MMP-10 Regulates Collagenolytic Activity of Alternatively Activated Resident Macrophages Maryam G

See related commentary on pg 2350 ORIGINAL ARTICLE

MMP-10 Regulates Collagenolytic Activity of Alternatively Activated Resident Macrophages Maryam G. Rohani1,2,3, Ryan S. McMahan1,4, Maria V. Razumova5, Angie L. Hertz1,2, Maryelise Cieslewicz5, Suzie H. Pun5, Michael Regnier5, Ying Wang1,2,3, Timothy P. Birkland1,2 and William C. Parks1,2,3

Matrix metalloproteinase-10 (MMP-10) is expressed by macrophages and epithelium in response to injury, but its functions in wound repair are unknown. We observed increased collagen deposition and skin stiffness in – – Mmp10 / wounds, with no difference in collagen expression or reepithelialization. Increased collagen – – deposition in Mmp10 / wounds was accompanied by less collagenolytic activity and reduced expression of speciﬁc metallocollagenases, particularly MMP-8 and MMP-13, where MMP-13 was the key collagenase. Ablation and adoptive transfer approaches and cell-based models demonstrated that the MMP-10-dependent collagenolytic activity was a product of alternatively activated (M2) resident macrophages. These data demonstrate a critical role for macrophage MMP-10 in controlling the tissue remodeling activity of macrophages and moderating scar formation during wound repair. Journal of Investigative Dermatology (2015) 135, 2377–2384; doi:10.1038/jid.2015.167; published online 28 May 2015

INTRODUCTION can also contribute to resolution of scarring and fibrosis Scar formation, the deposition of fibrous connective tissue indirectly by shaping the proteolytic phenotype of cells within a wound bed, is a normal consequence of injury and (Giannandrea and Parks, 2014). provides temporary strength to damaged tissue. However, In response to injury, stromelysin-2 (MMP-10) is induced by excessive, persistent scarring can lead to a variety of clinical epithelial cells and macrophages in many organs, including problems, such as keloids, hypertrophic scars, and severe the skin (Saarialho-Kere et al., 1994; Kassim et al., 2007; life-threatening fibrosis of lung, kidney, and liver. In these Koller et al., 2012). In collaboration with others, we reported conditions, excess scarring is seemingly due to both an that overexpression of active MMP-10 in basal keratinocytes overabundance of collagen production and inadequate did not affect reepithelialization and led to only a minor resolution (McKleroy et al., 2013). phenotype in basement membrane deposition (Krampert Although far from being fully understood, resolution of et al., 2004). The function of MMP-10 during wound repair fi scarring and brosis appears to be the responsibility of remains to be elucidated. macrophages and, in particular, alternatively activated (M2) Similar to most tissues, the skin contains a population of macrophages (Atabai et al., 2009; Madsen et al., 2013). resident macrophages, and following injury additional macro- Current models indicate that matrix turnover involves two phages influx into the wound bed (Jameson et al., 2005). sequential steps: limited extracellular proteolysis followed by Macrophages have distinct roles during different phases of uptake and lysosomal degradation (Madsen et al., 2011). For wound repair, but it is not clear how these roles are split the first step, some matrix metalloproteinases (MMPs) cleave between resident and recruited cells. Macrophages contribute the large collagen fibrils into fragments that can be fi endocytosed and degraded intracellularly (Lee et al., 2006; to scar formation by producing pro brotic cytokines that fi Madsen et al., 2007; Wagenaar-Miller et al., 2007). MMPs activate resident broblasts and pericytes into extracellular matrix–producing myofibroblasts (Duffield, 2003; Wynn and

1Center for Lung Biology, University of Washington, Seattle, Washington, USA; Barron, 2010). Depletion of macrophages soon after injury 2Department of Medicine, University of Washington, Seattle, Washington, USA; impairs formation of vascularized granulation tissue and 3 Department of Medicine, Cedars-Sinai Medical Center, Los Angeles, California, delays repair (Goren et al., 2009; Lucas et al., 2010). On USA; 4Department of Environmental and Occupational Health Sciences, University of Washington, Seattle, Washington, USA and 5Department of the other hand, as demonstrated in liver injury, depletion Bioengineering, University of Washington, Seattle, Washington, USA of macrophages during later phases leads to an inability to Correspondence: Maryam G. Rohani, Department of Medicine, Cedars-Sinai resolve fibrotic matrix (Duffield et al., 2005; Ramachandran Medical Center, 8700 Beverly Boulevard, AHSP A9403, Los Angeles, et al., 2012). Early-phase macrophages are predominately California 90048, USA. E-mail: [email protected] classically activated proinflammatory cells (M1 biased), Abbreviations: BMDM, bone marrow–derived macrophage; MMP, matrix metalloproteinase whereas the resolution-phase cells tend to be alternatively Received 4 January 2015; revised 8 April 2015; accepted 20 April 2015; activated, remodeling-competent macrophages (M2 biased) accepted article preview online 30 April 2015; published online 28 May 2015 (Mosser and Edwards, 2008; Lichtnekert et al., 2013).

