Inhibition of NETosis via PAD4 alleviated inflammation in giant cell myocarditis

Summary Giant cell myocarditis (GCM) is a rare, usually rapidly progressive, and potentially fatal disease. Detailed inflammatory responses remain unknown, in particular the formation of multinucleate giant cells. We performed single-cell RNA sequencing analysis on 15,714 Cd45+ cells extracted from the hearts of GCM rats and normal rats. NETosis has been found to contribute to the GCM process. An inhibitor of NETosis, GSK484, alleviated GCM inflammation in vivo. MPO (a marker of neutrophils) and H3cit (a marker of NETosis) were expressed at higher levels in patients with GCM than in patients with DCM and healthy controls. Imaging mass cytometry analysis revealed that immune cell types within multinucleate giant cells included CD4+ T cells, CD8+ T cells, neutrophils, and macrophages but not B cells. We elucidated the role of NETosis in GCM pathogenesis, which may serve as a potential therapeutic target in the clinic.


INTRODUCTION
Giant cell myocarditis (GCM) is a rare and highly fatal disease; its typical pathological characteristics include infiltration of inflammatory cells, myocardial necrosis, and multinucleate giant cell formation. 1 In 1905, Saltykow described and named the first case of GCM, and approximately 100 GCM cases were reported worldwide in the next 100 years. 2,3 The development of GCM is rapid, with approximately 2-3 weeks passing from symptom onset to hospitalization, and the major clinical presentations are acute congestive heart failure and ventricular arrhythmia. 4 Clinically, almost 20% of patients with GCM also present with autoimmune diseases such as inflammatory bowel disease, thyroiditis, celiac disease, and rheumatoid arthritis, and there have been a few patients with thymoma or lymphoma. [5][6][7] One typical characteristic of patients with GCM is multinucleate giant cell formation, which is the basis of GCM diagnosis. 8 The prognosis of GCM is extremely poor before the usage of immunosuppression, as almost all patients need heart transplantations. 9 Early treatment with cyclosporin-based immunosuppression, improved the 1-year survival rate of giant cell myocarditis to 90%. 10 Moreover, the recurrence of GCM after heart transplantation is high; long-term immunosuppressant therapy can reduce the risk of recurrence. 4,11 The pathogenesis of GCM is still unclear, but it is considered an autoimmune disease. Animal models indicate that GCM can be induced by autoimmunity, and inflammation cell analysis showed that the major inflammatory cells included macrophages and Cd4 + T cells. 12, 13 Lassner D et al. examined gene expression profiles in endomyocardial biopsy samples from 10 patients with histopathologically proven idiopathic giant cell myocarditis, 10 with cardiac sarcoidosis (CS), 18 with active myocarditis, and 80 inflammation-free control subjects by quantitative real-time PCR. 14 Their findings suggested that genes such as CCR5 and CCR6 related to T cells were expressed at higher levels in patients with GCM, which supports the opinion that patients with GCM suffer from massive T cell infiltration. 14 The detailed cardiac immunological environment of GCM is unknown. In addition, infiltration of neutrophils in GCM has rarely been reported due to the consensus that macrophages and Cd4 + T cells are predominant components of infiltrating mononuclear cells in GCM. 8 Neutrophil extracellular traps (NETs) are widely reported in a large range of inflammatory infectious and noninfectious diseases, and NETs can also activate other immune cells, such as B cells, antigen-presenting cells, and T cells. 15 The process of NET generation, called NETosis, is a specific type of cell death that differs from necrosis and apoptosis. It was reported that NETosis participated in autoimmunity as an inflammation inducer. 16 The roles of NETosis in the pathogenesis of GCM are unknown. Currently, a variety of approaches to therapeutically target neutrophils have emerged, including strategies to enhance, inhibit, or restore neutrophil function, with several agents entering clinical trials. 17

