This study was carried out in strict accordance with the recommendations of the Guide for the Care and Use of Laboratory Animals of the Brazilian National Council of Animal Experimentation (http://www.sbcal.org.br/) and Federal Law 11.794 (October 8, 2008). The Institutional Committee for Animal Ethics of Fiocruz (CEUA-Fiocruz-L004/09; LW-10/14) approved all experimental procedures used in the present study. All presented data were obtained from two or three independent experiments.

C57BL/6 (H-2^d) and CCL3-deficient (B6.129P2-Ccl3^tm1Unc/J) mice were originally purchased from Jackson Laboratories (Sacramento, CA, USA) and matched and maintained in the animal facilities of the Oswaldo Cruz Foundation (CECAL/ICTB/Fiocruz, Rio de Janeiro, Brazil). All mice were maintained under specific pathogen-free conditions with drinking water and chow food ad libitum. The wild-type condition (C57BL/6) and CCL3-deficiency were confirmed by genetic characterization of tail samples and using the primers and protocols described at The Jackson Laboratories home page (https://www.jax.org/), thereafter the mice were referred as ccl3^+/+ and ccl3^−/−, respectively.

In all sets of experiments, three to five sex- and age-matched non-infected (NI) controls were analyzed in parallel with 5–10 infected mice according to the experimental protocol. Five- to seven-week-old female mice were infected with the Colombian T. cruzi DTU I strain (22) by intraperitoneal injection of 100 blood trypomastigotes, obtained from passage mice to mice every 35 days post-infection (dpi). Parasitemia was estimated in 5 μL of tail vein blood. After the peak of parasitemia, detection of rare circulating trypomastigotes marked the onset of the chronic phase of infection, as previously described (6). Mortality was weekly registered.

Groups of three to five mice were subcutaneously inoculated daily with 0.1 mL of in vivo injection-grade saline (BioManguinhos, Rio de Janeiro, RJ, Brazil) or saline containing 10 μg of Met-RANTES, a CCR1/CCR5 partial antagonist (23), kindly provided by Dr. Amanda Proudfoot (Serono Pharmaceuticals, Geneva, Switzerland), for 30 consecutive days (from 120 to 150 dpi) and analyzed at 150 dpi.

For immunohistochemistry (IHC), the polyclonal antibody recognizing T. cruzi antigens was produced in our laboratory (LBI/IOC-Fiocruz, Brazil). Purified anti-F4/80 antigen (clone F4/80) antibody was purchased from CALTAG Laboratories (Burlingame, CA). Supernatants were home made with anti-mouse CD8a (53-6.7) and anti-mouse CD4 (GK1.5) hybridomas. Anti-CCL3 polyclonal antibody produced in rabbit was a kind gift of Dr Mauro Teixeira (Universidade Federal de Minas Gerais, Brazil). In our IHC studies, we also used the monoclonal antibodies anti-IFNγ (R4-6A2, BD PharMingen, USA) and anti-perforin (CB5.4, Alexis Biochemicals, San Diego, CA, USA), a polyclonal anti- inducible nitric oxide synthase (iNOS/NOS2) antibody (Cayman Chemical, USA), anti-rat immunoglobulin from DAKO (Glostrup, Denmark), and the biotinylated anti-rabbit immunoglobulin and peroxidase-streptavidin complex from Amersham (England). Appropriate controls were prepared by replacing the primary antibodies with purified rat immunoglobulin or rabbit normal serum (Sigma, USA). For in vivo cytotoxicity assays we used a cell trace ^TMCFSE cell proliferation kit (C34554) for flow cytometry (Invitrogen, Carlsbad, CA, USA). For flow cytometry studies using mouse cells, FITC- or PECy7-conjugated anti-TCR (clone H57.597) were purchased from Southern Biotech (Birmingham, AL, USA). PE-conjugated anti-TCR (clone H57.597), FITC- and APC-conjugated anti-mouse CD8a (53-6.7), PE-cy7-conjugated anti-LFA-1 (CD11a/CD18b, clone 2D7), FITC-conjugated anti-LFA-1 (CD11a/CD18b, clone M17/4), PECy-7-conjugated anti-TNF (MP6-XT22), FITC-conjugated anti-Pfn (clone 11B11), PECy-7-conjugated anti-IFNγ (XMG1.2) and PE-conjugated anti-CCR5 (clone C34-3448) were purchased from BD PharMingen (San Diego, CA, USA). Anti-CCR1-PerCp (clone sc-6125) was obtained from Santa Cruz Biotechnology (Dallas, TX, USA). APC-conjugated anti-IL-10 (clone LRM9104) was obtained from CALTAG (Burlingame, CA, USA). Appropriate controls were prepared by replacing the primary antibodies with their respective isotypes, which were also obtained from BD PharMingen (San Diego, CA, USA) or from Southern Biotech (Birmingham, AL, USA). All the antibodies (final concentrations: 1.5–20 μg/mL) and reagents were used according to the manufacturers' instructions.