Here we report that scarring was increased in wounded –/– 400 skin of Mmp10 mice. We found that excess extracellular * matrix was not associated with increased expression of collagen but rather with reduced metallocollagenolytic 300 g / wound)

activity of resident macrophages. As MMP-10 lacks collage- µ WT nolytic activity, these ﬁndings indicate that this proteinase 200 regulates the extracellular matrix degradative potential of Mmp10 –/– macrophages. 100

100 µm

RESULTS ( Hydroxyproline 0 MMP-10 does not affect reepithelialization WT Mmp10 –/– Using RNA from uninjured wild-type mouse skin, we detected a Ct range for Mmp10 mRNA between 36 and 37 0.5 1.5 WT (Supplementary Figure S1 online), indicating essentially no or Mmp10 –/– * very low levels of expression. Similar to that reported earlier 0.4 –1 1.0 (Krampert et al., 2004), Mmp10 was induced by day 3 after 0.3 wounding, remained elevated at day 8, and began to decline by day 12 (Supplementary Figure S1 online). To examine the 0.2 0.5 role of MMP-10 in reepithelialization, we compared the rate – – mN mm Force, of wound closure between wild-type and Mmp10 / mice 0.1 Collagen SA / total SA (Supplementary Figure S2 online). However, wound closure 0.0 0.0 rates (i.e., slope of the lines) did not differ between wild-type –/– – – WT Mmp10 0 5 10 15 20 and Mmp10 / wounds, and both visual (Supplementary % Length change Figure S2 online) and histologic (not shown) analyses showed complete reepithelialization in both genotypes by day 8 after Figure 1. Matrix metalloproteinase-10 (MMP-10) moderates collagen injury. accumulation at the site of injury. (a) Uniform 8 mm wound samples were collected on day 12 after injury, and hydroxyproline was quantified and normalized per wound sample (n = 8 mice per genotype); *Po0.05. MMP-10 moderates collagen deposition (b) Representative images of day 12 wounds stained with Masson’s trichrome. During wound healing, the granulation tissue is replaced with Scale bar = 100 μm. (c) Quantification of collagen density in Masson’s newly deposited type I collagen forming a scar. Net collagen trichrome–stained samples. Data are the ratio of collagen-positive (blue) area deposition is the sum of syntheses minus turnover. In skin to total surface area (SA). Each point represents samples from an individual wounds, collagen production begins 3–5 days after wounding mouse. (d) Tissue stiffness of wounds was measured using 12-day wound samples. Data are presented as the amount of force change per and peaks around day 10 (Wu et al., 2003). Between days 3 – – percent of length change. (n = 4 WT, 3 Mmp10 / ), *Po0.05. WT, wild type. and 8, we observed a trend for more deposition of collagen in – – Mmp10 / wounds compared with wild types (not shown) that led to significantly elevated levels of accumulated collagen deposition by day 12 in null mice (Figure 1a–c). Consistent stimulated expression of α-smooth muscle actin, or collagen with more collagen deposition, we found, using an estab- uptake between primary dermal fibroblasts from wild-type or – – lished method that measures the dynamic mechanical Mmp10 / mice (Supplementary Figure S4 online). strength of skin wounds (Jørgensen et al., 1993), that day 12 – – Mmp10 / wounds were significantly stiffer than wild-type MMP-10 promotes collagen degradation wounds (Figure 1d). These data indicate that MMP-10 The above data indicate that increased deposition in – – moderates scar formation. In intact skin, collagen levels did Mmp10 / wounds was due to impaired turnover. In defined, not differ significantly between wild-type (155 ± 26.8 μg per in vitro degradation assays, MMP-10 does not cleave native – – biopsy) and Mmp10 / (165.7 ± 26.0) mice. type I collagen (Sternlicht and Werb, 2001; Parks et al., 2004). Thus, we assessed whether MMP-10 influences the overall MMP-10 does not affect collagen mRNA levels collagenolytic activity in wounds. For this, wound samples We assessed whether increased collagen deposition in were incubated with quenched-fluorescent type I collagen – – Mmp10 / wounds was due to increased synthesis. Basal (DQ collagen), and release of fluorescence was measured as levels of Col1a1 mRNA and its increase and decline in res- an indicator of total collagenase activity. Indeed, significantly – – ponse to injury did not differ between genotypes (Supple- less collagenase activity was released from Mmp10 / wounds mentary Figure S3a online). In addition, we found no compared with wild-type samples (Figure 2a). significant difference in α-smooth muscle actin protein levels, We used biopsies of intact skin as an ex vivo injury model a marker of myofibroblasts, the major cell source of collagen (Dumin et al., 2001). In wild-type explants, induction of – – synthesis in scars, between wild-type and Mmp10 / wounds Mmp10 mRNA was detected at 72 hours after biopsy (Supple- at days 3 (not shown), 8, or 12 (Supplementary Figure S3b mentary Figure S5a online), similar to the in vivo pattern online). Furthermore, we found no difference in the prolifera- (Supplementary Figure S1 online). Using both DQ and native tion rate, the basal or transforming growth factor-β1– type I collagens, we found that significantly less collagenase

2378 Journal of Investigative Dermatology (2015), Volume 135 MG Rohani et al. MMP-10 Controls Scar Formation

– – activity was released from Mmp10 / explants compared with 20 6 20 * wild-type samples (Figure 2b and c). The levels of collagenase ** **** – – activity released from wild-type and Mmp10 / tissues and 15 15 ** 4 the differences between genotypes were comparable for both *** wound samples and explants (Figure 2a and b). Because 10 NS 10 recruited inﬂammatory cells are not present in explants, the 2 5 5 close similarity in the data between these models indicates

Collagenase activity that the MMP-10-dependent collagenase activity was a 0 0 0 product of cells residing in intact skin. Time(hours) 24 4872 24 48 72 3-D Collagen To identify the nature of the reduced collagenolytic activity, we treated wound samples with GM6001, a broad-spectrum metalloproteinase inhibitor. In the presence of GM6001, WT 40 Ctrl 25 * collagenase activity was reduced to the same level in both Mmp10 –/– rh-MMP-10 20 30 genotypes (Figure 2d), indicating that MMP-10 promoted the 15 expression and/or activity of a metallocollagenase(s). Addition 20 of active recombinant human MMP-10 increased the level of – – 10 collagenase activity released from Mmp10 / explants but had 10 5 no effect on the activity from wild-type samples (Figure 2e).

