Self-reactive CD4⁺ IL-3⁺ T cells amplify autoimmune inflammation in myocarditis by inciting monocyte chemotaxis

Myocarditis is an inflammatory heart disease and a primary cause of sudden death and dilated cardiomyopathy in children and young adults (Cooper, 2009; Sagar et al., 2012b; Trachtenberg and Hare, 2017; Swirski and Nahrendorf, 2018). The disease is often triggered by cardiotropic infections and is subsequently enhanced by an autoimmune response, in which inappropriately activated T cells specific to cardiac antigens damage the myocardium and propagate heart failure. A recent clinical study showing that patients with fulminant myocarditis have a more adverse outcome compared to those without the drastic inflammatory phenotype also suggests that augmented cellular immunity confers irreversible tissue injury, which causes fatal arrhythmia and cardiac dysfunction (Ammirati et al., 2017). Thus, investigating the immunopathogenic mechanisms involved in the development of autoimmune cardiac inflammation might identify new therapeutic interventions for patients with myocarditis.

An enduring concept in immunology emphasizes two discrete phases linked to the development of T cell–mediated autoimmune inflammation (Marrack et al., 2001; Rosenblum et al., 2015). During the sensitization phase, which typically occurs in the LNs, naive T cells specific for a self-antigen encounter APCs and, consequently, mature and expand. During the later effector phase, these activated T lymphocytes, along with other lymphoid and myeloid cell subsets, accumulate in their cognate antigen-expressing tissues. When the inciting antigen is a cardiomyocyte-derived peptide from a protein like the α myosin heavy chain (αMHC), inflammatory cells invade the heart, leading to myocarditis. While the first phase generates a lymphocyte pool with pathogenic potential, it is during this second phase that inflammation and irreversible organ damage actually occur.

The cellular and molecular sequelae that promote the generation of autoreactive T cells have been studied extensively in various autoimmune disease models (Goodnow et al., 2005; Kronenberg and Rudensky, 2005; Mohan and Unanue, 2012; Bluestone et al., 2015; Theofilopoulos et al., 2017). Less is known, however, about the cytokine networks that control mobilization, recruitment, and function of inflammatory cells during the effector phase. The multifunctional cytokine IL-3 promotes hematopoiesis, leukocyte differentiation, and survival (Williams et al., 1990; Weber et al., 2015). Although the discovery of IL-3 dates back several decades (Hapel et al., 1981; Ihle et al., 1981; Yang et al., 1986), relatively little is known about the scope of IL-3’s action in the context of autoimmunity, with a handful of studies on its role in type I diabetes (Ito et al., 1997), lupus nephritis (Renner et al., 2015), collagen-induced arthritis (CIA; Brühl et al., 2009; Srivastava et al., 2011), and experimental autoimmune encephalomyelitis (EAE; Hofstetter et al., 2005; Renner et al., 2016), but no studies have linked IL-3 with myocarditis. Moreover, studies that have connected IL-3 to autoimmune diseases have yielded potentially conflicting results and were not able to determine the cytokine’s mechanism of action. This study shows that IL-3 is dispensable to T cell sensitization but essential to the effector phase, during which it orchestrates a tricellular amplification loop consisting of self-reactive T cells, macrophages, and monocyte-derived dendritic cells (moDCs).

To test whether IL-3 plays a role in autoimmune inflammation observed in myocarditis, we adopted a mouse model of EAM in which myocardial inflammation is induced by immunizing genetically susceptible mice with αMHC peptide emulsified in CFA using two s.c. injections 7 d apart (Cihakova and Rose, 2008; Fig. 1 A). At 21 d after the first injection, the hearts of WT female mice developed considerable inflammation comprising intense leukocyte infiltration (Fig. 1 B), a high inflammatory score (Fig. 1 C), and elevated cardiac serum troponin I (Fig. 1 D). Il3^−/− female mice, however, were profoundly protected from EAM, developing comparably minimal cardiac inflammation. Male Il3^−/− mice were likewise protected from cardiac inflammation, indicating that sex was not a contributing factor to IL-3–dependent inflammation (Fairweather et al., 2013; Fig. S1 A). These dramatic differences prompted us to enumerate leukocyte recruitment to the heart at 10 d (beginning of the inflammation) and 21 d (peak of the inflammation) after the first injection. We adapted a flow cytometric procedure (Anzai et al., 2017) to identify and quantify Ly-6C^high monocytes/macrophages, moDCs, macrophages, plasmacytoid DCs (pDCs), classical DC1 (cDC1), cDC2, neutrophils, T cells, B cells, eosinophils, and other leukocytes (Fig. S1, B and C). We also confirmed monocytes’ bipotential capacity to differentiate into both F4/80⁺ CCR2⁺ MHCII⁺ macrophages and CD172a (SIRPα)⁺ XCR1⁻ CCR2⁺ moDCs in the inflamed heart (Epelman et al., 2014; Clemente-Casares et al., 2017; Van der Borght et al., 2017; Fig. S1, D–F). We show that compared with WT mice, Il3^−/− mouse hearts accumulated fewer leukocytes (Fig. 1 E), with notable decreases on day 21 among Ly-6C^high monocytes/macrophages, moDCs, MHCII⁺ macrophages, pDCs, neutrophils, and T cells (Fig. 1 F), including both CD4⁺ and CD8⁺ cells (Fig. 1 G). Realizing that myocardial inflammation can progress to inflammatory dilated cardiomyopathy, we then evaluated cardiac fibrosis and heart function. Azan staining of mouse hearts 45 d after disease induction revealed minimal development of cardiac fibrosis among Il3^−/− mice in comparison to WT controls (Fig. 1 H). Moreover, echocardiography indicated that Il3^−/− mice developed less myocardial dysfunction than WT animals (Fig. 1 I). Together, these data show that IL-3 contributes decisively to acute inflammation and chronic heart failure during autoimmune myocarditis.