scRNA-seq analysis of the GCM model
The rat GCM model was established according to a previous report, 23 and the clinicopathological phenotype was evaluated on day 21. The left ventricle end-diastolic dimension (LVEDD) was larger while the left ventricular ejection fraction (LVEF) was lower in the GCM group (LVEDD: 5.08 G 0.16 vs. 5.82 G 0.15, p = 0.008; LVEF: 86.70 G 2.95 vs. 77.60 G 2.32, p = 0.035; Supplementary material online, Figures S1A-S1D). The classic histopathologic features can be seen in GCM groups, including the presence of multinucleated giant cells, a lymphocytic inflammatory infiltrate, and myocyte necrosis (Supplementary material online, Figure S1E). 24 Following the histological examination, we pooled 5 hearts from each group to isolate single cells. The chosen GCM rat hearts shared similar immune states.
The harvested cardiac Cd45 + cells from GCM and normal controls were sequenced on a 10x Genomics platform ( Figure 1A; Supplementary material online, Figure S2 Table S2 and Figure S4). The immunological constitution was completely different between the two groups. The top 5 immune cell clusters were NK cluster 1, T cell cluster 1, macrophage cluster 3, neutrophils cluster 2, and cDCs in the control group, while the top 5 immune cell clusters were macrophage cluster 1, neutrophil cluster 3, macrophage cluster 2, neutrophil cluster 1, and T cell cluster 2 ( Figure 1C). Each cell cluster was derived from different phases and had different cell numbers and transcriptional activities, as determined by unique molecular identifiers ( Figure 1D; Supplementary material online, Table S2). Some immune cell clusters were derived from different groups, indicating that these cell clusters may contribute to certain pathophysiological processes.

Two specific macrophage clusters involved in GCM have roles in phagocytosis
Macrophages were reported as the major immune cells that make up giant cells. 8 In this study, 5,284 macrophages were detected as the largest cell population and were clustered into 5 clusters (Figure 2A). According to the macrophage distribution, macrophage clusters 1 and 2 were almost all derived from GCM, indicating that these two cell clusters may be related to GCM (Figure 2A). The macrophage polarization pattern is vital in macrophage biology 25 ( Figure 2B); in this case, classical macrophage polarization was not suitable to study the polarization patterns of macrophages at single-cell resolution. 26 The gene expression of all 5 clusters was distinct ( Figure 2C). Prdx1 and Prdx5 were the hallmark genes of macrophage cluster 1 ( Figure 2C), and these genes were reported to be associated with autoimmunity. 27,28 In addition, macrophage cluster 1 expressed arginine synthesis-associated genes such as Arg1 and Ass1. 29 Arginine, whose homeostasis is severely disturbed and can be triggered by infiltrating neutrophils as well as bacterial components, 29 supports continual efferocytosis in macrophages. 30 Macrophage cluster 2 expressed Ms4a family genes such as Ms4a7, Tmem176a, and Tmem176b ( Figure 2C). In addition, macrophage cluster 3, the major macrophage cluster under normal conditions, expressed Nr4a1, which inhibited the polarization of iScience Article macrophages toward a proinflammatory M1 phenotype. 31 Macrophage cluster 4 expressed Pf4, a pleiotropic inflammatory chemokine, which limited the activation of resident macrophages. 32 Macrophage cluster 5 was characterized by the expression of Fcnb, Plac8 (Onzin), and Vcan ( Figure 2C). Macrophage cluster iScience Article 1/2 had high scores for interferon-stimulated genes, phagocytosis, and inflammation ( Figure 2D; Supplementary material online, Table S3). The immunohistochemistry (IHC) results showed that macrophage clusters 1 (ARG1 + macrophages) and 2 (C1QA + macrophages) were the hallmark macrophage clusters of GCM. Cardiac macrophages consist of tissue-resident cells and recruited monocyte-derived macrophages from the systemic circulation. 33 Macrophage cluster 1 was monocyte-derived macrophages, and macrophage cluster 2 was mixed macrophages according to the expression of the marker gene (monocyte-derived macrophages: Gpnmb 34 , Ly6c; resident macrophages: Mrc1, C1qa; Supplementary material online, Figure S5).

