotting.

Macrophages were co-cultured with ISO-1(HY-16692,20uM, MCE) and human recombinant MIF factor (300–69,20 ng/ml, PeproTech) for 24 h. Ferrostatin-1 (Fer-1) and RSL3 were purchased from MCE.

First, the cells were washed with precooled PBS three times to remove the serum. Then the remaining liquid was aspirated, TRIzol (Thermo Fisher Scientific, 15596018) was added, the cells were gently scraped off with a cell scraper, chloroform was added, followed by vortex and centrifugation to remove the top layer, and isopropanol was added followed by centrifugation and isolation of the precipitate, and reverse transcription was performed. The reverse transcription system was 20 μL, and qRT-PCR was performed after cDNA was obtained.

To establish the lung metastasis model, 2.0 × 10⁶ luciferase labeled CNE2 cells were injected via the tail vein of mice in 100 µL of PBS (10 per group). All mice were divided into three groups, each with 10 mice. Three days after the injection of cells, the first group was intraperitoneally injected with DMSO every 3 days, the second group was injected with PBS, and the third group was injected with ISO-1 (MedChemExpress, HY-16692). Then, IVIS Lumina Series III was used to measure the photon flux of the whole body of the mouse every week to determine whether the model was successful. To monitor the lung metastasis during the 6 weeks, fluorescence measurements were performed on each mouse weekly.

All animal experiments carried out in this project followed the NIH Guidelines with the ethical approval of the Administration Committee of Experimental Animals, Jiangsu Province, China (Approval ID: S20200323-089). In vivo studies were approved by the committee on the Ethics of Animal Experiments of Nantong University (RDD number: 20180227-008).

After the cells were lysed on ice for 30 min and centrifuged to collect the supernatant, the concentration was measured with the BCA protein assay (Thermo Scientific Pierce™ BCA protein assay, Waltham, MA), 5× sample buffer was added, and the cells were boiled at 100°C for 10 min. Samples were loaded on a 12% SDS-PAGE gel, and then the proteins were separated in running buffer. The protein sample was transferred from the gel to a PVDF membrane. The transmission system was used at 300 mA on ice for 1 h. The membrane was blocked with 5% skimmed milk at room temperature. The membranes were incubated with the specific primary antibody overnight at 4°C. The next day, membranes were washed with TBST and then incubated with the secondary antibody in 1 × TBS for 1 h at room temperature. Then, the membranes were rewashed in TBST and incubated with ECL from Millipore (Billerica, MA). Finally, the films were exposed to radiographic film in a dark room. Use of antibodies for western blotting: Anti-MIF antibody (ab7207), ABCAM, the dilution concentration is 1/500; Anti-Glutathione Peroxidase 4 antibody (ab231174), ABCAM, the dilution concentration is 1/500; Beta Actin Monoclonal Antibody, CloneNo:2D4H5,Proteintech, the dilution concentration is 1/600. Anti-Flotillin 1 antibody (EPR6041) (ab133497), the dilution concentration is 1/200. Anti-CD9 antibody (EPR23105-121) (ab236630), ABCAM, the dilution concentration is 1/500; Anti-TSG101 antibody [EPR7130(B)] (ab125011), ABCAM, the dilution concentration is 1/500.

After the cells were prepared on the plate, they were fixed with 4% paraformaldehyde, blocked, and incubated with anti-MIF antibody (ab7207, ABCAM, 1/200) overnight. The sections were stained with anti-rabbit or anti-mouse Alexa Fluor 488 IgG (Proteintech), counterstained with DAPI, washed 3 times with PBS, mounted with an anti-fluorescence attenuation mounting plate, and observed with a fluorescence microscope.

Standard immunohistochemistry (IHC) methods were used to analyze human nasopharyngeal carcinoma specimens and lung metastasis specimens. Briefly, 5 μm thick sections were deparaffinized and then hydrated, and endogenous peroxidase activity was blocked with 3% hydrogen peroxide. The antigen recovery process was carried out in boiling water, blocked by peroxidase for 10 min, and then blocked with 10% BSA blocking buffer. The sections were incubated with anti-MIF (ab7207, ABCAM, 1/200), anti-INOS (ab115819, ABCAM, 1/100), anti-F4/80 (ab111101, ABCAM, 1/100) and anti-GPX4 (ab231174, ABCAM, 1/100), followed by secondary antibody and DAB. Then, hematoxylin was used to stain the nuclei, rinsed with running water for 40 min, finally, the gradient was reversed and observed under a microscope. Stained slides were assessed according to the staining intensity (extremely strong: four; strong: three; moderate: two; and weak: one) and the abundance of positive cells (≤5%: 0; 6–25%:1; 26–50%: 2; 51–75%:3, and ≥76%: 4) by two pathologists blind to the patient’s clinicopathological information. Scores of 0–8 were defined as low expression, and 9–16 were defined as high expression.

