Islet amyloid disrupts MHC class II antigen presentation and delays autoimmune diabetes in NOD mice

TL;DR

Islet amyloid disrupts immune presentation in mice, delaying diabetes onset despite causing inflammation.

Contribution

Islet amyloid is shown to delay autoimmune diabetes by inhibiting MHC class II antigen presentation in macrophages.

Findings

Islet amyloid formation downregulates MHC class II antigen presentation genes in macrophages.

NOD mice expressing hIAPP transgene or knockin show delayed diabetes onset compared to controls.

IAPP aggregates reduce MHCII surface expression and T cell activation in dendritic cells.

Abstract

Islet amyloid contributes to beta cell failure in type 2 diabetes through several mechanisms, one being the potent induction of local islet inflammation through activating inflammatory pathways in islet macrophages. As islet amyloid has recently been reported in pancreases of people with type 1 diabetes, and islet macrophages are thought to play a role in the pathogenesis of type 1 diabetes, we sought to understand the impact of islet amyloid on islet macrophages and beta cell autoimmunity.

We performed an unbiased phenotypic investigation of islet macrophages in the early stage of islet amyloid formation using single-cell RNA-seq of resident islet macrophages in mice with and without the amyloidogenic form of human islet amyloid polypeptide (hIAPP). The role of islet amyloid in autoimmune diabetes and antigen presentation was assessed in hIAPP-expressing NOD mice and in antigen-presenting cells ex vivo.

MHC class II (MHCII) antigen presentation genes were strongly downregulated in islet macrophages during islet amyloid formation. NOD mice expressing an hIAPP transgene had delayed diabetes relative to littermate controls (median onset 30.3 vs 19.5 weeks, p=0.016). Likewise, physiological expression of hIAPP by genetic knockin also delayed diabetes in NOD mice relative to littermate controls (median onset 28.2 vs 18.0 weeks, p=0.049), corresponding with decreased markers of antigen presentation and activation, as well as decreased immune cell infiltration in islets. Adoptive transfer studies showed that systemic autoimmune function remained intact and beta cells from hIAPP transgenic mice did not evade immune recognition by diabetogenic T cells, collectively indicating the protection from diabetes was mediated by decreased antigen presentation in the pancreas. Consistent with this, incubation of dendritic cells with islet amyloid polypeptide (IAPP) aggregates decreased MHCII surface expression and diminished antigen-specific T cell activation through a phagocytosis-dependent mechanism.

Collectively, our data reveal a novel role for IAPP aggregates in decreasing MHCII antigen presentation and show that despite the well-established proinflammatory response of macrophages to IAPP aggregates, the uptake of IAPP aggregates during early amyloid formation also disrupts beta cell autoimmunity and delays diabetes in NOD mice.

The online version contains peer-reviewed but unedited supplementary material available at 10.1007/s00125-025-06622-0.

Figures

The MHCII pathway is downregulated in islet myeloid cells of hIAPP Tg/0 mice. (a–g) Data from female control (black, solid circle) and hIAPP Tg/0 (teal, open square) transgenic mice on an F1 C57BL/6J × FVB/NJ background. (a) Overview of animal model breeding strategy. C57BL/6J mice were crossed with hIAPP Tg/0 transgenic FVB/NJ mice to generate F1 littermate hIAPP Tg/0 transgenic hIAPP 0/0 control mice on a mixed C57BL/6J × FVB/NJ background. (b, c) Glucose tolerance data (1.5 g glucose/kg body weight, i.p.) from chow-fed 8-week-old mice before (b) and 2 weeks after HFD (c), *n*=5/group. Data

c, d), mirroring the well-established beta cell phenotype observed in hIAPP Tg/0 mice on non-autoimmune backgrounds. A trend towards higher insulin levels in vivo was apparent in female NOD.hIAPP Tg/0 mice (Fig. [3](#Fig3)c). Glucose-stimulated insulin secretion (Fig. [3](#Fig3)d) and insulin content (Fig. [3](#Fig3)e) were increased ex vivo in female NOD.hIAPP Tg/0 islets compared with littermate islets. This difference was abolished upon normalisation to insulin content (Fig. [3](#Fig3)f). Male NOD.hIAPP Tg/0 mice also showed perturbations of in vivo and ex vivo stimulated insulin secretion

hIAPP expression does not delay islet allograft rejection. (a) Schematic overview of islet transplantation studies (Created in BioRender. Skovsoe, S. (2025) [https://BioRender.com/9jpxtdu](https://BioRender.com/9jpxtdu). In brief, isolated islets from 25-week-old hIAPP 0/0 or hIAPP Tg/0 male mice on the FVB/NJ background were transplanted under the kidney into STZ-induced diabetic 8-week-old male BALB/cJ mice. (b) Diabetes incidence of male BALB/cJ mice with transplanted islets isolated from either hIAPP 0/0 control (black, open diamonds, *n*=11) or hIAPP Tg/0 control (green, open diamonds, *n

IAPP aggregates disrupt MHCII antigen presentation. (a–e) Data from BMDCs pre-treated overnight with or without LPS in addition to vehicle (black), rIAPP (orange) or hIAPP (pink). (a) Representative flow cytometry plots showing MHCII fluorescence in treated BMDCs. (b) Frequency of treated BMDCs expressing high (MHCII^bright^) levels of MHCII. (c, d) Modal normalised count vs MHCII fluorescence (c) and median fluorescence intensity (MFI) of treated BMDCs (d). (e) Frequency of treated BMDCs expressing low (MHCII^dim^) levels of MHCII. (f) Proliferation (% of OT-II CD4^+^ cells) when co-cultured

