A Novel Role of Matrix Metalloproteinase-8 in Macrophage Differentiation and Polarization*

Background: Macrophages differentiation and/or polarization have been implicated in many inflammatory diseases. Results: Matrix metalloproteinase-8 (MMP8) promotes M2-macrophage differentiation and polarization through modulating TGF-β bioactivity. Conclusion: MMP8 plays a novel role in macrophage differentiation and polarization. Significance: The findings of this study significantly increased our understanding to biological functions of MMP8, the mechanisms for macrophage differentiation and polarization, and the pathogenesis of pathological conditions in which MMP8 is involved. Matrix metalloproteinase-8 (MMP8) has been shown to influence various cellular functions. As monocytes and macrophages (Mφ) express MMP8, we investigated if MMP8 played a role in macrophage differentiation and polarization. MMP8 expression was significantly increased during monocyte differentiation into Mφ. Monocyte-derived Mφ from MMP8-deficient mice expressed higher levels of M1-Mφ markers but lower levels of M2-Mφ markers than monocyte-derived Mφ from wild-type mice. Although Mφ from either MMP8-deficient or wild-type mice were inducible by interferon-γ into M1-Mφ, only wild-type Mφ but not MMP8-deficient Mφ could be induced into M2-Mφ by interleukin-4. However, MMP8-deficient Mφ exposed to conditioned culture media of wild-type Mφ developed a M2-Mφ phenotype. Compared with conditioned culture media of wild-type Mφ, conditioned culture media of MMP8-deficient Mφ contained a lower concentration of active transforming growth factor-β (TGF-β), an M2-Mφ inducer. Moreover, evidence also showed that the degradation of the TGF-β sequester, fibromodulin, was modulated by MMP8. The data indicate a previously unknown role of MMP8 in M2-Mφ polarization by cleaving fibromodulin and therefore increasing the bioavailability of the M2-Mφ inducer TGF-β.

Matrix metalloproteinase-8 (MMP8), a relatively less studied member of the MMP family, has received increasing attention in recent years. The best known substrates of MMP8 are types I, II, and III collagens. However, a growing number of other proteins have been reported to be also susceptible to cleavage by MMP8, including angiotensin-1 (10,11), chemokine (C-X-C motif) ligand 5 (CXCL5) (12), CXCL11 (13), and macrophage inflammatory protein-1 (14). Previously it was thought that only neutrophils produce MMP8, reflected by its alias names neutrophil collagenase and polymorphonuclear leukocyte collagenase. However, it is now known that various other types of cells also express this protease. Studies have shown that MMP8 can modulate the behavior and function of several cell types including neutrophils (14,15), eosinophils (16), smooth muscle cells (17), endothelial cells (18), and stem/progenitor cells (19), underlain by its proteolytic activity on matrix and non-matrix proteins. Furthermore, studies have shown that MMP8 is involved in a number of pathological conditions including atherosclerosis (11,19), acute hepatitis (20), acute lung injury (14), airway inflammation (16), and cancers (21). Because M play important roles in many pathological conditions, and given that M express MMP8, we investigated in this present study if MMP8 had a role in M differentiation and/or polarization.

