Limited extent and consequences of pancreatic SARS-CoV-2 infection

Abstract

Concerns that infection with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the etiological agent of coronavirus disease 2019 (COVID-19), may cause new-onset diabetes persist in an evolving research landscape, and precise risk assessment is hampered by, at times, conflicting evidence. Here, leveraging comprehensive single-cell analyses of in vitro SARS-CoV-2-infected human pancreatic islets, we demonstrate that productive infection is strictly dependent on the SARS-CoV-2 entry receptor ACE2 and targets practically all pancreatic cell types. Importantly, the infection remains highly circumscribed and largely non-cytopathic and, despite a high viral burden in infected subsets, promotes only modest cellular perturbations and inflammatory responses. Similar experimental outcomes are also observed after islet infection with endemic coronaviruses. Thus, the limits of pancreatic SARS-CoV-2 infection, even under in vitro conditions of enhanced virus exposure, challenge the proposition that in vivo targeting of β cells by SARS-CoV-2 precipitates new-onset diabetes. Whether restricted pancreatic damage and immunological alterations accrued by COVID-19 increase cumulative diabetes risk, however, remains to be evaluated.

Keywords: COVID-19; SARS-CoV-2; human coronaviruses; human islets; pancreas; type 1 diabetes; type 2 diabetes; viral infection.

Graphical abstract

Figure 1

Productive but limited SARS-CoV-2 infection of human islet cell populations (A) Left: gating strategy for identification of α (GCG⁺), β (INS⁺), and “other” (GCG⁻INS⁻) cells; adjacent plots depict SARS-CoV-2 NP staining of the indicated subsets in 48-h mock- and SARS-CoV-2-infected samples. Right: frequencies of SARS-CoV-2 NP⁺ islet cell subsets (n = 7 donors). (B) SARS-CoV-2 infection of all islet cells and donor stratification according to higher (>2.5% of total cells) and lower (<2.5%) extent of infection. (C) Infectious virus titers in TCS (initial inoculum, 4 × 10⁴ plaque-forming units (PFUs)/mL; 48-h culture; dotted line, limit of detection [LOD] = 67 PFU/mL; n = 6 donors). (D) SARS-CoV-2 infection of islet cell subsets after 48-h and 120-h culture (n = 4 donors). (E and F) Quantification of secreted proteins in UV-inactivated TCS of 48-h mock- or SARS-CoV-2-infected samples (NPX, normalized protein expression; n = 6 donors). (G) Gating strategy for distinction of live and dead islet cells. (H) Frequencies of live and dead SARS-CoV-2 NP⁺ islet cell subsets (48-h culture, n = 4 donors). (I) Survival of islet cells distinguished according to infection condition and absence/presence of viral NP (n = 4 donors). (J) INS expression levels (geometric mean of fluorescence intensity [GMFI]) by NP− and NP⁺ beta cells were normalized to respective INS GMFI values in donor-matched mock-infected cultures (n = 6 donors). All summary bar diagrams represent mean ± SD and scatter for indicated number of donors; statistical analyses were conducted by paired t test or repeated measures ANOVA with Tukey’s multiple comparisons (asterisks) or non-parametric Friedman test (hashtags) where applicable (∗ or #, p < 0.05; ∗∗ or ##, p < 0.01; ∗∗∗, p < 0.001). All summary bar diagrams represent mean ± SD and scatter for the indicated number of donors; statistical analyses were conducted by paired t test or repeated-measures ANOVA with Tukey’s multiple comparisons where applicable (#, non-parametric Friedman test).

Figure 2

Stringent ACE2 requirement for pancreatic islet cell infection with SARS-CoV-2 (A) Representative contour plots gated on live α, β, and “other” cells pre-treated with IgG (irrelevant polyclonal goat antibody AF7197) or the anti-ACE2 blocking antibody AF933 prior to SARS-CoV-2 infection (48 h). (B) Summary of SARS-CoV-2 NP expression by live islet cell subsets as a function of IgG treatment or ACE2 blockade (n = 6 donors). (C) Percent infection inhibition for β and “other” cells (inhibition for α cells is not shown because the very low extent of α cell infection in IgG-treated cultures for 2 of 6 donors substantially skews such calculations). (D) Infectious SARS-CoV-2 titers and extent of infection inhibition following ACE2 blockade (n = 3 donors). (E) Quantification of chemokines and cytokines in UV-inactivated TCS of SARS-CoV-2-infected islet cell cultures under conditions of IgG treatment or ACE2 blockade (48-h infection, n = 3 donors). (F) Infectious SARS-CoV-2 titers in TCS as a function of glucose concentration in islet culture medium (n = 3 donors). (G) Quantification of CXCL10 and CXCL11 in TCS as a function of glucose concentration. All summary bar diagrams represent mean ± SD and scatter for the indicated number of donors; statistical analyses were conducted by paired t test or repeated-measures ANOVA with Tukey’s multiple comparisons where applicable (∗, p < 0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001). All summary bar diagrams represent mean ± SD and scatter for the indicated number of donors; statistical analyses were conducted by paired t test or repeated-measures ANOVA with Tukey’s multiple comparisons where applicable.