RNA was isolated from heart of mice by acid guanidinium thiocyanate-phenol-chloroform extraction: RNA STAT-60^TM. Reverse transcriptase-polymerase chain reaction conditions, primer sequences used for detection of CCL3/MIP-1α, and PCR product size have been published elsewhere (24). The PCR product and molecular weight marker were electrophoresed in 6% polyacrylamide gel and stained with silver nitrate. Densitometry of gels was carried out on a Densitometer CS-9301PC (Shimadzu, Japan). The PCRs were standardized using hypoxanthine-guanine phosphoribosyl transferase (HPRT). Data are shown as relative CCL3 mRNA expression.

Leukocyte migration was measured as previously described (25), using a 24-well chemotaxis chamber (6.5 mm of diameter and 3-μm of pore-Falcon-BD, USA). Lower wells were loaded with 0.6 mL (1000 ng/mL) of the recombinant mouse CC-chemokines (CCL3 #450MA, CCL4/MIP-1β #451MB and CCL5/RANTES #478MR, R&D Systems, USA) prepared in RPMI 0.5% BSA. The polyethylene terephthalate membrane filter (#3492, B&D System, USA) pre-coated with 50 μL of purified fibronectin (10 μg/mL, #12173–019, GibcoBRL, USA) was placed between the upper and lower wells. Suspensions of peripheral blood mononuclear cells were prepared by pooling 1.5 mL heparinized individual samples (three animals per group) and performing Ficoll–Hypaque (d = 1.077 g/mL) separation, as previously described (24). A suspension containing 10⁶ cells/100 μL was added to the upper wells and the chemotaxis chamber was incubated at 37°C in humidified air with 5% CO₂ for 3 h. After the incubation period, the lower chamber migrating cells were collected, centrifuged, resuspended in 50μL and counted. The results are expressed as percentage of migrating cells compared with migrating cells in the absence of chemokines as controls.

Groups of five infected and three age-matched control mice were sacrificed under anesthesia at 28 and 120 dpi. The hearts and spleens of the mice were removed, embedded in tissue-freezing medium (Tissue-Tek, Miles Laboratories, Elkhart, IN, USA), and stored in liquid nitrogen for analysis by immunohistochemistry. Serial 3-μm thick cryostat sections were fixed in cold acetone and subjected to indirect immunoperoxidase staining, as previously described (6, 26). For visualization of the reaction we used the Dako liquid DAB plus substrate chromogen System - PT (K346811-2, Dako, Carpinteria, CA, USA). The positively stained areas for T. cruzi antigens⁺ areas and the numbers of positive CD4⁺, CD8⁺, F4/80⁺, IFNγ⁺, and perforin (Pfn)⁺ cells in 100 microscopic fields (50 mm²) were counted in heart sections. Tissue sections were stained and numbers of iNOS/NOS2⁺ cells in 100 microscopic fields (50 mm²) were counted in heart sections, as previously shown (26). The stained area was analyzed for CCL3 expression and cell types (CD4⁺, CD8⁺, F4/80⁺) in 25 microscopic fields (12.5 mm²) in two sections per heart tissue and spleens were evaluated with a digital morphometric apparatus. The images were digitized using a Sight DS-U3 color-view digital camera adapted to an Eclipse Ci-S microscope and analyzed with NIS Elements BR version 4.3 software (Nikon Co., Tokyo, Japan). In a set of experiments, the free software ImageJ (NIH, USA) was used to analyze the percentage of stained area for CCL3 and cell types, as well as the percentage of overlay for IHC-stained areas. According to the analyzed parameters, the data are presented as numbers of positive areas (parasite nests), cells per 100 microscopic fields or percentage of stained area.

Heart tissue was immersed in PBS solution (50 mg/0.5mL) containing complete protease inhibitor cocktail and Nonidet-P40 1% (Sigma, St. Louis, MO, USA). Tissue extract was obtained after homogenization using tissue grinder (T10BS1-5G, Ultra Turrax IKA, Germany) speed 8,000 rpm for 2 min, on ice. The homogenate was centrifuged at 14,000 × g 4°C for 10 min and the supernatant was aliquoted and frozen until use for quantification of cytokines.

The concentrations of cytokines in the cardiac tissue were evaluated by enzyme-linked immunosorbent assay (ELISA) DuoSet kits (R&D Systems, Minneapolis, MN, USA) for mouse IFNγ, TNF, IL-10, and CCL3 were used according to manufacturer's instructions. Diluted (1:2 and 1:10 in PBS) tissue extracts were analyzed in duplicates. Standards were 1/2 log dilutions of the recombinant cytokines from 1 pg/mL to 100 ng/mL.