Macrophage Culture and Polarization
Bone Marrow Monocyte-derived Macrophage Culture (BMM)-Bone marrow monocytes were isolated from the tibias and femurs of mice 6 -12 weeks of age. The bone marrow was triturated using an 18-gauge needle and passed through a 70-m nylon mesh cell strainer (Becton Dickinson, Franklin Lakes, NJ) to make a single cell suspension in Roswell Park Memorial Institute (RPMI)-1640 supplemented with 2% serum, 10 mM Hepes. Red blood cells in bone marrow cells were lysed with a lysis buffer (8.3 g of NH 4 Cl, 1.0 g of KHCO 3 , and 1.8 ml of 5% EDTA in 1000 ml). Bone marrow mononuclear cells were washed three times and cultured in the RPMI-1640 supplemented with 10% serum and 5 ng/ml of M-CSF for 7-14 days to induce BMM differentiation and maturation. For each independent experiment, cells isolated from 6 -8 WT (wildtype, ApoE Ϫ/Ϫ /MMP8 ϩ/ϩ ), MMP8KO (MMP8 knock-out, ApoE Ϫ/Ϫ /MMP8 Ϫ/Ϫ ), or C57BL/6 mice were pooled together, and subjected to different treatments. MMP8 WT and MMP8KO mice were generated in our previous study (11). C57BL/6 mice were purchased from Charles River. All the animal procedures were approved by Queen Mary University of London ethics review board, and performed to conform the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes or the NIH guidelines (Guide for the Care and Use of Laboratory Animals).
Peritoneal Naive Tissue Resident M (pM)-pM were isolated as described in the previous study (22). Briefly, 4 days before harvesting pM, 1.5 ml of 4% (w/v) Brewer thioglycollate medium were injected into the peritoneal cavity of each mouse to induce macrophage infiltration and accumulation into peritoneal cavity. Mice were euthanized, cleaned with 70% ethanol, and mounted on the styrofoam block on its back. 5 ml of ice-cold PBS (with 3% serum) were injected into the peritoneal cavity using a 27-guage needle, and carefully collected using a plastic Pasteur pipette. Cells in the suspension were spun down for direct RNA extraction or re-suspended in the RPMI-1640 supplemented with 10% serum for macrophage culture. For each independent experiment, cells isolated from 6 -8 WT or MMP8_KO mice were pooled together, and subjected to different treatments. In some experiments, an equal amount of PBS containing cells and peritoneal cavity fluid (4 ml) were collected from WT or MMP8KO mice and spun down to obtain cell-free supernatant. After mixing with a protease inhibitor mixture, the supernatant were concentrated using centrifugal concentrators (Millipore, UFC901024).
M1-and M2 Macrophage Polarization-Both BMM or pM were incubated with 10 ng/ml of IFN-␥ or IL-4 for 48 h to induce M1-and M2 macrophage polarization, respectively, as described previously (23). TGF-␤ and MMP8 Treatments-In separate experiments, BMM or pM were cultured for 48 h in RPMI culture medium in the presence or absence of 10 ng/ml of TGF-␤ or 50 ng/ml of MMP8, respectively. Conditioned Culture Medium Swap-BMM or pM were differentiated or isolated as described previously, followed by IL-4 polarization for 24 h. The culture medium (3 ml per well for six-well plate) was collected and swapped as indicated in the figures, followed by another 24 h of culture.

Immunoblotting
Cells were harvested and lysed in lysis buffer (50 mM Tris-Cl, pH 7.5, 150 mM NaCl, 1 mM EDTA, pH 8.0) supplemented with protease inhibitors and 0.5% Triton and sonicated to obtain whole cell lysate. 40 g of protein was separated by SDS-PAGE with 4 -20% Tris glycine gel (Invitrogen) and subjected to standard Western blot analysis. In some experiments, the blots were subjected to densitometric analysis with ImageJ software. Relative protein expression level was defined as the ratio of cleaved fragment to their respective full-length of fibromodulin protein expression level, of mature to precursor of TGF-␤ protein expression levels, of total TGF-␤ protein expression levels to GAPDH, or pSMAD3 to total SMAD3 expression levels, with that of the respective control sample set as 1.0.

Real-time Quantitative PCR (RT-qPCR)
Real-time quantitative PCR (RT-qPCR) was performed as previously described (24). Briefly, total RNAs were extracted from cells using TRI Reagent (Sigma) according to the manufacturer's instructions. Reverse transcription was performed using an Improm-II TM Reverse Transcription kit (Promega, Madison, WI) with RNase inhibitor (Promega), and Random primers (Promega). The resultant cDNA was diluted to a working concentration of 5 ng/l and stored at Ϫ20°C. Primers were designed using Primer Express software (Applied Biosystems) and the sequence for each primer was shown in Table 1. Relative mRNA expression level was defined as the ratio of target gene expression level to 18S expression level, respectively, with that of the control sample set as 1.0.

RAW264.7 Cell Culture and shRNA-mediated Stable Mmp8 Gene Silencing
Murine macrophage cell line, RAW264.7, was purchased from ATCC (ATCC TIB-71 TM ) and cultured in DMEM containing 10% fetal bovine serum. Non-target and Mmp8 shRNA lentiviral particles were produced as described in our previous study (19). For lentiviral infection, RAW264.7 cells were plated 24 h prior to infection in 6-well plates at 37°C. One transducing unit per cell (or 2-3 ϫ 10 6 /well) of Mmp8 shRNA or control virus were added with 10 g/ml of hexadimethrine bromide (H9268; Sigma). Viral constructs were incubated 24 h with the cells before the media was replaced with complete media containing 2-4 g/ml of puromycin (P9620, Sigma). For selection of transductants, fresh media containing puromycin was added at 2-3-day intervals for 10 -14 days. Stably infected cells were split and frozen for future experiments.