Full text

Introduction

Type 1 diabetes is caused by autoreactive T cell-mediated destruction of pancreatic beta cells leading to insufficient insulin levels and hyperglycaemia. Islet macrophages play a key role in the initiation and progression of type 1 diabetes pathogenesis, and depletion of macrophages protects NOD mice from developing diabetes [1–4]. Proinflammatory cytokines from macrophages contribute to beta cell loss in type 1 diabetes, exemplified by the protective effects of deleting NLR family pyrin domain containing 3 (NLRP3) inflammasome in NOD mice and inhibiting TNF in humans with recent-onset type 1 diabetes [5–7]. In addition, islet macrophages are also important antigen-presenting cells (APCs), phagocytosing secreted peptides, insulin granules and crinosomes from beta cells and presenting beta cell antigens to autoreactive T cells [1–3, 8–10].

Islet amyloid, formed by the aggregation of the beta cell peptide hormone islet amyloid polypeptide (IAPP), is a strong proinflammatory stimulus, triggering islet inflammation and contributing to beta cell dysfunction [11–15]. Islet macrophages, residing in close contact with beta cells and sampling insulin granule content and secretions [10, 16], are poised to directly interact with IAPP aggregates as they form. We and others have shown in vitro that IAPP aggregates directly interact with Toll-like receptor 2 (TLR2) and are phagocytosed by macrophages and dendritic cells (DCs), eliciting a proinflammatory response and NLRP3 inflammasome activation [15, 17, 18]. Phagocytosis of islet amyloid has also been observed in islet macrophages in situ [11, 13, 19]. By systemically depleting macrophages or inhibiting inflammatory cytokine signalling, we have shown macrophage inflammatory responses are an essential link between IAPP aggregates and beta cell dysfunction [11–13]. Nevertheless, the direct impact of islet amyloid on resident islet macrophages in situ has not been studied.

In this study, we investigated the effects of islet amyloid on islet macrophages in situ in an unbiased manner by performing single-cell RNA-seq (scRNA-seq) on islet myeloid cells from a mouse model of early islet amyloid formation (hIAPP Tg/0 mice). This subsequently led us to examine the effect of islet amyloid on autoimmune-mediated diabetes in the NOD mouse model. While the proinflammatory effects of islet amyloid are well studied in type 2 diabetes, islet amyloid has only recently been reported in islets from people with type 1 diabetes [20–22], and its role in beta cell autoimmunity has not been investigated.

Methods

Animals

All animal experiments were approved by the Animal Care Committee at the University of British Columbia and performed in agreement with the ethical guidelines and policies provided by the Canadian Council on Animal Care. Mouse strains are described in the electronic supplementary material (ESM) Methods and were purchased from The Jackson Laboratory (Bar Harbor, USA) except for human islet amyloid polypeptide (hIAPP) knockin mice, which were developed previously [23] and provided by S. Kahn (University of Washington Diabetes Institute, Seattle, USA). Mice were genotyped from ear notch-extracted DNA with primers in ESM Table 1. Refer to the ESM Methods for randomisation.

scRNA-seq

Refer to the ESM Methods for more details. Dispersed islet cells from high-fat diet (HFD)-fed hIAPP Tg/0 and hIAPP 0/0 littermates on an F1 C57BL/6J × FVB/NJ background were pooled (five mice per sample) and sorted into CD45^+^ and CD45^−^ fractions via BD FACS ARIA II (BD Biosciences), and reconstituted to a final composition of 30% CD45^−^ cells and 70% CD45^+^ cells. Samples were loaded onto the Chromium Controller and underwent library preparation scRNA-seq using the Chromium Single Cell 3′ V2 Reagent Kit (10X Genomics). Sample libraries were then sequenced on an Illumina NextSeq500 to a read-depth minimum of 80,000 reads per cell. Transcriptome analysis is described in the ESM Methods.

Flow cytometry

Islets were isolated via collagenase injection into the pancreatic duct, and dispersed for flow cytometry as previously published [24, 25] and described in the ESM Methods. Processing of spleen and lymph nodes is described in the ESM Methods. Antibodies and staining reagents are provided in ESM Table 2.

NOD diabetes progression studies

hIAPP Tg/0 transgenic mice on the FVB/NJ background were backcrossed (>10 generations) with NOD mice to generate NOD.hIAPP Tg/0 and NOD.hIAPP 0/0 littermates. Iapp^h/h^ knockin mice on a C57BL/6 background [23] were backcrossed (>10 generations) with NOD mice to create NOD.Iapp^h/m^ mice. All hIAPP Tg/0 strains described in this paper were maintained as hemizygous carriers, by crossing male carriers to wild-type females. NOD.Iapp^h/h^, NOD.Iapp^h/m^ and NOD.Iapp^m/m^ littermates were generated via NOD.Iapp^h/m^ heterozygous breeding. Blood glucose and body weight were monitored 1–2× weekly up to 30 weeks of age.

