Tumor-associated macrophages (TAM) contribute greatly to hallmarks of cancer. Notch blockade was shown to arrest TAM differentiation, but the precise role and underlying mechanisms require elucidation. In this study, we employed a transgenic mouse model in which the Notch1 intracellular domain (NIC) is activated conditionally to define the effects of active Notch1 signaling in macrophages. NIC overexpression had no effect on TAM differentiation, but it abrogated TAM function, leading to repressed growth of transplanted tumors. Macrophage miRNA profiling identified a novel downstream mediator of Notch signaling, miR-125a, which was upregulated through an RBP-J–binding site at the first intronic enhancer of the host gene Spaca6A. miR-125a functioned downstream of Notch signaling to reciprocally influence polarization of M1 and M2 macrophages by regulating factor inhibiting hypoxia inducible factor-1α and IRF4, respectively. Notably, macrophages transfected with miR-125a mimetics increased phagocytic activity and repressed tumor growth by remodeling the immune microenvironment. We also identified a positive feedback loop for miR-125a expression mediated by RYBP and YY1. Taken together, our results showed that Notch signaling not only supported the differentiation of TAM but also antagonized their protumorigenic function through miR-125a. Targeting this miRNA may reprogram macrophages in the tumor microenvironment and restore their antitumor potential. Cancer Res; 76(6); 1403–15. ©2016 AACR.
Tumor-associated macrophages (TAM) play pivotal roles in tumor microenvironment to facilitate tumor growth and metastasis (1–3). TAMs inhibit antitumor immunity by recruiting myeloid-derived suppressor cells (MDSC) and regulatory T cells (Tregs), and by repressing CD8+ cytotoxic T cells (1–3). TAMs are characterized by a molecular signature reminiscent of alternatively activated (M2) macrophages. These macrophages, with IL4-activated macrophages as a prototype, express higher levels of immunosuppressive cytokines such as TGFβ and IL10, together with arginase-1 (Arg-1), mannose receptor (MR) and other molecules involved in anti-inflammatory and/or tissue remodeling (4–6). TAMs also secret EGF and VEGF and other growth factors to promote cancer cell proliferation and tumor vascularization, respectively. In contrast to TAMs or M2 macrophages, lipopolysaccharide (LPS)- and IFNγ-stimulated macrophages or M1 macrophages upregulate IL12, inducible nitric oxide synthase (iNOS), and TNFα, accompanied by increased antigen presentation capacity. Macrophages with the M1 phenotype repress tumor growth through phagocytosis and enhanced antitumor immunity (1–3). Therefore, it is possible to reeducate TAMs to elicit antitumor activities, given that the regulation and mechanisms of macrophage polarization are established (7–9).
The Notch-RBP-J (recombination signal-binding protein Jκ) signaling pathway plays critical roles in cell fate specification and cell plasticity (10, 11). Notch signaling is involved in macrophage activation and polarization (12–19). Recently, Franklin and colleagues have demonstrated that Notch signal is required for TAMs differentiation in a mouse mammary tumor model (20). However, the role and mechanisms of Notch in TAMs after differentiation remain to be elucidated, although NF-κB, MAPK, STAT3, interferon regulatory factor (IRF) 8, and cylindromatosis (CYLD), as well as pyruvate dehydrogenase phosphatase 1 (Pdp1)-mediated mitochondrial metabolism reprogramming, have been implicated (17, 19, 21, 22).
miRNAs participate in myeloid differentiation and macrophage activation (23–25). miR-125a is enriched in myeloid progenitors but at low level in monocytes (25, 26), and is upregulated in M1 and downregulated in M2 macrophages (23, 24). miR-125a is involved in differential activation of macrophages and other immune cells, as well as in myeloproliferative neoplasm (27–30). In this study, we show that forced Notch activation in macrophages by conditional overexpressing Notch intracellular domain (NIC) was sufficient to support TAM differentiation but abrogate TAM functions, most likely through miR-125a, leading to repressed tumor growth in mice.