Th17 cell activation in the GCM
A total of 2,619 T cells were detected and clustered into 4 clusters ( Figure 3A). Although GCM has been reported as a T cell-mediated autoimmune disease, 8 no reports have investigated the details of the infiltration of T lymphocytes. T helper (Th) cells are the major effector T cells, 25 so we investigated Th-related cytokine expression in T cell clusters ( Figure 3B). Th1 cytokines (Ifna, Ifng) were expressed at low levels in all four T cell clusters, Th2 cytokines (Il4, Il5) were not detected in T cell clusters, and Th17 cytokines (Il17) were highly expressed in T cell cluster 2 ( Figure 3B). According to the important surface marker (Cd4, Cd8), cluster 3 T cells were cytotoxic T lymphocytes (CTLs), while cluster 2 T cells were Th17 cells ( Figure 3B). In the GCM, CD4 + T cells were the majority of the total T cell population (more than 90%), which may explain why researchers agree that Cd4 + T cells are the major T cell population in GCM. 35 In this study, we found that Th17 cells were the major T cell population of GCM (more than 80%), which was the first report to reveal the detailed T cell population of GCM at the single-cell level. There were different gene expression patterns in T cell clusters ( Figure 3C). T cell cluster 1 was Ccr7 + Lef1 + Sell + T cells, which belonged to central memory T cells. 36 T cell cluster 2 was characterized by the expression of Il17a, Il22, and Cd74. T cell cluster 3 expressed cytotoxicity factors such as Nkg7, Gzmm, Klrb1b, and Gzmk, which suggested that these cluster cells were cytotoxic T cells. T cell cluster 4 was Cd4 + Il2ra (Cd25) + Foxp3 + T cells, suggesting regulatory T cells (Treg cells). The mechanism of T cell activation is important in T cell biology, 37 and we found that different activation factors were detected in different T cell clusters. Cd27 was detected in T cell cluster 1, CTLs, and Treg cells but not in Th17 cells. Tnfrsf25 and Cd40lg were only expressed in Th17 cells, while Tnfrsf9 was only expressed in Treg cells. In addition, some activation factors, such as Ctla4, Icos, and Lag3, were expressed at higher levels in Treg cells. For CTLs, Gzmm was only expressed in this cluster. Together, under normal conditions, the major T cells were central memory T cells, while Th17 cells were the hallmark T cell population in GCM, which cluster expressed Cd74 and Il22.

NETosis involved in the pathogenesis of GCM
The role of neutrophils has always been ignored in GCMs. A total of 3,237 neutrophils were detected and clustered into 3 clusters ( Figure 4A). Neutrophils were the second largest cell population of GCM, with 32.86% of the total population from GCM, while neutrophils only accounted for 10.37% of the total population of normal controls (Supplementary material online, Table S2). This result indicated increased neutrophil infiltration into the infiltrates of myocardial rats with GCM. Considering that Il1 plays a pivotal role in the pathogenesis of myocarditis, we investigated its expression in neutrophils. 38 As shown in Figure 4B, Il1b was expressed in all three clusters, while Il1a was only expressed in neutrophil cluster 3. Il1a is a potential early trigger of acute inflammation, 39 so neutrophil cluster 3 may be the trigger of inflammation in GCM. In addition, neutrophil cluster 1 was mostly from GCM, which may be associated with the acute inflammatory response. The three neutrophil clusters shared different gene expression patterns ( Figure 4C): neutrophil cluster 1 expressed Socs3, Slpi, Irf1, and Alas1; neutrophil cluster 2 expressed S100a9/8; and neutrophil cluster 3 expressed Cxcl3, Ccl2, Cxc1, Ccl4, and Cxcl11. The functions of different neutrophils suggested that NETosis-related pathways were enriched in neutrophil cluster 1, which was the hallmark cell cluster in GCM, such as immune effector process, endocytosis, positive regulation of cell death, and apoptosis ( Figures 4A and 4D). Consistent with this finding, the genes related to those pathways were expressed at higher levels in neutrophil cluster 1 ( Figure 4E). IHC of MPO (a marker of neutrophils) and H3cit (a marker of NETosis) showed that NETosis was involved in the GCM process ( Figure 4F). Combining the functions of iScience Article GCM-related neutrophils, including myeloid leukocyte migration and production of molecular mediators involved in the inflammatory response, we speculated that NETosis contributed to the pathogenesis of GCM by recruiting other infiltrating immune cells into the heart, forming multinucleate giant cells. As reported, the transmembrane protein Ccdc25 is a NET-DNA receptor on cancer cells that senses extracellular DNA and subsequently activates the ILK-b-parvin pathway to enhance cell motility, 40 so we inferred that NETosis recruited immune cells via Ccdc25. To validate this hypothesis, Ccdc25 was expressed in immune cells in the GCM group, especially macrophage Cluster 1/2 and Th17 cells, and IHC staining was conducted iScience Article to suggest that the expression of Ccdc25 was higher in the GCM group ( Figure 4H). Taken together, NE-Tosis was involved in the pathogenesis of GCM.

