Islet Macrophages Shift to a Reparative State following Pancreatic Beta-Cell Death and Are a Major Source of Islet Insulin-like Growth Factor-1

Summary

Macrophages play a dynamic role in tissue repair following injury. Here we found that following streptozotocin (STZ)-induced beta-cell death, mouse islet macrophages had increased Igf1 expression, decreased proinflammatory cytokine expression, and transcriptome changes consistent with macrophages undergoing efferocytosis and having an enhanced state of metabolism. Macrophages were the major, if not sole, contributors to islet insulin-like growth factor-1 (IGF-1) production. Adoptive transfer experiments showed that macrophages can maintain insulin secretion in vivo following beta-cell death with no effects on islet cell turnover. IGF-1 neutralization during STZ treatment decreased insulin secretion without affecting islet cell apoptosis or proliferation. Interestingly, high-fat diet (HFD) combined with STZ further skewed islet macrophages to a reparative state. Finally, islet macrophages from db/db mice also expressed decreased proinflammatory cytokines and increased Igf1 mRNA. These data have important implications for islet biology and pathology and show that islet macrophages preserve their reparative state following beta-cell death even during HFD feeding and severe hyperglycemia.

Graphical Abstract

Highlights

• •Macrophages are a major source of IGF-1 protein within mouse pancreatic islets
• •Post-beta-cell death islet macrophages shift to a reparative state
• •Beta-cell death causes macrophage transcriptome changes consistent with efferocytosis
• •This change can occur even in the presence of HFD feeding or severe hyperglycemia

Diabetology; Immunology; Specialized Functions of Cells

Introduction

Macrophages are versatile, plastic, innate immune cells essential to numerous biological processes. They participate in host defense, recognition of pathogens, initiation and resolution of inflammation, and maintenance of tissue homeostasis (Okabe and Medzhitov, 2016). Following tissue injury, macrophages are involved in the three main stages of tissue regeneration: inflammation, repair, and resolution. During the inflammatory phase, macrophages disrupt the basement membrane, secrete chemotactic factors to recruit inflammatory cells, and act as scavengers to phagocytose cellular debris. This is followed by a period of wound repair, wherein macrophages produce numerous growth factors, including insulin-like growth factor-1 (IGF-1), platelet-derived growth factors, and vascular endothelial growth factors to stimulate blood vessel development and proliferation of neighboring parenchymal and stromal cells. Transforming growth factor-β is also produced during this stage and activates tissue fibroblasts to facilitate wound closure and extracellular matrix deposition. In the last stage, macrophages assume an anti-inflammatory or pro-resolving phenotype characterized by anti-inflammatory cytokines (interleukin [IL]-10) and immune checkpoint inhibitor expression (PD-L2) (Vannella and Wynn, 2017).

In recent years, pancreatic islet macrophages have become increasingly well characterized in their resting state. Islet macrophages have an M1-like phenotype; they express Il1b and Tnf transcripts, express major histocompatibility complex (MHC) class II, present antigens to T cells, are negative for CD206/CD301, and are derived from definitive hematopoiesis (Calderon et al., 2015, Ferris et al., 2017). In the presence of aggregates of islet amyloid polypeptide (IAPP) (Masters et al., 2010, Westwell-Roper et al., 2016), or when exposed to toll-like receptor (TLR) ligands (Nackiewicz et al., 2014), the proinflammatory state of islet macrophages is enhanced, leading to IL-1 secretion that causes beta-cell dysfunction (Nackiewicz et al., 2014, Westwell-Roper et al., 2014). In contrast, in transgenic models of pancreatic beta-cell regeneration, islet macrophages can produce factors that support beta-cell replication (Brissova et al., 2014, Riley et al., 2015).

Pancreatic beta-cell death is a feature of both type 1 and 2 diabetes, contributing to inadequate insulin secretion and clinical hyperglycemia in both diseases. In type 1 diabetes, apoptotic and necrotic beta-cell death occurs. The immunological consequences of apoptotic beta-cell death are unexplored, whereas necrotic beta-cell death is thought to initiate or further enhance the activation of antigen-presenting cells in response to released beta-cell factors, causing T cell priming and activation and promoting autoimmunity (Wilcox et al., 2016). In contrast, in type 2 diabetes apoptotic beta-cell death is mainly associated with disease pathology (Halban et al., 2014).

Very little is known about the dynamic role of islet macrophages following beta-cell death. We tested the hypothesis that islet macrophages could be skewed to a tissue repair phenotype in response to beta-cell death, because apoptotic cells promote a tissue repair program in macrophages (Bosurgi et al., 2017) and other tissue macrophages have been shown to be locally programmed for silent clearance of apoptotic cells (Roberts et al., 2017). Here, we thoroughly characterized resident islet macrophage and recruited monocyte cell populations and gene signatures in response to streptozotocin (STZ)-induced cell death, in high-fat diet (HFD)-STZ-treated mice and db/db mice. Macrophages were the major source of IGF-1 protein within pancreatic islets, and transcriptome changes post STZ indicated an enhanced state of cellular metabolism and lysosome activity important in efferocytosis. Adoptive transfer of macrophages maintained circulating insulin levels following beta-cell death in vivo, and IGF-1 neutralization resulted in reduced second-phase glucose-stimulated insulin secretion post STZ in vivo.

