This cross-sectional study used plasma samples of 84 individuals to measure circulating molecules. These individuals were divided in the following groups: (1) patients with CCC (CCC; n= 40) with positive serology for Trypanosoma cruzi, left ventricular enlargement accompanied by impaired systolic function and previous or current symptoms of heart failure (3); (2) patients in the indeterminate clinical form (IND; n= 32) of Chagas disease, with positive serology for T. cruzi and without clinical manifestation related to Chagas disease (21); (3) healthy donors (CTRL; n= 12), with negative serology for T. cruzi and normal ventricular function. The patients are from the estate of Minas Gerais, and were recruited at the outpatient clinic of the Hospital das Clínicas at the Federal University of Minas Gerais, where they are examined, characterized, and receive all care indicated by the responsible clinician and team. Exclusion criteria included: diabetes mellitus, thyroid dysfunction, renal insufficiency, chronic obstructive pulmonary disease, rheumatic heart disease, as well as any autoimmune disease or chronic inflammatory disorders. The demographic and clinical data of groups are represented in the Supplementary Table 1 .

Quantification of left ventricular size and function were obtained in accordance to the recommendations of the American Society of Echocardiography guidelines (22). Left ventricular diastolic diameter (LVDD) was measured at the end-diastole in the long-axis view, and left ventricular ejection fraction (LVEF) was derived by calculating the variance between end-diastolic and end-systolic volumes and then dividing it by the “end-diastolic volume”. We have expanded our analysis to encompass not only patients with Chagas cardiomyopathy but also those in the indeterminate phase of the disease. Correlation of cytokine expression with LVEF and LVDD from IND did not present any statistical significance, likely because these patients have normal left ventricular dimension with preserved LVEF. This additional information has been included in Supplementary Table 1 of the Supplementary Material .

We performed a second set of experiments measuring a higher number of molecules, as described below. This confirmatory experiment was performed in a subset of patients following the same clinical characterization described above, from the same region and recruited and cared for at the Hospital das Clínicas (UFMG). For this set, we enrolled 17 CTRL, and 30 CCC patients.

All individuals were informed about the objectives of study and signed the Informed Consent Form. Research Ethics Committee of Universidade Federal de Minas Gerais (COEP 93968218.7.0000.5149) approved this study.

Gene expression profile analysis included transcripts of human left ventricular free wall heart tissues from end-stage heart failure CCC patients (n= 10) and healthy individuals (CTRL, n= 7), using the accession number GSE84796 at the publicly available Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/).

Blood from all patients was obtained and used as a source of plasma, following protocol routinely used by us (23). Briefly, blood was collected in vacutainers containing heparin as anticoagulant. Material was processed within 4 hours after collection, submitted to centrifugation (10min, 400g at room temperature) for removal of plasma. Plasma was aliquoted and stored in cryotubes at -80 °C. Levels of soluble molecules were evaluated in the plasma of the different study groups using CBA assay (BD™ Human Th1/Th2/Th17 CBA; Cat # 560484). Data were acquired by the FACScantoII, according to manufacturer’s instructions and results expressed as the mean fluorescence intensity (MFI). The second set of experiments was performed using the 48-plex (Bio-Plex Pro ™ Human Cytokine Screening Panel; Cat #12007283) and read in the Bio-Plex 200 instrument equipped with the Manager software and the results were expressed as MFI.

The pattern of differentiation of soluble cytokines and chemokines between groups was analyzed using Clustvis software (R-version 0.7.7 package). Heat map analysis allows to identify clusters within the sample and to identify homogeneity or heterogeneity in the distribution of soluble molecules between the groups. Rows and columns are grouped using the correlation distance and the mean link. Principal component analysis (PCA) was performed, and the X and Y-axes show the percentage of the total variance. Prediction ellipses indicates a probability of 0.95 that a new observation will fall inside the ellipse (24).

Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) was used to determine gene expression profile in heart tissue from CCC and CTRL. GEO2R (http://www.ncbi.nlm.nih.gov/geo/geo2r/) web tool was used for comparing the two groups. The p-values using Benjamin and Hochberg false discovery rate method were applied in order to correct the occurrence of false positive results (25). An adj. P<0.05 and a logFC ≥ 1 were set as the cut-off criteria. ExpressAnalyst algorithm was used for visualization of functional categories enriched in a network (26).