The isolated exosomes were mixed with 4% paraformaldehyde. Then, the exosomes were dropped onto an electron microscope grid coated with formvar-carbon and fixed with 1% glutaraldehyde for 10 min. The samples were negatively stained with 2% uranyl acetate solution. Images were acquired by a Tecnai transmission electron microscope at 163 kV.

The detailed protocol of continuous ultracentrifugation for extracting exosomes can be shown in our previous work (Bao et al., 2018). The exosomes were labeled with a PKH-67 labeling kit (Sigma-Aldrich) to observe the uptake of exosomes. After CNE2 cells were cocultured with the labeled exosomes for 1 h, CNE2 cells were fixed, and the nuclei were stained with DAPI. Finally, images were collected with a Leica TCS-SP5 LSM microscope. In the subsequent uptake experiment of co-cultured exosomes, we added 80 ug of exosomes per 1 × 10^6 cells.

Macrophages were seeded on a 96-well plate in a 37°C incubator. The cells were washed with the serum-free medium three times. Then, the serum-free medium was added to the cells. Ammonium iron (II) sulfate prepared with ddH₂O (10 mmol/L) was added to the cells (final concentration: 100 μmol/L). The cells were incubated for 30 min and washed with HBSS three times. FerroOrange (1 μmol/L) was added to the cells as HBSS solution, and then the cells were incubated for 30 min in a 37°C incubator equilibrated with 95% air and 5% CO₂. The cells were observed under a confocal fluorescence microscope.

Intracellular ROS production in macrophages was measured by a dihydroethidium probe kit (DHE, BestBio, Beijing, China). Adherent macrophages were digested with trypsin and resuspended in 500 µL of prechilled PBS, probes were added and incubated for 1 h at 37°C in the dark, and 480–535 nm wavelength excitation was used to detect cellular ROS content.

First, we seeded 10,000 macrophage cells on a 96-well plate, then added different drugs and incubated them in a cell culture incubator (37°C, 5% CO2). After 18 h, aspirated the medium and washed it with PBS. Then Added 1/10 volume of CCK8 solution to the medium without serum, incubated for 2 h in the dark, and then detected its absorbance at 450 nm in a microplate reader, and recorded the value.

Migration assay was performed using Transwell inserts (Corning, 3422) with a pore size of 8 µm. A total of 5 × 10⁴ cells suspended in medium without serum were added in the upper chambers, and the medium containing 10% fetal bovine serum was added to the lower chambers. After 16 h of incubation, the cells attached to the upper side were removed, and the cells attached to the underside of the membrane were fixed and stained with crystal violet. The invasion experiment is based on the migration experiment. Matrigel (Corning, 356234) was spread before the cells were seeded in the upper chamber. Digital images were obtained from the membranes, and five random fields were counted per chamber.

Statistical analysis was performed with GraphPad Prism 8 software. Data are presented as the mean ± SD. The significance of mean values between the two groups was analyzed using Student’s t-test. The differences between individual groups were analyzed by one- or two-way ANOVA. The differences were considered significant when the p value was less than 0.05.

To identify possible genes related to NPC, we extracted data on HNC from TCGA. The level of MIF in HNC is higher than normal controls (Figure 1A), patients with higher MIF levels had a poor clinical prognosis relative to patients with lower MIF levels in HNSCC (Figure 1B), and MIF can be an important biomarker in various malignant tumors such as breast cancer, pancreatic ductal adenocarcinoma, and bladder cancer. We chose MIF for further research. Then, to further explore the level of MIF in NPC, immunohistochemistry (IHC) was performed. The results showed higher expression of MIF in NPC patients vs. normal controls and higher expression in clinical stage III-IV NPC patients than in clinical stage I-II patients (Figures 1C,D), which suggested that the upregulation of MIF level may be closely related to end-stage nasopharyngeal carcinoma. The expression level of MIF was further verified by immunofluorescence (IF), showing that MIF was highly expressed in NPC cell lines (Figures 1E,F). In addition, we found that CNE2 cells had the highest MIF level in NPC cell lines, so CNE2 cells were used in subsequent experiments.

