Enterovirus VP1 protein and HLA class I hyperexpression in pancreatic islet cells of organ donors with type 1 diabetes

Abstract

Aims/hypothesis: Earlier studies of pancreases from donors with type 1 diabetes demonstrated enteroviral capsid protein VP1 in beta cells. In the context of a multidisciplinary approach undertaken by the nPOD-Virus group, we assessed VP1 positivity in pancreas and other tissues (spleen, duodenum and pancreatic lymph nodes) from 188 organ donors, including donors with type 1 diabetes and donors expressing autoantibody risk markers. We also investigated whether VP1 positivity is linked to the hyperexpression of HLA class I (HLA-I) molecules in islet cells.

Methods: Organ donor tissues were collected by the Network for Pancreatic Organ Donors with Diabetes (nPOD) from donors without diabetes (ND, n=76), donors expressing a single or multiple diabetes-associated autoantibodies (AAb⁺, n=20; AAb⁺⁺, n=9) and donors with type 1 diabetes with residual insulin-containing islets (T1D-ICIs, n=41) or only insulin-deficient islets (T1D-IDIs, n=42). VP1 was assessed using immunohistochemistry (IHC) and HLA-I using IHC and immunofluorescence, in two independent laboratories. We determined assay concordance across laboratories and overall occurrence of positive assays, on a case-by-case basis and between donor groups.

Results: Islet cell VP1 positivity was detected in most T1D-ICI donors (77.5%) vs only 38.2% of ND donors (p<0.001). VP1 positivity was associated with HLA-I hyperexpression. Of those donors assessed for HLA-I and VP1, 73.7% had both VP1 immunopositivity and HLA-I hyperexpression (p<0.001 vs ND). Moreover, VP1⁺ cells were detected at higher frequency in donors with HLA-I hyperexpression (p<0.001 vs normal HLA-I). Among VP1⁺ donors, the proportion with HLA-I hyperexpression was significantly higher in the AAb⁺⁺ and T1D-ICI groups (94.9%, p<0.001 vs ND); this was not restricted to individuals with recent-onset diabetes. Critically, for all donor groups combined, HLA-I hyperexpression occurred more frequently in VP1⁺ compared with VP1^- donors (45.8% vs 16%, p<0.001).

Conclusions/interpretation: We report the most extensive analysis to date of VP1 and HLA-I in pancreases from donors with preclinical and diagnosed type 1 diabetes. We find an association of VP1 with residual beta cells after diagnosis and demonstrate VP1 positivity during the autoantibody-positive preclinical stage. For the first time, we show that VP1 positivity and HLA-I hyperexpression in islet cells are both present during the preclinical stage. While the study of tissues does not allow us to demonstrate causality, our data support the hypothesis that enterovirus infections may occur throughout the natural history of type 1 diabetes and may be one of multiple mechanisms driving islet cell HLA-I hyperexpression.

Keywords: Autoimmunity; Enterovirus; HLA-I molecules; Immunofluorescence; Immunohistochemistry; Pancreas; Type 1 diabetes.

Fig. 1

Distribution of pancreas samples, summary of the types of assays performed and the number of donors assessed/assay or assay combination. FFPE sections were provided to laboratories in Exeter (UK) and Tampere (Finland) and frozen sections to La Jolla (USA) and Munich (Germany). For both enterovirus VP1 and HLA-I assays, serial sections (enterovirus VP1) or sections from comparable pancreas regions (HLA-I) from a subset of donors were sent to Exeter/Tampere laboratories or to Exeter and La Jolla/Munich laboratories, respectively. All spleen, PLN and duodenum FFPE sections were distributed to Tampere and a subset of spleen samples (n=18) were sent to Exeter (ESM Tables 1 and 2). IF, immunofluorescence; IHC, immunohistochemistry; LJ, La Jolla; OCT, optimal cutting temperature compound

Fig. 2

(a) Representative images of enterovirus VP1 staining in serial sections of an islet from nPOD donor 6084. The purple outline defines the islet. (b) Agreement of enterovirus VP1 immunostaining between laboratories. Samples taken from the same pancreatic block were evaluated in 91 donors in Exeter (Lab 1) and Tampere (Lab 2). Negative (N) or positive (P) agreement are coloured electric blue, and discordant results are displayed in black. A total of 48 samples were found to be negative for VP1 in both laboratories while 31 were positive. Three were negative in Exeter and positive in Tampere, whereas nine were positive in Exeter and negative in Tampere. (c) Comparison of blinded VP1 immunohistochemistry in two different laboratories (Lab 1, Exeter; Lab 2, Tampere) performed on serial sections. nPOD donor numbers are shown along the bottom. Grey, VP1⁻; pink, VP1⁺

