Trypanosoma cruzi Induces the PARP1/AP-1 Pathway for Upregulation of Metalloproteinases and Transforming Growth Factor β in Macrophages: Role in Cardiac Fibroblast Differentiation and Fibrosis in Chagas Disease

Cardiomyopathy is the most important clinical manifestation of T. cruzi-driven CD. Recent studies have suggested the detrimental role of the matrix metalloproteinases MMP2 and MMP9 in extracellular matrix (ECM) degradation during cardiac remodeling in T. cruzi infection. Peripheral TGF-β levels are increased in clinically symptomatic CD patients over those in clinically asymptomatic seropositive individuals. We provide the first evidence that during T. cruzi infection, Mϕ release of MMP2 and MMP9 plays an active role in activation of TGF-β signaling of ECM remodeling and cardiac fibroblast-to-myofibroblast differentiation. We also determined that PARP1 signals c-Fos- and JunB-mediated AP-1 transcriptional activation of profibrotic gene expression and demonstrated the significance of PARP1 inhibition in controlling chronic fibrosis in Chagas disease. Our study provides a promising therapeutic approach for controlling T. cruzi-driven fibroblast differentiation in CD by PARP1 inhibitors through modulation of the Mϕ signaling of the AP-1–MMP9–TGF-β pathway.

differentiation during T. cruzi infection and chronic disease development. We recently documented that PARP1 [poly(ADP-ribose) polymerase 1], through signaling NF-B transcriptional activation, contributes to proinflammatory activation of M (12). We therefore also determined PARP1's role in regulation of M and cardiac fibrosis in CD. For this, we used cultured M and primary M isolated from bone marrow (BM) of wild-type (WT) and Parp1 Ϫ/Ϫ mice and employed classical approaches to evaluate the PARP1 signaling of transcriptional programs relevant to profibrotic M response during T. cruzi infection.

RESULTS
Macrophages express and release MMPs in response to T. cruzi infection. We first evaluated the mRNA levels for matrix metalloproteinases (MMPs) in RAW 264.7 M infected with T. cruzi. For this, M were incubated with T. cruzi (with or without gamma interferon [IFN-␥] and C-C motif chemokine ligand 2 [CCL2], individually or in combination) for 3 h and 18 h and analyzed by reverse transcription-quantitative PCR (RT-qPCR). We included IFN-␥ and CCL2 because these molecules are detected in the serum of infected mice and humans (13,14), IFN-␥ is shown to regulate MMPs through suppression of the ATF-3/AP-1 transcription program in lipopolysaccharide (LPS)treated M (15), while CCL2 is shown to enhance MMP9 expression (16). RAW M infected with T. cruzi (versus noninfected M) exhibited 2.7-to 3.4-fold and 4.3-to 8.1-fold increases in mRNA levels for Mmp2 and Mmp9 (which encode gelatinases), respectively, and the maximal increase in Mmp mRNAs was observed at 18 h ( Fig. 1A and B) (P Ͻ 0.001). Further, we noted an 11.1-fold increase in mRNA for Mmp12 (which encodes macrophage metalloelastase) at 18 h ( MMP12 degrades soluble and insoluble elastin. We noted 2-fold and 9-fold increases in pro-MMP12 (54 kDa) and cleaved MMP12 (45 kDa) levels in cell lysates and spent medium, respectively, of infected (versus noninfected) M at 24 h ( Fig. 1G and H) (P Ͻ 0.001). Coincubation with IFN-␥ and CCL2 had nonsignificant or suppressive effects on T. cruzi-induced MMP12 levels at 24 h ( Fig. 1G and H). By 48 h, intracellular levels of MMP12 were decreased in infected (with or without IFN-␥) M, while a 1.35-fold increase in secreted MMP12 was observed only when infected M were incubated with IFN-␥ and CCL2 ( Fig. S1B and C). Together, the results presented in Fig. 1 and Fig. S1 suggest that T. cruzi infection of M elicits a significant increase in the expression of MMP2, MMP9, and MMP12 metalloproteinases that is predominantly independent of the presence of IFN-␥ and CCL2. Most of the MMP2 and MMP9 that exhibited gelatinase activity and the cleaved form of MMP12 were released from infected M within 24 h postinfection.

Macrophage production of metalloproteinases and TGF-␤ in Chagas disease.
