Requirement for Pathogenic IL-23 Signaling Is Restricted to Initiation of Autoimmune Myocarditis

Abstract

Using a mouse model of experimental autoimmune myocarditis (EAM), we showed for the first time that IL-23 stimulation of CD4⁺ T cells is required only briefly at the initiation of GM-CFS-dependent cardiac autoimmunity. IL-23 signal, acting as a switch, turns on pathogenicity of CD4+ T cells, and becomes dispensable once autoreactivity is established. Il23a^−/− mice failed to mount an efficient Th17 response to immunization, and were protected from myocarditis. However, remarkably, transient IL-23 stimulation ex vivo fully restored pathogenicity in otherwise nonpathogenic CD4⁺ T cells raised from Il23a^−/− donors. Thus, IL-23 may no longer be necessary to uphold inflammation in established autoimmune diseases. In addition, we demonstrated that IL-23 induced GM-CSF mediates the pathogenicity of CD4⁺ T cells in EAM. The neutralization of GM-CSF abrogated cardiac inflammation. However, sustained IL-23 signaling is required to maintain IL-17A production in CD4⁺ T cells. Despite inducing inflammation in Il23a^−/− recipients comparable to WT, autoreactive CD4⁺ T cells downregulated IL-17A production without persistent IL-23 signaling. This divergence on the controls of GM-CSF-dependent pathogenicity on one side and IL-17A production on the other side may contribute to the discrepant efficacies of anti-IL-23 therapy in different autoimmune diseases.

Introduction

Myocarditis is the inflammation of myocardium, the cardiac muscle [1]. In human, myocarditis is usually triggered by cardiotoxic chemicals or infections, such as parvovirus B19, Coxsackie B3 virus, Borrelia burgdorferi, and Trypanosoma cruzi [2]. Myocarditis patients may progress to inflammatory dilated cardiomyopathy (DCMi), during which scar tissue replaces damaged cardiac muscle, compromising cardiac function [3],[4]. The myocardium undergoes irreversible fibrosis and remodeling, making myocarditis and its sequela DCMi among the most common causes of non-congenital heart failure in young individuals [5].

Regardless of the etiology, evidence points to autoimmunity being involved in the pathogenesis of many myocarditis patients [1],[6]. The involvement of cardiac autoimmunity may lead to prolonged pathology and poor prognosis [7]. To investigate the immunopathologic mechanisms responsible for myocarditis in humans, we have developed a mouse model of experimental autoimmune myocarditis (EAM). In BALB/c mice, EAM is induced by immunization with a peptide derived from the cardiac myosin heavy chain alpha (MyHCα_614-629) in Complete Freund's adjuvant (CFA). Immunized mice develop myocarditis characterized by inflammatory infiltration, starting around day 10 and peaking around day 21. By day 45, mice progress to DCMi characterized by cardiac fibrosis and remodeling [8].

EAM is a CD4⁺ T helper cell-dependent disease, as depletion of CD4⁺ T cells protects mice from EAM [9]. Early studies have shown that neither Th1 nor Th2 cells are the exclusive driver of EAM disease. Mice with deficiencies in the Th1 program, such as Ifng^−/− and Tbx21^−/− mice, suffer from even more severe disease than wild-type (WT) counterparts. In addition, the severity of EAM in Il12a^−/− mice is comparable to WT controls [10]. These results demonstrate that Th1 polarization is not required for the development of EAM, and may play a protective role in its pathogenesis. As for the Th2 program, IL-4 was found to aggravate EAM in A/J but not BALB/c mice [11]. However, IL-13 was shown to be protective in BALB/c mice by skewing the phenotype of macrophages [12]. Thus, neither Th1 nor Th2 cells are the primary driver of EAM.

In the past decade, Th17 cells have been characterized and found to be critical in several autoimmune diseases. The Th17 program is initiated by transforming growth factor β (TGF-β) and interleukin 6 (IL-6), and is further promoted by interleukin 1β (IL-1β) and interleukin 23 (IL-23) [13]. This combination of stimuli leads to the expression of master transcription factor retinoic acid receptor-related orphan receptor γ (RORγt) in T cells encoded by gene Rorc [14]. RORγt drives transcription of multiple Th17 effector molecules that play critical roles in autoimmunity [15]–[17]. Mice with deficiency in the Rorc gene were resistant to several autoimmune diseases [18], demonstrating the pivotal role Th17 cells play in autoimmunity. Although Th17 cells were named after the hallmark cytokine IL-17A, they produce an array of other effector molecules including IL-17F, IL-22, and GM-CSF [16]. The pathogenic effects of Th17 cytokines other than IL-17A in autoimmunity have been demonstrated in mouse model of EAE [19]–[21].

