microRNA-21 upregulates YAP by inhibiting transcription factor RUNX1 to regulate immunosuppressive ability of myeloid-derived suppressor cells in lung cancer CURRENT STATUS: UNDER REVIEW

Background Lung cancer is one of the most frequently fatal cancers. The microRNA-21 (miR-21) has a known oncogenic function on immune cells in tumors, but its biological role and clinical significance in immunosuppression of lung cancer remains largely enigmatic. Our study aims to explore the role and molecular mechanisms of miR-21 in lung cancer. Methods We observed that miR-21 and YAP were highly expressed, while RUNX1 was poorly expressed in lung cancer tissues and cell lines. The loss- and gain- function approaches were performed to determine the roles of miR-21, RUNX1 or YAP in the immunosuppressive ability of myeloid-derived suppressor cells (MDSCs) in lung cancer. Results MiR-21 inhibition, YAP knockdown or RUNX1 overexpression reduced the proportion of MDSCs in lung cancer tissue and peripheral blood, but increased the proportion of T helper (Th) and CTL. Furthermore, miR-21 inhibition, YAP knockdown or RUNX1 overexpression increased apoptosis of MDSCs, more cells at G0/G1 phase, fewer cells at G2/M phase, but reduced IL-10, TGF-β and GM-CSF levels. The effect of miR-21 and RUNX1 on tumor growth was verified by xenograft tumors in nude mice. Conclusions Taken together, miR-21 upregulates YAP expression by inhibiting RUNX1 and consequently promotes the immunosuppressive ability of MDSCs against lung cancer. Th: T Gene Expression the Cancer Genome Atlas; GEPIA: Gene Expression Profiling Interactive Analysis; Eagle Medium; phosphate buffer NC: negative control; CFSE: succinimidylester; PI: iodide; RIPA:

threshold (|logFC|>1, p < 0.01). The miRNA with the lowest p value was selected for following analysis. The downstream target genes of the miRNA were predicted by mirDIP (Integrated Score > 0.2) (http://ophid.utoronto.ca/mirDIP/) and starbase (clipExpNum ≥ 3) (http://starbase.sysu.edu.cn), and differential expression analysis of the lung cancer dataset GSE74706 also was performed in R language (|logFC|>1, p < 0.01). There were 36 samples in the GSE74706 dataset, including 18 normal and 18 lung cancer samples. Then, the human transcription factors were obtained from the Cistrome database (http://cistrome.org), and these results were intersected to determine the downstream target gene of miR-21 based on the existing studies. Next, the related genes of the target gene were containing 10% fetal bovine serum, 100 U/mL penicillin and 100 U/mL streptomycin at 37℃ in a humidified atmosphere with 5% CO 2 . The cells were observed daily as they grew into a monolayer adhering to the chamber walls. When the cells were in a logarithmic growth phase, they were trypsinized and passaged with 0.25% trypsin solution once every 2-3 days. The ratio of living cells was assessed by trypan blue staining. The cell concentration was diluted to 1 × 10 7 cells/mL in phosphate buffer saline (PBS), and the cell suspension was prepared for the following experiments.
Animal model establishment A total of 72 C57BL/6 mice with half males and half females (aged 5-6 weeks, purchased from Beijing Vital River Laboratory Animal Technology Co., Ltd., Beijing, China) were selected. The mice of the average weight of (20 ± 0.5) g were numbered after acclimation for 1 week, and 8 of them (half male and half female) were randomly selected as the blank control, and subcutaneously injected with 0.4 mL normal saline into the armpit of the right forelimb. Then the remaining 64 mice were inoculated with 0.2 mL Lewis lung cancer cell suspension (about 2 × 10 6 tumor cells). The tumor growth was observed and measured three days later.
