Changes in macrophage phenotype in injured muscle profoundly influence regeneration. In particular, the shift of macrophages from a proinflammatory (M1 biased) phenotype to a proregenerative (M2 biased) phenotype characterized by expression of CD206 and CD163 is essential for normal repair. According to the current canonical mechanism regulating for M1/M2 phenotype transition, signaling through PPARδ is necessary for obtaining the M2-biased phenotype. Our findings confirm that the murine myeloid cell–targeted deletion of Ppard reduces expression in vitro of genes that are activated in M2-biased macrophages; however, the mutation in mice in vivo increased numbers of CD206+ M2-biased macrophages and did not reduce the expression of phenotypic markers of M2-biased macrophages in regenerating muscle. Nevertheless, the mutation impaired CCL2-mediated chemotaxis of macrophages and slowed revascularization of injured muscle. In contrast, null mutation of IL-10 diminished M2-biased macrophages but produced no defects in muscle revascularization. Our results provide two significant findings. First, they illustrate that mechanisms that regulate macrophage phenotype transitions in vitro are not always predictive of mechanisms that are most important in vivo. Second, they show that mechanisms that regulate macrophage phenotype transitions differ in different in vivo environments.
The inflammatory response to skeletal muscle injury or disease has important influences on the rate and extent of muscle repair. Although the initial population of myeloid cells that enter muscle following injury is dominated by neutrophils that can amplify muscle damage, their numbers rapidly decline, and they are replaced by phenotypically complex populations of macrophages that can remain at elevated concentrations in the regenerating muscle for weeks (1–3). Over that prolonged period of muscle regeneration, the predominant macrophage phenotypes shift from a population that expresses proinflammatory cytokines and increases proliferation of myogenic cells to a population that produces anti-inflammatory cytokines, promotes muscle fiber growth, and increases connective tissue production (4–9). The importance of the shifts in macrophage phenotypes on the course of muscle repair and regeneration has been well established by investigations that have shown that disrupting the normal recruitment or phenotype specification of macrophages amplifies muscle damage and slows regeneration (10–15).
The initial characterization of macrophage phenotypes as M1 (proinflammatory) or M2 (anti-inflammatory) was based upon changes in gene expression that occurred in macrophages in vitro following stimulation with Th1 cytokines (IFN-γ and TNF) or Th2 cytokines (IL-4 and IL-13) (16–18). However, in the complex inflammatory environment in vivo, numerous other immunomodulatory molecules can influence macrophage activation to produce a broad and continuous spectrum of phenotypes that range between the M1 and M2 designations (19). Whether macrophage activation is biased along the spectrum toward the M1 or M2 phenotype is assessed by determining their expression of numerous phenotypic markers. Within injured and diseased muscles of mice, macrophages that are biased toward the M1 phenotype express inducible NO synthase, CD68, and high levels of CD11b and Ly-6C. The intramuscular macrophages biased toward the M2 phenotype express CD163, CD206, arginase (Arg) 1, and low levels of CD11b and Ly-6C (8, 10, 20–22).
Multiple signaling pathways contribute to activating macrophages to an M2-biased, proregenerative phenotype. For example, macrophage activation to a CD163+/CD206+, M2-biased phenotype in inflamed muscle is strongly influenced by IL-10, TGF-β, and insulin-like growth factor-1 (IGF1) that are expressed by macrophages at the site of muscle damage (7, 8, 21, 23). Furthermore, diminishing expression of these regulators of M2-biased macrophage phenotype in injured or diseased muscle causes impaired muscle repair and retarded muscle growth (8, 21, 23), which emphasizes the functional importance of intramuscular macrophages biased to an M2-biased phenotype. However, recent discoveries also demonstrated that changes in the relative prevalence of lipid mediators that influence macrophage phenotype change over the course of muscle repair following injury (24). For example, lipid mediators that reduce inflammation (specialized, proresolving lipid mediators [SPMs]) are more prevalent in Ly-6Clo macrophages, and intramuscular injection of one of the SPMs, resolvin D2, produced increases in Ly-6Clo cell numbers in injured muscles and improved muscle function (24). Those findings suggest that SPM induction of an M2-biased phenotype may also be able to improve muscle regeneration.
Although untested, the collective findings of numerous investigations support the hypothesis that lipid mediators may influence the shift of macrophages to an M2-biased phenotype in injured muscle by activating peroxisome proliferator-activated receptor-δ (PPARδ). PPARδ is a ligand-activated nuclear receptor that is activated by polyunsaturated fatty acids and other lipid mediators that can be produced during breakdown of apoptotic cells that are engulfed by macrophages (25). Activated PPARδ then dimerizes with retinoic acid receptors (RXR), translocates to the nucleus, and activates expression of M2-associated transcripts, including Mrc1 (which encodes CD206), Arg1, Chi3l3, Retnla, and Ppard (26, 27). The importance of PPARδ in activating macrophages to an M2-biased phenotype was demonstrated clearly in Kupffer cells in the liver and in adipose tissue macrophages (28, 29) in which the deletion of Ppard in LysM-expressing cells or in bone marrow–derived cells prevented activation of macrophages to an M2-biased phenotype. However, fatty acid ligation of PPARδ in macrophages can also cause an increase in IL-10 expression that requires PPARδ (25), and induction of IL-10 expression is increased by PPARδ in endothelial cells (30). Those findings show that IL-10–mediated signaling and PPARδ-mediated signaling pathways can interact, suggesting the possibility that IL-10 and lipid mediators may affect intramuscular macrophage phenotype through overlapping mechanisms.