Collagenase activity ﬁ 0 0 Con rming its lack of collagenase activity, recombinant Ctrl GM6001 WT Mmp10 –/– human MMP10 did not release ﬂuorescence from DQ

– – collagen (not shown). Figure 2. Reduced collagenase activity in Mmp10 / wounds. (a) Day 8 wounds were incubated with DQ collagen for 72 hours. Collagen degradation was measured every 24 hours (n = 3 per genotype). (b) Collagenase activity MMP-10 controls expression of collagenolytic MMPs – – from explants of normal skin (n = 7 wild type (WT), 6 Mmp10 / ). (c) Skin These data indicate that MMP-10 moderates scar formation explants were seeded on FITC-conjugated type I collagen gels. Collagenase and promotes resolution by controlling metallocollagenase activity was measured in conditioned medium at 72 hours (n = 5 per activity. To identify the speciﬁc enzyme(s), we assessed genotype). (d) Explants were incubated with 25 μM GM6001, and collagenase – – the expression of mRNAs for Mmp2, 8, 9, 13, 14, and 16 activity was assessed 72 hours later (n = 4 WT, 3 Mmp10 / ). (e) Explants were − 1 that encode known or suspected collagenases (Shi et al., treated with recombinant human (rh)-MMP-10 (100 ng ml ) or vehicle (control (Ctrl)), and collagenase activity was assessed 72 hours later 2008; Sabeh et al., 2009). In day 3 wound samples, we (n = 5 per genotype). *Po0.05, **Po0.01, ***Po0.001, and ****Po0.0001. found reduced expression of Mmp8 and Mmp9 mRNAs in – – MMP-10, matrix metalloproteinase-10; NS, not signiﬁcant. Mmp10 / samples (Figure 3). In day 8 wound samples, the

Mmp2 Mmp8 Mmp9 6 2,000 50 **** *** 1,500 * 1,000 40 4 500 WT 30 Mmp10 –/–

RQ 60 20 2 40 10 20 0 0 0 0 3 8 12 18 0 3 8 12 18 0 3 8 12 18

Mmp13 Mmp14 Mmp16 1,000 **** 4 10 **** 800 8 3 600 6

RQ 2 400 4 1 200 2

0 0 0 0 3 8 12 18 0 3 8 12 18 0 3 8 12 18 Days after injury

Figure 3. Matrix metalloproteinase-10 (MMP-10) influences expression of collagenolytic MMPs. Expression of mRNAs for Mmp2, 8, 9, 13, 14, and 16 was quantified in wound samples harvested at the indicated days. Data shown are the ratios compared with nonwounded wild-type (WT) samples and normalized to Hprt (n ≥ 4 per genotype). *Po0.05, ***Po0.001, and ****Po0.0001. RQ, relative quantification.

www.jidonline.org 2379 MG Rohani et al. MMP-10 Controls Scar Formation

fi levels of Mmp2, Mmp13, and Mmp14 mRNAs were signi - PBS ScrM2pepKLA –/– 40 25 ** cantly lower in Mmp10 mice (Figure 3). Similarly, ** Clodronate M2pepKLA in explants, we found that Mmp8, Mmp9, and Mmp13 20 NS – – 30 NS mRNAs were significantly reduced in Mmp10 / samples 15 (Supplementary Figure S5 online). These results suggest that 20 fi reduced expression of speci c collagenolytic MMPs con- 10 –/– tributed to increased collagen accumulation in Mmp10 10 5

