Colony stimulating factor-1 receptor signaling networks inhibit mouse macrophage inflammatory responses by induction of microRNA-21

Abstract

Macrophage polarization between the M2 (repair, protumorigenic) and M1 (inflammatory) phenotypes is seen as a continuum of states. The detailed transcriptional events and signals downstream of colony-stimulating factor 1 receptor (CSF-1R) that contributes to amplification of the M2 phenotype and suppression of the M1 phenotype are largely unknown. Macrophage CSF-1R pTyr-721 signaling promotes cell motility and enhancement of tumor cell invasion in vitro. Combining analysis of cellular systems for CSF-1R gain of function and loss of function with bioinformatic analysis of the macrophage CSF-1R pTyr-721-regulated transcriptome, we uncovered microRNA-21 (miR-21) as a downstream molecular switch controlling macrophage activation and identified extracellular signal-regulated kinase1/2 and nuclear factor-κB as CSF-1R pTyr-721-regulated signaling nodes. We show that CSF-1R pTyr-721 signaling suppresses the inflammatory phenotype, predominantly by induction of miR-21. Profiling of the miR-21-regulated messenger RNAs revealed that 80% of the CSF-1-regulated canonical miR-21 targets are proinflammatory molecules. Additionally, miR-21 positively regulates M2 marker expression. Moreover, miR-21 feeds back to positively regulate its own expression and to limit CSF-1R-mediated activation of extracellular signal-regulated kinase1/2 and nuclear factor-κB. Consistent with an anti-inflammatory role of miRNA-21, intraperitoneal injection of mice with a miRNA-21 inhibitor increases the recruitment of inflammatory monocytes and enhances the peritoneal monocyte/macrophage response to lipopolysaccharide. These results identify the CSF-1R-regulated miR-21 network that modulates macrophage polarization.

Figure 1

The CSF-1R pTyr721-regulated macrophage transcriptome. (A) Schematic depiction of the CSF-1Rs in the macrophage cell lines used to assess the necessity (loss of function, upper panels) and sufficiency (gain of function, lower panels) of pTyr-721 signaling in macrophages. The CSF-1R domains (extracellular [ECD]; transmembrane [TM], and tyrosine kinase [TK]) and the intracellular CSF-1R Tyr (Y, black) residues mutated to Phe (F, red) are indicated. (B) Pairwise comparisons of gene expression data for all 28 361 transcripts detected on microarrays, showing high Pearson correlation coefficients between biological replicates and the expected hierarchical clustering of samples, as a function of genotype and CSF-1 treatment. Results for M−/−.WT and M−/−.Y721F (upper panel) and M−/−.3ABY721 and M−/−.3AB (lower panel) are shown as heat maps illustrating the significant differences in gene expression between Tyr-721–expressing and Tyr-721–nonexpressing cells, each with an associated color key and histogram (−, without CSF-1; +, with 120 ng/mL of CSF-1; biological replicates are indicated by the numerals 1 and 2). (C-E) Venn diagrams, representing the size of the pTyr-721–regulated transcriptome, identified by each approach. To obtain the pTyr-721–regulated genes, the CSF-1–regulated transcripts expressed in either M−/−.WT or M−/−.3ABY721 cells were compared with those of the corresponding Y721F mutant cells. (C) The Y721F mutation (1904 CSF-1–regulated mRNAs) reduced the number of CSF-1–regulated mRNAs of M−/−.WT macrophages (2174) by 270 transcripts. However, of the CSF-1–regulated genes in both cell types, 1713 (∼9% of 19 000 total mouse genes) were differentially regulated in both data sets. (D) The addition of CSF1R Y721 to the 3AB background increased the number of CSF-1–regulated transcripts by 250 (from 870 to 1120). Of the CSF-1–regulated genes in both cell types, 244 (∼1.3% of 19 000 total mouse genes) were differentially regulated in both data sets. (E) Only 34 genes are common between the 2 datasets.