Materials and Methods
Mice were maintained on C57BL/6 background in a specific pathogen–free (SPF) facility. RBP-J-floxed (RBP-Jf) mice (31) or ROSA-Stopf-NIC transgenic mice (a gift from H.L. Li) were mated with Lyz2-Cre (#019096, Jackson Laboratory) or Mx1-Cre (provided by K. Rajewsky) mice to obtain mice with appropriate genotypes. Mx1-Cre-RBP-Jf mice were further injected intraperitoneally with 300 μg poly(I)-poly(C) (Sigma) to induce RBP-J gene disruption in bone marrow (inducible conditional RBP-J knockout, RBP-JicKO; ref. 31). The ROSA-Stopf-NIC transgenic mice contain a sequence encoding NIC (amino acids 1749-2293, lacking the C-terminal PEST domain) of the mouse Notch1 followed by a IRES-GFP cassette, which is inserted into the GT(ROSA)26Sor locus. Mice were genotyped with tail DNA by PCR using the primers listed in Supplementary Table S1. All animal experiments were approved by the Animal Experiment Administration Committee of the Fourth Military Medical University.
Lewis lung carcinoma (LLC) cells and B16F melanoma cells were purchased from the authenticated ATCC repository in 2014, and cell lines were tested by PCR for the absence of mycoplasma contamination and C57BL derivation during the experiment and injected subcutaneously on rear back of C57BL/6 mice. Tumor growth was monitored by measuring tumor length (L) and short (S) with a sliding caliper (tumor size = L × S2 × 0.51). Mice were sacrificed 2 (B16F) or 4 (LLC) weeks after inoculation.
Bone marrow–derived macrophages (BMDM) were cultured from bone marrow monocytes as described previously (17). BMDMs were stimulated with LPS (50 ng/mL, Sigma) and IFNγ (20 ng/mL) or IL4 (20 ng/mL, PeproTech) for 24 hours. In some experiments, γ-secretase inhibitor IX (GSI, 30 μmol/L, Sigma) was included in the medium, with DMSO as a control. The transfection of BMDMs with synthetic miR-125a mimics or antisense oligonucleotides (ASO, Ribio) and siRNAs was performed by using Lipofectamine LTX (Invitrogen), according to the recommended protocol. The sequences of siRNAs for murine factor inhibiting hypoxia-inducible factor-1 α (FIH1), IRF4, Yin and Yang 1 (YY1), and Notch1 are shown in Supplementary Table S1. RAW264.7 cells were cultured in DMEM supplemented with 10% FCS and 2 mmol/L l-glutamine and transfected with plasmids using Lipofectamine 2000 (Invitrogen).
Recombinant mD1R, which is composed of the DSL domain of mouse Dll1 and an arginine-glycine-aspartic acid (RGD) motif, was manufactured in E. coli (32). Wells of 12-well dishes were coated with 400 μL of mD1R (50 μg/mL) at 4°C overnight, with PBS as a control. BMDMs (5 × 105) were then seeded in the wells and cultured in the presence of LPS and IFNγ for 24 hours.
Allogenic T-cell stimulation assay was performed as described previously (17).
Immunofluorescence and flow cytometry
Cells were stained with antibodies listed in Supplementary Table S2, and observed under a laser scanning confocal microscope (FV-1000, Olympus). FACS analysis was performed using a FACSCalibur and FACSAriaII flow cytometer (BD Immunocytometry Systems). Data were analyzed with the FlowJo vX.0.6 software (FlowJo, LLC). Dead cells were excluded by propidium iodide (PI) staining.
L1210 murine leukemia cells (ATCC; 1 × 106 cells/mL) were labeled with carboxyfluorescein succinimidyl amino ester (CFSE; Dojindo Molecular Technologies, Inc.), and incubated with BMDMs (1 × 105 cells/mL) on coverslips at 37°C for 2 hours. After washing, samples were stained with anti-F4/80, and visualized under a fluorescence microscope (BX51, Olympus). The average number of engulfed L1210 cells per macrophage was calculated. The engulfment of bacteria by BMDMs was determined in a similar way by using E. coli BL21 transformed with an EGFP-expressing plasmid.