Intercellular communication network analysis revealed the roles of NETosis in GCM
We performed an intercellular communication network analysis of GCM samples by CellChat. 41 The biologically significant cell-cell communication network was inferred using CellChat, which assigns each interaction with a probability value and performs a permutation test. From the cell-cell communication network, we found some significant signaling pathways, such as APRIL, BAFF, TWEAK, THY1, THBS, SPP1, SN, IL1, CXCL, and CSF (Supplementary material online, Figure S6).
From the source sets to the target sets (Neu_1, Neu_2, Neu_3, Supplementary material online, Figure S6A), there are significant signaling pathways, such as APRIL, BAFF, TWEAK, THY1, THBS, SPP1, and SN, especially IL-1. Furthermore, IL-1 plays a pivotal role in the pathogenesis of myocarditis. 26 In addition, the IL1B-IL1R2 iScience Article pairs were found primarily between macrophage clusters and neutrophil clusters. This suggests that macrophages may induce NETosis, which was consistent with reports that macrophage-derived IL-1b enhances NET formation. 42,43 From the source sets (Neu_1, Neu_2, Neu_3) to the target sets (Supplementary material online, Figure S6B), there are significant signaling pathways, such as CXCL and CSF. CSF1-CSF1 receptor (CSF1R) pairs play a pivotal role in the CSF pathway. As reported, M2 polarization is known to be triggered by CSF1/CSF1R signaling, 44 supporting that macrophage clusters (Macro_1, Macro_2) could be triggered by infiltrating neutrophils (Neu_3), supporting continual efferocytosis in macrophages. The CXCL signaling pathway indicates that there is communication between neutrophil clusters (Neu_3) and T cells (T_2, T_3, T_4), and CXCR3 is the major receptor in T cells. CXCR3 was reported to guide effector and memory T cell migration to inflammatory lesions and contribute to disease pathogenesis. The expression levels of CXCR3 ligands (Cxcl9, Cxcl10, and Cxcl11) were highest in Neu_3 among neutrophils (Supplementary material online, Figure S7). In addition, these ligands were also expressed in some monocyte-derived macrophage clusters in GCM rats compared to control rats (Supplementary material online, Figures S5 and S7). CXCR3 ligands might attract these T cells, thereby promoting disease. 45 Together, NETosis may contribute to the pathogenesis of GCM by recruiting other infiltrating immune cells into the heart but with different pairs.

Inhibition of NETosis in GCM
As described above, macrophage cluster 1/2 and Th17 cells were the major inflammatory cell clusters of GCM, which expressed Ccdc25 ( Figure 4G), the receptor of NET-DNA. As the roles of neutrophils in the pathogenesis of GCM were reported for the first time, we further investigated the contribution of NETosis to GCM through an inhibitor of PAD4 with a drug, such as GSK484. 22 Thus, we established another GCM model (n = 10 rats) after two inductions on day 0 and day 7 to investigate the therapeutic effects of GSK484 on GCM ( Figure 5A). GSK484 was injected into each GCM rat for 2 weeks (from day 8 to day 20) daily before the rats were sacrificed on day 21. First, the intraperitoneal injection of GSK484 into GCM rats attenuated leukocyte and giant cell accumulation in the hearts and ameliorated inflammation on day 21 ( Figure 5B). As described previously, the infiltration of neutrophils in the myocardium of GCM increased significantly, and abundant NETosis was also observed in the myocardium of GCM, with a significant difference on day 21 after two sensitizations according to the IHC staining results ( Figures 4F and 5C). Strikingly, we treated GCM rats with this compound to inhibit PAD4 22 and observed that NETosis, neutrophil infiltration, and the inflammatory response in the myocardium of GCM were significantly inhibited ( Figures 5B and 5C). Taken together, NETosis participates in the pathogenesis of GCM, and inhibitors of NETosis could alleviate inflammation in GCM, which could be applied as new targets for treatment.
Given the role of NETosis in GCM, we next wondered whether this process could be targeted therapeutically to ameliorate GCM in clinical practice. Observations of human heart specimens highlighted the potential clinical importance of this question. According to the 2013 ESC Task Force, 46 we collected heart samples from patients with GCM and from patients with chronic heart failure with dilated cardiomyopathy (DCM) who had undergone HTx and compared them with those from healthy controls. The demographic data are presented (Supplementary material online, Table S4). Multifocal inflammatory infiltrates consisting of lymphocytes with multinucleated giant cells were observed in GCM ( Figure 5D). We found that MPO and H3CIT were significantly expressed at higher levels in patients with GCM than in patients with DCM and healthy controls ( Figure 5D). This result indicated that neutrophil infiltration was also detected and NETosis occurred in patients with GCM. Together, these data demonstrate the clinical relevance of NETosis to patients with GCM and may serve as a novel therapeutic target for clinical usage.