Results

Islet Macrophages in STZ-Treated Mice Exhibit a Gene Set Shift Indicative of Enhanced Metabolism and Lysosome Activity and Secrete IGF-1

STZ is a toxin that specifically kills pancreatic beta-cells (Lenzen, 2008) and is commonly used to study islet inflammation. We recently used STZ to investigate the role of gp130 cytokine signaling in pancreatic alpha-cells in a model mimicking type 2 diabetes (STZ + HFD; Chow et al. [2014]). When given at a repeated low dose (<30 mg/kg) it induces mild effects on insulin secretion, glucose tolerance, and beta-cell mass (Chow et al., 2014). Here we used STZ to study the dynamic role of islet macrophages and monocytes following beta-cell death.

We initially analyzed islet macrophages and recruited monocytes at various time points post STZ treatment. Body weight and non-fasting blood glucose were unchanged up to 3 weeks post STZ (Figures S1A and S1B). Islet CD45⁺ cells, islet macrophages, and recruited monocytes were increased following STZ, with significant increases starting at 1 week and peaking at 2 weeks (Figures 1A–1D). By 3 weeks, no further increases were detected (Figures 1A–1D). Changes in macrophage numbers correlated with gene expression changes, which showed the most difference at 2 weeks post STZ (Figure 1E). At 2 weeks post STZ, Tnf mRNA expression was decreased and Il1rn, Igf1, and Tgfbi mRNA expression were increased in islet macrophages (Figure 1E). No differences in mRNA expression of these genes were detected in recruited monocytes (Figure S1C), and Igf1 was consistently detected only in islet macrophages (see also Figures 1E and S1C).

Figure 1.

Islet Macrophages in Mice Challenged with Multiple Low-Dose STZ Exhibit a Gene Shift toward Enhanced Metabolism and Lysosome Activity and Secrete IGF-1

C57BL/6J male mice were given multiple low-dose STZ (30 mg/kg, 5 times daily intraperitoneal [i.p.] injections) or acetate buffer as an injection control (referred to as “control”) at 16–20 weeks of age.

(A) Representative flow cytometry plots and gating strategy for cell sorting of dispersed islets from mice treated with multiple low-dose STZ (right panel) or control treatments (left panel). Islets shown here were harvested 2 weeks after the first i.p. injection.

(B–D) Fractions of (B) CD45+ cells, (C) islet macrophages, and (D) recruited monocytes.

(E) qPCR of islet macrophages. Relative mRNA expression levels of Il1b, Tnf, Il1rn, Igf1, Pdgfa, and Tgfbi expressed as fold over islet macrophage control.

(B–E) n = 3 for 0.5-, 2-, and 3-week treatments, and n = 5 for 1-week treatment. For each sorting sample (n), islets were pooled from 2 to 4 mice (average of 911 ± 198 islets). *p < 0.05, **p < 0.01, ***p < 0.001 STZ versus control, Student's t test.

(F–H) Transcriptome analysis of islet macrophages from mice treated with multiple low-dose STZ or control. (F) Minus over average (MA) plot of islet macrophage gene expression post STZ with the mean of gene counts on the x axis and Log₂ fold change of up- and downregulated genes on the y axis based on DEseq2 analysis. Significantly up- and downregulated genes are shown in red (Log₂ fold change >1 and FDR <0.05). (G) Enrichment map generated with Cytoscape of top-ranking clusters of genes enriched in STZ islet macrophages taken from GSEA analysis. Nodes represent gene sets, and edges represent mutual overlap. Highly redundant gene sets are grouped together as clusters. Gene sets involved in similar biological processes are shown in the same color. (H) Heatmap of GSEA results showing top 25 enriched genes in STZ (left panel) and top 25 enriched genes in control (right panel) islet macrophages (red, pink, light blue, dark blue corresponds to the range of expression values: high, moderate, low, lowest), n = 3. For each sorting sample (n), islets from 10 mice were pooled (average of 2,314 ± 200 islets).

(I) Representative IGF-1 enzyme-linked immune absorbent spot (ELISPOT) images from cells sorted from dispersed islets. Two weeks after the first injection of buffer or multiple low-dose STZ, 2,500 cells per group were sorted and analyzed by IGF-1 ELISPOT assay. On the right, outlined regions are enlarged 4 times.

(J) Quantification of (I) IGF-1 ELISPOT area in pixels; n = 6, ***p < 0.001, islet macrophages from control mice versus non-immune cells; ###p < 0.001 islet macrophages from STZ-treated mice versus non-immune cells, one-way ANOVA with Dunnett's multiple comparisons test. All data represent mean ± SEM.

To obtain a broader unbiased view of the changes in islet macrophages following STZ, we performed transcriptome analysis of isolated islet macrophages at 2 weeks post STZ. The obtained gene expression profile was consistent with phagocytic immune cells, which are known to express high levels of Cd74 (encoding part of MHC class II), Lyz2 (also known as LysM, associated with lysozyme), and Ctsd (cathepsin D gene, associated with lysozyme; highlighted in Figure 1F). According to DEseq2 analysis 128 genes were found to be upregulated and 45 were downregulated (thresholds of Log₂ fold change >1 and false discovery rate [FDR] <0.05). Gene set enrichment analysis (GSEA) highlighted gene sets from various pathways enriched in macrophages from STZ-treated mice (Table S1). There were no gene sets downregulated with an FDR q value <0.05. A Cytoscape enrichment map (Cline et al., 2007) of the main altered pathways highlighted multiple overlapping gene sets involved in increased oxidative phosphorylation, increased p450 metabolism, and increased lysosome and protein degradation activity (Figure 1G). Increased lipid metabolism and PPAR signaling genes were also enriched in macrophages from STZ-treated mice. There were no inflammatory gene sets enriched, and prototypical anti-inflammatory genes were not differentially expressed (Il10, Arg1, Mrc1, Jag1, Il4ra). A number of growth factors showed consistent upregulation in all three samples based on fragments per kilobase of transcript per million mapped reads (FPKM) values (Igf1, Ngf, Pdgfc, Tgfbi, Vegfb, Vgf). Interestingly, a heatmap of the top 25 enriched genes resulted in Igf1 having the highest score (3.47) among all enriched genes (Figure 1H). Cathepsin D (Ctsd) and matrix metalloproteinase-2 (Mmp2), both involved in the proteolysis of IGF-binding proteins (Mutgan et al., 2018), were also among the top 25 enriched genes. Taken together, the transcriptome of islet macrophages at 2 weeks post-beta-cell death indicates a heightened state of metabolism, increased lysosome activity, and a state characterized by strong induction of Igf1.