The mice were sacrificed by blood draining under anesthesia, the harvested spleens were minced, and red blood cells were removed using lysis buffer (Sigma-Aldrich, St. Louis, MO, USA). For ex vivo analysis, splenocytes were incubated with 5 mg/mL brefeldin A (Sigma, St. Louis, MO, USA) for 4 h at 37°C. The cells were collected, washed, resuspended in PBS containing 2% fetal calf serum and labeled as previously described (6). In all experiments, one color labeled samples were prepared for establishment of the compensation values. The controls for specific labeling were prepared using isotype-matched antibodies. For analysis, 300,000 events per sample were acquired using the Beckman Coulter CyAn 7-Color flow cytometer (Fullerton, CA, USA) or 13-Color CytoFLEX-S (Beckman-Coulter, Houston, TX, USA). After gating in singlets and exclusion of dead cells, the mononuclear cell populations were analyzed using Summit v.4.3 Build 2445 software (Dako, Carpinteria, CA, USA) or CytExpert 2.3 software for Windows (Beckman-Coulter, Indianapolis, IN, USA), as described elsewhere (6). The fluorescence gates were cut in accordance with the labeling controls, respecting the curve inflecions. To identify CD8⁺ T cells bearing CCR5 and LFA-1, the gating strategy was as follows: singlets (R1), dead-cell exclusion (FSC-A × SSC-Lin, R2), TCR × CD8 dot plot (gating on TCR⁺ CD8⁺ cells, R3), LFA1 × CCR5 dot plot were analyzed. To characterize the CD8⁺ T cells expressing IFNγ and Pfn, the gating strategy was: singlets (R1), dead-cell exclusion (FSC-A × SSC-Lin, R2), TCR × CD8 dot plot (gating on TCR⁺ CD8⁺ cells), IFNγ × Pfn dot plot (% in quadrants of single-positive or double-positive cells), as previously shown (6). To identify CD8⁺ T cells expressing CC-chemokine receptors and cytokines, the gating strategy was as follows: singlets (R1), dead-cell exclusion (FSC-A × SSC-Lin, R2), TCR × CD8 dot plot (gating on TCR⁺ CD8⁺ cells), CCR1 × CCR5 (gating on single- and double positive cells), and TNF × IL-10 dot plot (% in quadrants of single- or double-positive cells), as described (16).

The ELISpot assay for the enumeration of IFNγ-producing cells was performed as previously described (27). The assays were performed in triplicate. The plates were coated (50 μL/mL) with anti-mouse IFNγ (clone R4-6A2; BD PharMingen, San Diego, CA, USA) antibody diluted in PBS (5 μg/mL). Further, total splenocytes used as antigen presenting cells (APC) were primed with H-2K^b-restricted VNHRFTLV peptide from ASP2 (10 μg/mL) for 30 min at 37°C. After three washings, 3 × 10⁵ primed APC in 100 μL were seed per well and incubated with freshly isolated splenocytes seeded at a suspension of 5 × 10⁵ cells/100 μL per well and were incubated for 20 h at 37°C and 5% CO₂. Concanavalin A (ConA, 5 μg/mL) was used as a mitogenic stimulant. After three-time washings with warmed PBS and three-times with 0.1% Tween-20 PBS. A biotin-conjugated anti-mouse IFNγ antibody (clone XMG1.2; BD PharMingen, San Diego, CA, USA) was used to detect the captured cytokines. The spots were revealed by respective incubation of the samples with a solution of alkaline phosphatase labeled streptavidin (BD PharMingen, San Diego, CA, USA) and a solution of nitro blue tetrazolium (NBT; Sigma, St. Louis, MO, USA) and 5-Bromo-4-chloro-3-indolyl phosphate (BCIP; Sigma, St. Louis, MO, USA) in Tris buffer (0.9% NaCl, 1% MgCl2, 1.2% Tris in H₂O). The mean number of spots in triplicate wells was determined for each experimental condition using a CTL OHImmunoSpot A3 Analyzer (Cleveland, OH, USA), and the number of specific IFNγ-secreting T-cells was calculated by estimating the stimulated spot count/10⁶ cells. Representative pictures of the assay have previously been published (6).

For the in vivo cytotoxicity assays, splenocytes collected from naive C57BL/6 mice were treated with red blood cells lysis buffer (Sigma, St. Louis, MO, USA). The cells were washed and divided into two populations that were labeled with the fluorogenic dye carboxyfluorescein diacetate succinimidyl diester (CFSE; Molecular Probes, Eugene, OR, USA) at a final concentration of 5 μM (CFSE^high) or 0.5 μM (CFSE^low). CFSE^high cells were coated with 2.5 μM of the VNHRFTLV ASP2 peptide for 40 min at 37°C. CFSE^low cells remained uncoated. Subsequently, CFSE^high cells were washed and mixed with equal numbers of CFSE^low cells before intravenous injection (1–2 × 10⁷ cells per mouse) into T. cruzi-infected C57BL/6 recipients that were sedated with diazepam (20 mg/Kg). Spleen cells from the recipient mice were collected at 20 h after adoptive cell transfer, red blood cells lysed, cells were washed in PBS and fixed with 1.0% paraformaldehyde. For analysis, 30,000 events were acquired with CyAn-ADP flow cytometer (Beckman-Coulter, Houston, TX, USA) and analyzed using the Summit v.4.3 Build 2445 program (Dako, Glostrup, Denmark). Representative pictures of the assay are shown, and the percentage of specific lysis was determined using the following formula:

