The human fibrosarcoma cell line HT-1080, synovial sarcoma cell line SW-982 and human skin fibroblast cell line HSF were obtained from iCell Bioscience (Shanghai, China). The stable transfection of HT-1080 cells was performed as we described (9). Briefly, HT-1080 cells were transfected with lentivirus-packaged recombinant plasmids expressing miR-939-3p or negative control (Tsingke Biotechnology, Shanghai, China), and resistant cells were selected by using Puromycin (Sigma, St. Louis, MO, USA) to establish stable cell lines.

Sarcoma, colorectal cancer (CRC) and the corresponding paratumor tissues from 12 patients were separately collected from the People’s Hospital of Xishui County (Zunyi, China). The study was carried out in accordance with the Declaration of Helsinki and approved by the People’s Hospital of Xishui County. Written informed consent has been obtained from all patients.

Four-week-old female nude mice were purchased from the Beijing Huafukang Biotechnology (Beijing, China). The mice were subcutaneously injected with 5 × 10⁶ HT-1080 cells in 0.2 mL PBS per mouse with or without stable miR-939-3p overexpression (n = 5). After 21 days, the xenografts were excised, measured for volume and weight, and further processed for RNA and protein extraction, which was approved by the Animal Welfare and Ethics Committee of Army Medical University.

Total RNA of HT-1080 cells and tumor tissues was extracted by using an RNA Purified Total RNA Extraction Kit (Beyotime), and then reversely transcribed into cDNA with RT Master Mix (MedChemExpress, Monmouth Junction, NJ, USA), as we previously described (24). qPCR was performed using a SYBR Green qPCR kit (MedChemExpress). Specific primer sets were provided in Supplementary Table S1 .

To quantify miR-939-3p, total RNA, including miRNA, was extracted using the RNAiso Plus reagent (TaKaRa, Shiga, Japan), which was reversely transcribed into cDNA by using a miRNA First Strand cDNA Synthesis kit from Sangon (Shanghai, China). qPCR was performed using a SYBR Green qPCR kit (MedChemExpress). The primer sets were listed in Supplementary Table S2 .

Western blot was performed as we described (24). Briefly, proteins were extracted from sarcoma cells and tissues by using RIPA lysis buffer (Beyotime). Proteins were separated and transferred onto PVDF membranes, which were then incubated with the antibodies against BATF2 (Abcam, Cambridge, MA, USA) or β-actin (Santa Cruz, Dallas, TX, USA), followed by the incubation of secondary antibody. Gray-scale of the bands were quantified by using ImageJ software.

Immunohistochemistry (IHC) staining was performed to detect the expression levels of BATF2 (Abcam) and Ki67 (Santa Cruz) in tumor xenografts as we previously described (9).

GEPIA (25) is a publicly available platform integrating data based on The Cancer Genome Atlas (TCGA) and Genotype-Tissue Expression (GTEx) project. It provides comprehensive pan-cancer analysis of RNA-sequencing data, including gene expression, correlation, and survival analysis.

The correlation between miR-939-3p expression levels and the prognosis of sarcoma patients, as well as the circulating miR-939-3p expression levels, were analyzed by using CancerMIRNome online database (26). CancerMIRNome is publicly available that enables analysis of miRNAs from TCGA projects and circulating miRNome datasets. Data are available from the Gene Expression Omnibus (GSE124158).

Potential miRNAs targeting BATF2 were predicted using miRDB, miRwalk and targetScan software programs. The potential binding sites of miR-939-3p in BATF2 3′ UTR were analyzed by using targetScan online software.

MiR-939-3p mimics or inhibitors, and negative controls were purchased from Tsingke Biotechnology, which were listed in Supplementary Table S3 . HT-1080 cells were harvested 48 h after transfection using Lipofectamine 3000 (Invitrogen).