IL-23, primarily secreted by antigen-presenting cells (APCs), promotes the maturation of Th17 cells [22]. IL-23 is a heterodimer of p19 and p40 subunits, encoded by the Il23a and Il12b genes, respectively [23]. The p19 subunit is unique to IL-23 and determines binding specificity of IL-23 to its receptor, IL-23R, while the p40 subunit is shared with IL-12. Upon activation by TGF-β and IL-6, CD4⁺ T cells upregulate IL-23R and respond to the IL-23 signal [24]. IL-23 induces activation of the STAT3 pathway, which stabilizes the Th17 program and enhances expression of major Th17 effector molecules [16]. Knockout of the IL-23 gene ameliorates experimental autoimmune encephalomyelitis (EAE) in mice [25], suggesting IL-23 is key in the pathogenicity of Th17 cells.

Our previous studies demonstrated that IL-17A, the hallmark cytokine of the Th17 program, was dispensable for initiation of cardiac inflammation and myocarditis [26],[27]. However, Th17 cells produce critical effector molecules other than IL-17A [13]. In this study, we employed Il23a^−/− mice to define the role that IL-23-induced Th17 polarization of CD4⁺ T cells plays in the pathogenesis of autoimmune myocarditis.

Results

Il23a^−/− Mice Are Resistant to EAM

Il23a^−/− mice, specifically deficient in the p19 subunit of IL-23 [28], and wild-type (WT) BALB/c mice were immunized with the MyHCα_614-629 peptide in CFA and sacrificed 21days post-immunization. Histopathologic study of mouse hearts by H&E staining showed that Il23a^−/− mice were fully protected from cardiac inflammation, whereas WT controls developed EAM (Figure 1A, 1B). Flow cytometric analysis (Figure S1A) revealed extensive infiltration of CD45⁺ leukocytes in the hearts of WT but not Il23a^−/− mice (Figure 1C, 1D). In addition, myocarditis in WT mice was accompanied by substantial expansion of myeloid effector cells, including Ly6G^hi neutrophils and CD11b⁺Ly6G⁻ monocytes (Figure S1B) in the spleen compared to Il23a^−/− mice (Figure 1E, 1F). In summary, Il23a^−/− mice were completely protected from cardiac inflammation, demonstrating that IL-23 is required for the induction of EAM.

Figure 1. Il23a^−/− Mice Are Protected from EAM.

EAM was induced in WT and Il23a^−/− mice by immunization. Mice were sacrificed 21 days post-immunization and myocarditis severity was established by cardiac histopathology. The compositions of heart-infiltrating cells and splenocytes were analyzed by flow cytometry (Figure S1A, S1B). Data are representative of 3 independent experiments.

(A) Representative cardiac histopathology of WT and Il23a^−/− mice. H&E staining, 4X and 20X. Bars represent 500 μm (4X) or 100 μm (20X).

(B) Cardiac histopathology score of WT and Il23a^−/− mice. Data points represent individual mice. Bars represent mean. Data are analyzed by Mann-Whitney U test. ***, p<0.001.

(C) Representative bivariate gating of cardiac CD45⁺ cells from all viable cells in WT and Il23a^−/− mice.

(D) Total cardiac CD45⁺ leukocytes in WT and Il23a^−/− mice.

(E) Proportion of Ly6G^hi neutrophils in viable CD45⁺ leukocytes in in the spleen of WT and Il23a^−/− mice.

(F) Proportion of Ly6G⁻CD11c⁻CD11b⁺F4/80⁻ monocytes in viable CD45⁺ leukocytes in the spleen of WT and Il23a^−/− mice.

(D ~ F) Data points represent individual mice. Bars represent mean. Data were analyzed by Student's t-test. ***, p<0.001; ****, p<0.0001; n.s., not significant.

Impaired Th17 Polarization in Il23a^−/− Mice during EAM

Since IL-23 has been shown to be required for the differentiation of Th17 cells, we investigated whether IL-23 deficiency had any effect on CD4⁺ T cell polarization during the induction phase of EAM. Il23a^−/− and WT BALB/c mice were immunized and sacrificed 14 days after immunization. Flow cytometric analysis (Figure S1A) showed that Il23a^−/− mice had significantly fewer heart-infiltrating CD45⁺ leukocytes and CD4⁺ T cells (Figure 2A, 2B). In addition, the percentages of IL-17A (Figure 2C, 2D), GM-CSF (Figure 2E, 2F), and in particular IL-17A/GM-CSF double producers (Figure 2G, 2H) in all cardiac CD4⁺ T cells were significantly lower in Il23a^−/− mice than WT controls. In contrast, the proportions of IFNγ-producing CD4⁺ T cells were comparable in Il23a^−/− and WT mice (Figure 2I). Similar patterns were observed in the spleens of Il23a^−/− and WT mice (data not shown). As a result, Il23a^−/− mice had significantly lower levels of IL-17A in the serum (Figure 2J). In summary, IL-23 deficiency impaired Th17 polarization of CD4⁺ T cells, and the protection from EAM in Il23a^−/− mice was strongly associated with diminished Th17 polarization. These results imply that IL-23-dependent polarization is needed for the induction of EAM.

