PARP1-cGAS-NF-κB pathway of proinflammatory macrophage activation by extracellular vesicles released during Trypanosoma cruzi infection and Chagas disease

Trypanosoma cruzi (T. cruzi) is the etiological agent of Chagas cardiomyopathy. In the present study, we investigated the role of extracellular vesicles (Ev) in shaping the macrophage (Mφ) response in progressive Chagas disease (CD). We purified T. cruzi Ev (TcEv) from axenic parasite cultures, and T. cruzi-induced Ev (TEv) from the supernatants of infected cells and plasma of acutely and chronically infected wild-type and Parp1-/- mice. Cultured (Raw 264.7) and bone-marrow Mφ responded to TcEV and TEv with a profound increase in the expression and release of TNF-α, IL-6, and IL-1β cytokines. TEv produced by both immune (Mφ) and non-immune (muscle) cells were proinflammatory. Chemical inhibition or genetic deletion of PARP1 (a DNA repair enzyme) significantly depressed the TEv-induced transcriptional and translational activation of proinflammatory Mφ response. Oxidized DNA encapsulated by TEv was necessary for PARP1-dependent proinflammatory Mφ response. Inhibition studies suggested that DNA-sensing innate immune receptors (cGAS>>TLR9) synergized with PARP1 in signaling the NFκB activation, and inhibition of PARP1 and cGAS resulted in >80% inhibition of TEv-induced NFκB activity. Histochemical studies showed intense inflammatory infiltrate associated with profound increase in CD11b+CD68+TNF-α+ Mφ in the myocardium of CD wild-type mice. In comparison, chronically infected Parp1-/- mice exhibited low-to-moderate tissue inflammation, >80% decline in myocardial infiltration of TNF-α+ Mφ, and no change in immunoregulatory IL-10+ Mφ. We conclude that oxidized DNA released with TEv signal the PARP1-cGAS-NF-κB pathway of proinflammatory Mφ activation and worsens the chronic inflammatory pathology in CD. Small molecule antagonists of PARP1-cGAS signaling pathway would potentially be useful in reprogramming the Mφ activation and controlling the chronic inflammation in CD.


Introduction
Chagas disease (CD) is an inflammatory, dilated cardiomyopathy caused by flagellated protozoa Trypanosoma cruzi (T. cruzi). The infection may be acquired through the vector-borne or transplacental routes, transfusion of contaminated blood components, or from a transplanted organ of an infected donor [1]. Exposure to pathogen results in acute parasitemia associated brief illness that in most cases is resolved without clinical intervention. Several years later, 30% of the infected patients progress into clinically symptomatic, chronic CD when they display cardiac insufficiency due to tissue fibrosis, ventricular dilation, and arrhythmia. Chagas cardiomyopathy continues to result in a loss of 2.74 million disability-adjusted life years, and 15,000 deaths due to heart failure per year [2]. Macrophages (Mφ) are the innate immune cells that play a critical role in modulating the host response to T. cruzi infection [3]. Classically activated Mφ, differentiated through the IL-12/IFN-γ axis, play a critical role in control of T. cruzi infection [4]. It has been documented that parasite killing is triggered in Mφ by autocrine TNF-α secretion. As antigen presenting cells, Mφ also contribute to the activation of Th1 CD4 + T cells and cytolytic CD8 + T cells that are essential for killing the intracellular, replicative form of T. cruzi [5]. A significant presence of Mφ is also noted during the progression of chronic Chagas disease. Stimulus for Mφ proliferation and activation and the role these cells may play in chronic CD is not fully understood [2,6].
Extracellular vesicles (Ev) are small vesicles harboring ligands, receptors, active lipids or RNA/DNA from the cell of their origin [7]. In pathological conditions, a stimulus that triggers Ev formation regulates the selective sorting of constituents and composition of Ev, and consequently, the biological information that they transfer. Recently, it was shown that Ev produced by T. cruzi trypomastigotes (infective form) fuse to host cell membranes and promote Ev release from THP-1 Mφ [8,9]. We have found that human peripheral blood mononuclear cells (PBMC) incubated with T. cruzi secreted Ev and the latter elicited a proinflammatory gene expression profile in human THP-1 Mφ [3]. A proinflammatory cytokine response was also noted when THP-1 Mφ were incubated with Ev isolated from peripheral blood of CD patients [10]. These findings indicate that exposure to T. cruzi influences the host cell Ev release, and the Ev have an impact on the surrounding infected or injured tissue [10]. The mechanism(s) of Ev-dependent Mφ activation and whether this is helpful or harmful to the infected host is not studied.
Poly(ADP-ribose) polymerase 1 (PARP1) is a 113-kDa protein (89-kDa active form) that belongs to the PARP family of seven known and ten putative members, and it accounts for >85% of the PARP activity in cellular systems [11]. PARP1 catalyzes the cleavage of NAD + into nicotinamide and ADP-ribose and uses the latter to synthesize poly (ADP-ribose) (PAR) polymers. The basal level activation of PARP1 by mild genotoxic stimuli causes PARylation of histone proteins (e.g. H1 and H2B) that mediates relaxation of the chromatin superstructure and recruitment of DNA-break repair enzymes, resulting in DNA repair and cell survival [12,13]. PARP1, by direct binding to or PARylation of enhancers and promoters, can also function as a transcriptional co-activator and modulate the expression of self and many other genes. However, excessive activation of PARP1 has been considered pathologic, and linked to a number of cancers, central nervous system disorders, and heart failure [12,14]. Accordingly, in recent years, significant efforts have been devoted to the development and testing of PARP1 targeted therapies. We have found that PARP1/PAR enhanced the mitochondrial production of reactive oxygen species (mtROS) and ROS-dependent NF-κB activation in cardiomyocytes infected by T. cruzi, and over-expression of PARP1/PAR might be of pathologic significance in chronic Chagas disease [15].
In this study, we aimed to determine the role of Ev released during T. cruzi infection and chronic CD in shaping the Mφ response and investigated the signaling mechanisms that mediate Ev-dependent Mφ activation. For this, we isolated Ev from media of cultured T. cruzi trypomastigotes, supernatants of immune and non-immune cells infected with T. cruzi, and plasma of acutely and chronically infected mice. We used cultured Mφ and primary Mφ isolated from bone marrow (BM) of wild type (WT) and Parp1 -/mice and employed classical approaches to evaluate the Ev-PARP1 signaling of Mφ activation. We also fractionated the Ev and used a variety of selective inhibitors to determine the role of DNA-and protein-recognizing innate immune receptors in Ev-PARP1 signaling of Mφ response in Chagas disease. We discuss the benefits of PARP1-targeted therapies in controlling the inflammatory pathology in CD.

