Adults (18 to 65 years old) with positive serology for Chagas disease in at least two distinct diagnostic methods, in accordance with the II Brazilian Consensus on Chagas disease (3), were included in this study at Alda Lima Falcão Outpatient Clinic at René Rachou Institute (IRR), Belo Horizonte, Minas Gerais, Brazil. Exclusion criteria comprised severe dilated cardiomyopathy, renal and/or liver failure, and comorbidities significantly impacting life expectancy.

Chagas disease patients were categorized by Chagas disease specialized physicians, according to the American Heart Association: Stage A (indeterminate, here denoted as asymptomatic form, n = 8) represents patients at risk for developing cardiac disease with no symptoms or altered electrocardiogram; Stage B1 (mild cardiomyopathy, n = 12) includes patients with electrocardiographic changes, but with normal global ventricular function and no current or previous heart failure symptoms; Stage B2/C/D (moderate/severe cardiomyopathy, n = 8) comprises patients with structural cardiomyopathy and global ventricular dysfunction and having or not heart failure symptoms. The control group (CTL) consisted of nine healthy donors from the National Institutes of Health blood bank, USA, and five from IRR, Brazil, testing negative for Chagas disease. Patients who were treated (32%) received benznidazole at 5 mg/Kg/day for 60 days according to the guidelines from Brazilian Consensus in Chagas Disease. Volunteers enrolled in different groups were similar in age and sex, but differences were found in both percentage of left ventricle ejection fraction, LVEF (%), and left ventricle diameter, LV (mm), between B1 versus B2-C-D. A concise overview of each patient’s metadata and demographics can be found in the ( Supplementary Table 1 ).

The protocols were approved by the Ethical Committee on Human Research at René Rachou Institute, Oswaldo Cruz Foundation (CEP-IRR- CAAE: 95998418.8.0000.5091). Patients provided written informed consent before their inclusion in the research.

Peripheral blood samples were collected in sodium heparin tubes, and after centrifugation at 1,000 x g, 10°C for 10 minutes, plasma samples were separated and stored at -80°C. Cells were reconstituted 1:1 (v/v) in RPMI 1640 (Sigma-Aldrich, St Louis, MO, USA) and the peripheral blood mononuclear cells (PBMC) were isolated by slowly transferring blood diluted in phosphate-buffered saline (PBS) (1:2) on top of Ficoll–Paque gradient (Sigma). Samples were centrifugated at 700 x g for 40 min without break. Subsequently, the PBMC layers were collected, washed, and cell concentrations and viability were determined using the Countess II FL automated cell counter (Invitrogen Thermo Fisher Scientific). PBMC were frozen in fetal bovine serum (FBS, Gibco) with 10% dimethyl sulfoxide (DMSO; SIGMA) at -80°C for 24 hours using a freezing container (Mr. Frosty™, Thermo Scientific) and then maintained in liquid nitrogen until use.