Adoptive transfers

NOD SCID.hIAPP Tg/0 and NOD SCID.hIAPP 0/0 littermates were generated as described in the ESM Methods. The Prkdc^scid^ allele was genotyped using the restriction fragment length polymorphism assay described by Quadros et al [26]. Adoptive transfers are described in the ESM Methods. Splenocytes from diabetic NOD/ShiLtJ donors were injected into NOD SCID.hIAPP Tg/0 and NOD SCID.hIAPP 0/0 female mice. Refer to the ESM Methods for more details. Blood glucose was measured 2× per week in recipients and diabetes defined as 2× blood glucose >13 mmol/l.

Islet allograft model

Diabetes was induced in 8-week-old BALB/cJ mice by injection of 200 mg/kg streptozocin (STZ) (S0130, Sigma, US) prepared as previously described [27]. Islets from 25-week-old hIAPP Tg/0 and hIAPP 0/0 male mice (to ensure amyloid formation) on the FVB/NJ background were isolated and transplanted under the kidney capsule of sex-matched diabetic BALB/cJ recipient mice following previously published protocols [28–30].

Blood glucose and insulin measurement

Tail blood glucose was measured using a OneTouch Ultra Glucometer (Life Scan, Burnaby, BC, Canada). Unless otherwise stated, diabetes was defined as 2× blood glucose >20 mmol/l. Refer to the ESM Methods for glucose tolerance tests and glucose-stimulated insulin levels.

T cell proliferation assay

Bone marrow-derived dendritic cells (BMDCs) were generated from C57BL/6J mice and treated with synthetic 10 µmol/l hIAPP, rodent islet amyloid polypeptide (rIAPP) (prepared as previously published [17]) or vehicle as indicated for 24 h. BMDCs were washed and co-cultured at a 1:5 ratio with 5(6)-carboxyfluorescein diacetate succinimidyl ester (CFDA-SE) (Stemcell Technologies, Canada)-labelled splenic CD4 T cells from a B6.Cg-Tg(TcraTcrb)425Cbn/J (OT-II) mouse. After 3 days, proliferation of CD4 cells was determined by flow cytometry (CytoFLEX). Refer to the ESM Methods for more details.

Nanostring

Islet gene expression was analysed in isolated islets from NOD.hIAPP Tg/0 mice and littermate controls via Nanostring nCounter XT (NanoString, Bothell, USA) with a custom Code Set following the manufacturer instructions. Briefly, islet cell lysates were prepared and added to hybridisation mixtures for 18 h at 65°C with a 70°C-heated lid according to the manufacturer instructions. Gene expression was measured with a Nanostring SPRINT Profiler, and data were analysed via nSolver 4.0 (NanoString).

Immunisation

Ovalbumin (Ova) mRNA with 5′ methoxy-uridine (5moU) base modification (Trilink BioTechnologies, San Diego, USA) was formulated into lipid nanoparticles (LNPs) (Integrated Nanotherapeutics, Canada) as previously published [31]. Refer to the ESM for LNP formulation. Mice were injected with 10 µg of mRNA in LNP formulation i.p. at 0 and 14 days, prior to collection of blood from the saphenous vein for measurement of anti-Ova IgG1 by ELISA (Cayman Chemical, Ann Arbor, USA).

Immunofluorescence

Pancreases were harvested immediately after euthanasia, and processed and stained as previously described [32]. Refer to the ESM Methods for more details.

Statistical methods

Except for gene expression data, all statistical analyses were performed using GraphPad Prism 10.0 (GraphPad Software, La Jolla, USA). Refer to the ESM Methods for more details.

