MyD88 activation in cardiomyocytes contributes to the heart immune response to acute Trypanosoma cruzi infection with no effect on local parasite control

Cardiomyopathy is the most serious consequence of Chagas disease, a neglected human disorder caused by Trypanosoma cruzi infection. Because T. cruzi parasites invade cardiomyocytes, we sought to investigate whether these cells recognize the parasite in vivo by receptors signaling through the MyD88 adaptor, which mediates the activation pathway of most Toll-like receptors (TLRs) and IL-1/IL-18 receptors, and influence the development of acute cardiac pathology. First, we showed that HL-1 cardiac muscle cell line expresses MyD88 gene and protein at resting state and after T. cruzi infection. To evaluate the role in vivo of MyD88 expression in cardiomyocytes, we generated Mer+MyD88flox+/+ mice in which tamoxifen treatment is expected to eliminate the MyD88 gene exclusively in cardiomyocytes. This Cre-loxP model was validated by both PCR and western blot analysis; tamoxifen treatment of Mer+MyD88flox+/+ mice resulted in decreased MyD88 gene and protein expression in the heart, but not in the spleen, while had no effect on littermates. The elimination of MyD88 in cardiomyocytes determined a lower increase in CCL5, IFNγ and TNFα gene transcription during acute infection by T. cruzi parasites of the Y strain, but it did not significantly modify heart leukocyte infiltration and parasitism. Together, our results show that cardiomyocytes can sense T. cruzi infection through MyD88-mediated molecular pathways and contribute to the local immune response to the parasite. The strong pro-inflammatory response of heart-recruited leukocytes may overshadow the effects of MyD88 deficiency in cardiomyocytes on the local leukocyte recruitment and T. cruzi control during acute infection.


Introduction
Chagas disease (American trypanosomiasis) is an illness caused by the flagellated protozoan Trypanosoma cruzi that affects 6-7 million people, mostly in Latin America [1]. One century after its discovery [2], Chagas disease remains an important socioeconomic and public health problem, representing the main cause of acute and chronic myocarditis in endemic countries [3]. The absence of spontaneous cure and limited therapeutic arsenal against T. cruzi, as well as the difficulty of eliminating the kissing bug vector, which is responsible for the main form of transmission [4], constitute aggravating factors.
In the first weeks after T. cruzi infection, a parasite-specific immune response is developed in the vertebrate host, in which the antibodies and T cells determine the control of acute parasitemia. This is, however, a non-sterile control and, consequently, small numbers of T. cruzi parasites persist for the life of the host. Although the majority of chronic patients remain asymptomatic, nearly 5% and 30% develop digestive and heart pathology, respectively. The most severe consequence is chronic chagasic cardiopathy (CCC) [5], a complication that yields high mortality rates due to arrhythmias and/or cardiac insufficiency [6]. Previously described as an autoimmune disease [7][8], heart pathology in CCC is currently considered the result of parasite persistence in the cardiac tissue [9][10][11][12]. In this context, heart tissue lesions result from cellular destruction caused by both the parasite and the local immune response [13][14].
The innate immune response is the first line of defense against T. cruzi, in which the main cell populations involved in parasite detection and destruction are macrophages, dendritic cells and natural killer cells [15]. Among the pattern-recognition receptors (PRRs) involved in parasite recognition by mononuclear phagocytes, Toll-like receptors (TLRs), and more specifically TLR2, 4, 7 and 9, play a major role [16][17][18][19][20]. TLR interaction with their ligands activates different molecular pathways that culminate in nuclear translocation of nuclear factor-κB (NF-κB) and activator protein 1 (AP-1) transcription factors, which stimulate the transcription of genes coding pro-inflammatory cytokines and chemokines. The MyD88 adaptor participates in the signaling cascade of TLRs with the exception of TLR3 that uses TIR-domain-containing adapter-inducing interferon-β (TRIF) (reviewed in [21]). Moreover, MyD88 is essential for signaling through the IL-1 and IL-18 receptors [22]. Other PRRs, such as nucleotide oligomerization domain (NOD)-like receptors (NLRs) [23] and inflammasomes [24][25] are also involved in T. cruzi detection.
Few studies have addressed the T. cruzi detection by cardiomyocytes, which is a structural cell population normally considered a passive target for parasite invasion. In vitro evidences suggest that cardiomyocytes detect T. cruzi parasites through TLRs. This was demonstrated using newborn mouse cardiomyocytes that express TLRs and produce cytokines and chemokines in response to the parasite [26][27][28]. However, these studies are not fully conclusive as fibroblast contamination may occur in the cardiomyocyte isolation process and no in vivo study has addressed this issue.
In the present study, we investigated whether MyD88 participates in T. cruzi detection by cardiomyocytes in vitro and in vivo. For the in vivo analysis, we developed the Mer + MyD88flox +/+ mice (C57BL/6 background), which are the second generation progeny of MerCre-Mer +/+ mice [29] and MyD88flox +/+ mice [30] that express two copies of the MyD88flox gene and at least one copy of the Mer gene. The MerCreMer +/+ mice carry gene sequences for both Cre recombinase and a modified estrogen receptor (hence tamoxifen sensitive) under the control of the heart myosin heavy chain specific promoter. The MyD88flox +/+ mice display the MyD88 gene flanked by loxP sites that can be recombined by Cre [31]. After tamoxifen treatment of Mer + MyD88flox +/+ mice, it would be expected that only cardiomyocytes could produce Cre and thus delete the MyD88 gene. Mice were infected with blood trypomastigotes of the Y strain of T. cruzi, which induce an acute myocarditis that is controlled in C57BL/6 mice. Our choice was based on the premise that sensing of the Y parasite by cardiomyocytes and resident macrophages initiates a protective immune response in the heart, which may be affected by elimination of MyD88 in cardiomyocytes. By using this Cre-loxP system we demonstrate that cardiomyocytes sense T. cruzi parasites in vivo through the MyD88 pathway, resulting in transcription of CCL5, IFNg and TNFa genes, with no effect on local parasite control.