Cytokine profile of macrophages elicited by T. cruzi-induced extracellular vesicles (TEv)
We first determined if T. cruzi infection promotes the release of Ev capable of programming Mφ response. For this, we isolated T. cruzi-induced Ev (TEv) from supernatants of Raw 264.7 Mφ at 24 h, 48 h, and 72 h post-infection. A new batch of Mφ were incubated for 48 h with TEv and supernatants were analyzed by an ELISA. TEv isolated at 72 h elicited the maximal activation of TNF-α release, while TEv isolated at 24 h were minimally active in inducing Mφ activation (S1A Fig). No effect of presence or absence of heat-inactivated FBS on TEv-induced Mφ release of TNF-α was observed (S1B Fig). Based on these observations, further experiments were conducted using TEv isolated at 72 h post-incubation.

PARP1 signals cytokine gene expression in Mφ incubated with T. cruziinduced Ev
We have previously shown that a PARP1 inhibitor (PJ34) repressed the gene expression for TNF-α and IL-1β cytokines in cardiomyocytes infected by T. cruzi [14]. PARP1 is found in both cytosolic and nuclear compartments and may influence the proinflammatory cytokine response of Mφ at the gene expression level and/or translational/post-translational level. To sort this out, we isolated primary BM-Mφ from WT and Parp1 -/mice, and incubated with TEv Raw and NEv Raw for 3 h, 18 h, and 48 h. WT BM-Mφ responded to TEv Raw by a potent cytokine gene expression, evidenced by 25-fold, 3.9-fold, and 6.6-fold increase in the Tnf, Il6, and Il1b mRNA levels, respectively, at 3 h post-incubation (vs. NEv, Fig 2A- In agreement with the transcriptional profile, WT BM-Mφ incubated with TEv Raw (± IFNγ) for 48 h exhibited 76-145%, 50-370%, and 46-80% increase in the release of TNF-α, IL-6, and IL-1β cytokines, respectively (Fig 2D-2F, all, + p<0.001). In comparison, a modest increase in cytokines' release was observed in Parp1 -/-BM-Mφ incubated with TEv Raw (± IFN-γ) ( Fig  2D-2F). T. cruzi infection also resulted in a potent increase in proinflammatory cytokines' release by WT ( i p<0.001), but not in Parp1 -/-, BM-Mφ (Fig 2D-2F).
To validate the in vitro findings, we incubated Raw Mφ with plasma TEv of chronically infected WT mice and showed 17.5-fold, 47-fold, and 7.6-fold increase in the secretion of TNF-α, IL-6, and IL-1β cytokines, respectively, as compared to that noted when Mφ were PLOS PATHOGENS PARP1-cGAS-NF-κB pathway in macrophages T. cruzi induced Ev elicit proinflammatory cytokines release in macrophages. Raw 264.7 Mφ and C2C12 muscle cells were incubated with media only or T. cruzi (cell: parasite ratio, 1: 3) for 72 h. Extracellular vesicles (Ev) released in supernatants were isolated as described in Materials and Methods. Next, cultured Mφ were incubated with Ev isolated from supernatants of Raw Mφ (A-C) and C2C12 cells (D-F) for 0-48 h (± 20 ng/mL IFN-γ) and an ELISA was performed to measure the release of TNF-α (A&D), IL-6 (B&E), and IL-1β (C&F) cytokines. Raw Mφ infected with T. cruzi or incubated with media alone were used as positive and negative controls, respectively. NEv: Ev isolated from non-infected cells; TEv: Ev isolated from supernatants of T. cruzi-infected cells. Data are representative of � 2 independent experiments (three biological replicates per treatment, duplicate observations per sample) and plotted as mean value ± SD. Significance is annotated as + TEv vs. NEv and � TEv+IFN-γ vs TEv, and p values of � 0.05, � 0.01, and � 0.001 are marked with one, two, and three symbol characters, respectively.
Together the results presented in Fig 2 and S3 Fig demonstrate that a) Ev produced during T. cruzi infection and chronic Chagas disease induce cytokines' gene expression and synergize with IFN-γ to elicit the proinflammatory cytokines' release in supernatants of primary and cultured Mφ. Further, b) chemical inhibition or genetic deletion of PARP1 arrested the transcriptional (and translational) activation of proinflammatory cytokine response in Mφ incubated with T. cruzi-induced Ev. We surmise that PARP1 is an essential transcriptional regulator that transmits the stimulus provided by Ev produced during Chagas disease to signal the proinflammatory cytokines' expression.  Fig 3D). Ev samples ranged from 70-260 nm in size with mean value of 128.7-137.2 nm and were within the microvesicles size range. Likewise, TcEv and TEv distribution per mL ranged from 1.5 1E+7-8.7 1E+7 (mean value: 3.51 1E+7-4.19 1E+7), thus suggesting that overall concentration used for NTA were similar for all samples. Extracellular vesicles harbor nucleic acids and proteins from the cell of their origin and may also uptake other components during transit from the site of their origin to secretory pathway. We performed western blot analysis to examine macrophage markers and parasite protein content in Ev (Fig 3F & 3G). CD11b (Mφ marker) was present in TEv released by infected cells and mice ( Fig 3F, top panel) similar to that noted in total lysates of Raw Mφ and splenic cells. Low, but detectable CD11b signal was also noted in NEv Raw . CD68 (marker of hematopoietic cells of monocytic lineage) was primarily detected in TEv mice ch and splenic lysate ( Fig 3F). Probing with polyclonal anti-T. cruzi sera detected several protein bands in TEv RAW and TEv mice ch , which were closely matched to those noted in T. cruzi lysate ( Fig 3G). Except for a 75 kDa band, no reactivity of T. cruzi polyclonal sera was noted with NEv Raw .