Mammalian BATF2 overexpression plasmids (pBATF2) and empty plasmid vector pcDNA3.1 (pControl) were purchased from Tsingke Biotechnology for transfection into HT-1080 cells by using Lipofectamine 3000 (Invitrogen). siRNA targeting BATF2 and scramble siRNA were bought from Santa Cruz.

HT-1080 cells were seeded into a 96-well plate and co-transfected with vector control or BATF2 overexpression plasmids and miR-939-3p mimic or inhibitor. Cell viability was determined by using Cell Counting Kit 8 (CCK-8) (Dojindo, Kumamoto, Japan) after 48 h or a Bromodeoxyuridine (BrdU) ELISA kit (Roche, Burgess Hill, UK) according to the manufacturer’s instructions.

The human BATF2 3’ UTR sequences containing putative binding sites of miR-939-3p were cloned into pGL4 vector (Promega, Madison, WI, USA). The primer sets and restriction enzymes were listed in Supplementary Table S4 . For point mutation, the binding sites (5′-GCCCAGG-3′) of miR-939-3p in BATF2 3′ UTR were mutated into 5′-ATTTCAA-3′ by using a MutanBEST kit (Takara).

HT-1080 cells were transiently co-transfected with the recombinant plasmids or pAP-1-Luc (BD Biosciences, Oxford, UK) and negative control or miR-939-3p mimic along with Renilla plasmids using Lipofectamine 3000 (Invitrogen) for dual-luciferase reporter gene assay (Promega), as we previously described (27).

Data were presented as mean ± SD and analyzed by using Graphpad Prism 8 software. Comparisons were statistically significant at P < 0.05 by using unpaired t-test or ANOVA.

We first explored the expression levels of BATF2 in human soft tissue sarcoma. It was found that BATF2 mRNA and protein levels were significantly decreased in 12 cases of sarcoma tissues, compared to the corresponding adjacent tissues ( Figures 1A, B ). Besides, BATF2 mRNA and protein expression was also downregulated in human fibrosarcoma cells (HT-1080) and synovial sarcoma cells (SW-982), compared with that in human skin fibroblast cells (HSF) ( Figures 1C, D ). Further, sarcoma patients were divided into two groups according to BATF2 expression levels. Compared with patients with high BATF2 expression (n = 131), patients with low BATF2 expression (n = 131) had obviously worse survival (P = 0.031) with a hazard ratio of 0.65 (P = 0.031) ( Figure 1E ). These findings indicated that BATF2 was downregulated in sarcoma and negatively associated with the prognosis of sarcoma patients.

Since miRNAs play an extensive role in modulating gene expression in sarcoma (21, 22), we hypothesized that BATF2 might be regulated by miRNA. To test this hypothesis, we utilized miRDB, miRwalk and targetScan software programs to identify miRNAs that had potential binding sites in the 3′ UTR of BATF2 ( Figure 2A ). We then screened the predicted nine miRNAs using quantitative PCR (qPCR) in both sarcoma tissues ( Figure 2B ) and cell lines ( Figure 2C ), which showed that miR-939-3p and miR-455-5p were upregulated in sarcoma tissues and cells compared to controls. Further, BATF2 expression was reduced by miR-939-3p mimic and increased by miR-939-3p inhibitor in both HT-1080 and SW-982 cells ( Figures 2D–G ). However, neither the mimic nor the inhibitor of miR-455-5p could regulate BATF2 expression ( Figures 2H–K ), suggesting that BATF2 expression was repressed by miR-939-3p in sarcoma.

To explore the role of miR-939-3p, we collected sarcoma tissues and observed an elevated expression of miR-939-3p compared to pericarcinoma tissues ( Figure 3A ). Similarly, sarcoma cell lines (HT-1080 and SW-982) also exhibited higher levels of miR-939-3p compared to a human skin fibroblast cell line (HSF) ( Figure 3B ). Moreover, circulating miR-939-3p levels were found to be upregulated in sarcoma patients with various histological subtypes, including glioma, malignant soft tissue tumor, pancreatic cancer, gastric cancer, intermediate soft tissue tumor, lung cancer, benign soft tissue tumor, esophageal cancer, colorectal cancer and hepatocellular carcinoma, compared to healthy individuals ( Figure 3C ).