Results

The MHCII pathway is downregulated in islet macrophages of hIAPP Tg/0 mice

We sought to examine the phenotype of islet macrophages exposed to islet amyloid in situ. Unlike hIAPP, rIAPP does not form amyloid due to differences in its protein sequence. Therefore, we used mice transgenic for the human form of IAPP (hIAPP Tg) which overexpress IAPP in beta cells, and, under obesogenic conditions, drives islet amyloid and beta cell dysfunction, modelling type 2 diabetes [12–14, 33, 34]. hIAPP Tg/0 mice were crossed with C57BL/6J mice to generate hIAPP Tg/0 and hIAPP 0/0 littermates on an F1 C57BL/6J × FVB/NJ background (Fig. 1a) which promotes islet amyloid formation [33]. To accelerate islet amyloid formation, 8-week-old male hIAPP Tg/0 and hIAPP 0/0 mice, which have similar baseline glucose tolerance (Fig. 1b), were placed on HFD for 2 weeks. Following this brief dietary change, there was a trend towards impaired glucose tolerance (p=0.01 by mixed effects model, incremental AUC [iAUC]=896 ± 186 in hIAPP 0/0 vs 1131 ± 69 in hIAPP Tg/0, p=0.095 by Mann–Whitney test; Fig. 1c) and fasting blood glucose was elevated (Fig. 1d) in hIAPP Tg/0 mice relative to littermate hIAPP 0/0 controls despite maintaining similar body weight (Fig. 1e), indicative of early islet amyloid formation disrupting beta cell function.Fig. 1. The MHCII pathway is downregulated in islet myeloid cells of hIAPP Tg/0 mice. (a–g) Data from female control (black, solid circle) and hIAPP Tg/0 (teal, open square) transgenic mice on an F1 C57BL/6J × FVB/NJ background. (a) Overview of animal model breeding strategy. C57BL/6J mice were crossed with hIAPP Tg/0 transgenic FVB/NJ mice to generate F1 littermate hIAPP Tg/0 transgenic hIAPP 0/0 control mice on a mixed C57BL/6J × FVB/NJ background. (b, c) Glucose tolerance data (1.5 g glucose/kg body weight, i.p.) from chow-fed 8-week-old mice before (b) and 2 weeks after HFD (c), n=5/group. Data were analysed using repeated measures mixed effects analysis. Individual iAUCs shown as mmol/l × min were analysed via Mann–Whitney test. (d, e) 4 h fasted blood glucose (d) and body weight (e) following 2 weeks of HFD (hIAPP Tg/0: n=15; hIAPP 0/0: n=16). Data were analysed using Mann–Whitney and unpaired t tests, respectively, depending on normality distribution. (f) Overview of scRNA-seq strategy of islet immune (CD45^+^) cells. (g) Percentage of CD45^+^ cells of total live islet cells (hIAPP Tg/0: n=3; hIAPP 0/0: n=3), five mice pooled per sample. (h) Bubble plot showing key marker gene expression in each major scRNA-seq population. (i, j) Uniform manifold approximation and projection (UMAP) plots of major scRNA-seq populations annotated by cluster (i) or genotype (j). (k) Enrichment score of Gene Ontology (GO) molecular function MHC class II protein complex binding gene set from hIAPP Tg/0 myeloid cell population vs control. (l) Bubble plot of select MHCII-associated genes in myeloid cells across samples. (m) Bubble plot of select markers of DC activation, T cell priming and phagocytosis, lysosomal maturation and tissue retention in myeloid cells across samples. Data are shown as mean ± SD. *p<0.05, **p<0.01. (a, f) Created in BioRender. Skovsoe, S. (2025) https://BioRender.com/ m17i879 and https://BioRender.com/ c48j305, respectively

We examined the islet macrophage phenotype in this early islet amyloid exposure model by scRNA-seq of single-cell suspensions from hIAPP Tg/0 and hIAPP 0/0 islets (Fig. 1f). To obtain sufficient islet immune cells (only ~2% of total islet cells), islet cells from five mice were pooled per sample, and viable CD45^+^ cells sorted by FACS. We observed a non-significant trend towards increased CD45^+^ cell frequency in hIAPP Tg/0 islets relative to controls (Fig. 1g), consistent with early islet inflammation [11, 13, 14]. Each sorted CD45^+^ sample was spiked with CD45^−^ islet cells to reconstitute samples of 75% CD45^+^, 25% CD45^−^ for scRNA-seq (Fig. 1f). As a result, samples consisted of a small proportion of endocrine cells (high ChgA expression, ESM Fig. 1) with distinct beta cell, delta cell and alpha cell clusters (Fig. 1h–j), along with a large population of immune cells (marked by Ptprc expression, ESM Fig. 1) and a small population of endothelial cells (marked by Plvap). Cd14-expressing myeloid cells composed the majority of islet immune cells, while small subsets of T and B cells were also present (high Cd3e, Cd79a, respectively*)*. These intra-immune proportions are consistent with studies confirming >90% of islet immune cells are macrophages [1, 24, 35] and our previous work quantifying 2–5% T and B cells in non-diabetic mouse islets [25].

Although islet amyloid triggers inflammation in macrophages, the MHC class II (MHCII) pathway was strongly downregulated in the myeloid population of hIAPP Tg/0 islets compared with wild-type controls (Fig. 1k). Both MHCII genes and those involved in MHCII antigen processing (e.g. Cd74) were downregulated (Fig. 1l). In addition, Ciita, the master transcriptional regulator of MHCII genes, was downregulated (Fig. 1l). In contrast, MHC class I genes were not altered, despite being proximal to MHCII genes in the H2 locus (ESM Fig. 1). These data suggest that despite its known inflammatory effects [11, 13–15], early islet amyloid exposure suppresses MHCII antigen presentation. Transcriptome profiles showed decreased markers of migration (e.g. Ccr2, Ccr7), activation and T cell priming (e.g. Cd86, Il12b), and a concomitant increase in markers of phagocytic activity (e.g. Fcgr1, Fcgr3), lysosomal processing (e.g. Rab7, Rab5a), degradation (Ctss, Ctsd, Ctsb) and tissue retention (Cxcr4 and Cx3cr1) (Fig. 1m). These transcriptomic changes suggest a functional shift in islet myeloid cells away from antigen presentation and T cell priming and towards a tissue-anchored, phagocytic and degradative role in the presence of islet amyloid.