Fibromodulin Knockdown by siRNA and TGF-␤ Inhibition
A pool of small interfering RNAs (siRNAs) for fibromodulin (MISSION esiRNA, esiRNA targeting mouse fibromodulin, EMU010381-20UG) and MISSION siRNA Universal Negative Control number 1 (SIC001-10NMOL) were purchased from Sigma. Bone marrow-differentiated macrophages were cultured in 6-well plates for 7 days, and 6 l of 10 M siRNA (final concentration of siRNAs: 60 nM) was introduced into cells with TransIT-X2 Transfection Reagent (Geneflow Limited) according to the protocol provided. 24 h post-transfection, cells were incubated with 10 ng/ml of IL-4 for another 24 h for M2 polarization. Cells were harvested and RT-qPCR analyses were performed. In additional experiments, bone marrow-differentiated macrophages were pre-treated with TGF-␤ signaling specific inhibitor (SB-431542, 10 M, Sigma) for 3 h, followed by 10 ng/ml of IL-4 incubation for another 24 h in the presence or absence of 10 M SB-431542.

Immunofluorescence Staining for Cells
Cells were treated with various conditions as indicated in the figures and fixed, followed by immunohistological analyses with the respective antibody as described before (25,26). Briefly, cells were blocked with 5% BSA in PBS for 1 h at room temperature in a humid chamber, and incubated with the respective primary antibody or IgG control overnight in the cold room. Followed by incubation with appropriate fluorescent dye-conjugated secondary antibodies, cells were then incubated with 4Ј,6-diamidino-2-phenylindole (DAPI) (1:1000, Sigma) for 5 min. Images were assessed with Axioplan 2 imaging microscope with Plan-NEOFLUAR ϫ20, NA 0.5, objective lenses, AxioCam camera, and Axiovision software (all Carl Zeiss MicroImaging, Inc.) at room temperature, or examined with Zeiss LSM 510 confocal microscope with Plan-NEOFLUAR ϫ40 objective lenses and Zeiss ZEN microscope software (Carl Zeiss, Germany) at room temperature. All images were processed with Photoshop software (Adobe). The cells positive for M1-or M2-M markers were counted over 200 total M, and the mean fluorescence intensity (MFI) of the cells that were positive for the respective markers were analyzed using Photoshop software by two well trained independent investigators blinded to the treatments, from 20 random selected cells.

ELISA Analyses
The levels of individual cytokine in the conditioned culture medium or concentrated peritoneal cavity fluid (for active TGF-␤) were measured using their respective ELISA kit purchased from Life Technologies, according to the manufacturer's instructions.

Statistical Analysis
Data were expressed as mean Ϯ S.E. and analyzed using a two-tailed Student's t test for two-group comparison or oneway analysis of variance followed by Newman-Keuls multiple comparison post-hoc test for comparing different groups. A value of p Ͻ 0.05 was considered as statistically significant.

MMP8 Is Significantly Up-regulated during Macrophage
Differentiation-To induce macrophage differentiation and maturation from BMMs, mouse BMMs were cultured in RPMI RT-qPCR ng/ml of M-CSF for the indicated experimental time points. Total RNA, conditioned culture medium, and proteins were harvested and subjected to RT-qPCR, ELISA, and Western blot analyses, respectively. The RT-qPCR analyses showed significant increases in the expression levels of M2-M markers (arginase (Arg) I, mannose receptor C type 1 (Mrc1)/Cd206, and Cd163), but much lower levels of M1-M markers (Arg II and monocyte chemotactic protein-1 (Mcp-1)), during M differentiation (Fig. 1A). Similarly, ELISA showed significant increases in several inflammatory cytokines (MCP-1, TNF-␣, IFN-␥, and IL-12) in conditioned culture media, over the course of M differentiation and maturation (Fig. 1B), and Western blot analyses showed significant induction in three M2-M markers examined in this study (Fig. 1C). Immunofluorescence staining showed that more than 95% of differentiated cells were positive for the M marker F4/80 after 10 -14 days (Fig. 1D). These data suggest that M-CSF promotes M differentiation and maturation from BMMs with a predominant M2-M phenotype, as suggested in previous studies (1).
Importantly, we also observed that MMP8 expression significantly increased during M differentiation and maturation ( Fig. 1, A and C). Double immunostaining demonstrated co-expression of MMP8 with M specific markers F4/80 (Fig. 1D).