Having established the importance of IL-3 in orchestrating myocarditis, we next sought to identify IL-3’s source. We measured Il3 by quantitative PCR (qPCR) in tissue sections collected at various time points after the first injection of αMHC/CFA. Il3 production was negligible to low in the steady state but increased substantially (>20-fold) in the heart on day 21, with only smaller increases in the draining LNs but not in other locations such as the bone marrow (BM), spleen, thymus, and lung (Fig. 2 A). Flow cytometry of cardiac single-cell suspensions on day 21 revealed CD3⁺ CD4⁺ T cells to be major sources of intracellular IL-3 (Fig. 2 B). Although ∼20% of the IL-3–producing CD4⁺ T cells were either IFN-γ⁺ or IL-17A⁺ and ∼4% were IFN-γ⁺ IL-17A⁺, most IL-3⁺ CD4⁺ T cells did not produce either cytokine (Fig. 2 C). In addition, none of the IL-3⁺ CD4⁺ T cells produced IL-4 (Fig. 2 C). Thus, while some IL-3–producing CD4⁺ T cells appear to be representative of the proinflammatory T helper (Th) 1 and Th17 cell lineages, which commonly associate with autoimmune inflammation (Dardalhon et al., 2008), the majority of IL-3⁺ CD4⁺ T cells do not actively secrete IFN-γ or IL-17A. Isolating T cells from sensitized animals and culturing them with BM-derived DCs (BMDCs) along with either αMHC or myelin oligodendrocyte glycoprotein, an antigen targeted in models of multiple sclerosis, confirmed that T cells sensitized to αMHC in vivo can secrete IL-3 protein in an antigen- and disease-specific manner upon recognizing their MHCII-restricted cognate peptide (i.e., αMHC; Fig. 2 D).

To determine the importance of IL-3–producing CD4⁺ T cells to establishing myocardial inflammation, we pursued a two-pronged strategy. First, we isolated CD4⁺ T cells from sensitized WT and Il3^−/− mice and adoptively transferred those cells to SCID mice lacking functional lymphocytes (Fig. 2 E). 10 d after transfer, we noted that hearts of SCID mice receiving WT CD4⁺ T cells accumulated more leukocytes than hearts of SCID mice receiving Il3^−/− CD4⁺ T cells (Fig. 2 F). Second, we generated mixed-chimeric Rag1^−/−Il3^−/− mice that specifically lacked IL-3 production by lymphocytes (Fig. 2 G). Compared with Rag1^−/− WT controls, hearts of mixed chimeras with Il3^−/− CD4⁺ T cells developed significantly less inflammation, as characterized by fewer Ly-6C^high monocytes/macrophages, moDCs, and T cells, 21 d after EAM induction (Fig. 2 H). Together, these data indicate that CD4⁺ T cells critically influence the inflammatory phenotype that occurs during autoimmune myocarditis.

Having determined that CD4⁺ T cell–derived IL-3 promotes inflammation during myocarditis, we next addressed the mechanism. We first tested IL-3’s role in sensitization, as this is the defining process during which APCs elicit expansion of antigen-specific T cells in LNs. Following a BrdU pulse, we observed that T cells in the draining LNs proliferated at higher rates in vivo 10 d after myocarditis induction, but their proliferative rate did not depend on IL-3 (Fig. 3 A). Likewise, CD4⁺ T cell proliferation in vitro, involving culture of T cells from sensitized animals with BMDCs, αMHC, and recombinant IL-3, indicated that proliferation required antigen, but not IL-3 (Fig. 3 B). Consequently, we failed to detect any IL-3–dependent differences in the number of CD4⁺ T cells in the draining LNs after myocarditis induction (Fig. 3 C). Furthermore, flow cytometric analyses showed no differences in the number of DC subsets isolated from the draining LNs, such as CD11c^high MHCII^int (resident cDCs), CD11c⁺ MHCII^high (migratory cDCs), and CD11b^high CD64⁺ CD11c⁺ MHCII⁺ (moDCs; Fig. 3, D and E). We also sought to determine whether IL-3 can alter DC APC function during sensitization, but we detected no differences in either DC expression of CD80, OX40L, or MHCII (Fig. S2 A), or T cell production of IFN-γ, IL-17A (Fig. 3, F and G), or other cytokines (Fig. 3 H and Fig. S2, B–D). The finding that adoptively transferring WT in vitro–generated αMHC-presenting BMDCs (Eriksson et al., 2003) to Il3^−/− mice did not give rise to fulminant inflammation (Fig. 3, I and J) further substantiated the idea that attenuated inflammation in Il3^−/− mice occurred downstream of DC sensitization capacity. Thus, T cell sensitization in autoimmune myocarditis does not require IL-3.

Our data indicate that IL-3 is critical to leukocyte accumulation in the heart during myocarditis, but is dispensable to sensitization. Given IL-3’s known role in leukocyte production (Weber et al., 2015), we enumerated leukocytes in the BM, blood, and spleen on days 0, 14, and 21 and found no differences between WT and Il3^−/− mice (Fig. S2, E and F), suggesting that the leukocyte differences in the heart did not depend on altered hematopoiesis in the BM. We therefore turned our attention to the heart. The first evidence supporting a local function for IL-3 was the rate of local CD4⁺ T cell proliferation in the hearts of WT and Il3^−/− mice: on day 21 after EAM induction, when WT hearts were producing abundant IL-3 (Fig. 2 A), we observed a marked reduction of local CD4⁺ T cell proliferation in Il3^−/− mice (Fig. 4, A and B), which is in line with reduced CD4⁺ T cell numbers on day 21 after EAM (Fig. 1 G). This finding contrasted with day 10, an early time point with minimal IL-3 production, on which we found similar CD4⁺ T cell numbers (Fig. 1 G) and less proliferation, with no differences between the groups (Fig. 4 B). Notably, the 2-h pulse-chase window can assess local proliferation, because no BrdU⁺ cells yet appear in the blood (Fig. 4 B). Local proliferation rather than increased survival appeared to drive cell expansion, as we did not detect any differences in T cell death, as assessed by caspase-3 (Fig. 4, C and D). Consequently, hearts of Il3^−/− mice accumulated fewer CD4⁺ T cells, corresponding to overall fewer IFN-γ⁺, IL-17⁺, and GM-CSF⁺ cells (Fig. 4 E), and thus diminished inflammation.

