Efferocytic remodelling of pancreatic islet macrophages by limited β-cell death

Abstract

The primary driver of type I diabetes is the autoimmune T cells that destroy insulin-producing β-cells within the islets of Langerhans in the pancreas¹. Pancreatic islet macrophages have also been variably linked to disease onset and progression. As macrophage-mediated removal of dying cells through efferocytosis regulates tissue homeostasis and immune responses², here we investigated how efferocytosis by intra-islet macrophages influences the immune environment of pancreatic islets. Using a series of complementary omics-based and functional approaches, we identify a subset of anti-inflammatory intra-islet efferocytic macrophages (e-Mac) within the pancreas of mice and humans. When limited β-cell apoptosis is induced in vivo in wild-type C57BL/6 mice and diabetic-prone NOD mice, islet macrophages adopt this e-Mac phenotype without an apparent increase in the total numbers of intra-islet macrophages. Such limited β-cell apoptosis and increase in e-Mac numbers led to long-term suppression of autoimmune diabetes in NOD mice. This e-Mac phenotype could also be recapitulated ex vivo by co-culturing macrophages with apoptotic β-cells. Mechanistically, the e-Mac-enriched populations imparted an anergic-like state on CD4⁺ T cells ex vivo and promoted accumulation of such anergic-like CD4⁺ T cells in vivo within the islets. Analysing macrophage-T cell interactions within pancreatic islets using NicheNet and targeted experimental validation, we identify the IGF-1-IGF1R axis as a contributor to the anergic-like T cell phenotype in the islets. Collectively, these data advance a concept that efferocytosis-associated reprogramming of the islet macrophages and the subsequent influence on the adaptive immune response could be beneficial in modulating diabetic autoimmunity.

Extended Data Fig. 1 |. Efferocytosis by islet macrophages following STZ^1-low treatment in C57BL/6 mice.

a. Immunofluorescent confocal microscopy imaging showing confocal planes of the pancreatic islets of C57BL/6 mice 12 h after STZ^1-low or control vehicle treatment. Green – apoptotic cells (cleaved caspase-3, CC3), blue – nuclei (Hoechst). Scale bar 30 μm. b. Quantification of fraction of apoptotic cells (CC3⁺) relative to the total cell number in the islets (Hoechst⁺). n = 9 islets in both conditions. c. Confocal images of islets taken 12 h after STZ^1-low treatment showing X-Y, Y-Z, and X-Z views of efferocytic events. Red – insulin, white – macrophages (F4/80), green – apoptotic cells (CC3), blue – nuclei (Hoechst). The CC3/insulin double positive areas inside the macrophages are considered as efferocytic events. The top row of images corresponds to (Fig. 1b). Scale bar – 30 μm. d. Quantification of efferocytic events via Imaris software automatic counting (see Methods). n = 9(Vehicle), n = 8(STZ^1-low). e. Percentage of islet macrophages (Live, CD45⁺, F4/80⁺, CD11c⁺) relative to all islet cells based on flow cytometric analysis 10 days after treatment with STZ^1-low or vehicle (C57BL/6 mice). n = 8(Vehicle), n = 11(STZ^1-low) mice. Statistics - two-tailed unpaired t-test, data shown as mean ± SEM.

Extended Data Fig. 2 |. scRNASeq of islet leukocytes following limited β-cell death.

a. The UMAP plot showing scRNASeq analysis of CD45⁺ cells from the islets of C57BL/6 J mice 10 days after STZ^1-low or vehicle treatment. b. Phenotypic markers expressed by cells from (a). c. Relative fractions of cells identified on (a) and (b).

Extended Data Fig. 3 |. Islet e-Mac shows similarity with pathology-associated macrophages from other tissues.

a. UMAP plots show scRNASeq of islet macrophages (C57BL/6 mice) and two publicly available scRNASeq datasets: i) microglia in the brain of an Alzheimer’s disease mouse model (GSM7049631 (ref. 36)), and ii) macrophages in adipose tissue of mice fed a high-fat-diet (GSE128518 (ref. 37)). Expression of some marker genes are shown on the right. b. Venn diagram illustrates overlaps among gene signatures from macrophage subsets in (a).