MMP8 Deficiency Leads to a M1 Macrophage Phenotype-To
investigate whether MMP8 plays a role in M differentiation and maturation from BMMs, BMMs isolated from MMP8 knock-out (MMP8KO) mice and WT control littermates were induced to differentiate into M by M-CSF stimulation for 14 days. RT-qPCR analyses showed that MMP8 deficiency significantly increased the expression levels of M1-M genes (Mcp-1, Arg II, Irf5, Tnf-␣, Il-6, and Il-12␤), but reduced M2-M gene expression (Tgf-␤, Arg I, Irf4, peroxisome proliferator-activated receptor-␥, Cd206, and Cd163), in differentiated M ( Fig. 2A). Western blot analyses showed corresponding changes in Arg I and Arg II protein levels, and that the MMP8 protein expression was absent in MMP8-deficient M (Fig. 2B). Conditioned culture media of MMP8KO M contained higher levels of inflammatory cytokines including TNF-␣, IFN-␥, IL-12, and MCP-1, as demonstrated by ELISA analyses (Fig. 2C). Importantly, at day 14, the percentage of M1-M was significantly higher and the percentage of M2-M was much lower, in M differentiated from MMP8KO BMMs than in M differentiated from WT BMMs, although there was no significant differences in the total numbers of M (F4/80positive cells, data not shown) (Fig. 2D). Expectedly, whereas MMP8 was clearly expressed in M differentiated from WT BMMs, no apparent signal for MMP8 was detected in M dif- ferentiated from MMP8KO BMMs (Fig. 2E). A similar effect of MMP8 deficiency was observed in naive pM (peritoneal cavity macrophages) (Fig. 2, F-H). Briefly, compared with WT pM, an increased expression level of M1-M markers, but a decreased expression level of M2-M markers, was observed in MMP8-deficient pM (Fig. 2, F and G). Similarly, more M1-M (percentage of MCP-1-positive cells) and less M2-M (the percentage of CD163-or CD206-positive cells) were observed in MMP8-deficient pM (Fig. 2G).
MMP8 Is Required for M2-M Polarization-To investigate whether MMP8 is involved in M polarization, BMM from MMP8KO and WT mice were incubated with M polarization inducers (10 ng/ml of IFN-␥ for M1 priming, and 10 ng/ml of IL-4 for M2, respectively) for 48 h. The M1 priming by IFN-␥ increased the expression of M1-M markers, but this effect was significantly greater in MMP8KO BMM than in WT BMM (Fig. 3, A and B). In contrast, the M2 inducer IL-4 significantly up-regulated the expression of M2-M markers in WT BMM, but such an effect was not observed in MMP8KO BMM (Fig. 3, C and D), indicating that MMP8 deficiency significantly reduces M2-M polarization. This was confirmed using the naive pM (Fig. 3

