Cytotoxic CD4+ T cells driven by T-cell intrinsic IL-18R/MyD88 signaling predominantly infiltrate Trypanosoma cruzi-infected hearts

Article activity feed

1. 

   eLife

   Jun 14, 2022

   Author Response:

   Reviewer #1:

   The manuscript by Bellio and colleagues is based on the experimental model of T. cruzi infection in WT, MyD88-/- and IL-18-/- mice previously described by the same group in a 2017 eLife publication. The main message of the current study is that, in addition to IFN-g+ Th1 effectors, T. cruzi infection induces an even larger population of cytotoxic CD4+ T cells.

   The characterization of the cytotoxic CD4+ T cells is well documented. The data shown are convincing. However, since Burel et al. (2012) described the existence of a similar population in humans infected with P. falciparum (an intracellular pathogen), the authors should modify the statement (line 35-36) in the abstract.

   First, we would like to thank Reviewer #1 for the positive comments on our work.

   Please note that our statement in the abstract …

   Author Response:

   Reviewer #1:

   The manuscript by Bellio and colleagues is based on the experimental model of T. cruzi infection in WT, MyD88-/- and IL-18-/- mice previously described by the same group in a 2017 eLife publication. The main message of the current study is that, in addition to IFN-g+ Th1 effectors, T. cruzi infection induces an even larger population of cytotoxic CD4+ T cells.

   The characterization of the cytotoxic CD4+ T cells is well documented. The data shown are convincing. However, since Burel et al. (2012) described the existence of a similar population in humans infected with P. falciparum (an intracellular pathogen), the authors should modify the statement (line 35-36) in the abstract.

   First, we would like to thank Reviewer #1 for the positive comments on our work.

   Please note that our statement in the abstract is: “Here, for the first time, we showed that CD4CTLs abundantly differentiate during mouse infection with an intracellular parasite” refers to mouse experimental models of parasite infection and not to human studies. We could not find any article with Burel JG as first author published in 2012; we believe that Reviewer# 1 is referring to a study published in 2016 (Burel et al. PLoS Pathog. 2016 Sep 23;12(9):e1005839), in which a population of CD4 T cells with cytotoxic properties was described in humans after primary exposure of blood-stage malaria parasites. Please note that the finding of the important role of T-cell intrinsic IL- 18R/MyD88 signaling for the development of a strong CD4CTL response is also part of the main message of our manuscript.

   Similarly, the title "Cytotoxic CD4+ T cells… predominantly infiltrate Trypanosoma cruzi-infected hearts" is an overstatement. If cytotoxic CD4+ T cells outnumber 10:1 IFN-g-secreting population (in lymphoid tissue) their higher representation in hearts of infected mice is not a selective phenomenon but rather expected.

   We would like to thank Reviewer #1 for this comment, giving us the opportunity to clarify this point. Of note, we were not referring to the ratio of CD4CTL to Th1 cells, but to the frequency of CD4CTL among all the CD4+CD44+ (activated/memory) T cells. In fact, as shown in Figure 7-figure supplement 2, (now added to the revised ms), we found that the frequency of GzB+ cells among all activated/memory CD4+CD44hi T cells is significantly increased in the heart compared to the frequency of GzB+ among CD4+CD44hi T cells found in the spleen. Please also note that the frequency of CD4+ T cells expressing both GzB and PRF also increases in the heart compared to the spleen (Fig. 7F, middle panel and Fig. 1D left panel). We are now including this information in the revised manuscript, clarifying this point.

   My major concern is that the function of these cells remains undefined. Are they beneficial or detrimental for the host? It appears that the authors themselves could not make up their minds. The GzB+ CD4+ T cells protect but do not decrease the parasite load (Fig 6G).

   Our results in the mouse model of infection with T. cruzi, employing the adoptive transfer of WT CD4+GzB+ T cells to the susceptible Il18ra-/- mouse strain, indicate a clear beneficial role of CD4CTLs in the acute phase of experimental T. cruzi infection. Significantly extended survival was observed in the group of mice receiving sorted CD4+GzB+ cells, without, however, decreasing parasite load (Figure 6G). We would like to comment here that in order to be beneficial to the host, an immune response does not always result in decreasing the pathogen load. In fact, in certain circumstances, to hinder the excessive inflammatory response (which can lead to host death), is an advantage for the host, even if this does not result in the reduction of the pathogen numbers. The advantage conferred to the host by regulating the inflammatory response was probably also explored in pathogen/host co-evolution, giving rise to chronic infections, where the host can survive for a longer period and the pathogen increases its chances of transmission (Schneider DS & Ayres JS., 2008, Nat Rev Immunol;8(11):889; Medzhitov R, et al, 2012, Science; 335(6071):936). Therefore, the results shown on Figure 6G are fully compatible with a potential regulatory role exerted by CD4CTLs, previously proposed by other authors (Mucida et al, Nat. Immunol. 2013), and point to the beneficial role of CD4CTLs for the host in the acute phase of infection with T. cruzi, probably by contributing to the decrease of immunopathology, the detrimental side of an exacerbated immune response, as discussed. Also favoring this hypothesis, the frequency of CD4CTLs expressing immunoregulatory molecules is increased when compared to other activated CD4+T cell subsets (Figure 3 and new Figure 7-figure supplements 3 and 4). Please see our complete discussion on this subject in the revised manuscript.