wounds. Collagenase activity 0 0 MMP-10 does not impact macrophage influx into skin wounds WT Mmp10 –/– WT Mmp10 –/– To analyze the cellular mechanism by which MMP-10 Figure 4. Matrix metalloproteinase-10 (MMP-10) regulates collagenase – – governs collagen cleavage, we focused on macrophages. activity of tissue macrophages. Wild-type (WT) and Mmp10 / skin explants Although the similar levels of activity released from wound were cultured in the upper chambers of a 24-transwell plate containing biopsies and explants (Figure 2) suggested that MMP-10- (a) phosphate-buffered saline (PBS) or clodronate liposomes (n = 3 per dependent collagenolysis was a function of resident cells, genotype) or (b) the apoptosis-inducing peptide M2pepKLA or − impaired influx of macrophage could also have contributed to ScrM2peptideKLA (20 μgml 1; n = 5 per genotype). FITC-conjugated type I – – reduced collagenase activity in Mmp10 / wounds. However, collagen was added to the medium in the lower chamber, and collagenase Po fi we found no difference in the number of resident macro- activity was measured 72 hours later. ** 0.01. NS, not signi cant. – – phages between wild-type and Mmp10 / skin (Supplementary Figure S6a and b online) or in the total number of macrophages (resident plus recruited) in wild-type and subtype–specific expression patterns did not differ between – – – – Mmp10 / wounds (Supplementary Figure S6c and d online). wild-type and Mmp10 / cells (Supplementary Figure S8 Furthermore, we detected no difference in the ability of wild- online). However, we did find reduced expression of Mmp8 – – type and Mmp10 / macrophages to migrate toward serum and Mmp13 mRNAs and elevated levels of Mmp9 in – – (Supplementary Figure S6e online) or in macrophage chemo- Mmp10 / M2 macrophages (Figure 5a). These data agree – – tactic activity in wild-type and Mmp10 / wounds with the reduced expression of Mmp8 and Mmp13 in – – (Supplementary Figure S6f online). These data indicate that Mmp10 / wounds (Figure 3) and explants (Supplementary – – the reduction in collagenase activity in Mmp10 / wounds Figure S5b online). Antibody clearance demonstrated that was not due to fewer macrophages. MMP-13 accounted for the bulk of collagenase activity produced by wild-type M2 macrophages (Figure 5b). MMP-10 modulates collagenolytic activity of alternatively Furthermore, polarization to M2-like macrophages activated macrophages increased the collagenase activity released from wild-type To assess a functional role for macrophages, we depleted macrophages but had no effect on the activity released from – – these cells in explants using clodronate liposomes Mmp10 / macrophages (Figure 5c). Expression of Arg1 and (Supplementary Figure S7a online). Whereas depletion of Fizz1 increased markedly in M2-differentiated macrophages macrophages in wild-type explants significantly reduced total (Gordon, 2003), but this well-characterized M2 response was – – – – collagenase activity to the levels seen in Mmp10 / samples, not altered in Mmp10 / cells (Supplementary Figure S9 ablation did not further lower the reduced activity in online). These data indicate that the reduced levels of – – – – Mmp10 / tissue (Figure 4a). Skin-resident macrophages are collagenase activity released from Mmp10 / M2 macrophage – – categorized as alternatively activated (M2) cells (Davies et al., were independent of the ability of Mmp10 / M0 macro- 2013), and M2 macrophages are considered to be the cells phages to respond to the M2 agonists. responsible for tissue remodeling (Madsen et al., 2013). To demonstrate that reduced collagenolytic activity of – – Hence, we selectively ablated M2-like macrophages using the macrophages contributed to enhanced scarring in Mmp10 / M2pepKLA fusion peptide (Cieslewicz et al., 2013). The wound, we transferred BMDMs from wild-type MacGreen – – effectiveness of M2pepKLA was confirmed by reduced levels mice into Mmp10 / recipients on day 10 after injury. Tissues of MRC-1 (Supplementary Figure S7b online), an M2 were harvested 2 days later (day 12 after wounding). EGFP+ – – macrophage marker. Similar to the clodronate data, ablation macrophages were detected in both wild-type and Mmp10 / of M2 macrophages with M2pepKLA significantly decreased wounds (Supplementary Figure S10 online), and similar collagenase activity in wild-type explants but not in numbers of F4/80+ macrophages were detected in wounds – – Mmp10 / samples (Figure 4b). from both genotypes (Figure 6a and b). We found that the The macrophage ablation data indicated that MMP-10 influx of wild-type macrophages normalized the increased – – influenced the expression of collagenolytic MMPs in M2 collagen deposition seen in Mmp10 / wounds (Figure 6c). macrophages. To test this idea, we isolated bone marrow– derived macrophages (BMDMs) from wild-type and DISCUSSION – – Mmp10 / mice and assessed the expression of collagenolytic We report that macrophage-derived MMP-10 serves a MMPs in naive (M0) and M1- and M2-differentiated cells. beneficial role in tissue repair by promoting resolution of Although the expression levels of metallocollagenases scar tissue. Our findings indicate MMP-10 shaped scar differed between M0 and M1 macrophages, these cell deposition indirectly by controlling the collagenolytic activity

2380 Journal of Investigative Dermatology (2015), Volume 135 MG Rohani et al. MMP-10 Controls Scar Formation

WT 4 –/– Mmp10 **** 3 ** WT

2 RQ ** Mmp10 –/– 1

0 1.5 1.5

Mmp2 Mmp8 Mmp9 Mmp13 Mmp14 1.0 1.0

1.5 * 50 M0 0.5 0.5 * M2 45 F4/80 signal/area

1.0 Hydroxyproline (AU) 0.0 0.0 40 WT Mmp10 –/– WT Mmp10 –/– 0.5 Figure 6. Adoptive transfer of wild-type macrophages normalizes collagen 35 – – – – deposition in Mmp10 / wounds. Wild-type (WT) and Mmp10 / mice