When the tumor diameter had reached about 5 mm (around the 5th day), every 8 mice (half male and half female) were infected with lentiviral particles containing miR-21 antagomir negative control (NC), miR-21 antagomir, miR-21 antagomir NC combined with sh-NC, miR-21 antagomir combined with sh-NC, miR-21 antagomir combined with sh-RUNX1, miR-21 antagomir NC combined with oe-NC, miR-21 antagomir combined with oe-NC, or miR-21 antagomir combined with oe-YAP, respectively. In brief, mice were intraperitoneally injected with 4 mg/kg normal saline (antagomir NC) or the same amount of miR-21 antagomir, and other mice were also intraperitoneally injected with the same amount of lentiviral particles or normal saline for 5 times, respectively. Then, on the 7th, 14th and 21st day after injection, the blood sample was collected from each mouse, and the tumor diameter was measured and recorded.

Preparation of mouse tumor and peripheral blood cell suspension
After about 3 weeks of subcutaneous injection of Lewis lung cancer cells, mice were anesthetized with intraperitoneal injection of 3% pentobarbital sodium at a dose of 50 mg/kg. Peripheral blood samples taken from orbit were put in 15 mL centrifuge tube (containing 1 mL anticoagulant). The mice were euthanized with cervical dislocation and displaced in 75% alcohol for 5 minutes. Then, the mice were fixed on a foam board, and the tumors were removed by scissors and forceps, before the tumors were placed into the 24-well plate with l mL PBS.
The tumor suspension was prepared as follows. In brief, the resected tumor was photograph and its length and width were measured with a vernier caliper, and the tumor volume was calculated as (length × width × width)/2. The tumor was cut with scissors into small portions in a 24 hole plate, transferred into a 15 mL centrifuge tube, added with 2 mL 1640 culture solution and 60 µL collagenase, and incubated in a shaker at 37℃ for 3 hours. After that, the centrifuge tube was mixed by vortexing, and added with PBS to a constant volume of 10 mL. After centrifugation, the supernatant was discarded and 2 mL PBS was added into the tube. Next, the solution was passed through a sieve into a 15 mL centrifuge tube, re-centrifuged and the supernatant was discarded.
Finally, PBS was added to make the tumor cell suspension.
The peripheral blood cell suspension was prepared. In brief, 1 mL PBS was added into the 15 mL centrifuge tube containing anticoagulant, centrifuged and the supernatant was discarded. Upon addition of 3 mL of hemolytic solution, the tube was put aside for 4-5 minutes, and added with PBS to obtain the peripheral blood cell suspension.

Flow cytometry for cell characterization
MDSCs were characterized as follows. A total of 50 µL of peripheral blood or tumor local cell suspension was transferred respectively into 2 mL centrifuge tubes, and 0.5 µL of CD45.2, CD11b and Gr-1 (BD Biosciences, Franklin Lakes, NJ, USA) was added to each tube. Then the tubes were then added with 48 µL of PBS, and stained in a refrigerator at 4℃ for 20 minutes. Then, the tubes were added with 1 mL of PBS and centrifuged at 300 rpm for 5 minutes, added with 100 µL PBS for resuspension, and finally analyzed with a flow cytometer.
T cell phenotype was detected. Briefly, 50 µL of peripheral blood and tumor local cell suspension was placed into 2 mL centrifuge tubes, and then mixed with 0.5 µL of CD45.2, CD4 and CD8a (BD Biosciences, Franklin Lakes, NJ, USA). Then, 48 µL of PBS was added to the tubes for staining in the refrigerator at 4℃ for 20 minutes. Then, the tubes were added with another 1 mL of PBS and centrifuged at 300 rpm. After removing the supernatant, 100 µL PBS was added for resuspension and analysis with a flow cytometer.
Flow cytometry for cell proliferation Lymphocyte isolating medium was adopted to separate cells in tumor tissues and peripheral blood samples. Then, the CD4 + or CD8 + T cells in the samples were separated using a CD4 + or CD8 + T cell separation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). According to the manufacturer's instructions (Invitrogen Inc., Carlsbad, CA, USA), CD4 + or CD8 + T cells were labeled with 5-(and 6)carboxyfluorescein diacetate, succinimidylester (CFSE), and stimulated with Con A (Sigma-Aldrich Chemical Company, St Louis MO, USA). MDSCs and T cells were derived from mice infected with lentiviral particles or control mice in the co-culture experiments. For a single co-culture, the T cells and the MDSCs were derived from the same mice. Then, the T cells were co-cultured with MDSCs at the proportions of 2:1, 4:1, 10:1 or 100:1 in 96 well plates for 96 hours. On the fourth day, the cells were analyzed by a flow cytometer. The proliferation data from MDSCs were obtained through the gradient experiments on average.

