The M2a Macrophage Phenotype Accompanies Pulmonary Granuloma Resolution in Mmp12 Knock-Out Mice Instilled with Multiwall Carbon Nanotubes

Sarcoidosis is a chronic disease with unknown etiology and pathophysiology, characterized by granuloma formation. Matrix Metalloproteinase-12 (MMP12) is an elastase implicated in active granulomatous sarcoidosis. Previously, we reported that oropharyngeal instillation of multiwall carbon nanotubes (MWCNT) into C57Bl/6 mice induced sarcoid-like granulomas and upregulation of MMP12. When Mmp12 knock-out (KO) mice were instilled with MWCNT, granuloma formation occurred 10 days post-instillation but subsequently resolved at 60 days. Thus, we concluded that MMP12 was essential to granuloma persistence. The aim of the current study was to identify potential mechanisms of granuloma resolution in Mmp12KO mice. Strikingly, an M2 macrophage phenotype was present in Mmp12KO but not in C57Bl/6 mice. Between 10 and 60 days, macrophage populations in MWCNT-instilled Mmp12KO mice demonstrated an M2c to M2a phenotypic shift, with elevations in levels of IL-13, an M2 subtype-regulating factor. Furthermore, the M2 inducer, Apolipoprotein E (ApoE), and Matrix Metalloproteinase-14 (MMP14), a promoter of collagen degradation, were upregulated in 60-day MWCNT-instilled Mmp12KO mice. In conclusion, alveolar macrophages express two M2 phenotypes in Mmp12KO mice: M2c at 10 days when granulomas form, and M2a at 60 days when granulomas are resolving. Findings suggest that granuloma resolution in 60-day Mmp12KO mice requires an M2a macrophage phenotype.


Introduction
Sarcoidosis is a multisystem granulomatous disease [1]. There are approximately 185,000 sarcoidosis cases in the United States, and approximately 1.2 million worldwide [1][2][3]. The pathogenesis of sarcoidosis is complex and not currently understood [1,4], and the current use of glucocorticoids as the primary treatment has shown mixed results along with severe adverse effects [5]. Sarcoid granulomas typically present in the eyes, skin, and lungs, with pulmonary granulomatous formations being associated with increased disease severity [6]. We previously described a multiwall carbon nanotube (MWCNT)-based murine model of pulmonary granulomatous inflammation in C57Bl/6 wild-type mice [7]. This model has been shown to induce chronic pulmonary granulomas with a histological presentation resembling that of human granulomas [7]. Alveolar macrophages, the predominant immune cell in the airways of healthy individuals, have been identified as key players in sarcoidosis pathogenesis [8]. Peroxisome 2 of 16 proliferator-activated receptor gamma (PPARγ) is a transcription factor that negatively regulates pro-inflammatory macrophage activation [9]. PPARγ is highly expressed in healthy individuals; however, its expression is decreased in alveolar macrophages from human sarcoidosis patients [10,11]. Alveolar macrophages from MWCNT-instilled mice also exhibit decreased PPARγ expression relative to control [12]. Correspondingly, interferon gamma (IFNγ), a pro-inflammatory cytokine, is upregulated in both human sarcoid patients and in the MWCNT model [12,13].
MMPs are a class of proteases that primarily function to degrade the extracellular matrix [14]. Additionally, this family of proteins functions in the immune and inflammatory responses, and are involved in tissue repair, tissue remodeling, and cell proliferation [14]. A transcriptional survey comparing alveolar macrophages from human sarcoid patients to alveolar macrophages from the MWCNT model found that a protein in this family, namely MMP12, was the most highly expressed gene in both groups [13]. MMP12 is an elastase produced by alveolar macrophages [14,15], and has previously been implicated in many chronic inflammatory diseases, both pulmonary and systemic [14]. In sarcoidosis patients, MMP12 gene expression is high in areas of active granulomatous inflammation [16]. Based on these observations, we analyzed pulmonary responses of Mmp12 KO mice to MWCNT instillation.
Previous results indicated that MWCNT instillation of Mmp12 KO mice resulted in granuloma formation at 10 days [17]. However, in a time course comparison, it was observed that while granulomas persist through 60 days in C57Bl/6 mice, granulomas in Mmp12 KO mice were resolving [17]. These results suggested that MMP12 was required for granuloma persistence. In MWCNT-instilled C57Bl/6 mice, PPARγ gene expression is decreased and IFNγ gene expression is elevated at 60 days [12]. In contrast, when Mmp12 KO mice are instilled with MWCNT, PPARγ expression is increased and IFNγ expression is decreased at 60 days [17].
PPARγ expression coincides with granuloma resolution in the Mmp12 KO model. In addition to the regulation of cytokines in the immune response, PPARγ can alter the phenotype of alveolar macrophages [18]. PPARγ promotes the M2 macrophage phenotype [19,20]. M2 macrophages function to heal, with anti-inflammatory activity in tissue repair and in the maintenance of tissue integrity [21,22]. In contrast, the M1 macrophages are pro-inflammatory in nature with their primary function being to kill or clear foreign antigens or pathogens by phagocytosis or microbicidal activity [21]. Interestingly, ApoE, a glycoprotein that may be induced by PPARγ through LXR activation [23,24], also induces an anti-inflammatory phenotype in macrophages, which drives production of factors such as IL-13 that promote the alternative M2 phenotype activation [25]. Based on these observations, we hypothesized that ApoE might also play a role in transforming the Mmp12 KO macrophage phenotype into a form responsible for granuloma resolution.