Isolated islet macrophages were confirmed to secrete IGF-1 protein ex vivo, whereas non-immune cells found in islets (CD45^-Ly6C⁺ and CD45^-Ly6C⁻ cells) did not (Figures 1I and 1J). Finally, to confirm that STZ was causing beta-cell apoptosis, numbers of TUNEL⁺Insulin⁺ cells were assessed at 1 and 2 weeks post STZ. STZ induced a 3-fold increase in beta-cell apoptosis at 2 weeks (Figures S1D and S1E).

Islet Macrophage Depletion Decreases Igf1 Expression following STZ-Induced Beta-Cell Death Ex Vivo

To isolate direct beta-cell STZ effects from indirect effects (e.g., elevated postprandial glucose) that may modulate islet macrophage gene expression in vivo, experiments were performed on isolated islets. Treatment of islets with increasing concentrations of STZ (0.25–4 mM) gave similar results to our in vivo studies, causing a decrease in Il1b mRNA expression, while significantly increasing Tgfbi and Pdgfa mRNA expression (Figure S2A). Whole-islet Il1rn and Igf1 mRNA expression also tended to increase. Beta-cell Ins1 and Ins2 expression were unchanged, whereas Pdx1 expression was increased by 4 mM STZ treatment (Figure S2A). To determine the contribution of islet macrophages to these gene expression changes, islet macrophages were depleted using islets isolated from CD11c-DTR mice treated with diphtheria toxin (DT). CD11C+ cells are exclusively macrophages in the islet (Ferris et al., 2017). Macrophage depletion following DT was confirmed by flow cytometry (Figure S2B). Depletion of islet macrophages completely abolished the STZ-induced increase in Igf1 mRNA expression (Figure 2A), whereas it had no effect on beta-cell mRNA levels (Ins1, Ins2, Pdx1). Our data also indicate that macrophages are the main source of Tgfbi transcript levels in islets (Figure 2A), which is interesting in the context of recent studies on TGFBI in islets (Han et al., 2014, Han et al., 2011). As previously shown (Ferris et al., 2017, Nackiewicz et al., 2014, Westwell-Roper et al., 2014), islet macrophages are also the main contributors to islet Il1b and Tnf expression. STZ-treated islets had increased numbers of TUNEL⁺Insulin⁺ cells (Figures 2B and 2C), with no effect on EdU⁺Insulin⁺ cells (Figures 2D and 2E). Depletion of islet macrophages reduced EdU⁺Insulin⁺ cells (Figure 2E).

Figure 2.

Islet Macrophage Depletion Decreases Igf1 Expression following STZ-induced Beta-Cell Death Ex Vivo

(A) qPCR of islets from male CD11c-DTR mice depleted (DT) or not (Ctrl) of islet macrophages followed by treatment with 4 mM STZ (STZ) or acetate buffer control (Control); n = 4–5, **p < 0.01, ***p < 0.001 DT versus Ctrl, ^##p < 0.01 STZ versus Control, two-way ANOVA with Bonferroni's multiple comparisons test.

(B) Representative sections of TUNEL+ CD11c-DTR islets depleted (+DT) or not (Ctrl) of islet macrophages followed by treatment with 4 mM STZ or acetate buffer control. Color scheme: DAPI, blue; insulin, green; TUNEL, red; colocalization of DAPI and TUNEL is shown in purple. Scale bar, 20 μm.

(C) Quantification of (B). Between 544 and 4,232 nuclei per section were counted. At least two sections from each sample (n) were counted; n = 4, *p < 0.05 STZ versus control, two-way ANOVA with Bonferroni's multiple comparisons test.

(D) Representative sections of EdU-treated CD11c-DTR islets depleted (+DT) or not (Ctrl) of islet macrophages followed by treatment with 4 mM STZ or acetate buffer control. Color scheme: DAPI, blue; insulin, green; EdU, red; colocalization of DAPI and EdU is shown in purple. Scale bar, 20 μm.

(E) Quantification of (D). Between 1,412 and 3,959 nuclei per section were counted. At least two sections from each sample (n) were quantified; n = 3–5, **p < 0.01 DT versus Ctrl, two-way ANOVA with Bonferroni's multiple comparisons test. All data represent mean ± SEM.

Finally, STZ did not increase Igf1 mRNA expression or protein secretion, or affect proliferation or apoptosis in bone marrow-derived macrophages (BMDMs; Figures S3A–S3D). In summary, these data support the conclusion that macrophages are the major source of IGF-1 within islets and that beta-cell death directly stimulates islet macrophage Igf1 mRNA expression.