Peritoneal cells were harvested in serum-free cold DMEM (LCG, Rio de Janeiro, Brazil) from non-infected ccl3^+/+ and ccl3^−/− mice and washed in serum-free medium at 4°C. Macrophages were resuspended in supplemented DMEM supplemented with fungizone (1 μg/mL), penicillin (1 μg/mL), streptomycin (1 μg/mL) and 10% heat-inactivated fetal calf serum (FBS; Gibco, USA) at 1 × 10⁶/mL, and 0.2 mL were dispensed onto 13 mm glass coverslips placed in a well of a 24-well plate (Falcon, EUA). Cells were allowed to adhere for 2 h at 37°C in the presence of 95% humidity and 5% CO₂, and then washed three-times with serum-free warmed DMEM, and 1.0 mL of supplemented DMEM was added to each well. The plates were incubated 2 h at 37°C in the presence of 95% humidity and 5% CO₂ and in the absence or presence of 10 ng/mL of murine recombinant IFNγ (eBioscience, USA). The macrophages were washed and then cultured in the absence or presence of live trypomastigotes obtained from Vero cell line as previously described (28), using a 10:1 parasite:macrophage ratio in a 1.0 mL of supplemented DMEM per well. After 4 h, the supernatants were collected and stored for cytokines and NO_x evaluation, the coverslips were gently washed in warmed PBS, fixed in methanol and stained using Giemsa (Merck, Germany), dehydrated in acetone/xylol, mounted in Entellan (Merck, Germany) and assessed using a light microscopy (Nikon, USA). The percentage of cells containing intracellular forms of T. cruzi were determined using a light microscope evaluating 200 cells per coverslips. In a set of experiments, peritoneal cells were harvested in serum-free cold DMEM from non-infected and chronically (120 dpi) infected ccl3^+/+ and ccl3^−/− mice and washed in serum-free medium at 4°C. Cells were allowed to adhere (2 × 10⁵/well) in a 24-well plate for 2 h at 37°C in the presence of 95% humidity and 5% CO₂, and then washed three-times with serum-free warmed DMEM. After addition of 1.0 mL of supplemented DMEM, the cells were cultivated for 24 h. The supernatants were collected and stored for NO_x evaluation.

Nitric oxide (NO_x) concentration was assessed by detection of nitrite/nitrate in macrophage culture supernatants using Griess reaction. Nitrate and NO_x levels in serum samples were determined by using the Griess reagent and vanadium chloride III (29). The sample concentrations were estimated based on standard curves from 0.8–200 μM of NaNO₂ and NaNO₃.

All mice were intraperitoneally tranquilized with diazepam (20 mg/Kg) and the transducers were carefully placed subcutaneously according to chosen preferential derivation (DII). The traces were recorded for 2 min using a digital system Power Lab 2/20 that was connected to a bio-amplifier at 2 mV for 1 s (PanLab Instruments, Barcelona, Spain). Filters were standardized to between 0.1 and 100 Hz and traces were analyzed using the Scope software for Windows V3.6.10 (PanLab Instruments, Barcelona, Spain). We measured the heart rate (beats per minute, bpm), the duration of the P wave and QRS, PR and QT intervals in milliseconds (ms). The relationship between the QT interval and the RR interval in the mouse was assessed in all animals. To obtain physiologically relevant values for the heart rate-corrected QT interval (QTc) in units of time (rather than time to a power that is not equal to 1), the observed RR interval (RR0) was first expressed as a unitless multiple of 100 ms, yielding a normalized RR interval, RR100 = RR0/ 100 ms. Next, the value of the exponent (y) in the relationship QT0 = QTc × RRy100 was assessed, with QT0 indicating the observed QT and the unit for QTc being ms. The natural logarithm was computed for each side of this relationship [(QT0) = In (QTc) + yln (RR100)]. Thus, the slope of the linear relationship between the log-transformed QT and RR100 defined the exponent to which the RR interval ratio should be raised to correct QT for heart rate (6).

For analysis of cardiac function through echocardiography mice were trichotomized in precordial region, sedated with 1.5% isoflurane gas in oxygen with flow 1 L/min, and examined with a Vevo 770 (Visual Sonics, Canada) coupled to a 30 MHz transducer. Cardiac geometry was made using two-dimensional mode images acquired for measurement of internal area of heart cavities (right and left ventricles). Long axis in systole and diastole and left ventricular ejection fraction (LVEF) were determined using Simpson's method.

