Excessive formation of macrophage extracellular trap (MET) has been implicated in several autoimmune disease pathogeneses; however, its impact on type 1 diabetes (T1D) and related mechanisms remains enigmatic. We demonstrated the pivotal role of peptidyl arginine deiminase 4 (PAD4) in driving profuse MET formation and macrophage M1 polarization in intestinal inflammation in NOD mice. Genetic knockout of PAD4 or adoptive transfer of METs altered the proportion of proinflammatory T cells in the intestine, subsequently influencing their migration to the pancreas. Combining RNA sequencing and CUT&Tag analysis, we found activated PAD4 transcriptionally regulated CXCL10 expression. This study comprehensively investigated how excessive PAD4-mediated MET formation in the colon increases the aggravation of intestinal inflammation and proinflammatory T-cell migration and finally is involved in T1D progression, suggesting that inhibition of MET formation may be a potential therapeutic target in T1D.

Article Highlights
  • The role of macrophage extracellular trap (MET) formation in the pathology of type 1 diabetes (T1D) remains unclear.

  • We aimed to investigate the impact of MET formation on intestinal inflammation and its subsequent effect on the pathogenesis of T1D in NOD mice.

  • Activated peptidyl arginine deiminase 4 (PAD4) facilitates MET formation and transcriptionally regulates CXCL10 expression, leading to accelerated T1D through the activation and migration of proinflammatory T cells via the gut/pancreas axis.

  • Findings suggest that targeting PAD4-mediated MET formation could emerge as a potential therapeutic strategy.

Mounting evidence indicates that increased intestinal inflammation and permeability are critical contributors to the onset of type 1 diabetes (T1D). Gut dysbiosis and barrier dysfunction were reported in susceptible individuals (1) and animal models of T1D, such as BB rats (2) and NOD mice (3), even before hyperglycemic symptoms and autoantibody positivity appeared. Disruption in innate immune cells, particularly macrophages, is known to induce intestinal inflammation (4). Originating from monocytes, macrophages are indispensable in intestinal tissue remodeling, repair, inflammation, and pathogen defense (5). Recent studies have demonstrated that macrophages stimulated by bacteria and other irritants can release macrophage extracellular traps (METs), which form a net-like structure predominantly composed of citrullinated histone H3 (H3cit) and capture and eliminate pathogens, representing a unique form of cell death distinct from apoptosis and necrosis (6). METs have been implicated in the pathology of various inflammatory diseases, including rheumatoid arthritis (7), acute kidney injury (8), and atherosclerosis (9). However, their specific role in the pathogenesis of intestinal inflammation leading to T1D has not been elucidated.

Peptidyl arginine deaminase 4 (PAD4), a crucial enzyme for extracellular trap (ET) formation (10–12), is remarkably increased in patients with T1D (13) and animal models (14). Our previous studies have demonstrated that oral administration of Cl-amidine (15), an irreversible inhibitor of PAD4, or use of PAD4 genetic knockout (KO) (14) can ameliorate intestinal inflammation and effectively prevent T1D development by suppressing neutrophil extracellular trap (NET) formation. However, PAD4 is also expressed in various myeloid lineage cells, in particular, monocyte and macrophages (16). Research has shown that PAD4 can promote the polarization of macrophages toward the proinflammatory M1 phenotype (17), which is more prone to MET formation (18). However, the direct role of PAD4 in mediating MET formation remains to be further investigated.

We used PAD4−/− NOD mice and integrated colon and pancreatic RNA sequencing (RNA-seq) with both predictive and validation approaches for immune cell infiltration to investigate the role of PAD4 in mediating MET formation and its subsequent involvement in the pathogenesis of T1D in vivo. We used RNA-seq and CUT&Tag methods to systematically explore the molecular mechanisms underlying PAD4-mediated MET formation in vitro and found that activated PAD4 transcriptionally regulated CXCL10 expression. Finally, we evaluated the effects of inhibition of the CXCL10/CXCR3 axis on disease progression in NOD mice. Our data support a novel role for PAD4-mediated MET formation in intestinal inflammation in the pathology of T1D.

Animals and AMG487 Treatment

Female C57/B6 mice and PAD4+/+ and PAD4−/− mice with an NOD/LtJ background were obtained from GemPharmatech Co., Ltd. (Nanjing, China). At age 5 weeks, NOD mice were administered 5 mg/kg AMG487 orally and normal saline daily for 20 weeks (n = 15). All experimental procedures were approved by the local ethics committee of China Pharmaceutical University (Nanjing, China).

Peritoneal Macrophage and Bone Marrow–Derived Macrophage Isolation and Polarization and MET Assay

Precooled PBS was injected into the mouse peritoneal cavity and rubbed for 5 min, followed by aspiration of PBS. The resulting cells were plated in 12-well plates (1 × 106 cells) containing DMEM supplemented with 10% FBS (HyClone Laboratories, Logan, UT), and the adherent cells were identified as peritoneal macrophages (PMs). Mouse thigh bone marrow was cultured in DMEM with 10% FBS and 20 ng/mL M-CSF (PeproTech, Cranbury, NJ) for 7 days to generate bone marrow–derived macrophages (BMDMs). To induce M1 macrophages, 20 ng/mL interferon-γ (IFN-γ) and 100 ng/mL lipopolysaccharide were added, followed by incubation for 24 h. For M2 macrophages, 20 ng/mL interleukin-4 (IL-4) was added, followed by incubation for 24 h. PMs were cultured in normal (RPMI 1640) or high glucose (DMEM) or pretreated with Cl-amidine (Selleck Chemicals, Houston, TX) for 1 h and then stimulated with 10 or 20 μmol/L ionomycin (Selleck Chemicals) for 0–24 h. The cells were stained with SYTOX green (KeyGen Biotech Co. Ltd., Nanjing, China), and fluorescence intensity was quantified using Image J software. To obtain METs, the suspension was centrifuged at 1,500g for 5 min, and the supernatant was collected as the MET sample.

Adoptive Transfer of METs Into the Peritoneal Cavity of NOD Mice

PMs from NOD and PAD4−/− NOD mice were stimulated with or without 20 μmol/L ionomycin. After washing with PBS twice, each group (2 × 106 cells) was suspended in 200 μL PBS and adoptively transferred into the peritoneal cavity of NOD or PAD4−/− NOD mice. After 24 h, mesenteric lymph nodes (MLNs), pancreatic lymph nodes (PLNs), and spleens were extracted and analyzed by flow cytometry.