Because the expression of MIF was higher in terminal patients, we speculated that MIF expression might be related to metastasis. To verify this conjecture, we compared the MIF staining intensity of patients with metastasis and patients without metastasis. We found that the MIF expression of patients with lymphatic metastasis was significantly higher than patients without lymphatic metastasis. Also, the MIF expression of patients with distant metastasis was notably higher than patients without distant metastasis (Figures 2A–C). In addition, the survival rate of patients was closely related to the expression of MIF. The higher the level of MIF expression was, the worse the prognosis of the patients (Figure 2D). The relevant clinical data were sorted in Table 1. Briefly, elevated MIF expression was significantly associated with tumor stage (p < 0.0001), lymph node metastasis (p = 0.0002), distant metastasis (p = 0.0175) and the survival rate of patients (p = 0.0447) but not with other clinicopathological parameters, including age and sex (Table 1).

Since the upregulation of MIF was significantly related to advanced disease stage, metastasis, and poor survival rate, we used a BALB/c mouse animal model to determine the function of MIF in vivo. Firstly, we injected luciferase labeled CNE2 cells into nude mice through the tail vein. Three days after the injection, the first group was intraperitoneally injected with DMSO every 3 days, the second group was injected with PBS, and the third group was injected with ISO-1 (an inhibitor of MIF). The growth of the tumor was then monitored by measuring the amount of bioluminescence (BLI). We found that the MIF inhibitor significantly inhibited the metastatic ability of CNE2 cells (Figures 3A,B). The number of lung metastasis nodules in this group of mice injected with the MIF inhibitor ISO-1 was significantly less than the other two groups (Figures 3C,D). This shows that MIF is a high-risk factor that promotes the metastasis of nasopharyngeal carcinoma.

MIF is also known as an inhibitory factor of macrophage migration and has been confirmed to be related to the migration function of macrophages. Then, we speculated that MIF was very likely to regulate the function of macrophages, thus affecting the metastasis of NPC. Stoppe C et al. found that MIF might be involved as a counter-regulatory factor to regulate tubular necroptosis or ferroptosis (Averdunk et al., 2020). This provided clues for us to explore the relationship between MIF and ferroptosis.

Tumor-associated macrophages (TAMs) are one of the most common host immune cells in the tumor microenvironment and can accelerate tumor growth, remodel the extracellular matrix, and promote cancer metastasis. Since macrophages are the most abundant iron storage cells in the human body, we speculated that MIF might regulate the ferroptosis of macrophages. We used a functional inhibitor of MIF (ISO-1) and MIF recombinant factors to treat macrophages, and the results showed that in M0, M1, or M2 macrophages, the expression of GPX4 decreased with the inhibition of MIF function. Still, the recombinant factor MIF restored the expression of GPX4 (Figure 4A). As an important part of the ferroptosis loop, ferrous iron plays the same important role as GPX4. In the experiment measuring divalent iron, after inhibiting the function of MIF, the content of divalent iron in macrophages decreased, while the recombinant factor MIF restored this effect (Figures 4B,C). Alexandr A. Kapralov et al. found that M1 polarized macrophages were the most resistant to RSL3-induced ferroptosis than M0 or M2 cells (Kapralov et al., 2020). Therefore, we were curious about the relationship between the ferroptosis and polarization of macrophages. We cocultured M0 macrophages with ISO-1 and found that after ISO-1 was added, the level of M0 macrophage differentiation into the M2 type decreased (Figure 4D). It suggested that MIF might promote the M2-type polarization of macrophages. Next, we endeavored to test the sensitivity of macrophages to a standard pro-ferroptosis treatment with a specific GPX4 inhibitor, RSL3 (1S,3R)-2-(2-chloroacetyl)-2,3,4,9-tetra-hydro-1-(methoxycarbonyl) phenyl-1H-pyrido (3,4-b) indole-3-carboxylic acid). We measured the ROS content of each group of macrophages by flow cytometry and we found that after inhibiting the MIF function of macrophages, the intracellular ROS content was significantly increased, and the recombinant factor MIF greatly inhibited ROS (Figures 4E,F). However, the increase in ROS caused by RSL3 could be effectively prevented by a selective ferroptosis inhibitor, ferrostatin-1 (Fer-1). The CCK8 experiment revealed that the ISO-1 treated cells were vulnerable and consistently responded with a loss of viability, whereas the MIF treated cells exerted high resistance to ferroptosis (Figure 4G). Above results proved that MIF could inhibit ferroptosis in macrophages. At the same time, we found that in the absence of RSL3, high MIF levels also decreased ROS content or cell death in macrophages. According to the decreasing or increasing rate, we suggested that MIF mainly regulated ferroptosis nor other forms of cell death.