Compositional analysis of extracellular vesicles produced by T. cruzi infection
We examined the DNA content of Ev samples by qPCR and traditional PCR. Representative gel images and average data from > 3 experiments plotted in bar graphs are shown in Fig

Components of TEv that elicit proinflammatory response in Mφ (± PARP1)
To determine if DNA or proteins carried by TEv stimulate PARP1-mediated cytokine response, we treated TEv with DNase I and protease, and used the treated samples to purify TEv protein and TEv DNA fractions, respectively. We confirmed that DNase I treatment degraded DNA in all Ev samples ( Fig 4A) but it did not interfere with protein content of TEv ( Fig 4B). Similarly, treatment with protease specifically resulted in significant reduction in protein bands intensity and number in TEv ( Fig 4B).
Next, we incubated the cultured Mφ with TEv and TEv DNA and TEv protein fractions (± iniparib), and examined the cytokines' gene expression by RT-qPCR at 3 h and cytokines' release by an ELISA at 48 h. Mφ incubated with DNase I-treated, TEv protein fractions exhibited 14-15.5-fold and 13-14.8-fold decline in Tnf and Il6 mRNA levels, respectively ( (Fig 4H, all, i p<0.05). Together these results suggest that DNA (but not protein) contents of TEv produced by T. cruzi infection provide the major stimulus for the activation of proinflammatory cytokine response in Mφ in a PARP1-dependent manner. NEv isolated from non-infected cells and mice carried similar amounts of host DNA as was noted in TEv (S4 Fig) and these NEv were non-inflammatory (Figs 1-3). These data suggest that parasite DNA, and likely not the host DNA, carried by TEv elicit the Mφ proinflammatory activation. However, these data do not rule out the potential role of epigenetically modified parasite and/or host DNA in signaling proinflammatory response, and this will be determined in future studies.

plots (D&E). (F&G) Protein markers by western blot analysis.
Purified TEv Raw , TEv mice ch and controls (NEv Raw , and Mφ, splenic, and T. cruzi lysates, 10 μg) were resolved by SDS-PAGE. Representative western blot images show CD11b, CD68 and, GAPDH levels (F) and reactivity to anti-T.cruzi polyclonal sera from infected mice (G). (H&I) PCR analysis of DNA in Ev. Total TEv DNA was purified from TEv isolated from supernatants of infected Raw Mφ and plasma of acutely (ac) and chronically (ch) infected WT and Parp1 -/mice. Real-time qPCR was performed to amplify T. cruzi-specific 18SrDNA and murine Gapdh sequences. The products of qPCR were resolved by 1.5% agarose gel electrophoresis (H, panel a). Tc18SrDNA levels normalized to mGapdh are presented as fold change ± SD (I, panels a-c) with two biological replicates each and triplicate observations per sample for panel a and n = 5 for panels b & c (vs. matched control NEv, + p<0.001). The conserved (330 bp) and variable (320 bp) regions of T. cruzi kinetoplast DNA minicircle were amplified by traditional PCR for 40 cycles and resolved by 1.5% agarose gel electrophoresis (I). NEv DNA purified from NEv of non-infected Mφ and mice were used as controls.

Signaling receptors involved in TEv-mediated up regulation of proinflammatory cytokine response in Mφ
Mφ express a variety of pattern recognition receptors (PRR) to recognize pathogen-and damage-associated molecular patterns (PAMPs and DAMPs) to signal the immune activation cascade. Among these, cytoplasmic toll-like receptors TLR3 and TLR7 recognize the singlestranded or double-stranded RNA and TLR9 recognizes DNA and can play a key role in activation of innate immune system. Cyclic GMP-AMP synthase (cGAS) is suggested to recognize genomic DNA damage and trigger innate immune responses through cGMP-mediated activation of STING adaptor protein [16]. To delineate whether TLRs or cGAS recognize T. cruziinduced Ev to signal the downstream cascade for cytokines gene expression, we incubated the cultured Mφ with TEv Raw in presence of specific inhibitors of TLR3/7/9, cGAS, and NF-κB transcription factor for 3 h or 18 h, and monitored the cytokines' gene expression by RTqPCR. As expected from Fig 4, incubation of Mφ with TEv Raw elicited a potent increase in cytokines' gene expression at 3 h and 18 h (vs. NEv control, all, p<0.001). Co-incubation with chloroquine (inhibits endosomal TLR3/7/9), quinacrine (inhibits TLR3/9) and ODN-2088 (specific inhibitor of TLR9) decreased the TEv-induced expression of Tnf and Il6 by 35-75% at 3 h and 52-95% at 18 h post-incubation (Fig 5A & 5B). Among these, ODN-2088 inhibitor of TLR9 was most effective in suppressing the Tnf and Il6 expression in TEv-stimulated Mφ (Fig 5A &  5B, all, + p<0.01). In comparison, short-term treatment with cGAS antagonist (PF-06928215) was sufficient to cause a potent decline in TEv-induced cytokines' expression in Mφ. This was evidenced by 230-fold, 148-fold, and 2.5-fold decline in Tnf, Il6, and Il1b mRNA levels, respectively, at 3 h post-incubation (Fig 5A-5C, all, + p<0.001) that was not further changed at 18 h ( Fig 5A-5C). In presence of 5 μM JSH-23 (inhibits NF-κB transcriptional activity), the expression of Tnf and Il6 was completely abolished in TEv-stimulated Mφ at 3 h and 18 h post-incubation ( Fig 5A & 5B, all, + p<0.001). NF-κB inhibition did not alter the TEv-induced Il1b gene expression in Mφ (Fig 5C). Together, these results suggest that a) TLR9 and cGAS signal cytokines gene expression (Tnf and Il6 >> Il1b) through NF-κB activation in TEv-treated Mφ. The pronounced inhibition of cytokines' expression by PF-06928215 (vs. ODN-2088) at 3 h suggests that cGAS-STING (and not TLR9-MyD88) pathway is the early responder in recognizing TEv stimulus and signaling Mφ proinflammatory cytokine gene expression.