, E-H).
It has been reported that ApoE can affect M polarization (27). The MMP8KO and WT mice used in the above experiments were apoE deficient. To investigate if the effect of MMP8 on M2-M polarization is apoE independent, murine macrophage cells, RAW264.7 (apoE wild-type), were infected with either Mmp8-specific shRNA lentivirus and non-target  (Fig. 4C), suggesting that the effect of MMP8 on M polarization is ApoE independent. To further confirm the functional importance of MMP8 in macrophage polarization in a truly wild-type genetic background, BMM from C57BL/6 were incubated with non-target shRNA or Mmp8 shRNA lentivirus and primed to M1-or M2-M as described previously. A similar effect of Mmp8 knockdown was observed in these BMM with a truly wildtype genetic background (Fig. 4, D-F   trations of molecules involved M polarization, WT BMM was cultured with conditioned medium of MMP8KO BMM, and MMP8KO BMM was cultured with conditioned medium of WT BMM. The experiment showed that culturing WT BMM with conditioned medium of MMP8KO BMM resulted in a significant decrease in M2-M genes/proteins (Fig. 5, A and B). In contrast, replacing the culture medium for MMP8KO BMM with conditioned medium of WT BMM almost restored the expression levels of M2-M genes/proteins in MMP8-deficient M (Fig. 5, A and B). An experiment using pM showed similar results (Fig. 5, C and D). These data suggest that the effect of MMP8 on M2-M polarization is mediated by certain biological molecule(s) in medium.
The Effect of MMP8 on M2 M Polarization Is Mediated by TGF-␤1-It has been reported that Tgf-␤ signaling plays a critical role in promoting IL-4-induced alternative M activation (M2-M polarization) (28). We found that TGF-␤1 mRNA expression levels were significantly lower in MMP8-deficient BMM and pM than in WT BMM and pM (Figs. 2, A and  F). To investigate whether the effect of MMP8 on M2 M polarization was mediated by TGF-␤1, the concentrations of total and active TGF-␤1 in conditioned culture media of BMM and pM were measured by ELISA. The experiment showed that the levels of total and active TGF-␤1 increased during M differentiation (Fig. 6A), and that active TGF-␤1 levels, but not total TGF-␤1 levels, were lower in conditioned media of MMP8-deficient BMM and pM than in condi-tioned media of WT BMM and pM (Fig. 6, B and C). Importantly, a higher level of active TGF-␤1 was also observed in the peritoneal cavity fluid of WT mice induced by Brewer thioglycollate medium injection compared with MMP8KO mice (Fig.  6D).
TGF-␤1 is synthesized as a latent precursor and secreted from cells as a latent complex (an inactive form) containing latent TGF-␤-binding protein and TGF-␤ pro-peptide called latency-associated peptide (LAP) (29,30). Activation of latent TGF-␤ involved cleavage of LAP (29,30). We found that IL-4 priming induced a significant increase in the levels of active TGF-␤1 in WT BMM and pM, but not in MMP8-deficient BMM and pM (Fig. 6, E and F). Immunostaining with an antibody against LAP showed that MMP8-deficient BMM and pM expressed higher levels of LAP protein than both WT BMM and pM (Fig. 6, G-J). Western blot analyses showed a much higher level of TGF-␤ cleavage (ratio of the mature to precursor of TGF-␤) occurred in the WT BMM, which was further increased by IL-4. No such induction by IL-4 was observed in MMP8-deficient BMM (Fig. 6, K and L). A moderate but significant lower level of total TGF-␤ was observed in the cell lysate of MMP8-deficient BMM (Fig. 6, K and L). Importantly, the protein level of phosphorylated SMAD3 (pSMAD3) was much lower in MMP8-deficient BMM than that of WT BMM, and IL-4 further increased the pSMAD3 protein level in WT BMM, but not in MMP8-deficient BMM (Fig. 6, K and L), suggesting that IL-4 induces M2-M polarization through activation of TGF-␤ signaling, and such signaling is at least partially responsible for MMP8-mediated M2-M polarization.
Furthermore, culturing MMP8-deficient BMM and pM with conditioned media of WT BMM and pM for 24 h increased active TGF-␤1 levels similar to those in conditioned media of WT BMM and pM (Fig. 7, A and B). To further determine whether TGF-␤1 mediates the effect of MMP8 on M2-M polarization, MMP8KO and WT BMM were incubated with a recombinant active TGF-␤1, followed by analyses of M2-M genes. The experiment showed that incubation with the recombinant active TGF-␤1 significantly increased the expression of M2-M genes in both WT and MMP8KO BMM (Fig. 7, C and D). Taken together, the above data suggest that the effect of MMP8 deficiency on M2-M polarization is, at least in part, due to reduced levels of active TGF-␤1.

MMP8 Increases TGF-␤1 Bioavailability and Regulates M2-M Polarization by Increasing Fibromodulin Cleavage-It
has been reported that the matrix protein fibromodulin can interact with TGF-␤1 and regulate TGF-␤1 bioavailability/activity through sequestering TGF-␤1 in the extracellular matrix (31). It has been suggested that MMPs can cleave fibromodulin (32). In this study, we found reduced cleavage of fibromodulin (lower ratio of the cleaved fragment to full-length fibromodulin) in MMP8-deficient M (Fig. 8A) and that IL-4 priming induced fibromodulin cleavage in WT BMM but not in MMP8-deficient M. Further experiments showed that incubation of M with 50 ng/ml of recombinant activated MMP8 significantly increased fibromodulin cleavage as well as the levels of active TGF-␤1, in conditioned culture media of both WT and MMP8-deficient M (Fig. 8, B and C). Furthermore, incubation of M with the recombinant activated MMP8 significantly up-regulated the expression of M2-M genes in both WT and MMP8-deficient M, as demonstrated by RT-qPCR (Fig. 8D) and immunostaining (Fig. 9), suggesting that MMP8 induces M2-M polarization via cleavage of fibromodulin that affects TGF-␤1 bioavailability.