To verify our findings from in vitro studies (Fig. 1) and determine if profibrotic M are stimulated in CD, we used a well-established mouse model of T. cruzi infection (17). Mice were euthanized at 150 days postinfection, which corresponds to the chronic disease phase. The splenic and cardiac CD11b ϩ M of infected and noninfected (control) mice were isolated and incubated for 24 h in complete medium, and cells and Cells were incubated with T. cruzi trypomastigotes (cell-to-parasite ratio, 1:3) in the presence and absence of 20 ng/ml IFN-␥ and 10 ng/ml CCL2 for 24 h. (E and F) Cell supernatants were collected after 24 h of incubation, concentrated 10-fold, and used for MMP2 (pro-form, 72 kDa; cleaved form, 66 kDa) and MMP9 (pro-form, 92 kDa) zymography. Bar graphs (F) show relative gelatin degradation scores. (G and H) Western blot images for MMP12 (pro-form, 54 kDa; cleaved form, 45 kDa [G]) and relative densitometry scores for MMP12 (normalized to GAPDH [H]) in cell homogenates and supernatants (obtained after 24 h of incubation). Data in bar graphs are means and SD and are representative of two independent experiments with two biological replicates per treatment per experiment (triplicate observations per sample for RT-qPCR analysis and duplicate observations per sample for gelatin zymography and Western blot analyses). Significance is annotated as follows: ϩ, control versus infected (Student's two-tailed t test);^, infected versus infected treated (one-way analysis of variance [ANOVA] followed by Tukey's post hoc correction test).^, P Յ 0.05; ϩϩ and^^, P Յ 0.01; ϩϩϩ and^^^, P Յ 0.001. Nor, normal (no infection).
Choudhuri and Garg ® supernatants were used for various studies. We detected 2.3-fold, 2.8-fold, and 8.1-fold increases in Mmp2, Mmp9, and Tgfb1 mRNA levels, respectively, in splenic M of CD-infected (versus control) mice and 5.6-fold, 7.2-fold, and 9.7-fold increases in Mmp2, Mmp9, and Tgfb1 mRNA levels, respectively, in myocardial M of chronically infected (versus noninfected) mice ( Fig. 2A to C) (P Ͻ 0.001). Splenic and cardiac M of Chagas (versus control) mice also exhibited 5-to 14.2-fold and 2.3-to 10.2-fold increases in TGF-␤, tumor necrosis factor alpha (TNF-␣), and monocyte-chemotactic protein 3 (MCP3) release, respectively, after 24 h of incubation ( Fig. 2D to F) (P Ͻ 0.001). Gelatin zymography showed 0.9-fold, 3.7-fold, and 4.0-fold increases in the gelatin degradation score for pro-MMP2 (72 kDa), cleaved MMP2 (66 kDa), and MMP9 (92 kDa), respectively, produced by splenic CD11b ϩ M of chronically infected (versus noninfected) mice ( Fig. 2G to J) (P Ͻ 0.001). Likewise, we observed 12-fold, 6.1-fold, and 15.0-fold increase in the gelatin degradation score of pro-MMP2 and cleaved MMP2 and MMP9 (92 kDa), respectively, released by heart CD11b ϩ M of chronically infected (versus noninfected) mice ( Fig. 2G and H) (P Ͻ 0.001). Together, these results show a profibrotic M response in chronic CD that is evidenced by an increase in the expression and release of active forms of MMP2 and MMP9 by splenic and heart M of CD-infected (versus noninfected) mice. The production of TGF-␤ by splenic and heart M provides further support for their profibrotic role in CD.