Macrophages and IGF-1 Positively Regulate Insulin Levels in Mice following STZ

We next set out to further investigate the role of islet macrophages in vivo. Because islet macrophages already expressed elevated Igf1 at 1 week post STZ, we depleted phagocytic cells with clodronate-loaded liposomes during, and immediately following, STZ (see study design in Figure 3A). We recently used this protocol to deplete islet macrophages in mice (Nackiewicz et al., 2014, Westwell-Roper et al., 2014). There were no differences in body weight (Figure 3B) between treatment groups, whereas non-fasting blood glucose was significantly elevated at the end of treatment in STZ-treated groups compared with their controls (Figure 3C). Liposome-treated mice all tended to have decreased non-fasting insulin levels (Figure 3D). Insulin secretion was significantly decreased in islets obtained from control mice treated with clodronate-loaded liposomes and in islets from STZ-treated mice (Figure 3E), whereas islet insulin content was significantly reduced only in mice receiving STZ (Figure 3F). Clodronate-liposome-treated mice also tended to show worse glucose tolerance (Figures S4A and S4B). No differences in TUNEL⁺ islet cells or pHH3⁺ islet cells were observed between groups (Figures S4C and S4D). Effects of PBS-liposomes were consistent with a previously described effect of liposomes themselves on macrophage function (Ma et al., 2011, Pervin et al., 2016), or might have been due to macrophage depletion with PBS-liposomes (Weisser et al., 2011). These data show that islet macrophages help maintain beta-cell insulin secretion.

Figure 3.

Macrophages Positively Regulate Insulin Levels in Mice during Multiple Low-Dose STZ

(A) Experimental design of macrophage depletion study. Multiple low-dose STZ (30 mg/kg, 5 times daily i.p. injections) or control (acetate buffer, 5 times daily i.p. injections) treatments were administered to C57BL/6J males two weeks before sacrifice; 200 μL clodronate-loaded liposomes (CLOD-lip) or PBS-loaded liposomes (PBS-lip) were injected i.p. on days 3, 6, and 10 from the first dose of STZ/buffer.

(B) Body weights; n = 5 mice/control, control + PBS-lip, STZ + PBS-lip, STZ + CLOD-lip groups; n = 4 mice/STZ group, and n = 3 mice/control + CLOD-lip group.

(C) Non-fasting blood glucose measurements; n = 5 mice/control, control + PBS-lip, STZ + PBS-lip, STZ + CLOD-lip groups; n = 4 mice/STZ group, and n = 3 mice/control + CLOD-lip group, *p < 0.01 for control versus STZ, #p < 0.01 for PBS-lip versus STZ + PBS-lip, $p < 0.01 for CLOD-lip versus STZ + CLOD-lip, two-way ANOVA with Tukey's multiple comparisons test.

(D) Cardiac puncture non-fasting insulin levels at the day of sacrifice; n = 5 mice/control, control + PBS-lip, STZ + PBS-lip, STZ + CLOD-lip groups; n = 4 mice/STZ group, and n = 3 mice/control + CLOD-lip group.

(E) Islet insulin secretion in 2.8 mM glucose; n = 5 mice/control, control + PBS-lip, STZ + PBS-lip, STZ + CLOD-lip groups; n = 4 mice/STZ group, and n = 3 mice/control + CLOD-lip group. *p < 0.05, **p < 0.01, versus control, one-way ANOVA with Tukey's multiple comparisons test.

(F) Islet insulin content; n = 5 mice/control, control + PBS-lip, STZ + PBS-lip, STZ + CLOD-lip groups; n = 4 mice/STZ group, and n = 3 mice/control + CLOD-lip group. *p < 0.05, ***p < 0.0001, **** p < 0.0001 versus control, one-way ANOVA with Tukey's multiple comparisons test.

(G) Design of experiments involving adoptive transfer of bone-marrow derived macrophages (BMDMs). Multiple low-dose STZ (50 mg/kg, 5 times daily i.p. injections) or control (acetate buffer, 5 times daily i.p. injections) treatments were administered to C57BL/6J males 4 weeks before sacrifice. BMDMs that were starved of L929-conditioned medium were injected i.p. on days 3 and 7 from the first dose of STZ/buffer.

(H) Qtracker-positive cells in dispersed pancreas collected one day after second BMDM transfer; n = 4–5 mice/group. *p < 0.05, versus control Student's t test.

(I) Body weights; n = 14–15 mice/group from two separate experiments.

(J) Non-fasting blood glucose measurements; n = 14–15 mice/group from two separate experiments, ^∗p < 0.01 STZ + BMDM versus STZ + DPBS, two-way ANOVA with Bonferroni's multiple comparisons test.

(K) Cardiac puncture serum non-fasting proinsulin levels at sacrifice; n = 6–12 mice/group. **p < 0.01, versus control one-way ANOVA with Dunnett's multiple comparisons test.

(L) Non-fasting proinsulin:insulin ratio at sacrifice; n = 6–12 mice/group. ***p < 0.001, versus control one-way ANOVA with Dunnett's multiple comparisons test.

(M) Cardiac puncture non-fasting serum insulin levels at sacrifice; n = 6–12, *p < 0.05, STZ + BMDM versus STZ + DPBS, Student's t test.

(N) Pancreatic insulin content at sacrifice; n = 5, **p < 0.01, ***p < 0.001 versus control + DPBS, one-way ANOVA with Dunnett's multiple comparisons test. All data represent mean ± SEM.