Figure 2

CSF-1R pTyr-721 controls expression of a specific subset of IEGs and DEGs associated with suppression of inflammation and promotion of an M2 phenotype. (A) Bar charts representing the relative abundance of a subset of pTyr-721–induced IEGs (left panel) and DEGs (center and right panels), illustrating pTyr-721–mediated decreased abundance of transcripts encoding proinflammatory (M1, *) molecules in conjunction with increased abundance of anti-inflammatory (M2, **) molecules. Known IEGs or DEGs are shown in bold. CSF-1–regulated transcript abundance in M−/−.WT cells (ie, log₂ fold change [FC] between the expression of the indicated RNA at the indicated time point and its measured expression at time 0) was normalized to CSF-1–regulated mRNA abundance in M−/−.Y721F macrophages calculated in the same manner. For a full set of pTyr-721–regulated genes, see supplemental Table 3. (B) Relative abundance of a subset of pTyr-721–induced genes in cells growing in CSF-1 (see supplemental Table 1) that encode secreted cytokines, illustrating the pTyr-721–mediated change in abundance of transcripts encoding M1 (*) and M2 (**) molecules (P < .05, FC > 1.5). (C-D) pTyr-721 suppresses secretion of proinflammatory molecules (*) and induces secretion of tumorigenic EGF, insulin-like growth factor-1 (IGF-1), and matrix metalloproteinase-3 (MMP-3). Conditioned medium from CSF-1–treated (120 ng/mL) and CSF-1–starved M−/−.WT and M−/−.Y721F macrophages was assayed for production of proinflammatory and proangiogenic factors using a mouse cytokine array permitting simultaneous measurements of 144 cytokines and soluble factors. For each cell line, signals detected for each cytokine produced by CSF-1–treated cells were integrated and normalized to the corresponding signal obtained from CSF-1–starved cells (duplicates ± range). (C) Molecules exhibiting pTyr-721–dependent increased secretion. (D) Molecules exhibiting increased secretion in the absence of pTyr-721 signaling. Error bars indicate the range of levels of biological duplicates.

Figure 3

CSF-1R pTyr-721 signaling suppresses M1 responses and enhances M2 responses. (A) qRT-PCR results showing the enhancement by pTyr-721 signaling of the IFN-γ and LPS stimulation of iNOS, TNF-α, and IL-6 gene transcripts in cells stimulated with CSF-1. Cells were either starved for CSF-1 overnight, or constitutively grown in CSF-1 (120 ng/mL), then treated with either IFN-γ and LPS (+) or vehicle (−) for 18 hours. HPRT gene expression levels were used as endogenous control. mRNA levels are expressed as the log₂ fold-change relative to CSF-1–starved cells. (B) Nitrate/nitrite levels in cell lysates from equivalent cultures. (C) Relative CSF-1–induced surface expression of the M1 polarization marker MHC II on CD11b⁺ cells, determined by flow cytometry in CSF-1–treated (18 hours, +CSF-1) vs CSF-1–starved (−CSF-1) cells and expressed as a ratio of the +CSF-1 to −CSF-1 responses. (D) qRT-PCR results showing the enhancement by pTyr-721 signaling of the IL-4–stimulated expression of arginase-1, IL-4Rα, and mannose receptor-1 (MRC1) transcripts. Experiments were carried out as described in panel A. (E) Total arginase 1 activity of supernatants of equivalent cultures. (F) Stacked bar chart showing relative contribution of combined sets of macrophage M2 (CD206 and IL-4Rα) and M1 (MHC II) cell surface–specific polarization markers in CSF-1–treated cells. The relative CSF-1 response ratio for each set of cell surface markers was calculated as in panel C and expressed as a percentage for each cell line. Data are representative of 3 independent experiments (error bars, ±standard deviation (SD); *P ≤ .05).

Figure 4

miR-21 expression is positively regulated downstream of CSF-1R pTyr-721 to decrease the abundance of proinflammatory molecules. (A) Downstream effector analysis performed by IPA on differentially regulated CSF-1 transcripts (FC > 1.5, P < .05), predicts miR-21 as the top statistically significant miRNA (supplemental Table 5) that is activated in a pTyr-721–dependent manner (z score >2, gray dotted line). The number of coexpressed mRNAs predicted as miR-21 targets (bold) and the statistical P values (italics) are indicated for each cell line at the top of each bar. (B) Validation of miR-21 as a pTyr-721–regulated molecule and of a miR-21 inhibitor. For miR-21 validation, qRT-PCR measurements were performed on complementary DNA prepared from M−/−.WT and M−/−.Y721F macrophages and from the M−/−.3ABY721 and M−/−.3AB cells that were either CSF-1–starved overnight, then treated with CSF-1 (120 ng/mL) for the indicated time points, or constitutively grown in CSF-1 (C). For inhibitor validation, cells were constitutively grown in CSF-1. The miR-21 inhibitor (C+I), or mismatch inhibitor control (C+M), were added 48 hours before harvesting the cells and determining the miR-21 levels by qRT-PCR. (C) Cytoscape representation of the IPA-predicted miR-21 targets that are regulated in a pTyr-721–dependent manner. Note that 80% of these molecules are associated with macrophage polarization toward an inflammatory phenotype (red circles), whereas only 10% are associated with the M2 phenotype (blue circles). (D-E) CSF-1 negatively regulates IL-1β mRNA levels and IL-1β secretion in a pTyr-721– and miR-21–dependent manner. (D) qRT-PCR measurements of IL-1β mRNA in Tyr-721– and Tyr-721F–expressing macrophages. Cells were treated as described in panel B. (E) Conditioned media from Tyr-721– and Tyr-721F–expressing macrophages treated as described in panel B were used to measure the amount of soluble mature IL-1β released in the medium, by enzyme-linked immunosorbent assay. (B,D,E) Key for cell lines as in panel A. Five biological replicates; error bars, ±SD; *P ≤ .05.