MDSC sorting of peripheral blood
The mice were euthanized, and the peripheral blood was collected, followed by removal of red blood cells with red blood cell lysis buffer. Then, the lysed blood was incubated at 4℃ for 15 minutes with the addition of biotion-conjugated anti-Gr-1 (BD Biosciences, Franklin Lakes, NJ, USA), and then added with anti-biotion beads and incubated in the dark at 4℃ for 15 minutes. After washing, the blood was resuspended with PBS and MDSC sorting was performed using a LS column from Miltenyi Biotec (Bergisch Gladbach, Germany), following the manufacturer's instructions.

Flow cytometry for cell cycle analysis
MDSCs were collected and the cell concentration was adjusted to 1 × 10 6 cells/mL. Then, the cells were seeded into 24 well plates, and treated with antagomir NC + sh-NC, miR-21 antagomir + sh-NC, miR-21 antagomir + sh-RUNX1, antagomir NC + oe-NC, miR-21 antagomir + oe-NC, or miR-21 antagomir + oe-YAP. After 24 hours, the cells were collected, centrifuged at 2000 rpm for 5 minutes, fixed with 70% ethanol precooled at 4℃, and stored at 4℃. Before staining, the fixed solution was washed off with PBS, and cells were resuspended with the addition of 200 µL PBS. A total of 100 µL RNase A was added into the cells in water bath at 37℃ for 30 minutes, whereupon 400 µL propidium iodide (PI) was added. The mixture was then dyed at 4℃for 30 minutes in the dark. Before analysis, the cells were screened through a 200-mesh cell sieve, and added with 300 µL PBS to adjust the cell density. Then the cell cycle was analyzed by a flow cytometer, with the red fluorescence recorded at 488 nm, and 10000 cells were counted.
Annexin V-FITC/PI analysis MDSCs were collected and the cell density was adjusted to 1 × 10 6 cells/mL. Then, the cells were seeded into 24 well plates, and infected with antagomir NC + sh-NC, miR-21 antagomir + sh-NC, miR-21 antagomir + sh-RUNX1, antagomir NC + oe-NC, miR-21 antagomir + oe-NC, or miR-21 antagomir + oe-YAP. After 24 hours, the cells were centrifuged at 2000 rpm for 5 minutes, washed twice with PBS, and suspended with the addition of 100 µL of binding buffer. The Annexin V-FITC kit was used for staining. A total of 5 µL Annexin V-FITC was added to the cell suspension, and then 5 µL PI was added and reacted for 15 minutes in the dark. Next, the cells were passed through a 200 mesh cell sieve.
Finally, the cells were analyzed by a flow cytometer with the excitation wavelength at 488 nm and the emission wavelength at 530 nm, and 10000 cells were counted.

Enzyme linked immunosorbent assay (ELISA)
The eyeballs of mice were removed to collect blood samples, which were stored overnight at 4℃, and then centrifuged at 3500 × g. The clear serum from the upper layer was stored at -80℃. The serum levels of interleukin 10 (IL-10), transforming growth factor-beta (TGF-β) and granulocyte-macrophage colony stimulating factor (GM-CSF) were measured by ELISA kit. The serum was cultured for 24 hours, whereupon the culture medium was collected, centrifuged at room temperature for 10 minutes at 1000 × g, and the supernatant was taken. The standard curve was drawn and the contents of IL-10, TGF-β and GM-CSF in the cell culture medium were measured in strict accordance with the instructions of the kit. The above kits were all purchased from Wuhan Xinqidi Biological Technology Co. Ltd. (Wuhan, China).

