4T1 mouse breast tumor and LLC mouse lung tumor cell lines were purchased from American Type Culture Collection (Manassas, VA, USA). 4T1-luc cells expressing luciferase were obtained from PerkinElmer (Waltham, MA, USA). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; HyClone, South Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS; ThermoFisher, Waltham, MA, USA) and 1% penicillin-streptomycin, at 37°C, in a humidified 5% CO₂-containing atmosphere. Analysis of mycoplasma contamination was performed for cultured cells.

According to previous studies (21, 22), BMDMs were prepared and polarized into M1- and M2-type macrophages. Bone marrow cells were isolated from tibias and femurs of Balb/c mice and cultured in DMEM supplemented with 10 ng/mL of colony-stimulating factor-1 (CSF-1; Gibco, Carlsbad, CA, USA) and 10% FBS for 7 d. The culture medium was changed every other day. Next, the cells were incubated with 100 ng/mL of lipopolysaccharide and 20 ng/mL of recombinant mouse IFN-γ (R&D Systems, Minneapolis, MN, USA), for 24–48 h to achieve M1 polarization. For M2 polarization, BMDMs were incubated with 20 ng/mL of recombinant mouse IL-4 for 24–48 h. M2-polarized macrophages were treated with either 50 nM 5-aza-dC (Merck, Rahway, NJ, USA) for 72 h or 25 nM TSA (Merck) for 48 h. For treatment with the combination, M2-polarized macrophages were treated with 5-aza-dC for 24 h followed by treatment with 5-aza-dC and TSA for 48 h.

Concentrations of cytokines secreted into the culture medium of macrophages were measured using ELISA. One hundred microliters of cell culture medium was added to culture plates pre-coated with antibodies against IL-4 and IL-10 (R&D Systems), as well as IL-6, IL-12p40, and TGF-β (ThermoFisher Scientific) and incubated at 25°C for 1–2 h. The plates were then incubated with biotin-conjugated antibodies against the corresponding cytokines for 30 min, followed by incubation with streptavidin-conjugated horseradish peroxidase at 25°C for 30 min. The plates were incubated with stabilized chromogen followed by incubation with stop solution at 25°C for 30 min, post which the absorbance was measured at the wavelength of 450 nm using a microplate reader (ThermoFisher Scientific).

Macrophages in cultures were incubated with fluorescence-labeled antibodies against CD43, CD86, and CD206 (BioLegend, San Diego, CA, USA) and colony-stimulating factor 2 receptor alpha (CSF2RA; Santa Cruz Biotechnology, Dallas, TX, USA) for 20 min in the dark. For analysis of immune cell populations in tissues, freshly excised tumors were fragmented into several pieces. The fragmented tissues were further minced into 2–3 mm³ pieces and incubated with collagenase D (Roche, Basel, Switzerland) and DNase (Merck), at 37°C for 40 min. The tissue samples were filtered through a 100 µm-cell strainer (Falcon, Corning, NY, USA) to collect the digested cells. Dead cells and cellular debris were removed using Ficoll^® (GE Healthcare, Buckinghamshire, UK) gradient centrifugation. Collected cells were re-suspended in phosphate-buffered saline containing FBS and incubated with fluorescence dye-labeled antibodies against CD3, CD4, CD8, CD11b, CD45, CD86, CD206, F4/80, and Gr1 (BioLegend) for 30 min in the dark. Primary antibodies were diluted at a ratio of 1:100. After gating for CD45⁺ cells, the cells were analyzed for the corresponding markers. At least 10,000 events were analyzed using the FACSCalibur™ cytometer (BD Biosciences, Franklin Lakes, NJ, USA). Data were evaluated using CellQuest™ Pro (BD Biosciences).