Synergistic role of cGAS and PARP1 in signaling NF-κB activity in Mφ (± TEv)
Since both cGAS and PARP1 are activated by DNA damage, we first evaluated the oxidized DNA content in Ev produced by T. cruzi infection of cells and mice. TEv released in supernatants of T. cruzi-infected Mφ and in plasma of chronically infected WT and Parp1 -/mice exhibited 2.2-fold, 2.25-fold, and 16.4-fold increase in the 8-OHdG contents, respectively, in comparison to matched NEv controls (Fig 6A, all, p<0.05). The Ev of chronically infected treated TEv DNA fractions (± 5 μM iniparib) for 3 h or 48 h. The mRNA levels (3 h) and protein levels (48 h) of TNF-α, IL-6 and IL-1β cytokines were determined by real time RT-qPCR (C-E) and ELISA (F-H), respectively. Data in bar graphs are representative of � 2 independent experiments (two biological replicates per treatment and triplicate observations per sample for RT-qPCR analysis; and three biological replicates per treatment and duplicate observations per sample for ELISA), and presented as mean ± SD. Statistical significance is annotated as + (TEv vs. TEv DNA or TEv protein fractions) and i (effect of iniparib on TEv-induced responses); and p values of � 0.05, � 0.01, and � 0.001 are presented by one, two, and three symbol characters, respectively. Horizontal bars denote the compared groups. ns: non-significant.  Then, we performed a dual luciferase reporter assay to determine if cGAS and PARP1 independently or synergistically signal NF-κB activation in Mφ. For this, Raw Mφ were transiently transfected with NFκB-Luc reporter plasmid and pREP7-Rluc (transfection efficiency control), and then incubated for 3 h with TEv isolated from supernatants of T. cruzi-infected cells, or from plasma of chronically infected WT and Parp1 -/mice. Mφ were incubated with Ev in presence of 5 μM and 10 μM concentrations of PF-06928215 (cGAS inhibitor) and/or iniparib (PARP1 inhibitor) to understand the role of these DNA sensing molecules in signaling NFκB transcriptional activation. Mφ incubated with TEv Raw (vs. NEv Raw ) exhibited 6.7-fold increase in NF-κB-luciferase activity (normalized to Renilla luciferase, p<0.001) that was inhibited by 65-80% and 60-75% in presence of cGAS inhibitor and PARP1 inhibitor, respectively (Fig 6B,   + p<0.05-0.01). In comparison, co-treatment with cGAS and PARP1 inhibitors (5 μM each) resulted in >85% inhibition of NF-κB activity in TEv Raw -stimulated Mφ (Fig 6B, + p<0.001). Likewise, TEv of chronically infected WT mice elicited 6-fold increase in NF-κB-luciferase activity (vs. NEv of uninfected mice, p<0.001), and TEv WT ch -induced NFκB activity was inhibited by 53-75% and 55-67%, respectively, in presence of cGAS and PARP1 inhibitors (Fig 6C, all, + p<0.05-0.01); and by >80% when TEv WT ch -stimulated Mφ were treated with both inhibitors (5 μM each, Fig 6C, + p<0.001). Unexpectedly, TEv isolated from peripheral blood of chronically infected Parp1 -/mice were also proinflammatory, evidenced by 6.5-fold increase in NF-κB-luciferase activity in comparison to that noted in Mφ incubated with NEv of Parp1 -/mice (Fig 6D, p<0.001). As above, we noted 65-75% and 55-75% decline in TEv-  (Fig 6D, p<0.001). Together, these results suggest that a) TEv released in response to T. cruzi infection and chronic disease carry oxidized DNA, and b) Mφ uptake of TEv carrying oxDNA is sensed by cellular DNA response element cGAS to signal NF-κB transcriptional activation. The findings that TEv Parp1 -/-ch carry similar amount of TcDNA and murine mtDNA as was noted in TEv WT ch (Fig 3H & 3I), and TEv Parp1 -/-ch also signaled NFκB activity in Mφ, while PARP1 inhibitor prevented the TEv WT ch-induced NF-κB activity, we surmise that c) PARP1 is not required for the generation of TEv of proinflammatory phenotype in CD. Instead, d) PARP1 synergizes with cGAS in signaling the NF-κB transcriptional activity in TEv stimulated Mφ.