To explore what accounted for local T cell proliferation, we first employed ex vivo approaches and found that CD4⁺ T cells proliferated robustly when cultured with antigen-loaded DCs sorted from myocarditis hearts, that absence of either DC or antigen reduced proliferation, and that recombinant IL-3 supplementation had no effect (Fig. 4 F). Indeed, T cells, which express negligible levels of CD123, the IL-3–specific receptor subunit α (IL-3Rα; http://www.immgen.org), did not use IL-3, as IL-3–deficient CD4⁺ T cells cultured with antigen-loaded DCs proliferated just as well as WT CD4⁺ T cells (Fig. 4 F). We therefore wondered whether local T cell proliferation depended on the number of DCs. Among the DC subsets accumulating in the heart, moDCs were particularly informative, as they were the most abundant in the inflamed heart and their numbers differed between WT and Il3^−/− mice (in contrast, cDC1 and cDC2 numbers between WT and Il3^−/− inflamed hearts did not differ, and cardiac inflammation did not require pDC, as determined by pDC depletion; Fig. 1 F and Fig. S3, A and B). Importantly, we found that both the extent of T cell proliferation (Fig. 4 G) and IL-3 supernatant levels (Fig. 4 H) correlated with moDC numbers, as proliferation and IL-3 production both rose with increased moDC frequency. We also found a strong positive in vivo correlation between CD4⁺ T cell proliferation in the heart and the number of cardiac moDC (Fig. 4 I), indicating that moDC are likely the primary drivers of local T cell proliferation. The noticeable decrease among MHCII⁺ macrophages, which were abundant in the inflamed hearts of WT mice and highly reduced in Il3^−/− mice (Fig. 1 F), suggested that MHCII⁺ macrophages could also contribute to local T cell proliferation within the inflamed cardiac tissue. Although sorted cardiac MHCII⁺ macrophages loaded with αMHC appeared to be less effective in stimulating autoreactive CD4⁺ T cell proliferation compared with sorted moDCs (Fig. 4 F), we noted a strong positive correlation between CD4⁺ T cell proliferation and MHCII⁺ macrophages in vivo (Fig. 4 I). Together, these data reveal that (i) in myocarditis hearts, CD4⁺ T cells produce IL-3; (ii) moDC and MHCII⁺ macrophage accumulation in the inflamed myocardium depends on IL-3; and (iii) T cells proliferate in response to accumulating antigen-loaded monocyte-derived APCs, but not directly in response to IL-3.

If moDC and macrophage accumulation in the heart depends on IL-3 (Fig. 1 F), then how does this occur? We first sought to identify the cells that can directly respond to IL-3. Analysis of leukocytes (Fig. 5 A) and non-leukocytes (endothelial cells, fibroblasts, and other stromal cells; Fig. S3 C) in the inflamed heart revealed the predominant IL-3R (CD123 and CD131)–expressing cells to be CD11b^high CD64^high MHCII⁺ CD11c⁻, a profile that is consistent with cardiac macrophages, while non-leukocytes showed negligible numbers of cells expressing either subunit. Indeed, MHCII⁺ macrophages, DCs, and Ly-6C^high monocytes/macrophages, but not T cells or non-leukocytes, phosphorylated STAT5 in response to IL-3, indicating direct signaling downstream of the growth factor (Fig. S3 D). Although basophils prominently expressed IL-3R in the periphery, their accumulation into cardiac tissue was negligible at the peak of EAM (0.10 ± 0.01% in WT vs. 0.07 ± 0.03% in Il3^−/− among CD45⁺ leukocytes; Fig. S3, E and F). Upon cell sorting, we noted that, among leukocytes, cardiac macrophages were by far the most abundant producers of the monocyte-attracting chemokines Ccl2, Ccl7, and Ccl12 (Fig. 5 B).

That said, we found that fibroblasts also produced monocyte chemoattractants in the inflamed myocardium, including Ccl7 and Ccl2 (Fig. S4 A), which agrees with previous studies showing that fibroblasts participate importantly in myocarditis (Lindner et al., 2014; Wu et al., 2014; Amoah et al., 2015). We therefore compared the relative abundance of cardiac macrophages and fibroblasts in naive and EAM mice and measured expression of Ccl2, Ccl7, and Ccl12 in equal numbers of sorted cardiac MHCII⁺ macrophages and fibroblasts isolated from day 21 hearts (Fig. S4 B). We found that while cardiac fibroblasts were more numerous than cardiac macrophages in naive hearts, cardiac macrophages outnumbered fibroblasts in inflamed hearts. In addition, we found that cardiac MHCII⁺ macrophages expressed more Ccl2 and Ccl12 than fibroblasts on a per-cell level (Fig. S4 B). Thus, fibroblasts may be important IL-3–independent sources of monocyte chemoattractants, but macrophage-derived chemokines outnumber fibroblasts during myocarditis.