In this investigation, we generated a mouse line in which there is a targeted deletion of Ppard in LysM-expressing myeloid cells to test the hypothesis that activation of macrophages to an M2-biased phenotype in acutely injured muscle is mediated by PPARδ. We also assayed whether Ppard mutation in macrophages influences macrophage function or affects muscle repair, growth, and revascularization following injury. Finally, we assayed whether the same parameters of macrophage activation and muscle repair are similarly influenced by ablation of Il10 expression to address the question of whether IL-10– mediated and PPARδ-mediated regulation of macrophage phenotype have similar effects on muscle inflammation, repair, and regeneration, which would indicate a potential interaction between the two regulatory pathways in acutely injured muscle. The direct comparison of the effects of the two mutations in the same injury model enabled us to demonstrate the distinct roles of myeloid cell PPARδ and IL-10 in regulating macrophage recruitment, phenotype, and regeneration following acute muscle injury.
Materials and Methods
Ethical approval and animals
All animals were handled according to guidelines provided by the Chancellor’s Animal Research Committee at the University of California, Los Angeles. C57BL/6, Ppardflox/flox (no. 005897), LysM-cre (no. 004781), and Il10−/− (no. 002250) mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and bred in specific pathogen-free vivaria at the University of California. Ppardfl/fl and LysM-cre mice were crossed to generate myeloid cell–specific Pparδ mutant mice. Knockdown of Pparδ was confirmed by quantitative real-time PCR (QPCR) assays on bone marrow–derived macrophages (BMDMs) from Ppardfl/fl LysM-cre (Pparδ mutants) and Ppardfl/fl (Ppard wild-type [WT]) mice. Il10 mutation was confirmed by QPCR of skeletal muscle from C57BL6 and Il10−/− mice. Mice were housed on a 12/12-h light/dark cycle and were provided food and water ad libitum. Following euthanasia by inhalation of isoflurane, muscles were collected, weighed, and flash-frozen for histological or biochemical analyses.
Acute muscle injury
Mice were anesthetized with isoflurane inhalation in a chamber (4–5% isoflurane with 100% oxygen) then moved to a nose cone (1–2% isoflurane). Anesthesia was checked by testing mice for a positive reflex response to a hind foot pinch and by monitoring respiration. The lower limb was wiped with 70% ethanol before intramuscular injection. Sterile muscle injury was induced by the intramuscular injection of 50 μl of a 1.2% barium chloride (BaCl2) solution into the midbelly of tibialis anterior (TA) muscles of healthy 4–6-mo-old male mice. Animals were monitored daily until they recovered.
RNA isolation and QPCR
RNA was isolated from cells and muscle homogenates and electrophoresed on agarose gels, and its quality assessed by 28S and 18S rRNA integrity (31). RNA samples (2 μg) were reverse transcribed with Super Script Reverse Transcriptase II using oligo(dT)s to prime extension (Invitrogen, Carlsbad, CA) to produce cDNA. Expression of selected transcripts was assayed using iTaq SYBR Green qPCR Supermix according to the manufacturer’s protocol (Bio-Rad Laboratories, Hercules, CA) and a QuantStudio 5 Real-Time PCR System (Applied Biosystems, Foster City, CA). Established guidelines for experimental design, data normalization, and data analysis for QPCR were used to maximize the rigor of quantifying the relative levels of mRNA (32, 33). We empirically tested reference genes and identified those with the least variability between experimental groups. Based on that analysis, the relative expression of transcripts of interest were normalized to the reference genes Srp14 and Rnps1 for muscle regeneration, Srp14 and Tpt1 for Pparδ BMDMs, and Tbp and Rplo for Il10−/− BMDM QPCR assays. The normalization factor for each sample was calculated by geometric averaging of the cycle threshold values of reference genes. Expression for each gene in control samples was set to 1, and the other values were scaled to the control. Primers used for QPCR are listed in Table I.
Muscle fiber cross-sectional area
Frozen cross-sections of TA muscles were sectioned at the midbelly and used for fiber cross-sectional area measurements. Sections were stained with hematoxylin, and the cross-sectional areas of fibers were measured using a digital imaging system (BioQuant, Nashville, TN). In muscle that had undergone acute injury, only centrally nucleated, regenerating muscle fibers (34) were measured from the area of the central lesion, identified as the region that was least regenerated, as previously described (35). The classification of small and large fibers was determined by setting three SD from the mean cross-sectional area for the Pparδ WT group at 15 d postinjury (dpi) time point and quantifying the proportion of myofibers measured that fell within these ranges (adapted from Ref. 36). The threshold for small and large fibers at 15 dpi was determined to be <719 μm2 or >2304 μm2, respectively.
Production of Pax7 Ab
Pax7 hybridoma cells were purchased from Developmental Studies Hybridoma Bank anti-Pax7 (Research Resource Identifier [RRID]: 2299243; Developmental Studies Hybridoma, Iowa City, IA). Cells were cultured, and Ab was isolated from the supernatant as previously described (37).
Muscles were dissected from euthanized mice and then rapidly frozen in liquid nitrogen–cooled isopentane. For identification of myogenic progenitor cells, cross-sections 10-μm thick were taken from the midbelly of muscles, air dried for 30 min, and then fixed in 4% paraformaldehyde for 10 min. Sections were then immersed in antigen retrieval buffer (10 mM sodium citrate and 0.05% Tween 20 [pH 6]) at 95–100°C for 40 min. Endogenous peroxidase activity in the tissue was quenched by immersion in 0.3% H2O2. Sections were then treated with blocking buffer from a Mouse on Mouse Immunohistochemistry Kit (Vector Laboratories, Burlingame, CA) for 1 h and immunolabeled with affinity purified mouse anti-Pax7 (1:500) Ab overnight at 4°C. Sections were washed with PBS solution and then incubated with biotin-conjugated anti-mouse IgG (1:250) for 30 min. Sections were subsequently washed with PBS and then incubated for 30 min with ABC reagents from the Mouse on Mouse Immunohistochemistry Kit. Staining was visualized with the peroxidase substrate 3-amino-9-ethylcarbazole (Vector Laboratories), yielding a red reaction product.
