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 [12,13,14, 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, 13,14,15], 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. 2

hIAPP 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 (Iapph/h) NOD/ShiLtJ backcrossing and diabetes incidence. Diabetes incidence of wild-type (NOD.Iappm/m) (black, solid circle, n=15), heterozygous (NOD.Iapph/m) (pink, open triangle, n=15) and homozygous knockin (NOD.Iapph/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.PrkdcScid/Scid recipient mice. (l) Diabetes incidence curve of NOD.CG.PrkdcScid/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.Iapph/h), and littermate controls homozygous or heterozygous for wild-type mouse IAPP (NOD.Iappm/m and NOD.Iapph/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.Iapph/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.PrkdcScid/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. 3

NOD.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. 4

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. 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 (MHCIIbright) 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 MHCIIbright 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 (MHCIIdim) BMDCs were observed between groups (Fig. 5e). To investigate whether IAPP aggregates reduced antigen presentation, we examined whether hIAPP pre-treated BMDCs could present Ova323–339 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. 5

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 (MHCIIbright) 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 (MHCIIdim) levels of MHCII. (f) Proliferation (% of OT-II CD4+ cells) when co-cultured with BMDCs pre-treated with Ova232–339 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 Ova232–339 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.