Ethics statement
All procedures were carried out in accordance with the Brazilian guidelines on animal welfare of the Conselho Nacional de Controle e Experimentação Animal-CONCEA, and were authorized by the ethics committee at the Biomedical Sciences Institute of the University of São Paulo (ICB/USP), under Comissão de Ética no Uso de Animais (CEUA) protocol number 140, approved on 2/23/2015.

T. cruzi parasites
For in vitro experiments, we used T. cruzi parasites of the Y strain [33] or of the Sylvio X10/4 clone [34], obtained from the supernatants of infected LLC-MK2 cells. For in vivo experiments, mice were infected by i.p. route with 1000 or 500 Y-strain trypomastigotes obtained from infected A/J mice.

HL-1 cell cultures
The HL-1 murine cardiac muscle cell line [35] was maintained in vitro by serial culture passages in FCS-supplemented CLAYCOMB media (Sigma-Aldrich). HL-1 cells were infected with 1 x 10 5 tissue-culture trypomastigotes of the Y strain or the Sylvio X10/4 clone. After 24h in culture at 37˚C and 5% CO 2 atmosphere, cells were detached, washed and the pellets were processed for RNA and protein analysis.

RNA extraction, cDNA synthesis and amplification by real-time PCR (RT-PCR)
The RNA from HL-1 cells was extracted using the RNeasy Mini Kit (Qiagen, USA). To obtain RNA from the heart and spleen, mice were perfused with phosphate buffered saline (PBS) to eliminate blood contaminants in tissues. In some experiments, after perfusion, the atrium and ventricle of each mouse were grossly separated by a transversal cut. The RNA was extracted from the organs crushed in liquid nitrogen, using the RNeasy Mini Kit (Qiagen), following the manufacturer's protocols. One microgram of RNA was converted to cDNA with reverse transcriptase (RT) (Applied Biosystems, USA). The platinum SYBR Green system (Invitrogen Life Technologies, USA) was employed to quantify mRNA levels using 100 ng of cDNA and 4 pmol/μl of the specific primers for T. cruzi 18S rRNA gene, MyD88, iNOS, CCL5, CXCL7, CXCL10, IFNγ, TNFα, IL10 and TGFβ (S1 Table) and the ABI 7500 Fast RT-PCR system equipment and software (Applied Biosystems,). The relative amount of mRNA was calculated using the GAPDH gene as endogenous control (S1 Table).

