C57BL/6 mice were purchased from the Academy of Military Medical Science (Beijing, China). All the mice used were 6–8 weeks old and housed in a specific pathogen-free animal facility at the Experimental Animal Center of Tianjin Medical University (Tianjin, China). The care and treatment for mice were performed according to guidelines for Laboratory Animal care and were approved by the Animal Ethics Committee of Tianjin Medical University (Tianjin, China).

Mouse pancreatic cancer cell line Panc02 and human pancreatic cancer cell line PANC1 were originally obtained from American Type Culture Collection. These cells were cultured in medium containing Dulbecco's Modified Eagle Medium(DMEM), 10% fetal bovine serum (FBS), 100 U/mL penicillin and 100 µg/mL streptomycin, and incubated at 37°C in a humidified atmosphere with 5% CO₂.

The 3′ UTR of wild type (wt) or mutant CD47 were synthesized and cloned into pmirGLO dual-luciferase miRNA target expression vector (Promega). The pmirGLO dual-luciferase 3′ UTR vectors were cotransfected with miR-340 overexpression vector into 293 T cells. The cells were harvested and lysed 48 hours after transfection. The luciferase activity was measured by a dual-luciferase assay system (Promega).

To establish the orthotopic model of PDAC, mouse abdomens were prepared with betadine solution, and an approximately 1 cm wide incision was performed in the upper left of each abdomen. The tip of the pancreas was gently grasped, and the pancreas and spleen were externalized in the lateral direction to expose them completely. A needle was inserted into the pancreas tail and positioned in the pancreatic head region. Then, 2×10⁶ Panc02 cells stable overexpression of miR-NC or miR-340 suspended in 50 µL phosphate buffered saline (PBS) were slowly injected using an insulin injection needle. The spleen was restored to the appropriate position in the abdomen, and the skin and peritoneum were closed with 3-0 vicryl sutures. After 3 weeks, all mice were euthanized, and the tumor tissues were collected for further study.

Panc02 cells stable overexpression of miR-NC or miR-340 were washed three times with PBS, and 2×10⁶ cells suspended in 100 µL PBS were subcutaneously injected into the dorsal part of the mice. From day 10, the tumor size was measured every 2 days and tumor volume was calculated as length × width²/2. Mice with tumors larger than 20 mm in size at the longest axis were euthanized due to ethical considerations. To analyze the effect of miR-340 on tumor growth, all mice were euthanized for further study.

For in vivo experiments, the tumors were weighed and digested at 37°C in 10 mL digestion solution (PBS supplemented with type I collagenase (200 U/mL), hyaluronidase and DNase I (100 µg/mL)) for 1 hour. Single cell suspensions were obtained by grinding the digested tissues and filtering through a 70 µm cell strainer (Becton and Dickinson, BD, USA). The immune cells were freshly isolated using density gradient centrifugation and stained with antibodies for 30 min at 4°C. The following monoclonal anti-mouse antibodies were used: CD45-PECy5.5 (eBioscience, USA), CD3-PECy7 (eBioscience, USA), CD4-APC (BioLegend, USA), CD8-APC (eBioscience, USA), F4/80-APC (Sungene, China), CD11b-PECy7 (eBioscience, USA) and MHC II-PE (eBioscience, USA). Flow cytometry was performed and the data were analyzed using FlowJo software (TreeStar, Ashland, Oregon, USA).