PBMC were thawed using thawsome adapted into a 15 mL tube containing RPMI 1640, 10% FBS, and benzonase nuclease (20 U/ml; Novagen, MilliporeSigma, Burlington, MA, USA) (32). Cell suspensions were washed in PBS and incubated with viability dye, diluted 1:1,000, for 15 min (Live/dead viability dye, UV blue, Thermo Scientific). After an additional wash in FACS buffer (PBS containing 0.1% BSA and 2 mM sodium azide), PBMC were incubated with two distinct antibody panels for 30 min (33). The panel designed for assessing CD4⁺ T cells and their subsets were based on (OMIP 58) (34) with modifications: anti-CCR6 (BB515, clone 11a90), CD57 (BB630, clone NK-1), CD244 (BB660, clone 2-69), CD127 (BB700, clone HIL-7R-M21), CXCR5 (BB790, clone RF8B2), TCR Vγ9 (PE, clone B3), TCR Vδ2 (PE-CF594, clone B6), CD161 (PE-Cy5, clone DX12), HLA-DR (PE-Cy5.5, clone TU36), ICOS (PE-Cy7, clone ISA-3), CD45RA (Ax700, clone HI100), CCR5 (APC-Cy7, clone 2D7/CCR5), CCR7 (BUV395, clone 150503), CD16 (BUV496, clone 3G8), CD56 (BUV563, clone NCAM16.2), CD279 (BUV661, clone EH12.1), CD95 (BUV737, clone DX27), CD4 (BUV805, clone SK3), CD122 (BV421, clone Mik-B3), CD3 (BV510, clone UCHT1), CD8 (BV570, clone RPA-T8), CD158 (BV605, clone DX27), CD28 (BV650, clone 28.2), CXCR3 (BV711, clone 1C6/CXCR3), CD38 (BV750, clone HB7), CD27 (BV786, clone L128). B cell subsets were assessed based on (OMIP 51) (35): anti-CD71 (FITC, clone M-A712), CD141 (BB630, clone 1a4), CD123 (BB660, clone 7G3), CD16 (BB700, clone 3G8), IgD (BB790, clone IA6-2), CD32 (PE, clone FUN-2), CD40 (PE-Dazzle594, clone 5C3), CD85j (PE-Cy5, clone GHI/75), CD11c (PE-Cy5.5, clone BU15), CXCR3 (PE-Cy7, clone G025H7), IgA (APC, goat polyclonal), CD27 (APC-R700, clone M-T271), CD19 (APC-H7, clone SJ25C1), CD1c (BUV395, clone F10/21a3), CD21 (BUV496, clone B-ly4), TACI (BUV563, clone 1a1-K21-M22), HLA-DR (BUV661, clone G46-6), IgG (BUV737, clone G18-145), CD20 (BUV805, clone 2H7), IL-21R (BV421, clone 2G1-K12), CD14 (BV510, clone MphiP9), IgM (BV570, clone MHM-88), BAFFR (BV605, clone 11C1), CD10 (BV650, clone HI10a), CD23 (BV711, clone M-L233), CXCR5 (BV750, clone RF8B2).

T and B cell frequencies were determined manually (supervised) within the parental population or using unsupervised approaches within CD4⁺ or CD19⁺ cells. Frequencies and mean fluorescence intensity were obtained using FlowJo (v10). Flow-Self-Organizing Map (SOM) (v2.1.10), Pheatmap (v1.0.12), and Rtsne (v0.15) packages in R (v4.1.2) were employed for unsupervised analyses. For FlowSOM training we defined 196 clusters for CD4+ T cell and 100 clusters for B cell. Those were clustered in 50 metaclusters or FSOM populations and their distribution was visualized as minimum spanning trees (MST) (36). Pheatmap package was employed to generate heatmaps with hierarchical clustering calculated using Manhattan or Euclidean distances. Dimensionality reduction was achieved through t-distributed Stochastic Neighbor Embedding (tSNE). In this process, tSNE was set to run 5,000 iterations with a learning rate (eta) of 10,000 and an exaggeration factor of 36 until 1,000 iterations, following which momentum was introduced. Cluster frequencies were exported for subsequent statistical analyses. The frequencies of cell populations within each clinical group were analyzed using GraphPad Prism (v.10.3). The Shapiro-Wilk test was employed to assess the parametric distribution of variables within each clinical group. Clinical groups were compared with the Kruskal-Wallis test, followed by the Dunn’s test for multiple comparisons with correction via the Bonferroni method. Data were represented using boxes and whiskers representing mean, maximum, and minimal values and symbols representing volunteers. Significant differences (p < 0.05) were represented by asterisks.