Collagenase activity + Collagenase activity 6 – 0.0 30 received 7 × 10 GFP bone marrow derived macrophages (BMDMs) via –1 –/– µg ml 0 0 1 2 4 WT Mmp10 retro-orbital injection at day 10 after wounding. Wounds were harvested on 2.5 25 0.25 day 12. (a) Representative images of 12-day wound sections stained for F4/80. MMP-8 Ab MMP-13 Ab Scale bar = 500 μm (left) and 60 μm (right). (b) The total area of F4/80-positive signal was quantified and normalized to total tissue surface area (n = 5 per fl Figure 5. Matrix metalloproteinase-10 (MMP-10) in uences expression of genotype). (c) Hydroxyproline was quantified, normalized per wound sample, collagenolytic MMPs and activity in M2 macrophages. (a) Total RNA was and expressed relative to wild-type samples (n = 5 mice per genotype). –/– – isolated from M2-polarized wild-type and Mmp10 bone marrow derived AU, arbitrary unit; MMP-10, matrix metalloproteinase-10. macrophages (BMDMs) and analyzed by quantitative real-time PCR (qRT-PCR). Data are presented as fold change relative to naive BMDMs from wild-type (WT) mice. (b) WT skin explants were cultured in the upper chambers of 24-transwell plates. DQ collagen was added to the lower (Uchinami et al., 2006). Regarding skin wounds, disparate fi –/– chamber containing antibody (Ab) against MMP-8 or antibody against ndings have been reported for phenotypes in Mmp13 MMP-13. Collagenase activity was measured 72 hours later. Values are mice. Whereas Hattori et al. (2009) concluded that MMP-13 presented as the ratio of untreated samples; n = 3. (c) Naive and M2-polarized promotes closure of skin wounds, Hartenstein et al. (2006) BMDMs were incubated with DQ collagen for 6 hours. Supernatants were found no difference in reepithelialization between wild-type –/– o o – – assayed for collagenase activity (n = 4 WT, 5 Mmp10 ). *P 0.05, **P 0.01, and Mmp13 / mice. Although neither skin wound study o fi ****P 0.0001. RQ, relative quanti cation. assessed scarring directly, Hattori et al. (2009) reported a reduction in both wound contraction and myofibroblast – – numbers in Mmp13 / mice, findings that would suggest of M2-biased resident macrophages without affecting col- that MMP-13 has profibrotic activities too. As MMP-13 is lagen production by fibroblasts/myofibroblasts. More specifi- expressed by many cells types, it is likely that this MMP cally, we found that MMP-10, which is not a collagenase, can affect multiple, distinct processes within and among promoted the expression of MMP-8 and MMP-13, two models. metallocollagenases released by macrophages. However, Our findings indicate that MMP-10 controls collagen we found that essentially all collagenolytic activity was due turnover by macrophages by promoting the expression of to MMP-13, suggesting that macrophage-derived MMP-8 specific metallocollagenases, primarily MMP-13. As MMP-13 – – functions in other processes. Indeed, data supporting that levels were reduced but not absent in Mmp10 / macro- MMP-8 functions as a collagenase are not compelling. Both phages, the residual enzyme likely contributed to the – – overexpression and deletion of Mmp8 reduce fibrosis (Siller- collagenase activity in Mmp10 / samples and cells and Lopez et al., 2004; Garcia-Prieto et al., 2010; Craig et al., may explain why we did not observe any differences in 2013). granulation tissue formation as seen in mice with a complete – – Studies with Mmp13 / mice complement our conclusions. lack of MMP-13 (Toriseva et al., 2012). Complementary to fi fl The resolution of CCl4-induced liver brosis is slowed down our conclusion, the transcription factor ANKRD1 in uences – – in Mmp13 / mice (Fallowfield et al., 2007) and accelerated expression of Mmp13 and Mmp10 (Almodovar-Garcia et al., in mice overexpressing this collagenase (Yang et al., 2014), 2014), suggesting a link in the synthesis of these two MMPs. and both of these studies linked MMP-13 production to scar- In our studies, we used a model of normal skin repair, and associated macrophages. Furthermore, excess type I collagen we observed no differences in macrophage influx or in the – – is deposited in the bones of Mmp13 / mice (Inada et al., expression of α-smooth muscle actin and type I collagen (or 2004; Stickens et al., 2004). On the other hand, liver fibrosis other markers of fibroblast activation) between wild-type and – – – – due to bile duct ligation is reduced in Mmp13 / mice Mmp10 / mice. In contrast, macrophage influx, α-smooth