Tissue inflammatory infiltrate in WT and Parp1 -/mice
Macrophage uptake of inflammatory TEv produced during chronic infection can sustain persistent inflammation, a key cause for left ventricular dysfunction in Chagas disease. We, therefore, first determined if inhibition of PARP1 would arrest chronic inflammation in Chagas disease. Histological evaluations showed the extent of inflammatory infiltrate in heart tissue of chronically infected WT mice (Fig 7D,

Discussion
Early studies have shown that parasite proteins are transferred from infected muscle, neuronal, epithelial, and fibroblast cells to uninfected host cells, though this antigen transfer was not observed in lymphocytes and erythrocytes [17]. Recent literature documented that this jumping of T. cruzi antigens occurs through the release of membrane vesicles (also called extracellular vesicles). It was shown that T. cruzi sheds compositionally different Ev depending on the developmental stage and virulence of the parasite strain [18,19], Ev shed by infective trypomastigote form of the parasite have high fusogenic potential with the host cell membranes [20], and contact with infective forms of the parasite also stimulated Ca 2+ -dependent shedding of membrane vesicles from THP-1 Mφ [21]. These and other studies did not explore the signaling cascades by which T. cruzi stimulates formation of membrane vesicles within itself or on the host cell membranes, though it was proposed that the host and parasite Ev may maintain cellular activation in CD [7]. Indeed, we recently showed that Ev released by human PBMCs exposed to T. cruzi infection activated a proinflammatory gene expression profile in THP-1 Mφ [3], a finding that strongly suggested that T. cruzi influences the host cell juxtacrine/paracrine Ev release and impacts the surrounding infected/non-infected cells and tissues. Herein, we provide the first evidence that Ev are released from immune and non-immune cells during T. cruzi infection and chronic CD progression, and these T. cruzi-induced Ev Seropositive individuals are categorized as indeterminate (C0) when they exhibit no symptoms of heart involvement, and progress through C1-C3 stages of clinical Chagas disease presented with increasing severity of structural and functional alterations in the heart [22,23]. While the indeterminate clinical form is biased towards an anti-inflammatory profile, the C1-C3 CD patients routinely present proinflammatory profile [6,[28][29][30] [6,[24][25][26] associated with a systemic increase in TNF-α + monocytes [27][28][29], oxidative stress (e.g. lipid hydroperoxides) [30,31], and an abundance of CD8 + T cells that express inflammatory cytokines and cytotoxic molecules [32]. Microscopic examination of tissues also routinely shows that T. cruzi infection causes extensive myocardial damage, characterized by vacuolization, myocytolysis, and myofibrillar degeneration and these changes are invariably associated with intense infiltration of macrophages followed by lymphocytes during acute infection. While parasite burden is controlled, tissue mononuclear cells remain active during chronic CD (reviewed in [2]). Our findings of T. cruzi DNA and proteins in TEv, and TEv stimulation of proinflammatory Mφ that are also excellent antigen presenting cells (APC) suggest that T. cruzi antigens carried by TEv might also shape the APC-T cell dependent response in patients with different forms of CD. If this is proven in future studies, it will solve the decades old question of how the stimulus is provided for chronic inflammatory state in CD and provide important information regarding the TEv's association with clinical disease progression.
While Mφ are the major innate immune cells that exert trypanocidal effects by producing ROS and NO, we have found that non-immune cells (skeletal muscle, cardiomyocytes) also respond to T. cruzi by increased release of ROS of mitochondrial origin [14]. The ROS/NO exert cytotoxic effects through oxidation of cellular components including DNA, proteins, and lipids and are not discriminatory of the parasite and the host cells [33]. Indeed, we have shown that 8-hydroxy-2'-deoxy guanosine (8-OHdG, marker of DNA oxidative damage) was enhanced in T. cruzi-infected cardiomyocytes [14] and myocardium of chronically infected Chagas mice and patients [34]. In this study, we demonstrate that these damaged DNA fragments are encapsulated in Ev (Fig 6) and provide stimulus for proinflammatory activation of macrophages (Fig 4). Indeed, Ev of different organisms have been described as PAMPs and promoters of the innate and adaptive immune responses [35]. The Ev shed by axenic cultures of T. cruzi were enriched in glycoproteins of the gp85/trans-sialidase (TS) superfamily and other α-galactosyl (α-Gal)-containing glycoconjugates, and stimulated TLR2, proinflammatory cytokines (TNF-α and IL-6), and NO in Mφ [18]. Our data provide first evidence that Ev produced during T. cruzi infection and chronic Chagas disease are DAMPs that promote NF-κB-mediated proinflammatory cytokines' production through the engagement of DNA-sensing innate immune receptors (Fig 5). While TLR9 is usually activated by unmethylated CpG sequences in ssDNA molecules, cGAS has emerged as a major sensor of genomic dsDNA damage and it elicits innate immune responses through cGMP-mediated activation of STING adaptor protein [16]. Our finding of an early and potent role of cGAS (than TLR9) in eliciting TEv DNA -dependent Mφ activation allows us to propose that genomic DNA damage of parasite and host cells (instead of CpG DNA content in the genomic DNA of the parasite or host cells) serve as the primary stimulus in engaging DNA sensing innate immune receptors and Mφ activation in the context of Chagas disease progression (Fig 5).
The catalytic activity of PARP1 promotes post-translational modification of self and a range of other proteins, and it is believed to be crucial for mediating multiple DNA damage repair pathways. PARP1 is also expressed by T. cruzi [36]; and PARP1 expression was increased in human cardiomyocytes [14] and in the myocardium of mice infected by T. cruzi [15]. PARP1 chemical inhibition or genetic deletion preserved the left ventricular function that otherwise PLOS PATHOGENS PARP1-cGAS-NF-κB pathway in macrophages was compromised in Chagas WT mice [15]. Our findings in the present study show that Parp1 deletion was beneficial in controlling the myocardial inflammatory infiltrate, especially the TNF-α-expressing Mφ, in Chagas disease (Figs 7 and 8). These studies imply that PARP1 contributes to Chagas cardiomyopathy through its effects on cardiomyocytes and Mφ. In cardiomyocytes, PARP1 was activated in response to T. cruzi induced mtROS/DNA damage, and PARP1 facilitated the assembly of the NF-κB transcription complex and cytokine gene expression through post-translational modification of RelA (p65)-interacting nuclear proteins [14]. In this study, we provide evidence that Ev oxDNA released in supernatants of infected cells and in plasma of chronically infected mice stimulate PARP1 activation in Mφ. Further, Mφ-PARP1 complemented the cGAS in stimulating the NF-κB-dependent cytokine gene expression in response to Ev oxDNA produced during T. cruzi infection. In this context, PARP1 likely served as a cytoplasmic sensor along with cGAS to activate innate signaling cascade (Fig 6). A recent study showed that while phosphorylation of cGAS at Tyr215 (by B-lymphoid tyrosine kinase) facilitates its cytosolic retention, DNA damage induced nuclear translocation of cGAS occurs in importin-α-dependent manner, and in the nucleus cGAS interacted with PARP1 and impeded the formation of the PARP1-Timeless complex and suppressed the homologousrecombination-mediated DNA repair [37]. Whether cGAS directly (or indirectly) interacts with PARP1 in cytosolic and/or nuclear fraction to stimulate Ev oxDNA -dependent, NF-κBmediated proinflammatory response in CD remains to be seen in future studies. However, we surmise that PARP1 is a potential target for controlling chronic inflammatory pathology and Chagas cardiomyopathy. Our proposal is supported by the findings that chemical inhibition or genetic deletion of PARP1 significantly decreased the myocardial inflammatory infiltrate (specifically the macrophages of proinflammatory phenotype) and improved the left ventricular function in Chagas mice.
In summary, we have shown that TcEv released by T. cruzi and TEv released by host during T. cruzi infection and disease progression shape the activation of Mφ that, in turn, augment the chronic proinflammatory state in Chagas disease. Damaged DNA fragments encapsulated within T. cruzi-induced Ev, are the key biocomponent that promote NF-κB-mediated proinflammatory cytokine production in Mφ through the engagement of cytosolic DNA sensors cGAS and PARP1. Whether TEv also provide antigenic stimulus and PARP1-cGAS induce antigen presenting capacity of the Mφ and support chronic activation of cytotoxic CD8 + T cells that are known to be pathologic in human Chagas disease remains to be determined in future studies. We propose that small molecule PARP1 inhibitors offer a potential therapy for controlling the pathologic chronic inflammation in Chagas disease through modulation of the Mφ signaling of cGAS-NF-κB pathway.