MMP2 and MMP9 promote TGF-␤ activation in macrophages infected with T. cruzi. MMP2 and MMP9 are (i) involved in degradation of ECM under normal physiologic and disease conditions (18), (ii) implicated in cell-to-cell communication independent of their ECM degradation function (19), and (iii) suggested to dock on the cell surface hyaluronic receptor CD44 to promote cleavage and activation of TGF-␤ (20). TGF-␤, in turn, downregulates the inflammatory processes and activates cell-proliferative responses (18). To determine whether MMPs promote M's profibrotic response by influencing TGF-␤ during T. cruzi infection, we incubated RAW M with T. cruzi in the presence or absence of specific inhibitors of MMP2, MMP9, MMP12, and urokinase-type plasminogen activator (uPA) (which activates MMP2 and MMP9) and monitored release of TGF-␤ and inflammatory cytokines/chemokines by an enzyme-linked immunosorbent assay (ELISA) at 24 h and 48 h. RAW M responded to T. cruzi infection with 9-to 10.5-fold, 20-to 27-fold, and 21-to 25-fold increases in TGF-␤, TNF-␣, and MCP3/CCL7 release, respectively, at 24 h and 48 h post-incubation ( Fig. 3A to C) (P Ͻ 0.001). T. cruzi-induced TGF-␤ production was decreased by 69 to 74%, 60 to 66%, and 79 to 82% when infected M were incubated with inhibitors of MMP2, MMP9, and uPA, respectively ( Fig. 3A) (P Ͻ 0.001). No effect of an MMP12 inhibitor on TGF-␤ production by infected M was observed (Fig. 3A). Likewise, no effect of any of the metalloproteinase inhibitors was observed on T. cruzi-induced TNF-␣ and MCP3/CCL7 production by M at any of the time points studied ( Fig. 3B and 3C). Together, these results suggest that (i) M respond to T. cruzi with a potent increase in profibrotic (TGF-␤) and proinflammatory (TNF-␣ and MCP3) responses and (ii) M activation of TGF-␤, but not TNF-␣ and MCP3, was dependent on MMP2 and MMP9. Our observation of inhibitory effects of a uPA antagonist on TGF-␤ production suggests that active forms of MMP2 and MMP9 signal TGF-␤ activation in M infected by T. cruzi. M MMPs and TGF-␤ contribute to cardiac fibroblast differentiation. Cardiac fibroblast differentiation to myofibroblasts and ECM remodeling are the major drivers of cardiac fibrosis in Chagas and other heart diseases. To determine if T. cruzi-induced M production of gelatinases and TGF-␤ stimulated cardiac fibroblast differentiation, we incubated human THP-1 M with T. cruzi in the presence and absence of inhibitors of MMP2 and MMP9 for 24 h and collected the supernatants. Next, human cardiac fibroblasts (HCF) were incubated for 48 h, 72 h, and 96 h with spent medium from infected M and subjected to immunostaining using fluorescence-tagged antibodies against S100A4 (fibroblast-specific) and alpha smooth muscle actin (␣-SMA) (myofibroblastspecific) proteins. HCF incubated with medium from T. cruzi-infected M exhibited a significant differentiation to myofibroblasts, as evidenced by a gradual increase in the frequency of ␣-SMA ϩ cells from 10% to 82% and a decline in the frequency of S100A4 ϩ cells from 89% to 17% during 48 to 96 h of incubation ( Fig. 4E to H). These findings were consistent with the observations made using recombinant TGF-␤1 that also led to a gradual increase in the frequency of ␣-SMA ϩ cells from 12% to 93% with a corre- ) are representative of two independent experiments with two biological replicates per treatment (triplicate and duplicate observations per sample for RT-qPCR and zymography, respectively). Significance was calculated by Student's two-tailed t test or a nonparametric Mann-Whitney U test. ϩϩϩ, P Յ 0.001.

Choudhuri and Garg
® sponding decline in the frequency of S100A4 ϩ cells from 87% to 7% during 48 to 96 h of incubation ( Fig. 4Q to T). In comparison, incubation of HCF with spent medium from noninfected M resulted in no differentiation ( Fig. 4A to D). When incubated with supernatants from THP-1 M previously treated with T. cruzi and an MMP2 antagonist, HCF continued to exhibit up to 67% differentiation to an ␣-SMA ϩ phenotype by 96 h (Fig. 4I to L), which indicates that MMP2 has a less significant impact on cardiac fibroblast-to-myofibroblast differentiation. In contrast, incubation of HCF with supernatant of THP-1 M previously treated with T. cruzi and an MMP9 antagonist led to only a 30% increase in the frequency of ␣-SMA ϩ cells, while up to 70% of the cells retained the S100A4-positive phenotype during the 96 h ( Fig. 4M to P). Together, the results presented in Fig. 4, along with those in Fig. 3, suggest that MMP9 plays a key role in the release of functional TGF-␤1 by M infected by T. cruzi, and M release of MMP9/TGF-␤ creates a microenvironment