www.jidonline.org 2381 MG Rohani et al. MMP-10 Controls Scar Formation

muscle actin expression, and fibrosis are elevated in MATERIALS AND METHODS – – Mmp10 / mice in models of severe liver (Garcia-Irigoyen Animals – – et al., 2013), colon (Koller et al., 2012), and lung injury Mmp10 / mice (C57BL/6J) (Kassim et al., 2007) and wild-type (Parks et al., unpublished findings). Although an explanation littermates (male and female) were used for these studies. All for these disparate findings is not yet apparent, we propose procedures were approved by the Office of Animal Welfare, that the mechanisms by which MMP-10 affects fibrosis and University of Washington, and the institutional animal care and use macrophage influx are context dependent. Thus, the profound committee, Cedars-Sinai. proinflammatory responses in liver/colon/lung models may lead to MMP-10-dependent roles in macrophage recruitment Wounds and explants and activation that, in turn, stimulate myofibroblast differ- Mice were anesthetized, and their dorsal surface was shaved. The entiation and collagen production. In contrast, because the skin was pulled into a fold and laid on a plastic surface. Two full- inflammatory response is less profound in skin wounds, thickness wounds were created by running a sterile 5-mm punch MMP-10-dependent controls on macrophage influx could biopsy (Miltex, New York, NY) through the fold. Four wounds were – – be modest and, hence, not easily detected in Mmp10 / made per mouse and left uncovered. Mice were housed in individual tissues. cages. On indicated days after injury, wounds were excised with an Compared with the evidence supporting their roles in 8-mm biopsy punch for further analysis. promoting fibrosis (Pellicoro et al., 2014), there are limited data on how macrophages function during resolution. We RNA found that MMP-10 shapes the tissue-remodeling function Total RNA was isolated, and specific transcripts were quantified of resident skin M2-like macrophages (Davies et al., 2013) by quantitative real-time PCR using TaqMan FAM-labeled probes by regulating expression of MMP-13. Reduced levels of (Applied Biosystems, Foster City, CA) as previously described (Rohani this metallocollagenase were consistently seen among wound et al., 2014). biopsies (Figure 3), explants (Supplementary Figure S5b online), and M2-polarized cultured macrophages (Figure 5a). Hydroxyproline Reparative or M2-like macrophages are the dominant popula- Excised wounds were weighed and homogenized in deionized tion of macrophages during the resolution phase (Lichtnekert water (20 μl per mg tissue). Equal volumes of homogenates were mixed et al., 2013; Duffield, 2014). M0 macrophages possessed with 12 N HCl (1:1) and hydrolyzed for 3 hours at 120 °C. Hydroxypro- baseline collagenolytic activity (Figure 5c), and treatment with line was measured using a colorimetric assay (BD BioVision, Milpitas, M2 agonists led to a significant increase in the overall CA). Data are presented as total μg hydroxyproline/biopsies. collagenase activity released by wild-type macrophages but – – had no effect on the levels from Mmp10 / cells. These findings Histology support our central conclusion that MMP-10 controls the Sections (4 μm) of paraffin-embedded tissues were stained with collagenolytic activity of M2-biased macrophages. In agree- Masson’s trichrome or processed for immunohistochemistry with anti ment with our findings, Madsen et al. (2013) demonstrated that F4/80 (BM8, eBioscience, San Diego, CA). VisioPharm software M2-like macrophages are the principal cells executing (Hørsholm, Denmark) was used to quantify the total area of F4/80 collagen turnover in the skin. signal or trichrome staining, normalized to total surface area. Our findings indicated that M2-biased resident macrophages were responsible for scar resolution in excisional skin Biomechanics wounds. Of note, induction of Arg1 and Fizz1, the commonly Skin stiffness was measured as previously described (Rohani et al., used M2 markers, was not altered in IL-4/IL-13-treated 2014). Briefly, 12-day wound samples were dissected into strips and – – Mmp10 / macrophages, but a key function––tissue remodel- suspended on hooks attached to a force transducer (Aurora Scientific, ing––attributed to this type of macrophage was markedly model 400A, Aurora, ON, Canada) and a length controller (Aurora impaired. These findings add to the many caveats of the M1/ Scientific, model 308B). Tissue stiffness (dF/dL) was determined by M2 classification scheme and underscore that function is imposing change in length of the skin strip and measuring the more important in defining macrophage subtype than resultant change in force normalized per mm of sample width. On expression of a few markers (Murray et al., 2014). each strip of tissue, 4 steps of 5% length stretches were taken every Excess scarring (fibrosis) can affect quality of life, with loss 30 seconds. Force and length signals were analyzed using custom of tissue function, restriction of movement, and adverse LabView software (National Instruments, Austin, TX). psychological consequences (Brown et al., 2008). Although fibrotic conditions are oft considered to be irreversible, an Collagenase understanding of how macrophage activation is controlled Wound and explant samples were placed in the upper chambers of and how excess extracellular matrix is cleared will shed light transwells in 50 μl, and 1% (v/v) quenched fluorescein-conjugated on the means to manipulate these severe diseases. Results DQ type I collagen (1 μgml− 1; Life Technologies, Grand Island, NY) from our study demonstrate a role for MMP-10 in controlling was added to the bottom chambers in 500 μl of phenol red-free RPMI the ability of M2-like macrophages to degrade fibrillar with 1% fetal bovine serum and 1% penicillin/streptavidin. collagen during repair of skin wounds. Whether MMP-10 Collagenase activity was defined as the fluorescence detected in has a similar, beneficial role in mitigating fibrosis in disease the bottom chamber (excitation 480 nm and emission 530 nm) minus settings has yet to be determined. fluorescence in blank controls (no sample in the upper chamber). For