Figure 2. Impaired Th17 Polarization in Il23a^−/− Mice during EAM.

EAM was induced in WT and Il23a^−/− mice by immunization. Mice were sacrificed 21 days post-immunization. The cytokine production profile of cardiac CD4⁺ T cells (Figure S1A) was analyzed by flow cytometry. Serum levels of IFNγ and IL-17A were measured by ELISA. Data are representative of 3 independent experiments.

(A) Total cardiac CD45⁺ leukocytes in WT and Il23a^−/− mice.

(B) Total cardiac CD4⁺ T cells in WT and Il23a^−/− mice.

(C) Representative histograms of IL-17A staining of viable cardiac CD4⁺ T cells.

(D) Proportion of IL-17A producers in cardiac CD4⁺ T cells in WT and Il23a^−/− mice.

(E) Representative histograms of GM-CSF staining of viable cardiac CD4⁺ T cells

(F) Proportion of GM-CSF producers in cardiac CD4⁺ T cells in WT and Il23a^−/− mice.

(G) Representative flow cytometry bivariate plot of IL-17A and GM-CSF single and double producers in cardiac CD4⁺ T cells. Numbers enumerate IL-17A/GM-CSF double producers.

(H) Proportion of IL-17A and GM-CSF double producers in cardiac CD4⁺ T cells in WT and Il23a^−/− mice.

(I) Proportion of IFNγ producers in cardiac CD4⁺ T cells in WT and Il23a^−/− mice.

(J) Levels of IL-17A in the serum of WT and Il23a^−/− mice.

(B, D, F, H ~ J) Data points represent individual mice. Bars represent mean. Data are analyzed by Student's t-test. *, p<0.05; **, p<0.01; ***, p<0.001; ****, p<0.0001; n.s., not significant.

IL-23 Signal Is Dispensable for Cardiac Inflammation Once Pathogenicity Is Established

Previous results suggested that the IL-23-dependent polarization was critical for the induction of pathogenic autoreactive CD4⁺ T cells required for EAM. To determine if IL-23 is needed to maintain the autoimmune response after pathogenic CD4⁺ T cells have been generated, we adoptively transferred CD4⁺ T cell to induce myocarditis. In this approach, splenocytes were collected from immunized Thy1.1⁺ WT donor mice 14 days after immunization, at the onset of cardiac inflammation. After culturing in Th17-polarizing media with recombinant IL-23 (rIL-23), recombinant IL-1β (rIL-1β), and sodium [29] ex vivo for 4 days, CD4⁺ T cells were magnetically isolated and transferred intravenously into sublethally irradiated Thy1.2⁺ WT or Il23a^−/− recipients. Heart inflammation was examined two weeks after the transfer (Figure 3A). Histopathology showed that even without IL-23 stimulation in the recipients, established autoreactive CD4⁺ T cells were able to induce heart inflammation in Il23a^−/− mice comparable to WT recipients (Figure 3B, 3C). Flow cytometric analysis (Figure S1A) showed that Il23a^−/− recipients had levels of CD45⁺ cell infiltration comparable to WT controls (Figure 3D), and the composition of infiltrating cells were similar in the two groups (Figure 3E-H). Thus, although IL-23 is essential in the induction of autoimmunity in EAM, it is no longer required once the pathogenic T helper cell population has been established.

Figure 3. IL-23-Elicited WT CD4⁺ T Cells Induce Myocarditis in Il23a^−/− Mice as Well as WT Controls.

WT Thy1.1 donor mice were immunized with MyHCα_614~629 peptide. Splenocytes were collected from donor mice 14 days post immunization and cultured ex vivo with 50μg/mL MyHCα_614~629, 20ng/mL rIL-23, 10ng/mL rIL-1β and 40mmol NaCl for 4 days. CD4⁺ T cells were isolated and transferred intravenously into WT or Il23a^−/− Thy1.2 recipients irradiated by 500 Rad γ-radiation. Myocarditis severity of WT and Il23a^−/− recipient mice was examined 2 weeks after the transfer by histopathology and flow cytometry (Figure S1A). Data are representative of 3 independent experiments.

(A) Schematic of experimental procedure.

(B) Representative cardiac histopathology of WT and Il23a^−/− recipient mice. H&E staining, 20X. Bars represent 50 μm.

(C) Cardiac histopathology score of WT and Il23a^−/− recipient mice. Data points represent individual mice. Bars represent mean. Data are analyzed by Mann-Whitney U test. n.s., not significant.

(D) Absolute count of viable cardiac CD45⁺ leukocytes in WT and Il23a^−/− mice that received transfer of WT CD4⁺ T cells.

(E) Proportion of Ly6G^hi neutrophils in total viable CD45⁺ leukocytes in the heart of WT and Il23a^−/− recipient mice.

(F) Proportion of Ly6G⁻CD11b⁺ MO/MΦs in total viable CD45⁺ leukocytes in the heart of WT and Il23a^−/− recipient mice.