Isolation and Coculture of MLNs With METs

PMs from NOD and PAD4−/− NOD mice were seeded into 12-well plates at a density of 1 × 106 cells per well and stimulated with 20 μmol/L ionomycin for 8 h. The control group was left unstimulated, and the groups were divided into wild type (WT), WT plus ionomycin, PAD4−/−, and PAD4−/− plus ionomycin. To collect macrophage supernatants and METs, MLNs were isolated from NOD mice and seeded at a density of 2 × 106 cells per well in 12-well plates containing either macrophage supernatants or METs of WT, WT plus ionomycin, PAD4−/−, and PAD4−/− plus ionomycin groups. After 12 h of culture, T-cell typing in MLNs was analyzed using flow cytometry.

Flow Cytometry

Single-cell suspensions were prepared from MLNs, PLNs, and spleens by spreading them into 24-well plates with 1 × 106 cells per well. The cells were then stimulated with Cell Activation Cocktail (BioLegend, San Diego, CA) for 4 h. Monoclonal antibodies against surface markers were used for staining and fixed and permeabilized with the Fixation/Intracellular Staining Perm Wash Kit (BioLegend), followed by staining for cytokines. Antibodies (BioLegend) used included BV510 anti-mouse CD3, FITC anti-mouse CD4, PerCP/Cyanine 5.5 anti-mouse CD8, PE anti-mouse α4β7, APC anti-mouse IFN-γ, and PE-Cyanine 7 anti-mouse IL-17. For PM analysis, cells were stained with FITC anti-mouse F4/80, PerCP/Cyanine 5.5 anti-mouse CD11b, and APC anti-mouse CD86.

Statistical Analysis

SPSS 22.0 software was used for statistical analysis. The statistical significance of differences between two or more samples was calculated using an unpaired two-tailed Student t test or one- or two-way ANOVA. All data are presented as mean ± SEM. P < 0.05 was used to indicate a statistically significant difference.

Data and Resource Availability

All data sets generated during the current study are available from the corresponding author on request.

PAD4 KO Decreases Colon Inflammation and Benefits Functional Pancreatic β-Cell Maintenance in NOD Mice

Our previous findings have revealed PAD4 KO markedly alleviates insulitis, leading to a 40% reduction in the incidence of diabetes when monitored for up to 40 weeks (14). We performed bulk RNA-seq on the colons and pancreases of 12-week-old PAD4+/+ NOD and PAD4−/− NOD mice at the initial stage of hyperglycemia (19) to detect T1D-dependent differentially expressed genes (DEGs) after PAD4 deletion. To our surprise, knocking out PAD4 did not directly decrease inflammation-related genes in the pancreas but did so in the colon. Colon sequencing results showed that inflammatory genes such as CXCL10 and IL-1β were significantly downregulated in PAD4−/− NOD samples compared with NOD control samples (Fig. 1A). Gene Ontology analysis showed that the downregulated genes mainly affected defense response and cellular response to IFN-β and CXCR3 chemokine receptor binding functions (Fig. 1B). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was performed on downregulated genes heavily involved in inflammation-related signaling pathways, such as tumor necrosis factor (TNF) signaling, cytoplasmic DNA sensing, and IL-17 signaling pathways (Fig. 1C). Consistently, mRNA expression levels of IL-1β, CXCL10, CXCR3, TNF-α, IFN-γ, and IL-17 were greatly reduced in the PAD4−/− NOD colon samples using quantitative RT-PCR validation (Fig. 1D), as were protein levels of TNF-α and IL-1β (Supplementary Fig. 1). These data suggest that PAD4 deletion in NOD mice reduced baseline colon inflammation. In addition, protein folding–related genes Pdia4, Hspa5, Hspa8, and Hsph1 and pancreatic β-cell regeneration genes Reg3b and Reg3g were significantly upregulated in the pancreases of the PAD4−/− group (Fig. 1E and H). These DEGs were primarily involved in responses to unfolded proteins, positive regulation of protein folding, and misfolded protein binding functions (Fig. 1F) and participated in protein processing in endoplasmic reticulum pathways (Fig. 1G). These data indicate that the correct folding of pancreatic proteins and islet regeneration ability in PAD4 KO mice were improved at least at the genetic level, which may benefit from the improved colon inflammation.

Figure 1

RNA-seq analysis and quantitative RT-PCR (qRT-PCR) validation of DEGs in colons and pancreases of NOD mice. A: Volcano plots depicting DEGs in colons of NOD mice compared with those of PAD4−/− mice. B and C: Gene Ontology (GO) (B) and KEGG (C) analyses of significantly downregulated genes in colon after PAD4 KO. D: Validation of DEGs in colon after PAD4 KO using qRT-PCR (n = 6). E: Volcano plots depicting DEGs in pancreases of NOD mice compared with those of PAD4−/− NOD mice. F and G: GO (F) and KEGG (G) analyses of significantly upregulated genes in pancreas after PAD4 KO. H: Validation of DEGs in pancreas after PAD4 KO using qRT-PCR (n = 6). All data presented as mean ± SEM. D and H: Statistical significance assessed by Student t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 1

RNA-seq analysis and quantitative RT-PCR (qRT-PCR) validation of DEGs in colons and pancreases of NOD mice. A: Volcano plots depicting DEGs in colons of NOD mice compared with those of PAD4−/− mice. B and C: Gene Ontology (GO) (B) and KEGG (C) analyses of significantly downregulated genes in colon after PAD4 KO. D: Validation of DEGs in colon after PAD4 KO using qRT-PCR (n = 6). E: Volcano plots depicting DEGs in pancreases of NOD mice compared with those of PAD4−/− NOD mice. F and G: GO (F) and KEGG (G) analyses of significantly upregulated genes in pancreas after PAD4 KO. H: Validation of DEGs in pancreas after PAD4 KO using qRT-PCR (n = 6). All data presented as mean ± SEM. D and H: Statistical significance assessed by Student t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Close modal

PAD4 KO Attenuates Proinflammatory T Cells in the Pancreas by Inhibiting Colonic MET Formation in NOD Mice

