Stromelysin-2 (MMP10) Moderates Inflammation by Controlling Macrophage Activation Free

Macrophages are critical effector cells of the immune system and play essential, yet distinct, roles in both promoting and resolving inflammation and in facilitating tissue repair and contributing to its destruction (1). That a single cell type can serve opposing functions may seem counterintuitive, but dramatic phenotypic changes occur when macrophages respond to local stimuli (1–4). Based on patterns of gene and protein expression and function, macrophages are commonly classified as classically activated (M1) or alternatively activated (M2) cells, as well as a variety of M2 subtypes (1, 4, 5). M1 macrophages are induced by infection and proinflammatory Th1 cytokines, are effective at killing bacteria, and release proinflammatory cytokines, such as IL-1β, IL-12, and TNF-α. M2 macrophages are induced by the Th2 cytokines IL-4 and IL-13 and other factors, release anti-inflammatory factors, such as IL-10, are weakly microbicidal, and promote repair. We recognize, however, that dividing macrophages into M1 and M2 classes oversimplifies the complex continuum of functional and reversible states in which these cells exist (6–9).

Several proteins influence macrophage behavior, including some members of the matrix metalloproteinase (MMP) gene family. For example, MMP12 and MMP28, both macrophage products, either promote or restrict macrophage influx into lung (10, 11), and MMP8 promotes M2 polarization (12). Additionally, we reported that MMP28 and TIMP3, an MMP inhibitor, moderate M1 activation of macrophages in models of lung infection and fibrosis (13, 14). As their name implies, MMPs are thought to degrade extracellular matrix proteins, a function that is indeed performed by some members (15–17). However, matrix degradation is neither the shared nor predominant function of these enzymes. Rather, individual MMPs have been shown to regulate specific immune processes, such as leukocyte influx and activation, antimicrobial activity, and restoration of barrier function, typically by processing of a range of non–matrix protein substrates (18–25).

MMP10 (stromelysin-2) is not expressed in developing or normal adult tissues, including lung (26, 27). However, in both human conditions and mouse models, MMP10 is induced in response to injury, infection, or transformation in essentially all tissues (28–33). In a meta-analysis of gene array experiments involving numerous different host–pathogen interactions, MMP10 was identified as a common host response gene (34). The widespread expression of MMP10 among tissues suggests that this proteinase serves critical roles in the host response to environmental insults. As we show in the present study, MMP10 serves a protective role in acute infection by moderating the proinflammatory activity of macrophages.

Recently, we reported that macrophage MMP10 promotes the ability of M2 macrophages to clear scar tissues in normal skin wounds by controlling the expression of collagenolytic MMPs, particularly MMP13 (35). However, because sterile excision wounds are not associated with a profound inflammatory response, the stimulus (i.e., a clean wound) may not have been sufficiently robust to reveal other MMP10-dependent roles in macrophage inflammation. Thus, for the present studies, we challenged Mmp10^−/− mouse lungs with Pseudomonas aeruginosa, an opportunistic pathogen and common cause of hospital-acquired infections and pneumonia (36) and the major pathogen in lungs of patients with cystic fibrosis (CF) (37). Compared to wild-type mice, we found that Mmp10^−/− mice were much more susceptible to airway infection and that MMP10 influenced activation of resident and infiltrated macrophages by curbing M1 polarization and P. aeruginosa–induced transcriptional changes. Our findings indicate that MMP10 is a critical cell-autonomous mediator controlling macrophage activation.

Mmp10^−/− mice and wild-type littermates (C57BL/6J, male and female, 8–16 wk old) were used for these studies. All procedures were approved by the Office of Animal Welfare at the University of Washington and the Institutional Animal Care and Use Committee at Cedars–Sinai Medical Center. Mmp10^−/− mice were generated by targeting exons 3–5 that code the catalytic domain as described (38). Mmp10^−/− mice breed and appear normal with similar litter sizes. We also generated Mmp10^−/− mice (Mmp10^tm1Lex/Mmucd) with embryonic stem cells from the Mutant Mouse Resource and Research Center (https://www.mmrrc.org/catalog/sds.php?mmrrc_id=11737). These cells were originally generated by Lexicon Pharmaceuticals (The Woodlands, TX), who used a targeting strategy distinct from ours.

Mice were challenged with live P. aeruginosa strain K either by aerosolization in a whole animal nebulizer chamber for 30 min to produce a deposition of bacteria between 10⁵ and 10⁶ CFU/left lung or by oropharyngeal aspiration as described (11, 39). Briefly, bacteria were grown in Luria–Bertani agar plates overnight, washed twice, and suspended in sterile PBS. For oropharyngeal aspiration, bacteria were quantified by OD₆₀₀ and suspended at 2 × 10⁸ cells/ml, and 50 μl was instilled per mouse. After exposure by either method, the lungs of three to five mice were immediately harvested and homogenized to determine 0 h bacterial counts (CFU). CFU were determined by plating serial dilutions of left lungs or spleens homogenized in 1 ml PBS on Luria–Bertani agar plates. Duplicate plates of each dilution were incubated 24–48 h at 37°C and counted using a Quebec colony counter (Reichert Technologies, Depew, NY).