(G) Proportion of Ly6C^hi population in Ly6G⁻CD11b⁺ MO/MΦs in the heart of WT and Il23a^−/− recipient mice.

(H) Proportion of CD4⁺ T cells in total viable CD45⁺ leukocytes in the heart of WT and Il23a^−/− recipient mice.

(D ~ H) Data points represent individual mice. Bars represent mean. Data are analyzed by Student's t-test. n.s., not significant.

Persistent IL-23 Signal Is Required to Maintain IL-17A Production

We next investigated the polarization of CD4⁺ T cells in the hearts of Thy1.2⁺ WT and Il23a^−/− recipients. The vast majority (more than 90%) of heart-infiltrating CD4⁺ T cells were of transferred Thy1.1⁺ donor origin (Figure 4A). Moreover, IL-17A or GM-CSF producing CD4⁺ T cells were almost exclusively of donor origin (Figure 4B-D). These results indicated that autoreactive Thy1.1⁺ donor CD4⁺ T cells used in the transfer drove the cardiac inflammation. Although transferred autoreactive CD4⁺ T cells were able to initiate cardiac inflammation of comparable severity in WT and Il23a^−/− recipients regardless of IL-23 signal, the cytokine production profiles differed between the two groups. Thy1.1⁺ donor CD4⁺ T cells in Il23a^−/− recipients produced significantly lower levels of IL-17A (Figure 4E, 4F) but higher levels of IFNγ (Figure 4E, 4G). This finding suggested that although IL-23 was not required for the augmentation of cardiac inflammation once the autoreactive CD4⁺ T cell population was established, it was crucial to maintain IL-17A production by the pathogenic CD4⁺ T cells and limit acquisition of IFNγ expression. These results further demonstrated that levels of IL-17A and IFNγ production did not affect the induction of EAM. Thus neither IL-17A nor IFNγ is the determining effector secreted by CD4⁺ T cells that directly initiates cardiac inflammation during EAM.

Figure 4. Continued IL-23 Stimulation Is Required to Maintain IL-17A Production.

WT Thy1.1 donor mice were immunized with MyHCα_614~629 peptide. Splenocytes were collected from donor mice 14 days post immunization and cultured ex vivo with 50μg/mL MyHCα_614~629, 20ng/mL rIL-23, 10ng/mL rIL-1β and 40mmol NaCl for 4 days. CD4⁺ T cells were isolated and transferred intravenously into WT or Il23a^−/− Thy1.2 recipients irradiated by 500 Rad γ-radiation. Origin and cytokine production profile of cardiac CD4⁺ T cells were examined 2 weeks after the transfer by flow cytometry (Figure S1A). Data are representative of 3 independent experiments.

(A) Proportion of Thy1.1⁺ donor cells in all CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(B) Proportion of Thy1.1⁺ donor cells in all IL-17A-producing CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(C) Proportion of Thy1.1⁺ donor cells in all GM-CSF-producing CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(D) Representative flow cytometry bivariate plot of IL-17A-producing CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(E) Representative flow cytometry bivariate plots of IL-17A and/or IFNγ - producing CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(F) Proportion of IL-17A producers in Thy1.1⁺ donor CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(G) Proportion of IFNγ producers in Thy1.1⁺ donor CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(H) Proportion of GM-CSF producers in Thy1.1⁺ donor CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(I) Proportion of RORγt⁺ population in Thy1.1⁺ donor CD4⁺ T cells in the heart of WT and Il23a^−/− recipient mice.

(A ~ C, F ~ I) Data points represent individual mice. Bars represent mean. Data are analyzed by Student's t-test. **, p<0.01; ****, p<0.0001; n.s., not significant.

Although Thy1.1⁺ donor CD4⁺ T cells in the hearts of WT and Il23a^−/− recipients differed in IL-17A and IFNγ production, the levels of GM-CSF producers were comparable in the two groups (Figure 4H). In addition, the levels of CD4⁺ T cells expressing the transcription factor RORγt were comparable in WT and Il23a^−/− recipients (Figure 4I). These results suggested that GM-CSF and RORγt, rather than IL-17A and IFNγ, may play important roles in determining the severity of cardiac inflammation during EAM.

Transient IL-23 Treatment Ex Vivo Restores Pathogenicity in CD4⁺ T Cells

Since Il23a^−/− mice are resistant to EAM, we next investigated whether recombinant IL-23 treatment ex vivo would be able to restore pathogenicity to CD4⁺ T cells elicited from Il23a^−/− mice. Splenocytes were collected from immunized Il23a^−/− donor mice, and cultured with or without IL-23 stimulation, before transfer into WT or Il23a^−/− recipients. Heart inflammation was examined two weeks after the transfer (Figure 5A).

Figure 5. Transient IL-23 Treatment Ex Vivo Restores Pathogenicity of CD4⁺ T Cells.