For identification of macrophages and endothelial cells, cross-sections 10-μm thick were taken from the midbelly of muscles and fixed in ice-cold acetone. Endogenous peroxidase activity in the sections was quenched by immersion in 0.3% H2O2. Sections were blocked for 1 h with blocking buffer (3% BSA, 2% gelatin, and 0.05% Tween 20 in 50 mM Tris-HCl [pH 7.6] containing 150 mM NaCl). Sections were then incubated with rat anti-mouse F4/80 (RRID: AB2314387, 1:100, overnight at 4°C, no. 14-4801; eBioscience, San Diego, CA), rat anti-mouse CD68 (RRID: AB322219, 1:100, 3 h at room temperature [RT], no. MCA1957; Bio-Rad Laboratories), rabbit anti-mouse CD163 (RRID: AB2074556, 1:100, 3 h at RT, no. 33560; Santa Cruz Biotechnology, Dallas, TX), rat anti-mouse CD206 (RRID: AB324622, 1:50, 3 h at RT, no. MCA2235; Bio-Rad Laboratories), and rat anti-CD31 (no. 01951D; BD Pharmingen, San Diego, CA). The sections were washed with PBS and probed with rabbit anti-rat IgG and goat anti-rabbit IgG (no. BA-4001 and no. BA-1000; Vector Laboratories). Sections were then washed with PBS and incubated with avidin D–conjugated HRP (1:1000, 30 min at RT; Vector Laboratories). Staining was visualized with 3-amino-9-ethylcarbazole.
The number of cells per volume of muscle was determined by measuring the total volume of each section using a stereological, point-counting technique to determine section area and then multiplying that value by the section thickness (10 μm) (38). The numbers of immunolabeled cells in each section were counted and expressed as the number of cells per unit volume of each section. In inflammatory lesions with high concentrations of macrophages, boundaries of individual cells were identified by using Nomarski optics (resolution ∼400 nm) and changing plane of focus in the z-axis.
Assay for ischemic muscle area
Ischemic areas of muscles were quantified by imaging the entire muscle section and merging the images using Image Composite Editor (Microsoft, Redmond, CA). BioQuant digital imaging software was used to calculate the entire area of the muscle followed by measuring and subtracting areas of the tissue containing artifacts or imperfections (tears or folds). Areas of ischemia that were sparsely occupied by CD31+ endothelial cells were quantified by an investigator blinded to the identity of tissue sections. Data were expressed as the proportion of the ischemic area divided by the total muscle area.
The immunofluorescence double-labeling protocol was similar to macrophage immunohistochemistry described above. Sections from the midbelly of TA muscles were fixed with 4% PFA for 10 min and then blocked with PBS containing 10% horse serum and 0.1% Tween 20 for 60 min. Muscle sections were incubated with rat anti-mouse F4/80 (1:100) and goat anti-Ki67 (RRID: AB2142374, 1:200; no. sc-7846; Santa Cruz Biotechnology) or rabbit anti-CCR2 (RRID: AB1603737, 1:50; no. 32144; Abcam, Cambridge, MA) overnight at 4°C. Sections were subsequently incubated with a combination of donkey anti-rat IgG Alexa Fluor 488 (no. ab102260; Abcam) and horse anti-goat IgG DyLight 594 (no. Dl-3094; Vector Laboratories) or horse anti-rabbit IgG DyLight 594 (no. Dl-1094; Vector Laboratories) and coverslipped with Prolong Gold Antifade reagent with DAPI (Invitrogen).
Assay for muscle fiber damage
Damaged muscle fibers were identified in sections that were blocked in a 1% gelatin solution in PBS for 30 min and then labeled with horse anti-mouse IgG DyLight 488 (1:100; no. Dl-2488; Vector Laboratories) and coverslipped with Fluoro-Gel (Electron Microscopy Sciences, Hatfield, PA). Intracellular labeling of muscle fibers denoted the presence of the extracellular mouse IgG protein in the muscle fiber, indicating damage to the muscle fiber membrane. The proportion of a muscle cross-sections that contained damaged fibers was determined by using a sampling grid that was superimposed over the tissue, after which the number of grid intercepts that overlaid IgG-positive fibers was counted and expressed as a percentage of the total number of intercepts (39).
Arg activity assay
Arg activity was measured in TA muscles at 7 dpi and in contralateral, noninjured TA muscles that were homogenized in five volumes of 25 mM Tris-HCl (pH 7.4) containing 5 mM MnCl2, 0.2% Triton X-100, and protease inhibitor mixture (Sigma-Aldrich, St. Louis, MO). Lysates were centrifuged, and the supernatant was heated to 56°C for 10 min to activate Arg. Substrate hydrolysis was performed by adding 0.5 M arginine (pH 9.7) to the tissue supernatant, followed by 1-h incubation at 37°C. The reaction was stopped by adding H2SO4 and H3PO4 in water (1:3:6 parts v/s). Samples were heated to 100°C for 45 min after adding 9% α-isonitrosopropiophenone. Urea content was then measured spectrophotometrically at 540 nm. Values were normalized to protein content of the tissue lysate. Six tissues were assayed at each per condition (adapted from Ref. 40).
Preparation of BMDMs
For preparation of BMDMs, bone marrow cells were aseptically flushed from femurs and tibiae and differentiated using RPMI 1640, 20% heat-inactivated FBS (Omega Scientific, Tarzana, CA), 100 U/ml penicillin and 100 μg/ml streptomycin, and 10 ng/ml M-CSF (no. RP2008; Cell Applications, San Diego, CA) in vitro to BMDMs (31). BMDMs were stimulated for 24 h with activation media consisting of DMEM with 0.25% heat-inactivated FBS, 100 U/ml penicillin and 100 μg/ml streptomycin, 10 ng/ml M-CSF and rIL-4 (25 ng/ml) with IL-13 (10 ng/ml) (no. 550067 and no. 554599; BD Pharmingen), and rTNF-α (10 ng/ml) with IFN-γ (10 ng/ml) (no. 554589 and no. 554587; BD Pharmingen) or rIL-10 (10 ng/ml) (no. 550070; BD Pharmingen).