RNA immunoprecipitation (RIP) assay
Lewis lung cancer cells were lysed with radio-immunoprecipitation assay (RIPA) cell lysis buffer (P0013B, Beyotime Biotechnology Co., Shanghai, China) on ice bath for 5 minutes, and centrifuged at 12000 × g and 4℃ for 10 minutes. One portion of the cell extract was removed to serve as input, and the other part was incubated with antibody for co-precipitation. Each co-precipitation reaction system was washed with 50 µL magnetic beads and resuspended in 100 µL RIP wash buffer, and then incubated with 5 µg antibody for binding. After washing, the magnetic beads-antibody complex was resuspended in 900 µL RIP wash buffer and incubated overnight with 100 µL cell supernatant at 4℃. Samples were then placed on magnetic pedestals to collect beads-protein complexes, whereupon the samples and input were detached by treatment with protease K to extract RNA for subsequent polymerase chain reaction (PCR) analysis. The antibodies used here were anti-RUNX1 (ab92336, Abcam Inc., Cambridge, UK) and immunoglobulin G (IgG, ab150077, Abcam Inc., Cambridge, UK), which served as NC.

Chromatin immunoprecipitation (ChIP) assay
The Lewis lung cancer cells were fixed with formaldehyde for 10 minutes to induce DNA-protein crosslinking. The, an ultrasonicator was used to break the chromatin into fragments for 15 cycles of ten seconds each, with intervals of ten seconds. After that, the supernatant was collected, divided into two equal portions, and centrifuged at 12,000 × g for 10 minutes at 4℃. The IgG (ab150077, Abcam Inc., Cambridge, UK) and protein specific antibody anti-RUNX1 (ab92336, Abcam Inc., Cambridge, UK) were added into the two tubes, respectively, which were incubated at 4℃ overnight. The DNA-protein complex was precipitated by Protein Agarose/Sepharose, and centrifuged at 12,000 × g for 5 minutes.
The supernatant was discarded and the nonspecific complex was washed to remove the cross-linking with incubation at 65℃ overnight. The DNA fragments were extracted and purified with phenol/chloroform, and the binding of RUNX1 and YAP promoter was measured by RT-qPCR with YAP promoter region specific primers.
Then, the tissues were immersed in 3% H 2 O 2 for 10 minutes and antigen retrieval was performed by high pressure in a pressure cooker for 90 seconds. The tissues were cooled at room temperature and cut into slices, which were blocked with 5% bovine serum albumin (BSA), incubated at 37℃ for 30 minutes, added with 50 µL RUNX1 rabbit polyclonal antibody (ab92336, Abcam Inc., Cambridge, UK) and YAP rabbit monoclonal antibody (ab52771, Abcam Inc., Cambridge, UK) for incubation at 4℃ overnight. The next day, after washing with PBS for 2 minutes, the slices were added with 50 µL HRP

RT-qPCR
Total RNA was extracted using the RNeasy Mini Kit (Qiagen company, Hilden, Germany). The cDNA was synthesized using the reverse transcription kit (RR047A, Takara Bio Inc., Otsu, Shiga, Japan) and the miRNA first strand cDNA synthesis first (tailing reaction) kit (B532451-0020, Shanghai Sangon Biotechnology Co. Ltd., Shanghai, China). RNA loading was performed using the SYBR® Premix Ex Taq™ II (Perfect Real Time) kit (DRR081, Takara Bio Inc., Otsu, Shiga, Japan). RT-qPCR was carried out in a real-time PCR instrument (ABI 7500, ABI Company, Oyster Bay, NY). The general negative primer for miRNA and the upstream primer for glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were provided by miRNA First Strand cDNA Synthesis (Tailing Reaction) kit, and other primers were synthesized by Shanghai Sangon Biotechnology Co. Ltd. (Shanghai, China) ( Table 1). The Ct value of each well was recorded. GAPDH or U6 were taken as internal references, and the relative expression of each gene was calculated by the 2 −ΔΔCt method.