Next, we investigated IL-3–dependent myeloid expansion in more detail. Compared with WT mice, we observed no changes in myeloid cell proliferation, death, or chemokine receptor expression in Il3^−/− mice, suggesting that IL-3–dependent myeloid expansion in the heart relied more on recruitment than on survival or local self-renewal (Fig. S4, C–E). This prompted us to ask whether IL-3 was stimulating IL-3R⁺ macrophages to produce monocyte-attracting chemokines. We elected to test for a direct IL-3–chemokine link with several approaches. First, we screened a macrophage cell line, J774, for IL-3R expression (Fig. S5 A) and cytokine production in response to IL-3 (Fig. S5 B), noting high induction of Ccl2, Ccl7, and Ccl12. Second, we sorted peritoneal macrophages, BM-derived macrophages, and cardiac MHCII⁺ macrophages from naive mice, stimulated the cells with IL-3, and again noted high increases of Ccl2, Ccl7, and Ccl12 gene expression (Fig. S5 C). Importantly, we did not observe any changes in the expression of Ccl2, Ccl7, and Ccl12 after stimulating sorted naive cardiac fibroblasts with IL-3 in vitro, which bolstered the idea that fibroblasts’ production of monocyte chemoattractants is independent of IL-3 in EAM (Fig. S5 C). Third, we generated IL-3Rα–deficient cells using the CRISPR-Cas9 technique (Heckl et al., 2014) to answer the specific question of whether IL-3 stimulated cardiac macrophages in vivo. We designed a small guide RNA (sgRNA) targeting the IL-3–specific mouse IL-3Rα and integrated the RNA guide into a lentiviral vector that contained Cas9 and enhanced GFP (eGFP; Fig. 5 C). We validated the efficacy of the lentivirus/CRISPR approach using J774 cells in vitro, finding suppressed phosphorylation of STAT5 (Fig. S5, D and E) and reduced expression of Ccl2, Ccl7, and Ccl12 in GFP⁺ cells, even after IL-3 stimulation (Fig. S5 F). We then transfected BM Lin⁻ cells with the vector, transplanted the cells into lethally irradiated WT mice, and induced myocarditis after BM reconstitution (Fig. 5 D). Though the transfection was only ∼15% efficient, we did find a population of GFP⁺ Ly-6C^high monocytes in the blood (Fig. 5 E) that, unlike their GFP^– counterparts, failed to phosphorylate STAT5 upon IL-3 stimulation (Fig. 5 F), thereby confirming that GFP⁺ cells lacked a functional IL-3Rα. Finally, upon sorting, Il3ra^−/− (GFP⁺) cardiac MHCII^high macrophages (Fig. 5 G) had reduced gene expression of the three key monocyte-attracting chemokines in the inflamed heart in comparison to Il3ra^+/+ (GFP⁻) cardiac MHCII^high macrophages (Fig. 5 H). This observation corresponded with considerably less overall Ccl2, Ccl7, and Ccl12 production in unsorted heart tissue of Il3^−/− mice 21 d after EAM (Fig. 5 I). Thus, T cell–derived IL-3 locally acts on cardiac MHCII^high macrophages to produce monocyte-attracting chemokines during the course of myocarditis.

Interestingly, despite the reduced accumulation of neutrophils and T cells in the inflamed myocardium at the peak of EAM (Fig. 1, F and G), neutrophil (Cxcl1 and Cxcl2) and T cell (Ccl19 and Ccl20) chemoattractant expression was unchanged in our in vitro screening experiments involving IL-3 stimulation (Fig. S5 B). However, expression of Cxcl1, Cxcl2, and Ccl19 was significantly decreased on day 21 in Il3^−/− hearts (Fig. S5, G and H), suggesting that additional, yet-unidentified IL-3–dependent mechanisms indirectly promote neutrophil and T cell recruitment to the inflamed myocardium.

Given IL-3’s role in stimulating inflammation, we next wondered whether the cytokine can be targeted therapeutically to ameliorate myocarditis. Observations on specimens obtained from human hearts highlighted this question’s potential clinical importance. We detected human IL3 in two specimens collected from patients with myocarditis, but not from two commercially acquired samples of healthy human hearts (4.718 × 10⁻⁴ ± 1.708 × 10⁻⁴ vs. undetectable by qPCR). Intraperitoneally injecting anti-IL-3 antibody to presensitized animals attenuated leukocyte accumulation in the heart (Fig. 6 A), ameliorated inflammation, and decreased fibrosis at day 28 after the development of BMDC-induced EAM (Fig. 6 B). In addition, animals treated with anti-IL-3 antibody had improved cardiac function based on reduced left ventricular (LV) end-systolic diameter (LVESD) and elevated LV ejection fraction (LVEF; Fig. 6, C and D). Altogether, these data demonstrate the clinical relevance of IL-3. Since, as these data show, anti-IL-3 antibody can ameliorate inflammation, even in sensitized animals, it could be a useful therapeutic target for treating myocarditis. Ultimately, these results position IL-3 within the inflammatory cascade that occurs during autoimmune myocarditis as a critical factor that escalates the effector phase by inducing myeloid cell recruitment (Fig. 7).

In this study, we identified and characterized an amplification loop, essential to autoimmune inflammation in myocarditis, that depends on the cytokine and growth factor IL-3. We show that IL-3–producing self-reactive effector CD4⁺ T cells incite tissue macrophages to produce monocyte-attracting chemokines, that newly recruited monocytes differentiate into monocyte-derived macrophages and DCs, and that the process repeats itself when DCs, upon encountering effector T cells in the tissue, stimulate local T cell proliferation.