Extended Data Fig. 4 |. Gal-3 upregulation in islet and peritoneal macrophages following apoptotic β-cell uptake ex vivo.

a. Isolation of islet macrophages using CD11c⁺ magnetic beads (MACS) (non-autoimmune C57BL/6 J mice). Flow cytometric analysis on dispersed islet cells before enrichment (left), and MACS-enriched Live, CD45⁺ F4/80⁺ CD11c⁺ macrophages (right). b. Schematic of the assay when primary islet macrophages co-cultured with live or apoptotic β-cells (UV-irradiated Min6 cells). c. Flow cytometric evaluation of e-Mac frequencies in the co-culture experiment (b). Quantification is shown on the right. n = 4(UT), n = 2(Live), n = 8(Apo) biological replicates over N = 4 independent experiments for “UT” and “Apo” and N = 2 for “Live”. Two-tailed unpaired t-test with Welch’s correction. d. Flow cytometric analysis showing peritoneal macrophages. Peritoneal lavage cells were collected from C57BL/6 mice, plated for 2 h, washed to remove unbound cells, and analysed by flow cytometry 24 h after the beginning of the incubation. e. Flow cytometric analysis showing engulfment of apoptotic β-cells by peritoneal macrophages. Apoptotic β-cells (Min6 cell line) were labelled with pHrodo red dye and co-cultured with pMacs for 2 h; unbound apoptotic cells were then washed away, and incubation continued. The macrophage-to-β-cell ratio was 1:3. Flow cytometry was done 16 h after beginning of the incubation. f. Flow cytometric analysis of peritoneal macrophages after co-culture with live or apoptotic β-cells (Min6). Two negative controls were used: cytochalasin D (CytoD, 1 μM) – inhibitor of cytoskeletal reorganization that prevents corpse uptake; and annexin V (20 μg/ml) that binds phosphatidylserine (PtdSer) and masks it from scavenger receptors on the phagocytic cells. g. Quantification of (f). n = 6 (No β-cell), n = 6 (Live β-cell), n = 6 (Apo β-cell, alone), n = 3 (Apo β-cell, +CytoD), n = 3 (Apo β-cell, +Ann V) biological replicates; N = 2. Statistics - two-tailed unpaired t-test. Data are representative of N = 3 independent experiments (a, b). The diagrams in a and b were created in BioRender. Ravichandran, K. (2025) https://BioRender.com/oc6ssn7; Ravichandran, K. (2025) https://BioRender.com/laier1m.

Extended Data Fig. 5 |. Both e-Mac and non-e-Mac islet macrophages can uptake apoptotic β-cells.

a. Representative confocal images of the pancreatic islets of NOD mice 12 h after STZ^1-low or control vehicle treatment. Green – apoptotic cells (cleaved caspase-3, CC3), blue – nuclei (Hoechst), red – Gal-3, white – F4/80, yellow – insulin. Scale bars: 20 μm (left, whole-islet view); 2 μm (right, zoomed-in view). b. Expression of Mer-TK by islet macrophages in Mertk^fl/fl/Cx3cr1^cre/wt (Mertk cKO) and Mertk^fl/fl/Cx3cr1^wt/wt, C57BL/6 mice. Data are representative of N = 3 independent experiments. The diagrams in a were created in BioRender. Ravichandran, K. (2025) https://BioRender.com/lwltf25.

Extended Data Fig. 6 |. Intra-islet macrophages are the primary cells engulfing apoptotic β-cells.

a. Representative confocal images of the pancreatic islets of NOD mice 12 h after STZ^1-low or control vehicle treatment. Green – apoptotic cells (cleaved caspase-3, CC3), blue – nuclei (Hoechst), red – CD11c, white – F4/80, yellow – insulin. Scale bars: 20 μm (left, whole-islet view); 2 μm (right, zoomed-in view). b. Quantification of the fraction of MФ and DC among all islet cells (top), and absolute number of each cell type per islet (bottom). c. Quantification of the fraction of MФ and DC containing apoptotic remnants (CC3⁺insulin⁺) among all islet cells (top), and absolute number of MФ and DC with internalized apoptotic material (bottom). Scale bars: 20 μm (left, whole-islet view); 2 μm (right, zoomed-in view). Statistics - paired t-test; N = 2. The diagram in a was created in BioRender. Ravichandran, K. (2025) https://BioRender.com/lwltf25.