The left lung hilum was ligated, and the left lung and spleen were removed for determination of bacterial load. Bronchoalveolar lavage (BAL) was collected by flushing the right lung via the trachea four times with 0.5 ml saline. The recovered lavage (typically 1.7 ml) was centrifuged at 400 × g, 4°C for 10 min. Supernatants were collected and frozen at −70°C for later analysis. Total protein in BAL was quantified using the Pierce BCA protein assay kit (Thermo Scientific, Rockford, IL) and IgM by ELISA (Bethyl Laboratories, Montgomery, TX). Pelleted cells were resuspended in DMEM with 10% FBS, counted in a hemocytometer, and a 100-μl aliquot was cytospun onto slides and differentially stained with LeukoStat (Fisher Scientific, Pittsburgh, PA). At least 300 cells were counted per lavage. Lungs from separate mice were used for RNA isolation or flow cytometry (see below).

Deidentified human lung specimens were obtained with approval of the University of Washington Institutional Review Board. Sections (5 μm) were stained with Abs specific for human MMP10 (Abcam, Cambridge, MA) as described (40). Mouse lungs were perfused-fixed with PBS-buffered formalin via the trachea under 25-cm pressure for 5 min, incubated in fixative at room temperature for 48 h, dehydrated through graded ethanol, and embedded in paraffin. Deparaffinized sections (5 μm) underwent heat-induced (95°C, 30 min) Ag retrieval in pH 7.0 citrate. Endogenous peroxidase activity was quenched with 3% H₂O₂ for 10 min. Sections were incubated with 4% normal rabbit serum for 1 h and then overnight at 4°C with anti-Mac2 Ab (1:500 dilution; clone M3/38.1.2.8 HL.2, American Type Culture Collection, Manassas VA). Bound Ab was detected using a Vectastain Elite ABC (rat IgG) kit (Vector Laboratories, Burlingame, CA) and 3,3′-diaminobenzidine (DAB) peroxidase substrate and counterstained with hematoxylin. For combined in situ hybridization/immunostaining, deparaffinized rehydrated sections were treated with 10 μg/ml proteinase K in RNase-free PBS for 20 min at ambient temperature, incubated with 0.1 M triethanolamine in RNase-free water, washed in RNase-free PBS, and then probed overnight at 55°C with either a 570-bp sense or antisense digoxigenin (DIG)-labeled RNA probe targeting bases 1050–1620 of murine Mmp10. Probes were labeled using DIG RNA labeling mix and SP6/T7 DIG labeling kit (Roche Diagnostics, Indianapolis, IN). After hybridization, slides were washed three times in 2× SSC/50% formamide for 15 min at 50°C, once in 2× SSC at ambient temperature for 20 min, and once in 0.5× SSC for 30 min then developed with anti–DIG-alkaline phosphatase using NBT/BCIP as substrate. Slides were washed with PBS and images were captured using an Olympus BX51 with a DP70 digital camera (Olympus America, Center Valley, PA). Sections were then processed for immunostaining with anti–Mac-2 as described above minus Ag retrieval and counterstaining.

Whole lungs were perfused with 10 ml ice-cold PBS through the left ventricle and then harvested, minced into tiny pieces using a razor blade, and digested in RPMI 1640 containing 2.5 mg/ml collagenase type IV (Life Technologies) for 45 min at 37°C with shaking. The digestion was stopped with addition of FBS to 10%. The lung cell suspension was applied to a 70-μm cell strainer and washed with cold RPMI 1640. The sample was subjected to RBC lysis (ACK lysing buffer; Thermo Fisher), followed by washing in RPMI 1640 and cell counting using a Bio-Rad cell counter. Lung cells were stained with fluorophore-conjugated Abs and analyzed using a BD LSRFortessa or sorted using a BD FACSAria III (BD Biosciences, San Jose, CA). Cell populations in the lungs were identified using the following combinations of cell surface markers: alveolar macrophages, F4/80⁺CD11c⁺Siglec-F⁺; interstitial macrophages, F4/80⁺CD11b⁺CD11c⁺MHC class II (MHC II)⁺Siglec-F⁻; Ly6C⁺ monocytes/macrophages, F4/80⁺Ly6C⁺CD11b⁺CD11c⁺MHC II⁻Siglec-F⁻; and neutrophils, F4/80⁻Ly6G⁺CD11b⁺. Sorted cells were kept at 4°C during sorting, centrifuged, and immediately processed for RNA isolation as described below. Siglec-F Ab was from BD Biosciences; all others were from eBioscience (San Diego, CA).