Islet amyloid delays autoimmune diabetes in NOD mice

We next examined the effect of islet amyloid in the NOD mouse model of type 1 diabetes by backcrossing hIAPP Tg/0 FVB/NJ mice >10 times to the NOD/ShiLtJ background to generate NOD.hIAPP Tg/0 and control NOD.hIAPP 0/0 littermates (Fig. 2a). NOD congenicity of >99% was confirmed by SNP analysis (ESM Fig. 2). Interestingly, female NOD.hIAPP Tg/0 mice had a significantly reduced diabetes incidence and delayed diabetes onset relative to littermate controls (median onset 30.3 vs 19.5 weeks and total incidence 50% vs 80%, respectively, p=0.016) (Fig. 2a). In males, the median onset and diabetes incidence were unaltered (ESM Fig. 3a, b).Fig. 2hIAPP delays spontaneous diabetes in female NOD mouse models. (a–g) Data from female control (black, solid circle) and hIAPP-expressing (pink, open symbols) mice on a NOD/ShiLtJ background. (a) Overview of hIAPP transgenic (hIAPP Tg/0) NOD/ShiLtJ backcrossing and diabetes incidence. Diabetes incidence of female NOD.hIAPP Tg/0 (n=12) and NOD.hIAPP 0/0 control (n=20) mice, analysed by logrank test. (b) Overview of hIAPP knockin (Iapp^h/h^) NOD/ShiLtJ backcrossing and diabetes incidence. Diabetes incidence of wild-type (NOD.Iapp^m/m^) (black, solid circle, n=15), heterozygous (NOD.Iapp^h/m^) (pink, open triangle, n=15) and homozygous knockin (NOD.Iapp^h/h^) (pink, open square, n=16) mice, analysed by logrank test. (c) Nanostring gene expression data from islets isolated from female NOD.hIAPP 0/0 (black) and NOD.hIAPP Tg/0 (pink) mice, n=6/group, data are mean + SEM of fold change relative to NOD.hIAPP 0/0. (d) Representative FACS plot of islet macrophages from NOD.hIAPP 0/0 and NOD.hIAPP Tg/0 mice. (e) Modal normalised count vs MHCII fluorescence in islet macrophages. (f, g) Islet immune cell populations as a frequency of all islet cells (f) and of CD45^+^ cells (g) from female NOD.hIAPP Tg/0 and NOD.hIAPP 0/0 mice. (h, i) CD4^+^ and CD8^+^ frequency of CD3^+^ cells (h) and Treg frequency (indicated by FOXP3) of CD4^+^ cells (i) in spleen (circular symbols) and pancreatic lymph nodes (square symbols) from female NOD.hIAPP Tg/0 (pink) and NOD.hIAPP 0/0 (black) mice. (j) Serum Ova IgG1 concentrations from NOD.hIAPP Tg/0 (black circles) and NOD.hIAPP 0/0 (pink circles) mice immunised with Ova mRNA. (k) Schematic of adoptive transfer of prediabetic splenocytes from either NOD.hIAPP Tg/0 or NOD.hIAPP 0/0 donor mice into NOD.CG.Prkdc^Scid/Scid^ recipient mice. (l) Diabetes incidence curve of NOD.CG.Prkdc^Scid/Scid^ recipient mice of splenocytes from NOD.hIAPP 0/0 (black solid circle, n=3) or NOD.hIAPP Tg/0 (pink open triangle, n=4) donor mice, analysed via logrank test. Unless otherwise stated, data are shown as mean ± SD. *p<0.05. FOXP3, forkhead box protein P3. (a, b, k) Created in BioRender. Skovsoe, S. (2025) https://BioRender.com/ e34p682, https://BioRender.com/ vu799d3, and https://BioRender.com/ u36v131, respectively

To control for artefacts that could result from the hIAPP transgene, we generated a complementary hIAPP knockin NOD/ShiLtJ mouse model. hIAPP knockin C57BL/6 mice carry the hIAPP protein coding sequence under the endogenous mouse Iapp promoter, resulting in physiological expression of amyloidogenic IAPP in islets [23]. These mice were backcrossed to NOD/ShiLtJ mice for >10 generations to generate littermates homozygous for hIAPP (NOD.Iapp^h/h^), and littermate controls homozygous or heterozygous for wild-type mouse IAPP (NOD.Iapp^m/m^ and NOD.Iapp^h/m^) (Fig. 2b). Similar to transgenic NOD.hIAPP mice, homozygous hIAPP knockin delayed diabetes onset relative to homozygous wild-type controls (median onset 28.2 vs 18.0 weeks and total incidence 50% vs 80%, respectively, p=0.049) (Fig. 2b). Interestingly, heterozygous hIAPP knockin NOD.Iapp^h/m^ mice were not protected from diabetes, indicative of a possible dependency on the level of hIAPP expression (median onset 19.9 and total incidence 85%). Thus, in two distinct NOD mouse models of islet amyloid, hIAPP delays autoimmune diabetes progression.

We next examined the immune phenotype of NOD.hIAPP Tg/0 mice. Whole islets isolated from NOD.hIAPP Tg/0 mice had reduced expression of MHCII antigen presentation genes and APC activation genes relative to littermate controls (Fig. 2c). Furthermore, macrophages in NOD.hIAPP Tg/0 islets had decreased MHCII surface expression (Fig. 2d, e). Consistent with reduced immune infiltration, CD45 (Ptprc) expression and T cell marker (Cd3d) expression were also reduced in whole hIAPP Tg/0 islets (Fig. 2c), as was the frequency of CD45^+^ and CD8^+^ T cells in hIAPP Tg/0 islets by flow cytometry (Fig. 2f, g). No significant differences in CD4^+^, CD8^+^ (Fig. 2h) or regulatory T (Treg) cell (Fig. 2i) frequency were observed in either the spleen or the pancreatic lymph node between groups, implying a lack of systemic perturbation in immune cell subsets. To further investigate whether NOD.hIAPP Tg/0 mice have a normal functioning systemic immune response, we immunised mice with Ova mRNA in an LNP and measured the generation of anti-Ova antibodies. No difference in serum anti-Ova IgG1 concentrations was observed, indicative of an intact immune response to systemically administered antigen (Fig. 2j). Furthermore, splenocytes from NOD.hIAPP Tg/0 mice and NOD.hIAPP 0/0 mice induced diabetes at a similar rate in immunocompromised NOD.Prkdc^Scid/Scid^ (NOD SCID) mice (Fig. 2k), again indicating functioning autoreactive T cells are present in NOD.hIAPP Tg/0 mice (Fig. 2l). Collectively, these data support that amyloid protects from autoimmune diabetes by disrupting MHCII antigen presentation locally in the pancreas, rather than disrupting systemic immune functions.