Official gene symbol from NCBI was used as the identifier for CCC-related genes. The downregulated and upregulated genes list was used for the pathway enrichment analysis by PathfindeR package on RStudio software (version 4.2.1). The Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses were available at the PathfindeR package (27). The NetworkAnalyst.ca. was used to construct interaction network of differentially expressed genes (DEGs). Pairwise correlation prediction was determined based on IMEx database. High-scoring genes were used to identify hub genes. Enriched pathway analyses obtained as a result of interconnections in the network were generated using KEGG. Biological processes (BP) pathways enriched were measured by NetworkAnalyst.ca.

To predict the correlation between DEGs in CCC and immune system cells, gene clusters were used as input to the Correlation Network Hub Prediction (CNHP) tool of the Immuno-Navigator database (sysimm.ifrec.osac.osaka-u.ac.jp/immune-navigator/) (28).

Comparisons between plasma levels of soluble molecules between the study groups and correlation analysis were performed using the GraphPad Prism (version 8.0.2) software (GraphPad Software, La Jolla – CA, USA). Parametric data were submitted to the one-way ANOVA multiple tests followed by the Tukey post-test and represented by the mean and standard deviation. Nonparametric data were analyzed using the Kruskal-Wallis test, followed by Dunn’s post-test. Correlation analysis between plasma cytokines and chemokines with left ventricular ejection fraction (LVEF) or left ventricular diastolic diameter (LVDD) parameters were performed using Pearson or Spearman test, based on parametric or non-parametric data, respectively.

We evaluated the levels of cytokines and chemokines in the plasma from individuals with different clinical forms of Chagas disease. First, we compared the expression of the soluble factors amongst the different clinical groups. Our data demonstrated an increase in inflammatory cytokines IL-6, IFN-γ and TNF-α in CCC compared to IND and to CTRL ( Figure 1A ). There were no differences in serum levels of IL12p70 and IL-1β between the distinct clinical forms of Chagas disease or comparing patients with CTRL ( Figure 1A ). While IL-10 expression was higher in IND as compared to CCC and CTRL, no significant change was observed comparing CCC and CTRL ( Figure 1A ). A decrease in IL-17 plasma levels was observed in CCC compared to IND and CTRL ( Figure 1A ). Regarding chemokines, CXCL10 and CCL5 were increased in IND and CCC patients compared to CTRL, while CCL2, CXCL9 and CXCL8 were reduced in both clinical forms of Chagas disease compared to CTRL ( Figure 1A ).

We then compared the expression of each molecule between CCC and CTRL in each compartment. In the blood, we found that levels of IL-6, IFN-γ, TNF-α, CXCL-10, CCL-5 were higher in CCC as compared to CTRL; levels of IL-12, IL-1β and IL-10 were similar when comparing CCC and CTRL; levels of IL-17, CCL-2, CXCL-9 and CXCL-8 were lower in CCC than CTRL. We then evaluated the gene expression for these factors in the heart of CCC patients as compared to CTRL. Our data showed that IFN-γ, TNF-α, CXCL10, CCL-5 and CXCL-9 genes were upregulated in CCC as compared to CTRL, while no significant changes were observed in the expression of IL-12, IL-1β, IL-10, IL-17 and CXCL-8 genes; CCL2 and IL-6 were downregulated in CCC as compared to CTRL ( Figure 1B ). Thus, we observed a similar behavior in blood and heart off CCC compared to CTRL in 9 out 12 molecules: IFN-γ, TNF-α, CXCL-10 and CCL-5 increased in blood and heart of CCC as compared to CTRL; IL-12, IL-1β and IL-10 levels were similar in CCC and CTRL in blood and heart; CCL-2 decreased and CXCL-8 had a clear tendency of decreased expression in blood and heart from CCC as compared to CTRL. These data show a similar behavior in 10 out of 12 molecules (~80%) in blood and heart of CCC compared to CTRL, and a statistically significant equivalence (67%) between expression of immune molecules in blood and heart tissues. A second set of experiments were performed, increasing the number of molecules analyzed. Our data showed that, of the 43 molecules that we had information in plasma and heart tissue (5 of them did not have available the gene expression in heart), 27 (63%) had a similar behavior in blood and heart, and 23 had statistically significant corresponding changes (54%) ( Supplementary Table 2 ). This data confirms the high correspondence of the molecules between blood and heart, even considering a higher number of molecules analyzed.

To evaluate if the expression of these soluble mediators in the blood allowed for discrimination between the groups of Chagas disease patients, we performed heatmap and PCA analysis. Heatmap and PCA data showed clear segregation between CTRL and Chagas patients, and a segregation with a few overlaps between CCC and IND groups ( Figures 2A, B ). A strong correlation was observed between TNF-α, IFN-γ, IL-6, and CCL5 in CCC ( Figure 2A ), while a strong correlation between IL-10 and IL-17, IL-12 and IL-1beta was observed in IND ( Figure 2A ).