PAD4 is primarily expressed in immune cells, such as macrophages, neutrophils, monocytes, and eosinophils. Neutrophil- and PAD4-mediated NET infiltration in the colon of NOD mice is closely associated with T1D onset and pathogenesis (14). We found abundant macrophages and associated ETs (shown by colocalization of F4/80 and H3cit+ cells) rather than neutrophils in the colon of diabetic NOD mice (Fig. 2A), suggesting that METs may play a more critical role in the pathogenesis of T1D. Interestingly, lower infiltration of METs was detected in the pancreas and not at all in islets, both in nondiabetic and diabetic NOD mice (Supplementary Fig. 2). To identify the immune cell subsets in colon and pancreatic tissues from NOD and PAD4−/− NOD mice, we used the cell-type identification by estimating relative subsets of RNA transcript (CIBERSORT) analysis method. The M1-type macrophage abundance in the colons of PAD4−/− NOD mice was decreased when compared with that in the colons of NOD mice (Fig. 2B). Immunofluorescence analysis further confirmed the decrease in inflammatory M1-type macrophages and MET levels in the colon after PAD4 KO (Fig. 2C and D). Additionally, macrophage and M1-type macrophage proportions were found to be reduced in the MLNs of PAD4−/− NOD mice; these were considerably higher in NOD mice than in C57/B6 mice (Fig. 2E). To further validate the impact of PAD4 KO on the progression of T1D, we analyzed the differences in T-cell phenotypes between NOD and PAD4−/− NOD mice in the MLNs, spleens, and PLNs. Flow cytometry results revealed that PAD4 KO decreased the proportion of total CD3 T cells in the MLNs and Th1 (CD4+ IFN-γ+), Th17 (CD4+ IL-17+), and Tc1 (CD8+ IFN-γ+) cells in the MLNs and spleen (Fig. 2F and Supplementary Fig. 3). Moreover, the proportion of Tc1 and gut-derived (α4β7+) Tc1 cells in the PLNs was significantly reduced (Fig. 2G). In summary, the data demonstrate that PAD4 deficiency leads to a reduction in colonic M1 macrophage and MET formation, which reduces the frequency of proinflammatory T cells in the colon and pancreas, suggesting a decreased migration of Tc1 cells from the colon to the pancreas.

Figure 2

PAD4 KO inhibits levels of M1-type macrophages and METs in colons of NOD mice. A: MET formation levels in colons of 15-week-old NOD mice with T1D. B: CIBERSORT analysis predicting immune cell subsets within colons and pancreases of NOD and PAD4−/− NOD mice. C and D: Immunofluorescence was performed to detect levels of M1 macrophages (C) and MET formation (D) in colons of 12-week-old NOD and PAD4−/− NOD mice. E: Flow cytometry was performed to detect proportions of macrophages and M1 macrophages in MLNs of 12-week-old C57/B6, NOD, and PAD4−/− NOD mice (n = 3). F and G: Flow cytometry was performed to detect proportions of T, Th1, Th17, Tc1, and α4β7+ Tc1 cells in MLNs (F) and PLNs (G) of NOD and PAD4−/− NOD mice (n = 3). All data presented as mean ± SEM. EG: Statistical significance was assessed by one-way ANOVA (E) or Student t test (F and G). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 2

PAD4 KO inhibits levels of M1-type macrophages and METs in colons of NOD mice. A: MET formation levels in colons of 15-week-old NOD mice with T1D. B: CIBERSORT analysis predicting immune cell subsets within colons and pancreases of NOD and PAD4−/− NOD mice. C and D: Immunofluorescence was performed to detect levels of M1 macrophages (C) and MET formation (D) in colons of 12-week-old NOD and PAD4−/− NOD mice. E: Flow cytometry was performed to detect proportions of macrophages and M1 macrophages in MLNs of 12-week-old C57/B6, NOD, and PAD4−/− NOD mice (n = 3). F and G: Flow cytometry was performed to detect proportions of T, Th1, Th17, Tc1, and α4β7+ Tc1 cells in MLNs (F) and PLNs (G) of NOD and PAD4−/− NOD mice (n = 3). All data presented as mean ± SEM. EG: Statistical significance was assessed by one-way ANOVA (E) or Student t test (F and G). *P < 0.05, **P < 0.01, ***P < 0.001.

Close modal

PAD4-Mediated Polarization of PMs Toward M1 Phenotype and MET Formation

Heterogeneity exists among macrophages from different origins (20). First, we compared the expression levels of PAD4 in three commonly used macrophage models, PMs, BMDMs isolated from C57/B6 mice, and the murine macrophage cell line RAW264.7 (20), and found that PAD4 was highly expressed only in the PMs (Fig. 3A). To explore an effective MET stimulator, we applied ionomycin, an ET formation inducer (14), in PMs, and this effectively induced MET formation in PMs in a dose- and time-dependent manner, shown by increased trap structures (Fig. 3B) and increased protein expression of PAD4, H3cit, and inducible nitric oxide synthase (iNOS; the functional marker of the M1 phenotype [21]) (Fig. 3C). Furthermore, the expression levels of iNOS and PAD4 proteins in PMs from NOD mice (Fig. 3D), as well as MET formation (Fig. 3E and F), were significantly higher at baseline and after ionomycin stimulation than those in C57/B6 mice. In contrast, the expression levels of iNOS and H3cit in PMs from NOD mice were decreased, along with a significant reduction in MET formation, when PAD4 activity was genetically disrupted (Fig. 3D–F) or inhibited using Cl-amidine (Supplementary Fig. 4A and B).