NOD.hIAPP Tg/0 beta cells are dysfunctional but do not evade autoimmunity

To further elucidate the mechanism of diabetes protection in hIAPP Tg/0 mice, we examined the beta cell phenotypes of prediabetic NOD.hIAPP Tg/0 mice. At 8 weeks of age, NOD.hIAPP Tg/0 female mice had impaired glucose tolerance compared with littermate controls (p=0.03, Fig. 3a) despite no differences in body weight (Fig. 3b). There was a non-significant trend towards increased beta cell area at this age, which could reflect protection from insulitis (ESM Fig. 4). Male NOD.hIAPP Tg/0 mice had a similar phenotype (ESM Fig. 3c, d), mirroring the well-established beta cell phenotype observed in hIAPP Tg/0 mice on non-autoimmune backgrounds. A trend towards higher insulin levels in vivo was apparent in female NOD.hIAPP Tg/0 mice (Fig. 3c). Glucose-stimulated insulin secretion (Fig. 3d) and insulin content (Fig. 3e) were increased ex vivo in female NOD.hIAPP Tg/0 islets compared with littermate islets. This difference was abolished upon normalisation to insulin content (Fig. 3f). Male NOD.hIAPP Tg/0 mice also showed perturbations of in vivo and ex vivo stimulated insulin secretion without altered insulin content (ESM Fig. 3e–h). Histology confirmed the development of islet amyloid in NOD.hIAPP Tg/0 female mice (Fig. 3g). Ruling out immune evasion of hIAPP-expressing beta cells, adoptive transfer of diabetic NOD splenocytes induced diabetes effectively in NOD SCID.hIAPP Tg/0 recipients (ESM Fig. 5a, b). In summary, these data indicate that NOD.hIAPP Tg/0 mice develop islet amyloid and beta cell dysfunction similarly to other hIAPP Tg/0 strains and that these perturbations per se do not evade autoimmune attack by diabetogenic T cells.Fig. 3NOD.hIAPP Tg/0 beta cells are dysfunctional but do not evade autoimmunity. (a–f) NOD.hIAPP Tg/0 (pink open circle) and NOD.hIAPP 0/0 (black circle) littermates were used for all experiments. (a) Glucose tolerance (1 g/kg, i.p.) of 8-week-old female NOD.hIAPP Tg/0 (n=17) and control NOD.hIAPP 0/0 (n=13) mice. Data were analysed using a repeated measures mixed effects analysis. Individual iAUCs (in mmmol/l × min) from experimental and control mice were analysed with the unpaired t test. (b) Body weights of 8-week-old NOD.hIAPP Tg/0 (n=17) and NOD.hIAPP 0/0 (n=13) mice. Data were analysed using the Mann–Whitney test. (c) In vivo glucose-stimulated (1 g/kg, i.p.) plasma insulin levels from 8-week-old NOD.hIAPP Tg/0 (n=9) and NOD.hIAPP 0/0 control (n=7) mice. Data were analysed using a repeated measures mixed effects analysis. (d) In vitro glucose-stimulated insulin secretion (GSIS) from islets isolated from 8-week-old NOD.hIAPP Tg/0 (n=6) and control NOD.hIAPP 0/0 (n=6) mice. Data were analysed using a mixed effects analysis. (e) Insulin content from islets isolated from 8-week-old NOD.hIAPP Tg/0 (n=6) and control NOD.hIAPP 0/0 (n=5) mice. (f) In vitro GSIS data normalised to insulin content from islets isolated from 8-week-old NOD.hIAPP Tg/0 (n=6) and control NOD.hIAPP 0/0 (n=5) mice. (g) Representative images of islet amyloid severity in newly diabetic and normoglycaemic 30-week-old NOD.hIAPP female mice. Data are shown as mean ± SD. *p<0.05, **p<0.01, ***p<0.001

hIAPP Tg/0 islets are not protected from alloimmune rejection

We next examined whether immune attack of transplanted allogeneic islets might also be delayed. Notably, in allotransplant rejection, antigen recognition occurs through APCs processing and presenting alloantigen to host cells and the direct recognition of allogeneic MHC molecules on transplanted cells. hIAPP Tg/0 and hIAPP 0/0 islets from F1 C57BL/6J × FVB/NJ donor mice were transplanted into STZ-induced diabetic BALB/cJ recipient mice (Fig. 4a). In direct contrast to the autoimmune models, rejection of allogeneic hIAPP Tg/0 islets was modestly accelerated relative to hIAPP 0/0 controls (Fig. 4b), suggesting that islet amyloid does not impair direct recognition of allogeneic MHC complexes.Fig. 4hIAPP expression does not delay islet allograft rejection. (a) Schematic overview of islet transplantation studies (Created in BioRender. Skovsoe, S. (2025) https://BioRender.com/9jpxtdu. In brief, isolated islets from 25-week-old hIAPP 0/0 or hIAPP Tg/0 male mice on the FVB/NJ background were transplanted under the kidney into STZ-induced diabetic 8-week-old male BALB/cJ mice. (b) Diabetes incidence of male BALB/cJ mice with transplanted islets isolated from either hIAPP 0/0 control (black, open diamonds, n=11) or hIAPP Tg/0 control (green, open diamonds, n=8) mice. Data were analysed via a logrank test