conducive to differentiation of cardiac fibroblasts to myofibroblasts. PARP1 signaling of profibrotic response during T. cruzi infection and Chagas disease. We have reported that PARP1 plays a regulatory role in signaling the NF-B- mediated proinflammatory gene expression in cardiac myocytes and M in response to T. cruzi infection (12,21) and in M stimulated with extracellular vesicles (EV) released by infected cells (12). Since NF-B can be involved in transcription of metalloproteinases, we determined if PARP1 also signals profibrotic gene expression during T. cruzi infection. For this, we incubated RAW M with T. cruzi in the presence or absence of 10 M iniparib (a selective PARP1 inhibitor) for 24 h and performed RT-qPCR analysis. We also incubated M with T. cruzi-induced EV (TEV) previously purified from the Next, human cardiac fibroblast (HCF) cells were incubated for 48 to 96 h with spent medium from THP-1 M that were noninfected (A to D), infected (E to H), or infected and treated with inhibitors of MMP2 (I to L) or MMP9 (M to P). Positive controls were HCF cells incubated with recombinant TGF-␤ (Q to T). Transition of HCF to myofibroblasts was determined by immunofluorescence staining for S1000A4 (red) and ␣-SMA (green), respectively. DAPI (4=,6-diamidino-2-phenylindole; blue) staining marks nuclei. Representative images are shown at a magnification of ϫ40. Bar graphs show percentages of S100A4 ϩ and ␣-SMA ϩ HCF at 48 to 96 h post-incubation. Data (means and SD) are derived from two independent experiments, with three biological replicates per treatment, and each sample was analyzed in 9 microscopic fields. Significance comparing the expression of S100A4 (ϩ) or ␣-SMA (*) at different time points was calculated by repeated-measures ANOVA or post hoc test. ϩ and *, P Յ 0.05; ϩϩ and **, P Յ 0.01; ϩϩϩ and ***, P Յ 0.001. culture supernatants of T. cruzi-infected M. As expected from the data in Fig. 2  exhibited only a 43% increase in the frequency of ␣-SMA ϩ cells (Fig. S2A to D). In comparison, up to 83% of the HCF had gained the ␣-SMA ϩ phenotype when incubated with supernatants of THP-1 M previously incubated with T. cruzi only (Fig. 4E to H). We observed no changes in Mmp3 and Mmp8 mRNA levels in M incubated with T. cruzi (with or without PARP1 inhibitor) (Fig. S2E and F) compared to matched controls. Likewise, incubation with TEV (with or without PARP1 inhibitor) resulted in no significant increase in mRNA expression for any of the metalloproteinases and TGF-␤ compared to that noted in normal controls (Fig. 5A to E; Fig. S2E and F).
Together, the results presented in Fig. 5 and Fig. S2 suggest that (i) PARP1 is required to activate the profibrotic transcriptional profile in M infected by T. cruzi and in heart tissues of Chagas mice and (ii) chemical inhibition or genetic deletion of Parp1 was beneficial in arresting M profibrotic gene expression and consequently TGF-␤driven cardiac fibroblast-to-myofibroblast differentiation in response to T. cruzi infection.
Transcriptional activation of profibrotic gene expression in M infected by T. cruzi (with or without PARP1). Finally, we examined transcriptional regulation of metalloproteinases in M. Genome-wide studies have identified transcription binding sites for activator protein 1 (AP-1), AP-2, NF-B, RUNX-2, and STAT-3 in promoter sequences of genes encoding metalloproteinases, especially MMP2 and MMP9, and TGF-␤ (22). To determine which of these transcription factors might be involved in upregulation of the profibrotic response to T. cruzi, we infected RAW M with T. cruzi, incubated them in the presence and absence of specific antagonists of transcription factors, and monitored gene expression by RT-qPCR at 3 h and 24 h. T. cruzi induced a 3-to 7-fold increase in Mmp2, Mmp9, and Tgfb1 mRNA levels that was not significantly changed in the presence of any of the inhibitors at 3 h (Fig. 7A to C). By 24 h, we noted an 8-to 20-fold increase in Mmp2, Mmp9, and Tgfb1 mRNA levels in infected (versus control) M (Fig. 7A to C). Coincubation with an AP-1 inhibitor resulted in 52%, 66.5%, and 68% declines in T. cruzi-induced Mmp2, Mmp9, and Tgfb1 mRNAs, respectively (P Ͻ 0.001 for all), while NF-B inhibition resulted in a 45 to 50% decline in Mmp2 and Mmp9 mRNAs (P Ͻ 0.001) but had no effect on T. cruzi-induced increases in Tgfb1 mRNA levels in infected M (Fig. 7A to C). Furthermore, inhibition of RUNX2 and STAT3 transcription factors had no effect on the mRNA levels of Mmp2, Mmp9, or Tgfb1 in infected M. These results suggest that AP-1 is involved primarily in transcriptional activation of the profibrotic response in M infected by T. cruzi.