BMDMs from RBP-JicKO and control mice were stimulated with LPS for 24 hours and used for miRNA expression profiling with an Agilent miRNA chip (Sanger miRBase v.12.0, 627 mouse miRNAs and 39 mouse viral miRNAs were represented) conducted by a commercial service (ShanghaiBio). The data were uploaded to Gene Expression Omnibus database (accession number GSE67364).
Total RNA preparation, reverse transcription, and real-time PCR were performed as described, with β-actin or U6 RNA (for miRNAs) as internal controls. Rapid amplification of cDNA ends (RACE) was performed using a SMARTer RACE cDNA Amplification Kit (Clontech). The PCR primers are shown in Supplementary Table S1.
The 3′-untranslated regions (UTR) of IRF4, FIH1, and RING1- and YY1-binding protein (RYBP) were amplified by PCR with a mouse cDNA library as a template. Point mutations were generated by PCR. Wild-type or mutant 3′-UTR fragments were inserted into pGL3-promoter (Promega) at the 3′ end of the luciferase gene to generate reporters (pGL3-IRF4, pGL3-HIF1, and pGL3-RYBP, respectively). The enhancer fragments of pri-miR-125a and MR genes were amplified by PCR with mouse genomic DNA as a template. These fragments and different truncations were inserted into pGL3-promoter to construct reporters (reporters 1–6 in Fig. 3A; pGL3-MR). A DNA fragment containing 4 × hypoxia responsive elements (HRE) was synthesized and inserted into pGL3-promoter to construct pGL3-HRE. pEF-BOS–NIC was as described previously (33). Full-length PU.1, IRF4, and RYBP cDNAs were amplified from a mouse cDNA library and inserted into pFlag-CMV2 to construct pFlag-PU.1, pFlag-IRF4, and pFlag-RYBP, respectively.
HeLa cells were transfected with different combinations of reporters, expression vectors, and miRNA using Lipofectamine 2000, with a Renilla luciferase vector (phRL-TK, Promega) as an internal control. Cells were harvested 24 or 48 hours after the transfection, and luciferase activity was measured with Dual Luciferase Reporter Assay using a Gloma X 20/20 Luminometer (Promega).
Chromatin immunoprecipitation (ChIP) assays were performed by using a kit (Merck Millipore) according to the manufacturer's instructions using anti-RBP-J or anti-NIC antibody. Collected immune complexes were extracted and analyzed by PCR using the primers listed in Supplementary Table S1.
Western blot analysis
Whole cell lysates were extracted with the RIPA buffer (Beyotime). Protein concentration was determined with a BCA Protein Assay kit (Pierce). Samples were separated by SDS–PAGE, blotted onto polyvinylidenefluoride (PVDF) membranes and probed with primary and secondary antibodies (Supplementary Table S2). Membranes were developed with chemoluminescent reagents (Pierce).
BMDMs were cultured, and 50 μL culture supernatants were added to 50 μL Griess reagent I and 50 μL Griess reagent II (Beyotime). Absorbance was measured at 540 nm with a microplate reader.
Images were imported into Image Pro Plus 5.1 software (Media Cybernetics Inc.) to quantify the densities of electrophoretic bands. Data were analyzed with Graph Pad Prism 5 software. Comparisons between groups were performed with unpaired Student t test or the paired t test. The results are expressed as the mean ± SD. P < 0.05 was considered significant.
Forced activation of Notch signaling in macrophages abrogated TAM phenotypes
Notch1 was the dominant type of Notch receptors expressed in macrophages (Supplementary Fig. S1). Blocking Notch signaling by RBP-J disruption arrests TAM differentiation (20). To investigate the role of Notch signaling in TAMs after differentiation, we employed mice with a NIC transgene controlled by Lyz2-Cre (hereafter named as NICcA), which was expected to have sufficient TAM differentiation. No obvious abnormal myeloid development was noticed in NICcA mice (data not shown). In NICcA mice, the growth of subcutaneously inoculated LLC and B16F tumors was delayed significantly (Fig. 1A; Supplementary Fig. S2A). The number of TAMs showed no significant difference between the NICcA and control mice (Supplementary Fig. S2B, S3A, and S3B).