Phagocytosis of IAPP aggregates by APCs decreases MHCII antigen presentation

To examine the mechanism of decreased antigen presentation in hIAPP-expressing mice, we modelled the direct effects of hIAPP on APCs by incubating BMDCs with either synthetic hIAPP or rIAPP. In contrast to rIAPP, hIAPP spontaneously forms IAPP aggregates in vitro, mimicking islet amyloid formation. hIAPP pre-treatment inhibited the lipopolysaccharide (LPS)-induced increase in high MHCII-expressing (MHCII^bright^) BMDCs (38.0 ± 0.8% vehicle + LPS vs 22.9 ± 1.7% hIAPP + LPS; Fig. 5a, b). This was not phenocopied by rIAPP (38.0 ± 0.8% vs 35.9 ± 1.2% rIAPP + LPS), indicating the reduction in MHCII^bright^ BMDCs was caused by IAPP aggregates rather than monomeric IAPP action. Similarly, hIAPP prevented the LPS-induced increase in the mean fluorescence intensity of MHCII on BMDCs (Fig. 5c, d). No differences in low MHCII-expressing (MHCII^dim^) BMDCs were observed between groups (Fig. 5e). To investigate whether IAPP aggregates reduced antigen presentation, we examined whether hIAPP pre-treated BMDCs could present Ova_323–339_ peptide as a model MHCII-restricted epitope to Ova-specific CD4^+^ T cells. Indeed, when hIAPP pre-treated BMDCs were pulsed with Ova peptide and then co-cultured with Ova-specific CD4^+^ T cells, antigen-specific T cell proliferation was dramatically reduced (Fig. 5f).Fig. 5IAPP aggregates disrupt MHCII antigen presentation. (a–e) Data from BMDCs pre-treated overnight with or without LPS in addition to vehicle (black), rIAPP (orange) or hIAPP (pink). (a) Representative flow cytometry plots showing MHCII fluorescence in treated BMDCs. (b) Frequency of treated BMDCs expressing high (MHCII^bright^) levels of MHCII. (c, d) Modal normalised count vs MHCII fluorescence (c) and median fluorescence intensity (MFI) of treated BMDCs (d). (e) Frequency of treated BMDCs expressing low (MHCII^dim^) levels of MHCII. (f) Proliferation (% of OT-II CD4^+^ cells) when co-cultured with BMDCs pre-treated with Ova_232–339_ peptide (Ova) and either vehicle (black) or hIAPP (pink). Data were analysed by one-way ANOVA. (g) Proliferation (% of OT-II CD4^+^ cells, normalised to Ova + vehicle) when co-cultured with BMDCs pre-treated with Ova_232–339_ peptide (Ova) and vehicle (black), rIAPP (orange), hIAPP (pink, no pattern), hIAPP and cytoD (pink, with pattern) or cytoD only (black, with pattern). Symbols of the same colour represent data generated within the same experiment. Data were analysed using two-way ANOVA. Data are shown as mean ± SD. *p<0.05. CytoD, cytochalasin D

Islet macrophages actively sample the islet microenvironment [10] and phagocytosis of IAPP aggregates by macrophages has been observed in islets [11, 13, 19] and causes proinflammatory activation of macrophages and DCs in vitro [15, 17, 18]. Thus, we tested whether hIAPP-impaired antigen presentation was phagocytosis-dependent. BMDCs were co-treated with both hIAPP and cytochalasin D, a reversible phagocytosis inhibitor, to prevent phagocytosis of hIAPP aggregates. Subsequently, BMDCs were washed and pulsed with Ova peptide antigen, then washed again and co-cultured with Ova-specific T cells. Indeed, phagocytosis inhibition during hIAPP exposure restored antigen presentation and normal levels of Ova-specific CD4^+^ T cell proliferation resulted (Fig. 5g). Furthermore, monomeric rIAPP had no impact on Ova-specific T cell proliferation, indicating that only IAPP aggregates disrupt antigen presentation. Collectively, these results show that direct phagocytosis of IAPP aggregates by APCs disrupts MHCII antigen presentation and subsequent activation of antigen-specific CD4^+^ T cells.