   On the other hand, during the chronic phase of the disease, the persistence of the immune response against the parasite might involve functional changes in the CD4 T cell response. This hypothesis could explain the association found between CD4CTLs and cardiomyopathy in chronic Chagas patients. Therefore, a beneficial role for CD4CTLs in the acute phase is totally compatible with the hypothesis that, during the chronic response in a persistent infection, CD4CTLs might acquire a detrimental role, contributing to immunopathology. Of note, several studies in the literature have shown a beneficial role for Th1 cells during the acute phase of infection with T. cruzi, while the Th1 response has also been associated to a pathologic outcome during the chronic phase of Chagas disease (reviewed in Ferreira et al, 2014 World J Cardiol 2014 6(8):7820 and in Fresno & Girones, 2018, Front.Immunol. 9;351). Therefore, it is not implausible that the CD4CTL subpopulation, could also display different roles in the acute versus the chronic phases of the infection with T. cruzi. However, at present, this hypothesis remains speculative as stated in the manuscript discussion. An extensive investigation of the role of CD4CTLs, as well as of immunoregulation mechanism acting in chronic Chagas patients need to be conducted to fully answer this question, which is beyond the scope of the present work. Nevertheless, we acknowledge that the alternative possibility remains, in which the higher levels of CD4CTLs in chronic patients reflect elevated parasite burden and/or inflammation in the heart, without a direct involvement of this cell subset in the pathology. Please see our answer to Review #2 on this topic and the inclusion of discussion clarifying this point in the revised manuscript.

   Are they terminally differentiated or "exhausted" effectors? GzB+ CD4+ T cells can be found in the hearts of chronically infected mice, but we do not know if they are specific for pathogen or self Ags. Do they express the markers of exhaustion on day 14 in the heart?

   1. We have commented in the first version of the manuscript that one of the limitations of our work is the fact that very few CD4 epitopes of T. cruzi presented by I-Ab have been described so far, and this limits the investigation on the specificity of CD4CTLs in our model. This is a very interesting and important question, which, however, is not possible to address in the present work.

   We would like to thank Reviewer#1 for the suggestion of performing a broader analysis on the expression of immunoregulatory markers associated with exhaustion and/or terminal differentiation, which adds for the comprehension of CD4CTL biology in the model of acute infection with T. cruzi. Whether GzB+CD4+ T cells are terminally differentiated or "exhausted" effectors is an interesting and debated question. It was initially hypothesized that since exhausted T cells share features with terminally differentiated T cells, this would suggest a developmental relationship between these cell states (Akbar, A.N. & Henson, S.M., 2011 Nat. Rev. Immunol.11:289; Blank, C.U. et al, 2018, Nat.Rev.Immunol,19:665). However, subsequent studies showed that exhausted T cells seem to be derived from effector cells that retain the capacity to be long-lived (Angelosanto, J.M. et al., 2012, J. Virol. 86: 8161). In the first version of our manuscript, we investigated the expression of several markers associated with exhaustion such as 2B4, Lag-3, Tim-3 and CD39, besides the downregulation of CD27 on GzB+ CD4+ T cells (Figures 1E, 3B, 3D-E and 5E). In general, cells losing the expression of CD27 have been characterized as Ag-experienced further differentiated cells (Takeuchi and Saito, 2017, Front.Immunol. 8:194). Our finding that, differently from GzB-negative cells, most GzB+CD4+ T cells had lost the expression of CD27, suggested to us that CD4CTLs present in the spleen of mice infected with T. cruzi might be further differentiated T cells (Figure 3E). The transcription factor Blimp-1 controls the terminal differentiation of cells in a variety of immunological settings and its high expression in CD4+ and CD8+ T cells is associated to the expression of immunoregulatory markers (Chihara, N. et al, 2018, Nature 558:454). The observed high expression of Blimp-1 by GzB+CD4+ T cells (Figure 5D) is also compatible with the hypothesis that CD4CTLs are terminally differentiated. Of note, most of the exhaustion studies were performed on CD8+ T cells and it is still not well established if this phenomenon is equally regulated in CD4+ T cells. We have now extended the investigation on the expression of terminal differentiation/exhaustion markers, including PD-1 staining, on GzB+PRF+ CD4+ T cells in the spleen and in the heart of infected mice. Results in Figure 7-figure supplement 3, show that CD44hiGzB+PRF+ CD4+ T cells compose the subset of activated cells among which the higher frequency of cells expressing these markers is found, both in the spleen and in the heart, at day 14 pi. The only exception was the equal ratio of cells expressing PD-1, and at equivalent levels, when comparing CD44hiGzB-PRF- and CD44hiGzB+PRF+ CD4+ T cells in the spleen. Non-significant differences in the percentages of cells expressing PD-1 among CD44hiGzB-PRF- and CD44hiGzB+PRF+ CD4+ T cells were found in the heart. However, the intensity of expression of the PD-1 marker (MFI) was significantly higher among CD44hiGzB+PRF+ compared to CD44hiGzB-PRF- CD4+ T cells infiltrating the heart. Furthermore, we also compared the frequency of CD44hiGzB+PRF+ CD4+ T cells expressing Lag-3, Tim-3, CD39 and PD-1, and their corresponding MFI values, between the spleen and the heart (Figure 7-figure supplement 4). Of note, while MFI values of Tim-3, CD39 and PD-1 expression were increased on CD4CTLs (CD44hiGzB+PRF+) in the heart compared to CD4CTLs in the spleen, Lag-3 expression levels were decreased on CD4CTLs infiltrating the cardiac tissue. Despite exhaustion being often seen as a dysfunctional state, it is important to note that the expression of these inhibitor molecules allows strongly activated T cells to persist and partially contain chronic viral infections without causing immunopathology and that highly functional effector T cells can also express such inhibitory receptors (reviewed in Wherry, E.J., 2011, Nat. Immunol.,12:492; Blank, C.U. et al, 2018, Nat. Rev. Immunol., 19:665). Interestingly, only PD-1, but not Lag-3, Tim-3 or CD39 expression is upregulated on CD8CTLs in the heart relatively to the spleen, an indication that the T. cruzi-infected cardiac tissue is a less so-called exhaustion-inducing environment compared to certain tumors (Figure 7- figure supplement 4). It is known that many immunomodulatory molecules, including Lag-3, Tim-3, PD-1 and CD39 are co-expressed as part of a module composing a larger co-inhibitory gene program, which is expressed in both CD4+ and CD8+ T cells under certain activation conditions, driven by cytokine IL-27 (Chihara, N. et al, 2018, Nature 558:454). The opposing behavior of Lag-3 expression, which is downmodulated on CD4CTLs in the heart in comparison to the spleen, indicate that CD4CTLs infiltrating the heart are not typically exhausted cells. Of note, a recent study has shown that exhausted CD8+T cells can partially reacquire phenotypic and transcriptional features of T memory cells, in a process that includes the downmodulation of Lag-3 expression (Abdel-Hakeem, M.S. et al, 2021, Nat.Immunol., 22:1008). As requested, these new data were included (Figure 7-figure supplements 3 and 4) and discussed in the revised manuscript.