The biological function of IL-3 has been predominantly studied in the context of mast cells, basophils, hematopoietic stem progenitor cells, and endothelial cells (Williams et al., 1990; Korpelainen et al., 1996; Lantz et al., 1998; Voehringer, 2013; Weber et al., 2015). However, the cytokine’s action on macrophages has been less explored. Likewise, although several studies have shown that IL-3 promotes certain kinds of autoimmune inflammation (Ito et al., 1997; Hofstetter et al., 2005; Brühl et al., 2009; Renner et al., 2015, 2016), its mechanism of action has remained elusive. For example, a previous study by Mack and colleagues using a mouse model of CIA reported IL-3’s effects to be pathogenic during the early phase of CIA, whereby blockade with anti-IL-3 at disease onset reduced myeloid cell and basophil infiltration into synovial tissue, proinflammatory cytokine production, and histological joint scores (Brühl et al., 2009). On the other hand, Wani and colleagues reported that IL-3 suppressed CIA development through indirectly inducing regulatory T (T reg) cell expansion by stimulating IL-2 production via IL-3:IL-3R signaling on non–T reg cells (Srivastava et al., 2011). While different animal models, study conditions, and mechanisms controlling T reg cell differentiation and function could account for these contrasting results, our data ultimately align with models that have revealed a pathogenic role for IL-3 in promoting autoimmune inflammation and leukocyte accumulation in the disease organ (Brühl et al., 2009; Renner et al., 2015, 2016).

The effector phase of autoimmune diseases such as autoimmune myocarditis is characterized by massive inflammation that involves recruitment of CCR2⁺ Ly-6C^high monocytes (Göser et al., 2005; Wu et al., 2014; Leuschner et al., 2015; Clemente-Casares et al., 2017). Our study places this phenomenon into a larger context of reciprocally controlled cellular function. On the one hand, the link between IL-3–producing T cells and monocyte recruitment reflects a functional hierarchy whereby effector T cells control monocyte recruitment in the effector tissue by stimulating resident macrophages. On the other hand, our data also show that accumulating monocytes, which are bipotential and can differentiate to fully functional moDCs and MHCII⁺ macrophages in cardiac tissue (Cheong et al., 2010; Epelman et al., 2014; Clemente-Casares et al., 2017; Van der Borght et al., 2017), influence T cells by stimulating their proliferation in a manner that is IL-3 dependent. This secondary reactivation in the effector organ is gaining interest and might play a role in disease relapse (Aloisi et al., 2000; Greter et al., 2005; Bailey et al., 2007; Becher et al., 2017). Indeed, in EAE, it has been shown that both CD11c⁺ DCs and Ly-6C⁺ monocyte precursors, which can differentiate into moDCs and macrophages, traffic to the central nervous system in a CCR2-dependent manner and are critical for disease progression (Izikson et al., 2000; King et al., 2009; Sagar et al., 2012a; Clarkson et al., 2015). While a role for central nervous system–infiltrating CCR2⁺ DCs in driving the accumulation of coinfiltrating, self-reactive T cells has been postulated in EAE (Wlodarczyk et al., 2014; Clarkson et al., 2015), how this process occurs upstream of CCR2-mediated recruitment in autoimmunity has remained unknown. Given our observations in EAM that IL-3 is responsible for directing the accumulation of moDCs and MHCII⁺ macrophages, which subsequently stimulate self-reactive CD4⁺ T cell proliferation and cytokine production during the peak of inflammation, it is plausible that IL-3 may also act analogously in other models of T cell–mediated autoimmune inflammation upstream of CCR2⁺ myeloid cell recruitment. Thus, uncontrolled inflammation occurs during the effector phase of autoimmune inflammation, our study suggests, not because one cell controls another but because of a positive IL-3–dependent reinforcement amplification loop that leads to a progressively larger leukocyte presence in the target organ.

It remains unclear whether IL-3–producing CD4⁺ T cells are a distinct Th cell lineage. On the one hand, it is possible the IL-3⁺ IFN-γ⁻ IL-17A⁻ CD4⁺ T cells identified in this study are a population of recently activated effector T cells that have yet to differentiate into other inflammatory Th lineages. On the other hand, they may instead be related to the recently described neuroinflammatory GM-CSF⁺ CD4⁺ T cells (Hartmann et al., 2014; Noster et al., 2014; Sheng et al., 2014). Future work is needed to identify any potential lineage-defining transcription factors that might support IL-3⁺ CD4⁺ T cell generation and lineage stability.

Our study’s finding that IL-3 is produced in human myocarditis hearts, coupled with the observation in mice that IL-3 is an essential amplifier of inflammation in the effector organ, identifies the cytokine as a potential therapeutic target to counteract T cell–driven cardiac tissue inflammation and fibrosis. Future work will show whether neutralizing IL-3 can be effective for treating viral myocarditis and other inflammatory cardiac complications, such as acute myocarditis and heart failure associated with checkpoint blockade immunotherapy (Swirski and Nahrendorf, 2018). IL-3, our study shows, contributes to this T cell–mediated inflammatory cascade, and though it may act downstream of sensitization, its profound effects on disease progression reveal it to be a major hub of influence.

BALB/c (Taconic), C.129S7(B6)-Rag1tm1Mom/J, CByJ.B6-Tg(UBC-GFP)30Scha/J, CBySmn.CB17-Prkdcscid/J were purchased from The Jackson Laboratory. IL-3–deficient (Il3^−/−) mice on a BALB/c background (nine generations) were bred in-house (Weber et al., 2015). All experiments were conducted with female mice unless otherwise indicated. All protocols were approved by the Animal Review Committee at Massachusetts General Hospital (Protocols 2011N000035 and 2015N000044) and by the Institutional Animal Care and Use Committee at the Keio University School of Medicine.