PCR for mouse genotyping
Mouse genotyping was carried out by PCR using DNA extracted from three tail blood drops collected in FTA CARDS paper (GE Healthcare Life Sciences, USA). The FTA paper was cut in three small discs of approximately 2 mm diameter that were placed in 0.2 ml eppendorf tubes. The samples were washed twice with FTA purification reagent (GE Healthcare Life Sciences) and once with Tris EDTA buffer followed by ultrapure water, and then dried at the oven for 15 minutes [37][38]. For MerCreMer and MyD88flox transgene amplification, the primers indicated in the S2 Table were added to a Master Mix solution (Applied Biosystems). The amplification product was evaluated by agarose gel electrophoresis.

Western blot analysis
To extract proteins from mouse hearts and spleens, tissue samples were homogenized using a Polytron homogenizer (Kinematica, USA) in extraction buffer (99 mM KCl, 10 mM Hepes, 3 mM MgCl 2 , 5 mM EDTA, 1% glicerol, 1 mM DTT, 0,04% SDS, pH7.4) with a mixture of protease inhibitors (PMSF, aprotinin, leupeptin and pepstatin from Amresco, ELITech Group, Benelux, and sodium orthovanadate from Sigma-Aldrich). The protein concentration was determined by Bradford Method [39]. Forty micrograms of protein were resolved by electrophoresis on 5% stacking/12% polyacrylamide-SDS gels, and then transferred to a nitrocellulose membrane (Bio-Rad, USA). The membrane was stained with Ponceau's solution to evaluate the protein transfer efficiency. The membrane was incubated overnight at 4˚C with rabbit anti-MyD88 monoclonal antibody (Cell Signaling Technology, USA) and mouse anti-GAPDH monoclonal antibody (Santa Cruz Biotechnology, USA). After washing, the membrane was incubated for 1h at room temperature with horseradish peroxidase-conjugated (HRP) antirabbit or anti-mouse IgG secondary antibodies (Jackson Immuno Research, USA) followed by washing and detection with ECL Western blotting substrate (Thermo Fisher Scientific Inc., USA). The chemiluminescence was detected using the Alliance-6.7 Imaging System (Uvitec, Cambrige, UK). The protein bands were quantified by optical densitometry with the ImageJ software.

Histopathological analysis
Mice were sacrificed and perfused by injecting PBS into the right ventricle using a Pro TPM peristaltic pump (Watson-Marlow Inc., USA) with the inferior vena cava sectioned. Half of the heart was fixed with 10% formaldehyde and included in paraffin. Heart tissue sections (5 μm) were hematoxylin-eosin stained using standard procedures and examined by optical microscopy. For each mouse, 3-6 heart tissue sections (90 μm apart) were evaluated by assigning scores from 0 to 5, considering the intensity of leukocyte infiltration in the pericardium, endocardium and myocardium, both in the ventricle and in the atrium. The number of amastigote nests in the ventricle and atrium was also estimated. The final score for each item in a given mouse was calculated by the mean of the scores of each of the slides.

Indirect heart parasitism evaluation
The presence of live T. cruzi parasites in PBS-perfused mouse hearts was evaluated by cultivating quintuplicates of 0.1 or 0.01 mg tissue homogenates in LIT medium, where the relative parasite load of each heart was determined by the number of positive cultures. Heart parasitism was also quantified indirectly by PCR using specific primers for T. cruzi 18S rRNA (AF303659) [36] in relation to GAPDH gene expression.

Statistical analysis
Mann-Whitney test or one-way and two-way ANOVA were used for statistical analysis and comparison of the experimental groups using GraphPad Prism Software version 5.00 (Graph-Pad Software Inc, USA). The specific statistical analysis in each situation is described in corresponding figure legends. Differences between groups were considered significant when p<0.05.