Bone marrow–derived macrophages (BMDM) and alveolar macrophages were isolated and cultured as described (11, 39). To generate BMDM, marrow cells were differentiated in CSF-1–containing medium for 7 d (referred to as M0 macrophages). To generate M1-biased macrophages, M0 BMDM (1.5 × 10⁶/well in six-well plates) were stimulated with 100 ng/ml Escherichia coli 0111:B4 LPS for 24 h or with live P. aeruginosa strain K at a multiplicity of infection of 5 bacteria per macrophage for 1 h before washing away bacteria and then culturing in medium with 75 μg/ml gentamicin to kill live extracellular bacteria. For M2 polarized macrophages, BMDM were stimulated with 10 ng/ml each of IL-4 and IL-13 for 48 h. For adoptive transfer, recipient mice were given 2 × 10⁶ wild-type or Mmp10^−/− M0 BMDM in 100 μl PBS retro-orbitally 24 h after infection. For the microarray experiments, M0 BMDM were exposed to P. aeruginosa strain K as described above. The cells were then cultured in medium containing gentamicin, 100 IU/ml penicillin, and 100 μg/ml streptomycin, and RNA was collected 6 or 24 h later. For alveolar macrophages, cells in BAL (98% macrophages) from four groups of five naive wild-type mice per group were pooled, and 2 × 10⁵ cells were cultured in 24-well plates overnight in RPMI 1640 containing 10% FBS and penicillin/streptomycin before stimulating with LPS or bacteria. The phagocytosis assay is described in the Supplemental Material.

Total RNA was isolated from lungs or macrophages and RT-PCR was performed and quantified as described (41). Unlabeled PCR primers and TaqMan probes (FAM dye labeled) were used to detect mRNAs for Nos2, Ifng, Ccl2, Ccl3, Ccl5, Il6, Tnf, Ccr2, Arg1, Il10, Mrc1, Clec10a, Retnla, Mmp3, 7, 8, 10, 12, and 28, and Hprt (Applied Biosystems, Foster City, CA). The data are expressed as relative quantification, which is the fold change and calculated as 2^−ΔΔCt.

Lung samples were homogenized in 2 ml lysis buffer: 150 mM NaCl, 15 mM Tris (pH 7.2), 1 mM MgCl₂, 1 mM CaCl₂, 1% Triton X-100, and cOmplete mini protease inhibitor cocktail (Roche Molecular Biochemicals, Mannheim, Germany). Homogenates were centrifuged at 10,000 × g for 15 min at 4°C, and the supernatants were collected. Multiplex reagents were from R&D Systems (Minneapolis, MN), including the mouse Fluorokine MAP kit and analytes for IFN-γ, IL-1β, IL-6, IL-10, IL-12, IL-17, TNF-α, CXCL1, CXCL2, and CCL2.

Total RNA from BMDM was purified using RNeasy mini kits (Qiagen, Hilden, Germany). All RNA samples were of high quality as assessed by an Agilent 2100 bioanalyzer, with RNA integrity number ≥ 9.8. Each RNA sample was converted to cDNA, labeled, and hybridized to GeneChip mouse gene 2.0 ST arrays (Affymetrix, Santa Clara, CA) following the manufacturer’s protocols. The whole-genome GeneChip mouse gene 2.0 ST is comprised of 35,240 transcripts including 26,515 unique genes. Hybridized arrays were scanned with an Affymetrix GeneChip 3000 scanner, and image generation and feature extraction were performed using Affymetrix GeneChip operating software. All arrays passed the manufacturer’s quality specifications with respect to background and percentage present call rates. Gene expression levels were estimated from probe intensities using the robust multiarray analysis method with quantile normalization and background adjustment (42). Differential gene expression between wild-type and Mmp10-null samples at each time point was determined using a Bayesian implementation of the parametric t test (43), and multiple hypothesis testing was controlled using false discovery rate analysis (q value) (44). A q value of <0.01 was used to identify significant differential gene expression. Functional enrichment analysis of differentially expressed genes was performed with WebGestalt program (45) based on multiple publicly available resources including Gene Ontology and Kyoto Encyclopedia of Genes and Genomes. Enrichment p values were determined using a Fisher exact test and corrected for multiple testing using the Benjamini–Hochberg method (46) with a significance adjusted p value of <0.01. Minimal information about microarray experiments–compliant data have been submitted to the National Institutes of Health Gene Expression Omnibus (accession no. GSE78175; http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?token=epkvmssmvhmrlmf&acc=GSE78175).

Statistical analysis was performed using two-way ANOVA with a Bonferroni posttest where appropriate. Data are presented as means ± SEM, with a p value ≤ 0.05 considered statistically significant. Mortality data were analyzed by the Kaplan–Meier log-rank test. Prism 5 software (GraphPad Software, La Jolla, CA) was used for all statistical analyses.

We stained de-identified samples of CF lungs, all colonized with P. aeruginosa, from transplant recipients (n = 5) and normal human lungs (n = 3) for MMP10 protein using a human-specific Ab. In all CF samples, signal for MMP10 was seen in most mononuclear cells (Fig. 1B, 1C). The large, pale nuclei of these cells and their accumulation within interstitial areas are characteristic features of macrophages. No signal for MMP10 was detected in normal human lung, including resident alveolar macrophages (Fig. 1A).