Establishment of IMC analysis for the heart tissue of patients with GCM
Based on the IMC protocols for liver cancer and breast cancer, 47,48 we established a 35-marker panel for heart tissue, including markers for endothelial cells, immune cells, and fibroblasts, the proliferation marker Ki-67, and the apoptotic marker cleaved caspase-3 (Supplementary material online, Figure S8 and Table S5).
We analyzed heart tissues from patients with GCM and revealed 21 typical structural markers, such as collagen I and aSMC, and immunological markers, such as CD4 and CD20 ( Figure 6A), among the 35 markers (Supplementary material online, Figure S8) iScience Article characterized by vimentin staining), could be detected ( Figure 6B). These features were consistent with those determined by H&E staining but were resolved in greater detail ( Figure 6B). We detected major cell types, including Cd45 + immune cells and aSMA + fibroblasts, and collagen I deposition in the extracellular matrix ( Figure 6C). Notably, we found two patterns of multinucleate giant cell formation ( Figure 6C): one kind was primarily composed of neutrophils, whereas the other was primarily composed of macrophages. Immune cell types within multinucleate giant cells included CD4 + T cells (2: CD4 + CD8a À CD20 À ), CD8 + T cells (3: CD20 À CD4 À CD8 + ), neutrophils (4: CD11b + CD68 À CD16 À ), and macrophages (5: CD68 + CD16 À CD11b À ) ( Figures 6D and 6E). B cells (1: CD20 + CD4 À CD8a À ) were not a component of multinucleate giant cells (Figure 6D). Interestingly, all multinucleate giant cells were characterized by the expression of vimentin ( Figure 6F). Vimentin is a citrullinated antigen in rheumatoid arthritis, which may underlie pathogenesis and be externalized during NETosis. 49 Together, NETs were involved in the inflammatory responses in GCM.