   The factors that control differentiation of cytotoxic CD4+ T cells are the same as for IFN-g- Th1 cells. MyD-88-/- and IL-18-/- mice significantly lack both populations and succumb to T. cruzi infection. In their 2017 eLife publication, this group reported that survival of infected MyD-88-/- and IL-18-/- mice can be rescued by adoptive transfer of purified total WT CD4+ T cells, which was attributed entirely to their ability to secrete IFN-g (at least in the case of MyD-88-/- recipients). In the current study, the authors only used infected IL-18-/- recipients and show that this time transfer of GzB+ CD4+ T cells is sufficient to confer the protection. When compared with the old data, the rescue of the infected IL-18-/- with only GzB+ CD4+ T cells looks weaker (2 surviving animals out of 10 pooled from 2 experiments), strongly suggesting that IFN-g Th1 cells do play a significant role. It is unclear when the parasite load in Fig G6 was evaluated. It would be good to show deltaCT values for individual mice.

   We thank Reviewer #1 for the opportunity to clarify the point on the protective role of Th1 and CD4CTLs cells during T. cruzi infection and to better discuss our data. Please note that we do not question the beneficial role of Th1 cells in this infection model. In our paper published in 2017 in eLife, we have shown that the adoptive transfer of IFN-g- deficient CD4+ T cells do not result in the decrease of parasite loads in susceptible recipient mice. These results are totally in agreement with the known beneficial role of Th1 cells during infection with T. cruzi, through the microbicidal action of IFN-g, which was also described by other groups.

   The new information that our present study brings is that the adoptive transfer of GzB+CD4+ T cells with poor (GzB-YFP+) or no (Ifng-/-) capacity of IFN-g secretion, also significantly extended survival of infected Il18r-/- mice, which have lower levels of both Th1 and CD4CTLs, compared to WT mice (Figure 6G and Figure 6-figure supplement 2). Please note that 3 (not 2) out of 10 mice that received GzB+CD4+ T cells survived. We stated in our discussion that, together, our present and past data demonstrate that both Th1 and CD4CTL are important for improving survival, although through different mechanisms, since adoptively transferred GzB+CD4+ T cells (as well as Ifng-/- CD4+ T cells) were not capable of reducing parasite load but, notwithstanding, extended survival.

   Following the guidelines of the Animal Care and Use Committee, in order to prevent/alleviate animal suffering, all laboratory animals found near death must be euthanized. Therefore, parasite load in the hearts was evaluated in mice found at the moribund condition (a severely debilitated state that precedes imminent death, as defined in Toth, L.,2000; ILAR J, 41:72), presenting unambiguous signals that the experimental endpoint has been reached. We have now included 2ˆDeltaCT values for individual mice in Figure 6G, as requested.

   Because donor IFN-g-/- CD4+ T cells do express IFN-gR (Supp Fig 6-2), IFN-g produced by IL-18-/- host cells could enhance the activity and/or help expand cytotoxic CD4+ T cells among the IFN-g-/- CD4+ donor population. To directly test the protective role of cytotoxic CD4+ T cells in the absence of IFN-g, the authors should treat infected IL-18-/- mice that have received IFN-g-/- CD4+ T cells with anti-IFN-gamma Ab.

   It is known that IFN-g is critically important for resistance against infection with T. cruzi. Accordingly, Ifng-/- mice are extremely susceptible, dying at early time points of infection (Campos, M. et al, 2004, J.Immunol, 172:1711). Of note, IFN-g production by other cell types, and not only derived from CD4+ T cells, is relevant for resistance against infection, as demonstrated for CD8+ T cells (Martin D & Tarleton R. Immunol Rev. 2004, 201:304). In our present work, we performed experiments where Ifng-/- CD4+ T cells were adoptively transferred to susceptible Il18ra-/- mice, with the goal of testing whether the transferred cells would be able to confer some increment in the survival time of infected mice, despite of not being able to decrease parasite loads, a direct consequence of their deficiency in IFN-g production, as previously shown (Oliveira et al., 2017, eLife). In fact, this turned out to be the case and we showed that the transfer of purified Ifng-/- CD4+ T cells extended survival (Figure 6-figure supplement 2). Of note, our data demonstrate that the percentage of GzB+CD4+ T cells is not affected in the total absence of IFN-g, since Ifng-/- mice display the same frequency of this cell population as found in WT mice (Figure 4B). The increased survival of adoptively transferred mice is compatible with a regulatory function of GzB+CD4+ T cells, which additionally express several immunoregulatory molecules, as shown. Whether IFN-g produced by the host is enhancing the activity and/or expanding cytotoxic CD4+ T cells among the transferred T cell population is not an essential point here, since we were not aiming to test the protective role of cytotoxic CD4+ T cells in the total absence of IFN-g in the host mice.

   The intracellular cytokine staining in this study appears to be suboptimal. Instead of stimulating with PMA/ionomycin in the presence of Golgi block, Roffe et al. (2012) stimulated lymphocytes with anti-CD3 prior to adding Brefeldin A, an important technical difference which may explain the rather low frequencies of IFN-g+ and IL-10+ cells in this study.