We used in situ hybridization coupled with immunostaining to identify the cell sources of Mmp10 mRNA expression in mouse lung. Similar to human lung, we detected no signal for Mmp10 mRNA in naive mouse lungs (Fig. 1E). However, at 72 h postinfection with P. aeruginosa, we detected a prominent signal for Mmp10 mRNA (Fig. 1F) in cells that stained positive for Mac-2, a pan-macrophage marker (11), demonstrating that all cells strongly positive for Mmp10 mRNA were macrophages (Fig. 1G, 1G′). No specific signal was seen in sections processed with sense RNA probe (Fig. 1H). These findings demonstrate that MMP10 is not expressed in healthy lung but is induced by macrophages in human and murine lung in response to bacterial infection.

To determine the role of MMP10 in the host response to infection, we exposed wild-type and Mmp10^−/− mice to live P. aeruginosa by aerosolization or oropharyngeal aspiration. Unchallenged Mmp10^−/− mice develop normally and reveal no overt phenotype in any tissue (33, 35, 38). Consistent with our in situ hybridization data (Fig.1E), Mmp10 mRNA levels were very low (average Ct = 38) in naive wild-type lung, indicating that this transcript was not expressed by any resident cell, including alveolar and interstitial macrophages in unchallenged lung. In response to P. aeruginosa infection, expression of Mmp10 mRNA was induced in wild-type lungs, peaked 3 d postinfection, and remained elevated at 7 d postinfection (Fig. 2A). The pattern and extent of MMP10 expression were equivalent between the two infection protocols and mirrored the 0–24 h induction response we reported in mice infected by nasal inoculation with P. aeruginosa strain PA51673, CF clinical isolate (38). In contrast, expression of other MMPs increased modestly (MMP3, MMP12), if at all (MMP28), or transiently (MMP7, MMP8) (Supplemental Fig. 1A). MMP8, which is produced by many cell types and is particularly abundant in neutrophils, quickly increased and then fell (Supplemental Fig. 1B), likely reflecting the rapid influx and efflux of neutrophils in this model (11). Thus, among MMPs, MMP10 revealed a potent and sustained response to infection.

Whereas infection with P. aeruginosa led to no mortality in wild-type mice, half of Mmp10^−/− mice had died by 48 h after P. aeruginosa exposure (Fig. 2B). Among the survivors, Mmp10^−/− mice were more susceptible to infection than were wild-type mice as manifested by a reproducible and significant delay in their ability to regain body weight (Fig. 2C). Lethality and delayed weight recovery in Mmp10^−/− mice coincided with the marked upswing in Mmp10 expression in wild-type mice (Fig. 2A). Additionally, we observed greater mortality (Supplemental Fig. 1C) and delayed weight gain (data not shown) in Mmp10^tm1Lex/Mmucd mice generated with embryonic stem cells in which the Mmp10 locus was targeted by a gene-trap approach. Thus, the phenotypes we observed in our Mmp10^−/− mice were not due to an off-target artifact. These data indicate that MMP10 serves a critical and protective role in the host response to infection.

Lethality and morbidity in Mmp10^−/− mice were not due to impaired bacterial clearance. By 48 h, both wild-type and Mmp10^−/− mice had effectively cleared the infection (Fig. 2D). Additionally, bacteremia, as measured by the recovery of live bacteria from spleen (Fig. 2E), and spleen weights (data not shown) before and during infection did not differ between genotypes. Furthermore, infection-induced increase in lung permeability, a marker of acute lung injury, also did not differ between genotypes (Supplemental Fig. 1D). We did not see any indication (edema, hemorrhage, inflammation) that other tissues (liver, intestines) were involved (data not shown).

Whereas mortality of Mmp10^−/− mice was more pronounced when P. aeruginosa was administered by aerosolization (≥50%; Fig. 2B, Supplemental Fig. 1C) compared with oropharyngeal aspiration (≤14%; n > 90 mice/genotype from several experiments; data not shown), morbidity as gauged by degree of weight loss and delay in weight recovery was similar by either method of P. aeruginosa infection. Thus, to minimize a potential survival bias, our subsequent in vivo data were obtained with mice infected by oropharyngeal aspiration.