PARP1 is shown to activate JNK and DNA binding of AP-1 in fibroblasts (23) and enhance NF-B/AP-1 binding to MMP9 promoter sequence in diabetic retinopathy (24), though its role in transcriptional activation of the profibrotic program in M, if any, is not known. To investigate whether PARP1 is engaged in AP-1 transcriptional activation of the profibrotic response, we incubated the primary bone marrow M of WT and Parp1 Ϫ/Ϫ mice with T. cruzi trypomastigotes for 24 h and utilized the nuclear extracts for an AP-1 transcription factor family assay. We found 2.64-fold, 2.86-fold, and 0.65-fold increases in c-Fos, JunB, and c-Jun AP-1 family proteins, respectively, in primary WT M infected with T. cruzi (versus no infection; P Ͻ 0.001) (Fig. 7D, G, and I). Parp1 Ϫ/Ϫ M infected with T. cruzi exhibited 1.3-fold and 1.5-fold declines in c-Fos and JunB levels, respectively ( Fig. 7D and 7G) (P Ͻ 0.001), but no decline in c-Jun level (Fig. 7I) compared to infected WT M. No activation of FosB, Fra-1, or JunD AP1 transcriptional family proteins was observed in either the WT or Parp1 Ϫ/Ϫ primary M in response to T. cruzi infection (Fig. 7E, F, and H). Together, the results presented in Fig. 7 suggest that PARP1 signals transcriptional activation of c-Fos and JunB AP-1 transcriptional family proteins to regulate the M profibrotic response to T. cruzi infection and CD.

DISCUSSION
There is a growing body of research endorsing the role of parasite persistence in facilitating the myocardial infiltration of immune and nonimmune cells and disturbing cardiac homeostasis (reviewed in reference 2). Dilated cardiomyopathy, the pathological manifestation of chronic CD, is characterized by diffused myocarditis, ECM remodeling, and interstitial fibrosis, which cause left ventricular dysfunction (5,6). Matrix metalloproteinases, especially MMP2 and MMP9, have emerged as modulators of cardiovascular inflammation as well as remodeling. In acute T. cruzi infection, MMP2 and MMP9 expression was associated with increased inflammation, and treatment with MMP2/MMP9 antagonists was beneficial in controlling inflammatory pathology and mortality in mice (25). In chronic CD, increased plasma levels of MMP2 predicted early cardiac remodeling in clinically asymptomatic CD, while MMP9 was identified as a risk factor for late fibrosis and severe cardiac remodeling in clinically symptomatic patients (11). By using flow cytometry evaluation of peripheral blood cells, Medeiros et al. (26) showed that intracellular MMP2 and MMP9 levels were increased in neutrophils and monocytes of seropositive patients. When cells were exposed to a T. cruzi antigenic stimulus, the frequency of MMP2-and MMP9-expressing monocytes was further increased in peripheral blood cells of clinically symptomatic patients with the cardiac form of the disease but not in the seropositive, clinically asymptomatic individuals, thus implying that immune cells may contribute to fibrosis in CD through MMP production. In the present study, we demonstrate that expression and activity of MMP2, MMP9, and MMP12 were increased in cultured and primary (splenic and cardiac) M in response to T. cruzi infection and in the myocardium of chronically infected mice ( Fig. 1 and 2). The M activation of metalloproteinases by T. cruzi was not critically influenced by CCL2 and IFN-␥, as is observed in other disease models (27)(28)(29). Furthermore, T. cruzi-induced M metalloproteinases (MMP9 more than MMP2) promoted TGF-␤ production and human cardiac fibroblast-to-myofibroblast differentiation in vitro (Fig. 4). The MMP2-  JunD (H), and c-Jun (I) by using AP-1 transcription factor family ELISA. Data (means and SD) are representative of two independent experiments, 2 or 3 biological replicates per treatment, and 2 or 3 observations per sample. Significance is annotated as follows: ϩ, noninfected versus infected (Student's two-tailed t test or nonparametric Mann-Whitney U test);^, infected versus infected and treated (one-way ANOVA followed by Tukey's post hoc correction test). ϩϩ, P Յ 0.01; ϩϩϩ and^^^, P Յ 0.001). and MMP9-producing M were also associated with myocardial fibrosis, which presented with increased tissue distribution of TGF-␤ (which indicates a profibrotic response), galectin-3 (a marker of fibrotic response), and vimentin (a marker of fibroblast differentiation to myofibroblasts) in chronically infected Chagas mice (Fig. 6). Our findings endorse the importance of cardiac M in driving the MMP2/MMP9-mediated, TGF-␤-dependent profibrotic-fibrotic outcomes in the heart during T. cruzi infection and CD development.