To obtain macrophages from the peritoneal cavity, thioglycollate-elicited peritoneal macrophages were collected 96 hours after an intraperitoneal injection of a 3% thioglycollate solution. The cells were harvested from the peritoneal cavity in 10 mL of PBS and cultured in Roswell Park Memorial Institute (RPMI)-1640 supplemented with 10% FBS and 1% penicillin–streptomycin for 2 hours at 37°C with 5% CO₂ to macrophages. Then the non-adherent cells were removed. To obtain macrophages from the bone marrow, bone marrow cells were isolated from the femur and tibia of C57BL/6 mice and cultured in complete RPMI-1640 medium supplemented with 10 µg/mL recombinant mouse M-CSF (PeproTech) in a CO₂ incubator for 5 days at 37°C to differentiate into macrophages. The cells were washed twice every other day with PBS and then cultured with fresh medium. The macrophages acquired from both the peritoneal cavity and bone marrow (5×10⁴ per well) were separately seeded in 24-well culture plates in complete RPMI-1640 medium for 24 hours prior to the experiment. The macrophages were incubated in serum-free medium for 2 hours, and 2×10⁴ GFP⁺ cancer cells were added. After co-cultivation at 37°C for 4 hours, the cells were harvested, and the macrophages were stained with macrophage antibody anti-mouse F4/80-APC (Sungene, China); then, the phagocytosis was inspected using a Fluoview FV1000 Laser Scanning Confocal Microscope (Olympus, Japan). The phagocytosis was calculated as the number of phagocytosed GFP⁺ cells per 100 macrophages. For the flow cytometry analysis of phagocytosis, 10,000 cells per sample were analyzed using a FACS Canto II flow cytometer (BD, USA), and unstained control and single stained cells were prepared for gating. Phagocytosis was calculated as the percentage of F4/80⁺GFP⁺ cells among F4/80⁺ macrophages.

The orthotopic model of PDAC was established in accordance with protocols described previously. In brief, C57BL/6 mice were injected with 2×10⁶ Panc02 cells stable overexpression of miR-NC or miR-340 into the pancreas. Tumors were allowed to grow for 7–10 days and treated with isotype control IgG (200 µg/day intravenously, Clone No. MPC-11, BioXcell) or anti-mouse CD47 mAb (200 µg/day intravenously, Clone No. MIAP301, BioXcell) in vivo. After 2 weeks of treatment, the mice were sacrificed, and tumors were removed and digested to perform flow cytometry. The phagocytosis was calculated as the percentage of F4/80⁺GFP⁺ cells among CD11b⁺F4/80⁺ macrophages.

The depletion of macrophages in C57BL/6 mice was performed as previously described23 with minor modifications. In brief, the mice were injected intravenously with 200 µL of either clodronate liposomes or control liposomes (Clodronateliposomes.com, Amsterdam, The Netherlands) 1 day before in situ injections of miR-NC or miR-340 overexpression Panc02 cells into the pancreas, and then the tumor-bearing mice were injected intravenously with 200 µL of the respective liposomes every 3 days. After performing the depletion protocol six times, the mice were sacrificed for further study. The effectiveness of the macrophage depletion was determined by a flow cytometry analysis of the percentage of macrophages (CD45⁺CD11b⁺F4/80⁺) among splenocytes or tumor-infiltrating cells.

The candidate targeting miRNAs of CD47 were derived from integrated miRNA target prediction algorithms (miRanda,24 DIANA-microT25 and TargetScan26). The clinical relevance of CD47 and miR-340 was confirmed by data from the Oncomine (https://www.oncomine.org)27 and LinkedOmics (http:// www.linkedomics.org).28

The statistical analysis was performed by two-tailed unpaired Student’s t-test or two-way Analysis of Variance (ANOVA) and presented as the mean±SEM using the Prism software (GraphPad, USA). The correlation was evaluated by Spearman’s correlation analysis. The log-rank Mantel–Cox test was used for survival analysis. The differences were considered significant at p<0.05.