CD4⁺ T cell subsets were manually defined within viable single CD3⁺ cells, lacking T cell receptors TCRVγ9 and TCRVδ2 ( Figures 1A, B ). We defined naïve and memory subsets based on the expression of CD45RA and CCR7. CD45RA⁺CCR7⁺ cells consisted of naïve (CD95^-CD27⁺), stem cell-like memory (TSCM, CD95⁺CD27⁺), and CD27^- central memory cells (CM, CD27^-). Effector cells (Eff) were defined as CD45RA⁺CCR7^-, which we further distributed into CD27^- and CD27⁺ Eff. CD45RA^-CCR7⁺ consisted of central memory cells and were further classified as CM (CD27⁺) and tissue-resident memory (TRM, CD27^-). Lack of CD45RA, CCR7, CD28 and CXCR5 defined effector memory (EM), and of CD45RA, CCR7, CXCR5, but presence of CD28, were transitional memory cells (TSM). We identified Tfh cells based on the co-expression of CXCR5, PD-1, and ICOS within CM and EM subsets ( Figures 1A, B ).

CM cells were decreased in asymptomatic patients (A) compared to healthy donors (CTL). Patients with mild CCC (B1) displayed expansion of EM compared to CTL. Both mild and moderate/severe (B2/C/D) CCC patients showed increased frequencies of effector CD27^- cells. No differences were observed in the frequencies of naïve cells; however, patients with moderate/severe CCC showed a lower frequency of CD27^- CM among CD45⁺CCR7⁺ cells. Expansion of Tfh was observed in mild CCC compared to asymptomatic patients ( Figure 1C ).

We applied unsupervised clustering using a self-organizing maps algorithm (FlowSOM - FS) to further explore distinct CD4⁺ T cell phenotypes among patients with different clinical forms of Chagas disease ( Figure 2 ). A heatmap displays the heterogeneity of 50 clusters considering the entire study population. These clusters are composed of a distinct number of events (middle panel) and by distinct cell subsets according to the manual supervised gating (right panel) ( Figure 2A ). FS3, 4, and 9 consisted mainly of effector CD27^- and, to a lesser extent, of CM in naïve. FS7-8, FS1, and 11 comprise EM and a small proportion of effector CD27^-. Most FS populations (17-29 and 39-49) are composed of TSM and/or CM cells. The majority of FS28 and 31 consisted of Tfh cells. FS6-21 contain cells expressing high levels of CCR7 and CD27, corresponding mainly to TSCM, some CM, and CM in CCR7⁺CD45RA⁺ cells. Effector CD27⁺ cells are a rare population concentrated in the FS14, along with naïve cells ( Figure 2A ).

We performed dimensionality reduction using tSNE and superimposed the data generated by supervised ( Figure 2B , top panel) and unsupervised analyses (FS; Figure 2B , bottom panel) to evaluate qualitative differences in the cellular composition from patients with distinct clinical forms of Chagas disease. As shown in Figure 1C , we observed an increase in the frequency of effector CD27^-, TRM, and EM cells with the progression of Chagas disease (groups A, B1, and B2/C/D) compared to CTL ( Figure 2B , top panel). The tSNE generated with the FS populations revealed several other differences among the clinical forms ( Figure 2B , bottom panel), including cells placed in the top area of the tSNE, where naïve subsets are present.

Among the 50 FS populations, 14 had frequencies altered in at least one of the three clinical forms compared to CTL ( Figures 3A – 4A ). In addition, FS22 revealed different frequencies between asymptomatic and mild CCC patients ( Figures 3A, B ). In contrast, 35 clusters were similar among groups ( Supplementary Figure 1 ). FS11 was composed almost exclusively of EM and was the only population increased in Chagas patients, independently of the clinical form, compared to CTL ( Figure 3A ). The frequencies of FS3 and 9, mainly composed of effector cells, were expanded in both mild and moderate/severe CCC patients compared to CTL. FS3, 9, and 11 have high expression of CD244, consistent with activation and effector functions, as depicted in the minimum spanning tree (MST) ( Figure 3C , Supplementary Figure 2 ) (37, 38). FS3 also expresses high levels of CD57, a marker of highly activated cells (39, 40). The frequencies of FS22 were higher in mild CCC than in asymptomatic patients and CTL, while moderate/severe CCC showed a higher abundance of FS18 compared to CTL. Although clusters FS18 and 22 were composed mainly of transitional memory cells, they are essentially different; FS22 expresses a prominently activated phenotype characterized by the expression of high levels of PD-1, ICOS, CD28, CD95, CD122, CCR5, CXCR3, and HLA-DR, while FS18 expresses only high levels of CD28 and CD127 ( Figure 3C ). A higher frequency of FS28 was found in mild CCC compared to asymptomatic patients. FS28 is a heterogeneous population, composed of Tfh, transitional, and CM cells and characterized by high levels of CXCR5, PD-1, ICOS, and CD28 ( Figure 3C ).