The M-specific matrix metalloproteinase, i.e., MMP12, is identified as an endogenous resolution promoting factor in the events of myocardial infarction (30), and it plays a protective role by arresting inflammatory pathways and neutrophil infiltration in various diseases (31). MMP13 is a member of the collagenase family, which degrades many types of collagens and other profibrotic molecules (32). In vitro studies have identified many chemokines (e.g., CCL2, -3, -5, -7, -8, -15, -16, -17, and -23) as putative MMP13 substrates (33). Studies in Mmp13-deficient mice have identified contradictory phenotypes, with finding of attenuated inflammation and pulmonary fibrosis after irradiation-induced injury (34), no effect on fibrotic responses and an increase in inflammatory profile after hypertoxic lung injury (35), and development of severe inflammation and lung fibrosis after bleomycin treatment (33). The findings of increased expression of Mmp12 and Mmp13 in M infected by T. cruzi ( Fig. 5C and D) and in the heart tissue of chronically infected mice ( Fig. 5H and I) in this study and in a previously published study (36) and increased MMP12 expression in infected M (Fig. 1G and H) suggest that these metalloproteinases may also play a key role in regulating pathological inflammation and fibrosis in CD. However, the mechanistic role of MMP12 and MMP13 in the context of CD was not addressed in this study or in previously published literature, and this remains to be examined in future studies.
The role of TGF-␤ signaling in remodeling heart is complex and not completely understood. This is because its diverse effects elicit multiple and often opposing cellular responses. In brief, the TGF-␤ superfamily transduces signal to downstream effectors primarily through the SMAD family of proteins, but it may also activate other signaling cascades, including extracellular signal-regulated kinase (Erk), c-Jun N-terminal kinase (JNK), TGF-␤-activated kinase 1 (TAK1), and p38 mitogen-activated protein kinase (MAPK) pathways (37,38). Depending on the time, extent, and types of surface receptors and intracellular signaling cascades engaged, TGF-␤ may play a crucial role in repression of the inflammatory response and mediate resolution of the inflammatory infiltrate after injuries caused by infectious or genotoxic stimuli (9). TGF-␤ may be cardioprotective or play a key role as a mediator of pathological hypertrophy and ventricular remodeling, depending on the extent to which it modulates the fibroblast phenotype, promotes ECM deposition by upregulating collagen and fibronectin synthesis and by decreasing matrix degradation through induction of protease inhibitors, or stimulates cardiomyocyte growth and induces interstitial fibrosis (7,39). In CD, TGF-␤ is suggested to serve as a growth factor, promoting parasite replication (40), and treatment with a TGF-␤ receptor inhibitor decreased the parasitemia, heart damage, and mortality in infected mice (41). Patients with the clinically symptomatic cardiac form of CD exhibit higher serum levels of TGF-␤ than clinically asymptomatic seropositive patients, and CD patients with higher TGF-␤ levels present a worse clinical outcome (42,43). Our study showed significantly higher Tgfb1 expression in the heart tissues of chronic CD-infected mice and T. cruzi-infected M. Thus, current literature and our data corroborate the importance of TGF-␤ in the development and maintenance of cardiac damage in response to T. cruzi infection.