Tumor tissues (50 mg) were homogenized in QIAzol ^® lysis reagent (Qiagen, Hilden, Germany). Macrophages in cultures were also incubated with the lysis reagent. RNAs were isolated from the tissue and cell lysates using an miRNeasy ^® Mini Kit (Qiagen) and subjected to qRT-PCR. Primers for actb, arg1, cd43, csf2ra, ifng, il-4, il-10, il-12p40, inos, and tgfb were obtained from Bioneer (Daejeon, Korea). cDNA was synthesized using PrimerScript™ 1st strand cDNA Synthesis Kit (Takara, Shiga, Japan). qPCR using SYBR™ Green (Qiagen) was performed on a real-time cycler (Bio-Rad, Hercules, CA, USA). Relative mRNA expression levels were analyzed according to the Livak method (23).

Tumor cells and M2-type macrophages were treated with 50 nM of 5-aza-dC and 25 nM of TSA for 24–72 h alone or in combination. To examine the effect of the conditioned medium (CM) of cultured macrophages, tumor cells were incubated with the CM of M2 macrophages in the absence or presence of 1 μM of paclitaxel for 24 h. After treatment, the cells were incubated with a fresh culture medium containing 10% Cell Counting Kit-8 solution (Dojindo, Kumamoto, Japan) for 1–4 h. Absorbance was measured at the wavelength of 450 nm and cell viability was calculated based on the absorbance.

Six to eight-week-old mice were purchased from Orient Bio (Seongnam, Korea). Mice were cared for and maintained in conformance with the Guidelines of the Institutional Animal Care and Use Committee of Kyungpook National University (permission no. 2015-0017). An orthotopic breast tumor model was prepared by injecting 1×10⁶ 4T1 cells into the lower left mammary fat pad of Balb/c female mice. A syngeneic lung tumor model was prepared by injecting 1×10⁶ LLC cells into the lower right dorsal flank of C57BL/6 male mice.

Whole-body bioluminescence images were captured using an IVIS^® imaging system (PerkinElmer), at 10 min after intraperitoneal injection of D-luciferin (150 mg/kg body weight) into the tumor-bearing mice. Images were taken every 3 d from the tenth day of tumor inoculation, following which tumor progression of the mice was monitored. Mice were treated by intraperitoneally injecting them with 5-aza-dC (1 mg/kg body weight) and TSA (0.3 mg/kg body weight) once a day for 5 d. Tumor sizes were measured using a digital caliper, and tumor volumes were calculated using the following formula: Volume = (L × W × H)/2 (L: length, longest dimension, W: width, shorter dimension, parallel to the mouse body, and H: height, perpendicular to the length and width). At the end of the treatment, blood was collected for analysis of hematologic parameters from a few of the mice before sacrifice, and their tumor tissues were prepared for further study. The remaining mice were maintained to check the survival rates.

For depletion of CD8⁺ T cells, female Balb/c mice were intraperitoneally injected with 400 µg of monoclonal antibody against CD8α (Bio X cell, West Lebanon, NH, USA), one day before the inoculation of 4T1 tumor cells and every 5 day thereafter. The effectiveness of CD8⁺ T cell depletion was determined by means of flow cytometric analysis of CD45⁺CD3⁺CD8⁺ cells among the splenocytes or tumor-infiltrating cells after treatments.

For depletion of macrophages, Balb/c mice were intraperitoneally injected with 200 µL of clodronate liposomes (Liposoma, Amsterdam, Netherlands) 2 days prior to start of treatments. The effectiveness of macrophage depletion was determined by flow cytometric analysis of CD45+F4/80+MHCII+ cells in the splenocytes. For adoptive transfer of macrophages, mice were pre-treated with clodronate liposomes and then treated with intravenously injection of M2 macrophages (1 × 10⁶ cells) pre-treated with combination of 5-aza-dC and TSA or with miR-7083-5p (once a day, twice a week for 3 weeks).

RNAs for miRNA profiling were isolated using the miRNeasy^® Mini Kit. Microarray analysis of miRNAs was performed using an Affymetrix GeneChip miRNA 4.0 array (Affymetrix, Santa Clara, CA, USA) by Macrogen (Seoul, Korea). Array data export, processing, and analysis were performed using Affymetrix GeneChip Command Console^® software. Hierarchical clustering heat map was obtained using the Euclidean method. Target genes of a miRNA were predicted using TargetScan analysis (https://www.targetscan.org/).