Ethics statement
All animal experiments were performed by following the NIH guidelines for Care and Use of Experimental Animals, and in accordance with protocols approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch, Galveston (protocol number: 0805029).

PARP1-cGAS-NF-κB pathway in macrophages
For cell cultures, fetal bovine serum (FBS, Invitrogen, Carlsbad, CA) was heat inactivated at 56˚C for 30 minutes with intermittent shaking before use in medium. Murine bone marrow (BM) cells were isolated from the femurs of mice by following a standard protocol, and either used immediately or stored at -80˚C in 80% heat-inactivated FBS (ha-FBS) with 20% DMSO. The BM cells were suspended in RPMI medium containing 10% ha-FBS, 2-mmol/L glutamine, 100 IU/mL penicillin, 100-μg/mL streptomycin (Corning, Corning, NY), added to 6-well plates (5 x 10 6 cells per ml per well), and incubated at 37˚C in 5% CO 2 in presence of 20 ng/ mL of macrophage colony stimulating factor (M-CSF; Millipore, Burlington MA) [38]. The culture medium and M-CSF were replenished every two days, and cells were incubated for nine days allowing the monocyte progenitor cells to mature as Mφ. Raw 264.7 murine Mφ (ATCC TIB-71) were cultured in complete high glucose Dulbecco's modified Eagle's medium (DMEM) containing 10% ha-FBS. The C2C12 mouse myoblast cells (ATCCC CRL-1772) were cultured in complete RPMI 1640 medium containing 5% or 10% ha-FBS.
T. cruzi (SylvioX10/4, ATCC 50823) trypomastigotes were propagated by in vitro passage in C2C12 cells. All chemicals used in the study were of molecular grade and purchased from Sigma-Aldrich unless otherwise specified. Protein levels in the samples were determined by using the Bradford Protein Assay (Bio-Rad, Hercules, CA).