2382 Journal of Investigative Dermatology (2015), Volume 135 MG Rohani et al. MMP-10 Controls Scar Formation

3D matrix, we mixed native type I collagen (PurCol, Advanced Cieslewicz M, Tang J, Yu JL et al. (2013) Targeted delivery of proapoptotic BioMatrix, Carlsbad, CA) with DQ collagen and seeded explants peptides to tumor-associated macrophages improves survival. Proc Natl Acad Sci USA 110:15919–24 onto the mixed matrix and assessed collagenase activity. For Craig VJ, Quintero PA, Fyfe SE et al. (2013) Profibrotic activities for matrix antibody depletion, we cultured samples with antibodies against metalloproteinase-8 during bleomycin-mediated lung injury. JImmunol MMP-8 (R&D Systems, Minneapolis, MN) or MMP-13 (Thermos 190:4283–96 fi Fisher Scienti c, Waltham, MA). For macrophages, cells were Davies LC, Jenkins SJ, Allen JE et al. (2013) Tissue-resident macrophages. Nat incubated with phenol red-free RPMI containing 1% DQ collagen Immunol 14:986–95 for 6 hours. Duffield JS (2003) The inflammatory macrophage: a story of Jekyll and Hyde. Clin Sci (Lond) 104:27–38 Macrophage depletion Duffield JS (2014) Cellular and molecular mechanisms in kidney fibrosis. JClinInvest124:2299–306 Explants were cultured for 72 hours with phosphate-buffered saline fi or clodronate liposomes (50 mg ml − 1; http://www.clodronatelipo Duf eld JS, Forbes SJ, Constandinou CM et al. (2005) Selective depletion of macrophages reveals distinct, opposing roles during liver injury and repair. somes.org) added to the upper chamber. To deplete M2 macro- JClinInvest115:56–65 − 1 phages, explants were cultured with 20 μgml M2pepKLA or a Dumin JA, Dickeson SK, Stricker TP et al. (2001) Procollagenase-1 (matrix control scrambled peptide (scrM2pepKLA) (Cieslewicz et al., 2013). metalloproteinase-1) binds the integrin α2β1 upon release from keratinocytes migrating on type I collagen. JBiolChem276:29368–74 Macrophage culture Fallowfield JA, Mizuno M, Kendall TJ et al. (2007) Scar-associated macro- Isolation and activation of BMDMs were done as previously phages are a major source of hepatic matrix metalloproteinase-13 and facilitate the resolution of murine hepatic fibrosis. J Immunol 178: fl described (Manicone et al., 2009). Brie y, marrow cells were 5288–95 cultured for 7 days in Mac medium. For M2 activation, BMDMs Garcia-Irigoyen O, Carotti S, Latasa MU et al. (2013) Matrix metalloproteinase- − were stimulated with 10 ng ml 1 IL-4 and IL-13 for 48 hours. For M1 10 expression is induced during hepatic injury and plays a fundamental − – activation, cells were cultured with 100 ng ml 1 E. coli lipopoly- role in liver tissue repair. Liver Int 34:e257 70 saccharide for 6 hours, washed, and cultured for another 24 hours. Garcia-Prieto E, Gonzalez-Lopez A, Cabrera S et al. (2010) Resistance to bleomycin-induced lung fibrosis in MMP-8 deficient mice is mediated by For adoptive transfer, macrophages were collected at day 5 after interleukin-10. PloS One 5:e13242 marrow harvest and suspended in phosphate-buffered saline. Giannandrea M, Parks WC (2014) Diverse functions of matrix metalloprotei- 6 Recipient mice received 7 × 10 BMDMs in 50 μl phosphate- nases during fibrosis. Dis Model Mech 7:193–203 buffered saline via retro-orbital injection. Gordon S (2003) Alternative activation of macrophages. Nat Rev Immunol 3:23–35 Statistical analysis Goren I, Allmann N, Yogev N et al. (2009) A transgenic mouse model of Statistical analysis was performed using Prism 5 (GraphPad Software, inducible macrophage depletion: effects of diphtheria toxin-driven lysozyme M-specific cell lineage ablation on wound inflammatory, La Jolla, CA). We used t-test to compare two groups or two-way angiogenic, and contractive processes. Am J Pathol 175:132–47 analysis of variance followed by Bonferroni post-test to test the effect Hartenstein B, Dittrich BT, Stickens D et al. (2006) Epidermal development and of two factors. Data are presented as mean ± SEM. A P-value of wound healing in matrix metalloproteinase 13-deficient mice. JInvest o0.05 was considered statistically significant. Dermatol 126:486–96 Hattori N, Mochizuki S, Kishi K et al. (2009) MMP-13 plays a role in keratinocyte migration, angiogenesis, and contraction in mouse skin – CONFLICT OF INTEREST wound healing. Am J Pathol 175:533 46 The authors state no conflict of interest. Inada M, Wang Y, Byrne MH et al. (2004) Critical roles for collagenase-3 (Mmp13) in development of growth plate cartilage and in endochondral fi – ACKNOWLEDGMENTS ossi cation. Proc Natl Acad Sci USA 101:17192 7 This work was supported by NIH grants AR060157 (to MR), HL098067 (to Jameson JM, Cauvi G, Sharp LL et al. (2005) Gammadelta T cell-induced WCP), and HL089455 (to WCP). We thank Dr Anne Manicone, Dr Peter hyaluronan production by epithelial cells regulates inflammation. JExp Chen, and Tyler Vandivort for helpful discussions. We also thank Erin McCarty Med 201:1269–79 and Cara Leigh Appel of the Histology and Imaging Core in the Center for Jørgensen PH, Bang C, Andreassen TT (1993) Mechanical properties of skin Lung Biology, University of Washington. graft wounds. Br J Plast Surg 46:565–9 Kassim SY, Gharib SA, Mecham BH et al. (2007) Individual matrix SUPPLEMENTARY MATERIAL metalloproteinases control distinct transcriptional responses in airway epithelial cells infected with Pseudomonas aeruginosa. Infect Immun 75: Supplementary material is linked to the online version of the paper at http:// 5640–50 www.nature.com/jid Koller FL, Dozier EA, Nam KT et al. (2012) Lack of MMP10 exacerbates experimental colitis and promotes development of inflammation- REFERENCES associated colonic dysplasia. Lab Invest 92:1749–59 Almodovar-Garcia K, Kwon M, Samaras SE et al. (2014) ANKRD1 acts as a Krampert M, Bloch W, Sasaki T et al. (2004) Activities of the matrix metallo- transcriptional repressor of MMP13 via the AP-1 site. Mol Cell Biol 34: proteinase stromelysin-2 (MMP-10) in matrix degradation and keratinocyte 1500–11 organization in wounded skin. Mol Biol Cell 15:5242–54 Atabai K, Jame S, Azhar N et al. (2009) Mfge8 diminishes the severity of tissue Lee H, Overall CM, McCulloch CA et al. (2006) A critical role for the fibrosis in mice by binding and targeting collagen for uptake by membrane-type 1 matrix metalloproteinase in collagen phagocytosis. Mol macrophages. JClinInvest119:3713–22 Biol Cell 17:4812–26 Brown BC, McKenna SP, Siddhi K et al. (2008) The hidden cost of skin scars: Lichtnekert J, Kawakami T, Parks WC et al. (2013) Changes in macrophage quality of life after skin scarring. J Plast Reconstr Aesthet Surg 61: phenotype as the immune response evolves. Curr Opin Pharmacol 13: 1049–58 555–64