Clusters FS1, 7, 8, 10, and 46 were expanded only in mild CCC patients compared to CTL ( Figures 5A, B , top to bottom). FS01, 07, and 08 mainly composed of EM cells, while FS10 and FS46 mainly contained TM, EM, and effector CD27^- cells. Most FS with EM phenotype (FS1, 2, 7, 10) express high levels of CD57. FS01, 08, and 10 also express CD244, and FS08 expresses high levels of CD56. FS46 expresses high levels of CD161, which is associated with a memory phenotype (41), CCR5, CD127 and CD28 ( Figure 5C ).

A few FS populations had lower frequencies in CCC patients than in CTL ( Figure 4A, B ). Frequencies of FS13, FS20, and FS32 were lower in mild CCC than in CTL. FS13 was the most heterogeneous population, composed of CM, TSCM, and, to a lesser extent, of CM in CD45RA⁺CCR7⁺ and TSM cells. FS20 was mainly composed of cells with TSCM phenotype and FS32 of TSM and CM cells. These FS express very low activation markers, suggesting their central memory and resting status. Similarly, FS30 and FS48 were composed of TSM and CM cells, but their frequencies were lower in moderate/severe CCC than CTL. Both express high levels of CCR6, but F30 expresses high levels of CD95 and is less activated, while FS48 shows high levels of CD161, CD127, and CD28 ( Figure 4C ).

Taken together, CD4⁺ T cells from patients with Chagas disease bear an effector and activated phenotype that is pronounced in the mild CCC clinical form.

Distinct T cell subsets differently impact the development and differentiation of B cells. We assessed the composition of B cell compartments based on most of the peripheral B cell differentiation stages. B cells were defined as CD19⁺CD20⁺CD21⁺, while plasma cells were defined as CD19⁺ lacking CD21, CD20, IgM, and IgD but expressing CD27 ( Figure 6A , Supplementary Figure 3 ). From B cells, we defined transitional B cells (CD10⁺IgD⁺) and mature B cells, which were further divided into unswitched and class-switched cells based on the expression of IgM and IgD. Class-switched B cells (IgM^-IgD^-) were further grouped based on IgG⁺, IgA⁺ or IgG^-IgA^-, and each of these isotype-expressing switched B cells consists of Activated Memory (AM, CD21^-CD27⁺), Atypical Memory (AtyM, CD21^-CD27^-), Intermediary Memory (IntM, CD21⁺CD27^-) and Resting Memory (RM, CD21⁺CD27⁺) ( Figure 6A ). Unswitched cells were classified as IgM memory (IgM⁺IgD^-), marginal zone (MZ, IgM⁺IgD⁺CD27⁺), IgD memory (IgM^-IgD⁺CD27⁺), and naïve (IgD⁺CD27^-CD21⁺ or CD21^-) ( Figure 6A , Supplementary Figure 3 ).