DISCUSSION
Giant cell myocarditis is a rare autoimmune disease with a high mortality rate and recurrence rate after heart transplantation, and it was reported that this disease is a type of T cell lymphocyte-mediated inflammation of the myocardium that typically affects young and middle-aged adults. 50 Although some clinical findings of this disease have helped to lead to a paradigm shift in the management of giant cell myocarditis resulting in an improvement in overall and transplant-free survival, such as combination immunosuppressive therapy, the immunological mechanism remains unknown. As scRNA-seq technology has been applied in the field of myocarditis, 19 we thought it could help to uncover the immunological mechanism of GCM. This study could be a reference for us to gain insight into the immunological environment of GCM.  iScience Article As we know, it is difficult to obtain fresh patient cardiac tissue samples, so we investigated them by the GCM model. 23 According to this study, we found some results that were consistent with those of a previous study, and we also obtained some new findings about the pathogenesis of GCM. First, it was reported that most of the infiltrating immune cells in GCM were macrophages and T cells. In our study, we found that macrophages and T cells were the top 3 cell populations. Second, we found that neutrophils were the top 2 cell populations in GCM, which was the first single-cell report that neutrophils did appear in the GCM. 8,23 Third, we further identified a subcluster of cells, such as Arg1 + macrophages and Th17 cells, that contributed to the pathogenesis of GCM. According to the immunological changes in GCM, innate immune cells received our attention, especially neutrophils. The recent RNA-seq study by Amancheria et al. was based on formalin-fixed paraffin-embedded human GCM heart samples, including 4 GCM, 3 lymphocytic myocarditis, 4 rejection, and 3 control samples. 51 They found that pathway enrichment analysis in GCM showed that upregulated pathways were enriched for neutrophil degranulation, multiple cytokine signaling pathways, and phagocytosis, but they did not pay much attention to the neutrophil roles in GCM. In addition, the study did not focus on the immune cell types and immunological formation of giant cells. In our study, we analyzed the immune cell types and found that Th17 cells were the main T cell types in GCM, and neutrophils may contribute to the process of GCM. In addition, querying the Drug Gene Interaction Database identified several novel biologically plausible therapeutic targets, including IL2RB, CD3G, CD3D, CD22, and CD20. These targets were mostly associated with the specific immunity but not innate immunity. We first came up with the inhibition of NETosis to treat GCM. Together, our study on cardiac immune cells of GCM was more comprehensive than the RNA-seq study. In addition, we compared macrophage clusters observed in giant cell myocarditis in our report with the macrophage clusters from cardiac sarcoidosis in the report by Liu et al. due to the lack of single-cell data of giant cell myocarditis in the report by Liu et al. 52 In the reports by Liu et al., spatial transcriptomics were reported in a myocarditis series that also included hearts from 3 patents with giant cell myocarditis, but not single-cell or single-nucleus data of giant cell myocarditis. We compared the results of the enrichment analysis (Supplementary material online, Figure S9). From the results, Macro_1 was proinflammatory. Macro_2 was similar to Mac_HLA, which was associated with antigen processing and presentation function. Macro_3 was similar to Mac_res, which is a resident macrophage that positively regulates the immune response. Macro_4 was similar to Mono, which was associated with the M2 process and recovery process. Macro_5 was similar to Mac_VCAN, which was related to the tumor necrosis factor signaling pathway, interferon signaling, and cytokine-mediated signaling pathway (Supplementary material online, Table S6). This result indicated that the macrophage cluster in GCM was similar to that in CS, but the detailed difference in single-cell levels between GCM and CS needs further investigation in the future.
GCM is a type of T cell-mediated autoimmunity, but it is not clear which kind of T cells contribute to the pathogenesis of GCM. Th17 cells have been reported in experimental autoimmune myocarditis and other autoimmune diseases. 19,53,54 Th17 cells have been seen as candidate therapeutics for the treatment of autoimmune diseases. 55 In the future, inhibitors that target Th17 cells could be investigated to determine whether they are helpful for patients with GCM.
Neutrophils are reported as the most abundant circulating leukocytes and the first line of defense against bacterial infections. 25 Meanwhile, neutrophils also contribute to tissue damage during various autoimmune and inflammatory diseases. The intimate but complex involvement of neutrophils in various diseases makes them exciting targets for therapeutic intervention. 17 It was also reported that NETosis may play a crucial were used to specify lymphocyte clusters. E CD68 (red), CD11b (green), and CD16 (blue) were used to specify myeloid cell clusters. F. CD68 (red), CD11b (green), and vimentin (cyan) were used to indicate the features of multinucleated giant cells. Arrows 1: B cells (CD20 + CD4 À CD8a À ), arrows 2: CD4 + T cells (CD4 + CD8a À CD20 À ), arrows 3: CD8 + T cells (CD20 À CD4 À CD8 + ), arrows 3: neutrophils (CD11b + CD68 À CD15 + ), arrows 4: neutrophils (CD11b + CD68 À CD16 À ), arrows 5: macrophages (CD68 + CD16 À CD11b À ), and arrows 6: giant cells positive for vimentin. See also Figure S8 and iScience Article role in inflammation and autoimmunity in a variety of autoimmune diseases, such as rheumatoid arthritis, systemic lupus erythematosus, and anti-neutrophil cytoplasmic antibody-associated vasculitis. NETosis can also activate other immune cells, such as B cells, antigen-presenting cells, and T cells. NETosis may be a central regulator of inflammation and autoimmunity, as well as a promising target for future therapeutics of inflammatory autoimmune diseases. 15 In our study, we first demonstrated neutrophil infiltration into the myocardium of GCM rats and we performed a quantitative analysis of neutrophil infiltration that showed a significant increase in neutrophil infiltration in GCM rats compared to normal rats ( Figures 4F  and 5C). Furthermore, we observed abundant NETosis in GCM myocardium ( Figure 5C), which implied that NETosis might play roles in the pathogenesis of GCM. Then, we found that GSK484, an inhibitor of NETosis via PAD4, could ameliorate the inflammatory response of GCM, indicating that inhibiting NETosis could treat GCM rats ( Figure 5B). In addition, both MPO and H3CIT expression also increased significantly in heart tissue samples of patients with GCM ( Figure 5D), suggesting that neutrophils infiltrated into the hearts of patients with GCM and NETosis was also observed in patients with GCM. This result indicated that GSK484 may be used to treat patients with GCM in the clinic and should be further investigated in clinical trials in the future. Neutrophils make up nucleated giant cells that do not account for normal giant cell phagocytosis activity, and the mechanism of giant cell formation and phagocytosis activity requires further investigation. In general, bone marrow chimeric Ldlr-deficient mice reconstituted with either wild-type or Pad4-deficient cells were used to delineate the major role of Pad4 as the main enzyme responsible for NE-Tosis development and potentially the main therapeutic target. 56 In the future, this approach should be applied to demonstrate that Pad4-dependent NETosis mechanisms are mainly involved in GCM pathophysiology in a rat model and that further strategies targeting PAD4 would be potentially relevant clinical options in patients with GCM. In addition, organoids represent a promising research model, helping us gain a more profound understanding of organs such as the intestine, brain, heart, and kidney. 57 Due to the application of organoids in cardiovascular disease, 58-60 organoid cell culture models would be of major interest to replace the proposed rat model in our group in the future. In addition, giant cells are thought to be involved in ''cleaning up'' and thus the appearance of neutrophils exhibiting phagocytic activity may not necessarily account for a pathogenic mechanism. The ''cleaning up'' role of neutrophils may be regarded as a pathogenic role of GCM. It was reported that neutrophils were quickly recruited to the ischemic region, where they initiate the inflammatory response, aiming at cleaning up dead cell debris. 61 However, excess accumulation and/or delayed removal of neutrophils is deleterious. Neutrophils can promote myocardial injury by releasing reactive oxygen species, granular components, and proinflammatory mediators. Thus, the two-alternative hypothesis of neutrophils contributing into the pathogenesis of GCM includes NETosis and the ''cleaning up'' role of neutrophils.
Intercellular communication network analysis indicated that macrophages could induce NETosis by macrophage-derived Il1b (Supplementary material online, Figure S6A). In addition, we found that neutrophils were able to recruit immune cells, specifically T cells, into the heart muscle via the CXCL-CXCR3 signaling pathway. Macrophages were also involved in expressing CXCL chemokines (Cxcl9, Cxcl10, and Cxcl11) to attract CXCR3 + T cells. 62 These CXCL chemokines were higher in macrophages in the GCM than in controls, suggesting that macrophages in the GCM induced T cells as well as neutrophils (Supplementary material online, Figure S7).
According to IMC analysis, all the multinucleated giant cells were positive for the expression of vimentin ( Figure 6F). This suggests that vimentin may be an autoantigen extended by NETosis, 63 which could induce many autoimmune diseases, such as lupus nephritis 64 and rheumatoid arthritis. 49 To back up this speculation, we will compare the difference in the level of vimentin autoantibodies in serum from GCM rats and GSK484-treated rats by ELISA or other methods in the future. (Supplementary material online, Table S2), but several genes with known functions were noted in DC cells, including Tap1 and Psmb9, which have a role in antigen processing and presentation. For NK cells, we found that all NK cell cluster numbers were decreased in GCM compared with normal controls, suggesting that the roles of NK cells in GCM were very small. This result was different from those in viral myocarditis.