Mice were subcutaneously injected with 100 µg cardiac αMHC (Ac-RSLKLMATLFSTYASADR-OH) peptide emulsified in CFA containing 1 mg/ml heat-killed Mycobacterium tuberculosis H37Ra (Sigma-Aldrich) on days 0 and 7 (Leuschner et al., 2015). Myocarditis was also achieved by αMHC-pulsed, activated BMDCs as previously described (Eriksson et al., 2003). In brief, BM cells were cultured in complete medium (RPMI-1640 medium supplemented with 10% FBS, 2 mM L-glutamine, 100 U/ml penicillin and streptomycin, 10 mM Hepes, 50 µM 2-mercaptoethanol, 1 mM sodium pyruvate, and 1× nonessential amino acids) supplemented with 10 ng/ml GM-CSF (PeproTech) for 6 d. After collecting nonadherent cells, CD11c⁺ cells were enriched by magnetic separation. The cells were then pulsed with 10 µg/ml αMHC overnight and activated by 1 µg/ml LPS (Sigma–Aldrich) and 5 µg/ml of antibody against CD40 (BD Biosciences) for 4 h. Subsequently, 3–6 × 10⁵ activated BMDCs were intraperitoneally injected into recipient mice on days 0, 2, and 4. In some experiments, spleen cells were collected from immunized WT or Il3^−/− mice on day 14 and pulsed with 20 µg/ml αMHC for 72 h. CD4⁺ T cells were isolated by magnetic separation, and 10⁷ of these T cells were adoptively transferred into CBySmn.CB17-Prkdcscid/J (Scid) mice to induce myocarditis.

BM Ly-6C^high monocytes were obtained from CByJ.B6-Tg(UBC-GFP)30Scha/J mice by flow-assisted cell sorting after enrichment of cells negative for CD3, CD19, B220, CD49b, and Ly6G using magnetic cell separation. 15 d after the first immunization, 2 × 10⁶ GFP⁺ Ly-6C^high monocytes were injected intravenously into WT BALB/c mice, and GFP⁺ cells in the inflamed heart were assessed 2 d later.

Naive WT BALB/c mice were lethally irradiated (500 cGy twice, 4 h apart) and reconstituted with a mixture of BM cells obtained from Rag1^−/− and either WT or Il3^−/− mice to generate Rag1^−/−/WT and Rag1^−/−/Il3^−/− BM mixed chimeras. After 6–7 wk to allow for reconstitution, the mice were subjected to EAM induction, and leukocyte populations in the heart were evaluated by flow cytometry on day 21.

To assess cell proliferation, 1 mg BrdU was injected intraperitoneally 2 h before euthanasia and subsequent organ harvest. A BrdU flow kit (BD Biosciences) was used to stain BrdU⁺ cells.

WT BALB/c mice were randomly assigned to either the control IgG or anti-CD317 antibody (300 µg; clone BX444; Bio X Cell)–injecting group and subjected to BMDC-induced myocarditis. The antibodies were intraperitoneally injected into mice every other day from 1 d before the first BMDC injection.

After injection of activated BMDCs pulsed with αMHC into WT mice on days 0, 2, and 4, mice were randomly allocated to either the anti-IL-3–treated group or the control IgG-treated group. Neutralizing antibody against IL-3 (100 µg; clone MP2-8F8; Bio X Cell) or control IgG1 (100 µg; clone HRPN; Bio X Cell) was intraperitoneally injected once a day for 7 d.

Transthoracic echocardiography was performed with a Vevo 2100 instrument (VisualSonics) equipped with an MS-400 imaging transducer. M-mode tracings were recorded through the anterior and posterior LV walls at the papillary muscle level to measure LV end-diastolic diameter (LVEDD), LVESD, and LVEF. LV mass was assessed to evaluate cardiac hypertrophy.

Peripheral blood was collected by retro-orbital bleeding, and erythrocytes were lysed in RBC lysis buffer (BioLegend). Spleen, draining LNs, femurs, tibia, and heart were excised after vascular perfusion with cold PBS. Minced spleen and LNs and flushed BM were strained through 40-µm nylon mesh (BD Biosciences) and further subjected to RBC lysis. The naive and inflamed hearts were minced and digested with 450 U/ml collagenase I, 125 U/ml collagenase XI, 60 U/ml DNase I, and 60 U/ml hyaluronidase (Sigma-Aldrich) in PBS for 1 h at 37°C while shaking. For cell sorting, single-cell suspensions of heart tissue from indicated animals were made as described above and stained to identify indicated cell populations. Cells were sorted on a FACS Aria II cell sorter (BD Biosciences) directly into either RLT lysis buffer (Qiagen) for subsequent RNA isolation or complete medium for cell culture experiments.

Single-cell suspensions were stained in PBS supplemented with sterile 2% FBS and 0.5% BSA. The following monoclonal antibodies were used for flow cytometric analysis: anti-CD45 (30-F11), anti-CD3 (clone 145-2C11), anti-CD4 (clone RM4-5), anti-CD8 (clone 53–6.7), anti-CD90.2 (clone 53–2.1), anti-B220 (clone RA3-6B2), anti-Ly6G (clone 1A8), anti-Ly-6C (AL-21), anti-I-A/I-E (MHCII; clone M5/114.15.2), anti-CD11b (clone M1/70), anti-CD11c (clone HL3), anti-CD64 (clone X54-5/7.1), anti-CD103 (clone 2E7), anti-PDCA1 (clone 927), anti-CD62L (clone MEL-14), anti-CD80 (clone 16-10A1), anti-OX40L (clone RM134L), anti-CD123 (clone REA114), anti-CD131 (clone REA193), anti-CD31 (clone 390), anti-Feeder (MEFSK4; clone mEF-SK4), anti-CD192 (CCR2; clone 475301; clone SA203G11), anti-F4/80 (clone BM8), anti-CD115 (CSF-1R; clone AFS98), anti-Siglec-F (clone E50-2440), anti-CD24 (clone M1/69), anti-CD26 (clone H194-112), anti-XCR1 (clone ZET), anti-CD172a (SIRPα; clone P84), anti-Zbtb46 (clone U4-1374), anti-CD117 (c-Kit; clone 2B8), anti-IgE (clone RME-1), anti-FcεRIα (clone MAR-1), anti-CD49b (clone DX5), anti-IL-3 (clone MP2-8F8), anti-IL-4 (clone 11B11), anti-IFNγ (clone XMG1.2), anti-IL-17A (clone TC11-18H10.1), anti-GM-CSF (clone MP1-22E9), anti-activated caspase-3 (clone C92-605), anti-pSTAT5 (clone pY694), and isotype controls. Antibodies were purchased from BioLegend, BD Biosciences, eBioscience, or Miltenyi Biotec. Viable cells were identified as unstained cells with Zombie Aqua (BioLegend).