First, we evaluated the expression of CD47 in primary human tissues from “normal” adjacent non-tumor pancreas and PDAC tissues from the Oncomine database. Although CD47 was still detectable in the normal (non-tumor) pancreatic tissues, the expression of CD47 in primary PDAC tissues was significantly higher (p<0.001) (figure 1A). Then, we explored whether CD47 was a prognostic factor of PDAC. In the Cox regression analysis performed with LinkedOmics, the stratification of patients into “CD47 high” and “CD47 low” groups based on the median expression revealed that high levels of CD47 expression were associated with decreased probability of survival in PDAC (figure 1B). To explore the potential regulation of CD47 by miRNAs, we used miRNA target-predicting algorithms (miRanda, DIANA-microT and TargetScan) based on the presence of binding sites in the 3′ UTR and found eight candidate miRNAs that overlapped among these algorithms (figure 1C). Then, we used LinkedOmics to identify miRNAs that were significantly inversely correlated with CD47 expression in three cancers, namely, PDAC, lung adenocarcinoma (LUAD) and skin cutaneous melanoma (SKCM), and explored three candidate miRNAs that overlapped among the three kinds of cancers (figure 1D). As shown in figure 1E–G, miR-340 was inversely correlated with CD47 in 32 LUAD samples, 97 SKCM samples and 178 PDAC samples. In addition, the heat map showed that miR-340 ranked as the fourth most significantly negatively correlated miRNA with CD47 expression in PDAC (figure 1H). These findings indicated that miR-340 may be closely related to pancreatic cancer development and progression. To determine the expression of miR-340 and its correlation with the prognosis of PDAC, we analyzed data from starBase and found that the expression of miR-340 in primary PDAC tissues was significantly lower than that in normal pancreatic tissues (figure 1I). Low expression of miR-340 was associated with a decreased probability of overall survival in PDAC (figure 1J). MiR-340 was significantly increased following radiation therapy in patients suffering from PDAC (figure 1K). These results indicated that miR-340 may be a clinically relevant prognostic factor and plays critical roles in PDAC.

Evaluation of the 3′ UTR sequence of CD47 revealed one binding site with a perfectly matched sequence for miR-340. To confirm the existence of a direct relationship between miR-340 and its predicted target CD47, we cloned the 3′ UTR of CD47 into the dual-luciferase vector pmirGLO and generated mutations in the binding site to abrogate the interaction between miR-340 and the CD47 3′ UTR. As expected, while the reporter with the intact CD47 3′ UTR was effectively suppressed by miR-340, the CD47 3′ UTR carrying a mutated binding site was resistant to suppression by miR-340 (figure 2A). Additionally, we overexpressed miR-340 by transfecting miR-340 mimics into the mouse pancreatic cancer cell line Panc02, and the efficiency of the overexpression was confirmed by Quantitative Real-time PCR (qRT-PCR). We found that the overexpression of miR-340 by mimic transfection significantly decreased CD47 at both the mRNA and protein levels in Panc02 cells (figure 2B). Then, we constructed Panc02 cells with stable miR-NC and miR-340 stable overexpression, and the high expression efficiency was confirmed by qRT-PCR. We determined that, consistent with the effect of the miR-340 mimics, the stable overexpression of miR-340 could also notably decrease the expression of CD47 in Panc02 cells (figure 2C). To further confirm the regulation of CD47 by miR-340 in human pancreatic cancer cells, we overexpressed miR-340 by transfecting miR-340 mimics into the human pancreatic cancer cell line PANC1. We confirmed the overexpression of miR-340 by qRT-PCR and found that the overexpression of miR-340 by mimic transfection significantly decreased the expression of CD47 in PANC1 cells (figure 2D). Taken together, these results indicate that miR-340 downregulates CD47 through a binding site in its 3′ UTR.