Although TAMs in NICcA mice expressed comparable levels of MHC II and VCAM-1, they expressed significantly lower level of MR (Supplementary Fig. S3A). qRT-PCR confirmed that several M1 markers increased and M2 markers decreased significantly in sorted TAMs from LLC tumors on NICcA mice (Fig. 1B), suggesting a loss of TAMs phenotypes. Consistently, CD11b+Ly6G+ MDSCs decreased whereas CD8+ cytotoxic T cells increased in LLC tumors on NICcA mice (Fig. 1C; Supplementary Fig. S3C and S3D). Tumor vasculature also decreased remarkably (Fig. 1D). These results suggested that forced Notch activation in macrophages subverted TAMs phenotypes.
Identification of miR-125a as a downstream molecule of Notch signaling in macrophages
To identify molecules downstream to Notch signaling in regulating TAMs, we compared miRNA profiles between LPS-activated BMDMs derived from RBP-JicKO and control mice. Thirteen miRNAs exhibited differential expression between RBP-J-deficient and control macrophages (Fig. 2A). miR-125a was chosen for further investigation because several recent reports have highlighted its role(s) in macrophages (26–29).
We examined miR-125a expression in RBP-J–deficient (RBP-JcKO) and Notch-activated (NICcA) macrophages stimulated with PBS (M0), LPS+IFNγ (M1), or IL4 (M2). The results showed that M1 polarization led to a significant upregulation of miR-125a whereas RBP-J deficiency led to its downregulation in macrophages (Fig. 2B). Constitutive Notch activation upregulated miR-125a, even in unstimulated macrophages (Fig. 2B). During the in vitro differentiation of monocytes into macrophages, the level of miR-125a increased steadily, in correlation with RBP-J and Hey1 expression (Fig. 2C). Blocking Notch signaling by disrupting RBP-J or with GSI significantly suppressed miR-125a upregulation during macrophage differentiation in vitro (Fig. 2D; Supplementary Fig. S4).
To access the regulation of miR-125a expression by Notch signaling more specifically, we knocked down Notch1 expression in BMDMs using siRNA (Supplementary Fig. S5A). Notch1 knockdown resulted in attenuated M1 and strengthened M2 polarization of BMDMs (Supplementary Fig. S5B; ref. 17). Notch1 knockdown also reduced miR-125a expression during in vitro macrophage differentiation and activation (Fig. 2E; Supplementary Fig. S5C). Moreover, we activated Notch signaling in BMDMs with immobilized Notch ligand mD1R, and found that Notch activation resulted in upregulation of miR-125a (Fig. 2F; Supplementary Fig. S6). These data suggest that miR-125a is a downstream molecule of Notch signaling in macrophages.
Notch signaling directly transactivated the enhancer of pri-miR-125a through an RBP-J–binding site
The pri-miR-125a gene is located upstream to the sperm acrosome-associated protein (Spaca) 6 gene (NC_000083.6; ref. 30). We performed 5′-RACE to determine the transcription starting site of Spaca6/pri-miR-125a using cDNA generated from BMDMs (Supplementary Fig. S7A). Sequencing the 5′-RACE-amplified fragment identified a novel exon (exon 1′) located 3.6 kb upstream to exon 1 of Spaca6 (GenBank accession number KP893886), and an alternative splicing acceptor within exon 1 (Supplementary Fig. S7B and S7C). The first intron of Spaca6A contains three pri-miRNA genes including pri-miR-99b, pri-let-7e, and pri-miR-125a, and a putative enhancer element containing recognition sites for YY1, MYB, Smad3, and RBP-J (Supplementary Fig. S8).