   We respectfully disagree from Reviewer #1 on this point. The frequency of IFNg+ CD4+ and IL-10+CD4+ T cells in the spleen of mice infected with T. cruzi Y strain obtained in our experiments is in the same range to what was previously described by other research groups investigating the immune response to this parasite, including studies that have employed anti-CD3 stimulation and brefeldin A, such as Jankovic, D. et al, 2007, JEM 204:273 (Fig.S1), cited in our manuscript (page 9, lines 218-219), among others (Nihei J et al, 2021, Front. Cell. Infect. Microbiol.11:758273; Martins GA et al, 2004, Microbes Infect 6:1133 – Fig.6B; Hamano S. et al, 2003, Immunity, 19:657- Fig. 2A). In the present work, we used the combination of monensin and brefeldin A after PMA/iono treatment, and found the same frequency of IFN-g+CD4+ T cells described in a previous study of our group, where staining was performed after incubation of splenocytes with parasite-derived protein extract and brefeldin A alone (Oliveira AC et al., 2010, PLoSPath 6(4):e1000870 –Fig. 8D). On the other hand, please note that the study cited by Rev. #1 (Roffe et al., JI 2012) employed a different strain of T. cruzi, the Colombiana strain, which differs in several aspects from the Y strain used in our work. Colombiana induces a different pathology, with distinct kinetics. In that study, intracellular IFN-g and IL-10 detection was performed at a much later time point of infection (day 30 pi), and in cells infiltrating the heart, not the spleen. In summary, frequencies of IFN-g and IL-10 secreting CD4+ T cells described in our manuscript are comparable to the ones found in the spleen of mice infected with the same or similar strains of T. cruzi and reported in articles of prestigious journals by other groups, cited above.

   Reviewer #2:

   In this work, Professor Bellio and her colleagues provide compelling evidence to show unusually strong induction of cytotoxic CD4 T cells (CD4CTLs) in Trypanosoma cruzi-parasitized mice. Using genetic models and mixed bone marrow chimeras they dissect the signals responsible for CD4CTL induction in this infection and identify T cell-intrinsic IL-18R/MyD88 signaling as the key inducer. The CD4CTLs that clonally expand in T. cruzi infection outnumber CD4 cells with typical Th1 profile (IFN-γ secretion) and bear the hallmarks of CD4CTLs described in other model systems and in humans. Utilizing GzmbCreERT2/ROSA26EYFP reporter mice, the authors show that adoptive transfer of CD4 cells that have made GzB can increase the survival of T. cruzi parasitized l18ra-/- mice. Finally, the authors describe a clear correlation between the frequency of CD4CTLs the circulation of patients with T. cruzi-induced chronic Chagas cardiomyopathy, implying a pathogenic role for these cells in chronic disease.

   The findings reported here are an important addition to the understanding of both the origin of CD4CTLs and their potential role in host protection or disease. The evidence provided in support of the main claims is very strong and the association between CD4CTLs and Chagas disease quite intriguing. There are, however, some aspects of the work that would benefit from further clarification or experimental support, so that alternative interpretations of the data can be excluded.

   The defining characteristic of CD4CTLs that separates them from other CD4 subsets is the production of granzymes and perforin and, by extension, the ability to kill target cells in a granzyme/perforin-dependent manner. In contrast, all T cells can kill target cells via alternative mechanisms that are not dependent on granzyme/perforin, for example through expression of TNF family members. It would appear that much, if not most, of the killing activity of T. cruzi-induced CD4CTLs can be attributed to FasL (Fig. 1B). FasL-mediated killing is not restricted to CD4CTLs and as the title of one of the cited studies (Kotov et al., 2018) states, "many Th cell subsets have Fas ligand-dependent cytotoxic potential". It would be important to ascertain if expression of granzyme/perforin by CD4CTLs in T. cruzi infection is also associated with granzyme/perforin-dependent cytotoxicity. This affects the direct and indirect in vitro cytotoxicity assays, as well as the interpretation of in vivo protection.

   Similarly, the protective effect of transferring GzmbCreERT2/ROSA26EYFP reporter-positive cells to Il18ra-/- mice may not be necessarily mediated in a granzyme/perforin-dependent manner or by CD4CTLs for that matter. The reporter will mark cells that express GzB at the time of tamoxifen administration but does not guarantee that these cells will continue to express GzB or that they will prolong survival of recipients in a granzyme/perforin-dependent manner.

   While the authors provide evidence that GzB-producing cells are largely distinct from IFN-γ-producing cells, the reporter-positive cells may still contain genuine Th1 cells. Given Th1 cells have been previously found necessary for protection of Il18ra-/- mice in the T. cruzi model, can a role for Th1 cells in this transfer model be formally excluded? The authors do convincingly demonstrate that IFN-γ itself is not essential for protection, but that does not leave granzyme/perforin-dependent as the only other alternative. For example, the experiment described in Fig. 6G lacks an important control, the transfer of reporter-negative cells. What would the conclusion be if reporter-negative (but T. cruzi-specific) cells proved as protective as reporter-positive cells?

   We would like to thank Reviewer #2 for the positive comments on our study and for giving us the opportunity to better discuss and clarify the relevant points raised in this review.

   (i) Concerning the role of GzB/PRF in cytotoxicity: as explained in more details in our next answer to Reviewer #2, we have now shown that the cytolytic activity of the CD4 T cell subset differentiating in the murine T. cruzi-infection model is totally dependent on a GzB- and PRF-mediated mechanism.

   (ii) Concerning a possible role for Th1 in the adoptive transfer experiments: please note that the parasite load is not decreased by the adoptive transfer of CD4+GzB+ T cells (Figure 6G); Additionally, we showed that the adaptive transfer of Ifng-/- CD4+ T cells also extend the survival of infected mice (Figure 6-figure supplement 2), but did not decrease parasite levels (Oliveira et al., 2017). These results exclude a role for Th1 cells, which are known to exert an important microbicidal function through the production of IFN-g, as previously demonstrated by us (Oliveira, 2017) and other groups. Together, our present and past data support the notion that both Th1 and CD4CTL are important for extending survival, although through different mechanisms. Our results are in accordance with an immunoregulatory role played by CD4CTLs, likely through the GzB/PRF/FasL-mediated killing of infected APCs in an IFN-g-independent manner, although it is not possible to attribute the beneficial role of the adoptively transferred CD4CTLs exclusively to their cytolytic function, as discussed in the revised manuscript. Of note, we also show here that most CD4+GzB+PRF+ T cells express high levels of immunomodulatory molecules, raising the possibility that the beneficial role of adoptively transferred CD4CTLs might rely on the concerted action of their cytolytic function and immunomodulatory activity. Please see the full discussion on this point in the revised version of the manuscript.