Il23a^−/− donor mice were immunized with MyHCα_614~629 peptide. Splenocytes were collected from Il23a^−/− donor mice 14 days post immunization and cultured ex vivo with 50μg/mL MyHCα_614~629, 10ng/mL rIL-1β and 40mmol NaCl for 4 days, with (+ rIL-23) or without (− rIL-23) rIL-23. CD4⁺ T cells were isolated and transferred intravenously into WT or Il23a^−/− recipients irradiated by 500 Rad γ-radiation. Cardiac inflammation in recipient mice was examined 2 weeks after the transfer by histopathology and flow cytometry (Figure S1A, S1B). Data are representative of 3 independent experiments.

(A) Schematic of experimental procedure.

(B) Representative cardiac histopathology of WT and Il23a^−/− mice that received transfer of rIL-23 stimulated (+ rIL-23) or control (− rIL-23) Il23a^−/− donor T cells. H&E staining, 20X. Bars represent 100 μm.

(C) Cardiac histopathology score of WT and Il23a^−/− recipient mice. Data points represent individual mice. Bars represent mean. Data are analyzed by Kruskal-Wallis test followed by Dunn's procedure. REC., recipients; CEL. cell treatment; *, p<0.05; **, p<0.01; n.s., not significant.

(D) Absolute count of viable cardiac CD45⁺ leukocytes in WT and Il23a^−/− recipient mice.

(E) Representative flow cytometry bivariate plots showing viable cardiac cells of WT and Il23a^−/− mice that received transfer of rIL-23 stimulated (+ rIL-23) or control (− rIL-23) Il23a^−/− donor T cells.

(F) Proportion of Ly6G^hi neutrophils in total viable CD45⁺ leukocytes in in the spleen of WT and Il23a^−/− recipient mice

(F) Proportion of Ly6G⁻CD11c⁻CD11b⁺F4/80⁻ monocytes in total viable CD45⁺ leukocytes in the spleen of WT and Il23a^−/− recipient mice.

(D, F ~ G) Data points represent individual mice. Bars represent mean. Data are analyzed by one-way ANOVA followed by Tukey's post-test. REC., recipients; CEL. cell treatment; ****, p<0.0001; n.s., not significant.

Histopathologic examination showed that rIL-23-stimulated CD4⁺ T cells were able to induce myocarditis when transferred into WT or even Il23a^−/− recipients (Figure 5B, 5C). This was reflected by the numbers of infiltrating CD45⁺ leukocytes in the heart (Figure 5D, 5E), and expansion of neutrophil and monocyte myeloid effector populations in the spleen (Figure 5F, 5G). In this experiment, CD4⁺ T cells transferred into Il23a^−/− mice did not receive any IL-23 signal before or after ex vivo culture. Thus, brief, transient IL-23 stimulation is sufficient to establish a pathogenic profile in CD4⁺ T cells. In other words, IL-23 signal acts as a switch of pathogenicity in CD4⁺ T cells. However, CD4⁺ T cells isolated from Il23a^−/− donors failed to induce severe cardiac inflammation without ex vivo IL-23 stimulation even in WT recipients (Figure 5B~G). These results implied that high intensity of activation, as in immunization with CFA and in ex vivo stimulation of splenocytes, was a prerequisite for IL-23 to take effect.

IL-23 Potentiates the Production of GM-CSF in CD4⁺ T Cells

As shown above, lower levels of GM-CSF and IL-17A production in CD4⁺ T cells were strongly associated with protection from myocarditis in Il23a^−/− mice. Since our previous studies established that IL-17A signaling was dispensable for the development of EAM [26],[27], we propose that GM-CSF serves as the IL-23-dependent pathogenic factor. While maintaining comparable GM-CSF production, WT CD4⁺ T cells were able to induce inflammation in Il23a^−/− recipients as well as in WT recipient, even though their IL-17A production was significantly downregulated (Figure 3C, 4F). This further indicates that GMCSF, not IL-17A, is the IL-23-dependent pathogenic factor during EAM.

To address this hypothesis, we collected splenocytes from immunized Il23a^−/− mice 14 days post immunization, and stimulated the cells with different cytokine cocktails. Flow cytometric analysis showed that rIL-23 in combination with rIL-1β and added sodium induced high levels of GM-CSF production from CD4⁺ T cells in vitro (Figure 6A), which correlated with their pathogenicity shown in the previous transfer experiment. Expression of the transcription factor RORγt, which drives GM-CSF production in IL-23-polarized CD4⁺ T cells, was also increased when CD4⁺ T cells were stimulated with rIL-23. (Figure 6B).

Figure 6. GM-CSF Is Required for the Induction of EAM.

(A ~ B) Il23a^−/− mice were immunized with MyHCα_614~629 peptide. Splenocytes were collected from Il23a^−/− mice 14 days post immunization and cultured in vitro with 50μg/mL MyHCα_614~629, 10ng/mL rIL-1β and 40mmol NaCl for 4 days, with or without rIL-23. GM-CSF production and RORγt expression in CD4⁺ T cells in vitro were analyzed by flow cytometry. Data are representative of 3 independent experiments.