The activity of the creatine kinase myocardial band (CK-MB) is a marker of myocardial injury associated with CCC severity in experimental models (29–32). CK-MB activity was measured with commercial kits (cat. 2010075K, Kovalent do Brasil Ltda, São Gonçalo, RJ, Brazil), as previously described (6). The assay was adapted for reading in a microplate spectrophotometer (ASYS Hitech GmbH, Eugendorf, Austria) to allow the study of small quantities of mouse serum according to the manufacturer's recommendations. The optical density at 340 nm was recorded every 2 min for 15 min.

For statistical analyses, we used the Student's t-test to compare two groups. Comparisons between groups were carried out by analysis of variance (ANOVA) followed by the Bonferroni post hoc test. The Kaplan-Meier method was employed to compare survival rates of the studied groups. All statistical tests were performed with GraphPad Prism 8.0.1 (La Jolla, CA, USA). Data are expressed as the arithmetic mean ± SE. Differences were considered statistically significant when p < 0.05.

In comparison with non-infected (NI) mice, increased expression of CCL3 mRNA transcripts was detected in the heart tissue of acutely infected C57BL/6 ccl3^+/+ mice at 28 dpi and persisted during the chronic phase of infection (120 dpi). Similarly, CCL3 mRNA expression was upregulated in the spleen of acute and chronically T. cruzi-infected mice (Figure 1A). Further, in the acute phase of infection CCL3 protein levels were increased in the cardiac tissue and, although reduced, CCL3 expression persisted elevated in the chronic infection (Figure 1B). The IHC study in serial sections of spleen tissues showed that in NI controls and chronically (120 dpi) C57BL/6 infected mice the expression of CCL3 protein preferential overlay with CD8⁺ and, mainly, F4/80⁺ cells (Figure S1). In addition, IHC analyses of cardiac tissues of NI controls revealed low expression of CCL3, which is upregulated in the tissues of chronically T. cruzi-infected mice, mainly associated with endothelial and cardiac cells (Figure S2A). Further, CD8⁺ and F4/80⁺ (macrophage marker) cells are mainly located in CCL3⁺ areas of the heart tissue (Figures S2A–C). Importantly, the intensity of heart tissue inflammation paralleled the CCL3 levels in the cardiac tissue in acute and chronically T. cruzi-infected mice (Figure 1C).

The migration potential of circulating cells of acutely (28 dpi) and chronically (120 dpi) T. cruzi-infected ccl3^+/+ mice was assessed by ex vivo migration assay using a modified Boyden chamber with a filter coated with fibronectin, allowing cell adhesion and transmigration. The peripheral blood mononuclear cells obtained from NI mice barely migrated toward CCL3, CCL4 and CCL5 (Figure 1D). However, circulating mononuclear cells of acutely infected mice migrated in the presence of CCL4 and CCL5 but did not migrate toward CCL3, which retained the cells on the upper chamber (Figure 1D). Conversely, peripheral blood mononuclear cells from chronically infected mice respond to the chemotactic property of CCL3, albeit did not respond to CCL4 and CCL5 (Figure 1D), supporting a role for CCL3 cell retention in a tissue in the acute phase of and in cell migration in the chronic T. cruzi infection.

To explore the role of CCL3 in T. cruzi parasitemia, survival and control of tissue infection, ccl3^+/+ and ccl3^−/− mice were infected with 10² parasites of the Colombian T. cruzi strain. Although ccl3^−/− mice showed higher number of circulating parasites in the early acute infection (28 dpi), independently of the CCL3 status all infected mice controlled the parasitemia and achieved the chronic phase of infection (Figure 2A). At 120 dpi, 100% of the ccl3^−/− and 75–80% of the ccl3^+/+ T. cruzi-infected mice survived (Figure 2A). In the acute phase of infection (28 dpi), the T. cruzi-infected ccl3^−/− mice showed an increase in the number of parasite⁺ areas in the cardiac tissue in comparison with infected ccl3^+/+ mice (Figure S3), supporting a role for CCL3 in resistance to acute T. cruzi infection. At 120 dpi, when parasite nests are rarely detected in the cardiac tissue of ccl3^+/+ mice, corroborating previous data (6), heart parasitism was lower in ccl3^−/− infected mice (Figure 2B).

To explore the participation of CCL3 in CCC, we assessed the intensity and cell composition of the T. cruzi-elicited myocarditis. At 120 dpi, the T. cruzi-infected ccl3^−/− mice showed reduced heart inflammation composed of CD8⁺ cells compared with infected ccl3^+/+ mice (Figure 3A). Indeed, infected ccl3^−/− mice presented reduced numbers of cells infiltrating the heart tissue. Although the numbers of CD4⁺ cells infiltrating the heart tissue remained similar, lower numbers of F4/80⁺ and, mainly, CD8⁺ cells were found in the heart tissue of T. cruzi-infected ccl3^−/− mice, in comparison with infected ccl3^+/+ mice (Figure 3B). In comparison with chronically T. cruzi-infected ccl3^+/+ mice, ccl3^−/− mice showed reduced numbers of Pfn⁺ and IFNγ⁺ cells infiltrating the heart tissue (Figure 3C). Further, compared with heart extracts of NI controls, IFNγ concentrations were increased in tissues of infected ccl3^+/+ and ccl3^−/− mice (Figure 3D). Nevertheless, IFNγ levels were lower in heart tissues of infected ccl3^−/− mice compared with ccl3^+/+ mice (Figure 3D).