Mmp12 KO MWCNT Alveolar Macrophages Express M2c Markers at 10 Days and Convert to M2a by 60 Days
The M2 macrophages are further characterized into subtypes: M2a, M2b, and M2c. Surface expression of the receptor for advanced glycation end-products, or RAGE, is unique to M2c macrophages [27], while M2a macrophages are defined by Dectin-1 expression [28]. Currently, no unique surface proteins on M2b macrophages have been identified [28]. Macrophages from Mmp12 KO mice were co-stained with anti-RAGE and CD206 to determine the presence of the M2c phenotype. Anti-Dectin-1 and CD206 were co-stained to delineate the M2a phenotype. As seen in Figure 2A, high levels of RAGE were observed in both Mmp12 KO sham-and MWCNT-instilled mice at 10 days, while Dectin-1 was not  The M2 macrophages are further characterized into subtypes: M2a, M2b, and M2c. Surface expression of the receptor for advanced glycation end-products, or RAGE, is unique to M2c macrophages [27], while M2a macrophages are defined by Dectin-1 expression [28]. Currently, no unique surface proteins on M2b macrophages have been identified [28]. Macrophages from Mmp12 KO mice were co-stained with anti-RAGE and CD206 to determine the presence of the M2c phenotype. Anti-Dectin-1 and CD206 were co-stained to delineate the M2a phenotype. As seen in Figure 2A, high levels of RAGE were observed in both Mmp12 KO sham-and MWCNT-instilled mice at 10 days, while Dectin-1 was not observed in either group ( Figure 2B). These results suggest that the M2c macrophage phenotype predominated at 10 days in both sham-and MWCNT-instilled Mmp12 KO mice. observed in either group ( Figure 2B). These results suggest that the M2c macrophage phenotype predominated at 10 days in both sham-and MWCNT-instilled Mmp12 KO mice. As granuloma resolution occurs at 60 days in Mmp12 KO, we evaluated the macrophage phenotype using the same markers described above. At 60 days, RAGE expression was high in sham-instilled Mmp12 KO but was absent in MWCNT-instilled Mmp12 KO mice ( Figure 3A). At 60 days, Dectin-1 was not expressed by sham-instilled Mmp12 KO but was expressed at high levels in MWCNT-instilled Mmp12 KO mice ( Figure 3B). These observations of macrophage phenotype changes (summarized in Table 1) suggested a shift in macrophage polarization from M2c to M2a between 10 and 60 days in Mmp12 KO mice instilled with MWCNT, coinciding with granuloma resolution. As granuloma resolution occurs at 60 days in Mmp12 KO, we evaluated the macrophage phenotype using the same markers described above. At 60 days, RAGE expression was high in sham-instilled Mmp12 KO but was absent in MWCNT-instilled Mmp12 KO mice ( Figure 3A). At 60 days, Dectin-1 was not expressed by sham-instilled Mmp12 KO but was expressed at high levels in MWCNT-instilled Mmp12 KO mice ( Figure 3B). These observations of macrophage phenotype changes (summarized in Table 1) suggested a shift in macrophage polarization from M2c to M2a between 10 and 60 days in Mmp12 KO mice instilled with MWCNT, coinciding with granuloma resolution.   [29,30]. We investigated whether IL-13 was increased at 60 days. IL-13 gene and protein expression were significantly increased in MWCNT-instilled Mmp12 KO at 60 days but not at 10 days (Figure 4 A,B). Increased IL-13 was consistent with a phenotype shift from M2c to M2a in MWCNT-instilled  [29,30]. We investigated whether IL-13 was increased at 60 days. IL-13 gene and protein expression were significantly increased in MWCNT-instilled Mmp12 KO at 60 days but not at 10 days ( Figure 4A,B). Increased IL-13 was consistent with a phenotype shift from M2c to M2a in MWCNT-instilled Mmp12 KO mice at 60 days, as seen in immunocytochemistry staining (Table 1). No change in IL-13 was observed in wild-type mice instilled with MWCNT (data not shown). Mmp12 KO mice at 60 days, as seen in immunocytochemistry staining (Table 1). No change in IL-13 was observed in wild-type mice instilled with MWCNT (data not shown).