Because lethality in Mmp10^−/− mice was not associated with impaired bacterial clearance, we predicted that Mmp10^−/− mice died due to altered or excessive inflammation. We examined BAL and lung sections at various times after infection. Consistent with the similar rate of bacterial clearance (Fig. 2D), neutrophils, which are critical for clearance of P. aeruginosa, were the predominant cell in BAL after infection (>95% of total cells), and their numbers did not differ between genotypes (Fig. 3B). Lymphocyte numbers, which increased in response to infection, also did not differ significantly between genotypes (data not shown). In contrast, we detected 3-fold more macrophages in BAL (Fig. 3C) at 48 h postinfection, coincident with the upswing in MMP10 expression (Fig. 2A), and even more in tissue sections from Mmp10^−/− mice (Fig. 3E). At 4 and 24 h postinfection, before the significant increase in MMP10 expression, the number of macrophages in BAL did not differ between genotypes (Fig. 3C). Flow cytometry of CD45⁺ cells from nonlavaged lung demonstrated that numbers of alveolar (F4/80⁺CD11c⁺Siglec-F⁺) and interstitial (F4/80⁺CD11b⁺CD11c⁺MHC II⁺Siglec-F⁻) macrophages did not differ between naive wild-type and Mmp10^−/− mice and that the influx of macrophages into lungs of both genotypes was predictably due to infiltrated cells (F4/80⁺Ly6C⁺CD11b⁺CD11c⁺MHC II⁻Siglec-F⁻; data not shown). Similarly, we observed a 2-fold increase in BAL macrophages in Mmp10^tm1Lex/Mmucd mice after infection (data not shown).

The number and differential of circulating leukocytes did not differ between wild-type and Mmp10^−/− mice (data not shown). Additionally, we reported no difference in the ability of wild-type and Mmp10^−/− macrophages to migrate toward serum or other chemotactic stimuli (35), indicating that MMP10 does not affect the migratory machinery of macrophages. We also assessed the expression of various chemokines and cytokines that could affect macrophage influx or that have proinflammatory activity. We found no differences in the levels for TNF-α, CXCL1/2, IL-6, or IL-17 in BAL or lung homogenates (data not shown). In contrast, we found elevated expression of CCL2/MCP1, a potent macrophage chemokine (47, 48), in Mmp10^−/− lungs (see Supplemental Fig. 3A), a finding that was mirrored in isolated macrophages (see below).

Macrophages that influx into the lung, or any tissue, in response to infection or injury are distinct from the resident population (49, 50). We found that both resident and recruited macrophages expressed MMP10 at fairly similar levels after infection (data not shown). We then assessed whether macrophages in culture respond as these cells did in vivo. Consistent with our in vivo data, Mmp10 mRNA was not expressed by alveolar macrophages (Fig. 4B) or bone marrow (Fig. 4C) freshly isolated from naive wild-type mice. Upon CSF-1–mediated differentiation into macrophages (M0), MMP10 was induced in bone BMDM (Fig. 4C, Supplemental Fig. 2A). MMP10 was not induced by marrow cells cultured with GM-CSF (Supplemental Fig. 2A). As GM-CSF promotes differentiation into dendritic cells (5), these negative data indicate further that macrophages are the source of MMP10. Expression of MMP10 by both alveolar macrophages and BMDM was stimulated with LPS and even more so by exposure to live P. aeruginosa (Fig. 4B, 4C), reaching levels that were quite similar between cell types and comparable to the levels detected for Mmp10 mRNA in lung in response to airway infection (see Fig. 2A). Taken together, these findings indicate that both resident alveolar and infiltrated macrophages express similar levels of MMP10 in response to infection.

We performed adoptive transfer experiments to determine whether wild-type macrophages could ameliorate morbidity in infected Mmp10^−/− mice. Indeed, adoptive transfer of wild-type BMDM 24 h postinfection with P. aeruginosa resulted in significantly more rapid recovery of weight loss (used as a surrogate for morbidity) comparable to that seen in wild-type mice (Fig. 5). Mmp10^−/− BMDM did not affect weight loss in either wild-type or Mmp10^−/− recipients (Fig. 5). These findings demonstrate that macrophages are the key cells mediating the MMP10-dependent phenotypes in response to P. aeruginosa lung infection.

We assessed whether MMP10 influences the state of macrophage activation by measuring the in vivo expression of M1 and M2 markers (5). Compared to levels in wild-type mice, we found elevated expression of mRNAs for the M1 markers IFN-γ, inducible NO synthase (iNOS), and CCL2 (Supplemental Fig. 3A) and reduced expression levels of M2 markers arginase-1 and IL-10 in infected Mmp10^−/− lungs (Supplemental Fig. 3C). In contrast, IL-12a, an M1 marker, was reduced in Mmp10^−/− lungs (Supplemental Fig. 3A), and protein levels of TNF-α and IL-1β did not differ between genotypes (Supplemental Fig. 3B). Additionally, as we found in a skin wound model (35), expression of the M2 marker resistin-like α (Retnla, FIZZ1) was not affected by MMP10 (Supplemental Fig. 3D). Overall, the in vivo phenotypes we saw in Mmp10^−/− mice, as summarized in Table I, support the idea that MMP10 mitigates the proinflammatory activity of macrophages.