Extended Data Fig. 7 |. Leukocytes from islets of NOD mice following limited β-cell death.

a. The UMAP plot showing CD45⁺ cells from the islets of NOD mice 10 days after inducing limited β-cell death with STZ^1-low treatment. b. Differentially expressed genes defining cell types from (a).

Extended Data Fig. 8 |. Efferocytosis induces an anti-inflammatory program in islet macrophages.

a. Heatmap showing differentially expressed genes defining macrophage subsets in the islets of NOD mice at 5.5 weeks of age (10 days after treatment). b. Overrepresentation pathway analysis based on the gene signatures of e-Mac, Mac-2, and Mac-3 subsets (hypergeometric test). c. Venn diagrams comparing gene signature of e-Mac with those of other macrophage subsets. Note that gene signatures of Mac-1 and e-Mac have overlap. d. Examples of islet infiltration scores based on hematoxylin and eosin (H&E) staining of pancreas sections of 30-week-old NOD female mice treated at 4 weeks with either STZ^1-low or vehicle. Scale bar – 20 μm. N = 2. e. Flow cytometric chart of CD45⁺ cells from pancreatic islets isolated from NOD.Rag1^−/− mice (aged 4–6 weeks; pooled from 5 mice). Representative of N = 4 independent experiments. f. Flow cytometric charts of experiment in which naive CD4⁺ T cells were activated by islet macrophages or by conventional dendritic cells (cDC) from spleens and lymph nodes. Islet macrophages from either STZ^1-low- or vehicle-treated NOD.Rag1^−/− mice were co-cultured with CD4⁺ T cells without adding cognate antigen for 96 h. When conventional dendritic cells were used as antigen-presenting cells, 0.5 μM BDC mimotope was added when indicated. Conventional dendritic cells were isolated from spleen and lymph nodes of NOD.Rag1^−/− mice. (See also Fig. 3b). The diagram in f was created in BioRender. Ravichandran, K. (2025) https://BioRender.com/mrggrwq.

Extended Data Fig. 9 |. CD4⁺ T cells acquire partial anergic-like phenotype after activation by efferocytosis-associated islet macrophages both ex vivo and in vivo.

a. Peritoneal macrophages isolated from NOD mouse using magnetic-activated cell sorting (MACS). CD11b^pos macrophages are enriched into Tim-4^hi large peritoneal macrophages (LPM, expressing low level of MHC-II) and Tim-4^low small peritoneal macrophages (SPM, MHC-II^hi). Representative of N = 3 independent experiments. b. Antigen presentation assay in which naive TCR transgenic BDC2.5 CD4⁺ T cells were co-cultured for 3 days with SPM or LPM at indicated concentrations of antigenic peptide; N = 2. c. Gene set enrichment analysis (GSEA) plot showing enrichment of gene signature of CD4⁺ T cells activated by “e-Mac enriched” macrophages interrogated against publicly available transcriptional dataset comparing anergic and naive CD4⁺ T cells (GSE143739). d. Summary plot of GSEA using CD4⁺ T cells gene signature induced by islet macrophages from STZ^1-low- or vehicle-treated mice (“e-Mac enriched” and “Vehicle” correspondingly). Differential expression comparisons between anergic cells and either naive or antigen-experienced CD4⁺ T cells were used as reference (GSE143739). The CD4⁺ T cells in the reference dataset were exposed to different levels of antigen. e. GSEA pathways comparing gene expression pathways upregulated in CD4⁺ T cells activated by islet macrophages from STZ^1-low- versus vehicle-treated mice. Gene Ontology MSigDB (GO:BP). f. Heatmap showing differentially expressed genes among CD4⁺ T cell subsets in the islets of NOD mice 10 days after the treatment with either STZ^1-low- or control vehicle. g. GSEA plot showing islet anergic-like CD4⁺ T cells gene signature interrogated against bulk transcriptional dataset (GSE143739) comparing anergic and naive cells. NES, normalized enrichment score. Statistics: weighted Kolmogorov-Smirnov test (c, e, g).