   (iii) Concerning the adoptive transfer of GzB-EYFP-negative cells: unfortunately, GzB-EYFP-negative cells cannot be employed as a control, since in the GzmBCreERT2/ ROSA26EYFP mouse line age, only 1 - 3 % of total splenic CD4+ T cells express EYFP after induction by tamoxifen (Figure 2-figure supplement 3). This contrasts to 10-40% of GzB+ and PFR+ cells among CD4+ T lymphocytes, observed by intracellular staining. Consequently, the majority of the CD4+GzB+ T population is EYFP-negative in this system and thus, sorted “GzB-EYFP-negative”, based on the absence of expression of EYFP, would not be bona-fide GzB-negative cells. If it were possible to sort GzB reporter-negative cells, Th1 cells would be among the sorted cells and upon adoptive transfer they would secrete IFN-g and, consequently, decrease the parasite load in recipient mice (Oliveira, 2017). However, in the absence of the proposed immunoregulatory action of CD4CTLs, Th1 cells transferred alone might also increase pathology and, consequently, it is possible that they would not extend survival, albeit diminishing parasite load. It is expected that higher levels of extended survival would be attained when both Th1 and CD4CTLs are transferred, as discussed in the manuscript and in answer (ii) above. Importantly, please note that one current hypothesis is that CD4CTLs differentiate from Th1 and, therefore, the adoptive transfer of Th1 cells will not guarantee that Th1-derived CD4CTLs would not be developing in vivo, unless special engineered mouse strains, not available at present, would be employed for these experiments.

   Reviewer #3:

   By modelling trypanosoma cruzi infection in mice, the authors highlighted the presence of a subsets of CD4 T cells expressing canonical markers and transcription factors of CTLs and capable of exerting antigen specific and MHC class II restricted cytotoxic activity. Mechanistically, using KO mice, the authors have shown that myd88 expression is required for strengthening the CD4 CTLs phenotype during the infection.

   Moreover, by investigating the presence of a previously published CD4 CTLs gene signature in a mixed bone marrow chimera settings they highlighted a cell intrinsic role for Myd88 in imprinting the signature. The study also identifies Il18R as a myd88 upstream receptor potentially responsible for CD4 CTLs development by showing that lack of IL18R phenocopied myd88 deficiency in failing to promote a CD4 CTLs phenotype.

   Finally, by showing the direct correlation between perforin expressing CD4 T cells in Chagas infected individuals and parameters of heart disfunction the authors hinted at a possible involvement of CD4CTLs in a clinical setting.

   -The core finding of the paper, providing the first evidence of CD4 CTLs development in a mouse model of intracellular parasite is well supported by the data. The expression of markers correlated to CD4 cytotoxicity in other settings and gene signatures fits well the phenotype described and suggests possible common features for CD4 CTLs development across infection with different pathogens.

   This manuscript will boost the knowledge over the involvement of non canonical CD4 types in the immune responses to parasites. Moreover the finding that CD4 CTLs are the predominant phenotype in organs importants for viral replication imply an involvement of these cells in the development of the pathology that will have to be taken into accounts in future studies.

   • The understanding of the parental relationship beteween CD4CTLs and Th1 remains unclear and it's complicated by the low numbers of IFNg (regarded as an hallmark of functional Th1) producing CD4 T cells detected in the model. IFN-g production by CD4 is lower than 10% even when achieved by PMA/Iono stimulation and half of Gzb+ CD4 stain positive for the cytokine. On the other hand the putative transcription factor of Th1 development, Tbet, is expressed by all Gzb positive CD4s. This discrepancy and the low number of IFNG+ should be better discussed by the authors.

   First, we would like to thank Reviewer #3 for the constructive criticism on our manuscript. Regarding the apparent discrepancy on the frequencies of IFN-g+ and Tbet+ CD4+ T cells in our model, please first note that the percentage of IFN-g+ CD4+ T cells detected in the present study is comparable to the ones found in the spleen of mice infected with the same or similar strains of T. cruzi and reported by other groups (please see our complete answer to Reviewer #1 on this topic). With this remark done, we think that the apparent discrepancy between the expression of T-bet and the low fraction of GzB+CD4+ T cells producing IFN-g is a very interesting question. It is known that T-bet is a key transcription factor associated with the development of IFN-g-producing CD4+ T cells and that it also coordinates the expression of multiple other genes in CD4+ T cells and in other cell types. Also, T-bet can interact with other proteins, resulting in the induction or inhibition of key factors in T cell differentiation (reviewed in Hunter, 2019, Nat. Rev. Immunol, 19:398). Importantly, it has been shown that during the late stages of Th1 cell activation, T-bet recruits the transcriptional repressor Bcl-6 to the Ifng locus to limit IFNg transcription (Oestreich, 2011, JEM, 208:1001) Therefore, T-bet action is not limited to transactivation of the Ifng gene, but can also act as part of a negative-feedback loop to limit IFN-g production in certain cells. We do not believe that Bcl-6 is playing a role in CD4+GzB+ T cells in our model, since we found that the majority of CD4+GzB+ T lymphocytes express Blimp-1 (Figure 5D), and Blimp-1 and Bcl-6 are known to be reciprocally antagonistic transcription factors.