Besides its complex role, the unusual biology of TGF-␤ activation is intriguing. Using a 3-dimensional collagen gel model loaded with TGF-␤ in conjunction with fibroblasts deficient in MMP9, Kobayashi et al. (44) showed that MMP9 was required to activate TGF-␤ and that MMP9 deficiency reduced the TGF-␤-driven responses, including activity of a Smad3 reporter gene and production of fibronectin. Using TA3 murine mammary carcinoma cells, Yu and Stamenkovic (20) showed that MMP9 is localized on the cell surface in a CD44-dependent manner and stimulates proteolytic cleavage of TGF-␤, and they proposed a novel role for MMP9 in TGF-␤-dependent tumorigenesis. Activation of TGF-␤1 was at least partially dependent on MMP2 activity in age-associated arterial remodeling (45). TGF-␤, in turn, can inhibit MMP9 activity through the synthesis of protease inhibitors. Our data show that inhibition of MMP2, MMP9, or uPA (but not MMP12) abolished the M release of TGF-␤ in response to T. cruzi infection. Whether MMP2 and MMP9 influence TGF-␤ release through regulation of transcriptional, translational, or posttranslational activation programs in M infected by T. cruzi remains to be examined in future studies. Nonetheless, our findings provide the first evidence that during T. cruzi infection, M MMP9 plays a critical role in the activation of latent TGF-␤ and that this in turn regulates cardiac fibroblast-tomyofibroblast differentiation.
PARP1 is widely regarded as an intracellular DNA damage sensor. In CD, innate immune cells responding to low-grade parasite persistence and mitochondrial dysfunction are recognized as major sources of free radicals (reviewed in references 2, 46, and 47), and the resultant oxidative DNA adducts were shown to cause PARP1 hyperactivation in cardiomyocytes and heart tissue infected by T. cruzi (21,48). PARP1 direct binding and PARP1-mediated PARylation of proteins can affect the expression of inflammation-related genes at the transcriptional and posttranscriptional levels (21), and PARP1-dependent NF-B-mediated proinflammatory gene expression was increased in cardiac myocytes infected by T. cruzi (48). Further, we have found that T. cruzi infection and chronic CD induce release of extracellular vesicles (EV) carrying oxidized DNA of host and parasite origin. When phagocytized by M, the oxidized DNA in these EV was sensed by the cytosolic DNA response element cGAS, which synergized with PARP1 to signal NF-B transcriptional activation (12). Our results in this study show that T. cruzi-induced EV (TEV) were not profibrotic, and invasion by live T. cruzi was required to induce the MMP/TGF-␤ expression in M. These findings suggest that TEV do not regulate the MMP and TGF-␤ expression in M by the involvement of cytosolic cGAS-PARP1 and the downstream NF-B signaling pathway. Instead, nuclear activation of Fos and JunB AP-1 transcription family proteins, in synergy with nuclear PARP1, signaled the increase in the expression and activity of MMP2, MMP9, and TGF-␤ in M infected with T. cruzi (Fig. 7). This was evidenced by our findings in this study that genetic deletion or chemical inhibition of PARP1 abolished the T. cruzi-induced increase in mRNAs for Mmp2, Mmp9, Mmp12, and Mmp13 in M and cardiac tissue of Chagas mice and decreased the expression of MMP9, TGF-␤, and profibrotic/ fibrotic differentiation markers in heart tissue of CD-infected mice. Likewise, we reported previously that PJ34 (a PARP1 inhibitor) significantly improved the mitochondrial biogenesis and antioxidant capacity in the heart tissues of T. cruziinfected mice and genetic deletion of PARP1 decreased the collagen deposition and the cardiac fibrosis and remodeling in chronically infected mice (49). Others have shown that treatment with olaparib (PARP1 inhibitor) ameliorated MMP2 and MMP9 expression in an elastase-induced mouse model of emphysema (50). Thus, based on our previous and present findings, we surmise that PARP1 plays an important role in progression of inflammation and fibrosis in CD through the engagement of cGAS-NF-B and AP-1 signaling pathways.
MMP promoters harbor several cis elements, which allow the regulation of MMP gene expression by various transcriptional factors, such as AP-1, NF-B, RUNX2, and STAT3, among others (reviewed in reference 22). PARP1 was shown to promote nuclear translocation of RelA (p65) in T. cruzi-infected cardiomyocytes (48) and to enhance NF-B/AP-1 binding to the MMP9 promoter sequence in diabetic retinopathy (24) and JNK-dependent activation of AP-1 in murine fibroblasts (23). In this study, NF-B only partially contributed to increases in Mmp2 and Mmp9 expression, and AP-1 was necessary to signal M expression of Mmp2, Mmp9, and Tgfb1 in response to T. cruzi infection. Further, Parp1 Ϫ/Ϫ (versus control) M exhibited significantly low activity of c-Fos and JunB AP-1 transcriptional family proteins. Our findings establish that PARP1 functions as a transcriptional coregulator of c-Fos and JunB members of the AP1 family to drive the profibrotic/fibrotic response of M, and they offer novel insights into the role of PARP1 in the transcriptional regulation and release of M gelatinases and profibrotic TGF-␤ in Chagas disease.