ApoE Is Upregulated in MWCNT-Instilled Mmp12 KO at 60 Days
PPARγ upregulation has been linked previously to upregulation of Apolipoprotein E (ApoE), a glycoprotein that is associated with numerous biological functions [31,32]. One of these functions involves the induction of the M2 macrophage phenotype [25,33,34].

ApoE Is Upregulated in MWCNT-Instilled Mmp12 KO at 60 Days
PPARγ upregulation has been linked previously to upregulation of Apolipoprotein E (ApoE), a glycoprotein that is associated with numerous biological functions [31,32]. One of these functions involves the induction of the M2 macrophage phenotype [25,33,34]. Additionally, ApoE deficiency has been associated with sarcoid-like granulomatous for-mations [35]. We hypothesized that at 60 days, high levels of ApoE in Mmp12 KO mice instilled with MWCNT might coincide with granuloma resolution.
Accordingly, ApoE gene expression was found to be elevated in MWCNT-instilled Mmp12 KO at 60 days in comparison to 10 days ( Figure 5A). ApoE gene expression was not elevated in MWCNT-instilled C57Bl/6 mice at 60 days ( Figure 5A). Correspondingly, ApoE protein was increased in BALF from MWCNT-instilled Mmp12 KO at 60 days relative to sham and 10-day ApoE levels ( Figure 5B). No change in BALF ApoE protein expression was seen in C57Bl/6 mice ( Figure 5B). At 10 days, low levels of ApoE protein were observed in BAL cells from both treatment groups of C57Bl/6 and Mmp12 KO mice. However, increased ApoE expression was observed in BAL cells from 60-day Mmp12 KO mice relative to C57Bl/6. These findings indicated that ApoE levels were low when granulomas were prevalent and increased when granulomas were resolving in 60-day MWCNT-instilled Mmp12 KO mice. Additionally, ApoE deficiency has been associated with sarcoid-like granulomatous formations [35]. We hypothesized that at 60 days, high levels of ApoE in Mmp12 KO mice instilled with MWCNT might coincide with granuloma resolution. Accordingly, ApoE gene expression was found to be elevated in MWCNT-instilled Mmp12 KO at 60 days in comparison to 10 days ( Figure 5A). ApoE gene expression was not elevated in MWCNT-instilled C57Bl/6 mice at 60 days ( Figure 5A). Correspondingly, ApoE protein was increased in BALF from MWCNT-instilled Mmp12 KO at 60 days relative to sham and 10-day ApoE levels ( Figure 5B). No change in BALF ApoE protein expression was seen in C57Bl/6 mice ( Figure 5B). At 10 days, low levels of ApoE protein were observed in BAL cells from both treatment groups of C57Bl/6 and Mmp12 KO mice. However, increased ApoE expression was observed in BAL cells from 60-day Mmp12 KO mice relative to C57Bl/6. These findings indicated that ApoE levels were low when granulomas were prevalent and increased when granulomas were resolving in 60-day MWCNT-instilled Mmp12 KO mice.  ApoE was not increased in MWCNT-instilled C57Bl/6 mice relative to sham at 60 days. ApoE expression in MWCNT-instilled Mmp12 KO was increased at 60 days relative to 10 days (* p ≤ 0.05; ** p ≤ 0.01; n ≥ 5/group). (B) ApoE BALF protein levels were increased in MWCNT-instilled 60-day Mmp12 KO relative sham (* p ≤ 0.05; ** p ≤ 0.01; n ≥ 9/group).