Extended Data Fig. 10 |. Macrophages from 12-week-old NOD mice and human islet leukocytes analysed by scRNASeq.

a. scRNASeq analysis (UMAP plots) showing macrophage subsets from pancreatic islets at 12 weeks of age (late pre-diabetic) from NOD mice treated with STZ^1-low or vehicle at 4 weeks of age (at early pre-diabetic stage). b. Markers of macrophage subsets from (a). c. Reanalysis of a publicly available dataset showing Igf1r expression across CD4⁺ T cell subsets from spleen and lymph node (GSE244278). (Wald statistics, DESeq2). Data are presented as mean values ± SEM, 3 biological replicates over N = 1 RNASeq experiment. d. Flow cytometric analysis of human donor pancreatic islet cells. Islet macrophages were identified as Live, CD45⁺ CX3CR1⁺ CD11c⁺ cells. Representative of N = 2. e. Human islet leukocytes analysed by scRNASeq, 3 individual donors. CD45⁺ cells were FACS-purified as shown on (d), left. Transcriptional data was integrated between the donors and subjected to clustering. f,g. Markers differentially expressed between the identified immune cell types in human pancreatic islets. The diagram in d was created in BioRender. Ravichandran, K. (2025) https://BioRender.com/ddu5zzj.

Fig. 1 |. Detection of intra-islet efferocytic macrophages in the mouse pancreas.

a, Representative 3D confocal microscopy images of whole islets from C57BL/6J mice. White, F4/80 (macrophages, Mac.); blue, Hoechst (nuclei); red, insulin. Scale bar, 30 μm. The brighter edges are due to imaging artefacts. b,c, Confocal images showing a macrophage engulfing an apoptotic β-cell (apo-β) 12 h after STZ^1-low treatment. Red, insulin (Ins); white, F4/80; green, CC3; blue, Hoechst. For b, scale bar, 3 μm. The CC3⁺insulin⁺ area within the F4/80⁺ cell denotes efferocytosis. 3D surface renderings are shown in b (bottom right) and c. d,e, The number of macrophages containing apoptotic β-cells per islet (d) and the total macrophages per islet (e) were manually counted from 3D z-stack confocal images. Eight islets were examined over two mice per condition (d); n = 8 (STZ^1-low) and 9 (vehicle) islets over 2 mice per condition (d, e). NS, not significant. f,g, Electron microscopy analysis of islets showing live β-cells (f) and apoptotic β-cell cargo in macrophages (g). Pseudo colours: green, live β-cells; blue, apoptotic β-cells; red, macrophages. For f and g, scale bars, 2 μm. h, Schematic of the scRNA-seq analysis of islet CD45⁺ cells FACS-purified from C57BL/6 mice at 6–8 weeks of age, 10 days after STZ^1-low or vehicle injection. i.p., intraperitoneal. i,j, Uniform manifold approximation and projection (UMAP) showing islet macrophage subsets from C57BL/6 mice identified by scRNA-seq (i), and expression of selected markers (differentially expressed genes, DEGs) upregulated in each subset (j). k, The relative subset fractions from i. l, The relative increase in the e-Mac subset fraction after limited β-cell death in vivo. Left, flow cytometry analysis of Gal-3⁺ islet macrophages 10 days after STZ^1-low or vehicle treatment. Right, quantification of e-Mac (CD9⁺, Gal-3⁺) frequencies. N = 2 experiments, n = 4 mice per condition. m, e-Mac frequencies in islet macrophages from Mertk^fl/flCx3cr1^cre mice 10 days after STZ^1-low injection. n, Quantification of m. n = 2 (vehicle, no Cre), n = 2 (vehicle, Mertk cKO), n = 9 (STZ, no Cre) and n = 5 (STZ, Mertk^fl/fl). N = 3 experiments (STZ^1-low, Cre⁺ and Cre⁻) and N = 2 experiments (vehicle, Cre⁺ and Cre⁻). Statistical analysis was performed using unpaired two-tailed t-tests. Data are mean ± s.e.m. The diagrams in h, l, and m were created in BioRender: Ravichandran, K. (2025) https://BioRender.com/kerdery; Ravichandran, K. (2025) https://BioRender.com/Ygattoy; Ravichandran, K. (2025) https://BioRender.com/0apcwb9; and https://BioRender.com/emk5kiz.