Western blot analysis
The total protein in tissues or cells was extracted by radioimmunoprecipitation assay (RIPA) containing phenylmethylsulfonyl fluoride (PMSF). Bicinchoninic acid (BCA) kit was used to measure the total protein concentration. A total of 50 µg protein was dissolved in 2 × sodium dodecyl sulfate (SDS) sample buffer and boiled at 100℃ for 5 minutes. The protein was separated by a sodium dodecyl sulfate-polyacrylamide electrophoresis (SDS-PAGE) and transferred onto a polyvinylidene fluoride (PVDF) membrane, and sealed with 5% skim milk solution at room temperature for 1 hour.
The membrane was incubated overnight at 4℃ with primary antibodies purchased from Abcam Measurement data were summarized by mean ± standard deviation. When data were in normal distribution and homogeneity, data between two unpaired groups were compared by unpaired t-test.
Measurement data among multiple groups were compared by one-way analysis of variance (ANOVA)

Downregulation of miR-21 inhibits the immunosuppressive ability of MDSCs in lung cancer
Through the R language differential analysis of GSE63805 in the GEO database, we found 19 differentially expressed miRNAs, among which miR-21 exhibited the most significant expression (Fig. 1A). We decided to explore the role and mechanism of miR-21 in the immunosuppressive ability of MDSCs in lung cancer.
First of all, in order to determine the expression of miR-21 in lung cancer, we drew a box line map by extracting the expression data of miR-21 of dataset GSE63805, which found overexpression of miR-21 in lung cancer samples (Fig. 1B). RT-qPCR analysis showed that, compared with normal tissue, the expression of miR-21 in lung cancer tissues was higher. Compared with the mouse lung epithelial cells-12 (MLE-12), miR-21 expression was upregulated in all of the lung cancer cell lines (p < 0.05) (Fig. 1C, D). Statistical analysis of the relationship between miR-21 expression and clinical indicators of lung cancer patients showed that the expression level of miR-21 was closely related to tumor size, tumor node metastasis (TNM) stage, smoking history, and presence of lymph node metastasis (Table 2). Large cell lung cancer 4 1 3 Expression of miR-21 in mice treated with miR-21 antagomir was lower than in mice treated with antagomir NC (p < 0.05) (Fig. 1E). In modeled mice, the proportion of MDSCs with CD11b + Gr-1 + in the peripheral blood was increased. Then, we selected peripheral blood and tumor tissues to further evaluate the effect of miR-21 on the immunosuppression of lung cancer. We used a flow cytometer to analyze the proportion of MDSCs, T helper (Th) and cytotoxic T lymphocyte (CTL) in the peripheral blood and tumor tissues of mice treated with miR-21 antagomir. The results indicated that, compared with the antagomir NC, the proportion of MDSCs was notably decreased, while the proportion of Th and CTL was increased in mice treated with miR-21 antagomir (p < 0.05) (Fig. 1F, G). Next, CD4 + or CD8 + T cells labeled with CFSC were detected by flow cytometry (Fig. 1H), which revealed that, compared with mice treated with antagomir NC, MDSCs in the peripheral blood and tumor tissues of mice injected with miR-21 antagomir markedly reduced the inhibition on the proliferation of Th and CTL (p < 0.05) (Fig. 1I, J). Then the expression of MDSCs functional markers ARG-1, iNOS was measured by RT-qPCR and western blot analysis, which showed markedly reduced expression of ARG-1, iNOS in the mice injected with miR-21 antagomir compared with the NC (p < 0.05) (Fig. 1K. L).