HL-1 cardiomyocytes infected with T. cruzi parasites express MyD88 mRNA and protein
Cardiomyocytes isolated from newborn hearts have been shown to respond to T. cruzi by producing cytokines and other inflammatory mediators [26][27]. These results could indicate the involvement of MyD88 in the cardiomyocyte response to T. cruzi infection. To demonstrate the presence of MyD88 in a cell population consisting exclusively of cardiomyocytes, we evaluated HL-1 cells, an atrial cardiomyocyte tumor line derived from C57BL/6 mice [35]. HL-1 cells were infected with Sylvio X10/4 and Y strain trypomastigotes and, after 24h in culture, assayed for MyD88 mRNA transcription and protein expression. In these experiments, the heart extract of MyD88-deficient mice was used as a negative control. PCR and western blot results showed that both uninfected and T. cruzi-infected HL-1 cells display MyD88 transcripts ( Fig 1A) and protein ( Fig 1B), with no difference between pre-and post-infection levels. From these results, we confirmed the presence of MyD88 in T. cruzi-infected cardiomyocytes and, consequently, confidently approached the in vivo experiments.

Tamoxifen treatment induces MyD88 gene deletion in cardiomyocytes of double transgenic Mer + MyD88flox +/+ mice
To obtain a mouse model in which the MyD88 was absent exclusively in cardiomyocytes, we developed a tamoxifen-induced Cre-LoxP gene conditional deletion system by crossing Mer-CreMer +/+ mice with MyD88flox +/+ mice. The MerCreMer +/+ mice have Cre and ER (estrogen receptor) sequences under the control of the cardiac alpha myosin heavy chain promoter [29]. MyD88flox +/+ mice have LoxP sites flanking the MyD88 gene that are sensitive to deletion by Cre [30]. Our goal was to obtain a double transgenic F2 progeny, named Mer + MyD88flox +/+ , in which the MyD88 gene would be selectively eliminated from cardiomyocytes after tamoxifen treatment. We opted to delete the MyD88 gene in the adult, avoiding undesired secondary effects due to an eventual participation of this molecule in the development of the embryonic/ fetal heart. The genotype of MerCreMer +/+ and MyD88flox +/+ mouse F1 progeny is shown in the S1A Fig To evaluate the effects of tamoxifen treatment on the MyD88 gene and protein expression in the heart, Mer -/-MyD88flox +/-, MerCreMer +/+ and Mer + MyD88flox +/+ mice were treated i. p. with this drug for five consecutive days. After 25 and 55 days of treatment, MyD88 gene and protein expression was estimated in the heart and spleen by RT-PCR and western blot (Fig 2). For comparison, these parameters were also evaluated in untreated C57BL/6 and MyD88 -/mice. The tamoxifen-treated Mer + MyD88flox +/+ mice showed significantly lower MyD88 gene transcription in the heart, but not in the spleen (Fig 2A), compared to tamoxifen-treated Mer -/-MyD88flox +/mice. In fact, MyD88 gene transcription in the heart of tamoxifen-treated Mer + MyD88flox +/+ mice was at the levels of that in MyD88 -/mice, but lower than those in C57BL/6 mice and tamoxifen-treated MerCreMer +/+ mice (Fig 2B). Lower MyD88 gene transcription was also observed at the ventricle and atrium of tamoxifen-treated Mer + MyD88flox +/+ mice when compared to corresponding regions in tamoxifen-treated MerCreMer +/+ mice (S2 Fig). Interestingly, for both mouse strains, MyD88 gene transcription was higher in the atrium than in the ventricle.
The low MyD88 gene expression in the heart of tamoxifen-treated Mer + MyD88flox +/+ mice was confirmed at the protein level ( Fig 2C). Moreover, as expected, tamoxifen-treated Mer + MyD88flox +/+ and MerCreMer +/+ mice showed similar levels of MyD88 protein in the spleen (Fig 2D). Remnants of the MyD88 protein were observed in the heart of tamoxifentreated Mer + MyD88flox +/+ mice, probably not from cardiomyocyte origin, as reported by Yan Feng and coworkers at the heart of a Mer/MyD88fl mouse lineage with constitutive MyD88 gene deletion restricted to cardiomyocytes [40]. Together, our results show that the Mer + MyD88flox +/+ mouse model is suitable to evaluate the ability of cardiomyocytes to sense T. cruzi infection through MyD88-mediated signaling pathways.