Figure 3

PAD4 mediates M1 polarization of peritoneal macrophages and MET formation. A: PAD4 protein expression in PMs, BMDMs, and Raw264.7 (n = 3). B: Levels of MET formation in PMs after stimulation with 10 and 20 μmol/L ionomycin for 8 h (n = 3). C: Protein expression levels of iNOS, PAD4, and H3cit in PMs after stimulation with 10 and 20 μmol/L ionomycin over a period of 0–24 h (n = 3). D: Protein expression levels of iNOS and PAD4 in PMs derived from C57/B6, NOD, and PAD4−/− NOD mice (n = 3). E: Levels of MET formation in PMs from C57/B6, NOD, and PAD4−/− NOD mice after stimulation with 20 μmol/L ionomycin for 8 h (n = 4). F: Protein expression levels of PAD4 and H3cit in PMs from C57/B6, NOD, and PAD4−/− NOD mice after stimulation with 20 μmol/L ionomycin for 8 h (n = 3). G: Protein expression levels of iNOS, PAD4, and H3cit in M0, M1, and M2 polarized PMs after stimulation with 20 μmol/L ionomycin for 8 h (n = 3). H: Levels of MET formation in M0, M1, and M2 polarized PMs after stimulation with 20 μmol/L ionomycin for 8 h (n = 3–4). I: Levels of MET formation in normal and high-glucose environments with 10 and 20 μmol/L for 8 h (n = 3). J: Protein expression levels of PAD4 and H3cit in PMs after stimulation with 20 μmol/L ionomycin for 8 h under normal and high-glucose environments (n = 3). All data presented as mean ± SEM. B, E, H, and J: Statistical significance assessed by two- (B) or one-way ANOVA (E and H) or Student t test (J). *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 3

PAD4 mediates M1 polarization of peritoneal macrophages and MET formation. A: PAD4 protein expression in PMs, BMDMs, and Raw264.7 (n = 3). B: Levels of MET formation in PMs after stimulation with 10 and 20 μmol/L ionomycin for 8 h (n = 3). C: Protein expression levels of iNOS, PAD4, and H3cit in PMs after stimulation with 10 and 20 μmol/L ionomycin over a period of 0–24 h (n = 3). D: Protein expression levels of iNOS and PAD4 in PMs derived from C57/B6, NOD, and PAD4−/− NOD mice (n = 3). E: Levels of MET formation in PMs from C57/B6, NOD, and PAD4−/− NOD mice after stimulation with 20 μmol/L ionomycin for 8 h (n = 4). F: Protein expression levels of PAD4 and H3cit in PMs from C57/B6, NOD, and PAD4−/− NOD mice after stimulation with 20 μmol/L ionomycin for 8 h (n = 3). G: Protein expression levels of iNOS, PAD4, and H3cit in M0, M1, and M2 polarized PMs after stimulation with 20 μmol/L ionomycin for 8 h (n = 3). H: Levels of MET formation in M0, M1, and M2 polarized PMs after stimulation with 20 μmol/L ionomycin for 8 h (n = 3–4). I: Levels of MET formation in normal and high-glucose environments with 10 and 20 μmol/L for 8 h (n = 3). J: Protein expression levels of PAD4 and H3cit in PMs after stimulation with 20 μmol/L ionomycin for 8 h under normal and high-glucose environments (n = 3). All data presented as mean ± SEM. B, E, H, and J: Statistical significance assessed by two- (B) or one-way ANOVA (E and H) or Student t test (J). *P < 0.05, **P < 0.01, ***P < 0.001.

Close modal

To understand MET formation in activated macrophages, we polarized PMs (M0) isolated from C57/B6 mice toward M1 (high expression of iNOS and IL-1β) and M2 (high expression of Arg1) phenotypes (Supplementary Fig. 5). Under ionomycin stimulation, the expression of PAD4 and H3cit protein was upregulated in three macrophage phenotypes, and the M1 subpopulation displayed the highest fold of upregulation in MET formation (Fig. 3G and H). Hyperglycemia is a major characteristic of T1D, and the high glucose–treated PMs significantly increased the expression of PAD4 and MET formation (Fig. 3I and J). In summary, PMs from NOD mice had higher levels of PAD4 expression and were inclined toward M1 polarization, the cell population acting with the greatest capability on MET formation. The high-glucose environment after T1D onset potentially further exacerbated MET formation, which might be greatly resolved by knocking out PAD4 or inhibiting PAD4 activity.

Adoptive Transfer of METs Increases Gut-Derived Proinflammatory T Cells in the Pancreases of NOD Mice

To investigate the active role of METs in the polarization and migration of T cells, we examined the migration of gut-derived Tc1 and Th1 cells to the PLNs by adoptively transferring ionomycin-stimulated PMs from NOD or PAD4−/− NOD mice into NOD mice (Fig. 4A). Results revealed that Tc1 cells in MLNs were significantly increased in NOD mice that received ionomycin-treated NOD PMs (Fig. 4B). Gut-derived Th1 and Tc1 cells were also significantly increased in the spleen (Fig. 4C) and PLNs (Fig. 4D) when mice received ionomycin-treated NOD PMs. However, when ionomycin-treated PMs isolated from PAD4−/− mice were transferred into NOD recipients, their Th1 and Tc1 cell populations in the spleen and PLNs were comparable to those in untreated PMs from NOD mice (Fig. 4C and D). These phenomena were also observed when PAD4−/− NOD mice served as the recipients (Supplementary Fig. 6AC). To further clarify the effect of METs on T-cell function in vitro, we established a coculture system using PMs and T cells from MLNs in NOD or PAD4−/− NOD mice (Fig. 5A). We found that ionomycin-stimulated PMs promoted T-cell migration but lost this ability when PAD4 was depleted (Fig. 5B). Meanwhile, these stimulated PMs induced T-cell polarization into Th1 and Tc1 phenotypes, which could also be reversed by PAD4 KO (Fig. 5C). Our findings suggest that PAD4-dependent METs accumulating in the colons of NOD mice can activate gut-derived proinflammatory T cells and facilitate their migration to the spleen and pancreas.

Figure 4

Autoimmune response in NOD mice promoted by adoptive transfer of METs. A: Diagram of adoptive transfer of METs into NOD mice. BD: Flow cytometry was performed to investigate impact of adoptive transfer of METs on proportions of Tc1, α4β7+ Tc1, Th1, and α4β7+ Th1 cells in MLNs (B), spleens (C), and PLNs (D) of NOD mice (n = 3). All data presented as mean ± SEM. BD: Statistical significance assessed by one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001.

Figure 4

Autoimmune response in NOD mice promoted by adoptive transfer of METs. A: Diagram of adoptive transfer of METs into NOD mice. BD: Flow cytometry was performed to investigate impact of adoptive transfer of METs on proportions of Tc1, α4β7+ Tc1, Th1, and α4β7+ Th1 cells in MLNs (B), spleens (C), and PLNs (D) of NOD mice (n = 3). All data presented as mean ± SEM. BD: Statistical significance assessed by one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001.