At 120 dpi, compared with NI control mice T. cruzi-infected ccl3^+/+ mice showed an augment in the number of circulating leukocytes. This leukocytosis was not detected in T. cruzi-infected ccl3^−/− mice (Figure 3E), raising the idea that these mice may present impairment in cell activation/migration process. Splenomegaly, characterized by increase in the relative spleen weight and cellularity, is detected in chronically T. cruzi-infected mice (6). Here, splenomegaly was similarly detected in ccl3^+/+ (3.29 ± 0.31 mg/g in NI vs. 9.68 ± 0.58 mg/g in infected ccl3^+/+ mice; p < 0.001) and ccl3^−/− (3.78 ± 0.17 mg/g in NI vs. 8.93 ± 0.49 mg/g in infected ccl3^−/− mice; p < 0.001) T. cruzi-infected mice. The frequency of CD8⁺ T-cells was alike in NI and chronically T. cruzi-infected ccl3^+/+ mice (Figure 3F). However, in comparison with NI (ccl3^+/+ and ccl3^−/−) and infected ccl3^+/+ mice, the frequency of CD8⁺ T-cells was increased in infected ccl3^−/− mice (Figure 3F), which may indicate retention of CD8⁺ T-cells in the spleen of chronically infected CCL3-deficient mice. In contrast with the high frequency of CCR5⁺ LFA-1⁺ among the splenic CD8⁺ T-cells (R1/R2/R3 gated) detected in T. cruzi-infected ccl3^+/+ mice, a remarkable decrease in the frequency of the CCR5⁺ LFA-1⁺ CD8⁺ T-cell population was detected in infected ccl3^−/− mice (Figure 3F). These data indicate that the reduced numbers of CD8⁺ cells in the heart tissue and circulating leukocytes may reflect the retention of CD8⁺ T-cells in the spleen associated with reduction in the frequency of the potentially able to migrate CCR5⁺ LFA-1⁺ CD8⁺ T-cells in T. cruzi-infected ccl3^−/− mice.

At 120 dpi, there was an increase in the frequency of TCR⁺ cells in the spleen of T. cruzi-infected ccl3^−/− mice compared with NI counterpart controls, what is not detected in ccl3^+/+ infected mice (Figure 4A). However, the intensity of TCR expression was reduced on cells of infected ccl3^+/+ and, mainly, ccl3^−/− mice (Figure 4A). The frequencies of single-positive Pfn⁺ and double-positive IFNγ⁺ Pfn⁺ were not modified after infection of ccl3^+/+ and, mainly, ccl3^−/− mice. Although the increase in the frequency of splenic IFNγ⁺ CD8⁺ T-cells has been shown in infected ccl3^+/+ and ccl3^−/− mice, compared with respective NI controls, this was lower in infected ccl3^−/− mice (Figure 4B). The recognition of the H-2K^b-restricted ASP2 peptide VNHRFTLV by specific CD8⁺ T-cells has been linked to protective immune response (6, 32). Thus, we evaluated the potential effector function studying IFNγ-producing using ex vivo ELISpot assay and the CD8⁺ T-cell cytotoxic activity in vivo. At 120 dpi, reduced numbers of VNHRFTLV-specific IFNγ-secreting CD8⁺ T-cells was detected in the spleen of T. cruzi-infected ccl3^−/− mice, compared with infected ccl3^+/+ mice (Figure 4C). This was a condition associated with the absence of CCL3, as ELISpot response to ConA stimulation of splenic cells also led to diminished numbers of IFNγ-secreting cells detected in both NI and T. cruzi-infected ccl3^−/− mice (Figure S4). Conversely, there was an increase in the specific lysis of target cells by the VNHRFTLV-specific cytotoxic CD8⁺ T-cells in T. cruzi-infected ccl3^−/− mice, compared with the cytotoxic activity in infected ccl3^+/+ mice (Figure 4C). Overall, the lack of CCL3 was associated with reduced IFNγ-secreting and enhanced cytotoxic effector activities by splenic CD8⁺ T-cells in chronic T. cruzi infection.