Oligonucleotide pairs containing the CSF2RA or CD43-3′-untranslated regions (3′-UTRs) were synthesized (Bioneer). The oligonucleotide sequences used were as follows: and 3′-TTTGATCGCCGGCGATCAGTTTAGGAGGAGATATCGGGGCGAGATC-5′ (CSF2RA) and and 3′-TTTGATCGCCGGCGATCAAGGTTAGACGAGGAATCGGGGCCAGATC-5′ (CD43). Four nanograms of the annealed oligonucleotides was incubated with 50 ng of pmirGLO dual-luciferase miRNA target expression vector (Promega, Madison, WI, USA), which was linearized by performing digestion with Pmel and Xbal restriction enzymes. After ligation, the miRNA target site was inserted into the 3′-end of the Firefly luciferase reporter gene. The humanized Renilla luciferase gene of the pmirGLO vector was used as a control reporter for the normalization of gene expression. HEK293T cells (5×10⁴ cells) were seeded into 24-well plates overnight and transfected with 2 µg of the pmirGLO plasmid containing the CSF2RA- or CD43-3′-UTR, in the absence or presence of 20 µM of miR-7083-5p or a scrambled control. Firefly luciferase activity normalized to Renilla luciferase activity was obtained 24 h after transfection of the dual-luciferase reporter expression vector using a SpectraMax^® L microplate reader (Molecular Devices, San Jose, CA, USA).

Data are expressed as the mean ± standard deviation (SD). All statistical analyses were performed using Prism 6 software (GraphPad Software Inc., San Diego, CA, USA). Statistical significance was determined using one-way ANOVA followed by Tukey’s multiple comparison post-hoc test or two-way ANOVA followed by Bonferroni multiple comparisons post-hoc test.

To examine whether epigenetic therapy by treatment with DNA methylation and histone deacetylation inhibitors reprograms the M2-type macrophages towards the M1-like phenotype, BMDMs isolated from mouse bone marrow were treated with CSF-1 and polarized into M2-type macrophages by treatment with IL-4, and then the M2-polarized macrophages were incubated with 5-aza-dC and TSA, alone or in combination ( Supplementary Figure 1A ). The M2-polarized macrophages displayed a more spindle-like shape as compared to the M1-polarized macrophages ( Supplementary Figure 1B ). Treatment of M2-polarized macrophages with 5-aza-dC and TSA, alone or in combination, did not affect the cell viability ( Supplementary Figure 1C ). Treatment of the M2-type macrophages with either 5-aza-dC or TSA decreased the secretion of immune-suppressive M2 cytokines such as IL-10, IL-4, and TGF-β into the culture medium ( Figure 1A ), while increasing the secretion of immune-stimulatory M1 cytokines such as IL-12 and IL-6 ( Figure 1B ). Interestingly, treatment of M2 macrophages with 5-aza-dC and TSA synergistically decreased the secretion of M2 cytokines (IL-10 and TGF-β) and increased the secretion of M1 cytokines (IL-12), as compared to treatment with either 5-aza-dC or TSA alone ( Figures 1A, B , respectively). In addition, compared to untreated M2 macrophages, treatment of M2 macrophages with the combination of 5-aza-dC and TSA significantly decreased the mRNA levels of M2 cytokines (il-10, il-4, and tgfb) and a marker (arg1) ( Figure 1C and Supplementary Figure 2A ), while increasing those of M1 cytokines (il-12p40 and ifng) and a marker (inos) ( Figure 1D and Supplementary Figure 2B ). Moreover, treatment of M2 macrophages with the 5-aza-dC and TSA decreased the surface expression of CD206 M2 marker and increased that of CD86 M1 marker, compared to untreated M2 macrophages ( Figure 1E ). The expression of M2- and M1-type cytokines and markers in the M2 macrophage, after treatment with the combination of 5-aza-dC and TSA were skewed towards that of M1 macrophages ( Figures 1A–E ).