We next assessed whether MMP10 plays a cell-autonomous role in regulating macrophage polarization. We treated wild-type and Mmp10^−/− M0 BMDM with live P. aeruginosa to stimulate M1 differentiation or with IL-4 and IL-13 for M2 differentiation and assessed expression of various markers. Expression of MMP10 by M0 macrophages was stimulated by M1 agonists (Fig. 4C, Supplemental Fig. 2B) but was not further elevated by M2 agonists (Supplemental Fig. 2B), consistent with findings in human monocytes (51). In M0- and M2-polarized BMDM, we saw no difference in the expression of M1 and M2 markers between wild-type and Mmp10^−/− BMDM (Supplemental Fig. 4, Table II), with one notable exception. IFN-γ expression was elevated ∼10-fold in Mmp10^−/− M0 BMDM compared with wild-type M0 cells (Supplemental Fig. 4A), suggesting that in the absence of MMP10 macrophages are primed toward a proinflammatory phenotype.

We did observe significant MMP10-dependent effects in M1-polarized macrophages (Table II). In particular, the induction of CCL2 in response to P. aeruginosa was markedly greater in Mmp10^−/− alveolar macrophages and BMDM compared with wild-type cells (Fig. 6). Additionally, expression of mRNAs for the M1 markers iNOS, IFN-γ, and CCL5 and the ability to phagocytize E. coli particles, an M1 property, were significantly elevated in M1-polarized Mmp10^−/− macrophages (Supplemental Fig. 4A). Expression of IL-6 did not differ between genotypes (Supplemental Fig. 4B, Table I). Furthermore, expression of the M2 markers arginase-1 and IL-10 was stimulated in P. aeruginosa–treated BMDM, a response that was significantly diminished in Mmp10^−/− BMDM (Supplemental Fig. 4C). CD206 (mannose receptor) was downregulated in M1-polarized wild-type BMDM and significantly more so in Mmp10^−/− BMDM (Supplemental Fig. 4C, Table I). The levels of other M2 markers, including CCR2 (Supplemental Fig. 4D) and resistin-like α (35), did not differ between M1-polarized wild-type and Mmp10^−/− BMDM. Collectively, these data indicate that MMP10 functions to moderate several characteristic responses of M1 macrophage activation.

To gain insight into potential mechanisms by which MMP10 influences the macrophage response to bacterial infection, we performed global gene expression analysis. Wild-type and Mmp10^−/− BMDM were exposed to P. aeruginosa for 1 h, and cells were harvested at 6 and 24 h later to assess acute and persistent gene expression changes. We identified many genes differentially expressed (q < 0.01) compared with untreated controls at 6 h postinfection in both wild-type and Mmp10^−/− BMDM. As expected, P. aeruginosa infection caused a robust transcriptional response in both wild-type (12,672 genes) and Mmp10^−/− (15,164 genes) BMDM. The vast majority of differentially expressed genes (n = 10,286) were common between wild-type and Mmp10^−/− BMDM (Fig. 7A). Moreover, the direction and pattern of differential expression of the common genes were similar in wild-type and Mmp10^−/− cells, indicating that P. aeruginosa infection was the dominant cause of early transcriptional changes (Fig. 7B). Functional enrichment analysis of the smaller subsets of genes that were differentially expressed only in wild-type or Mmp10^−/− at 6 h did not reveal coherent phenotypes (data not shown).

Because gene expression is a temporally dynamic process, we postulated that the role of Mmp10 in regulating transcriptional programs induced by bacterial challenge may be revealed by examining a later time point. Thus, we focused on the large set of common differentially expressed genes at 6 h and assessed whether their levels remained significantly altered at 24 h postinfection. Of the 10,286 common genes, 5,609 remained differentially expressed 24 h postinfection (Fig. 7C). Notably, we observed that most (3981) genes that remained differentially expressed at the 24 h time point were unique to Mmp10^−/− BMDM, implying that MMP10 plays a distinct role in moderating the postinfection transcriptional response. Of the 3981 genes that remained differentially expressed only in Mmp10^−/− BMDM, 1386 were upregulated and 2505 were downregulated at 24 h. We applied functional analysis using WebGestalt to identify biological processes enriched within the subset of persistently upregulated genes (Fig. 7D). We found that the absence of Mmp10 in P. aeruginosa–exposed BMDM was associated with persistent activation of a wide repertoire of pathways mapping to immunity, inflammation, apoptosis, and cell signaling. In contrast, functional analysis of the persistently downregulated genes showed enrichment of processes involved in metabolism, transcription, and translation.

In this study, we report on the expression and function of macrophage MMP10 in the host response to infection. Our findings indicate that MMP10 is not expressed in healthy, unchallenged human and mouse lung but is markedly induced in both resident and infiltrated macrophages following infection with P. aeruginosa. With a combination of in vivo and primary macrophage models, we demonstrate that MMP10 protects mice from morbidity and mortality induced by acute P. aeruginosa lung infection. By assessing the expression of various markers and overall gene expression, our findings indicate that MMP10 functions to drive the activation of macrophages from proinflammatory (M1-like) cells into immunosuppressive (M2-like) cells. Our findings complement observations we reported with others showing greater mortality and delayed weight gain in Mmp10^−/− mice following acute liver injury (52) and more macrophages biased toward M1 activation and more tissue injury in Mmp10^−/− mice subjected to acute colon injury (33). Furthermore, Mmp10^−/− mice are much more susceptible to the lethal effects of acute lung injury caused by bleomycin toxicity than are wild-type animals (T.C. Vandivort, T.P. Birkland, and W.C. Parks, unpublished observations). Thus, across organ systems, MMP10 appears to play a beneficial role in mitigating the deleterious responses to injury and infection by promoting the conversion of proinflammatory macrophages toward immunosuppressive cells. Our findings support the general concept that in response to acute injury or infection macrophage activation evolves from potentially detrimental M1-biased cells to suppressive M2-biased macrophages (6, 39, 50).