Surface Expression of MMP14 Is Increased in the Absence of MMP12 at 60 Days
Recently, ApoE was implicated in the uptake and degradation of collagen fragments [36]. Atabai et al. demonstrated that collagen is degraded into fragments by surface proteins, including MMP14, and that these fragments are then bound by ApoE, marking the fragments for phagocytosis and intracellular degradation [37]. Collagen is a component of mature human sarcoid granulomas [6]. When immunocytochemistry was performed on BAL cells, an increase in MMP14 protein expression was seen at 60 days in MWCNT-instilled Mmp12 KO mice relative to sham, while no expression was observed at 10 days ( Figure 6). Thus, MMP14 protein expression coincides with BALF levels of ApoE, suggesting the activation of this pathway.

Surface Expression of MMP14 Is Increased in the Absence of MMP12 at 60 Days
Recently, ApoE was implicated in the uptake and degradation of collagen fragments [36]. Atabai et al. demonstrated that collagen is degraded into fragments by surface proteins, including MMP14, and that these fragments are then bound by ApoE, marking the fragments for phagocytosis and intracellular degradation [37]. Collagen is a component of mature human sarcoid granulomas [6]. When immunocytochemistry was performed on BAL cells, an increase in MMP14 protein expression was seen at 60 days in MWCNTinstilled Mmp12 KO mice relative to sham, while no expression was observed at 10 days ( Figure 6). Thus, MMP14 protein expression coincides with BALF levels of ApoE, suggesting the activation of this pathway.

Discussion
As previously reported, granulomas resolved at 60 days in Mmp12 KO mice while granuloma formation persisted in C57Bl/6 wild-type mice [17]. Thus, MMP12 is essential Figure 6. MMP14 protein increased in Mmp12 KO mice at 60 days. ICC staining showed increased MMP14 expression on BAL cells from Mmp12 KO mice at 60 days relative to 10 days. Additionally, Mmp12 KO mice instilled with MWCNT expressed MMP14 at higher levels than in sham-instilled Mmp12 KO mice. Representative images of n ≥ 3 animals.