Fig. 2 |. Limited β-cell death affects islet macrophages in the NOD mouse model.

a, The fraction of e-Mac cells in NOD islets decreases during diabetogenesis, as assessed using flow cytometry (macrophages: live, CD45⁺F4/80⁺CD11c⁺; e-Mac: Gal-3⁺CD9⁺ macrophages). n = 3, 4 and 3 mice at 5, 9 and 12 weeks, respectively; N = 2 experiments. Data are mean ± s.e.m. One-phase decay curve. b, Schematic of the experiment: STZ^1-low was injected into NOD mice at 4 weeks; islets were collected 10 days later for scRNA-seq analysis. c,d, Flow cytometry analysis of islet immune cells from NOD mice at 4 weeks of age (c) and 5.5 weeks of age (d). In d, mice received STZ^1-low or vehicle at 4 weeks and were analysed 10 days later. e,f, Representative flow cytometry charts (e) and quantification (f) of e-Mac cells in the islets of STZ^1-low- or vehicle-treated NOD mice 10 days after the treatment. n = 5 (STZ) and n = 6 (vehicle) mice; N = 2 experiments. Statistical analysis was performed using unpaired two-tailed t-tests. Data are mean ± s.e.m. g, UMAP of islet macrophage subsets from NOD mouse scRNA-seq data. h, The relative fractions of subsets from g. i, The expression of markers delineating islet macrophage subsets. j, The diabetes incidence in NOD female mice treated at 4 weeks with STZ^1-low or vehicle showing that STZ^1-low protects against autoimmunity. n = 20 (STZ) and n = 17 mice (vehicle); N = 2 independent experiments. Statistical analysis was performed using the log-rank test (Mantel–Cox); P = 0.0004. k, The islet infiltration score was evaluated using haematoxylin and eosin (H&E) staining in 30-week-old NOD female mice that were treated at 4 weeks of age with STZ^1-low or vehicle (non-diabetic mice only). n = 30 (vehicle) and n = 32 (STZ) islets from 4 mice per condition; N = 2 experiments. Statistical analysis was performed using two-sided χ² test. l,m, Flow cytometry (l) and quantification (m) of islets from 30-week-old NOD mice treated at 4 weeks of age with STZ or vehicle. m, Quantification of immune cell infiltration into the islets: total leukocytes (CD45⁺), CD4⁺ T cells, CD8⁺ T cells and macrophages (Methods). n = 5 (vehicle) and n = 8 (STZ); N = 2 experiments. Statistical analysis was performed using unpaired t-tests. Data are mean ± s.e.m. The diagram in b was created in BioRender. Ravichandran, K. (2025) https://BioRender.com/jjo6f1x.

Fig. 3 |. Islet β-cell death affects CD4⁺ T cells in a mouse model of T1D.

a, Primary CD4⁺ T cells were co-cultured with islet macrophages isolated from NOD.Rag1^−/− mice 10 days after STZ^1-low or vehicle treatment. b, Quantification of the fraction of CFSE-diluted CD4⁺ T cells 96 h after co-culture. Data are mean ± s.d. The number of biological replicates is indicated on the graph. N = 3 experiments. cDC, conventional dendritic cell. c, RNA-seq analysis of CD4⁺ T cells activated ex vivo by islet macrophages from STZ^1-low- or vehicle-treated NOD.Rag1^−/− mice (PCA plot). n = 5 (STZ^1-low), n = 4 (vehicle) and n = 4 (naive). d, DEGs in CD4⁺ T cells activated by islet macrophages from STZ^1-low- and vehicle-treated mice. Statistical analysis was performed using Wald tests with Benjamini–Hochberg adjustment. e, Heat maps showing genes (fold change ≥ 2, P_adj < 0.05) in anergic-like T cells in both a previous study (GSE244278) and in our study, in which T cells were co-cultured with a e-Mac-enriched macrophage population. Genes in red are highlighted in d. f–h, scRNA-seq analysis of islet CD4⁺ T cells, 10 days after STZ^1-low treatment in NOD mice (5.5 weeks old). UMAP plot showing CD4⁺ T cell subsets (f), dot plot depicting phenotypic markers (g) and the relative subset fractions (h). T_reg, regulatory T cells. i, GSEA using DEGs upregulated in anergic-like and TCM subsets in islet CD4⁺ T cells. Reference dataset: GSE244278 (ref. 44). ES, enrichment score. j, Islet CD4⁺ T cells, 10 days after STZ^1-low treatment. n = 5 mice per condition. Data are mean ± s.e.m. N = 3 experiments. k,l, The relative fractions of islet macrophages from 12-week-old NOD mice treated at 4 weeks of age; two independent biological replicates (k). A representative replicate is shown (l). m–o, UMAP analysis of CD4⁺ T cells of 12-week-old NOD mice treated at 4 weeks of age (m), and quantification (n) and the relative subset fractions (o) across two independent biological replicates. Statistical analysis was performed using unpaired t-tests (b and j), Wald tests with Benjamini–Hochberg adjustment (d and e) and weighted Kolmogorov–Smirnov test (i). The diagrams in a and b were created using BioRender. Ravichandran, K. (2025) https://BioRender.com/jjo6f1x ; Ravichandran, K. (2025) https://BioRender.com/mrggrwq.