Functional Involvements of Fibromodulin and TGF-␤ Signaling in IL-4-mediated M2-M Polarization-
To further establish the causal link between fibromodulin and M2-M polarization induced by IL-4, fibromodulin knockdown in BMM was conducted by using specific Fibromodulin siRNA. As expected, the endogenous expression level of Fibromodulin in BMM was successfully knocked down by Fibromodulin-specific siRNAs (Fig. 10A). Consequently, M2-M-specific gene expression levels (Arg I, Cd163, and Cd206) were significantly up-regulated (Fig. 10A, 3rd bar versus 1st bar), such inductions were further increased in the presence of IL-4 (Fig. 10A, 3rd bar  versus 4th bar), suggesting a functional role of fibromodulin in M2-M polarization. Moreover, to investigate if TGF-␤ signaling plays a causal role in IL-4-induced M2-M polarization, the TGF-␤ signaling specific inhibitor, SB-431542, was used to FIGURE 6. Bioavailability of TGF-␤ was mediated by MMP8 during macrophage differentiation and polarization. Cells were cultured and treated as described previously. Conditioned culture medium (CM) were harvested and subjected to ELISA analyses. A, both total and active TGF-␤ levels were significantly increased during bone marrow macrophage differentiation from monocytes. B and C, active, not total TGF-␤ levels in MMP8-deficient BMM (B) and pM (C) were lower than that of WT macrophages. D, active TGF-␤ levels in the peritoneal cavity fluid of MMP8-deficient mice were lower than that of WT mice. E and F, active TGF-␤ levels in CM of WT and MMP8-deficient BMM (E) and pM (F) in response to IL-4 polarization. The data presented here are an average of three to four independent experiments. *, p Ͻ 0.05 (day 14 versus 7 or MMP8KO versus WT); #, p Ͻ 0.05 (M inducers versus control). G-J, MMP8-deficiency results in higher amount of LAP on BMM (G and H) and pM (I and J). Cells were fixed and subjected to immunofluorescence staining analyses with antibody against LAP (N terminus of TGF-␤1). Shown in the figure are representative images each from three independent experiments, and column charts of mean fluorescence intensity (mean Ϯ S.E., n ϭ 20). *, p Ͻ 0.05 (versus controls). K and L, Western analyses of the expression levels of TGF-␤ and pSMAD3. Proteins were harvested and subjected to Western blot analyses with the antibodies against the C terminus of TGF-␤ and pSMAD3 (phospho-Ser 423 /Ser 425 ). Note: the molecular mass for the bands of TGF-␤1 (precursor) and TGF-␤1 (mature) in panel K are ϳ37.5 and ϳ12.5 kDa, respectively. GAPDH and total SMAD3 were included as internal controls. Shown in the figure are representative images each from three independent experiments, and column charts of relative protein levels (mean Ϯ S.E., n ϭ 3). *, p Ͻ 0.05 (versus WT/vehicle). block activation of the TGF-␤ signal pathway in the BMM. As shown in Fig. 10B, the expression level of phosphorylated SMAD3 (pSMAD3), one of the key downstream effect genes of the TGF-␤ signal pathway, was significantly up-regulated by IL-4, whereas its induction was almost abolished by incubation with SB-431542. Expectedly, incubation with SB-431542 in BMM resulted in a decrease of M2-M gene expressions, and importantly inhibition of the TGF-␤ signaling pathway almost abrogated M2-M gene expressions induced by IL-4 (Fig. 10C), implying a functional role of TGF-␤ signaling in IL-4-induced M2-M phenotype.