The full-length and truncated pri-miR-125a enhancer fragments were inserted into a pGL3-promoter to construct serial reporter genes (Fig. 3A, left). HeLa cells were cotransfected with pEF-BOS–NIC and different reporter constructs. All constructs with the RBP-J–binding site (reporters 1, 3, 4, and 5) were transactivated by NIC, and disruption of this site by mutagenesis (reporter 6) prevented NIC-mediated transactivation (Fig. 3A, right). ChIP assay indicated that, consistent with the miR-125a expression pattern, occupation of the RBP-J–binding site by RBP-J and NIC increased significantly on day 3 of BMDM differentiation in vitro (Fig. 3B). miR-99b was regulated coordinately by Notch signaling in BMDMs (Fig. 2E, 2F, 3C; Supplementary Fig. S5C). These data indicated that Notch signaling directly transactivated the enhancer of pri-miR-125a through the RBP-J–binding site.
miR-125a functioned downstream to Notch signaling to promote M1 and suppress M2 polarization
BMDMs from normal mice were transfected with miR-125a mimics or control oligonucleotides and stimulated with PBS, LPS+IFNγ, or IL4 for 24 hours. qRT-PCR showed that miR-125a upregulated the M1 markers iNOS, IL12, and TNFα, and downregulated the M2 marker MR in BMDMs (Fig. 4A). Moreover, transfection of miR-125a into BMDMs enhanced NO production (Fig. 4B). We also cocultured miR-125a–transfected BMDMs with allogenic naïve T cells, and found that miR-125a–overexpressing BMDMs promoted stronger T-cell proliferation (Fig. 4C). In addition, miR-125a–transfected BMDMs exhibited enhanced bacterial phagocytosis, consistent with enhanced M1 polarization (Supplementary Fig. S9). These results suggested that miR-125a promoted M1 and suppressed M2 polarization. In line with this finding, transfection of an ASO of miR-125a downregulated M1 markers and upregulated MR (Fig. 4D).
In cultured BMDMs, Notch signaling enhanced M1 while suppressed M2 polarization (Supplementary Fig. S5B and S10; ref. 17). BMDMs were prepared from RBP-JcKO and control mice and transfected with miR-125a mimics or control oligonucleotides, and stimulated with PBS, LPS+IFNγ, or IL4. Although RBP-J disruption downregulated iNOS and TNFα and upregulated MR, transfection of miR-125a reversed these changes (Fig. 4E). These results indicated that miR-125a acted downstream to Notch signaling to regulate macrophage polarization.
miR-125a–targeted FIH1 and IRF4 to enhance M1 and attenuate M2 polarization simultaneously
The 3′-UTRs of FIH1 and IRF4 are potential miR-125a targets (Supplementary Fig. S11A and S12A). BMDMs were transfected with miR-125a, and the expression of FIH1 and IRF4 was determined with Western blot analysis. FIH1 and IRF4 expression decreased significantly in BMDMs transfected with miR-125a (Fig. 5A). Reporter assay showed that miR-125a reduced luciferase activity in cells transfected with reporters containing the wild-type 3′-UTR of FIH1 or IRF4, and disruption of the proximal seed sequence (302–309 bp) in the FIH1 3′-UTR or the unique seed sequence in the IRF4 3′-UTR abrogated this effect (Fig. 5B and 5C). These data suggested that miR-125a downregulated FIH1 and IRF4 in macrophages through their 3′-UTRs.
FIH1 suppresses HIF1α activity, which promotes M1 polarization through glycolysis and iNOS (34). Culturing BMDMs under hypoxia or knockdown of FIH1 with siRNA upregulated iNOS in BMDMs (Supplementary Fig. S11B and S11C). Transfection with miR-125a enhanced pGL3-HRE transactivation mildly in BMDMs, suggesting that miR-125a enhanced HIF1α activity (Supplementary Fig. S11D). On the other hand, IRF4 enhances M2 polarization (35, 36). IRF4 knockdown downregulated the M2 marker MR (Supplementary Fig. S12B). IRF4 binds to PU.1, and there are four PU.1-binding sites in or near the MR promoter (37, 38; Supplementary Fig. S12C). ChIP assay confirmed that IRF4 could be recruited to the PU.1 site in the first intron of the MR gene (Supplementary Fig. S12C), and transactivated the MR enhancer reporter (pGL3-MR) in RAW264.7 cells (Supplementary Fig. S12D), which was dependent on PU.1 in NIH3T3 cells (Supplementary Fig. S12E). Consistently, transfection of miR-125a suppressed pGL3-MR transactivation in RAW264.7 cells (Supplementary Fig. S12F).