Generation, isolation, and fractionation of extracellular vesicles (Ev)
The differential centrifugation protocol for the enrichment of Ev consisting microvesicles (100-1000 nm) and apoptotic bodies (1000-5000 nm) is described previously [39]. Briefly, T cruzi trypomastigotes (1X 10 7 /10 mL) were incubated in serum free RPMI media for 72 h at 37˚C at 5% CO 2 . Cells (C2C12 or Raw 264.7) were seeded in T25 or T75 cell culture flasks, and at 70% confluency, infected with T. cruzi trypomastigotes (cell: parasite ratio: 1:3). Cells were incubated at 37˚C at 5% CO 2 for 24 h, 48 h, or 72 h in serum free RPMI or DMEM medium. The culture supernatants were centrifuged at 4000 g for 10 minutes to pellet the cell debris and parasites. Then culture supernatants were subjected to three series of centrifugation at 4˚C for 30 min each at 20,000 g. The pelleted Ev samples from each centrifugation were washed, resuspended at 10-fold concentration in serum free RPMI medium, and stored at -80˚C.
Mice (WT and Parp1 -/-, 6-weeks old, n = 10 per group) were infected with T. cruzi trypomastigotes (10,000/mouse), and euthanized at 30 days and 150 days post-infection (pi) corresponding to acute parasitemia and chronic disease phase, respectively [40,41]. The EDTA blood samples were centrifuged for 20 minutes at 2000 g to separate plasma. Plasma samples were centrifuged at 4˚C for 30 min at 20,000 g, and the pelleted Ev samples were washed and stored at 10-fold concentration in serum free RPMI medium, as above.
For fractionation, Ev samples were treated with DNase I (1 U/μL, EN0521, Thermo Fisher Scientific) at 37˚C for 30 min to degrade Ev-associated membrane bound contaminant DNA followed by inactivation of DNase with EDTA. Then Ev samples were treated with 0.5% Triton X-100 in 1X PBS for 10 min at 25˚C. The permeabilized Ev samples were incubated for 5 min each with protease cocktail (0.03 U/g, P-311, Sigma-Aldrich) to degrade proteins and 1 mM phenylmethanesulfonyl fluoride (PMSF) to inactivate the proteases, and Ev DNA was extracted by using DNeasy blood and tissue Kit (Catalog; 69504, Qiagen, Hilden, Germany). In other studies, permeabilized Ev samples were sequentially treated with DNase I (1 U/μL, EN0521, Thermo Fisher Scientific) at 37˚C for 30 min to degrade Ev DNA and 50 mM EDTA for 10 min to inactivate DNase I and used as a source of Ev protein fraction. In all cases, Ev and Ev fractions were stored at -80˚C at 10-fold concentration of the original volume and used at 1:10 ratio (Ev: culture medium, v/v) to obtain biological levels.

Size, distribution and molecular characterization of Ev
To quantify size and distribution, Ev purified from different samples were subjected to Nanoparticle Tracking Analysis (NTA) by using PMX-120-12B R2 ZetaView (Particle Metrix, Meerbusch, Germany). Briefly, the Brownian motion of each Ev particle was visualized by a laser PLOS PATHOGENS PARP1-cGAS-NF-κB pathway in macrophages light scattering method (at 488 nm) and tracked over 30-45 sec to calculate particle size and concentration. Each measurement was performed for two cycles, scanning 11 cell positions and capturing 60 frames per position per cycle with camera sensitivity 90 volt/μ joule/cm 2 , shutter time 70 milli-sec. The videos were analyzed by ZetaView Software 8.05.05. SP2. Ev samples were diluted 1:10-1:20 in 1X PBS to ensure that the concentration and size distribution of Ev in each sample was optimal for ZetaView analysis (range: 30-60 particles/frame, 15-2000 nm size).
To determine the origin of Ev DNA, real time qPCR and traditional PCR were performed. Real time qPCR was performed as described above using 2 μL of Ev DNA and oligonucleotides pairs to amplify murine (COII, Cytb, Gapdh) and T. cruzi (Tc18SrDNA) DNA sequences. Traditional PCR was performed in a 25 μL reaction containing 2.5 μL of Ev DNA , 12.5 μL Go Taq Green master mix (M7122, Promega) and 20 μM of T. cruzi kDNA (kinetoplast DNA, minicircle)-specific oligonucleotides. The cycling program included an initial denaturation at 95˚C for 2 min, 40 cycles of 95˚C for 30 sec, 57˚C for 30 sec, 72˚C for 30 sec, and a final extension at 72˚C for 5 min. All oligonucleotides used for qPCR and traditional PCR are listed in S1 Table. The Ev DNA fractions and qPCR and traditional PCR products were resolved on 1.5% agarose gels, stained with 1 μg/mL of ethidium bromide and imaged using Fluor Chem HD2 UV transilluminator (Protein Simple, San Jose, CA).
The levels of 8-hydroxy-2'-deoxy guanosine (8-OHdG, ubiquitous marker of oxidative DNA damage) in Ev DNA fractions from normal and infected cells and mice (WT and Parp1 -/-) were measured by using an 8-OHdG DNA Damage ELISA kit (STA320, Cell Biolabs, San Diego, CA). For this, Ev DNA samples (1 mg/mL) were denatured at 95˚C for 5 min and digested with nuclease P1 at 37˚C for 2 h to form nucleosides. Samples were then treated with 5U alkaline phosphatase for 1 h at 37˚C, centrifuged at 6000 g for 5 min, and supernatants containing Ev DNA fragments were used in an ELISA. Then, 50 μL of supernatant containing Ev DNA fragments were added in triplicate to 96-well plates, and plates were incubated at room temperature for 1 h each with 50 μL of anti-8-OHdG antibody (1:500 dilution) and 100 μL of HRP-conjugated secondary antibody (1:1000 dilution). The color was developed with TMB substrate and change in absorbance was recorded at 450 nm by using a Spectra Max M2 microplate reader (Molecular Devices, Sunnyvale, CA). Standard curve was prepared by using 8-OHdG (100 pg-20 ng/ mL).
To examine the protein content, Ev samples were subjected to protein extraction with 1X RIPA buffer. Ev and Ev protein fractions (10 μg) were electrophoresed on a 10% polyacrylamide gel by using a Mini-PROTEAN electrophoresis chamber (Bio-Rad). Gels were stained with Coomassie blue and imaged using an Image Quant LAS4000 system (GE Healthcare, Pittsburgh, MA). For Western blotting, proteins were transferred to PVDF membrane using a Criterion Trans-blot System (Bio-Rad) and membranes were blocked for 2 h with 20 mM Tris-HCl (pH 7.4), 136 mM NaCl, 0.1% Tween 20 (TBST) containing 0.5% BSA. Membranes were incubated overnight at 4˚C with primary antibody to Mφ markers CD11b (ab133357, 1: 1000 dilution) and CD68 (ab31630, 1:1000 dilution), GAPDH (ab9485, 1:2500 dilution, loading control) and polyclonal sera (1: 50 dilution) from chronically infected mice. All antibodies were purchased from Abcam, MA, USA; and dilutions were made in TBST-0.5% BSA. Membranes were washed with TBST (three times at each step) and incubated for 1 h with HRP-conjugated secondary antibody (1:5000 dilution, Southern Biotech, Birmingham AL). Color was developed by pierce ECL western blot substrate, images were acquired as above, intensity analysis of protein bands was performed by using Image J software (NIH, Bethesda, MD).