Close modal
Figure 5

METs promote T-cells migration and induce Th1 and Tc1 polarization in vitro. A: Transwell assay was performed to detect ability of METs to promote T-cell migration in MLNs. B: Quantitative analysis of T-cell migration in MLNs (n = 3). C: Flow cytometry was performed to detect activation ability of METs on T cells in MLNs (n = 3). All data presented as mean ± SEM. B and C: Statistical significance assessed by one-way ANOVA. *P < 0.05, ***P < 0.001.

Figure 5

METs promote T-cells migration and induce Th1 and Tc1 polarization in vitro. A: Transwell assay was performed to detect ability of METs to promote T-cell migration in MLNs. B: Quantitative analysis of T-cell migration in MLNs (n = 3). C: Flow cytometry was performed to detect activation ability of METs on T cells in MLNs (n = 3). All data presented as mean ± SEM. B and C: Statistical significance assessed by one-way ANOVA. *P < 0.05, ***P < 0.001.

Close modal

PAD4 Directly Promotes CXCL10 Expression During MET Formation at the Transcriptomic Level

To elucidate the influence of PAD4 on MET formation at the transcriptomic level, we isolated PMs from NOD and PAD4−/− NOD mice and stimulated them with ionomycin, resulting in four experimental groups: WT, KO, WT plus ionomycin, and KO plus ionomycin. Transcriptomic differences among these groups were assessed via bulk RNA-seq, showing that 458 genes were downregulated after PAD4 depletion regardless of ionomycin treatment (marked as PAD4+-related genes), including CXCL10 and iNOS (Fig. 6A). The decreased expression of CXCL10 and iNOS was also validated using quantitative RT-PCR (Fig. 6B). Of all DEGs, CXCL10 was the only gene that overlapped in the following three comparisons: upregulated genes in WT plus ionomycin versus WT, downregulated genes in KO versus WT, and downregulated genes in KO plus ionomycin versus WT plus ionomycin (Fig. 6D). We measured the supernatant CXCL10 secretion in all four PMs using ELISA and found that ionomycin treatment significantly increased the secretion of CXCL10, which could be countered by PAD4 KO (Fig. 6C). To delineate the biologic significance and functional roles of PAD4+ genes, we performed KEGG analysis on all 39 genes overlapping between pairwise comparisons from the Venn diagram, which revealed their involvement in cytokine-cytokine receptor interaction, IL-17 signaling, and TNF signaling pathways (Fig. 6E).

Figure 6

PAD4-mediated transcriptional activation of CXCL10 expression during MET formation. A: Heat map showing gene expression in WT, KO, WT plus ionomycin, and KO plus ionomycin groups (n = 3). B: mRNA expression levels of CXCL10 and iNOS in four groups (n = 3). C: ELISA was performed to detect CXCL10 secretion levels in supernatant of four groups (n = 3). D: Venn diagram showing intersection of upregulated genes in WT plus ionomycin vs. WT, downregulated genes in KO vs. WT, and downregulated genes in KO plus ionomycin vs. WT + ion groups. E: KEGG analysis of DEGs in Venn diagram. F: CUT&Tag was performed to investigate enrichment levels of PAD4 and H3cit in CXCL10 TSS region. G: Motif enrichment analysis of PAD4 binding sequences related to transcription factors. All data presented as mean ± SEM. B and C: Statistical significance assessed by one-way ANOVA. *P < 0.05, **P < 0.01.

Figure 6

PAD4-mediated transcriptional activation of CXCL10 expression during MET formation. A: Heat map showing gene expression in WT, KO, WT plus ionomycin, and KO plus ionomycin groups (n = 3). B: mRNA expression levels of CXCL10 and iNOS in four groups (n = 3). C: ELISA was performed to detect CXCL10 secretion levels in supernatant of four groups (n = 3). D: Venn diagram showing intersection of upregulated genes in WT plus ionomycin vs. WT, downregulated genes in KO vs. WT, and downregulated genes in KO plus ionomycin vs. WT + ion groups. E: KEGG analysis of DEGs in Venn diagram. F: CUT&Tag was performed to investigate enrichment levels of PAD4 and H3cit in CXCL10 TSS region. G: Motif enrichment analysis of PAD4 binding sequences related to transcription factors. All data presented as mean ± SEM. B and C: Statistical significance assessed by one-way ANOVA. *P < 0.05, **P < 0.01.

Close modal

Next, we conducted CUT&Tag assays and used PAD4 and H3cit proteins to clarify how PAD4 regulates the expression of CXCL10 during MET formation at the epigenetic level. PAD4 and H3cit enrichment was observed at the transcription start site (TSS) of CXCL10, which was further elevated during ionomycin-induced MET formation (Fig. 6F). Recent research has demonstrated that PAD4-mediated H3cit promotes gene transcription (22). Motif enrichment analysis of PAD4 demonstrated concurrent enrichment of transcriptional activation factors SP1, SP2, and SP5 within its binding region (Fig. 6G). Subsequent analysis using the TFBIND tool (https://tfbind.hgc.jp/) (23) helped uncover multiple SP1 binding sites within the enriched region of the PAD4 and H3cit sequences proximal to the CXCL10 TSS (Fig. 6F and G), suggesting a potential synergistic relationship between PAD4 and SP1 in facilitating CXCL10 transcription. In conclusion, PAD4 directly promoted the transcription of proinflammatory chemokine CXCL10 during MET formation.