To further evaluate the contribution of FIH1 and IRF4 to miR-125a–mediated macrophage polarization, we cotransfected BMDMs with a miR-125a ASO and siRNA targeting FIH1 or IRF4. The miR-125a ASO downregulated M1 markers IL12, iNOS, and TNFα, and upregulated the M2 marker MR. FIH1 knockdown partially rescued the effect of the miR-125a ASO by increasing IL12, iNOS, and TNFα expression (Fig. 5D). On the other hand, IRF4 knockdown reduced MR expression and nearly reversed miR-125a ASO-mediated MR upregulation (Fig. 5E). These results suggested that miR-125a promoted M1 and suppressed M2 polarization simultaneously through FIH1-HIF1α pathway and IRF4, respectively.
Macrophages overexpressing miR-125a exhibited strong antitumor activity
The expression of miR-125a was upregulated in sorted TAMs overexpressing NIC (Fig. 6A). We transduced BMDMs with a lentivirus overexpressing miR-125a and EGFP, with lentivirus expressing EGFP only as a control. These BMDMs were mixed with LLC cells and inoculated in normal mice. miR-125a-overexpressing macrophages strongly repressed tumor growth (Fig. 6B). FACS analysis of tumoral macrophages indicated that in the EGFP+ compartment, macrophages overexpressing miR-125a expressed a higher level of iNOS and a lower level of MR, suggesting that they were M1-polarized (Fig. 6C). Interestingly, in the EGFP− compartment most, if not all, host-derived macrophages were also M1-polarized (Fig. 6C, right). The expression of pri-miR-125a, iNOS, TNFα, and IL12 increased whereas the expression of MR, IL10, and TGFβ decreased significantly in tumors containing miR-125a-overexpressing macrophages (Fig. 6D). In addition, the number of CD11b+Ly6G+ MDSCs decreased in tumors and spleens, whereas CD8+ T cells increased in tumors and lymph nodes, of mice bearing tumors containing miR-125a–overexpressing macrophages (Fig. 6E). These results suggested that macrophages overexpressing miR-125a skewed the immune microenvironment of the tumors into an antitumor state. Moreover, BMDMs transfected with miR-125a exhibited stronger phagocytic activity to L1210 leukemia cells in vitro, suggesting that miR-125a enhanced direct antitumor activities of macrophages (Fig. 6F).
miR-125a amplified its own expression through RYBP
RYBP mRNA was another predicted potential target of miR-125a (Supplementary Fig. S13A). RYBP binds YY1 to suppress transcription (39, 40). Two YY1 recognition sites exist in the pri-miR-125a enhancer (Supplementary Fig. S7 and S8). Therefore, it is likely that miR-125a augments its own expression in macrophages through a negative feedback loop composed of miR-125a–RYBP/YY1–pri-miR-125a enhancer. BMDMs were transfected with miR-125a. Western blot analysis showed that miR-125a efficiently downregulated RYBP expression (Fig. 7A). Consistently, reporter assay indicated that miR-125a suppressed a reporter encoding the 3′-UTR of RYBP mRNA (Fig. 7B). RYBP overexpression significantly downregulated miR-125a and miR-99b in RAW264.7 cells (Supplementary Fig. S13B), and knockdown of YY1 with siRNA reversed the downregulation of miR-125a resulted from ectopic RYBP overexpression (Fig. 7C). Reporter assay showed that RYBP overexpression suppressed transactivation of a reporter containing the full length pri-miR-125a enhancer fragment (reporter 1 in Fig. 3A), and this suppression was reversed by knockdown of YY1 with siRNA (Fig. 7D), suggesting that RYBP inhibited miR-125a expression through YY1. In BMDMs transfected with miR-125a, the expression of both pri-miR-125a and miR-99b was enhanced (Fig. 7E). These data suggested that miR-125a could positively regulate its enhancer to augment its expression in macrophages.