   However, the possibility remains that another repressor factor is downregulating Ifng gene transcription in the majority of T-bet+ CD4+GzB+ T cells, with the participation of T-bet or not. Of note, Blimp-1 was shown to be a critical regulator for CD4 T cell exhaustion during infection with T. gondii, and CD4+ T cells deficient in Blimp-1 produced higher levels of IFN-g in infected mixed-bone marrow chimeric mice reconstituted with WT and Blimp-1 conditional knock-out cells (Hwang, S., 2016, JEM 213:1799). Furthermore, Blimp-1 attenuates IFN-g production in CD4 T cells activated under nonpolarizing conditions and chromatin immunoprecipitation showed that Blimp-1 binds directly to a distal regulatory region in the Ifng gene (Cimmino, L. et al. 2008, JI 181:2338). We have also shown that, like Blimp-1, Eomes is expressed by around 60% of the GzB+CD4+ T cells (Figure 2G). It is known that Eomes controls the transcription of cytotoxic genes and promotes IFN-g production in CD8+ T cells, binding to the promotor of the Ifng gene. Interestingly, Eomes was also shown to participate in the induction of immunoregulatory/exhaustion receptors, such as PD-1 and Tim-3. Furthermore, deficiency of Eomes led to increased cytokine production (Paley, M.A. et al., 2012, Science 338: 1220). More recently, evidence in favor of the participation of Eomes in the repression of IFN-g production in TCR-gamma-delta T cells was also published (Lino, C. et al.,2017, EJI 47:970). Therefore, these studies indicate the complex control of Ifng gene, in which T-bet, Eomes, Blimp-1 and possible other TFs might play concerted roles. We think it would be interesting to investigate the role of Eomes and/or Blimp-1 in the repression of the Ifng gene in GzB+CD4+ T cells. Kinetics studies on the expression of these TFs, may contribute for the better understanding of the parental relationship between CD4CTLs and Th1 cells, a fundamental question, not completely understood yet. A comment on this subject was included in the revised manuscript.

   On the same note, while the confirmation of a CD4 CTLs gene signature in the model is very convincing, it must be noted that the one used as a reference was obtained by performing single cell RNA seq , taking into account only IFNg+ CD4 cells and then comparing Gzb+ and Gzb- negative in the setting. The authors are instead using bulk RNA seq and comparing populations of cells that would have none VS low levels of Th1. In this view, while the confirmation of the CD4 CTLs signature is striking, addressing the relative relationship with Th1 cells is complicated. Using Gzb YFP reporters in the setting could help improving the resolution between the 2 subsets.

   Our analysis clearly demonstrated the presence of the CD4CTL signature among WT CD4+ T cells, and its absence among Myd88-/- CD4+ T cells from the same mixed-BM chimeric mice. Together with our past work (Oliveira, 2017) and results included in the present manuscript, this analysis strongly contributes to demonstrate the importance of T-cell intrinsic IL-18R/MyD88 signaling for the development of a robust CD4CTL response to infection with an intracellular parasite. Although these results argue in favor of a common origin for CD4CTLs and Th1 cells during infection, an interesting point is that Ifng-/- mice display the same percentage of GzB+CD4+ T cells as WT mice (Figure 4B), suggesting that GzB+CD4+ T cells might emerge independently of IFN-gdependent Th1 cells. Therefore, the possibility remains that not all CD4CTLs are derived from the putative terminal differentiation of Th1 cells but that, instead, a divergence between the Th1 and CTL differentiation programs might occur at an earlier step. Although addressing this fundamental question goes beyond the possibilities of the present study, we believe that our results bring an important and substantial contribution for the understanding of the biology of CD4CTLs in response to infection and highlights the importance of IL-18R/MyD88 signaling for the reinforcement and/or stabilization of CD4+ T cell commitment into the CD4CTL phenotype. Regarding the use of GzB-YFP reporters, please see our answer below.

   • The dependancy on the Myd88/IL18r axis to promote CD4 CTLs is well characterized and the prolonged survival rate of IL18r-/- after the adoptive transfer of Gmb YFP+ CD4 is very convincing. However instead of using PBS as control the authors could have used YFP- or total CD4 cells for the task. While in previous publication it was already showed that protection was achieved by transferring the total CD4 population; comparing GzB + VS GzB- would have added useful insights over the amount of protection conferred by the subtypes and relative roles of CD4 CTLs and Th1 in the model. Parasitemia could also be reassessed in this view.

   We have already discussed the impossibility of sorting bona-fide GzB-negative cells from the reporter mouse strain available. Please see our complete answer to Reviewer 2 on this issue (iii) in this point-by-point letter. Moreover, due to the low percentage of GzB-EYFP cells labeled in the tamoxifen-treated reporter mice, a high number of mice is necessary for performing these adoptive transfer experiments. Unfortunately, due to the COVID-19 pandemic and its consequences on our animal facility, at present it is impossible to repeat this experiment including total CD4+T cells within a reasonable time. However, we have already shown in our past study (Oliveira, 2017), that the transfer of total WT CD4+T cells to Il18ra-/- mice, increased survival and lowered parasite load. On the other hand, our current data demonstrate that the adoptive transfer of GzB+CD4+ T cells increases survival but does not change the parasite load (Figure 6G). Therefore, these data strongly support that GzB+CD4+ T cells act in an IFN-g-independent way and, hence, differ from Th1 in the effector mechanism employed for extending survival of the recipient mice. In summary, our results favor the notion that CD4CTLs and Th1 cells have complementary roles, both being able to extend survival of recipient mice, although only Th1 are effective in lowering parasite load.

   Read the original source
2. 

   eLife

   Jun 14, 2022

   Evaluation Summary:

   This is an interesting study, conducted in mice, that demonstrates for the first time the presence of a large population of cytotoxic CD4+ T lymphocytes in infection with Trypanosoma cruzi, a relevant human pathogen. At present, the relevance of these cells in protective immunity engendered by the host remains unclear. Additional experiments are needed to characterize the functionality of these cytotoxic CD4 T cells vis-a-vis the canonical Th1 T cells. This paper can be of interest to scientists interested in immune responses to parasitic infections.

   (This preprint has been reviewed by eLife. We include the public reviews from the reviewers here; the authors also receive private feedback with suggested changes to the manuscript. Reviewer #3 agreed to share their name with the authors.)

   Read the original source
3. 

   eLife

   Jun 14, 2022

   Reviewer #1 (Public Review):

   The manuscript by Bellio and colleagues is based on the experimental model of T. cruzi infection in WT, MyD88-/- and IL-18-/- mice previously described by the same group in a 2017 eLife publication. The main message of the current study is that, in addition to IFN-g+ Th1 effectors, T. cruzi infection induces an even larger population of cytotoxic CD4+ T cells.