(A) Proportion of GM-CSF producers in Il23a^−/− CD4⁺ T cells after rIL-23 stimulation.

(B) Percentage of Il23a^−/− CD4⁺ T cells expressing RORγt after rIL-23 stimulation.

(A ~ B) Data are shown as Mean+SD. Data are analyzed by Student's t-test. *, p<0.05; **, p<0.01.

(C ~ H) WT donor mice were immunized with MyHCα_614~629 peptide. Splenocytes were collected from WT donor mice 14 days post immunization and cultured ex vivo with 50μg/mL MyHCα_614~629, 20ng/mL rIL-23, 10ng/mL rIL-1β and 40mmol NaCl for 4 days. CD4⁺ T cells were isolated and transferred intravenously into WT recipients irradiated by 500 Rad γ-radiation. On and 7 days after T cell transfer, recipient mice receives 0.5mg anti-GM-CSF (αGM-CSF), same amount of rat IgG1 isotype antibody (Isotype), or same volume of vehicle (PBS). Myocarditis severity in recipient mice was examined 14 days after the transfer by histopathology and flow cytometry (Figure S1A, S1B). Data are representative of 2 independent experiments.

(C) Schematic of experimental procedure.

(D) Representative cardiac histopathology of WT recipient mice treated with anti-GM-CSF (αGM-CSF), rat IgG1 (Isotype), or vehicle (PBS). H&E staining, 20X. Bars represents 100 μm.

(E) EAM score of recipient mice. Data points represent individual mice. Bars represent mean. Data are analyzed by Kruskal-Wallis test followed by Dunn's procedure. *, p<0.05.

(F) Absolute count of viable cardiac CD45⁺ leukocytes in recipient mice.

(G) Proportion of Ly6G^hi neutrophils in viable CD45⁺ leukocytes in the spleen of recipient mice.

(H) Proportion of Ly6G⁻CD11c⁻CD11b⁺F4/80⁻ monocytes in viable CD45⁺ leukocytes in the spleen of recipient mice.

(F ~ H) Data points represent individual mice. Bars represent mean. Data are analyzed by one-way ANOVA followed by Tukey's post-test. *, p<0.05; **, p<0.01.

GM-CSF is required in the induction of EAM

In order to investigate whether GM-CSF is directly responsible for the initiation of inflammation in the heart, we studied the effects of GM-CSF neutralization during EAM. To exclude the possibility that GM-CSF affects antigen presentation and the initial polarization of CD4⁺ T cell in the lymph node, myocarditis was induced by the transfer of WT donor CD4⁺ T cells into sublethally irradiated WT recipients. Recipients received 0.5mg anti-GM-CSF antibody, an isotype control or PBS control, on day 0 (the day of transfer) and day 7 after transfer. Heart inflammation was examined two weeks after the transfer (Figure 6C). Histopathology showed that GM-CSF neutralization significantly reduced cardiac inflammation in recipient mice (Figure 6D, 6E). In addition, GM-CSF neutralization resulted in significantly reduced CD45⁺ inflammatory infiltration in the heart (Figure 6F) and diminished expansion of myeloid populations in the spleen (Figure 6G, 6H). These results strongly suggest that the IL-23-driven pathogenicity of autoreactive CD4⁺ T cells is dependent on GM-CSF.

Discussion

Th17 polarization of CD4⁺ T cells has been implicated in the pathogenesis of various autoimmune and autoimmune-related diseases [22],[30]–[33]. However, in the mouse model of EAM, our previous studies have established that IL-17A/IL-17RA signaling is not needed for the induction of cardiac inflammation [26],[27]. This seeming discrepancy reveals that the functions of IL-23-induced Th17 cells and their hallmark cytokine IL-17A are not synonymous. Using Il23a^−/− mice, we have provided definitive evidence that IL-23 is required for the induction of cardiac inflammation in EAM. This directly contrasts with the role of IL-12p35, which is dispensable for the induction of EAM [10],[34]. EAM has long been established as a CD4⁺ T cell-dependent disease [9]. In addition, CD4⁺ T cell transfer experiments showed that IL-23 was able to restore pathogenicity to otherwise non-pathogenic Il23a^−/− CD4⁺ T cells. These results demonstrated that the IL-23–dependent effector molecules of CD4⁺ T cells are directly responsible for the induction of autoimmunity in the heart.

There are recently two main hypotheses regarding the role of IL-23 in Th17 cells development. The first underscores that IL-23 is essential for cell commitment, and the second stresses that IL-23 is able to induce distinct combination of cytokines produced by T cells [33]. We show here evidence for both hypotheses playing a role in cardiac autoimmunity.