ELISA analysis showed that IL-10, TGF-β and GM-CSF levels in the mice injected with miR-21 antagomir were much lower than the NC (Fig. 1M). Therefore, downregulated miR-21 can retard the immunosuppressive ability of MDSCs on lung cancer. miR-21 promotes the expression of YAP by inhibiting RUNX1 To further examine the downstream regulatory mechanism of miR-21, we obtained 2290 downstream target genes of miR-21 through mirDIP and 1273 through starBase, and 4036 differentially expressed genes in lung cancer by analysis of the GSE74706 dataset in the GEO database ( Fig. 2A). We compared the predicted downstream target gene of miR-21 with the differentially expressed gene and the human transcription factors in Cistrome, and selected 8 transcription factors with significant differences in the downstream target genes of miR-21 in lung cancer (Fig. 2B). GeneMANIA found of 20 genes related to RUNX1 (Fig. 2B), and Cistrome predicted 485 highly correlated target genes of RUNX1 in lung adenocarcinoma. We took the intersection of GeneMANIA and Cistrome results to identify the key downstream target of RUNX1, which proved to be YAP1 (Fig. 2D). Through GEPIA analysis of lung adenocarcinoma and lung squamous cell carcinoma data in the TCGA database, we found that RUNX1 and YAP1 were negatively correlated (Fig. 2E). Besides, the starbase website predicted that miR-21 could inhibit the expression of transcription factor RUNX1 (Fig. 2F). Therefore, we speculated that miR-21 might regulate YAP expression by inhibiting RUNX1 expression in mice.
The microarray-based analysis suggested the presence of binding sites between miR-21 and RUNX1 ( Fig. 2F). At the same time, the targeting relationship between miR-21 and RUNX1 was confirmed by the dual luciferase reporter gene assay, which showed that the fluorescence intensity of cells cotransfected with miR-21 inhibitor and RUNX1-WT was higher than that in cells treated with miR-21 NC only (p < 0.05), while there was no such significant difference in the cells co-transfected with miR-21 inhibitor and RUNX1-WT (P > 0.05) (Fig. 2G). The combination of miR-21 and RUNX1 detected by RIP assay indicated greater enrichment of miR-21 and RUNX1 in cells treated with Ago2 group than cells treated with IgG (p < 0.05) (Fig. 2H).
Immunohistochemistry results showed that, compared with the NC, the expression of RUNX1 in tumor bearing mice was lower, while the expression of YAP was increased (p < 0.05) (Fig. 2I). Then, RUNX1 was knocked-down to interfere specifically with the small RNAs, and the interference efficiency was detected. Results showed that, compared with the cells treated with sh-NC, the expression of RUNX1 was notably decreased in cells treated with sh-RUNX1-1, sh-RUNX1-2, and sh-RUNX1-3 (p < 0.05), indicating that the low expression vector was successfully transfected. The expression of RUNX1 in cells treated with sh-RUNX1-3 decreased most significantly, so it was selected for the following experiments (Fig. 2J).
The results of RT-qPCR showed that the mRNA expression of YAP in the cells transfected with sh-RUNX1 was higher than in the NC cells (p < 0.05) (Fig. 2K). The results of ChIP assay indicated that, compared with the cells treated with IgG, the DNA of YAP gene promoter in cells transfected with RUNX1 was remarkably increased (p < 0.05), indicating that the transcription factor RUNX1 regulated YAP expression in the gene promoter region. In addition, when RUNX1 was knocked down in Lewis lung cancer cells, the enrichment of RUNX1 on the YAP gene promoter was reduced (p < 0.05) (Fig. 2L). According to the dual luciferase reporter gene assay, the fluorescence intensity of cells cotreated with sh-RUNX1 and YAP-WT was much higher than in the cells treated with sh-NC (p < 0.05), while there was no significant difference in the cells co-treated with sh-RUNX1 and YAP-MUT (p > 0.05) (Fig. 2M). Then the expression of miR-21 was measured by RT-qPCR (Fig. 2N) miR-21 regulates the immunosuppressive ability of MDSCs against lung cancer via promoting the expression of YAP mediated by RUNX1 (p < 0.05). Furthermore, cells co-treated with miR-21 antagomir and oe-NC, or miR-21 antagomir and oe-YAP exhibited lower miR-21 expression than cells treated with antagomir NC and oe-NC (Fig. 3A).
Western blot analysis showed that the level of RUNX1 was notably increased and the level of YAP was lower in the cells co-transfected with miR-21 antagomir and sh-NC compared to cells co-treated with antagomir NC and sh-NC. Compared with the cells co-treated with miR-21 antagomir and sh-NC, the level of RUNX1 was decreased while the level of YAP was dramatically increased in the cells co- oe-YAP compared with mice treated with both miR-21 antagomir and oe-NC (p < 0.05) (Fig. 3G-H).