Treatment of 4T1 tumor cells with 5-aza-dC and TSA, alone or in combination, for 24–72 h reduced the cell viability ( Figure 2A ). There were no significant differences between the single and combination treatments. Next, we examined whether the CM of M2-type macrophages after the treatments caused tumor cell cytotoxicity. Treatment of 4T1 tumor cells with the CM of M2-type macrophages treated with the combination of 5-aza-dC and TSA reduced the survival of 4T1 tumor cells at higher levels than upon treatment with the CM of M2-type macrophages treated with either therapy alone ( Figure 2B ). Moreover, the CM of M2-type macrophages treated with the combination of 5-aza-dC and TSA enhanced the sensitivity of 4T1 tumor cells to paclitaxel given with the CM, resulting in about 65% cytotoxicity ( Figure 2B ). In contrast, the CM of M2-type macrophages treated with either 5-aza-dC or TSA alone did not enhance the sensitivity of 4T1 tumor cells to paclitaxel ( Figure 2B ). The CM of M1-type macrophages used as a control also reduced the viability of the tumor cells. The CM of untreated M2 macrophages rather increased the tumor cell survival ( Figure 2B ). These results suggested that, like the M1-type macrophages, M2-type macrophages treated with the combination of 5-aza-dC and TSA secrete active inflammatory or anti-tumoral factors.

An orthotopic breast tumor was established by inoculating 4T1 mouse syngeneic breast tumor cells into the fat pad of the mammary gland of mice. Tumor-bearing mice were treated with 5-aza-dC and TSA and monitored using whole-body luminescence imaging ( Figure 3A ). Treatment with 5-aza-dC and TSA synergistically inhibited primary tumor growth ( Figures 3B–D ) as well as lung metastasis ( Figure 3E ), and increased mouse survival rate ( Figure 3F ) in 4T1 tumor-bearing mice, as compared to treatment with either 5-aza-dC or TSA alone and that with the saline control. Treatment with 5-aza-dC alone inhibited tumor growth and metastasis and increased the survival rate of 4T1 tumor-bearing mice to a lesser degree than upon treatment with the combination, while treatment with TSA alone showed no significant effects on these parameters ( Figures 3B–F ). These treatments did not show significant changes in the body weight ( Figure 3G ) and hematologic parameters including the number of white blood cells ( Supplementary Figure 3 ).

To examine whether treatment with the combination of 5-aza-dC and TSA promoted M2 reprogramming in the tumor microenvironment, the levels of macrophage cytokines/markers and the immune cell population in tumor tissues were analyzed. Treatment with 5-aza-dC and TSA synergistically decreased the mRNA levels of M2 cytokines (il-10, il-4, and tgfb) and a marker (arg1) in tumor tissues ( Figure 4A and Supplementary Figure 4A ), while increasing those of M1 cytokines (il-12p40 and ifng) in tumor tissues ( Figure 4B and Supplementary Figure 4B ), as compared to either therapy alone and in the saline-treated control. In addition to that of cytokines, the population of CD206⁺ M2 macrophages ( Figure 4C ), myeloid derived suppressor cells (MDSCs) ( Figure 4D ), and CD4⁺ T cells ( Figure 4E ) reduced more efficiently after treatment with the combination of 5-aza-dC and TSA, while increasing the population of CD86⁺ M1 macrophages ( Figure 4C ) and CD8⁺ T cells ( Figure 4F ) among the tumor-infiltrating leukocytes in the 4T1 tumor tissues, as compared to those upon treatment with 5-aza-dC or TSA alone and those in the saline-treated group. As the number of CD8⁺ T cells was considerably higher in the 4T1 tumor tissues after the combination treatment, we examined whether the increase in the number of CD8⁺ T cells contributes to the anti-tumor growth activity. Pre-treatment of tumor-bearing mice with an anti-CD8 neutralizing antibody almost depleted the population of CD8⁺ T cells ( Supplementary Figures 5A, B ). However, depletion of CD8⁺ T cells did not significantly decrease the anti-tumor growth activity after the combination treatment ( Figure 4G ). In contrast, depletion of macrophages by pre-treating mice with clodronate almost reduced the anti-tumor growth activity by the combination treatment ( Figure 4H and Supplementary Figures 5C, D ). These results suggested that a decrease in the M2 macrophage population but not an increase in the CD8⁺ T cell population is the primary factor for the therapeutic effect of the treatment with the combination of 5-aza-dC and TSA.