A novel mechanism mediating the regulation of TAMs by Notch signaling
Conditional deletion of RBP-J abrogated CD11c+ TAMs while maintained MHCIIhiCD11bhi macrophages in the MMTV-PyMT mammary tumor model, suggesting that Notch signaling is specifically required for the differentiation of TAMs (20). In this study, to overcome this RBP-J disruption-mediated developmental arrest of TAMs, we employed conditionally activated NIC transgenic mice. FACS analyses of macrophages in tumors suggested that TAMs existed in a similar number as in the control. This provided us a chance to evaluate the consequence of Notch activation on TAMs phenotypes. Our results indicated that TAMs with forced Notch activation exhibited M1 phenotype and antitumor activity. Therefore, in addition to supporting TAMs differentiation, Notch signaling represses the tumor-promoting activity of TAMs (Supplementary Fig. S14). However, due to the limitations of the cancer model and gene-modified mice used in this study, more efforts are required to clarify other potential mechanisms underlying the discrepancy between the phenotypes of CD11c-Cre–mediated RBP-J cKO (20) and Lyz2-Cre–mediated NICcA mice (this study). First, the consequence of NIC overexpression may not be exactly compatible with the phenotype of RBP-J disruption due to the existence of noncanonical Notch signaling that is independent of RBP-J. Second, different Cre transgenic mice were used in the two studies, which might impact differentially on other myeloid cells such as neutrophils in addition to macrophages. Third, LLC cells used in this study might exert different influence on macrophages from that of the MMTV-PyMT mammary cancer cells used in the RBP-J cKO study. Finally and most importantly, it could not be excluded that subcutaneously inoculated LLC tumors might contain different subsets of macrophages from the MMTV-PyMT mammary tumors. How Notch signaling regulates these different subsets of macrophages remains an open question.
Notch signaling modulates the activation of macrophages by targeting a variety of downstream molecules (12–19, 41). Moreover, macrophages, regardless of their embryonic or bone marrow origins, proliferate in situ at sites of inflammation, such as cancer (4, 42). This raises the question of how polarized macrophages maintain their phenotypes. miRNAs are a class of noncoding RNAs involved in epigenetic regulation and macrophage activation (43). We identified miR-125a as a novel target of canonical Notch signaling in macrophages. As a downstream molecule of Notch signaling that regulates macrophage activation, miR-125a has several important properties. First, it enhances M1 polarization and suppresses M2 polarization simultaneously. We have previously shown that disrupting Notch signaling reduces M1 and increases M2 macrophage polarization, even in the presence of M1 inducers (17). We and others have also identified IRF8, CYLD, and SOCS3 as downstream targets of Notch signaling in the regulation of macrophage activation (17–19). However, these models consider that M1 polarization is a result of Notch activation and M2 polarization represents a “default” state. In this study, we showed that Notch signaling led to the upregulation of miR-125a, which may actively enhance M1 and suppress M2 polarization through FIH1 and IRF4, respectively (Supplementary Fig. S14). Second, Foldi and colleagues have shown that the Notch pathway amplifies its own signaling during macrophage activation (44). miR-125a may participate in the self-amplification of Notch signaling strength during macrophage activation by upregulating its own expression through suppressing RYBP, which functions as a transcriptional repressor of the Spaca6A/pri-miR-125a enhancer via interaction with YY1. Last but not least, it has been widely accepted that macrophages proliferate in situ in solid tumors to elicit long-lasting protumor activities (4, 42). Notch-miR-125a signaling may play a role in the epigenetic memory associated with the immune-suppressing and tumor-promoting capacities of TAMs. Further experiments are required to clarify this possibility.