MyD88 expression in cardiomyocytes is not crucial for acute parasitemia control
To investigate if tamoxifen treatment affects the parasitemia, C57BL/6 mice were inoculated with this drug, or with the sunflower oil used for its solubilization, for five days at the prescribed dose, and infected with 10 3 T. cruzi parasites of the Y strain a week later. An untreated infected group was also included as control. Parasitemias, evaluated from 5 to 32 days postinfection (p.i.), increased in tamoxifen-treated mice in relation to the untreated group (S3A Fig), while treatment with sunflower oil had no effect. Moreover, tamoxifen-treated mice died up to day 15 p.i., while untreated mice or those treated with sunflower oil were alive at day 32 p.i., when they were sacrificed. Next, we evaluated if this adverse tamoxifen effect was diminished by postponing the time of infection in relation to that of drug administration. For this purpose, C57BL/6 mice were inoculated with tamoxifen for five days and then infected with 10 3 T. cruzi parasites one, two or three weeks later. No differences were observed between the different tamoxifen-treatment protocols, except for second peak reduction in mice treated after three weeks of infection (S3B Fig). From these results, we concluded that tamoxifen treatment should be performed more than three weeks after T. cruzi infection to reduce its immunosuppressive effect. In addition, in the infection studies, tamoxifen-treated and infected Mer + MyD88flox +/+ mice must always be compared to tamoxifen-treated and infected MerCre-Mer +/+ or Mer -/-MyD88flox +/control groups.
To determine whether MyD88 signaling in cardiomyocytes contributes to systemic T. cruzi control, we compared the parasitemia curves and survival rates of Mer + MyD88flox +/+ and MerCreMer +/+ mice that were both treated with tamoxifen and infected with 5 x 10 2 T. cruzi parasites four weeks later. As shown in Fig 3A, the parasitemia curves of tamoxifen-treated Mer + MyD88flox +/+ and MerCreMer +/+ mice were similar up to day 21 of infection. Accordingly, these mouse groups displayed comparable survival rates (Fig 3B). Globally, we concluded that MyD88 gene deletion in cardiomyocytes does not change the amount of parasites in the blood nor the mouse survival.

MyD88 deficiency in cardiomyocytes does not affect heart pathology at the early and late acute T. cruzi infection
Next, we evaluated the impact of the selective MyD88 elimination in cardiomyocytes on the intensity of inflammatory leukocyte infiltration and parasite load at the heart of acutely T. cruzi-infected mice. For this purpose, tamoxifen-treated MerCreMer +/+ and Mer + MyD88flox +/+ mice were sacrificed from day 10 to 33 p.i., for histopathology and parasite load analyses of the heart tissue. The end-point of our evaluation was the moment of the infection when direct parasitemia became negative, no amastigote nests were observed by screening of heart histopathology slides, and T. cruzi RNA positivity at the heart dropped to very low levels. In Figs 4 and 5, we show two independent experiments, respectively evaluated at early and late acute phase, which are representative of the whole acute phase. At day 13 p.i., when the local parasite load and intensity of leukocyte infiltrates at the heart attained high levels, no differences in these parameters were noticed between tamoxifen-treated MerCreMer +/+ and Mer + MyD88flox +/+ mice (Fig 4A and 4B). This was patent both at the atrium and the ventricle, as well as in the pericardium, endocardium and myocardium. Considering that the examination of a few sections does not totally represent the whole heart, T. cruzi presence was also evaluated in the heart tissue of each mouse by analyzing in vitro parasite growth (Fig 4C). This approach corroborated the analysis of T. cruzi amastigote nests in histopathology studies, reinforcing the conclusion that, at day 13 p.i., the absence of MyD88 in cardiomyocytes does not largely impact on heart parasite load. At day 28 p.i., when parasites were rarely found in the heart, the intensity of cellular infiltration (myocarditis, pericarditis) and the number of nests in the ventricle and atrium were also not different in tamoxifen-treated, Mer + MyD88flox +/+ and MerCreMer +/+ mice (Fig 5A and 5B). In S4 Fig we show the histopathology pictures of the atrium and ventricle of tamoxifen-treated T. cruzi-infected Mer + MyD88flox +/+ and MerCre-Mer +/+ mice at day 27 p.i.. In these pictures, we observed no difference in the intensity of inflammatory infiltrates at the atrium and ventricle, or in the diameter of cardiac muscle fibers, as a consequence of the MyD88 elimination from cardiomyocytes.
Confirming the optical microscopy analysis of heart parasite nests, in Fig 6 we show the comparative RT-PCR results for T. cruzi RNA detection at days 13, 27 and 33 p.i. No difference was observed in heart parasite load between tamoxifen-treated Mer + MyD88flox +/+ and Mer-CreMer +/+ mice, both at the early and late acute infection.
Because of our failure to detect any effect on cardiac pathology as a result of the deletion of the MyD88 gene in cardiomyocytes, and considering that heart infection determines a strong infiltration of MyD88-expressing leukocytes, we compared MyD88 gene transcription in tamoxifen-treated Mer + MyD88flox +/+ and Mer -/-MyD88flox +/mice infected for 4 weeks, or left uninfected. MyD88 gene transcription at the heart was notably higher in infected mice (S5 Fig), an expected result that confirms the intense MyD88 expression of inflammatory cells. More important, the MyD88 gene transcription levels at the heart of tamoxifen-treated