To explore the effect of miR-340 on the biological function of mouse pancreatic cancer cells, we detected the effect of both miR-NC and miR-340 stable overexpression on Panc02 cell apoptosis with Annexin-V/7-AAD and flow cytometry, and no significant difference was found between the two groups (figure 2E). We also explored the proliferation of Panc02 cells with stable overexpression of miR-NC or miR-340 with the CCK-8 assay at 0, 24, 48, 72 and 96 hours and found no clear difference in proliferation (figure 2F). To detect the impact of miR-340 on the migration of Panc02 cells, we performed a wound healing assay. As shown in figure 2G, there were no obvious changes in wound healing. To explore the effect of miR-340 on the biological function of human pancreatic cancer cells, we overexpressed miR-340 by transfecting miR-340 mimics into PANC1 cells and detected cell apoptosis with Annexin-V/PI by flow cytometry. No significant difference was found between the two groups (figure 2H). We also explored the proliferation of miR-NC-transfected or miR-340 mimic-transfected PANC1 cells with the CCK-8 assay at 0, 24, 48, 72 and 96 hours and there was no clear difference (figure 2I). Similarly, no obvious impact of miR-340 on the migration of PANC1 cells was observed (figure 2J). As 2D culture does not reproduce important tumor-enriched metabolic states, we explored the impact of miR-340 on tumor cell growth in VitroGel 3D, a hydrogel similar to Matrigel. We detected the proliferation of Panc02 cells with stable overexpression of miR-NC or miR-340 with the CCK-8 assay at 0, 24, 48 and 72 hours in VitroGel 3D. Representative images of cell growth in 3D culture are shown in figure 2K, and there was no clear difference in cell viability in 3D culture (figure 2L). Then we evaluated the ability of cells in VitroGel to migrate through the 24-well cell culture inserts with or without the hydrogel VitroGel 3D. The 3D migration and invasion assays showed that overexpression of miR-340 did not impact either the migration or invasion of Panc02 cells in VitroGel 3D (figure 2M,N). Overall, the high expression of miR-340 did not directly affect apoptosis, proliferation, migration or invasion of pancreatic cancer cells.

CD47-mediated CD47–SIRPα signaling plays important roles in phagocytosis by human and mouse macrophages in different types of hematologic and solid cancers.3 29 As we have demonstrated that CD47 is the direct target of miR-340, we attempted to explore the impact of miR-340 on the ability of macrophages to engulf cancer cells. First, we isolated macrophages from the mouse peritoneal cavity and then cocultured these macrophages with Panc02 cells stably overexpressing miR-NC or miR-340. After 4 hours, the cells were harvested and stained with anti-mouse F4/80-APC, and flow cytometry was performed. The results clearly showed that compared with that of miR-NC Panc02 cells, the phagocytosis of Panc02 cells with stable overexpression of miR-340 by mouse peritoneal cavity-derived macrophages was significantly increased (figure 3A,B). The gating strategy is shown in online supplementary figure S1. Then, we performed in vitro phagocytosis assays by incubating bone marrow-derived macrophages (BMDMs) with Panc02 cells stably overexpressing miR-NC or miR-340. As expected, we found that the forced expression of miR-340 increased the level of phagocytosis by approximately twofold compared with that in the miR-NC group (figure 3C,D). We also used an anti-CD47 antibody to block CD47, and the results showed that overexpression of miR-340 significantly increased the phagocytosis by macrophages with CD47 blockade (figure 3E,F). Furthermore, we performed in vitro phagocytosis assays by incubating BMDMs with Panc02 cells stably overexpressing miR-NC or miR-340, with or without CD47 blocking, and measured phagocytosis by fluorescence microscopy. Representative images of Panc02 cell coculture groups with stable overexpression of both miR-NC and miR-340 were acquired. The white arrows indicate macrophages that engulfed cancer cells. The overexpression of miR-340 significantly increased the phagocytosis by macrophages, and phagocytosis was further increased when CD47 was blocked (figure 3G,H, online supplementary figure S2). These results indicated that miR-340 may serve as a potent therapeutic target against PDAC given its close relationship with macrophage phagocytosis.

To determine whether the forced expression of miR-340 might exert suppressive effects on cancer development, we further performed an experiment using an orthotopic model of PDAC. C57BL/6 mice were implanted with Panc02 cells stably overexpressing miR-NC or miR-340 through direct injection into the pancreas. Compared with the miR-NC Panc02 cell-implanted mice, the miR-340 stable overexpression Panc02 cell-implanted mice exhibited a significantly reduced tumor size (figure 4A–C) and tumor weight (figure 4D). Furthermore, we performed an experiment using a subcutaneous mouse model of PDAC. C57BL/6 mice were subcutaneously inoculated with Panc02 cells stably overexpressing miR-NC or miR-340 into the shaved right lateral flanks. Consistent with the results obtained using the orthotopic mouse model, this model showed that compared with the tumors in the miR-NC Panc02 cell-implanted mice, the overexpression of miR-340 significantly reduced both the tumor volume (figure 4E,F) and tumor weight (figure 4G). In addition, we detected the expression of miR-340 and CD47 in orthotopic tumors from both the miR-NC and miR-340 stable overexpression Panc02 cell-bearing mice and found that CD47 was obviously downregulated in the tumors of the miR-340 overexpression group both at the mRNA (figure 4H) and protein levels (figure 4I). We also performed immunohistochemistry and flow cytometry, and the mean intensity and the mean fluorescence intensity (MFI) of CD47 in mice bearing Panc02 cells stably overexpressing miR-NC or miR-340 confirmed that the expression of CD47 was decreased in tumors in the miR-340 overexpression group (figure 4J,K). These results indicated that miR-340 may function as a tumor suppressor by targeting CD47.