Chagas disease triggers the expansion of AM B cells in asymptomatic and mild CCC patients, independently of the Ig isotype ( Figure 6B ). In addition, AM IgA cells were elevated in moderate/severe CCC. Chagas patients showed higher frequencies of AtyM IgG compared to CTL. AtyM IgA and IgG^-IgA^- B cells display higher frequencies in mild CCC and asymptomatic patients, respectively, compared to CTL ( Figure 6B ). No differences occurred in the frequencies of IgM and IgD memory B cells ( Figure 6C ), naïve CD21^- ( Figure 6D ), marginal zone, and plasma cells ( Figure 6E ). However, naïve CD21⁺ ( Figure 6D ) and transitional ( Figure 6E ) B cells were decreased in patients with mild and moderate/severe CCC, respectively, compared to CTL.

As for T cells ( Figure 2A ), we applied FlowSOM to explore distinct B cell phenotypes in patients with Chagas diseases ( Figure 7 ). We thus identified several heterogeneous clusters consisting of plasma cell (FS 45-44), transitional (FS7), and class-switched (FS29-46) B cell subsets ( Figure 7A , Supplementary Figure 3 ). Furthermore, FS15-9 predominantly represent CD21⁺ naïve B cells while FS16 and 32 are CD21^- naïve B cells. FS20, 26-34 are mostly constituted of marginal zone B cells. FS43-49 and FS48-47 predominantly consist of IgG⁺ and IgA⁺ B cells, respectively ( Figure 7A ).

In the dimensionality reduction obtained by supervised ( Figure 7B , top panel) analysis, we observed an increase in the frequency of class-switched subsets: in IgG, IgA, and IgG^-IgA^- AM and AtyM B cells with the progression of Chagas disease (groups A, B1 and B2/C/D). On the other hand, naïve and transitional B cells were decreased in patients with CCC compared to CTL, especially in moderate/severe CCC group. tSNE generated with the FS populations ( Figure 7B , bottom panel) revealed several differences among the clinical forms and CTL ( Figure 7B , bottom panel).

Eight FS were altered in at least one of the clinical forms compared to CTL; one was altered between asymptomatic and mild CCC patients, and another was altered between asymptomatic and moderate/severe CCC ( Figure 8 ). Forty FS were similar among groups ( Supplementary Figure 4 ). Higher frequencies of FS 30 and 41 were observed in Chagas patients compared to CTL, regardless of the clinical form ( Figures 8A, B ). Both FS populations are composed of AtyM cells, but express different levels of co-receptors and activation markers such as CD20, CD85j, CD40, TACI, CD32, among others ( Supplementary Figures 5, 6 ). Moderate/severe CCC patients have lower frequencies of FS15 and FS18 than CTL and FS25 than asymptomatic patients. Those are composed of naïve and transitional cells. FS15 expresses high levels of IgD and intermediate levels of CD20, CD40, and HLA-DR; FS25 expresses high levels of IgM and IL-21R and intermediate levels of CD20, CD40 and BAFFR; and FS18 expresses high levels of both IgM and IgD, CD20, CD40, CD32, HLA-DR, BAFFR and TACI, and does not express IL-21R ( Figures 8A, B , Supplementary Figure 6 ).

FS32 was expanded in asymptomatic and mild CCC patients compared to CTL and was characterized by the expression of IgD and high levels of CD20, HLA-DR, CD85j, and CD32 ( Figures 8C, D , Supplementary Figure 6 ).

FS37, 40, and 45 were exclusively increased in mild CCC compared to CTL. FS37 consists mostly of IgA cells and comprises AM, AtyM, RM, and IntM cells. This cluster is characterized by intermediate levels of BAFFR, CD20, HLA-DR and CD40. FS40 contains IgG-expressing AM and AtyM cells and displays high levels of CD20, CD85j, CD11c, TACI, HLA-DR, CD32, and intermediate levels of BAFFR and CD40. FS45 represents plasma cells expressing high levels of CD27 and CD71. The only FS that decreased in mild CCC compared to CTL was FS12, a naïve subset with increased expression of IgD, CD40, HLA-DR, CD23, BAFFR, and CD64 ( Figures 8C, D , Supplementary Figure 6 ).

In general, Chagas disease leads to expansion of activated, class-switched B cell subsets, which is more prominent in mild CCC clinical form.