Functional experiments further revealed that the viabilities of HT-1080 and SW-982 cells were enhanced by miR-939-3p mimic, while suppressed by miR-939-3p inhibitor, as evidenced by CCK-8 and BrdU incorporation assay ( Figures 3D–G ). Consistent with our hypothesis, Kaplan Meier survival analysis showed a negative correlation between miR-939-3p expression levels and the prognosis of sarcoma patients, as analyzed using both the CancerMIRNome database (P = 0.014, with a hazard ratio of 2.69) ( Figure 3H ) and UALCAN databases (P = 0.0036) ( Supplementary Figure S1 ). These data revealed that miR-939-3p was significantly upregulated in sarcoma and was inversely associated with the prognosis of sarcoma patients.

Statistical analysis showed a significantly negative correlation between miR-939-3p and BATF2 expression levels in both sarcoma tissues (r = -0.8600, P = 0.0003) ( Figure 4A ) and colon cancer tissues (r = -0.7162, P = 0.0088) ( Supplementary Figure S2 ), implying the involvement of a potential miR-939-3p/BATF2 signaling pathway in sarcoma. Based on the predicted binding sites of miR-939-3p in the 3’ UTR of BATF2 mRNA ( Figure 4B ), we postulated that miR-939-3p could suppress BATF2 expression through binding to its 3’ UTR. As shown in Figure 4C , luciferase reporter assays confirmed that overexpression of miR-939-3p suppressed the luciferase activity of wild-type (WT) recombinant BATF2 3’ UTR. As expected, miR-939-3p mimic had no effect on the fluorescence intensity in mutant BATF2 3’ UTR, where potential binding sites were mutated into ATTTCAA ( Figures 4B, C ). To validate the miR-939-3p/BATF2 axis, we further examined the downstream genes of BATF2 (13). Luciferase reporter gene assay and qPCR analysis showed that both AP-1 activity and MET proto-oncogene receptor tyrosine kinase (MET) expression were induced by miR-939-3p mimic, while repressed by miR-939-3p inhibitor ( Figures 4D, E ). These results demonstrated that miR-939-3p suppresses BATF2 expression through directly binding to its 3’ UTR.

To investigate whether miR-939-3p-mediated regulation of BATF2 is involved in sarcoma proliferation, HT-1080 cells were transfected with BATF2 overexpression plasmids (pBATF2) or control plasmids (pControl) ( Figure 5A ). It was found that overexpression of BATF2 significantly alleviated the miR-939-3p-mediated suppression of BATF2 expression and promotion of sarcoma cell proliferation ( Figures 5B–E ), as well as cell migration and invasion ( Supplementary Figure S3 ). Besides, miR-939-3p could neither further enhance sarcoma cell proliferation nor regulate AP-1 activity or MET expression, when BATF2 was knocked down by a large amount of siRNA in HT-1080 cells ( Supplementary Figures S4A–E ). These findings collectively suggest that miR-939-3p could induce sarcoma cell proliferation via inhibiting the expression of BATF2.

To further examine the role of miR-939-3p, in vivo experiments utilizing xenograft nude mouse models were conducted. HT-1080 cells stably infected with lentivirus expressing miR-939-3p or control were injected into the mice separately. Compared to the control mice, the miR-939-3p group exhibited larger tumor volumes and weights ( Figures 6A–C ). As anticipated, miR-939-3p expression was significantly higher in the xenografts of miR-939-3p group, whereas the expression levels of BATF2 mRNA and protein were much lower, compared to the control mice ( Figures 6D–G ), while immunohistochemical staining of Ki67, a biomarker of cell proliferation, showed that Ki67 expression was much higher in tumor xenografts from miR-939-treated mice than that in control mice ( Figure 6H ). These data revealed that miR-939-3p enhanced sarcoma growth in vivo.