MyD88 activation in cardiomyocytes induces gene transcription of the CCL5 chemokine and pro-inflammatory IFNγ and TNFα cytokines genes at the heart of T. cruzi-acutely infected mice
Because the results of this study did not reveal significant differences in the levels of heart pathology and parasite load between tamoxifen-treated, T. cruzi-infected, MerCreMer +/+ and Mer + MyD88flox +/+ mice, it was important to evaluate if MyD88 in cardiomyocytes is activated along the heart infection by T. cruzi. To indirectly approach this issue, we evaluated the transcription of various effector molecules that could be induced through the MyD88 adaptor after TLR activation. We first choose to measure the mRNA levels for iNOS, a gene that is transcribed in vivo by cardiomyocytes in response to LPS inoculation [40], as well as those for CCL5, CCL7 and CXCL10, three chemokines transcribed in newborn cardiomyocytes after in vitro infection with T. cruzi [27]. While no changes were observed in the mRNA levels of iNOS, CCL7 and CXCL10 in the hearts of infected mice, as a consequence of the MyD88  In vivo cardiomyocyte MyD88 signaling by Trypanosoma cruzi deletion in cardiomyocytes, a significant reduction in transcriptional levels of CCL5 was observed (Fig 7), both at early and late time points in the acute phase. We also evaluated the mRNA levels of pro-inflammatory (IFNγ and TNFα) and anti-inflammatory (IL10 and TGFβ) cytokines at the hearts of tamoxifen-treated MerCreMer +/+ and Mer + MyD88flox +/+ T. cruziinfected mice (Fig 8). Interestingly, the transcription of IFNγ in Mer + MyD88flox +/+ mice was reduced at 13 dpi, and that of TNFα both at 13 and 27 dpi, but not at day 33, when the parasite was almost eliminated from the cardiac tissue. Yet, transcription of anti-inflammatory cytokines IL10 and TGFβ was not different in tamoxifen-treated infected MerCreMer +/+ and Mer + MyD88flox +/+ mice along the acute infection period. In these experiments we can also observe that, similar to that found for parasite load and intensity of leukocyte infiltrates, the heart transcription levels of iNOS, chemokines and cytokines were considerably reduced at day 33 dpi, in relation to day 13.
Together, the chemokine and cytokine results indirectly demonstrate that cardiomyocytes are signaled to respond through MyD88 along the heart infection by T. cruzi parasites.