Discussion
As previously reported, granulomas resolved at 60 days in Mmp12 KO mice while granuloma formation persisted in C57Bl/6 wild-type mice [17]. Thus, MMP12 is essential for the persistence of granulomatous formation through unknown mechanisms. In the MWCNT-instilled Mmp12 KO, increased PPARγ expression coincided with this granuloma resolution [17]. PPARγ can function to change the phenotype of macrophages from M1 to M2 [19,20]. Functionally, this shift is from an inflammatory macrophage that acts by killing and clearing pathogens to an anti-inflammatory phenotype that acts in tissue repair [21]. As alveolar macrophages are the predominant immune cell in the lower airways where granulomas develop [8], we aimed to determine whether this phenotypic change could be involved in granuloma resolution. Our results using CD206, an M2 macrophage marker, in immunofluorescence staining clearly showed that the M2 phenotype was elevated in Mmp12 KO mice with very low levels in C57Bl/6. Therefore, the M2 macrophage phenotype may be involved in granuloma resolution.
The M2 macrophage population can be further classified into M2a, M2b, and M2c subpopulations. We characterized the Mmp12 KO macrophage population using M2a and M2c markers, Dectin-1 and RAGE, respectively [27,28]. Results from immunofluorescence analyses indicated the macrophage phenotype shifted from M2c to M2a between the acute (10 days) and chronic (60 days) stages of MWCNT-induced granulomas. The M2c macrophage phenotype was predominant during granuloma formation in MWCNTinstilled Mmp12 KO mice, while the M2a phenotype was present during granuloma resolution. The M2a phenotype facilitates the release of M2-related cytokines, along with enhancing endocytic activity and promoting tissue repair [38]. This M2a function differs from that of the M2c, which functions in the phagocytosis of apoptotic cells [38]. The M2a phenotype and its corresponding functions may be responsible for the granuloma resolution at 60 days in Mmp12 KO mice.
As a shift from M2c to M2a was observed in Mmp12 KO mice, we attempted to determine whether IL-13 could be involved in this change. IL-13 is a cytokine that induces the M2a phenotype [30,31]. We observed increased levels of IL-13 in the Mmp12 KO MWCNT-instilled mice at 60 days, which also corresponds with a shift to the M2a phenotype. At 10 days in these mice, granulomas formed in the presence of M2c macrophages. Between 10 and 60 days, the increased IL-13 resulted in a M2a macrophage phenotype when granulomas were resolving in MWCNT-instilled Mmp12 KO mice.
PPARγ, which led us to explore macrophage polarization, has also been implicated in the regulation of ApoE [23,24]. Previously, Samokhin et al. showed that ApoE-deficient mice on a high cholate diet form sarcoid-like granulomatous formations [35]. Additionally, ApoE has also been shown to induce the M2 macrophage phenotype [25,33,34]. Correspondingly, ApoE expression is increased in MWCNT-instilled Mmp12 KO mice at 60 days, when granulomas are resolving. The relationship between ApoE and granuloma resolution is not clear. However, ApoE was recently implicated in the uptake and degradation of collagen fragments [36]. ApoE binds these fragments, marking them for phagocytosis [37]. MMP14 is a surface protein that functions to degrade collagen [39]. We observed an increase in surface expression of MMP14 at 60 days in the Mmp12 KO mice instilled with MWCNT. Taken together, these results led us to propose a model for granuloma resolution (Figure 7). At 60 days in MWCNT-instilled Mmp12 KO mice, surface expression of MMP14 on the M2a macrophages degrades the collagen present in and around the granulomatous formation. Increased PPARγ expression leads to ApoE release by macrophages which bind collagen fragments. The bound fragments are then degraded by the M2a macrophages. We theorize that the role of M2a macrophages in granuloma resolution is through increased MMP14 surface expression which will function to degrade collagen, a defining feature of mature granulomas. These M2a cells also secrete increased ApoE, which can bind to collagen fragments, marking them for uptake. Through this collagen degradation and uptake, the stability of the granuloma will be diminished, leading to granuloma resolution.
Even though, Mmp12 KO mice instilled with MWCNT are intrinsically an M2 phenotype, our data suggest that the necessary M2c to M2a shift for granuloma resolution might be induced by IL-13. As MMP12 is necessary for granuloma persistence, our previous studies with a PPARγ agonist rosiglitazone have shown a decrease in granuloma size and formation in C57Bl/6 MWCNT-instilled mice [40], as well as downregulated MMP12 gene expression ( Figure S1). These findings may have significant translational impact on granuloma resolution in sarcoidosis, as there is spontaneous remittance in 30% of the patients [1]. The exact pathogenesis of sarcoidosis is unknown. Similarly, why the disease in some patients resolves spontaneously is unknown. We are postulating that these spontaneous remissions may involve a shift in macrophage phenotypes and that potential therapies may involve agents which drive macrophages to an M2a phenotype. Future studies in patients are needed to better define macrophage phenotypes in patients with resolving disease versus patients with progressive disease. Interestingly, several studies have suggested that alveolar macrophages from sarcoidosis patients have a mixed phenotype with both M1 and M2 being present [41,42]. However, the M2 subtypes have not been investigated in sarcoidosis. In conclusion, future studies are needed to define macrophage subtypes, and involvement of matrix metalloproteinases along with ApoE, in patients with chronic non resolving granulomas and those whose disease has resolved. This shift in M2 macrophage phenotype correlates with the resolution of granulomas at 60 days in Mmp12 KO mice. We theorize that the role of M2a macrophages in granuloma resolution is through increased MMP14 surface expression which will function to degrade collagen, a defining feature of mature granulomas. These M2a cells also secrete increased ApoE, which can bind to collagen fragments, marking them for uptake. Through this collagen degradation and uptake, the stability of the granuloma will be diminished, leading to granuloma resolution.
Even though, Mmp12 KO mice instilled with MWCNT are intrinsically an M2 phenotype, our data suggest that the necessary M2c to M2a shift for granuloma resolution might be induced by IL-13. As MMP12 is necessary for granuloma persistence, our previous studies with a PPARγ agonist rosiglitazone have shown a decrease in granuloma size and formation in C57Bl/6 MWCNT-instilled mice [40], as well as downregulated MMP12 gene expression ( Figure S1). These findings may have significant translational impact on granuloma resolution in sarcoidosis, as there is spontaneous remittance in 30% of the patients [1]. The exact pathogenesis of sarcoidosis is unknown. Similarly, why the disease in some patients resolves spontaneously is unknown. We are postulating that these spontaneous remissions may involve a shift in macrophage phenotypes and that potential therapies may involve agents which drive macrophages to an M2a phenotype. Future studies in patients are needed to better define macrophage phenotypes in patients with resolving disease versus patients with progressive disease. Interestingly, several studies have suggested that alveolar macrophages from sarcoidosis patients have a mixed phenotype with both M1 and M2 being present [41,42]. However, the M2 subtypes have not been investigated in sarcoidosis. In conclusion, future studies are needed to define macrophage subtypes, and involvement of matrix metalloproteinases along with ApoE, in patients with chronic non resolving granulomas and those whose disease has resolved.