Fig. 3

(a) Enterovirus VP1⁺ islet cells are more frequently observed in T1D-ICI donors. The bar graph shows significantly higher VP1 immunopositivity in the pancreas of T1D-ICI donors compared with ND, AAb⁺ and T1D-IDI donors (***p<0.001, two-sided Fishers Exact Test; significant after FDR corrections for multiple comparisons). (b) The frequency of enterovirus VP1⁺ islet cells is increased in donors with type 1 diabetes and residual beta cells (T1D-ICI). AI-based image analysis was performed on a subset of donor pancreas sections stained in Exeter to determine the proportion of islet cells that were VP1⁺ within each pancreatic section. Each data point represents an individual donor. The median, first and third quartile for the percentage of VP1⁺ islet cells are shown. *p<0.05, ***p<0.001 (one-way ANOVA with Kruskal–Wallis post-test). (c–g) Enterovirus VP1 is found more frequently in islet cells than in acinar cells in all donor groups. AI-based image analysis was performed on a subset of donor pancreas sections stained in Exeter to determine the proportion of islet or acinar cells that were VP1⁺. Each data point represents an individual donor with a line linking the proportion of islet and acinar VP1⁺ cells. Blue (c), ND (n=39); green (d), AAb⁺ (n=18); yellow (e), AAb⁺⁺ (n=8); orange (f), T1D-ICI (n=33); red (g), T1D-IDI (n=18). *p<0.05, **p<0.01, ***p<0.001 (Wilcoxon matched-pairs test)

Fig. 4

(a) Agreement of HLA staining by region of pancreatic sample. HLA staining from 118 donors was performed in Exeter and La Jolla/Munich laboratories using FFPE and optimal cutting temperature compound (OCT) frozen samples, respectively. Of those, samples were examined from the same block in 66 cases, a different block but the same region (i.e. head, body or tail) in 47 cases, different regions in three cases and unspecified blocks or regions in two cases. Normal expression, elevated expression or hyperexpression was noted by both laboratories. Complete agreement is coloured in electric blue, complete disagreement in black and partial agreement in violet. (b) Comparison of blinded HLA-I immunostaining in two different laboratories (Lab 1, Exeter; Lab 2, La Jolla/Munich) on FFPE vs OCT material respectively. Blue, normal expression of HLA-I; yellow, elevated expression of HLA-I; pink, hyperexpression of HLA-I. (c) HLA-I hyperexpression was more frequently observed in T1D-ICI and AAb⁺⁺ donors compared with ND and T1D-IDI donors. Blue bars, proportion of donors with normal expression; yellow bars, elevated expression; pink bars, hyperexpression ***p<0.001 (two-sided Fishers Exact Test; two-sided significance after FDR corrections for multiple comparisons). (d) Donors with type 1 diabetes and HLA-I hyperexpression had overall shorter disease duration than those categorised as having elevated and normal expression. Each data point shows individual donor values for disease duration (in years) in each of the three HLA-I expression categories (normal, elevated and hyperexpression). The median, first and third quartile are shown. **p<0.01, ***p<0.001

Fig. 5

(a) Among VP1⁺ donors, HLA-I hyperexpression is more frequently observed in T1D-ICI and double AAb⁺⁺ donors compared with ND donors. Bar graphs show the proportion of donors that are VP1⁺ and have normal expression (blue), elevated expression (yellow) or hyperexpression of HLA-I (pink) (N=72; ND n=24; AAb⁺ n=5; AAb⁺⁺ n=4; T1D-ICI n=29; T1D-IDI n=10). *p<0.05, **p<0.001, ***p<0.001 (Kruskal–Wallis test). (b) HLA-I expression is more frequently normal in VP1⁻ ND, AAb⁺ and T1D-IDI donors (N=75; ND n=30; AAb⁺ n=14; AAb⁺⁺ n=5; T1D-ICI n=9; T1D-IDI n=17). **p<0.01, ***p<0.001; two-sided Fishers Exact Test; two-sided significance after FDR corrections for multiple comparisons). (c–g) Dot plots show the proportion of VP1⁺ islet cells in relation to HLA-I expression category in each donor group and for all donors combined. The median, first and third quartile are shown. Blue (c), ND; green (d), AAb⁺/AAb⁺⁺; orange (e), T1D-ICI; red (f) T1D-IDI; (g) all donors. *p<0.05, **p<0.001, ***p<0.001 (Kruskal–Wallis test)