Treatment with the combination of 5-aza-dC and TSA significantly inhibited the growth of the subcutaneous LLC mouse syngeneic lung tumor ( Supplementary Figure 6A ) and increased the mouse survival rate ( Supplementary Figure 6B ), as compared to treatment with either 5-aza-dC or TSA alone and saline-treated control. Similarly to 4T1 breast tumor, in the LLC lung tumor as well, treatment with 5-aza-dC and TSA synergistically decreased the levels of M2 cytokines and markers ( Supplementary Figure 6C ), while increasing those of M1 cytokines ( Supplementary Figure 6D ), as compared to treatment with either therapy alone. In addition, it decreased the population of CD206⁺ M2 macrophages ( Supplementary Figure 6E ), MDSCs ( Supplementary Figure 6F ), and CD4⁺ T cells ( Supplementary Figure 6G ), while increasing that of CD86⁺ M1 macrophages ( Supplementary Figure 6E ) and CD8⁺ T cells ( Supplementary Figure 6H ), as compared to treatment with either therapy alone. These results showed that the combination of 5-aza-dC and TSA efficiently reprograms M2-type TAMs into an M1-like phenotype and leads to an increase in the anti-tumor immunity in the tumor microenvironment, as well as inhibition of tumor growth.

To identify miRNAs that are involved in the M2 to M1 reprogramming of macrophages upon treatment with the combination of 5-aza-dC and TSA, we analyzed miRNA profiles in M2-type macrophages, before and after the treatments. Hierarchical clustering heat map showed the upregulation and downregulation of miRNAs in M2-type macrophages, after treatment with the combination of 5-aza-dC and TSA ( Figure 5A ). When treated with 5-aza-dC and TSA, the levels of 139 miRNAs were upregulated in the M2-type macrophages, as compared to those in the untreated M2 macrophages ( Figure 5B ). In addition, the levels of 97 miRNAs were higher in the M1-type macrophages than in the untreated M2 macrophages, and 60 of these miRNAs overlapped with the miRNAs found in M2 macrophages upon treatment with the combination of 5-aza-dC and TSA ( Figure 5B ). In contrast, the levels of 116 miRNAs were downregulated in the M2-type macrophages treated with 5-aza-dC and TSA in combination than in the untreated M2 macrophages ( Figure 5B ). The levels of 110 miRNAs were lower in the M1-type macrophages than in the untreated M2 macrophages, and 70 of these overlapped with the miRNAs found in M2 macrophages upon treatment with 5-aza-dC and TSA ( Figure 5B ). The miRNAs whose levels were upregulated and downregulated by more than three-fold in M2 macrophages treated with the combination of 5-aza-dC and TSA and in M1 macrophages, as compared to those in untreated M2 macrophages, are listed in Supplementary Tables 1 , 2 , respectively. Among the upregulated miRNAs, miR-7083-5p level was remarkably (approximately 36-fold) higher in M2 macrophages treated with the combination of 5-aza-dC and TSA ( Supplementary Table 1 ), whereas, it was not upregulated in M1 macrophages ( Supplementary Table 2 ), as compared to that in untreated M2 macrophages. In contrast, the expression of certain miRNAs such as miR-7043-5p and miR-184-3p was upregulated both in the M2 macrophages treated with the combination of 5-aza-dC and TSA ( Supplementary Table 1 ) as well as in M1 macrophages ( Supplementary Table 2 ), as compared to that in untreated M2 macrophages. qRT-PCR analysis showed the upregulation of miR-7083-5p in 4T1 tumor tissues ( Figure 5C and Supplementary Figure 7A ) as well as M2-type macrophages in culture ( Figure 5D and Supplementary Figure 7B ) after treatment with 5-aza-dC and TSA. Transfection of M2 macrophages with miR-7083-5p decreased the mRNA levels of M2 cytokines and markers ( Figure 5E and Supplementary Figure 7C ) and secretion of M2 cytokines into the culture medium ( Figure 5F ). In contrast, upregulation of miR-7083-5p in M2 macrophages increased the mRNA levels of M1 cytokines and markers ( Figure 5G and Supplementary Figure 7D ) and secretion of M1 cytokines into the culture medium ( Figure 5H ). Upregulation of miR-7083-5p in M2 macrophages, as well as treatment of M2 macrophages with 5-aza-dC and TSA, decreased the surface expression of CD206 M2 marker, while increasing that of CD86 M1 marker ( Figure 5I ).