Macrophage mobility and chemotaxis
Cell migration and chemotaxis were evaluated using a modified Boyden chemotaxis chamber according to the manufacturer’s protocol (Neuro Probe, Gaithersburg, MD). Briefly, the lower chambers were filled with a volume of DMEM with 1% BSA or DMEM with 1% BSA and 100 ng/ml rCCL2 (no. 479-JE-010; R&D Systems, Minneapolis, MN), resulting in a slight positive meniscus over the well. Next, a 5-μm porous membrane, silicone gasket, and the upper plastic chambers were placed over top of the lower chambers. BMDMs were prepared as described above. Cells were dissociated from the plate and resuspended at a concentration of 2 × 106 cells/ml, after which 1 × 105 cells were loaded into the upper chambers. The filled chambers were incubated for 2 h in an incubator at 37°C in humidified air with 5% CO2. After 2 h, the remaining cells were aspirated from the upper chamber, the chamber was disassembled, and the cells on the nonmigrated side of the membrane were cleared. The membrane was stained with hematoxylin and mounted to a slide, and the number of cells that migrated across the membrane were quantified.
Transcriptional regulation of macrophage activation by myeloid PPARδ in vitro
We generated myeloid-specific Ppard mutant mice by crossing Ppardfl/fl mice (Ppard WT) with lysozyme Cre mice to assess the regulatory roles of PPARδ in muscle inflammation and repair following acute injury. First, we assayed the influence of Ppard mutation on myeloid cell phenotype and function in vitro by assaying changes in gene expression in BMDMs. QPCR analysis confirmed >90% knockdown of Ppard expression in Ppard mutant macrophages (Fig. 1A; primers listed in Table I). The mutation also resulted in a 93% increase in the expression of the M1-biased transcript Nos2 while reducing the expression of M2-biased transcripts Arg1, Mrc1, and Chil3 by 61, 30, and 31%, respectively (Fig. 1B). These findings generally agree with previous investigations showing that PPARδ signaling promotes the M2 phenotype in macrophages (28, 29).
|Gene .||Forward .||Reverse .|
|Gene .||Forward .||Reverse .|
Because the activation state of macrophages in vitro along the M1/M2 spectrum of phenotypes is largely determined by the presence of Th1 or Th2 cytokines (17, 19), we assayed the influence of Ppard mutation on changes in gene expression that were induced by cytokines that may be modulated in muscle following acute injury or disease. Stimulation of WT BMDMs with the Th2 cytokines IL-4 and IL-13 showed the anticipated increases in expression of M2-biased genes (Arg1, Arg2, Mrc1, Retnla, and Chil3), and Ppard mutation diminished the increase for each of those transcripts, except for Arg2 (Fig. 1C). These observations are consistent with the previously reported, Ppard-dependent increase in the expression of Mrc1, Retnla, and Chil3 in BMDMs stimulated with IL-4 only (29). Stimulation of WT BMDMs with the Th1 cytokines TNF and IFN-γ produced expected increases in some M1-biased genes (Nos2 and Tnf) while reducing expression of some M2-biased genes (Mrc1 and Chil3). Unexpectedly, TNF/IFN-γ stimulation of WT BMDMs also increased expression of Arg1 and Arg2. Ppard mutation also caused a small increase in Nos2 expression in TNF/IFN-γ–stimulated BMDMs and diminished the increased expression of Arg1 that was caused by stimulation with the Th1 cytokines (Fig. 1D). Those observations show that Ppard mutation somewhat amplifies the M1-biased phenotype in TNF/IFN-γ–activated cells. In addition, the findings show that the influence of Ppard mutation on the expression of some genes, such as the increase in Nos2 and reduction in Arg1, occurs regardless of the Th1 versus Th2 cytokine environment. However, reductions in Mrc1 and Chil3 expression that were caused by Ppard mutation in unstimulated BMDMs or IL-4/13-stimulated BMDMs did not occur in TNF/IFN-γ–stimulated BMDMs. This suggests that the regulatory role of PPARδ signaling may vary with the stage of inflammation in injured tissue as the cytokine environment changes.
Modulation of Th1 and Th2 cytokine gene expression following acute muscle injury
Because in vitro observations showed that the response of BMDMs to stimulation with either Th1 or Th2 cytokines in vitro was influenced by changes in Ppard expression, we assayed whether the expression of those cytokines changed over the time course of muscle repair following acute injury caused by BaCl2 injection. In addition, IFN-γ, TNF, and IL-4 have all been shown previously to influence the muscle inflammation and repair following acute injury or disease (5, 8, 41, 42) Although changes in the expression of IL-4 and IFN-γ have been observed following muscle injury by toxin (5, 41), the inflammatory responses to muscle injury by cardiotoxin and BaCl2 differ (43). QPCR data showed that Tnf expression was greatly elevated at 3 dpi, remained elevated at 7 dpi, and then declined to basal levels by 15 dpi (Fig. 2A), which was consistent with a previous report that BaCl2-induced injury increases TNF expression in muscle (44). However, we also observed that Ifng, Il4, and Il13 expression were elevated at 7 dpi, with Il4 and Il13 returning to levels that did not differ from noninjured muscle by 15 dpi (Fig. 2B–D). Notably, the increase in Il4 and Il13 expression at 7 dpi coincided to the time postinjury at which CD206+ macrophage populations are elevated (23, 45, 46). Because IL-4 and IL-13 can increase expression of Mrc1 (which encodes CD206) through a PPARδ-mediated mechanism (29), the observations support the expectation that elevations in IL-4 and IL-13 expression in muscle at 3–7 dpi could lead to increased activation of macrophages to the CD206+ M2-biased phenotype through a PPARδ-mediated process.