Inhibition of the CXCL10/CXCR3 Axis Ameliorates Disease Progression in NOD Mice

To determine whether CXCL10 accumulation during MET formation drives T1D progression, we first assessed the tissue distribution of CXCR3 (CXCL10 receptor), CXCL9, CXCL10, CXCL11 ligand, and PAD4 expression in NOD mice. Results showed that compared with other tissues, MLNs displayed higher CXCR3 and CXCL10 mRNA expression, and PAD4 also demonstrated relatively high expression (Fig. 7A). Serum CXCL10 concentrations in NOD mice gradually increased with age, peaking at 16 weeks, followed by a slight decrease through week 20 but remaining above baseline levels (Fig. 7B). Moreover, elevated mRNA expression of CXCR3 and CXCL10 was observed in the colons of diabetic NOD mice postonset (Fig. 7C). Based on these findings, we applied the CXCR3 inhibitor AMG487 in NOD mice by daily oral administration starting from age 5 weeks. Consequently, AMG487 administration postponed the onset of diabetes by 3 weeks and decreased the incidence of diabetes to some extent before 19 weeks in NOD mice (Supplementary Fig. 7A). Notably, by 15 weeks, a significant difference in blood glucose levels was evident between the two groups of mice (Supplementary Fig. 7B). Furthermore, NOD mice treated with AMG487 showed improved glucose tolerance, reduced islet inflammation, and enhanced islet function (Supplementary Fig. 7CE) compared with NOD mice receiving saline (vehicle). Additionally, colon inflammation in NOD mice treated with AMG487 was characterized by relatively intact colon villi and crypt structures, with no apparent inflammatory cell infiltration in the mucosal layer (Supplementary Fig. 8A). Lower mRNA expression levels of proinflammatory IFN-γ and IL-17 were detected in the colon and pancreatic tissues (Supplementary Fig. 8B), along with a reduced proportion of Th1 and Tc1 cells in the MLNs and spleen after treatment, consistent with levels in spleen tissue (Supplementary Fig. 8CE). Because CXCR3 is primarily expressed in T cells, we then cocultured T cells with PMs or METs, with or without AMG487, and examined T-cell migration in vitro under chemotactic signals. Results showed that the chemotactic ability of METs on T cells was greatly impaired when CXCR3 was inhibited (Supplementary Fig. 8F). In summary, inhibition of MET-induced CXCL10/CXCR3 signaling, to a certain extent, alleviated intestinal inflammation, as well as proinflammatory T cells in peripheral tissue, resulting in the improvement of insulitis in NOD mice.

Figure 7

Expression of CXCR3/CXCL10 axis in progression of T1D in NOD mice. A: mRNA expression levels of CXCL10/CXCR3-related chemokines and PAD4 in various tissues (n = 6). B: Quantification of CXCL10 levels in serum of NOD mice at 8, 12, 16, and 20 weeks (n = 10). C: mRNA expression levels of CXCL10 and CXCR3 in colons of NOD mice before and after onset of T1D (n = 4–5; nondiabetic 12–15 weeks; diabetic 15–22 weeks). All data presented as mean ± SEM. AC: Statistical significance assessed by one-way ANOVA (A and B) or Student t test (C). *P < 0.05, **P < 0.01.

Figure 7

Expression of CXCR3/CXCL10 axis in progression of T1D in NOD mice. A: mRNA expression levels of CXCL10/CXCR3-related chemokines and PAD4 in various tissues (n = 6). B: Quantification of CXCL10 levels in serum of NOD mice at 8, 12, 16, and 20 weeks (n = 10). C: mRNA expression levels of CXCL10 and CXCR3 in colons of NOD mice before and after onset of T1D (n = 4–5; nondiabetic 12–15 weeks; diabetic 15–22 weeks). All data presented as mean ± SEM. AC: Statistical significance assessed by one-way ANOVA (A and B) or Student t test (C). *P < 0.05, **P < 0.01.

Close modal

This study demonstrates the crucial role of PAD4 in driving profuse MET formation in the pathogenesis of intestinal inflammation in T1D. First, integrated colon and pancreatic RNA-seq results from WT and PAD4 KO mice revealed the impact of PAD4 on colon inflammatory responses not directly in the pancreas. Then, we observed that PMs in NOD mice exhibited an inclination toward an M1 phenotype, which significantly increased the likelihood of MET formation upon stimulation. MET formation was markedly elevated in the colons of diabetic NOD mice, which is instrumental in activating intestinal proinflammatory T cells and facilitating their migration to the pancreas. Consequently, proinflammatory T-cell migration resulted in damage to pancreatic islet cells and the onset of T1D.

Increasing evidence suggests intestinal inflammation and barrier dysfunction are critical triggering factors for T1D (24). Our laboratory previously demonstrated that excessive NET formation mediated by PAD4 as a result of intestinal leakage could activate intestinal T cells, which then migrated to the pancreas through the gut/pancreas axis, causing islet inflammation and T1D disease (14). In this study, two intriguing experimental findings brought macrophages to our attention. First, immunofluorescence analysis revealed a substantial infiltration of macrophages and METs in the colons of NOD mice after T1D onset, significantly higher even than that of neutrophils and NETs. Second, CIBERSORT analysis of colon transcriptome data showed a significant decrease in M1-type macrophage abundance in the colons of PAD4−/− NOD mice compared with NOD mice. These results suggest that macrophages and MET formation may play an important role in T1D-associated intestinal inflammation.

Tissue-resident macrophages are generally involved in fighting invading pathogens and repairing intestinal inflammation (25). It is imperative to note that this repair mechanism involves a temporal transition from the antimicrobial M1 phenotype to the tissue-reparative M2 phenotype, a process that spans several days. During this period, PMs act as a critical substitute in intestinal tissue repair (26). Proteomic analysis of macrophages from different mouse tissues showed significant heterogeneity in PAD4 expression detected only in PMs (20). Consistently, our study detected extremely high PAD4 expression in the PMs of mice, with notably elevated levels observed in the T1D model NOD mice. METs, produced by a unique cell death program (called METosis) in macrophages, trigger inflammatory cascades and cause tissue damage (18,27). Ionomycin, a classical Ca2+-dependent PAD4 agonist (28), has been deemed suitable for inducing MET formation in vitro. This induction elicited an upregulation of PAD4 expression, concomitant with macrophage polarization toward the M1 subtype. Upon PAD4 KO, the net-like structure of METs completely disappeared, and M1 macrophage polarization was reduced, indicating that PAD4 is critical for MET formation and also related to M1 transition. Benjamin S et al. (18) found that the METosis level in human monocyte–derived macrophages differed depending on their polarization state, where M1 macrophages were more susceptible to stimulation-induced MET formation. These results, together with our findings, indicate that proinflammatory M1 macrophages may be the intermediary in MET formation. In addition, we found that high-glucose environments in vitro could further promote MET formation in a PAD4-dependent manner, suggesting high glucose could be the promotor of MET formation in patients with diabetes.