Fig. 4 |. Efferocytosis-induced IGF-1 links e-Mac and anergic-like CD4⁺ T cells.

a,b, NicheNet analysis of ligand–receptor interactions between intra-islet macrophages (sender cells) and CD4⁺ T cells (receiver cells) based on the scRNA-seq data of 5.5 week old NOD mice islets. Circos plot (a), and scaled expression of top-ranked ligands, their predicted target genes and the corresponding regulatory potential scores (b). c, Expression of Igf1r by intra-islet anergic-like CD4⁺ T cells, and Igf1 by e-Mac cells (scRNA-seq, 5.5-weeks-old NOD mice; see also Figs. 2g and 3f). d, Schematic of the potential connection between efferocytosis by islet macrophages and CD4⁺ T cells through IGF-1 signalling. e, IGF-1 secretion by peritoneal macrophages 12 h after exposure to apoptotic β-cells (Min6), as measured by ELISA. n = 5 biologically independent samples; N = 2 experiments. f, Intracellular staining of IGF-1 among intra-islet macrophages (CD45⁺F4/80⁺CD11c⁺), comparing the e-Mac (Gal-3^highCD9^high) and non-e-Mac (Gal-3^lowCD9^low) subsets. Mice 1, 3 and 5 received STZ^1-low treatment; mice 2 and 4 received vehicle treatment. Islets from n = 5 individual mice; N = 2 experiments. g, Flow cytometry analysis of IGF1R staining and markers of anergic-like cells expressed by intra-islet CD4⁺ T cells (10 days after STZ^1-low). Data are representative of n = 5 mice. h,i, NOD mice (4-weeks of age) were treated with recombinant IGF-1 or vehicle twice daily for 10 days before islets were collected and analysed. The representative flow cytometry plots (h) and quantification (i) show the increase in the frequency of anergic-like CD4⁺ T cells in the IGF-1-treated conditions. Islets from n = 6 individual mice per group; N = 2 experiments. Statistical analysis was performed using unpaired t-tests (e and i) and paired t-tests (f). s.c., subcutaneous. The diagrams in d and h were created using BioRender. Ravichandran, K. (2025) https://BioRender.com/y6jwwiv; Ravichandran, K. (2025) https://BioRender.com/rbl2ugb.

Fig. 5 |. Islet e-Mac cells are conserved between mouse and human.

a, Confocal imaging of pancreatic islets from a human donor. Red, Gal-3; yellow, CD68 (myeloid cells); blue, Hoechst (nuclei). Scale bars, 20 μm. b, Subsets of islet macrophages from three human donors captured by scRNA-seq. c, DEGs among three macrophage subsets from b. d, The expression of several DEGs defining the subsets of human islet macrophages. e, The relative expression of the e-Mac gene signature in mouse and human samples. The hypergeometric P = 3.3 × 10⁻¹¹⁵ shows the significance level of the overlap between human and mouse e-Mac gene signatures (orthologue genes). f, Flow cytometry analysis of e-Mac frequencies in a human donor based on CD9 and Gal-3 staining. The diagrams in a, e and f were created using BioRender. Ravichandran, K. (2025) https://BioRender.com/ddu5zzj; Ravichandran, K. (2025) https://BioRender.com/go6od0r.