Myeloid PPARδ affects resident macrophage phenotype in healthy muscle without affecting muscle mass
We tested whether Ppard mutation in myeloid cells affected their numbers or phenotype in healthy muscle in vivo. Using Abs to the pan-macrophage marker F4/80, we found that deletion of Ppard in myeloid cells did not affect total numbers of macrophages in healthy adult muscle (Fig. 3A, 3B, 3I). However, the mutation yielded significant reductions of CD163+ (38%; Fig. 3E, 3F, 3I) and CD206+ (55%; Fig. 3G–I) M2-biased macrophages, and we observed a strong trend (p = 0.06) for an increase in CD68+ macrophages per volume of muscle (Fig. 3C, 3D, 3I). The reduction in CD206+ macrophages in Ppard mutant mice is consistent with our in vitro observations showing that the mutation reduced expression of Mrc1 (Fig. 1). Despite the established role of M2-biased macrophages in promoting muscle growth following injury (10, 14, 22), we did not observe any influence of the reduction of intramuscular CD206+ M2-biased macrophages on the mass in healthy adult TA muscles (Fig. 3J–L).
The influence of myeloid cell deletion of Ppard differs between healthy and injured muscles
We tested whether PPARδ served similar regulatory roles for regulating macrophage phenotype in healthy and injured muscle by assaying numbers of macrophages in muscle that expressed CD68, CD163, or CD206 over the course of muscle inflammation following injury. Contrary to our observations on resident macrophages in healthy muscle, numbers of CD68+ macrophages were significantly reduced at 3 dpi, returning to levels present in nonmutant mice at 7 dpi (Fig. 4A–C). Also contrary to the effect of the mutation in healthy muscle, numbers of CD163+ macrophages were not affected at any stage of inflammation following muscle injury (Fig. 4D–F). However, in most distinct contrast to healthy muscle in which the mutation caused large reductions in CD206+ M2-biased macrophages, numbers of CD206+ cells were significantly elevated in mutant mice TAs at 7 dpi (Fig. 4G–I).
We also observed that the influence of Ppard deletion on the expression of transcripts related to the M2-biased phenotype differed greatly in macrophages in vitro compared with myeloid cells in injured muscles. For example, Ppard mutation in macrophages in vitro reduced Arg1 expression, regardless of the cytokine environment (Fig. 1), but the mutation increased Arg1 expression in inflamed muscle at 7 and 15 dpi (Fig. 4J), which was reflected by elevated Arg activity in the injured muscle (Fig. 4K). Similarly, Retnla, a stereotypical indicator of the M2-biased phenotype, was significantly elevated at 15 dpi (Fig. 4L). Thus, ablation of Ppard in macrophages within acutely injured muscle increased their expression of transcripts associated with activation of the M2-biased phenotype.
Ppard mutation in myeloid cells impairs macrophage accumulation in injured muscle and reduces chemotaxis
We assayed whether Ppard mutation in myeloid cells affected the kinetics of macrophage accumulation in injured muscle by assaying changes in total macrophage numbers, using Abs to the pan-macrophage marker F4/80. The concentration of F4/80+ cells was substantially reduced in Ppard mutant mice at 3 dpi (Fig. 5A–C). However, the number of F4/80+ cells at 7 dpi was significantly greater in injured muscles of mutant mice than in WT muscles, and a strong trend (p = 0.08) for sustained elevations of F4/80+ cells in muscle persisted through 15 dpi (Fig. 5C).
Because previous investigators have shown that Ppard-deficient macrophages have a markedly reduced mitogenic response to IL-4 stimulation in vitro (29), we tested whether the delayed accumulation of macrophages in injured muscles of Ppard mutants was attributable to impaired macrophage proliferation in vivo. However, the proportion of F4/80+ cells that coexpressed Ki67, a marker of cell proliferation, did not differ between injured muscles of control and mutant mice (Fig. 5D, 5E), indicating that suppression of proliferation did not underlie the slowed kinetics of macrophage accumulation in injured muscles.
Alternatively, the delayed accumulation of macrophages in injured muscles of mutant mice could result from impaired recruitment because several previous investigations have shown that perturbations of CCL2/CCR2 signaling are particularly important in disrupting macrophage recruitment to injured muscle (6, 47–49). We observed that the proportion of F4/80+ cells that expressed detectible levels of CCR2 was significantly reduced in Ppard mutant muscles at 3 dpi (Fig. 3F–H), suggesting that reductions in recruitment via CCR2 can contribute to the reduced numbers of total macrophages at this stage. In addition, we confirmed that Ppard mutation caused large reductions (41%) in Ccr2 expression by BMDMs in vitro (Fig. 5I), consistent with previous reports (50). Furthermore, Ppard mutant BMDMs exhibited impaired chemotaxis toward CCL2 but did not affect cell mobility independent of the CCR2 ligand (Fig. 5J). Finally, we also tested whether expression of either CCL2 or CCR2 differed in mutant and WT muscles at 7 dpi, when numbers of F4/80+ cells in mutant muscles increased, suggesting a recovery in a defect in CCR2-mediated chemotaxis. Although CCL2 expression did not differ between WT and mutants at 7 dpi (Fig. 5K), CCR2 expression was significantly elevated and greater in mutant muscles at this stage (Fig. 5L). Thus, our data indicate that myeloid cell PPARδ is important for influencing macrophage recruitment into muscle after acute injury, possibly by regulating CCR2 expression and reducing chemotaxis at early stages postinjury.