For intracellular cytokine staining, cell suspensions were stimulated in 2% FBS RPMI-1640 medium with 100 ng/ml PMA and 1 µg/ml ionomycin (Sigma-Aldrich) in the presence of GolgiStop and GolgiPlug (BD Biosciences) for 3–4 h at 37°C, 5% CO₂ before fixation and permeabilization. Staining for pSTAT5 was performed as previously described (Anzai et al., 2017). In brief, cells were stimulated with or without recombinant murine IL-3 (20 ng/ml; PeproTech) for 30 min at 37°C to induce phosphorylation of STAT5, followed by addition of 4% PFA solution (pH 7.4) to obtain a final concentration of 2%. After fixation for 20 min, cells were resuspended in 1 ml ice-cold methanol to permeabilize. After 30 min of incubation at 4°C, cells were washed twice and stained with anti-pSTAT5 antibody as well as antibodies for extracellular antigens for 45 min at room temperature before a final wash and acquisition. For proliferation assays, CD4⁺ T cells were labeled with Cell Tracer Violet (Thermo Fisher Scientific), and proliferating cells were identified as Cell Tracer Violet-dim. Data were acquired on a BD LSRII and analyzed with FlowJo (Tree Star).

WT and Il3^−/− mice were subjected to CFA-induced EAM induction, and autoreactive T cells were isolated from draining LNs on day 14. Cardiac APCs such as moDCs, cDC1, cDC2, and pDCs, as well as MHCII^high macrophages and Ly-6C^high monocytes/macrophages, were sorted from the heart at peak inflammation, and BMDCs were generated as described above. 5 × 10⁵ T cells with 10⁵ BMDCs or 5 × 10⁴ T cells with 10⁴ sorted cardiac APCs were cultured in 200 µl complete medium with or without 10 µg/ml αMHC for 72 h in a humidified 5% CO₂ incubator at 37°C. The supernatants were collected, and indicated cytokines were measured by ELISA as described below. For the T cell proliferation assay, T cells were stained with Cell Trace Violet before incubation according to the manufacturer’s instruction. Recombinant murine IL-3 (20 ng/ml; PeproTech) was added to some culture wells. To evaluate the effect of IL-3 on macrophages, J774 macrophage cell lines (Sigma-Aldrich), peritoneal macrophages, BMDMs, and sorted cardiac MHCII⁺ macrophages (CD64⁺ MHCII⁺ Ly-6C⁻ F4/80⁺) from naive mice were used. Peritoneal macrophages were obtained as previously described (Rauch et al., 2012). To prepare BMDMs, BM cells were isolated from adult WT BALB/c mice and cultured in complete medium supplemented with 10 ng/ml recombinant murine M-CSF (PeproTech) for 8 d. The macrophage cell lines and primary cells were stimulated with or without 20 ng/ml recombinant IL-3 for 8 h and subjected to mRNA isolation as described below.

LentiCRISPRv2GFP was a gift from David Feidser (University of Pennsylvania, Philadelphia, PA; plasmid 82416; http://n2t.net/addgene:82416; RRID:Addgene_82416; Addgene). sgRNA targeting mouse IL-3Rrα (5′-GGTCCTGGGCTGCGGCCTCG-3′) was cloned into the vector using Golden Gate assembly with BsmBI restriction enzyme. Validation of sgRNA was performed by flow cytometric analysis for phosphorylation of STAT5 after IL-3 stimulation. To generate lentivirus particles, plasmids (LentiCRISPRv2GFP-IL-3Rα, pMD2G, psPAX2) were cotransfected to HEK 293T cells, and supernatant was collected 48 h after the transfection. Following filtration (40 µm), virus particles were concentrated with ultracentrifugation at the speed of 20,000 rpm for 3 h. Subsequently, virus pellet was suspended with medium without aeration. To generate IL-3Rα–deficient cells in vivo, BM lineage–negative cells were isolated from adult BALB/c mice using EasySep Mouse hematopoietic progenitor cell isolation kit (STEMCELL Technologies) according to the manufacturer’s instruction. The isolated progenitor cells were cultured in StemSpan (STEMCELL Technologies) supplemented with 50 ng/ml murine Tpo and 50 ng/ml murine Scf (both PeproTech) for 1 h and then transduced with concentrated lentiviral supernatant. 16 h after transduction, cells were collected and transplanted into lethally irradiated (500 cGy, twice, 4 h apart) WT BALB/c mice. After 6 wk of recovery, either EAM was induced.

Immunized mice were euthanized, and hearts were removed, embedded in Tissue-Tek O.C.T. compound (Sakura Finetek), frozen in 2-methylbutane (Fisher Scientific) cooled with dry ice, and sectioned into 6-µm slices. H&E staining was used to assess acute cardiac inflammation, which was evaluated by histopathological microscopic approximation of the percent area of myocardium infiltrated with mononuclear cells. To assess cardiac fibrosis, fibrotic area was quantified in 5–10 randomly selected fields of view per Azan-stained sample by ImageJ software.

Mouse total RNA was isolated using the RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. cDNA was generated from 1 µg of total RNA per sample using High Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Human myocardial tissue samples were obtained from two patients with confirmed myocarditis and one patient with normal myocardium, frozen, and kept in the Brigham Research Institute tissue bank at Brigham and Women’s Hospital. Those tissues were approved by Institutional Review Board Protocol 1999P991348/BWH. Total RNA and cDNA extraction were performed as described above. Two additional control heart total RNA were also obtained from Thermo Fisher Scientific (lot number 1906770 and 1866106), followed by generation of cDNA.