miR-125a expression has been shown to be higher in M2 macrophages and to promote their polarization by targeting KLF13 (26, 27). However, in the current study, we showed that miR-125a was induced in M1 macrophages after LPS+IFNγ stimulation and that miR-125a overexpression promoted M1 polarization by targeting FIH1 while inhibiting M2 polarization by targeting IRF4. This inconsistency might be a consequence of different experimental conditions employed during the induction of macrophage activation and polarization. In contrast with granulocyte-macrophage colony-stimulating factor (GM-CSF) and M-CSF, which have been previously used to induce M1 and M2 macrophage polarization, respectively (26), we employed LPS+IFNγ or IL4 stimulation. Other studies have also shown that miR-125a expression is upregulated after stimulation with LPS, consistent with our observation and other reports (24). In addition, it should be noted that FIH1 inhibits both HIF1α and HIF2α, which play opposite roles in macrophage polarization (34). Therefore, it cannot be formally excluded that under specific conditions, miR-125a might also enhance M2 macrophage activation. A recent report has unveiled other activities of miR-125a (30), also prompting further investigations.
Notch-miR-125a signaling might also be involved in monocyte-to-macrophage differentiation in vitro (Fig. 2C; Supplementary Fig. S1A). Schroeder and Just have reported that Notch signaling promotes myeloid differentiation through RBP-J (45). However, RBP-J KO does not significantly change the number of myeloid colony-forming units in bone marrow (31). Notch signaling has been implicated in various myeloid malignancies (46). Disruption of this signaling pathway leads to myeloproliferation in mice (47, 48). Additional studies are required to elucidate the mechanism by which Notch signaling regulates myeloid development.
Education of macrophages to elicit antitumor activities using miR-125a
TAMs have been widely recognized as a major tumor-promoting population during tumor initiation, growth, invasion, and metastasis (1–3). Macrophage depletion shows tumor-repressive effects in several experimental systems (7, 8). However, as an important cell population in innate immunity, macrophages may also elicit antitumor activities (7). It has been speculated that TAMs can be “educated” to perform antitumor functions, if their phenotypes are reversible. Indeed, blocking CSF1R signaling in TAMs leads to the conversion of M2-like macrophages to M1-like macrophages (49). Moreover, a recent study has shown that low-dose irradiation of tumors induces massive increases in M1 macrophages in the tumors (9). The results of this study indicated that miR-125a–overexpressing macrophages exhibited strong antitumor activities. These macrophages possessed M1 characteristics and enhanced phagocytosis of tumor cells. More importantly, miR-125–overexpressing macrophages exhibited M1-polarized characteristics with increased secretion of TNFα and IL12. The immune microenvironment molded by miR-125a–overexpressing macrophages had higher levels of TNFα and IL12 and lower levels of IL10 and TGFβ, which mediate M1 and M2 polarization, respectively (1, 2, 7). This cytokine milieu would polarize newly recruited macrophages into M1 directly or indirectly, accompanied by enhanced CD8+ T-cell infiltration and diminished MDSC recruitment. Moreover, these macrophages proliferated in tumors while retaining their M1 characteristics, which strengthened their antitumor activities. Therefore, our findings may facilitate the development a new therapeutic strategy for tumors based on Notch- or miR-125a–modified macrophages in the future.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Conception and design: H. Han, H.-Y. Qin
Development of methodology: J.-L. Zhao, F. Huang
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J.-L. Zhao, F. Huang, F. He, C.-C. Gao, S.-Q. Liang, P.-F. Ma, G.-Y. Dong
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.-L. Zhao, F. Huang, F. He, S.-Q. Liang, P.-F. Ma, G.-Y. Dong, H. Han, H.-Y. Qin
Writing, review, and/or revision of the manuscript: H. Han, H.-Y. Qin
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.-L. Zhao, F. Huang, H. Han, H.-Y. Qin
Study supervision: H. Han, H.-Y. Qin
The authors thank H.L. Li and K. Rajewsky for mice. The study was performed at the Graduates Innovation Center of the Fourth Military Medical University.
This work was supported by grants from Ministry of Science and Technology (2015CB553702) and National Natural Science Foundation of China (31130019, 31371474, 31570878, 81530018, 31301127, 81170963, 31071291).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.