Although more macrophages emigrated into Mmp10^−/− lungs in response to infection, we found no difference in the ability of wild-type and Mmp10^−/− macrophages to migrate toward serum or wound-tissue homogenate (35), indicating that MMP10 does not affect the migratory machinery of macrophages. With others (53), we reported that Mmp10^−/− macrophages have an impaired ability to migrate over fibronectin and to invade into Matrigel. Although these findings are seemingly opposed to our in vivo observations, there are two important caveats with these in vitro studies. First, it is not clear whether macrophages migrate over fibronectin on their way into and through the interstitial space. Although fibronectin would be present in the area, it appears that other matrix components, particularly versican and hyaluronan (54), serve as the interstitial substrata that macrophages follow as they migrate through tissue (55, 56). Second, Matrigel is a highly dense material that does not mirror the more porous architecture of interstitial matrix (57, 58). In fact, Matrigel does not contain interstitial matrix components; it is comprised of misassembled basement membrane components (57). Thus, unlike how they move through the interstitium in vivo, macrophages may require proteases to burrow through a thick plug of Matrigel in a culture dish. As we reported that MMP10 activates matrix-remodeling programs in macrophages (35, 59), Mmp10^−/− cells may not possess the enzymes needed to invade through the artificial setting of Matrigel. We conclude that MMP10 does not affect macrophage migration per se in physiologic settings.

An open question is whether increased morbidity and mortality in Mmp10^−/− animals was due to more macrophage numbers, to a bias toward M1 activation, or a combination of these two processes. We propose that altered activation of both resident and infiltrated macrophages toward an M1, proinflammatory state led to worse outcomes for infected Mmp10^−/− mice and this deleterious effect was augmented by the presence of more M1-biased cells. Consistent with this conclusion, Johnston et al. (39) reported that the macrophage phenotypes begin to evolve toward an M2 state ∼48 h after lung infection with P. aeruginosa, and D’Alessio et al. (60) recently demonstrated that therapeutically promoting macrophage differentiation toward M2 cells (via instillation of IL-4) markedly reduces lethality and morbidity in response to acute P. aeruginosa infection. In Mmp10^−/− mice and macrophages, we found increased expression levels of M1 markers and decreased levels of M2 markers (Tables I, II), such as the increased and prolonged expression of CCL2 (MCP-1), which was also seen in injured Mmp10^−/− colons (33). CCL2, a potent macrophage chemokine, is expressed predominantly by resident lung macrophages (47, 48), and overexpression of this factor may account for the excess macrophage inflammation seen in Mmp10^−/− mice. The diminished expression of IL-10 and Arg1, both M2 markers, in infected Mmp10^−/− lungs and BMDM provides further evidence that MMP10 broadly affects macrophage polarity to moderate inflammation.

Neutrophil influx, which accounted for ∼95% of the cells recovered from BAL, and the levels of CXCL1 and CXCL2, critical neutrophil chemokines, did not differ between wild-type and Mmp10^−/− mice in response to infection. Similarly, Koller et al. (33) found no difference in neutrophil numbers between wild-type and Mmp10^−/− mice during the first 14 d after colon injury. However, they did find that elevated neutrophil counts persisted in Mmp10^−/− mice at 21 d after injury, at which time their numbers had dropped in wild-type colons. Although sustained neutrophil influx may have been secondary to worse injury of Mmp10^−/− colons, it is also possible that a bias toward M1 macrophages maintained neutrophil influx that, in turn, led to more tissue damage. In our lung infection model, the lack of difference in neutrophil numbers and indices of acute lung injury between genotypes indicates that neutrophils did not contribute to worse outcomes in Mmp10^−/− mice. A potential caveat to this conclusion is whether MMP10 influences the state of neutrophil activation. For example, in other studies, we found that MMP7 shedding of syndecan-1/CXCL1 complexes from injured lung epithelium permits both neutrophil migration through the mucosal barrier and their subsequent activation (20, 61). Without this processing mechanism, Mmp7^−/− mice are markedly protected from the lethal effects of neutrophil-mediated lethality in response to acute lung injury (20). However, we propose that the similar rate of bacterial clearance in wild-type and Mmp10^−/− mice does not support an MMP10-dependent role in neutrophil activation.