Next, we investigated if adoptively transferred macrophages could protect mice from STZ-induced hyperglycemia. Similar to islet macrophages, BMDMs are a rich source of IGF-1 protein (Figure S3B) and secrete increased IGF-1 in response to phagocytosis of apoptotic cells (Han et al., 2016). We injected BMDMs intraperitoneally during, and immediately following, STZ (see study design in Figure 3G). Injected macrophages homed to the pancreas (Figure 3H), but could not be found in the spleen (data not shown). Macrophages had no effect on body weight (Figure 3I) but significantly decreased non-fasting blood glucose (Figure 3J). STZ-treated mice had elevated proinsulin levels and proinsulin:insulin ratios both with and without injected macrophages (Figures 3K and 3L), but only mice receiving macrophages had significantly increased non-fasting insulin levels versus STZ controls (Figure 3M). Insulin content was severely reduced due to STZ (Figure 3N), whereas no differences in TUNEL⁺ islet cells or EdU⁺ islet cells were observed between groups (Figures S4E and S4F). Mice receiving macrophages also showed improved glucose tolerance (Figure S4G). These data support the conclusion that macrophages that home to the pancreas can increase insulin secretion in vivo following STZ-induced beta-cell death and implicate a role for IGF-1.

Finally, because islet macrophages are the major, if not the sole, source of IGF-1 in islets and its expression is upregulated following beta-cell death, we investigated if IGF-1 neutralization affects glucose homeostasis during STZ-induced beta-cell death. We injected an IGF-1 antibody intraperitoneally during, and immediately following, STZ (see study design in Figure 4A). IGF-1 neutralization had no effect on body weight, whereas non-fasting blood glucose was increased in the STZ + IgG group only (Figures 4B and 4C). During a glucose challenge, only STZ + IGF-1 Ab mice had significantly impaired glucose tolerance versus IgG control mice (Figures 4D and 4E). Significantly lower insulin levels at 30 min in STZ + IGF-1 Ab mice versus STZ controls coincided with the time point where the glucose tolerance test curves separated (Figures 4F and 4D). Similar to the adoptive transfer experiments, both groups of STZ-treated mice had elevated proinsulin levels and proinsulin:insulin ratios versus their controls (Figures 4G and 4H), whereas non-fasting insulin levels tended to be lower in IGF-1 Ab-treated mice (Figure 4I). Growth hormone [GH] levels were unchanged between groups (Figure 4J). Finally, no differences in TUNEL⁺ islet cells or EdU⁺ cells were observed between groups (Figures S5A and S5B). Thus, post STZ, IGF-1 signaling helps maintain second-phase insulin secretion in vivo.

Figure 4.

IGF-1 Neutralization Decreases Insulin Secretion in Mice following Multiple Low-Dose STZ

(A) Design of IGF-1 neutralization experiments. Multiple low-dose STZ (30 mg/kg, 5 times daily i.p. injections) or control (acetate buffer, 5 times daily i.p. injections) treatments were administered to C57BL/6J males 2 weeks before sacrifice. IGF-1-neutralizing antibody (0.1 μg/g body weight) or control IgG was injected i.p. on days 3, 7, and 11 following the first dose of STZ/buffer.

(B) Body weights; n = 5 mice/group.

(C) Non-fasting blood glucose measurements; n = 5 mice/group, **p < 0.01, ****p < 0.0001 STZ + IgG versus Control + IgG, two-way ANOVA with Dunnett's multiple comparisons test.

(D) Intraperitoneal glucose tolerance test (IPGTT, 1.5 g glucose/kg body weight) 13 days following the first dose of STZ or acetate buffer; n = 5 mice, most of the blood glucose readings in STZ + IGF-1. Ab group were above the range of the glucose meter and were recorded as 33.3 mmol/L, the highest reading within the range of the glucose meter.

(E) Incremental area under the curve (AUC) for mice in (D); n = 5, *p < 0.05 Control + IgG versus STZ + IgG, Kruskal-Wallis test with Dunnett's multiple comparisons test.

(F) Serum insulin levels. Blood was collected at time 0, 15, and 30 min during IPGTT showed in (D); n = 5, *p < 0.05 STZ + IgG versus STZ + IGF1 Ab at 30 min, Student's t test.

(G) Cardiac puncture non-fasting serum proinsulin levels from day 11–14; n = 5–7 mice/group from two separate experiments. **p < 0.01, STZ versus control one-way ANOVA with Tukey's multiple comparisons test.

(H) Non-fasting proinsulin:insulin ratio from day 11–14; n = 5–7 mice/group from two separate experiments. *p < 0.05, **p < 0.01 STZ versus control one-way ANOVA with Tukey's multiple comparisons test.

(I) Cardiac puncture non-fasting serum insulin levels from day 11–14; n = 8–9 from two separate experiments.

(J) Cardiac puncture non-fasting serum growth hormone from day 11–14; n = 5. All data represent mean ± SEM.

High-Fat Diet Further Increases Islet Macrophage Numbers and Growth Factor Gene Expression following Beta-Cell Death

Beta-cell death and increased islet macrophages are usually associated with obesity in individuals with type 2 diabetes (Butler et al., 2003, Ehses et al., 2007). Therefore, we made mice obese by feeding them HFD. All mice were sacrificed at the same time point, 2 weeks post STZ, and at the same age (see study design in Figure 5A). At 12 weeks, HFD and HFD + STZ mice had increased body weight compared with chow-fed mice (Figure 5B). Non-fasting blood glucose was increased in 12-week HFD + STZ mice (Figure 5C), and glucose tolerance was impaired (Figures S6A and S6B). Interestingly, numbers of CD45⁺ cells in islets were significantly increased in 12-week HFD + STZ mice versus STZ mice, due to increased numbers of islet macrophages and other CD45⁺Ly6C⁺CD11B⁻ cells (Figures 5D–5G and S6C–S6E). Numbers of CD45^-Ly6C⁻ cells were also higher in 12-week HFD + STZ compared with STZ islets (Figure S6D), whereas no differences in islet CD45^-Ly6C⁺ cell numbers were found (Figure S6E). HFD did not increase islet monocytes versus STZ mice, similar to findings from Ying and colleagues (Ying et al., 2018).