   The characterization of the cytotoxic CD4+ T cells is well documented. The data shown are convincing. However, since Burel et al. (2012) described the existence of a similar population in humans infected with P. falciparum (an intracellular pathogen), the authors should modify the statement (line 35-36) in the abstract. Similarly, the title "Cytotoxic CD4+ T cells... predominantly infiltrate Trypanosoma cruzi-infected hearts" is an overstatement. If …

   Reviewer #1 (Public Review):

   The manuscript by Bellio and colleagues is based on the experimental model of T. cruzi infection in WT, MyD88-/- and IL-18-/- mice previously described by the same group in a 2017 eLife publication. The main message of the current study is that, in addition to IFN-g+ Th1 effectors, T. cruzi infection induces an even larger population of cytotoxic CD4+ T cells.

   The characterization of the cytotoxic CD4+ T cells is well documented. The data shown are convincing. However, since Burel et al. (2012) described the existence of a similar population in humans infected with P. falciparum (an intracellular pathogen), the authors should modify the statement (line 35-36) in the abstract. Similarly, the title "Cytotoxic CD4+ T cells... predominantly infiltrate Trypanosoma cruzi-infected hearts" is an overstatement. If cytotoxic CD4+ T cells outnumber 10:1 IFN-g-secreting population (in lymphoid tissue) their higher representation in hearts of infected mice is not a selective phenomenon but rather expected.

   My major concern is that the function of these cells remains undefined. Are they beneficial or detrimental for the host? It appears that the authors themselves could not make up their minds. The GzB+ CD4+ T cells protect but do not decrease the parasite load (Fig 6G). Are they terminally differentiated or "exhausted" effectors? GzB+ CD4+ T cells can be found in the hearts of chronically infected mice, but we do not know if they are specific for pathogen or self Ags. Do they express the markers of exhaustion on day 14 in the heart?

   The factors that control differentiation of cytotoxic CD4+ T cells are the same as for IFN-g- Th1 cells. MyD-88-/- and IL-18-/- mice significantly lack both populations and succumb to T. cruzi infection. In their 2017 eLife publication, this group reported that survival of infected MyD-88-/- and IL-18-/- mice can be rescued by adoptive transfer of purified total WT CD4+ T cells, which was attributed entirely to their ability to secrete IFN-g (at least in the case of MyD-88-/- recipients). In the current study, the authors only used infected IL-18-/- recipients and show that this time transfer of GzB+ CD4+ T cells is sufficient to confer the protection. When compared with the old data, the rescue of the infected IL-18-/- with only GzB+ CD4+ T cells looks weaker (2 surviving animals out of 10 pooled from 2 experiments), strongly suggesting that IFN-g Th1 cells do play a significant role. It is unclear when the parasite load in Fig G6 was evaluated. It would be good to show deltaCT values for individual mice.

   Because donor IFN-g-/- CD4+ T cells do express IFN-gR (Supp Fig 6-2), IFN-g produced by IL-18-/- host cells could enhance the activity and/or help expand cytotoxic CD4+ T cells among the IFN-g-/- CD4+ donor population. To directly test the protective role of cytotoxic CD4+ T cells in the absence of IFN-g, the authors should treat infected IL-18-/- mice that have received IFN-g-/- CD4+ T cells with anti-IFN-gamma Ab.

   The intracellular cytokine staining in this study appears to be suboptimal. Instead of stimulating with PMA/ionomycin in the presence of Golgi block, Roffe et al. (2012) stimulated lymphocytes with anti-CD3 prior to adding Brefeldin A, an important technical difference which may explain the rather low frequencies of IFN-g+ and IL-10+ cells in this study.

   Read the original source
4. 

   eLife

   Jun 14, 2022

   Reviewer #2 (Public Review):

   In this work, Professor Bellio and her colleagues provide compelling evidence to show unusually strong induction of cytotoxic CD4 T cells (CD4CTLs) in Trypanosoma cruzi-parasitized mice. Using genetic models and mixed bone marrow chimeras they dissect the signals responsible for CD4CTL induction in this infection and identify T cell-intrinsic IL-18R/MyD88 signaling as the key inducer. The CD4CTLs that clonally expand in T. cruzi infection outnumber CD4 cells with typical Th1 profile (IFN-γ secretion) and bear the hallmarks of CD4CTLs described in other model systems and in humans. Utilizing GzmbCreERT2/ROSA26EYFP reporter mice, the authors show that adoptive transfer of CD4 cells that have made GzB can increase the survival of T. cruzi parasitized l18ra-/- mice. Finally, the authors describe a clear …

   Reviewer #2 (Public Review):

   In this work, Professor Bellio and her colleagues provide compelling evidence to show unusually strong induction of cytotoxic CD4 T cells (CD4CTLs) in Trypanosoma cruzi-parasitized mice. Using genetic models and mixed bone marrow chimeras they dissect the signals responsible for CD4CTL induction in this infection and identify T cell-intrinsic IL-18R/MyD88 signaling as the key inducer. The CD4CTLs that clonally expand in T. cruzi infection outnumber CD4 cells with typical Th1 profile (IFN-γ secretion) and bear the hallmarks of CD4CTLs described in other model systems and in humans. Utilizing GzmbCreERT2/ROSA26EYFP reporter mice, the authors show that adoptive transfer of CD4 cells that have made GzB can increase the survival of T. cruzi parasitized l18ra-/- mice. Finally, the authors describe a clear correlation between the frequency of CD4CTLs the circulation of patients with T. cruzi-induced chronic Chagas cardiomyopathy, implying a pathogenic role for these cells in chronic disease.

   The findings reported here are an important addition to the understanding of both the origin of CD4CTLs and their potential role in host protection or disease. The evidence provided in support of the main claims is very strong and the association between CD4CTLs and Chagas disease quite intriguing. There are, however, some aspects of the work that would benefit from further clarification or experimental support, so that alternative interpretations of the data can be excluded.