First, IL-23 signal acts as an on-switch for pathogenicity. Even transient IL-23 stimulation ex vivo was sufficient to fully restore pathogenicity to CD4⁺ T cells, as CD4⁺ T cells from Il23a^−/− donors, briefly conditioned by rIL-23 signaling, were functionally comparable to WT counterparts. For the ex vivo conditioning we used a protocol published by Kuchroo et al. using the new finding that the sodium through salt-sensitive kinase, SGK1, regulates IL-23R expression and stabilizes Th17 cells [29]. Furthermore, even without continued IL-23 stimulation upon transfer into Il23a^−/− recipients, autoreactive CD4⁺ T cells were still able to induce cardiac inflammation comparable to WT recipients, demonstrating that IL-23 signal is no longer required once pathogenicity is established.

Second, IL-23 signaling is essential for the persistent production of IL-17A in CD4⁺ T cells. Although WT or rIL-23-conditioned autoreactive CD4⁺ T cells induced comparable levels of cardiac inflammation in Il23a^−/− recipients compared with WT recipients, they produced significantly lower levels of IL-17A. This result suggested that, contrary to its action in inducing CD4⁺ T cell pathogenicity, continued IL-23 signaling is critical in maintaining IL-17A production. Thus, IL-23 has dual effect in EAM. Transient IL-23 stimulation of CD4⁺ T cells is sufficient to initiate cardiac autoimmunity; however, sustained IL-23 signaling in CD4⁺ T cells is required to maintain IL-17A production. This focus on the divergent controls of pathogenicity and IL-17A production by IL-23 has not been comprehensively interrogated across different autoimmune diseases, and we believe our study provides a paradigm in investigating this intricate topic.

Recent studies underscore the significant plasticity of CD4⁺ T cells. Using YFP reporter mice, it was demonstrated that most IFNγ–producing CD4⁺ T cells in the CNS during EAE were formerly IL-17A-producing cells [35]. Our results indicated that a similar process occurs in EAM. As autoreactive CD4⁺ T cells produced lower levels of IL-17A in Il23a^−/− recipients, they expressed higher levels of IFNγ, suggesting that in the absence of the maintenance of IL-23 signal, IL-17A producers converted to IFNγ producers. However, this reduction of IL-17A production had no significant effect on the general severity of cardiac inflammation, further supporting our previous finding that IL-17A/IL-17RA signaling is dispensable for initiation of cardiac autoimmunity.

Agents blocking IL-23 have been developed for the treatment of autoimmune and autoimmune related diseases. These agents have shown promising efficacy in the treatment of psoriasis and Crohn's disease [36],[37]. However, they did not reach expected efficacy in multiple sclerosis patients [38]. The dual effects of IL-23 on CD4⁺ T cells revealed in our study may help explain this phenomenon. We showed that continued IL-23 signal was required to support IL-17A production in CD4⁺ T cells. However, even transient IL-23 stimulation was sufficient to restore pathogenicity in otherwise non-pathogenic CD4⁺ T cells, and the lack of IL-23 signaling did not affect the initiation of EAM once autoreactivity was already established in CD4⁺ T cells. Considering that pathogenic CD4⁺ T cells in human patients most likely already had IL-23 stimulation before treatment can be initiated, anti-IL-23 therapy is more likely to be effective in diseases where IL-17A plays an essential role in driving adverse immunopathology.

GM-CSF has been implicated in playing critical roles in the initiation of autoimmune responses [20],[21],[39]. Our experiments employing the CD4⁺ T cell adoptive transfer model of EAM confirmed these findings. Although pathogenic CD4⁺ T cells produced lower levels of IL-17A in WT and Il23a^−/− recipients, they induced comparable levels of inflammation. This parity in disease severity correlated with the parity of GM-CSF production from CD4⁺ T cells, irrespective of the background of the recipients. Moreover, GM-CSF production in CD4⁺ T cells appears to be independent of their IL-17A or IFNγ production, as no statistical correlation was found among the several variants of our model. These results were further indications that GM-CSF is the key pathogenic factor secreted by autoreactive CD4⁺ T cells. We verified this hypothesis by blocking GM-CSF in the recipients of T cell adoptive transfer EAM. GM-CSF neutralization protected recipients from developing myocarditis, thus successfully abrogating the pathogenicity of transferred autoreactive CD4⁺ T cells, demonstrating that GM-CSF mediates the initiation of the autoimmune response in the heart. Studies in the EAE model have shown similar results [20].

Our experiments clearly illustrate that IL-23 promotes GM-CSF production in CD4⁺ T cells ex vivo. However, IL-23 signal is not required for the continued production of GM-CSF in vivo as autoreactive CD4⁺ T cells produced similar levels of GM-CSF following transfer into Il23a^−/− recipients. This finding was reflected by studies in EAE [20],[21]. Our previous study demonstrated that, during late stages of the development of EAM and inflammatory dilated cardiomyopathy, cardiac fibroblasts respond to IL-17A stimulation to secret GM-CSF, shaping the polarization of myeloid lineage effector cells [26]. Our present data add to this model that GM-CSF is decisive in both the initiation of autoimmunity and the progression of its effector mechanisms. Therefore, targeting GM-CSF may prove to be a promising approach in the treatment of myocarditis and other autoimmune diseases.