Figure 5.

High-Fat Diet Further Increases STZ-Induced Islet Macrophages and Growth Factor Gene Expression

(A) Experimental design. C57BL/6J male mice were fed regular chow or HFD for 12 weeks. Multiple low doses of STZ (30 mg/kg, 5 times daily i.p. injections) or acetate buffer (referred to as “Control”) were administered 2 weeks before sacrifice.

(B) Body weights; n = 12–22 mice/group, ^###/***p < 0.001 HFD/HFD + STZ group versus Control, two-way ANOVA with Dunnett's multiple comparisons test.

(C) Non-fasting blood glucose measurements; n = 12–22 mice/group, ***p < 0.001 HFD/HFD + STZ group versus Control, two-way ANOVA with Dunnett's multiple comparisons test.

(D) Representative flow cytometry plots and gating strategy for cell sorting of dispersed islets from mice that received HFD for 12 weeks with acetate buffer injections (left panel) or with multiple low-dose STZ injections (right panel).

(E–G) Fractions of (E) CD45+ cells, (F) islet macrophages, (G) recruited monocytes from mice described in (A); n = 5 for control, STZ groups; n = 5–6 for 12 weeks HFD, 12 weeks HFD + STZ groups. For each sorting sample (n), islets from three mice were pooled together (average of 828 ± 164 islets). *p < 0.05, Student's t test.

(H) qPCR of islet macrophages. Relative mRNA expression levels of Il1a, Il1b, Tnf, Il6, Il1rn, Igf1, Pdgfa, and Tgfbi expressed as fold over islet macrophage control; n = 4–6. For each sorting sample (n), islets from three mice were pooled together (average of 828 ± 164 islets). *p < 0.05, **p < 0.01, test group versus indicated control, one-way ANOVA with Tukey's multiple comparisons test. All data represent mean ± SEM.

Similar to islet macrophage gene expression post STZ alone, 12-week HFD + STZ macrophages also had increased Igf1 mRNA expression versus HFD control (Figure 5H). However, HFD did not further increase Igf1 expression. Interestingly, HFD did further increase Tgfbi mRNA versus STZ alone (Figure 5H). HFD + STZ macrophages also had significantly increased Pdgfa mRNA versus chow controls (Figure 5H). No differences in mRNA expression of these genes were detected in recruited monocytes, CD45^-Ly6C⁻ cells, or CD45^-Ly6C⁺ cells (Figures S6F–S6H). Igf1 mRNA was consistently detected only in islet macrophages (see also Figures 5H and S6F–S6H). In summary, HFD combined with STZ further increased numbers of islet macrophages and skewed islet macrophages to a state of increased growth factor expression.

Islet Macrophages in Diabetic db/db Mice Express Increased Igf1 and Decreased Proinflammatory Cytokines

Because STZ is a chemical toxin that might not be relevant for human disease, we also studied islet macrophages in a genetic rodent model of type 2 diabetes, the db/db mouse. At age 6 weeks, db/db mice had elevated body weight, were hyperglycemic, were hyperglucagonemic, and were hyperinsulinemic compared with BKS controls (Figures 6A–6D). However, between 8 and 11 weeks of age insulin levels declined (Figure 6D), indicative of beta-cell dysfunction and death (Medarova et al., 2005, Puff et al., 2011). Therefore, we investigated islet macrophages and monocytes at 8 and 11 weeks of age. A trend toward increased numbers of CD45⁺ cells in db/db islets at 8 weeks of age was mainly due to significantly increased numbers of islet macrophages (Figures 6E–6G). Similar to islets post STZ, monocytes also tended to be increased (Figure 6H). CD45^-Ly6C⁻ cell numbers were increased and CD45^-Ly6C⁺ cell numbers were significantly reduced (Figures S7A and S7B).

Figure 6.

Islet Macrophages in Diabetic db/db Mice Express Igf1 and Decreased Proinflammatory Cytokines

(A) Body weights of 6- to 11-week-old male BKS and db/db mice.

(B) Non-fasting blood glucose levels of 6- to 11-week-old BKS and db/db mice.

(A and B) n = 17–18 mice for 6- to 8-week-old groups, n = 4 mice for 11-week-old group; ***p < 0.001 db/db versus BKS, Student's t test.

(C) Non-fasting glucagon levels; n = 8 mice for 6- to 8-week-old groups, n = 4 mice for 11-week-old group; **p < 0.01 6-week-old db/db versus BKS control, 8- and 11-week-old db/db versus 6-week-old db/db, one-way ANOVA with Tukey's multiple comparisons test.

(D) Non-fasting insulin levels; n = 8 mice for 6- to 8-week-old group, n = 3–4 mice for 11-week-old group; **p < 0.01, ****p < 0.0001 6-week-old db/db versus BKS control, 8- and 11-week-old db/db versus 6-week-old db/db, one-way ANOVA with Tukey's multiple comparisons test.

(E) Representative flow cytometry profiles and gating strategy for cell sorting of dispersed islets from 8-week-old BKS and db/db mice.