www.jidonline.org 2383 MG Rohani et al. MMP-10 Controls Scar Formation

Lucas T, Waisman A, Ranjan R et al. (2010) Differential roles of macrophages in Saarialho-Kere UK, Kovacs SO, Pentland AP et al. (1994) Distinct populations of diverse phases of skin repair. JImmunol184:3964–77 keratinocytes express stromelysin-1 and -2 in chronic wounds. JClinInvest – Madsen DH, Engelholm LH, Ingvarsen S et al. (2007) Extracellular collagenases 94:79 88 and the endocytic receptor, urokinase plasminogen activator receptor- Sabeh F, Li X, Saunders TL et al. (2009) Secreted versus membrane-anchored associated protein/Endo180, cooperate in fibroblast-mediated collagen collagenases: relative roles in fibroblast-dependent collagenolysis and degradation. JBiolChem282:27037–45 invasion. JBiolChem284:23001–11 Madsen DH, Ingvarsen S, Jurgensen HJ et al. (2011) The non-phagocytic route Shi J, Son MY, Yamada S et al. (2008) Membrane-type MMPs enable of collagen uptake: a distinct degradation pathway. JBiolChem286: extracellular matrix permissiveness and mesenchymal cell proliferation 26996–7010 during embryogenesis. Dev Biol 313:196–209 Madsen DH, Leonard D, Masedunskas A et al. (2013) M2-like macrophages are Siller-Lopez F, Sandoval A, Salgado S et al. (2004) Treatment with human responsible for collagen degradation through a mannose receptor- metalloproteinase-8 gene delivery ameliorates experimental rat liver mediated pathway. J Cell Biol 202:951–66 cirrhosis. Gastroenterology 126:1122–33 Manicone AM, Birkland TP, Lin M et al. (2009) Epilysin (MMP-28) restrains Sternlicht MD, Werb Z (2001) How matrix metalloproteinases regulate cell early macrophage recruitment in Pseudomonas aeruginosa pneumonia. J behavior. Annu Rev Cell Dev Biol 17:463–516 – Immunol 182:3866 76 Stickens D, Behonick DJ, Ortega N et al. (2004) Altered endochondral bone McKleroy W, Lee TH, Atabai K (2013) Always cleave up your mess: targeting development in matrix metalloproteinase 13-deficient mice. Development collagen degradation to treat tissue fibrosis. Am J Physiol Lung Cell Mol 131:5883–95 – Physiol 304:L709 21 Toriseva M, Laato M, Carpen O et al. (2012) MMP-13 regulates growth of Mosser DM, Edwards JP (2008) Exploring the full spectrum of macrophage wound granulation tissue and modulates gene expression signatures activation. Nat Rev Immunol 8:958–69 involved in inflammation, proteolysis, and cell viability. PLoS One 7: Murray PJ, Allen JE, Biswas SK et al. (2014) Macrophage activation and e42596 polarization: nomenclature and experimental guidelines. Immunity 41: Uchinami H, Seki E, Brenner DA et al. (2006) Loss of MMP 13 attenuates 14–20 murine hepatic injury and fibrosis during cholestasis. Hepatology 44: – Parks WC, Wilson CL, Lopez-Boado YS (2004) Matrix metalloproteinases as 420 9 modulators of inflammation and innate immunity. Nat Rev Immunol 4: Wagenaar-Miller RA, Engelholm LH, Gavard J et al. (2007) Complementary 617–29 roles of intracellular and pericellular collagen degradation pathways – Pellicoro A, Ramachandran P, Iredale JP et al. (2014) Liver fibrosis and repair: in vivo. Mol Cell Biol 27:6309 22 immune regulation of wound healing in a solid organ. Nat Rev Immunol Wu N, Jansen ED, Davidson JM (2003) Comparison of mouse matrix 14:181–94 metalloproteinase 13 expression in free-electron laser and scalpel incisions – Ramachandran P, Pellicoro A, Vernon MA et al. (2012) Differential Ly-6C during wound healing. JInvestDermatol121:926 32 expression identifies the recruited macrophage phenotype, which orches- Wynn TA, Barron L (2010) Macrophages: master regulators of inflammation and trates the regression of murine liver fibrosis. Proc Natl Acad Sci USA 109: fibrosis. Semin Liver Dis 30:245–57 – E3186 95 Yang L, Kwon J, Popov Y et al. (2014) Vascular endothelial growth factor Rohani MG, Chow YH, Razumova MV et al. (2014) uPARAP function in promotes fibrosis resolution and repair in mice. Gastroenterology 146: cutaneous wound repair. PLoS One 9:e92660 1339–50

2384 Journal of Investigative Dermatology (2015), Volume 135