In summary, we used in vitro and in vivo models of T. cruzi infection and CD to show that M MMP2 and MMP9 released during T. cruzi infection are active and shape the ECM profibrotic response by the release of active TGF-␤. Chemical inhibition or genetic deletion of PARP1 arrested c-Fos-and JunB-mediated AP-1 transcriptional activation of MMP/TGF-␤ responses in M, and this was beneficial in arresting the T. cruzi-driven cardiac fibroblast-to-myofibroblast differentiation in chronic CD. One limitation of the present study is that Parp1 Ϫ/Ϫ mice express a shortened transcript lacking enzymatic activity in all cells and tissues, and it is possible that the observed decline in myocardial infiltration of profibrotic M and cardiac fibrosis in Parp1 Ϫ/Ϫ mice was an outcome of the cumulative effects of PARP1 on AP-1-MMP9 -TGF-␤ signaling pathway (this study) and cGAS-NF-B signaling of proinflammatory responses (our previous study [12]). Future studies in mice with M-and cardiomyocyte-specific knockdown of PARP1 will be needed to examine the relative importance of PARP1 signaling of cGAS-NF-B and AP-1-MMP9-TGF-␤ pathways in CD pathogenesis. Nevertheless, our findings make PARP1 an attractive target for CD therapy and provide an impetus for testing the efficacy of a variety of PARP1 inhibitors in controlling chronic inflammatory infiltrate and cardiac fibrosis, the two hallmarks of CD severity, in multiple mammalian hosts.

MATERIALS AND METHODS
Six-week-old wild-type and Parp1 Ϫ/Ϫ (both male and female) mice were infected with trypomastigotes of T. cruzi strain Sylvio X10 obtained from ATCC (10,000 parasites per mouse) and euthanized at ϳ150 days postinfection. RAW 264.7 (ATCC TIB-71), THP-1, and primary murine M and human cardiac fibroblasts (HCF) were cultured by standard protocols. Myocardial and splenic M were purified using the MagniSort mouse CD11b positive-selection kit. M were incubated with T. cruzi (cell-to-parasite ratio, 1:3), extracellular vesicles (EV) released from the infected cells, and/or various reagents for the time periods stated in the figure legends.
Gene and protein expression. Changes in the expression of target genes was examined by real-time RT-qPCR (22). Protein levels were determined by Western blotting. Zymography was performed to examine the release of MMP2 and MMP9 using Novex 10% Zymogram Plus gels following the manufacturer's instructions (Thermo Fisher Scientific). Images were analyzed by using ImageJ software. Quantitative ELISA was performed to monitor the release of active TGF-␤1 (no. 88-8350; eBioscience), TNF-␣ (no. 88-7324; eBioscience), and MCP3/CCL7 (no. 205571; Abcam) in culture supernatants.
Immunostaining. HCF were incubated with spent cell culture supernatants of THP-1 M incubated with T. cruzi trypomastigotes and various inhibitors. Cells were sequentially incubated with anti-mouse S100A4 and rabbit anti-mouse alpha smooth muscle actin (␣-SMA) primary antibodies and fluorescenceconjugated secondary antibodies, and signals were analyzed on an Olympus BX-15 fluorescence microscope equipped with Simple PCI software (12).
AP-1 transcriptional activity. BM-derived primary M (WT and Parp1 Ϫ/Ϫ ) were incubated with T. cruzi and specific inhibitors, and AP-1 transcriptional activity was monitored using the 96-well TransAM AP-1 family transcription factor assay kit.
Statistical analysis. Data are presented as means and standard deviations (SD) (P Ͻ 0.05). Details are given in the figure legends.
Detailed materials and methods are presented in Text S1.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. TEXT S1, DOCX file, 0.04 MB.

ACKNOWLEDGMENTS
We are thankful to Nandadeva Lokugamage for breeding Parp1 Ϫ/Ϫ mice and Imran Hussain Chowdhury for helping with the protocol for macrophage purification from cardiac tissues.