(F–H) Fractions of (F) CD45+ cells, (G) islet macrophages, and (H) recruited monocytes in islets of 8-week-old BKS and db/db mice; (F–H) n = 4, 2–4 mice pooled to obtain 556 ± 52 islets per sample (n); ***p < 0.001 db/db versus BKS, Student's t test.

(I) qPCR of islet macrophages in (G). Relative expression levels of Il1a, Il1b, Tnf, Il6, Il1rn, Igf1, Pdgfa, and Tgfbi expressed as fold control (BKS); n = 4, 2–4 mice pooled per sample (n); *p < 0.05 db/db versus BKS, Student's t test. All data represent mean ± SEM.

Assessment of cytokine (Il1a, Il1b, Il6, Tnf, Il1rn) and growth factor (Igf1, Pdgfa, Tgfbi) mRNA expression in islet macrophages showed significantly reduced Il6 and Tnf expression with 6-fold increased Igf1 mRNA expression (Figure 6I). No differences in mRNA expression of these genes were detected in monocytes (Figure S7C). Igf1 mRNA was consistently detected only in islet macrophages (see also Figures 6I and S7C).

At 11 weeks of age, absolute numbers of CD45⁺ cells in islets also tended to be increased in db/db mice (Figures S7D and S7E). Interestingly, this difference was no longer due to differences in islet macrophage or monocyte numbers (Figures S7F and S7G) and was mainly due to an increase in other immune cell populations (CD45⁺LY6C⁻CD11B⁻CD11C⁻, CD45⁺LY6C⁺CD11B⁻, Figure S7D). Trends in other non-immune cells were similar to those seen at 8 weeks (Figures S7H and S7I). Islet macrophage cytokine and growth-factor mRNA expression showed a similar trend to data from 8-week-old db/db mice (Figure S7J) with elevated Igf1 mRNA.

In summary, similar to STZ-treated and HFD + STZ mice, islet macrophage numbers are increased in 8-week-old db/db mice, and gene expression indicates a state of increased Igf1 expression, and decreased proinflammatory cytokine expression.

Discussion

A number of studies, including our own, have shown that islets in humans with type 2 diabetes have increased numbers of macrophages (Ehses et al., 2007, Lundberg et al., 2017, Marchetti, 2016, Richardson et al., 2009). These cells express markers of both classical pro-inflammatory macrophages (CD68) and non-inflammatory macrophages (CD163) (Ehses et al., 2007). However, their functional role is unclear at present and can only be inferred from preclinical studies and clinical studies targeting the pro-inflammatory cytokine, IL-1 (Larsen et al., 2007, Everett et al., 2018).

Our findings here show that beta-cell death results in a dynamic increase in islet macrophage and recruited monocyte cells within 1 week, returning back to normal levels by 3 weeks. This is kinetically similar to effects seen in other tissues, such as cardiac tissue following injury (Walter et al., 2018). Selected pro-inflammatory, anti-inflammatory, and growth factor genes showed changes at 2 weeks, mainly indicative of a state of wound repair (Vannella and Wynn, 2017), and this was confirmed in our whole-transcriptome analysis. Changes in macrophages paralleled significant increases in TUNEL+ beta-cells, potentially providing insight into how they shift to this state.

Phagocytosis of dead cells, also called efferocytosis, is known to induce an anti-inflammatory, reparative state in macrophages and has been increasingly studied in cardiovascular diseases (Brophy et al., 2017). Macrophages rapidly recognize and engulf apoptotic cells via the so-called eat-me signals, the most fundamental of which is phosphatidylserine (PtdSer) (Lemke, 2019). Efferocytotic macrophages are characterized by a state of increased lysosome activity coupled with a state of increased energy needs (Henson, 2017, Voll et al., 1997). Recent studies have shown that fatty acid oxidation fuels the energy requirements of macrophages undergoing efferocytosis (Zhang et al., 2019). Macrophages undergoing phagocytosis of apoptotic cells are also known to secrete increased levels of IGF-1, which increases phagocytosis of local non-professional phagocytes and minimizes inflammation (Han et al., 2016). Our islet macrophage transcriptome data at 2 weeks post beta-cell death (increased expression of genes involved in oxidative phosphorylation, lysosome and protease activity, lipid transport and oxidation, and increased Igf1) fits well with these known effects of efferocytosis on macrophages in other tissues. Future mechanistic studies should determine the phagocytic receptor responsible for shifting islet macrophages to this reparative state.

The energy requirements of macrophages undergoing efferocytosis may also help explain the changes seen in islet macrophages when beta-cell death was combined with HFD feeding. Enhanced skewing toward a reparative state under HFD feeding could be the result of an increased lipid energy source. Indeed, changes in cellular metabolism that lead to functional programming of phagocytic macrophages are a subject of considerable current interest; our data highlight potential pathways that could be targeted to promote islet macrophages with tissue regenerative properties.

The liver is the main source of circulating IGF-1; however, it was recently proposed that macrophages could be the main source of extrahepatic IGF-1 (Gow et al., 2010). Indeed, numerous studies have identified macrophages as major producers of local tissue IGF-1 in the brain, in skeletal muscle, in the lung, and in various other tissues in studies investigating tumor-associated macrophages in cancer biology (Forbes, 2016, Han et al., 2016, Ireland et al., 2016, Tonkin et al., 2015). Here we found that islet macrophages are the major source, if not the sole source, of IGF-1 within pancreatic islets. This has important implications for islet biology.