The infiltration of Tc1 into the pancreas is the direct cause of T1D onset by destruction pancreatic β-cells (29). Sodré et al. (19) applied the PAD4 inhibitor BB-Cl-amidine in NOD mice and found significantly decreased neutrophil citrullination, Tc1 cell–mediated pancreatic inflammation, and slower T1D progression. The contribution of PAD4-mediated METs has not yet been well examined. Our investigation revealed that the deletion of PAD4 significantly reduced MET levels in the colon, as well as the proportions of M1 macrophages and T cells in MLNs. The heterodimeric integrin α4β7 expressed on T cells is a biologic marker for gut-derived T cells (30). Interestingly, PAD4 KO reduced the number of Tc1 cells in the PLNs, particularly α4β7+ Tc1 cells, indicating the migration of gut-derived T cells was suppressed. Additionally, our coculture experiments and adoptive transfer experiments directly demonstrated that METs could stimulate T-cell differentiation toward Th1 and Tc1, enhance their migration, and subsequently increase gut-derived T cells in the pancreas.

Generally, cytokines derived from macrophages potentiate T-cell activation (31). CXCL10, produced by proinflammatory macrophages, binds to the homologous receptor CXCR3 on T cells, inducing their chemotaxis and activation (32,33). Our results revealed that macrophages produced significantly proinflammatory cytokines such as CXCL10 during METosis. Moreover, bulk RNA-seq combined with CUT&Tag analysis in PMs showed that PAD4 and H3cit were specifically enriched in the promoter region of CXCL10 and positively correlated with its expression. PAD4 citrullinates histone arginine residues and opens up spatial constraints to promote transcription factor binding (22). Sp1 is an activating transcription factor that enhances RNA polymerase II binding in the gene promoter region (34). PAD4-related motif enrichment analysis indicated that Sp1 might cooperatively interact with PAD4 in the regulation of gene expression, particularly in the case of CXCL10. Serum CXCL10 levels are significantly elevated in patients with newly diagnosed T1D (35). In our rodent T1D model, we also found elevated serum CXCL10 levels after T1D onset and abnormally high CXCL10/CXCR3 expression in the colon and MLNs, along with colonic MET accumulation. Using AMG487, a CXCR3 antagonist, we found that T1D progression could be mitigated to some extent by inhibition of the CXCL10/CXCR3 axis, although little long-term effect was demonstrated, suggesting that early intervention, such as inhibition of MET formation using the PAD4 inhibitor (15), may play a better role in preventing diabetes development.

Some limitations should be acknowledged. The adoptive transfer of METs into NOD mice primarily focused on changes in proinflammatory T cells. Future research should include long-term transfer experiments to comprehensively assess its impact on pancreatic inflammation and T1D progression. Additionally, clinical samples are required to confirm the presence of METs in the pancreases or colons of patients with T1D. In conclusion, the current study demonstrates the pivotal role of PAD4 in governing MET formation. Using a T1D animal model, we revealed how PAD4-mediated METs in the colon increase the aggravation of intestinal inflammation and proinflammatory T-cell migration, ultimately exacerbating T1D progression. This work sheds light on the therapeutic potential of targeting PAD4 or METs in interventions in T1D.

This article contains supplementary material online at https://doi.org/10.2337/figshare.26515963.

Acknowledgments. The authors thank Jia Li and Yuting Ye for pathologic technical support.

Funding. J.W. has received grants from the National Natural Science Foundation of China (81673340 and 81973224) and the Priority Academic Program Development of Jiangsu Higher Education Institutions.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. Y.S., R.S., Y.W., and Z.Z. conducted the experiments, analyzed the data, and drafted the manuscript. Y.S., S.L., H.F., and J.W. critically revised the manuscript. Y.S. and J.W. conceptualized and devised the study. C.W. and Q.Y. conducted bioinformatic analysis of the sequencing data. H.F. and J.W. supervised the research. J.W. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