To determine the association between cytokine plasma levels with clinical parameters of heart dysfunction, left ventricular ejection fraction (LVEF) and left ventricular diastolic diameter (LVDD), we performed a correlation analysis between these variables in CCC group. We observed an inverse and statistically significant correlation between the inflammatory cytokines TNF-α, IL-6 and IFN-γ with LVEF, a worse prognosis predictor in Chagas disease (3), while IL-17 plasma level was positively correlated with LVEF and LVDD, indicating better heart function ( Figure 2C ). In addition, TNF-α, IL-6 and IFN-γ plasma levels were positive and statistically significant correlated with LVDD ( Figure 2C ), reinforcing that the levels of systemic inflammatory cytokines are correlated with worse heart function.

The circulating cytokines and chemokines that were differentially regulated in CCC were used to build a connectivity network and key enriched signaling pathways. Network analysis shows that inflammatory and regulatory cytokines present in the plasma of CCC compared to CTRL are intrinsically connected ( Figure 3A ). In addition, we found that soluble mediators differentially modulated in CCC as compared to CTRL mainly contribute to the enrichment of inflammatory signaling pathways such as: (i) cytokine-cytokine receptor interaction, (ii) IL-17 signaling, (iii) Chagas disease, (iv) Toll-like receptors (TLR), (v) TNF signaling and (vi) chemokine signaling ( Figure 3A ).

Genes that were differentially expressed in the heart of CCC as compared to CTRL and were concordant to expression in blood were included in the network and signaling pathway analysis, which also showed intrinsic connection between the analyzed molecules ( Figure 3B ). Similar to observed in blood, signaling pathway analysis also showed an enrichment of inflammatory pathways in the heart ( Figure 3B ). Fifteen out of the 20 enriched signaling pathways were similar between blood and heart (marked in bold in Figures 3A, B ).

To verify the overall profile of interactions and signaling pathways in the heart, and gain insight into in situ immunological events, analysis of the GSE84796 dataset for all differentially expressed genes (DEG) in CCC as compared to CTRL was performed. 2166 DEGs were identified, which included 1562 upregulated and 604 downregulated genes ( Figure 3C , red and blue dots, respectively). Of the 20 top enriched signaling pathways derived from all DEGs, 10 are directly linked to inflammatory pathways ( Figure 3C , marked in green). Strikingly, despite the large number of DEGs included in this overall analysis, 20% of the top pathways were similar to the ones observed in blood ( Figure 3C , marked with boxes), showing a clear biological correspondence between blood- and heart-derived molecule pathways in human Chagas disease.

To obtain insights into biological processes associated with the differentially expressed molecules in blood and heart tissue of CCC, we performed gene ontology enrichment analysis. We found that biological processes related to activation of the immune response and inflammatory responses were enriched in blood ( Supplementary Figure 1A ) and heart analysis ( Supplementary Figure 1B ). Importantly, of the top 15 biological processes, 10 were similar between blood and heart ( Supplementary Figure 1A , marked in bold).

Given that the T-cell receptor signaling pathway appeared as the top hit of the overall DEG analysis in the heart ( Figure 3C ), we employed the Immuno-Navigator tool to predict what T-cell populations were associated with the upregulated genes. For this purpose, we examined the interaction between the genes of TCR signaling pathway and CD4+, CD8+, regulatory (Treg) and memory (Tmem) T cells. Interaction network hub prediction indicates that genes upregulated in TCR signaling pathway are mainly correlated with CD8+ T cells, followed by CD4+, regulatory (Treg) and memory (Tmem) T cells ( Figure 4A ). Network analysis showed greater number of interactions within the CD8 as compared to the CD4 compartment ( Figure 4B ; Supplementary Figure 2 ). We also performed interaction network between NK cell-mediated cytotoxicity pathway, which is upregulated in the heart of CCC ( Figure 3C , third hit), with CD8+, CD4+ and mature NK cells, and observed that this pathway was mainly correlated with CD8+ T cells, followed by mature NK and CD4+ cells ( Figure 4C ). Although the inputted pathway was the NK cell-mediated cytotoxicity pathway, network analysis showed greater interaction with CD8+ cells even when compared with NK cells ( Figure 4D ; Supplementary Figure 2 ).