To further examine whether miR-7083-5p is involved in the M2 macrophage reprogramming, we performed adoptive transfer of M2 macrophages pre-treated with miR-7083-5p or combination of 5-aza-dC and TSA into mice following treatment with clodronate. The adoptive transfer of pre-treated M2 macrophages inhibited the tumor growth while that of untreated M2 macrophages enhanced tumor growth compared to saline-treated control ( Figure 6A ). The adoptive transfer of pre-treated M2 macrophages decreased the levels of M2 cytokines ( Figure 6B and Supplementary Figure 8A ) and CD206 ( Figure 6D ) and the population of MDSCs ( Figure 6E ) and CD4+ T cells ( Figure 6F ), with no significant changes in that of CD8+ T cells ( Figure 6G ), compared to saline-treated control. In contrast, the adoptive transfer increased the levels of M1 cytokines ( Figure 6C and Supplementary Figure 8B ) and CD86 ( Figure 6D ) compared to saline-treated control.

TargetScan database analysis was carried out to predict candidate target genes of miR-7083-5p ( Supplementary Table 3 ). Functional enrichment analysis of the 33 predicted genes using DAVID software (24) indicated four genes (cd43, csf2ra, parp3, and satb1) involved in immune/inflammatory response ( Figure 7A ). Of the four genes, we further focused on the genes with high expression in the monocytes/macrophages from The Human Protein Atlas (25) and selected two genes, cd43 and csf2ra ( Figure 7B ). CSF2RA, also known as granulocyte-macrophage colony-stimulating factor receptor alpha, and CD43, also known as sialophorin or leukocyte marker, are expressed in macrophages and are involved in tumor progression (26–31). To examine the targeting of the mRNAs of csf2ra and cd43 genes by miR-7083-5p, HEK293T cells were transfected with a luciferase reporter vector inserted with csf2ra- or cd43-3′-UTR, in the absence or presence of miR-7083-5p or a scrambled control ( Figure 7C ). Transfection of miR-7083-5p reduced both CSF2RA and CD43 luciferase reporter activities, as compared to transfection with a scrambled control in HEK293T cells ( Figure 7D ). To further examine the role for miR-7083-5p in the regulation of CSF2RA and CD43 expression, M2 macrophages were transfected with miR-7083-5p. Flow cytometry and qRT-PCR analysis showed that the upregulation of miR-7083-5p reduced the protein ( Figure 7E ) and mRNA ( Figure 7F and Supplementary Figure 9A ) levels of CSF2RA and CD43 in the transfected M2 macrophages, as compared to those in the M2 and M1 macrophages). Moreover, the mRNA levels of csf2ra and cd43 were significantly lower in the tumor tissues of mice bearing 4T1 ( Figure 7G and Supplementary Figure 9B ) and LLC ( Figure 7H and Supplementary Figure 9C ) tumors after treatment with the combination of 5-aza-dC and TSA than those seen upon either therapy alone.