Murine MWCNT Model
All studies were conducted in conformity with Public Health Service (PHS) policy on human care and use of laboratory animals and were approved by the institutional animal care and use committee (J207). C57Bl/6 wild-type and Mmp12 KO mice (Jackson Laboratories, Bar Harbor, ME, USA) were administered a single oropharyngeal dose of freshly prepared MWCNT (100 µg) (900-1201, lot-GS1802, SES Research, Houston, TX, USA) in PBS/35% surfactant (gift: Ony Inc, Amherst, NY, USA) or PBS/35% surfactant alone (sham control) as previously described and characterized [43]. Animals were sacrificed at either 10 or 60 days post-instillation and evaluated as previously described [7].

Characterization of Bronchoalveolar Lavage (BAL) Cells
BAL cells were collected as previously described [7,44]. BAL cells were characterized by total cell counts and differential analysis ( Table 2). Cells and aliquoted BAL fluid were stored at −80 • C and used for gene expression and protein analysis.

Statistical Analysis
Using Prism 7 software (GraphPad Inc., San Diego, CA, USA), the data were analyzed by Student's t-test or one-way analysis of variance (ANOVA) with Tukey's multiple comparison test. A p-value of ≤0.05 was considered significant.

Histological Analysis
Lungs were dissected and fixed in 10% buffered formalin. Paraffin embedded blocks were sectioned at 7 µm and stained with hematoxylin and eosin (H&E).