To determine whether miR-340 could enhance antitumor immunity in vivo, tumor-infiltrating immune cells from mice bearing Panc02 cells stably overexpressing miR-NC or miR-340 were stained and assessed via flow cytometry. Compared with that in the miR-NC Panc02 cell-bearing mice, the percentage and MFI of the M1-like (CD11b⁺F4/80⁺MHC-II⁺) phenotype in the tumors were significantly increased in the miR-340 stable overexpression group (figure 5A). The ratio of M1-like and M2-like cells in the tumors from the miR-340 stable overexpression group was dramatically increased (figure 5B). Consistent with the changes in tumors, the percentages and MFI of the M1-like phenotype in the spleens of mice bearing Panc02 cells overexpressing miR-340 were obviously increased (figure 5C). The M1-like and M2-like cell ratio in the spleens in the miR-340 stable overexpression group was markedly increased (figure 5D). Although the CD4⁺ T lymphocytes showed no statistically significant difference between the tumors from mice bearing Panc02 cells with stable overexpression of miR-NC or miR-340 (figure 5E), the CD8⁺ T lymphocytes were significantly upregulated in the miR-340- overexpressing tumors (figure 5F). In the spleens from the tumor-bearing mice, both CD4⁺ T lymphocytes and CD8⁺ T lymphocytes were significantly upregulated in the miR-340 overexpression groups (figure 5G,H). In the blood from the tumor-bearing mice, although the CD4⁺ T lymphocytes showed no statistically significant difference between the miR-NC and miR-340 stable overexpression Panc02 cell-bearing mice (figure 5I), the CD8⁺ T lymphocytes were significantly upregulated in the miR-340-overexpressing mice (figure 5J), which was consistent with the results obtained in the tumors and spleens. In addition, to further demonstrate that miR-340 regulated phagocytosis by macrophages in vivo, we established an orthotopic model of PDAC and detected the changes in phagocytosis by macrophages using flow cytometry. An anti-CD47 antibody was used as a positive control. Our research showed that the phagocytosis of macrophages increased significantly in miR-340-overexpressing tumor-bearing mice and further increased with CD47 blocking (figure 5K), which showed that miR-340 could effectively enhance phagocytosis by macrophages in vivo.

To investigate whether macrophages contribute to the inhibitory effect of miR-340 on tumor formation by CD47-expressing cancer cells, we depleted mice of macrophages by administering clodronate liposomes prior to performing in situ injections of cancer cells into the pancreas throughout the experiment (figure 6A). The depletion of macrophages (CD11b⁺F4/80⁺) in the tumors was effective with an approximate decrease of 50% after the intravenous injection of clodronate liposomes (figure 6B,C) and clearly decreased after the intravenous injection of clodronate liposomes in the spleens (figure 6D,E). We found that clodronate liposome treatment partially inhibited tumor growth, but the combination had no additive effects on either tumor growth (figure 6F,G) or tumor weight (figure 6H), confirming that macrophages indeed contribute, at least partially, to the antitumor effect of miR-340. Therefore, considering our in vitro and in vivo studies, miR-340 directly inhibited the expression of CD47, resulting in the increased phagocytosis of tumor cells by macrophages, consequently enhancing antitumor immunity (figure 6I).