In models of acute infection and in in vitro experiments, CCL3 is crucial for control of T. cruzi infection, in a NO-dependent manner (11, 12). Here, at 28 dpi there is a similar increase in the numbers of iNOS/NOS2⁺ cells in the heart tissue of T. cruzi-infected ccl3^+/+ and ccl3^−/− mice, compared with NI counterpart controls. At 120 dpi, a similar decay in the numbers of iNOS/NOS2⁺ cells were seen (Figure S5A). Comparable high levels of NO_x were detected in the serum of infected ccl3^+/+ and ccl3^−/− mice, at 120 dpi (Figure S5B). However, peritoneal macrophages obtained from chronically infected ccl3^−/− mice produced higher levels of NO_x, compared with macrophages of infected ccl3^+/+ mice (Figure S5C). Therefore, we checked the role of macrophages obtained from CCL3-deficient mice to control T. cruzi and to produce the inflammatory mediator NO_x and the cytokines TNF and IL-10 (Figure 5A). Peritoneal macrophages obtained of NI ccl3^+/+ mice were susceptible to in vitro infection by trypomastigote forms of the Colombian strain (Figure 5B), which triggered the production of NO_x, TNF and IL-10 (Figures 5C–E). In the presence of IFNγ these macrophages showed reduced frequency of infected macrophages in a milieu with high concentrations of NO_x, TNF and IL-10 in the supernatants of cultures (Figures 5B–E). Conversely, macrophages isolated from ccl3^−/− mice were more susceptible to infection by the Colombian strain trypomastigotes (Figure 5B). These ccl3^−/− macrophages produced similar amounts of NO_x and TNF, but lower concentrations of the regulatory cytokine IL-10, compared with macrophages isolated from ccl3^+/+ mice (Figures 5B–E). Notably, when macrophages obtained from ccl3^−/− mice were stimulated with IFNγ a significant reduction in the frequency of parasitized macrophages (Figure 5B) paralleled the reduction in the concentrations of NO_x and TNF (Figures 5C,D). However, the levels of IL-10 were not modified (Figure 5E). Together, these data support that CCL3 is important for macrophage control of parasite infection and formation of an IFNγ-triggered TNF- and NO-enriched inflammatory milieu.

Next, we analyzed the consequences of CCL3 deficiency for CCC. At 120 dpi, similar body weight was detected in all analyzed mice groups (ccl3^+/+: 21.9 ± 1.3 g in NI vs. 20.6 ± 1.8 g in infected mice, p > 0.05; ccl3^−/−: 24.5 ± 3.3 g in NI vs. 21.4 ± 1.6 g in infected ccl3^−/− mice, p > 0.05). Increased relative heart weight was detected in ccl3^+/+ mice compared with NI controls (Figures 6A,B), corroborating previous data (6). However, this heart alteration was not found in T. cruzi-infected ccl3^−/− mice (Figures 6A,B). Further, compared with counterpart NI controls, infected ccl3^+/+ mice showed alterations in heart geometry, revealed as augment in the heart longitudinal axis during systole (Figure 6C). Importantly, this abnormality was not detected in ccl3^−/− infected mice (Figure 6C). When compared with sex- and age-matched NI counterparts, ccl3^+/+mice chronically infected by the Colombian strain showed decrease in LVEF (Figure 6D). Notably, LVEF values of T. cruzi-infected ccl3^−/− mice resembled their counterpart NI controls (Figure 6D). Altogether, our data support that infected ccl3^−/− mice were protected of the abnormalities of these geometrical and functional abnormalities detected in infected ccl3^+/+ mice.

Inspection of CK-MB isoenzyme activity serum, a marker of cardiomyocyte lesion and cardiomyopathy severity in experimental CCC (30, 32), revealed that CK-MB levels were increased in infected ccl3^+/+ mice compared with NI mice. Importantly, T. cruzi-infected ccl3^−/− mice had lower CK-MB activity levels in serum as compared to infected ccl3^+/+ infected mice (Figure 6E). To explore the participation of CCL3 in T. cruzi-elicited cardiac electrical conduction, ECG analyses were carried out in chronically T. cruzi-infected ccl3^+/+ and ccl3^−/− mice. At 120 dpi, bradycardia and increased PR interval were similarly detected in T. cruzi-infected ccl3^+/+ and ccl3^−/− mice, compared with respective NI control groups. In comparison with NI matched controls, the QRS interval remained unaltered in both ccl3^+/+ and ccl3^−/− mice during the chronic T. cruzi infection (Figure S6). Corroborating previous data (16), prolongment of the QTc interval was demonstrated in chronically T. cruzi-infected ccl3^+/+ mice. However, the QTc interval was not augmented in T. cruzi-infected ccl3^−/− mice, in comparison with their age- and sex-matched NI controls. Moreover, CCL3 deficiency hampered the augment of the QTc interval in infected ccl3^−/− mice, compared with infected ccl3^+/+ mice (Figure 6F).

Next, we set out an experiment to analyze the status of the pro-inflammatory TNF and the regulatory IL-10 in heart tissue extracts of T. cruzi-infected ccl3^+/+ and ccl3^−/− mice. In comparison with ccl3^+/+ mice, chronically infected ccl3^−/− mice showed reduced concentrations of TNF, but similar concentrations of IL-10 in extracts of heart tissue (Figure 6G).