Among other cells, DC cells and NK cells were low in number
Recently, some single-cell RNA sequencing studies on viral myocarditis have been reported. 65,66 The cell populations reported in viral myocarditis were roughly the same but also different. In viral myocarditis, we found that the major T cell subtypes were Th17 cells, CTLs, and Treg cells, which were similar to the T cell ll OPEN ACCESS iScience 26, 107162, July 21, 2023 iScience Article subtypes in GCM. However, the proportion of CTLs in GCM decreased, which was different from the increase in CTLs in viral myocarditis. For macrophages, we found that the transcriptomes of myeloid cells were mainly of the M2 phenotype in viral myocarditis, while the transcriptomes of macrophages were mainly of the M1 phenotype in GCM. Neutrophils participate in the pathogenic inflammatory and cardiac fibrosis process in viral myocarditis, with Il1b being the major driver of this process. Consistently, Il1b was also upregulated in neutrophils in GCM. This result indicated that myocarditis (viral myocarditis and giant cell myocarditis) is increasingly being considered an Il1b-mediated disease process so that neutralization of Il1b renders the mice more resistant to the development of myocarditis. 67 However, NETosis was not reported in these viral myocarditis single-cell data, 65,66 and the roles of NETosis need more investigation in viral myocarditis. Spatiotemporal transcriptomics has been applied in the study of viral myocarditis, 66 but our study on GCM is lacking. In addition, single-cell RNA sequencing data of viral myocarditis include more cell types, including immune cells and nonimmune cells, which could help us to gain insight into the interaction between immune systems and other cell types to reveal the immune injury mechanism.
In summary, these data described the changes in the immunological environment of cardiac tissues from a GCM model using scRNA-seq. Some cell clusters were identified in the pathogenesis of GCM, such as macrophage and T cell subpopulations. NETosis was identified in GCM at the single-cell level. Furthermore, an inhibitor of NETosis alleviated immune cell infiltration in the GCM, which suggested that NETosis could be a potential candidate target for the treatment of GCM in the clinic.