Discussion
The key novel finding of this study is that MMP8 deficiency has an effect on M differentiation and polarization, favoring M1 over M2, suggesting a particular role of MMP8 in M2-M differentiation and polarization. This study indicates that this effect is related to fibromodulin cleavage and therefore increased bioavailability of TGF-␤1 that induces M2-M. This finding is relevant for the understanding of regulatory mechanisms for M differentiation and polarization, and for the understanding of the pathogenesis of inflammatory conditions in which MMP8 is implicated.
Previous studies have shown that MMP8 has a protective effect against acute hepatitis (20), acute lung injury (14), airway inflammation (16), and cancers (21). These studies have demonstrated that the protective effect of MMP8 in these conditions is related to its influence on the behavior and function of neutrophils (14,16,21) or of eosinophils in the case of airway inflammation (16). The finding of the present study raises the possibility that the effect of MMP8 in some of these conditions might also be related to its role in M2-M polarization, which warrants further investigations.
There is also substantial evidence from experimental and clinical studies indicating MMP8 promotes atherosclerosis and a number of atherosclerosis-related conditions including atherosclerotic plaque rupture leading to myocardial infarction, heart failure after myocardial infarction, neointima formation following angioplasty, and abdominal aortic aneurysm. The pathogenesis of atherosclerosis involves a number of cell types, with endothelial cells, macrophages, lymphocytes, and smooth muscle cells having been studied most. The finding of the present study that MMP8 promotes M2-M polarization seems at odds with the finding that MMP8 promotes atherosclerosis, because inflammation plays an important role in atherogenesis and because M2-M are anti-inflammatory. However, athero-  JULY 31, 2015 • VOLUME 290 • NUMBER 31 genesis also involves a number of other cell types, and previous studies have indicated that MMP8 can promote atherosclerosis partly via its influences on migration and proliferation of stem/ progenitor cells (19), smooth muscle cells (17), and endothelial cells (18), as well as leukocyte recruitment (11). Thus, it appears that MMP8 can influence atherosclerosis via multiple mechanisms.