Ppard mutation in myeloid cells slows muscle repair and revascularization following acute injury The intramuscular macrophages can contribute to amplifying muscle damage or promoting muscle repair following injury or disease (3), which led us to examine the influence of Ppard mutation in myeloid cells on muscle injury and repair. We assayed muscle injury by measuring the proportion of total muscle cross-sections that contained injured fibers with large membrane lesions that allowed the unregulated influx of IgG (Fig. 6A, 6B), which does not cross the intact cell membrane of healthy muscle fibers. The proportion of the muscle that contained injured fibers exceeded 73% at 3 dpi and did not differ between WT and mutant mice at this stage (Fig. 6C), indicating that PPARδ signaling is not an essential mediator of macrophage cytotoxicity in this injury model. Furthermore, because the numbers of macrophages in muscles at 3 dpi were greatly reduced (Fig. 5C) although muscle injury did not differ between WT and mutant muscles at that stage (Fig. 6C), the observations show that the magnitude of the inflammatory response is not an important determinant of the extent of muscle damage at this stage in this injury model. In addition, we observed that the regions of the most extensive muscle damage at 3 dpi contained few macrophages, and regions of muscle fiber repair and regeneration were relatively enriched in macrophages (Fig. 6D, 6E). Collectively, these observations show that neither the magnitude of macrophage accumulation, the localization of macrophages, or their expression of PPARδ is an important determinant of the initial magnitude of muscle damage in this injury model. However, at 7 dpi, the area of injury in WT muscles returned almost to levels of noninjured muscle, although injury in muscles of mutant mice remained greatly elevated (Fig. 6C), coinciding with the large elevation of macrophages in the muscle at this stage (Fig. 5C). Repair of injury in muscles of mutant mice was complete by 15 dpi (Fig. 6C), indicating that Ppard mutation slowed the rate of muscle repair.
Previous investigators demonstrated that treatments that reduce macrophage numbers in injured muscles can disrupt revascularization, which was associated with impaired muscle regeneration (6, 48, 51). Using immunohistochemistry for vascular endothelial cells that expressed CD31, we tested whether the myeloid cell targeted deletion of Ppard impaired revascularization. We found that areas of muscle where there was extensive muscle fiber damage contained few CD31+ cells, although areas where regenerative muscle fiber displayed intact membranes were richly vascularized (Fig. 6F, 6G). We tested whether Ppard mutation affected the extent of ischemia in muscle during the repair phase following acute injury by assaying for the proportion of the muscle cross-section that was devoid of CD31+ vascular endothelial cells and found that mutant muscles showed over 4-fold greater area of ischemia than WT at 7 dpi (Fig. 6H–J). In addition, we assayed levels of expression of the hypoxia-inducible gene Hif1a as an independent index of ischemia and oxidative stress and found significantly elevated levels of Hif1α at each stage of muscle repair in Ppard mutants (Fig. 6K).
Ppard mutation in myeloid cells slows myogenesis following acute injury
Macrophage-derived factors can influence the expansion of myogenic satellite cell populations and affect the growth of regenerative muscle fibers following muscle injury or disease (7, 23, 31). We tested whether PPARδ in myeloid cells played a significant role in these promyogenic roles of macrophages by assaying the effect of Ppard mutation on the numbers of Pax7+ myogenic cells and the size of regenerative muscle fibers over the course of muscle repair. Ppard mutants showed large reductions in the number of Pax7+ cells at 3 dpi, which persisted in mutants at 7 dpi (Fig. 7A–C). However, by 15 dpi, the numbers of Pax7+ cells in WT and mutant muscles were greatly reduced and did not differ between the two genotypes (Fig. 7C). The negative influence of the Ppard mutation on myogenesis also manifests as a reduction in the cross-sectional area of centrally nucleated regenerative fibers (Fig. 7D–F), which was 23% less in mutant mice at 15 dpi. In addition, the mutation produced a significant reduction in the proportion of large diameter muscle fibers (>2304 μm2) and a nearly significant increase (p = 0.06) in small diameter fibers (Fig. 7G).
IL-10 is a greater effecter of macrophage activation to a CD163+/CD206+ M2-biased phenotype in injured muscle than PPARδ
Our unexpected finding that Ppard mutation did not diminish macrophage activation to a CD163+/CD206+ M2-biased phenotype in injured muscle led us to assay for other effector molecules that activate the CD163+/CD206+ phenotype in injured muscle. Our previous observation that ablation of Il10 in mice experiencing increased muscle use greatly reduced CD163+ M2-biased macrophages while increasing CD68+ macrophages (21) implicated IL-10 as a significant modulator of intramuscular macrophage phenotype. In the present investigation, we showed that Il10 expression is greatly elevated in muscles following acute injury, reaching expression levels at 3 dpi that are more than 150 times the expression in uninjured muscle (Fig. 8A). We also confirmed that IL-10 is a strong activator of the M2-biased phenotype in BMDMs in vitro, causing large increases in the expression of Arg1, Arg2, and Mrc1, although Cd163 expression was not affected (Fig. 8B).
In contrast to the effects of Ppard gene ablation in myeloid cells, Il10 mutation greatly influenced macrophage phenotype without affecting macrophage accumulation during repair of injured muscle. Assays for changes in F4/80+ total macrophages showed that Il10 mutation did not affect the numbers of macrophages or the time course of their accumulation in injured muscle (Fig. 9A–C), which indicates that expansion and recruitment of macrophage populations were unaffected by the mutation. However, the mutation caused significant increases in the numbers of CD68+ macrophages at 15 dpi (Fig. 9D–F) and caused large reductions in numbers of CD163+ cells at earlier stages of repair (Fig. 9G–I). The most prominent treatment effect that we observed was the greatly reduced numbers of CD206+ macrophages that occurred at every stage of repair (Fig. 9J–L), showing that IL-10–mediated signaling rather than PPARδ signaling is an important regulator of the CD206+ M2-biased phenotype in injured muscle. Preventing expansion of CD206+ macrophages was not accompanied by a reduction in either Arg1 or Arg2 in injured muscles of Il10 mutant mice (Fig. 9M, 9N), in contrast to the tremendous induction of Arg expression in IL-10–stimulated BMDMs (Fig. 8B).