   The defining characteristic of CD4CTLs that separates them from other CD4 subsets is the production of granzymes and perforin and, by extension, the ability to kill target cells in a granzyme/perforin-dependent manner. In contrast, all T cells can kill target cells via alternative mechanisms that are not dependent on granzyme/perforin, for example through expression of TNF family members. It would appear that much, if not most, of the killing activity of T. cruzi-induced CD4CTLs can be attributed to FasL (Fig. 1B). FasL-mediated killing is not restricted to CD4CTLs and as the title of one of the cited studies (Kotov et al., 2018) states, "many Th cell subsets have Fas ligand-dependent cytotoxic potential". It would be important to ascertain if expression of granzyme/perforin by CD4CTLs in T. cruzi infection is also associated with granzyme/perforin-dependent cytotoxicity. This affects the direct and indirect in vitro cytotoxicity assays, as well as the interpretation of in vivo protection.

   Similarly, the protective effect of transferring GzmbCreERT2/ROSA26EYFP reporter-positive cells to Il18ra-/- mice may not be necessarily mediated in a granzyme/perforin-dependent manner or by CD4CTLs for that matter. The reporter will mark cells that express GzB at the time of tamoxifen administration but does not guarantee that these cells will continue to express GzB or that they will prolong survival of recipients in a granzyme/perforin-dependent manner.

   While the authors provide evidence that GzB-producing cells are largely distinct from IFN-γ-producing cells, the reporter-positive cells may still contain genuine Th1 cells. Given Th1 cells have been previously found necessary for protection of Il18ra-/- mice in the T. cruzi model, can a role for Th1 cells in this transfer model be formally excluded? The authors do convincingly demonstrate that IFN-γ itself is not essential for protection, but that does not leave granzyme/perforin-dependent as the only other alternative. For example, the experiment described in Fig. 6G lacks an important control, the transfer of reporter-negative cells. What would the conclusion be if reporter-negative (but T. cruzi-specific) cells proved as protective as reporter-positive cells?

   Read the original source
5. 

   eLife

   Jun 14, 2022

   Reviewer #3 (Public Review):

   By modelling trypanosoma cruzi infection in mice, the authors highlighted the presence of a subsets of CD4 T cells expressing canonical markers and transcription factors of CTLs and capable of exerting antigen specific and MHC class II restricted cytotoxic activity. Mechanistically, using KO mice, the authors have shown that myd88 expression is required for strengthening the CD4 CTLs phenotype during the infection.

   Moreover, by investigating the presence of a previously published CD4 CTLs gene signature in a mixed bone marrow chimera settings they highlighted a cell intrinsic role for Myd88 in imprinting the signature. The study also identifies Il18R as a myd88 upstream receptor potentially responsible for CD4 CTLs development by showing that lack of IL18R phenocopied myd88 deficiency in failing to promote a …

   Reviewer #3 (Public Review):

   By modelling trypanosoma cruzi infection in mice, the authors highlighted the presence of a subsets of CD4 T cells expressing canonical markers and transcription factors of CTLs and capable of exerting antigen specific and MHC class II restricted cytotoxic activity. Mechanistically, using KO mice, the authors have shown that myd88 expression is required for strengthening the CD4 CTLs phenotype during the infection.

   Moreover, by investigating the presence of a previously published CD4 CTLs gene signature in a mixed bone marrow chimera settings they highlighted a cell intrinsic role for Myd88 in imprinting the signature. The study also identifies Il18R as a myd88 upstream receptor potentially responsible for CD4 CTLs development by showing that lack of IL18R phenocopied myd88 deficiency in failing to promote a CD4 CTLs phenotype.

   Finally, by showing the direct correlation between perforin expressing CD4 T cells in Chagas infected individuals and parameters of heart disfunction the authors hinted at a possible involvement of CD4CTLs in a clinical setting.

   -The core finding of the paper, providing the first evidence of CD4 CTLs development in a mouse model of intracellular parasite is well supported by the data. The expression of markers correlated to CD4 cytotoxicity in other settings and gene signatures fits well the phenotype described and suggests possible common features for CD4 CTLs development across infection with different pathogens.

   This manuscript will boost the knowledge over the involvement of non canonical CD4 types in the immune responses to parasites. Moreover the finding that CD4 CTLs are the predominant phenotype in organs importants for viral replication imply an involvement of these cells in the development of the pathology that will have to be taken into accounts in future studies.

   - The understanding of the parental relationship beteween CD4CTLs and Th1 remains unclear and it's complicated by the low numbers of IFNg (regarded as an hallmark of functional Th1) producing CD4 T cells detected in the model. IFN-g production by CD4 is lower than 10% even when achieved by PMA/Iono stimulation and half of Gzb+ CD4 stain positive for the cytokine. On the other hand the putative transcription factor of Th1 development, Tbet, is expressed by all Gzb positive CD4s. This discrepancy and the low number of IFNG+ should be better discussed by the authors.

   On the same note, while the confirmation of a CD4 CTLs gene signature in the model is very convincing, it must be noted that the one used as a reference was obtained by performing single cell RNA seq , taking into account only IFNg+ CD4 cells and then comparing Gzb+ and Gzb- negative in the setting. The authors are instead using bulk RNA seq and comparing populations of cells that would have none VS low levels of Th1. In this view, while the confirmation of the CD4 CTLs signature is striking, addressing the relative relationship with Th1 cells is complicated. Using Gzb YFP reporters in the setting could help improving the resolution between the 2 subsets.

   - The dependancy on the Myd88/IL18r axis to promote CD4 CTLs is well characterized and the prolonged survival rate of IL18r-/- after the adoptive transfer of Gmb YFP+ CD4 is very convincing. However instead of using PBS as control the authors could have used YFP- or total CD4 cells for the task. While in previous publication it was already showed that protection was achieved by transferring the total CD4 population; comparing GzB + VS GzB- would have added useful insights over the amount of protection conferred by the subtypes and relative roles of CD4 CTLs and Th1 in the model. Parasitemia could also be reassessed in this view.

   Read the original source
6. 

   Version published to 10.7554/elife.74636 on eLife

   Jun 7, 2022
7. 

   Version published to 10.1101/2021.10.21.465260v2 on bioRxiv

   May 13, 2022
8. 

   Version published to 10.1101/2021.10.21.465260v1 on bioRxiv

   Oct 22, 2021