ELISA assay showed that the expression of IL-10, TGF-β and GM-CSF in cells co-treated with miR-21 antagomir and sh-NC was dramatically lower than that in cells co-transfected with antagomir NC and sh-NC, while opposite results were seen in cells treated with miR-21 antagomir and sh-RUNX1 in comparison to miR-21 antagomir and sh-NC. The levels of IL-10, TGF-β and GM-CSF were decreased in the cells treated with miR-21 antagomir and oe-NC than that in the cells treated with both antagomir NC and oe-NC, but opposite results were witnessed in cells treated with both miR-21 antagomir and oe-YAP than that co-treated with miR-21 antagomir and oe-NC (p < 0.05) (Fig. 3I). The results of RT-qPCR showed that the expression of ARG-1 and iNOS in the mice treated with both miR-21 antagomir and sh-NC was lower than that in the mice treated with both antagomir NC and sh-NC, but that in the mice co-treated with miR-21 antagomir and sh-RUNX1 was notably higher than the former (p < 0.05). The levels of ARG-1 and iNOS in the mice co-treated with miR-21 antagomir and oe-NC were much lower than in the mice injected with antagomir NC and oe-NC in combination. However, but the expression of ARG-1 and iNOS in the mice co-treated with miR-21 antagomir and oe-YAP were much higher than in the mice co-treated with miR-21 antagomir and oe-NC (p < 0.05) (Fig. 5D).
RT-qPCR results showed that the expression of miR-21 in the mice co-treated with miR-21 antagomir and sh-NC, miR-21 antagomir and sh-RUNX1 was lower than that in the mice co-treated with antagomir NC and sh-NC (p < 0.05). Compared with the mice co-treated with antagomir NC and oe-NC, the level of miR-21 in the mice inoculated with both miR-21 antagomir and oe-NC, miR-21 antagomir and oe-YAP was unaffected (p < 0.05) (Fig. 5E).
Western blot analysis indicated that the protein level of RUNX1 was notably increased but the protein level of YAP, ARG-1 and iNOS was remarkably decreased in the mice co-treated with miR-21 antagomir and sh-NC compared to that in the mice co-treated with antagomir NC and sh-NC, while it was reciprocal in mice co-treated with miR-21 antagomir and sh-RUNX1 compared with miR-21 antagomir and sh-NC (p < 0.05). Compared with the mice co-treated with antagomir NC and oe-NC, the protein level of RUNX1 was increased while the protein levels of YAP, ARG-1 and iNOS were notably decreased in the mice co-transfected with miR-21 antagomir and oe-NC, but the results were opposite in the mice co-treated with miR-21 antagomir and oe-YAP than the latter (p < 0.05) (Fig. 5F).
In conclusion, miR-21 promotes tumor development through elevating the expression of RUNX1mediated YAP.

Discussion
As a main cause of cancer-related death among men and women, the occurrence of lung cancer has a close link with smoking and the use of tobacco products, as well as such environmental factors as air pollution [17]. Although chemotherapy and the integration of targeted therapy for lung cancer have made some progress, the overall effect is still not satisfactory with respect to long term survival [18].
At the same time, the understanding of the molecular mechanisms of tumor immunology, especially the inhibition of anti-tumor immune response mediated by immune synapses or immune checkpoints, has grown rapidly in the past decade [19]. Therefore, the purpose of this study is to explore the specific molecular mechanism of miR-21 on the immunosuppressive ability of MDSCs against lung cancer.