5 × 10³ Ly-6C^high monocytes/macrophages, moDC, MHCII^high macrophages, pDC, cDC1, cDC2, neutrophils, T cells, eosinophils, B cells, and other leukocytes and 2 × 10⁴ total leukocytes, endothelial cells, fibroblasts, and other stromal cells were sorted from inflamed heart, and total RNA was extracted using the RNeasy Micro Kit (Qiagen) followed by cDNA transcription. Quantitative real-time TaqMan PCR was performed using following TaqMan primers (Applied Biosystems): Il3 (Mm00439631_m1), Il1α (Mm00439620_m1), Il1β (Mm01336189_m1), Il6 (Mm00446190_m1), Il10 (Mm01288386_m1), Il22 (Mm01226722_g1), Il23a (Mm00518984_m1), Tnf (Mm00443258_m1), Tgfβ (Mm01227699_m1), Mmp2 (Mm00439498_m1), Mmp9 (Mm00442991_m1), Csf2 (Mm01290062_m1), Vegf (Mm00437306_m1), Cxcl1 (Mm04207460_m1), Cxcl2 (Mm00436450_m1), Ccl2 (Mm00441242_m1), Ccl3 (Mm00441259_g1), Ccl5 (Mm01302427_m1), Ccl7 (Mm00443113_m1), Ccl12 (Mm01617100_m1), Ccl19 (Mm00839967_g1), Ccl20 (Mm01268754_m1), Ccl24 (Mm00444701_m1), Bax (Mm00432051_m1), Bim (Mm00437796_m1), Bcl2 (Mm00477631_m1), Il3rα (Mm00434273_m1), and housekeeping gene Gapdh (Mm99999915_g1). PCR was run on a 7500 Fast Real-Time PCR system (Applied Biosystems). Gene expression was calculated relative to Gapdh and normalized to controls. Human IL3 gene expression data were quantified by 2^−ΔCT, and mouse gene expression data were calculated by the 2^−ΔΔCT method.

IL-3 levels were measured in cell culture supernatants with the Mouse IL-3 ELISA Kit (R&D Systems or Boster Biological). IL-2, IL-4, IL-6, IL-17A, IFN-γ, TNF-α, and GM-CSF levels were also measured in cell culture supernatants with ELISA Kits from R&D Systems according to the manufacturer’s instructions. High-sensitivity mouse troponin I ELISA kit (Life Diagnostics) was used to measure serum cardiac troponin I after EAM induction.

GraphPad Prism 8.0 (GraphPad Software) was used for statistical analyses for all experiments. Results are shown as mean ± SEM. Data were analyzed for normality using the D’Agostino–Pearson test. Statistical tests included unpaired, two-tailed nonparametric Mann–Whitney U tests (when Gaussian distribution was not assumed) and two-tailed unpaired t tests for normally distributed data. For multiple comparisons, nonparametric multiple comparisons test comparing mean rank of each group (when Gaussian distribution was not assumed) or one-way ANOVA followed by Tukey’s test was performed. P values of ≤0.05 were considered significant.

Fig. S1 shows that Il3^−/− male mice are likewise protected from EAM inflammation and depicts flow cytometric profiling for identifying inflammatory cell subsets in the inflamed heart. Fig. S2 shows that IL-3 does not affect T cell sensitization and peripheral leukocyte numbers in EAM. Fig. S3 shows that pDC depletion does not affect cardiac leukocyte numbers during EAM and depicts the cell sorting strategy for non-leukocytes, pSTAT5 activity in leukocyte subsets by flow cytometry, and assessment of IL-3 receptor expression and basophil quantification in the periphery and inflamed heart by flow cytometry. Fig. S4 shows the profiling chemokine expression in different cell subsets in myocarditis and IL-3’s effects on proliferation, apoptosis, and chemokine receptor expression in myeloid cells. Fig. S5 shows how IL-3 enhances monocyte-attracting chemokine expression in macrophages.

The authors thank S. Ko and Y. Miyake (Keio University School of Medicine) for technical assistance, the Massachusetts General Hospital (MGH) Peptide/Protein Core Laboratory for αMHC peptide synthesis, and the MGH Harvard Stem Cell Institute–Center for Regenerative Medicine Flow Cytometry Core Facility for cell sorting.

This work was supported in part by the National Institutes of Health (grants R35 HL135752, R01 HL128264, and P01 HL131478); the American Heart Association’s Established Investigator Award; and the Patricia and Scott Eston MGH Research Scholar Award (to F.K. Swirski). A. Anzai was supported by grants from the Uehara Memorial Foundation, the Kanae Foundation for the Promotion of Medical Science, and the Japan Society for the Promotion of Science (KAKENHI grant 18K08048). L. Halle was supported by a Boehringer-Ingelheim Fonds MD fellowship. S. Sano was supported by an American Heart Association postdoctoral fellowship. C.S. McAlpine was supported by Canadian Institutes of Health Research and Banting Research Foundation postdoctoral awards. C. Valet was supported by a Fondation pour la Recherche Medicale postdoctoral fellowship. D. Fairweather was supported by the National Institutes of Health (grant R01 HL111938). K. Walsh was supported by the National Institutes of Health (grant R01 HL131006).

The authors declare no competing financial interests.

Author contributions: A. Anzai and J.E. Mindur conceived the project, designed and performed experiments, analyzed and interpreted data, made the figures, and drafted the manuscript; L. Halle, S. Sano, J.L. Choi, S. He, C.S. McAlpine, C.T. Chan, F. Kahles, C. Valet, A.M. Fenn, M. Nairz, S. Rattik, and Y. Iwamoto performed experiments; D. Fairweather, K. Walsh, P. Libby, and M. Nahrendorf provided intellectual input and edited the manuscript; D. Fairweather, K. Walsh, and P. Libby provided materials; and F.K. Swirski conceived the project, designed experiments, interpreted data, and wrote the manuscript.