Figure 5

CSF-1R pTyr-721/PI3K signaling regulates the amplitude and duration of ERK_1/2 and NF-κB p65 activation and induces miR-21 expression. CSF-1–starved macrophages were stimulated with CSF-1 (120 ng/mL) for the indicated times and processed for western blotting (WB) or RNA extraction. (A) M−/−.WT and M−/−.Y721F macrophages and (B) M−/−.3ABY721 and M−/−.3AB macrophages were subjected to WB analysis with antibodies to the activated phosphorylated forms of ERK_1/2 and NF-κB p65 and to the total ERK_1/2 and NF-κB p65. (C) Relative qRT-PCR quantitation of miR-21 levels in M−/−.WT macrophages treated with PI3K inhibitor LY 2940002 alone (100 μM), in combination with either the ERK_1/2 inhibitor PD98059 (50 μM) or the IKK inhibitor PS1145 (2 μM), or vehicle alone (1% dimethylsulfoxide). (D) miR-155 levels in M−/−.WT macrophages. Experiments performed as in panel A. Relative expression values in panels C-D indicate the fold-change of miRNA levels in CSF-1–treated cells relative to CSF-1–starved cells at the indicated times. Data are representative of 3 independent experiments (error bars, ±SD; *P ≤ .05). (E) M−/−.WT and M−/−.3ABY721 macrophages, treated for 48 hours with LNA-miR-21 inhibitor (I), or (F), an inhibitor mismatch control (M), were CSF-1–starved before CSF-1 stimulation for the indicated times and processed as described in panels A-B. Comparisons between matching cell lines (panels A-B; top 2 panels and bottom 2 panels of panels E-F) were made on the same blots (loading control, β-actin). (G) Schematic representation of pTyr-721–mediated signaling events leading to induction of miR-21 and suppression of inflammatory networks. Filled lines, relationships demonstrated in the present study; dashed lines, suggested interactions from the literature; arrows, activation; blunt arrows, inhibition; point arrows, kinetically regulated activation and inhibition. Ovals indicate molecule activity, not expression level. SIRPb1 and IL-1β are directly suppressed by miR-21 (supplemental Table 8), but are not necessarily the exclusive mediators of macrophage miR-21 effects on ERK_1/2 and NF-κB. ERK_1/2 activation is positively regulated through CSF-1R tyrosines 544, 559, and 807.

Figure 6

miR-21 suppresses the peritoneal monocyte/macrophage inflammatory response to LPS. Mice were intraperitoneally injected with endotoxin-free LNA-miR-21 inhibitor (LNA-miR-21-I) or mismatch inhibitor control (MMC) at days 1 and 3, followed by intraperitoneal injections of LPS or PBS at days 3 and 4. Peritoneal lavages were collected on day 5 at 18 or 48 hours post-LPS. Total viable CD11b⁺SSCA^loLy6C⁺Ly6G⁻ monocytes (A), CD11b⁺SSCA^loLy6C⁻Ly6G^lo/−F4-80⁺ macrophages (B), SPMs and large peritoneal macrophages (LPMs), (C) and the percent expression levels of the cell surface marker MHC II on the CD11b⁺SSCA^loLy6C⁻Ly6G^lo/-F4-80⁺ macrophages (D) were determined by flow cytometry. Data are representative of 2 independent experiments, 3 to 4 mice per group (error bars, ±SD; *P ≤ .05).