Certain miRNAs can regulate the differentiation, maturation and function of immune cells, and thus act as important contributors in maintaining cellular homeostasis and the development of distinct physiological systems [20]. Initially, our study found that miR-21 was upregulated in the lung cancer, which was in line with findings provided by Wu et al. [21]. Similarly, miR-21 was reported to be highly expressed in NSCLC and regulates invasion and chemo-sensitivity of the cancer by mediating SMAD7 [22]. In addition, downregulation of miR-21 inhibits proliferation and migration of non-small cell lung cancer cells by mediating programmed cell death 4 [23]. However, it is not clear whether and how miR-21 participates in the differentiation and functional regulation of MDSCs. ELISA, T cell proliferation analysis and immunohistochemistry suggested that downregulation of miR-21 can inhibit the immunosuppressive ability of MDSCs to lung cancer. Similarly, others have shown that miR-21 can regulate the immune resistance of myelogenous suppressor cells to tumor [24].
On the other hand, RUNX1 was predicted the downstream target gene of miR-21 by mirDIP and starBase. Indeed, RUNX1 is an important regulator of hematopoiesis, and is known to be related to the enhancement of metastasis [25]. Furthermore, miR-9 regulated MDSCs differentiation by targeting RUNX1, an essential transcription factor in regulating MDSC differentiation and function [20]. In our study, RUNX1 was downregulated in the lung cancer. Similarly, RUNX1 is downregulated and negatively correlated with MDSC-mediated immunosuppression in lung cancer [26]. Also, Rasip1 is regulated by the transcription factor RUNX1 to promote migration of NSCLC [25]. According to the intersection of GeneMANIA and Cistrome results, YAP is the downstream target of RUNX1 in lung cancer. Previous research has shown that RUNX1 can inhibit YAP expression to accelerate the occurrence of tumors and inhibit the expression of its target gene [13]. YAP is a highly expressed tumor protein in NSCLC, which plays an important role in regulating the growth and invasion of tumor cells [27]. Furthermore, YAP promotes tumor development through regulating the infiltration of MDSCs [28,29]. Hence, our study suggested that miR-21 can upregulate the level of YAP by targeting RUNX1 to regulate the immunosuppressive ability of MDSCs in lung cancer.
A previous study has indicated that overexpression of miR-21 increased invasion of MCF-7 cancer cells, and enhanced EGF-mediated invasion and TGF-β-mediated invasion in breast cancer [30]. In addition, miR-21 functions as an upstream regulator of IL-10 by targeting the 3' untranslated region of IL-10 mRNA [31]. Recently, it was demonstrated that miR-21 targets IL-10 mRNA and plays a proinflammatory role by retarding IL-10-expressing regulatory B cell (B10) differentiation [32].
Similarly, the ELISA results in our study indicated that silencing miR-21 can inhibit the expression of IL-10, TGF-β and GM-CSF, which was reserved by silencing RUNX1. All in all, miR-21/RUNX1/YAP axis can promote the cycle and apoptosis of MDSCs, and inhibit the effect of key immunosuppressive molecules.
Taken together, the upregulation of miR-21 inhibits the expression of the downstream target gene RUNX1 in lung cancer. RUNX1 can bind to the promoter region of YAP to downregulate its expression.
We conclude that miR-21 upregulated YAP expression by inhibiting the transcription factor RUNX1 to regulate the immunosuppressive ability of MDSCs against lung cancer (Fig. 6). Our findings not only improved the understanding of how miR-21 modulates the immunosuppressive ability of MDSCs in lung cancer, but also offered a potential prognostic marker and a therapeutic target for lung cancer.   The effect of miR-21/RUNX1/YAP axis on immunosuppressive ability of MDSCs in lung cancer in vitro. A, the proportion of MDSCs in peripheral blood cells before and after magnetic bead sorting; B, the cell cycle of MDSCs analyzed by a flow cytometer; C, apoptosis map of measured by ELISA; E, the expression of ARG-1 and iNOS detected by RT-qPCR and western blot analysis. Measurement data were expressed as mean ± standard deviation. * p < 0.05 vs. cells co-treated with antagomir NC and sh-NC or antagomir NC and oe-NC, # p < 0.05 vs. cells co-treated with miR-21 antagomir and sh-NC or miR-21 antagomir and oe-NC.
Measurement data among multiple groups were compared by ANOVA with Tukey's post hoc test. The experiment was repeated three times.