Materials and Methods

Mice

Il23a^−/− founder mice were provided by Pfizer Inc. and backcrossed to the BALB/cJ background for 10 generations. C.By.PL(B6)-Thy1a/ScrJ (Thy1.1⁺) and WT BALB/cJ mice were acquired from the Jackson Laboratory (Bar Harbor, Maine, U.S.A.). Mice were housed in the Johns Hopkins University School of Medicine in specific-pathogen free conditions. We used 6-10 weeks old male mice for all experiments. All experiments were done in compliance with the Animal Welfare Act and approved by the Animal Care and Use Committee of The Johns Hopkins University.

Induction of EAM

The myocarditogenic peptide of cardiac myosin heavy chain, MyHCα_614-629 (Ac-SLKLMATLFSTYASAD), was produced by fMOC chemistry and purified by HPLC to no lower than 90% purity (Genscript). Mice, on day 0 and day 7, received subcutaneous axillary immunizations of 100 μg of MyHCα_614-629 emulsified in complete Freund's adjuvant (CFA) (Sigma) supplemented to 5 mg/mL with heat-killed Mycobacterium tuberculosis strain H37Ra (Difco). Mice were also injected with 500 ng of pertussis toxin ip on day 0 (List Biologicals).

Induction of EAM by Adoptive Transfer of CD4⁺ T Cells

Donor mice were immunized with MyHCα_614~629 peptide as above. Splenocytes were collected from donor mice 14 days post-immunization and cultured ex vivo in Dulbecco's modified Eagle's medium (DMEM) with 4.5g/L glucose, 2mM L-Glutamine, 1mM sodium pyruvate, 25mM HEPES, 100U/mL penicillin G, 100μg/mL streptomycin, 55μM 2-mercaptoethanol and 10% fetal bovine serum (FBS) supplemented with 25μg/mL MyHCα_614~629, 20ng/mL rIL-23, 10ng/mL rIL-1β, and 40mM NaCl for 4 days. CD4⁺ T cells were isolated and transferred iv into recipients irradiated by 500 cGy γ-radiation. Heart inflammation was examined 2 weeks following transfer.

Assessment of EAM Severity by Histopathology

Hearts from mice with EAM were fixed in SafeFix (Fisher Scientific). 5μm serial sections were cut from longitudinally embedded tissue and stained with hematoxylin and eosin (H&E) (HistoServ, Gaithersburg, MD). EAM severity was evaluated as follows: grade 0, no inflammation; grade 1, less than 10% of the heart section is involved; grade 2, 10-25%; grade 3, 25-50%; grade 4, 50-75%; grade 5, more than 75% [40]. Scoring was performed by evaluating five sections per heart by three independent investigators in a blinded manner and scores were averaged.

Flow Cytometry Analysis

Mouse hearts were perfused for 3 min with 1x PBS + 0.5% FBS, and processed in GentleMACS C Tubes with 5mL digestion buffer (10,000 units/mL collagenase II (Worthington) and 1,000 units/mL DNase I (Worthington) in HBSS) according to manufacturer's instructions (Miltenyi Biotec). The resulting single cell suspension was stained with LIVE/DEAD Aqua according to manufacturer's instructions (Life Techonologies). Cells were then blocked with anti-CD16/32 (eBiosciences), and surface markers were stained with fluorochrome-conjugated mAbs (eBioscience, BD Pharmingen, BioLegend). For intracellular cytokine staining, cells were fixed and permeablized with Fix/Perm buffer (BD Pharmingen). For transcription factor staining, cells were fixed and permeablized with Transcription Factor Buffer Set (BD Pharmingen). Samples were acquired on the LSR II cytometer running FACSDiva 6 (BD Immunocytometry). Data were analyzed with FlowJo 7.6 (Treestar Software). Gating strategy is shown in Supplementary Figure 1.

ELISA

Quantitative sandwich ELISA for mouse IL-17A was performed on mouse serum samples with colorimetric ELISA kits according to manufacturer's protocol (R&D Systems, cat # M1700).

GM-CSF Blockade

Adoptive transfer EAM was elicited as described above. On days 0 and 7 post-immunization, 0.5mg anti-mouse GM-CSF antibody M250 (Amgen) or rat IgG1 isotype control clone HRPN (Bio × Cell) were injected ip. Heart inflammation was examined 2 weeks following transfer.

Statistics

Normally distributed data were analyzed by two-tailed Student's t-test or one-way ANOVA followed by Tukey's post-test for multiple group comparisons. EAM severity scores were analyzed by Mann-Whitney U test (up to two groups) or Kruskal-Wallis test followed by Dunn's procedure. Values of p < 0.05 were considered statistically significant.

List of Abbreviations Used

EAM

Experimental autoimmune myocarditis

DCMi

inflammatory dilated cardiomyopathy

MyHCα_614-629

cardiac myosin heavy chain alpha 614~629

RORγt

retinoic acid receptor-related orphan receptor γt