Discussion
This study was designed to assess whether cardiomyocytes are active participants in acute myocarditis that follows T. cruzi invasion of the heart tissue. In this scenario, we thought that cardiomyocytes could sense T. cruzi infection through MyD88-mediated molecular pathways, such as TLR signaling by parasite components and IL-1/IL-18 receptor engagement by cytokines produced locally by leukocytes or other cells. In response to these stimuli, cardiomyocytes could secrete chemokines, cytokines and other inflammatory mediators and contribute to leukocyte recruitment and local control of the parasite. To test this hypothesis, we investigated whether MyD88 gene deletion in cardiomyocytes affects heart inflammation and parasite burden during acute T. cruzi infection. Before addressing this issue, we confirmed that cardiomyocytes express MyD88. Although it has been shown that cardiomyocyte-enriched cells isolated from the heart of newborn mice express TLR molecules and produce proinflammatory cytokines in response to T. cruzi [26][27][28], the presence of a few contaminant fibroblasts was not fully discarded. Therefore, we analyzed MyD88 expression in HL-1 cells, a C57BL/6 cardiomyocyte cell line. Corroborating the results obtained with newborn cardiomyocytes, we found that HL-1 cells transcribe and express MyD88. Interestingly, as reported for neonatal heart cells [27], there was no difference in MyD88 gene and protein levels between uninfected and T. cruzi-infected HL-1 cells.
To evaluate the contribution of MyD88 expressed by cardiomyocytes to the heart pathology caused by T. cruzi infection, we developed a conditional knockout model in which the MyD88 gene can be selectively eliminated in these cells. The suitability of this murine model was demonstrated by data showing absence of MyD88 transcription and 60% reduction of MyD88 protein, in the heart, but not in the spleen, from tamoxifen-treated Mer + MyD88flox +/+ mice, a phenomenon that was not observed for tamoxifen-treated MerCreMer +/+ mice. The presence of MyD88 protein remnants in the heart of tamoxifen-treated Mer + MyD88flox +/+ mice is not surprising considering that the heart tissue contains fibroblasts, endothelial cells and resident In vivo cardiomyocyte MyD88 signaling by Trypanosoma cruzi phagocytes, in which the MyD88 protein should not be affected by tamoxifen treatment. A partial reduction in heart MyD88 gene expression was also reported in a constitutive Mer +-MyD88fl mouse model [40]. In addition, we observed greater amounts of MyD88 transcripts in the atrium than in the ventricle of Mer + MyD88flox +/+ and MerCreMer +/+ mouse hearts, and tamoxifen treatment of Mer + MyD88flox +/+ mice reduced MyD88 gene expression in both heart compartments. Although these data suggest that cardiomyocytes exhibit more MyD88 mRNA in the atrium than in the ventricle, other cell populations (endothelial cells, fibroblasts, and even resident macrophages) can also contribute to this difference.
The MyD88 deficiency in cardiomyocytes does not affect the systemic control of the parasite, as similar parasitemia curves were observed in tamoxifen-treated Mer + MyD88flox +/+ and MerCreMer +/+ mice. Of note, in relation to untreated mice, tamoxifen-treated Mer + MyD88flox +/+ and MerCreMer +/+ mice had a lower systemic T. cruzi control, as indicated by higher and extended parasitemias that culminated in some mouse death at the end of acute phase. These effects probably arise from an inhibitory effect of tamoxifen on endogenous estrogen receptor signaling, which is known to up-regulate the immune response [41]. Due to this effect, and to reduce the negative impact of tamoxifen on parasite control, mice were infected 4-5 weeks after drug administration, a delay that attenuates its immunosuppressive effects. A long interval also allows the complete recovery of functional cardiac abnormalities, which may occur following tamoxifen treatment in MerCreMer +/+ mice, but fully revert 9 days after drug discontinuation [32,42].
When the effects of MyD88 depletion in cardiomyocytes on T. cruzi-induced acute heart pathology were investigated from 10 to 33 days p.i., we found no differences between tamoxifen-treated Mer + MyD88flox +/+ and MerCreMer +/+ mice. That is, no change was observed in the intensity of heart cell infiltrates and number of parasite nests in the histopathological examination, as well as in heart parasite load estimated by PCR and by heart tissue culture in LIT medium. In addition, we did not observe large differences in the diameter of cardiac muscle fibers as a consequence of MyD88 elimination in cardiomyocytes, an unexpected result considering that hypertrophy of newborn cardiomyocytes is induced in vitro by T. cruzi signaling through TLR2 and IL-1R, two activators of the MyD88 pathway [43]. From these results, we can conclude that the MyD88 in cardiomyocytes does not largely contribute to local inflammation and parasite elimination during the acute T. cruzi infection by Y strain parasites, its role being obscured by the effector activity of resident and recruited leukocytes, as well as by redundancy of parasite sensors in cardiomyocytes and other cell populations. At this respect, it has been shown that purified trans-sialidase of T. cruzi activates newborn cardiomyocytes in vitro through a MyD88-independent pathway, resulting in production of the chemokines MCP-1 and Fractalkine [44].