Discussion

The current study reveals that the phagocytosis of IAPP aggregates by APCs inhibits MHCII antigen processing and presentation, resulting in delayed autoimmune-mediated diabetes. The findings of this paper are strengthened using multiple mouse models to triangulate the consistent effect of amyloidogenic IAPP on APCs, including an unbiased scRNA-seq approach on an F1 FVB/NJ × C57BL/6J genetic background, and similar protective phenotypes in two independent strains of congenic hIAPP-expressing NOD mice. This controls for artefacts that could result from overexpression of the hIAPP transgene or the proximity of any Idd genes around the hIAPP transgene insertion site (chromosome 15) or Iapp locus (chromosome 6, Iapp also being an Idd locus itself and important for the formation of a hybrid insulin peptide antigen, HIP6.9) [36–39]. In vitro studies allowed us to confirm a direct effect of IAPP aggregates on MHCII antigen presentation. Supporting an aggregation-dependent effect, a previous study found deletion of the endogenous, non-aggregating Iapp gene in NOD mice did not alter diabetes progression [38].

While scRNA-seq analysis was conducted in Cd14-expressing islet myeloid cells, which are consistent with islet macrophages [1, 40], and flow cytometry confirmed reduced MHCII in islet macrophages from NOD mice, functional assays were not performed in macrophages. The paucity of islet macrophages creates technical challenges to directly test the functional effects of IAPP aggregates on islet macrophage antigen presentation, and therefore we made the assumption that the effects observed in BMDCs will generalise to islet macrophages. There is ample literature demonstrating the proinflammatory effects of IAPP aggregates are consistent in BMDCs, bone marrow-derived macrophages (BMDMs) and islet macrophages alike, and evidence that all are capable of phagocytosing IAPP aggregates [11, 13, 15, 18, 19]. As islet macrophages are unique amongst tissue-resident macrophages in their haematopoietic origin, basal proinflammatory state, surface expression of the DC marker CD11c and effective APCs, we selected BMDCs to model antigen presentation by these cells. Furthermore, the transcriptomic evidence of decreased antigen presentation and increased phagocytosis from islet myeloid cells supports the in vitro findings. An additional limitation of this study is that we did not confirm our results in human cells, which would be valuable in future studies as islet amyloid has recently been found in pancreases from people with type 1 diabetes [20–22]. Finally, the expression of hIAPP delays but does not prevent diabetes onset. Diabetes incidence was monitored up to 30 weeks of age, but with the median onset of diabetes ranging from 28 to 30 weeks in hIAPP-expressing mice, diabetes incidence is expected to continue to increase beyond this age.

The protective effect of hIAPP in the NOD mouse model of type 1 diabetes was surprising given the extensive evidence that IAPP aggregates induce islet inflammation, impair beta cell function and contribute to beta cell failure and loss in type 2 diabetes (reviewed in [41]). IAPP aggregates interact directly with macrophages through TLR2 and phagocytosis and induce proinflammatory cytokine secretion which contributes to beta cell dysfunction [11, 13, 15, 17–19]. Indeed, we recently showed that targeting IAPP aggregates specifically with a monoclonal antibody improves the function of human islet transplants in mice, again highlighting the deleterious effects of islet amyloid on beta cell function [42]. The induction of beta cell dysfunction by islet amyloid is evident in both the presence and absence of autoimmunity, confirmed here by glucose intolerance in both hIAPP Tg/0 C57BL/6J × FVB/NJ and prediabetic NOD.hIAPP Tg/0 mice. The impact on islet inflammation was also still apparent in the absence of autoimmunity, evidenced by the increased CD45^+^ cells in islets of hIAPP Tg/0 C57BL/6J × FVB/NJ mice and the acceleration of allogenic islet transplant rejection in hIAPP Tg/0 islets. Interestingly, the mechanism by which IAPP aggregates induce inflammation, particularly the induction of IL-1β secretion, is dependent upon phagocytosis of IAPP aggregates and subsequent disruption of the lysosomes [15, 17, 18]. Since MHCII antigen processing occurs throughout the phagolysosomal system, we speculate that the disruption of lysosomes by IAPP aggregates may directly interfere with MHCII antigen processing. In addition, IAPP aggregate phagocytosis may also cause a homeostatic decrease in the MHCII pathway, as evidenced by the reduction of Ciita expression in islet myeloid cells. Although MHCII processing genes were broadly downregulated, the use of Ova_323–339_ peptide in antigen presentation studies does not measure antigen processing, as it can directly load MHCII on the cell surface.

We propose the following model to account for the protective effect of IAPP aggregates in autoimmune diabetes in the face of extensive evidence, including our own, that IAPP aggregates elicit islet inflammation and beta cell loss in type 2 diabetes. The phagocytosis of IAPP aggregates by islet macrophages elicits two opposing effects: (1) increased proinflammatory cytokine secretion which promotes islet inflammation and beta cell dysfunction; and (2) decreased MHCII antigen presentation which reduces induction of adaptive immune responses. The outcome of these opposing effects on immunity, regarding beta cell protection or loss, depends on whether this occurs in the presence of islet autoimmunity. When autoimmunity is present, decreased antigen presentation outweighs the induction of inflammation, delaying diabetes. In the absence of autoimmunity, as in models of type 2 diabetes, antigen presentation is inconsequential, and the islet inflammation drives beta cell dysfunction and loss. Similarly, in allogeneic islet transplants where multiple pathways to alloantigen recognition exist, decreased MHCII antigen presentation does not delay rejection and the deleterious impact of islet amyloid is apparent. Collectively, our data reveal a surprising impact of islet amyloid on autoimmune diabetes in mice and beg the question, what is the effect of islet amyloid in people living with type 1 diabetes?

Supplementary Information

Below is the link to the electronic supplementary material.ESM (PDF 1678 KB)