Because our observations of the effects of Ppard mutation on muscle repair showed PPARδ played a significant role in regulating the expansion of satellite cell numbers, reducing the extent of muscle damage and revascularization of muscle, we also assayed whether IL-10 contributed to these features of muscle repair. Unlike the effects of Ppard mutation, Il10 mutation had no effect on the time course or magnitude of expansion of satellite cell populations following injury (Fig. 9O–Q). Also, unlike Ppard mutants in which the initial magnitude of muscle damage was unaffected by the mutation, the Il10 mutants experienced a small decrease in the proportion of the muscle that showed damage at 3 dpi (Fig. 9R–T). However, at 15 dpi, Il10 mutants had more muscle damage than WT mice (Fig. 9T). Finally, in further contrast to Ppard mutants, neither the extent of revascularization of muscle at 7 dpi (Fig. 9U–W) nor the levels of Hif1a expression were affected by Il10 mutation (Fig. 9X).
Landmark discoveries by previous investigators demonstrated the importance of Ppard expression and activation in regulating the phenotype of resident macrophages in liver and adipose tissue in vivo and in BMDMs in vitro (28, 29). In particular, the reduced expression of markers of M2-biased macrophage activation (e.g., Arg1, Mrc1 [or CD206], Retnla, Chi3l3, and Il10) in myeloid-targeted Ppard mutants in these models (28, 29) served as the foundation for the broad view that PPARδ-mediated signaling provides a crucial control for the switch of macrophages from a proinflammatory, M1-biased phenotype to an anti-inflammatory, M2-biased phenotype (52). The present investigation shows that this regulatory function for PPARδ in myeloid cells does not exist in inflamed, regenerating muscle following acute injury. Conversely, we observed that the myeloid cell–specific ablation of Ppard caused either increases or no change in the numbers of CD206+ macrophages over the course of muscle inflammation and caused either an increase or no change in the expression of phenotypic markers of M2-biased macrophages. These findings show that regulation of macrophage phenotype in vivo differs with the tissue environment in which phenotype switching occurs.
Our findings also confirm that regulatory controls of macrophage phenotype that are apparent in vitro are not necessarily predictive of controls that are most important in vivo, as emphasized previously by luminaries in the field of macrophage biology (19). Although our in vivo observations showed that myeloid cell–targeted ablation of Ppard enhanced expression of markers of the M2-biased phenotype in injured muscle, our in vitro observations showed that the same mutation in BMDMs in vitro diminished their expression of M2-biased phenotypic markers. In addition, the mutation diminished the elevated expression of M2-biased phenotypic markers in BMDMs activated with the Th2 cytokines IL-4 and IL-13, as reported previously (29).
The distinct differences between the roles of PPARδ in controlling macrophage phenotype in vivo may reflect differences in the tissue in which activation occurred (i.e., fat or liver versus skeletal muscle); however, our observations indicate that tissue type is not the most important variable for determining the influence of PPARδ on macrophage phenotype. Contrary to our findings in injured muscle in which the myeloid-targeted mutation of Ppard increased CD206+ M2-biased macrophage number, the mutation reduced CD206+ resident macrophages in healthy, noninjured muscle. This effect of Ppard mutation on resident macrophages in healthy muscle resembles the influence of Ppard ablation in adipose tissue macrophages in healthy mice (29). Together, the observations suggest that differences between the regulatory roles of Ppard on macrophages in injured muscle versus liver or adipose tissue may be attributable, in part, to differences in phenotype regulation in injured versus noninjured tissue rather than differences in tissue type. In contrast to the evidence that Ppard mutation does not prevent macrophage activation to an M2-biased phenotype in injured muscle, we observed that Il10 mutation produced large reductions in the numbers of CD206+ and CD163+ macrophages over the entire course of muscle regeneration that we studied. This is consistent with previous observations that IL-10 plays a significant role in macrophage phenotype transitions in muscles of dystrophic mice and mice experiencing modified muscle loading (8, 21) and that treatment of infected muscles with exogenous IL-10 increases the proportion of macrophages that express high levels of CD206 (53). We also observed that the rate of repair of the damaged muscle was impaired in Il10 mutant mice, which is consistent with previous findings that showed that diminishing macrophage phenotype transitions during muscle regeneration increased damage or slowed repair (7, 8, 21, 23). Thus, our findings collectively show that IL-10 contributes significantly to the shift of macrophages to a CD206+, CD163+, M2-biased phenotype that is independent of PPARδ signaling and accelerates muscle repair.
Although our findings show that IL-10 expression is mandatory for normal acquisition of a CD206+ M2-biased phenotype in acutely injured muscle and PPARδ is not required, it is feasible that PPARδ can contribute to acquisition of CD163+/CD206+ M2-biased phenotype in injured muscle but that function is assumed by other molecules in mutant mice. Many other molecules could potentially serve that function, including other members of the PPAR family. For example, PPARγ has a well-established role in regulating macrophage function (54), although it is not essential for development of the macrophage lineage (55). Although numerous findings indicate that PPARγ-mediated signaling can promote an M2-biased phenotype in macrophages (56–60), other findings indicate that inhibition or loss of PPARγ can bias macrophages to a CD163+/CD206+ macrophage phenotype (61). These different outcomes may reflect differences in the environment in which macrophage activation occurred and suggest that the role of PPARγ signaling in regulating macrophage phenotype may also vary in vivo according to the tissue and inflammatory context. However, there is no current evidence to show a role for PPARγ in regulating macrophage phenotype in injured muscle. Specifically, in skeletal muscle injured by snake toxin, ablation of PPARγ in myeloid cells did not affect the proportion of Ly-6c+ macrophages (associated with an M1-biased phenotype) relative to Ly-6c− macrophages (associated with an M2-biased phenotype) (62) in skeletal muscle injured by snake toxin.