Here, we showed that the intensity of the CD8-enriched myocarditis and electrical abnormalities were related to the concentrations of TNF in the cardiac tissue. Thus, we analyzed the cytokine profiles, focused on TNF and IL-10, in CD8⁺ T-cells and, particularly, in CD8⁺ T-cells bearing the CCL3 receptors CCR1 and CCR5 on the cell surface. At 120 dpi, the frequencies of CD8⁺ T-cells were similar in NI controls (12.35 ± 1.3%) and T. cruzi-infected (12.35 ± 1.3%) C57BL/6 mice. Compared with NI controls, we observed an increase in the frequency of TNF⁺, but not in IL-10⁺ and TNF⁺ IL-10⁺, among the CD8⁺ T-cells in the spleen of infected mice (Figure 7A). The frequency of CCR1⁺ was reduced (5.9 ± 1.7% in NI vs. 3.7 ± 1.9% infected mice; p < 0.05), whereas the frequencies of CCR1⁺ CCR5⁺ (0.34 ± 0.12% in NI vs. 1.2 ± 0.6% infected mice; p < 0.001) and CCR5⁺ (3.6 ± 1.2% in NI vs. 10.6 ± 1.5% infected mice; p < 0.001) were augmented among the CD8⁺ T-cells (Figure 7B). In NI mice, CCR1⁺ and CCR5⁺ CD8⁺ T-cells showed a predominance of IL-10-expressing cells, while the CCR1⁺ CCR5⁺ were mostly TNF⁺ (Figure 7C). At 120 dpi, we detected a decrease in the frequency of IL-10⁺ cells and an increase in the frequencies of TNF-expressing cells among the CCR1⁺, CCR1⁺ CCR5⁺ and CCR5⁺ CD8⁺ T-cells (Figure 7C). Indeed, TNF/IL-10 ratio tend to be elevated in CCR1⁺ (0.39 ± 0.17% in NI vs. 2.0 ± 1.3% infected mice; p = 0.092) and was increased in CCR1⁺ CCR5⁺ (1.6 ± 0.12% in NI vs. 6.8 ± 1.6% infected mice; p < 0.01) and CCR5⁺ (0.69 ± 0.16% in NI vs. 4.8 ± 1.9% infected mice; p < 0.05) CD8⁺ T-cells. Therefore, T. cruzi infection fuels with TNF expression the CD8⁺ T-cells expressing CCR1 and/or CCR5, inflammatory cells potentially able to respond to CC-chemoattractants as CCL3.

In chronic T. cruzi infection, macrophages (CD14⁺ CD45R⁺ F4/80⁺ cells) are mostly segregated in CCR5⁺ expressing TNF, while the CCR1⁺ mainly express IL-10 (16). Here, we showed an increased frequency of TNF⁺ cells among CCR1- and CCR5-bearing CD8⁺ T-cells. Therefore, we explored whether the blockage of CCR1/CCR5 with Met-RANTES affects T. cruzi-induced ECG abnormalities and the relation with CCL3 and TNF status in the heart tissue. For that, C57BL/6 mice were infected, analyzed at 120 dpi for ECG abnormalities, divided in two groups and treated with vehicle or Met-RANTES. At 150 dpi, animals were analyzed for ECG alterations and cytokine concentrations in the cardiac tissue (Figure 8A). At 120 dpi, average heart rates were reduced and PR and QTc intervals were prolonged, compared with NI controls (Figure 8B). At 150 dpi, in vehicle-treated mice ECG alterations progressed (Figure 8C). Conversely, blockage of CCR1/CCR5 with Met-RANTES more than hampered progression of ECG alterations as partially reversed the bradycardia and the prolonged PR in infected mice. Moreover, Met-RANTES administration almost completely reversed the prolonged QTc intervals present in chronically infected mice (Figure 8C). At 150 dpi, compared with saline-injected mice Met-RANTES therapy reduced chronic myocarditis (774 ± 146 cells/100 fields in saline vs. 547 ± 62 cells/100 fields in Met-R; p < 0.01), mostly due to reduction in the numbers of CD8⁺ (338 ± 93 cells/100 fields in saline vs. 228 ± 49 cells/100 fields in Met-R; p < 0.05) and F4/80⁺ cells (158 ± 39 cells/100 fields in saline vs. 97 ± 57 cells/100 fields in Met-R; p = 0.0744) and less to alteration in the numbers of CD4⁺ cells (278 ± 89 cells/100 fields in saline vs. 221± 39 cells/100 fields in Met-R; p > 0.05). Compared with age- and sex-matched NI controls, CCL3 concentrations were increased in the heart tissue of chronically T. cruzi-infected vehicle-injected mice, while CCL3 levels were diminished by Met-RANTES therapy (Figure 8D). Importantly, the TNF and, mainly, CCL3 concentrations in the heart tissue were directly correlated with QTc prolongation (Figure 8E).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.