Limitations of the study
Although we have compared the differences in single-cell levels of macrophages between GCM and CS, we should emphasize two points. First, myocarditis and cardiac sarcoidosis are different diseases with different transcriptomic profiles, so it is not appropriate to compare macrophage clusters from cardiac sarcoidosis with those from our study. Last, these are also different species, as one is from the rat and the other from the human. Therefore, more investigations are needed to uncover the differences between the GCM rat model and the patient with GCM. In addition, the mechanism of antigen processing and presentation of antigen-presenting cells such as DCs in GCM needs more investigation, especially its mechanism of autoimmunity, which is also a limitation of this study. Many chemotactic pathways in neutrophils and macrophages were assumed to be generative and inconclusive, and additional investigations by inhibition of these chemotactic pathways are needed to confirm them, which is also a limitation.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Human heart sample collection
The use of human tissue in the present study was approved by the Human Ethics Committee of Fuwai Hospital, Chinese Academy of Medical Sciences. Written informed consent was obtained from each patient. Human heart samples were collected from patients who had undergone heart transplantation (HTx) in the operating room. The patients were divided into 2 groups: the GCM group (n = 3) and the chronic heart failure group (dilated cardiomyopathy [DCM] without myocarditis, n = 9). Healthy heart samples (n = 5) were obtained from brain-dead donors with a normal circulatory supply who were not suitable for transplantation due to technical or noncardiac reasons, such as body weight mismatch, according to the guidelines of China Transplant Services. All heart samples were obtained after fixation and fresh/frozen.

Echocardiography and electrocardiogram (ECG) examination
Echocardiography was performed 3 days before the first immunization and 1 day before sacrifice. All rats were anesthetized with an intraperitoneal injection of ready-to-use anesthetic (isoflurane, 0.2 mL/10 g), and then the chest hair was removed. Each rat was placed and fixed in a dorsal position on a heated pad (37 C). Ultrasound gel was applied to the chest to place the electrodes. During the examination, the isoflurane concentration was reduced to the minimal amount (1-2%) required to achieve constant and comparable heart rates. M-mode images were acquired from the parasternal long axis and parasternal short axis to evaluate the morphology, including end-systolic and end-diastolic ventricular inner diameter of the left ventricle (LV), and cardiac functions, such as the ejection fraction and fractional shortening of the LV. Images were analyzed using the dedicated software package VevoLAB (Version 1.7.2).

Collection of heart tissues and histology
Surgically excised hearts were collected in Dulbecco's Modified Eagle's Medium (DMEM, Gibco, 11965-092) containing 10% fetal bovine serum (FBS) on ice. In each group, an approximately 1 mm-thick cross-section of the myocardium was removed from the middle of the heart and fixed with 4% paraformaldehyde overnight. Tissues were processed into paraffin sections and stained with hematoxylin and eosin. The remaining heart tissues were prepared as single-cell suspensions. Following published data, 68 the heart tissue inflammatory condition was graded from 0 to 4: ''0'' means no inflammation; ''1'' means 1-5 distinct mononuclear inflammatory areas, with the involvement of 5% or less of the cross-sectional area of the heart; ''2'' means more than 5 distinct mononuclear inflammatory areas, or the involvement of over 5% but not over 20% of the cross-sectional area of the heart; ''3'' means profound mononuclear infiltration involving over 20% of the area, without necrosis; and ''4'' means diffuse inflammation with necrosis in the heart.

Preparation of the single cardiac cell suspension
Following the histological examination, we pooled 5 hearts from each group to isolate single cells. The remaining heart tissues were cut into small pieces in prechilled phosphate-buffered saline (PBS) and digested with 400 U/ml collagenase type II (Worthington, 43J14367B) in a 37 C water bath with mild shaking. The

OPEN ACCESS
Quantification of gene expression in single cells, determination of the major cell types, identification of marker genes, enrichment pathway analysis and cell-cell interaction (CCI) analysis Single-cell RNA-seq data were quantified using the 10x software package CellRanger 2.2.0 to map the data to the Rat (Rnor6) reference genome. We next removed the cells that (1) had greater than 10% expression originating from mitochondrial genes, (2) expressed less than 200 genes or greater than 5,000 genes, and (3) had less than 1,000 unique molecular identifier (UMI) counts or greater than 40,000 UMI counts to filter out low-quality cells and doublets (Supplementary material online, Table S1). Clustering was performed using a deep embedding algorithm for single-cell clustering (DESC) based on the variable genes identified with the filter_genes_dispersion function in Scanpy. 72,73 DESC is an unsupervised deep learning-based clustering method for single-cell data. DESC iteratively learns cluster-specific gene expression signatures and cluster assignments to improve clustering accuracy and remove batch effects. First, DESC pretrains an ll OPEN ACCESS iScience 26, 107162, July 21, 2023