Although mutation of Ppard in myeloid lineage cells did not prevent activation of macrophages to a CD206+ M2-biased phenotype in injured muscle, our data indicate that the mutation affected the kinetics of the inflammatory process. In particular, the amplification of F4/80+ macrophage numbers in injured muscles of mutant mice was delayed and was accompanied by reductions in the expression of CCR2 by intramuscular macrophages. Further, we observed an impaired chemotactic response of mutant macrophages to CCL2 in vitro. Because discoveries by other investigators showed that signaling via CCL2/CCR2 is tremendously important in promoting the traffic of macrophages into diseased or injured tissue (6, 47, 48, 63, 64), disrupting that signaling pathway by Ppard mutation would affect the time course and magnitude of inflammation. However, elevations in PPARδ signaling can also decrease CCL2/CCR2-mediated chemotaxis, at least in other tissue types and states of tissue damage. For example, activation of PPARδ by agonists suppresses atherogenic inflammation and atherosclerosis, which was attributed to the ability of PPARδ to repress the expression of CCL2 (65, 66) and thereby reduce transendothelial migration of monocytes, at least in vitro (65). These unpredictable relationships between changes in PPARδ signaling and signaling via CCL2/CCR2 and chemotaxis can be explained in part by opposing influences of PPARδ on CCL2 expression. Both pharmacological activation and genetic deletion of PPARδ can inhibit CCL2 promoter activity (66, 67), which indicates that any perturbation of Ppard expression or activity has the potential to reduce CCL2/CCR2-driven chemotaxis.
Notably, CCL2/CCR2 signaling does not only influence macrophage chemotaxis; it can also influence macrophage polarization. For example, disruption of CCL2/CCR2 signaling in human macrophages in vitro impairs their activation to a CD163+ M2-biased phenotype (68), and direct activation of human monocytes with CCL2 shifts them to a CD206+ M2-biased phenotype (69). However, in contrast with those observations, other investigators report that CCL2/CCR2 signaling in macrophages suppresses the M2 phenotype and promotes the M1 phenotype. For example, ablation of CCR2 in mice on a high-fat diet increased expression of M2 phenotypic markers in macrophages (70). Those observations suggest that the elevated numbers of CD206+ M2-biased macrophages that we observed in Ppard muscles at 7 dpi could reflect reductions of CCR2-mediated regulation of macrophage phenotype.
Because we found that Ppard mutation in myeloid cells produced either an increase or no change in numbers of macrophages that expressed markers of the M2-biased, proregenerative phenotype, we were surprised that the mutation greatly slowed muscle repair between 3 and 15 dpi. However, we observed that the mutation affected the distribution of macrophages in the injured muscles and that macrophages were more broadly distributed in injured WT muscles than in injured mutant muscles. In addition, macrophages were located in areas of regeneration and absent in areas of necrosis in both genotypes. These observations are consistent with a defect in macrophage recruitment to sites of muscle injury in the mutant mice as a consequence of reduced expression of CCR2. Associated with the reduced dispersal of macrophages in the injured muscle of mutant mice, we found significantly larger areas of muscle lacking vascularization. Although we cannot definitively address the cause of this defect, the failure to recruit proregenerative macrophages to the sites of muscle damage may contribute. Previous investigators have established that macrophages play a significant role in affecting muscle revascularization following injury (71) and that reduction in numbers of macrophages in regenerative muscle is associated with impaired revascularization and slowed repair of injured muscle (6, 48, 51). This is similar to the outcomes that we observed. Interestingly, the systemic delivery of PPARδ agonists to experimental animals increases angiogenesis in synthetic implants in vivo (72) and in skeletal muscle (73). Because endothelial cells express PPARδ and signaling through PPARδ increases proliferation of endothelial cells (72, 74), the provascular effects of PPARδ signaling were partially attributed to more rapid expansion of endothelial cell populations. However, our observations suggest the possibility that part of the increased angiogenesis in experimental animals treated with PPARδ agonists may result from an increase in PPARδ signaling in myeloid cells in addition to endothelial cells.
Collectively, our findings indicate that PPARδ signaling in myeloid cells and IL-10–mediated signaling both influence muscle inflammation and improve muscle repair following acute injury. However, their effects appear to be mediated by distinct influences on the inflammatory process, in which PPARδ signaling affects the chemotactic response of monocytes and macrophages to the injured muscle, and IL-10 signaling influences the shift in macrophages to an M2-biased phenotype, reflected by the numbers of cells that expressed the phenotypic markers CD206 and CD163. In addition to contributing to our understanding of processes that regulate muscle inflammation and repair following acute injury, these findings illustrate that the transcriptional control of macrophage phenotype switching in vivo varies with tissue type and is not always reliably predicted by in vitro findings.
Confocal laser scanning microscopy was performed at the California NanoSystems Institute Advanced Light Microscopy/Spectroscopy Shared Resource Facility at the University of California, Los Angeles. The Pax7 hybridoma developed by T.M. Jessell, Columbia University, was obtained from the Developmental Studies Hybridoma Bank, created by the National Institute of Child Health and Human Development of the National Institutes of Health, and maintained at the Department of Biology of University of Iowa (Iowa City, IA).
This work was supported by National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Institutes of Health Award R01AR066036 and National Institute on Aging, National Institutes of Health Award R01AG041147 (both to J.G.T.).
Abbreviations used in this article:
bone marrow–derived macrophage
peroxisome proliferator-activated receptor-δ
quantitative real-time PCR
Research Resource Identifier
specialized, proresolving lipid mediator
The authors have no financial conflicts of interest.