Although MMP10 can also be expressed by injured/infected epithelium, including in lung (38), we do not think that the enzyme produced by these cells contributed to the key phenotypes and responses we report in the present study. Indeed, adoptive transfer of wild-type macrophages into Mmp10^−/− recipients reduced postinfection morbidity to levels seen in wild-types. Similarly, using bone marrow transplantation, Koller et al. (33) concluded that the lack of MMP10 in macrophages led to worse tissue damage in a model of acute colon injury. Furthermore, the macrophage polarization changes we observed in vivo (Table I) were largely duplicated with isolated macrophages in culture (Table II). These findings support the conclusion that MMP10 is a cell-autonomous regulator of macrophage activation. In contrast, the function of epithelial MMP10 remains unanswered. In skin wound models, we found no re-epithelialization phenotype in Mmp10^−/− mice (35) or in mice overexpressing active MMP-10 in basal keratinocytes (32).

We recently contributed to a multicenter genome-wide association study that identified MMP10 as a candidate gene for chronic obstructive pulmonary disease (59). Using a model of chronic (6 mo) exposure to cigarette smoke, we found that Mmp10^−/− mice are fully resistant to the development of emphysema. Additionally, MMP10 is produced by macrophages from human smokers with emphysema (62) and is one of two genes whose expression correlates with reduced lung function (i.e., forced expiratory volume in 1 s) in smokers (63). These findings indicate that macrophage MMP10 contributes to disease progression in emphysema, which is seemingly opposed to the protective role for this MMP in acute infection we reported in the present study. However, there are important differences between these models, especially with respect to macrophage biology. Macrophages that function early in inflammation are functionally distinct from those that function late in inflammation or in a persistent inflammatory response (4, 6, 50, 64, 65), and this concept has been demonstrated in models of liver (66), vascular (67), cardiac (68), and kidney disease (69). Additionally, whereas acute infection and injury bias toward an M1 phenotype (4), cigarette smoke promotes expansion of M2 macrophages (70). Macrophages are considered to be the destructive cell in emphysema (71, 72), and our studies in skin wounds indicate that MMP10 promotes expression of matrix-degrading proteinases in M2 macrophages (35). Thus, in an acute setting, MMP10 is beneficial by moderating the proinflammatory activity of M1-biased macrophages and by stimulating the ability of M2-biased macrophages to remodel scar tissue. However, in a chronic setting, MMP10-driven matrix remodeling could be excessive and detrimental, as suggested in our smoke-exposure studies. Still, an important and shared conclusion among these models is that MMP10 functions to control macrophage behavior.

Significant differences in expression of M1 and M2 markers were only observed between infected (M1-biased) wild-type and Mmp10^−/− macrophages but not between M0 or M2 cells (Table II). Consistent with these data, transcriptomic analysis did not reveal significant enrichment of any inflammation-related pathways in genes differentially expressed between M0 wild-type and Mmp10^−/− BMDM (data not shown). Thus, the effects of MMP10, at least in terms of polarization, are more significant in M1-activated macrophages. However, MMP10 does shape some functional properties of M2-biased macrophages, such as their ability to clear collagen (35). The P. aeruginosa–induced transcriptional changes between wild-type and Mmp10^−/− BMDM were quite similar at 6 h postinfection but diverged significantly at 24 h. Nearly 4000 of the genes that were differentially expressed in both genotypes at 6 h remained significantly changed only in Mmp10^−/− BMDM at 24 h. These data indicate that MMP10 functions to broadly moderate or quench the temporal transcriptional response induced by bacteria in infiltrating macrophages. The genes that remained persistently upregulated only in Mmp10^−/− macrophages at 24 h were significantly enriched for inflammation and immune-related pathways, consistent with our observations in the infected Mmp10^−/− lung.

In summary, our data demonstrate a novel role for MMP10 in moderating lung inflammation by altering the balance of M1/M2 transcripts and broadly suppressing immune-related pathways in macrophages. Importantly, the patterns of MMP10 expression are conserved between human and mouse lung and macrophages. The critical functions of MMP10 in lung inflammation are largely attributable to infiltrated macrophages and may therefore be relevant in other organs and inflammatory conditions. Given the importance of macrophage polarization in the initiation and resolution of inflammation, further study of MMP10 function is warranted in other disease models, especially those involving bacterial infection. Although beyond the scope of this study, identification of the physiologic substrates cleaved by MMP10 will be essential to determine the mechanisms by which MMP10 regulates macrophage polarization. Schlage et al. (73) have identified multiple MMP10 substrates in primary mouse fibroblasts, but we predict that macrophage-derived MMP10 may cleave a limited and distinct set of substrates, some of which may only be expressed in an inflamed microenvironment. Determining the relevant MMP10 cleavage events in the lung is a significant challenge but a priority for our future work.

We thank Drs. Ann Chen and Gail Deutsch for help acquiring human lung specimens, Ying Wang for maintaining mouse strains, Dr. Elaine Raines for providing Mac-2 Ab, and Fred Farin, Theo Bammler, and Jesse Thai for assistance with microarray experimental design and processing.

The authors have no financial conflicts of interest.