IGF-1 has long been known to have a beneficial role in diabetes and cardiovascular disease (Higashi et al., 2019). Beta-cell IGF-1 overexpression protects from STZ-induced diabetes (George et al., 2002, Robertson et al., 2008) and exogenous IGF-1 protects NOD mice from developing type 1 diabetes, likely via effects on T cells (Bergerot et al., 1995, Kaino et al., 1996). However, overexpression of IGF-1 locally within islets does not lead to an overt phenotype or effects on beta-cell mass, suggesting that its actions on beta-cells in diabetes are mainly indirect (George et al., 2002, Robertson et al., 2008). The only beta-cell effect observed in transgenic RIP-IGF-1 mice was increased second-phase (30-min) insulin secretion in response to a glucose challenge (Guo et al., 2005). This insulin secretory effect is in agreement with studies knocking out the IGF-1 receptor (IGF-1R) from beta-cells. The absence of a beta-cell IGF-1R did not impair beta-cell development or have effects on beta-cell mass, but it did result in increased fasting insulin levels and impaired first- and second-phase insulin secretion in mice (Kulkarni et al., 2002, Xuan et al., 2002). These data agree well with our current findings, where the primary effect of neutralizing IGF-1 was to decrease second-phase insulin secretion post beta-cell death. We propose that macrophages are the major local source of IGF-1 within pancreatic islets and that paracrine and autocrine effects of macrophage IGF-1 are critical in dampening islet inflammation and maintaining insulin secretion under pathological conditions (Figure 7).

Figure 7.

Proposed Hypothesis for How Beta-Cell Death Leads to Macrophage IGF-1 Production

Following beta-cell death, islet macrophages rapidly recognize and phagocytose dead cells via specific phagocytosis receptors. This results in increased expression of genes involved in oxidative phosphorylation, ATP production, lysosome activity, and lipid transport and oxidation. During this process islet macrophages change their polarization state, decrease expression of pro-inflammatory cytokines, and increase IGF-1 secretion. These reparative macrophages then attempt to maintain beta-cell insulin secretion via IGF-1. IGF-1 may also have effects on other islet cell types or macrophages themselves to help limit inflammation and maintain beta-cell function under pathological conditions.

The present findings may seem counterintuitive given the established deleterious role of islet inflammation and IL-1 in humans with type 2 diabetes (Donath and Shoelson, 2011). Indeed, our own work has shown that IAPP induces IL-1β in islet macrophages and TLR ligands elevated in type 2 diabetes activate islet macrophages and have negative effects on beta-cell Ins gene expression (Nackiewicz et al., 2014, Westwell-Roper et al., 2014). However, in the absence of IAPP aggregates or certain TLR ligands islet macrophages clearly skew to a reparative phenotype following beta-cell death. We also observed a pro-resolution/anti-inflammatory phenotype in islet macrophages in ex vivo studies on GK rat islets isolated when beta-cell function was declining in vivo (not shown).

When considering our data in the context of a chronic disease such as type 2 diabetes, it is important to keep in mind how dynamic inflammation and the function of a macrophage is. This is supported by our studies on the db/db mouse, where increased numbers of islet macrophages at 8 weeks where no longer evident at 11 weeks of age, when increases in other immune cell subsets were observed. It is also important to bear in mind that in our own work and that of others, global unbiased transcriptome analysis of islet macrophages was never conducted. It is entirely possible that islet macrophages exhibit a “mixed” pro-inflammatory and reparative phenotype in type 2 diabetes in response to phagocytosis of dead beta-cells following cell death induced by IAPP and other factors (glucose, free fatty acids, endoplasmic reticulum stress, local and/or circulating cytokines). Regardless, our findings are in agreement with a recent study in db/db mice, wherein an NLRP3 inhibitor had no effects on glycemia (Kammoun et al., 2018). Ying and colleagues also found a mixed phenotype of islet macrophages during HFD feeding, not simply defined by M1 or M2 skewing (Ying et al., 2018). Similar to our findings, they also saw increased Pdgfa expression during HFD feeding (Ying et al., 2018).

Our data may have important therapeutic implications relevant to type 2 diabetes. Despite the observance of a significant decrease in HbA1c, IL-1β inhibition has not shown prolonged effects on glycemic control in the studies conducted until now (Everett et al., 2018, Larsen et al., 2007). This could be due to the study population investigated until now or due to the dynamics of islet inflammation. Because inflammation is dynamic and islet macrophages can still change their phenotype to a reparative state even in the presence of HFD feeding or extreme hyperglycemia, increasing their beneficial effects might be an alternative approach to preserving functional beta-cell mass in type 2 diabetes.

Limitations of the Study

A few limitations should be considered when interpreting our data. Currently available techniques do not allow for manipulation of islet-specific macrophages in vivo due to lack of tissue-specific gene expression. Thus, macrophage depletion and adoptive transfer experiments should be interpreted with this in mind. To minimize this limitation, we made use of a toxin that specifically kills beta-cells (Lenzen, 2008), to enable study of the effects of macrophages in islet tissue. We also cannot completely exclude a role for liver IGF-1 in our neutralization experiments, despite effects on insulin secretion that were only evident following beta-cell death. GH levels were also not changed so there did not appear to be any impact of short-term IGF-1 neutralization on insulin sensitivity (Kim and Park, 2017). Although the data suggest that islet macrophage-derived IGF-1 helps maintain beta-cell insulin secretion post beta-cell death, this cannot be conclusively proven in the absence of macrophage IGF-1 knockout studies.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.

Published: January 24, 2020

Data and Code Availability

All relevant data are available from the authors upon request. RNA sequencing data have been deposited in the ArrayExpress database at EMBL-EBI (www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-7234.