Transfection and NFκB activity by dual luciferase assay
Transfection and dual luciferase assays were conducted by using a Transfection Collection NFκB Transient Pack (79268, BPS Biosciences, San Diego, CA). Briefly, Raw Mφ (30,000 cells/ 100 μL BPS medium) were seeded in 96-well, clear bottom, tissue culture plates, and allowed to adhere for 24 h. For transfection, 1 μL of NFκB reporter (consists NFκB reporter vector + constitutively expressing Renilla luciferase vector) or negative control reporter (non-inducible luciferase vector + Renilla luciferase vector) were diluted in 15 μL of Opti MEM I medium, mixed with 0.35 μL of Lipofectamine 2000, and added to each well [19]. After incubation for 24 h at 37˚C / 5% CO 2 , cells were replenished with fresh BPS medium. Then cells were loaded with Ev isolated from supernatants of normal and infected Mφ or from plasma of non-infected and chronically infected WT and Parp1 -/mice. Cells were incubated with Ev for 3 h in the presence or absence of 5 μM and 10 μM of iniparib (PARP1 inhibitor) and/or PF-06928215 (cGAS inhibitor). To measure luciferase activity, equal volumes of firefly luciferase followed by renilla luciferase working solution provided in the Dual Luciferase Assay System (BPS Biosciences) were added, and the release of luminescence was recorded by using a Glomax 96 microplate luminometer (Promega, Madison, WI). The relative luminescence for NFκB reporter (firefly luciferase) was normalized to renilla luciferase (determines transfection efficiency).

Histology
Heart tissue sections of chronically infected WT and Parp1 -/mice were fixed in 10% buffered formalin for 24 h, dehydrated in absolute alcohol, cleared in xylene, and embedded in paraffin. Paraffin-embedded 5-micron tissue-sections were stained with hematoxylin and eosin (H&E) and evaluated by light microscopy. Tissue section slides (three mice per group, at least two slides per tissue, ten microscopic fields per slide) were analyzed by light microscopy, and the presence of inflammatory cells was scored as (0)-absent/none, (1)-focal or mild with � 1 foci, (2)-moderate with � 2 inflammatory foci, (3)-extensive with generalized coalescing of inflammatory foci or disseminated inflammation, and (4)-severe with diffused inflammation, interstitial edema, and loss of tissue integrity [44].

Statistical analysis
All experiments were repeated at least twice. In general, in vitro experiments were conducted with duplicate or triplicate biological replicates per group with two or three observations per sample per experiment. Murine samples (n = 10 per group for Ev analysis and n = 3 per group for histology studies) were analyzed in duplicate. All data were analyzed by using an InStat version 5 (GraphPad, La Jolla, CA) software. Mean values were compared by unpaired Student's two tailed t-test (for comparison of two groups) and one-way analysis of variance (ANOVA) with post hoc correction test (for comparison of multiple groups). Data are presented as mean ± standard deviation (SD). A p value of < 0.05 was considered as minimum level of significance for the comparison of minimum two variables.

PLOS PATHOGENS
PARP1-cGAS-NF-κB pathway in macrophages C2C12 muscle cells were incubated with media only or T. cruzi (cell: parasite ratio, 1: 3) for 72 h, and Ev were isolated from supernatants of normal (NEv) and Tc-infected (TEv) cells. Next, Raw Mφ were incubated with NEv C2C12 or TEv C2C12 in presence or absence of 20 ng/mL IFNγ and 5 μM iniparib (inib, selective PARP1 inhibitor) for 48 h, and release of TNF-α, IL-6, and IL-1β cytokines was monitored by an ELISA. Mφ incubated with T. cruzi and IFN-γ (± iniparib) were used as controls. Data are representative of � 2 independent experiments (2-3 biological replicates per treatment, and 2-3 observations per sample) and presented as mean ± SD. Horizontal bar indicates the compared groups. Statistical significance is captured with + NEv vs. TEv, � effect of IFN-γ on TEv, and i effect of Parp1 knockdown on TEv+IFN-γ. The p values of � 0.05, � 0.01, and � 0.001 are presented by one, two, and three symbol characters, respectively. Horizontal bar indicates the compared groups. Myocardial tissue sections of non-infected and infected mice were subjected to immunohistochemistry staining. Shown is the myocardial expression of CD11b, presented as semiquantitative immunohistochemistry quick score ± SD (n = 3 mice per group, two tissue sections per mouse, 9 microscopic fields per tissue section, 20X magnification). Significance is annotated as +++ infected vs. non-infected (p<0.001) and �� WT.Tc vs. Parp1 -/-.Tc (p<0.01). (TIF) S1 Table. Oligonucleotides used in this study. (DOCX)