1.
Maffeis
C
,
Martina
A
,
Corradi
M
, et al
.
Association between intestinal permeability and faecal microbiota composition in Italian children with beta cell autoimmunity at risk for type 1 diabetes
.
Diabetes Metab Res Rev
2016
;
32
:
700
709
2.
Watts
T
,
Berti
I
,
Sapone
A
, et al
.
Role of the intestinal tight junction modulator zonulin in the pathogenesis of type I diabetes in BB diabetic-prone rats
.
Proc Natl Acad Sci U S A
2005
;
102
:
2916
2921
3.
Joesten
WC
,
Short
AH
,
Kennedy
MA
.
Spatial variations in gut permeability are linked to type 1 diabetes development in non-obese diabetic mice
.
BMJ Open Diabetes Res Care
2019
;
7
:
e000793
4.
Abraham
C
,
Medzhitov
R
.
Interactions between the host innate immune system and microbes in inflammatory bowel disease
.
Gastroenterology
2011
;
140
:
1729
1737
5.
Na
YR
,
Stakenborg
M
,
Seok
SH
,
Matteoli
G
.
Macrophages in intestinal inflammation and resolution: a potential therapeutic target in IBD
.
Nat Rev Gastroenterol Hepatol
2019
;
16
:
531
543
6.
Rasmussen
KH
,
Hawkins
CL
.
Role of macrophage extracellular traps in innate immunity and inflammatory disease
.
Biochem Soc Trans
2022
;
50
:
21
32
7.
El Shikh
MEM
,
El Sayed
R
,
Nerviani
A
, et al
.
Extracellular traps and PAD4 released by macrophages induce citrullination and auto-antibody production in autoimmune arthritis
.
J Autoimmun
2019
;
105
:
102297
8.
Okubo
K
,
Kurosawa
M
,
Kamiya
M
, et al
.
Macrophage extracellular trap formation promoted by platelet activation is a key mediator of rhabdomyolysis-induced acute kidney injury
.
Nat Med
2018
;
24
:
232
238
9.
Pertiwi
KR
,
de Boer
OJ
,
Mackaaij
C
, et al
.
Extracellular traps derived from macrophages, mast cells, eosinophils and neutrophils are generated in a time-dependent manner during atherothrombosis
.
J Pathol
2019
;
247
:
505
512
10.
Papayannopoulos
V
.
Neutrophil extracellular traps in immunity and disease
.
Nat Rev Immunol
2018
;
18
:
134
147
11.
Thiam
HR
,
Wong
SL
,
Qiu
R
, et al
.
NETosis proceeds by cytoskeleton and endomembrane disassembly and PAD4-mediated chromatin decondensation and nuclear envelope rupture
.
Proc Natl Acad Sci U S A
2020
;
117
:
7326
7337
12.
Hemmers
S
,
Teijaro
JR
,
Arandjelovic
S
,
Mowen
KA
.
PAD4-mediated neutrophil extracellular trap formation is not required for immunity against influenza infection
.
PLoS One
2011
;
6
:
e22043
13.
You
Q
,
He
DM
,
Shu
GF
, et al
.
Increased formation of neutrophil extracellular traps is associated with gut leakage in patients with type 1 but not type 2 diabetes
.
J Diabetes
2019
;
11
:
665
673
14.
You
Q
,
Shen
Y
,
Wu
Y
, et al
.
Neutrophil extracellular traps caused by gut leakage trigger the autoimmune response in nonobese diabetic mice
.
Front Immunol
2021
;
12
:
711423
15.
Shen
Y
,
You
Q
,
Wu
Y
,
Wu
J
.
Inhibition of PAD4-mediated NET formation by Cl-amidine prevents diabetes development in nonobese diabetic mice
.
Eur J Pharmacol
2022
;
916
:
174623
16.
Sabnis
RW
.
Novel peptidylarginine deiminase type 4 (PAD4) inhibitors
.
ACS Med Chem Lett
2022
;
13
:
1537
1538
17.
Cheng
Y
,
Si
Y
,
Wang
L
, et al
.
The regulation of macrophage polarization by hypoxia-PADI4 coordination in rheumatoid arthritis
.
Int Immunopharmacol
2021
;
99
:
107988
18.
Rayner
BS
,
Zhang
Y
,
Brown
BE
,
Reyes
L
,
Cogger
VC
,
Hawkins
CL
.
Role of hypochlorous acid (HOCl) and other inflammatory mediators in the induction of macrophage extracellular trap formation
.
Free Radic Biol Med
2018
;
129
:
25
34
19.
Sodré
FMC
,
Bissenova
S
,
Bruggeman
Y
, et al
.
Peptidylarginine deiminase inhibition prevents diabetes development in NOD mice
.
Diabetes
2021
;
70
:
516
528
20.
Qie
J
,
Liu
Y
,
Wang
Y
, et al
.
Integrated proteomic and transcriptomic landscape of macrophages in mouse tissues
.
Nat Commun
2022
;
13
:
7389
21.
Garaicoa-Pazmino
C
,
Fretwurst
T
,
Squarize
CH
, et al
.
Characterization of macrophage polarization in periodontal disease
.
J Clin Periodontol
2019
;
46
:
830
839
22.
Wang
Y
,
Lyu
Y
,
Tu
K
, et al
.
Histone citrullination by PADI4 is required for HIF-dependent transcriptional responses to hypoxia and tumor vascularization
.
Sci Adv
2021
;
7
:
eabe3771
23.
Yang
L
,
Shen
W-W
,
Shao
W
, et al
.
MANF ameliorates DSS-induced mouse colitis via restricting Ly6ChiCX3CR1int macrophage transformation and suppressing CHOP-BATF2 signaling pathway
.
Acta Pharmacol Sin
2023
;
44
:
1175
1190
24.
Sorini
C
,
Cosorich
I
,
Lo Conte
M
, et al
.
Loss of gut barrier integrity triggers activation of islet-reactive T cells and autoimmune diabetes
.
Proc Natl Acad Sci U S A
2019
;
116
:
15140
15149
25.
Mowat
AM
,
Scott
CL
,
Bain
CC
.
Barrier-tissue macrophages: functional adaptation to environmental challenges
.
Nat Med
2017
;
23
:
1258
1270
26.
Honda
M
,
Kadohisa
M
,
Yoshii
D
,
Komohara
Y
,
Hibi
T
.
Directly recruited GATA6+ peritoneal cavity macrophages contribute to the repair of intestinal serosal injury
.
Nat Commun
2021
;
12
:
7294
27.
Yin
K
,
Wang
D
,
Zhang
Y
, et al
.
Polystyrene microplastics promote liver inflammation by inducing the formation of macrophages extracellular traps
.
J Hazard Mater
2023
;
452
:
131236
28.
Neeli
I
,
Radic
M
.
Opposition between PKC isoforms regulates histone deimination and neutrophil extracellular chromatin release
.
Front Immunol
2013
;
4
:
38
29.
Zhou
L
,
He
X
,
Cai
P
, et al
.
Induced regulatory T cells suppress Tc1 cells through TGF-β signaling to ameliorate STZ-induced type 1 diabetes mellitus
.
Cell Mol Immunol
2021
;
18
:
698
710
30.
Xu
B
,
Cook
RE
,
Michie
SA
.
Alpha4beta7 integrin/MAdCAM-1 adhesion pathway is crucial for B cell migration into pancreatic lymph nodes in nonobese diabetic mice
.
J Autoimmun
2010
;
35
:
124
129
31.
Guerriero
JL
.
Macrophages: their untold story in T cell activation and function
.
Int Rev Cell Mol Biol
2019
;
342
:
73
93
32.
Zhang
X
,
Fan
L
,
Wu
J
, et al
.
Macrophage p38α promotes nutritional steatohepatitis through M1 polarization
.
J Hepatol
2019
;
71
:
163
174
33.
Karin
N
.
CXCR3 ligands in cancer and autoimmunity, chemoattraction of effector T cells, and beyond
.
Front Immunol
2020
;
11
:
976
34.
Revyakin
A
,
Zhang
Z
,
Coleman
RA
, et al
.
Transcription initiation by human RNA polymerase II visualized at single-molecule resolution
.
Genes Dev
2012
;
26
:
1691
1702
35.
Meyers
AJ
,
Shah
RR
,
Gottlieb
PA
,
Zipris
D
.
Altered Toll-like receptor signaling pathways in human type 1 diabetes
.
J Mol Med (Berl)
2010
;
88
:
1221
1231
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/journals/pages/license.