In a septic shock model, Feng and coworkers observed that Mer + MyD88fl mice (a mouse lineage that exhibits constitutive deficiency of MyD88 in cardiomyocytes) display lower mortality after inoculation of LPS compared to wild-type mice, a result that supports the strong participation of cardiomyocytes in the heart response [40]. The discrepancy between these and our results could be due to the fact that in the sepsis model the complete cardiomyocyte population responds concomitantly to soluble LPS molecules, whereas in our model of T. cruzi infection, not only the number of responding cardiomyocytes should be very small, but cardiomyocyte infection is not synchronic.
Importantly, our results clearly demonstrate that cardiomyocytes sense T. cruzi infection though MyD88-mediated pathways. This was concluded from data showing significant reduction of CCL5, IFNg and TNFa gene transcription in the heart of tamoxifen-treated Mer + MyD88flox +/+ mice. To our knowledge this is the first evidence of the in vivo functionality of this molecular pathway in cardiomyocytes during T. cruzi infection. Interestingly, among more than 7500 genes investigated, the CCL5 gene has been described as the most intensively up-regulated in newborn cardiomyocytes after in vitro T. cruzi infection. [27]. Thus, the reduced transcription of the CCL5 gene in the heart of tamoxifen treated Mer + MyD88flox +/+ mice in which the MyD88 cardiomyocyte molecule was deleted is probably a direct effect on the chemokine-inducing capacity of these structural cells, although the possibility also exists that infiltrating leukocytes could also transcribe this gene in response to an unknown cardiomyocyte signal. Because CCL5 is a chemokine involved in T cell recruitment, we can extrapolate that its production by cardiomyocytes could be particularly important to attract tissue-infiltrating CD8 + T cells that express high levels of the CCL5 receptor CCR5 [45]. By detecting extracellular T. cruzi through TLR2 and TLR4 at the moment of the infection, or intracellular T. cruzi through TLR9 and TLR7, the cardiomyocytes should activate the MyD88 pathway and induce CCL5 that will focus the recruitment of cytotoxic CD8 + T cells towards the infected cell. For IFNγ, produced by CD4 + , CD8+ and natural killer (NK) lymphocytes, and for TNFα, produced by different myeloid and lymphoid cell populations, these are pro-inflammatory cytokines crucial not only for systemic control of T. cruzi but also for the local control of the parasite at the acutely-infected heart [46,47]. In this context, differently from CCL5, it is probable that the reduced transcription of IFNγ (and eventually of TNFα) in tamoxifen-treated Mer + MyD88flox +/+ infected mice could be an indirect effect, secondary to the elimination of the MyD88 adaptor in cardiomyocytes.
In relation to the anti-inflammatory cytokines IL10 and TGFβ, because we did not observe any significant change at the heart of tamoxifen-treated Mer + MyD88flox +/+ mice we can speculate their transcription is not induced after MyD88 signaling in cardiomyocytes.
In summary, our results indicate that cardiomyocytes respond to T. cruzi infection by molecular pathways involving the MyD88 adaptor, and this results in increased transcription of CCL5, IFNg and TNFa genes. Nevertheless, and presumably because of the limited number of cardiomyocytes that are directly or indirectly activated by T. cruzi in our infection model, the protective role of MyD88 in cardiomyocytes becomes a minor one, not adding much to the effects of resident phagocytes and recruited leukocytes for the resolution of the acute T. cruzi infection at the heart. Future work will need to evaluate the role of MyD88 in cardiomyocytes in other murine models of the acute infection where there is only a partial control of T. cruzi at the heart. Moreover, unraveling the role played by cardiomyocytes in chronic phase models of CCC is essential for our understanding of Chagas disease physiopathology.
The roles played by cardiomyocytes in T. cruzi infection need to be considered for future prospects of Chagas disease therapy. Among other factors, polymorphisms in genes involved in parasite sensing by cardiomyocytes may have an impact on the immune response of the heart and on the outcome of the disease. In addition, since these polymorphisms may be predictive of CCC development and evolution, they could be used for the selection of therapeutic protocols.
Supporting information S1