Then, how did highly expressed MIF in NPC reach macrophages? We had previously successfully extracted and identified exosomes derived from NPC and found that exosomes in the microenvironment of NPC played a significant role in cell-to-cell communication, thus promoting angiogenesis, invasion, and metastasis (Bao et al., 2018; Shan et al., 2018; Duan et al., 2019; Zhang et al., 2021). To determine whether exosomes mediate the transfer of MIF to macrophage, we isolated exosomes from the serum of normal people and NPC patients. The characteristic features of exosome lipid bilayer membranes can be identified by transmission electron microscopy (TEM) (Figure 5A). In addition, after M0, M1, and M2 macrophages were cocultured with PKH67-labeled exosomes for 2 h, the uptake of PKH67-labeled exosomes was observed by fluorescence microscopy (Figure 5B). This indicated that exosomes derived from NPC could indeed be taken up by macrophages. We have also verified the serum-derived exosomes markers (Figure 5C). Next, we quantified the expression of MIF in normal nasal mucosal cells, NPC cell lines, and macrophages. We found that the MIF level of NPC cell lines was significantly higher than that of the normal nasal mucosal cell line NP69 and M0, M1, and M2 macrophages (Figures 5D,E), and GPX4 (glutathione peroxidase 4) expression was positively correlated with MIF (Figures 5D,F). The expression level of MIF in the exosomes of NPC cell lines was also significantly higher than normal (Figures 5G,H). To quantify the specific level of MIF in exosomes, the Western blotting results proved that the MIF level in patients with NPC was higher than that in healthy volunteers (Figures 5I,J). We also conducted the CCK8 experiment to compare the effects of soluble MIF and exosomal MIF on the viability of macrophages (under RSL3 treatment). Results showed that soluble MIF in the supernatant could also inhibit ferroptosis, but their influence on ferroptosis effectiveness was weaker than exosomal MIF (Figure 5K). This indicates that MIF derived from NPC may be delivered to macrophages through exosomes to play a role in regulating ferroptosis.

To confirm the possible regulatory effect of MIF in exosomes on the ferroptosis of macrophages. We collected exosomes secreted by CNE2 cells transfected with knockdown MIF lentivirus and CNE2 cells cocultured with recombinant MIF. The results showed that in M0, M1, and M2 cells, the exosomes derived from the MIF knockdown group significantly reduced the content of divalent iron in the macrophages, and adding exosomes derived from the recombinant MIF group reversed this result, indicating that exosomal MIF can inhibit ferroptosis of macrophages (Figures 6A,B). It was also observed by transmission electron microscopy that after coculture with CNE2-derived exosomes and MIF knockdown, the mitochondria of macrophages decreased, the membrane density increased, and the outer mitochondrial membrane was broken. After adding CNE2-derived exosomes that were cocultured with recombinant factor MIF, the effect was reversed (Figure 6C). Similarly, the expression of GPX4 was the same as indicated above. After macrophages were cocultured with CNE2-derived exosomes that knocked down MIF, the expression of GPX4 decreased. However, after coculture with recombinant factor MIF, CNE2-derived exosomes increased GPX4 expression (Figure 6D). Next, we tested the ROS content in macrophages. Consistent with the above results, MIF in exosomes reduced the ROS content in macrophages (Figures 6E,F). Once again, the results of CCK8 also indicated that MIF strengthened the resistance of macrophages to ferroptosis (Figure 6G).

To verify the effect of macrophages on the migration and invasion of CNE2 after ingesting the exosomes secreted by CNE2, we did a chamber assay. The results showed that macrophages exposed to exosomes with low MIF expression significantly reduced the migration and invasion ability of CNE2 (Figures 7A,B). To further verify the relationship between MIF, macrophage ferroptosis, and metastasis, we performed immunohistochemistry of lung metastases in mice from Figure 3. Immunohistochemistry and iron staining showed that the expression of MIF was positively correlated with the number of M2 macrophages but was opposite to the number of M1 macrophages. After inhibiting MIF function, ferric iron in tissue increased significantly, but the expression of GPX4 was the lowest (Figures 7C–H). This proved that MIF promoted tumor metastasis by inhibiting ferroptosis in macrophages.