MMP8 in Macrophage Differentiation and Polarization
Several other MMPs have been reported to have an effect on M polarization (23,33). A recent study showed that Mmp28 gene inactivation impaired M2-M polarization and resulted in an aggravated cardiac dysfunction after myocardial infarction in mice (23). Another study showed that loss of MMP28 reduced M2 polarization and protection from bleomycin-in-duced fibrosis (33). It has also been shown that loss of MMP7 resulted in M1-M polarization within Helicobacter pylori-infected stomachs, and M isolated from MMP7-deficient mice infected with H. pylori expressed significantly higher levels of the M1-M marker IL-1␤ compared with M isolated from WT mice (34). Furthermore, MMP1, MMP3, and MMP10 have been found to be highly expressed in M1-M, whereas MMP12 has been found to be strongly expressed in M2-M (35). It has also been shown that classical activation of mouse M increased the expression of MMP13, MMP14, and MMP25 but decreased MMP19 and TIMP2, whereas alternative activation with IL-4 increased MMP19 expression (36). Taken together, findings from previous and present studies indicate that various MMPs play important and divergent roles in macrophage polarization. It would be very interesting to study the relationships between these reported MMPs and MMP8 in IL-4-induced M2-M polarization. To this aim, we first examined if the expression levels of these MMPs were regulated by IL-4 in our cell culture system. Our data showed that both Mmp12 and Mmp19 were significantly up-regulated by IL-4 treatment, whereas the expression levels of Mmp7 and Mmp28 were not dramatically changed upon IL-4 treatment (data not shown). Interestingly, the expression levels of Mmp12 and Mmp28 were much lower in MMP8-deficient M than that of WT M, whereas Mmp7 and Mmp19 were mildly increased in MMP8deficient M compared with that in WT M (data not shown), suggesting a potential role for these MMPs in MMP8-mediated M2-M polarization, which warrants further investigation. TGF-␤ signaling has recently been suggested to play a central role in promoting M2-M activation (28). TGF-␤1 is synthesized as a latent precursor and secreted from cells as a latent complex. Once secreted from cells, the TGF-␤1 latent complex is sequestered in the extracellular matrix through binding to matrix proteins (29,30). Proteolytic degradation of the LAP or the bound matrix proteins by specific proteolytic enzymes such as MMPs, or mechanical stretching of cell-surface integrins, release active TGF-␤ and allow it to interact with TGF-␤ receptors and induce transcription of TGF-␤-responsive genes by initiating the canonical Smad-dependent pathway and/or non- canonical pathways involving MAPK and RhoA (37). In the current study, we provided several lines of evidence to support the notion that MMP8 mediated M2-M differentiation and polarization through activating the TGF-␤ signal pathway. First, our data showed that the gene expression levels (Fig. 2, A and F) and total cellular protein levels (Fig. 6, K and L) of TGF-␤ were much higher in WT M than that of MMP8KO M. Second, compared with WT, a decreased level of TGF-␤, particularly the active form of TGF-␤, was observed in the conditioned culture medium of MMP8KO M (Fig. 6, B and C) and in the peritoneal cavity of MMP8KO mice (Fig. 6D). Third, compared with the WT M the level of pSMAD3 was much lower in MMP8-deficient M (Fig. 6, K and L), suggesting that TGF-␤ signaling is suppressed in MMP8-deficient M. Fourth, exogenous TGF-␤ could restore the inhibitory effects of Mmp8 gene inactivation on M2-M gene expression (Fig. 7, C and D). Finally, our data also revealed that the TGF-␤ signalling pathway at least partially was responsible for IL-4-induced M2-M polarization (Fig. 10, B and C). Interestingly, we observed a more significant inhibitory effect of Mmp8 gene inactivation on Tgf-␤ mRNA expression (Fig. 2, A and F) than the total TGF-␤ protein level in conditioned culture medium (Fig. 6, B and C), indicating a role for MMP8 in TGF-␤ post-transcription and/or post-translational regulation, or in modulating TGF-␤ protein secretion from cells, which remains to be investigated in a future study. Furthermore, it would be interesting to investigate if MMP8 could affect TGF-␤ dimmer formation. For such a purpose, we have applied different non-reduced and/or nondenatured conditions in our Western blot analyses (e.g. omitting ␤-mercaptoethanol and/or SDS in loading buffer or running buffer, boiling or not boiling samples), which is prerequisite for detecting the protein dimers in whole cell lysate, but we were unsuccessful to obtain any meaningful data under these conditions. Although we have failed to show the effects of MMP8 deficiency on TGF␤1 dimmer formation, theoretically, the TGF␤1 protein dimer level will be much lower in the MMP8KO macrophage due to the facts that 1) a much lower level of TGF␤1 monomer has been observed in MMP8KO macrophages (Fig. 6K); and 2) there has been no report to suggest that MMP8 plays a role in protein dimmer formation through disulfide bond(s).
Studies have indicated that the matrix protein fibromodulin can interact with TGF-␤1 and regulate TGF-␤1 bioavailability/ activity through sequestering TGF-␤1 in the extracellular matrix (31, 38 -40). Other studies have shown that MMP2, MMP8, MMP9, and MMP13 can cleave fibromodulin (32). In line with these findings, our present study indicates that MMP8 induces M2-M differentiation and polarization via degrading fibromodulin and increasing the bioavailability of the M2-M inducer TGF-␤1. It is noteworthy to mention that one most recent study (41) has suggested an opposite role for MMP8 in regulating TGF-␤ signaling in breast cancer cells and metastasis progression. Soria-Valles et al. (41) reported that MMP8 inhibited tumor growth and breast cancer cell lung metastasis through supressing TGF-␤ signaling via cleaving and releasing another matrix protein, decorin. Subsequently, the cleaved decorin by MMP8 can bind to soluble TGF-␤ and prevent its interaction with TGF-␤ receptors. Therefore, it would be plausible to speculate that MMP8 may play a distinct role in the TGF-␤ signaling pathway in different cellular contexts or the regulatory role of MMP8 in the TGF-␤ signal pathway is likely substrate-dependent. It also would be very interesting to investigate the expression levels and cleavage of decorin by MMP8 in M, however, such investigations are beyond the remit of the current study.
In conclusion, we have identified a novel role of MMP8, i.e. in M2-M differentiation and polarization. The findings of this study are useful for understanding the biological functions of MMP8, the mechanisms for M differentiation and polarization, and the pathogenesis of pathological conditions in which MMP8 is involved.
Author Contributions-G. W. designed and performed the experiments and analyzed the data. C. Z. and Q. C. performed and analyzed RT-qPCR experiments. L. A. L. performed microscopy experiments. A. M. isolated and cultured peritoneal M. S. Y. supervised the study and wrote the manuscript. Q. X. conceived the study, analyzed the data, supervised the study and wrote the manuscript.