Figure 3

scRNA-seq analysis of pancreatic islets infected in vitro with SARS-CoV-2 (A) UMAP (Uniform Manifold Approximation and Projection) of 21,728 single cells integrated across samples from 3 donors, each infected or mock-infected with SARS-CoV-2. Points are colored by assigned cell type (left) or assigned infection state (right). (B) Relative cell type composition of each sample, with total number of cells per sample indicated at the top. (C) Number of infected cells per donor (SARS-CoV-2-infected samples only) by cell type. (D) Differentially expressed genes in pairwise comparisons between infected, bystander, or mock-infected β cells; selected genes have an adjusted p value of less than 0.01 and absolute log2 fold change greater than log2(1.5) in any contrast within any donor. Dot size indicates the percentage of β cells in the indicated donor and infection state expressing the gene; dot color indicates the average expression of that gene in β cells scaled across infection state within each donor. Filled boxes (below) indicate gene membership in the indicated hallmark gene sets. For infected cells, additional annotation (right) indicates the percentage of β cells infected per donor, and the violin plot indicates percent of all detected transcripts identified as viral transcripts per cell. Because of insufficient numbers of infected β cells in donor 1, this sample was excluded from differential expression contrasts involving infected cells.

Figure 4

MC-based pancreatic cell annotation and quantification of SARS-CoV-2 infection (A) Hierarchical gating strategy for identification of major pancreatic cell subsets. Starting with a distinction of α (GCG⁺) and β (proinsulin [proINS]⁺) cells (top row, center), the panel represents a flow chart where demarcated regions and associated arrows connect successive plots to designate and characterize pancreatic cell populations (PPY, pancreatic polypeptide; SST, somatostatin; GHRL, ghrelin). Regions, arrows, and cell type names rendered in blue highlight final cell annotations. Red regions/arrows refer to subsets selected for further phenotypic stratification. The bottom left and right plots with a gray background feature comparative α and β cell properties. (B) Left: tSNE visualization of pancreatic cell populations in 48-h mock- and SARS-CoV-2-infected samples. Right: relative abundance of non-hematopoietic pancreatic cell subsets in mock- and SARS-CoV-2-infected samples (n = 5 donors). (C) Left: co-detection of SARS-CoV-2-S/NP in live non-hematopoietic cells. Center: distribution of viral burden visualized by projection onto tSNE plots. Right: α-like, β-like, and unannotated cell projections onto tSNE clusters and associated viral burden (viral infection of other cell types are excluded here). (D) Representative SARS-CoV-2-S and -NP expression in mock- and SARS-CoV-2-infected pancreatic cell types (48-h culture). (E) Extent of SARS-CoV-2 infection in all pancreatic cell types (dotted line, average infection extent for all non-hematopoietic cells; n = 5 donors; asterisks indicate statistical differences calculated between mock- and SARS-CoV-2-infected samples by paired t test; ∗, p < 0.05; ∗∗, p < 0.01). (F) Relative distribution of the viral burden across pancreatic cell subsets.

Figure 5

Phenotypic alterations of SARS-CoV-2-infected islet cells (A) Normalized hormone expression by α, β, γ, and δ cells, comparing respective GMSI (geometric mean of signal intensity) values in S/NP⁺ (red) and S/NP⁻ (red/gray hatched) cell fractions of infected cultures with matched populations from mock-infected cultures (gray). (B) MC contour plots are gated on S/NP⁺ α cells (top) or β cells (bottom) and depict expression of viral S protein (left), respective hormone content (center), and HLA-ABC expression levels (right) as a function of viral NP expression level (NP^int versus NP^hi). (C) Modulation of HLA-ABC expression levels in major pancreatic cell subsets in response to SARS-CoV-2 infection. (D) Identification of ductal cells as the major HLA-DR-expressing pancreatic cell subset, distribution of the viral burden in relation to HLA-ABC/HLA-DR expression (values indicate percentage HLA-DR⁺ cells), and summary of HLA expression regulation as a function of SARS-CoV-2 infection and viral S/NP expression. (E) Phenotypic alteration of ACs and α cells by SARS-CoV-2 infection. (F) β Cell TF expression across mock-infected and infected S/NP⁻ and S/NP⁺ populations. All bar diagrams represent mean ± SD and scatter of 4–7 donors (B–E) or 3–4 donors (F); statistics were calculated by repeated-measures ANOVA (∗, p<0.05; ∗∗, p < 0.01; ∗∗∗, p < 0.001).

Figure 6

Permissiveness of human islet cells to endemic coronavirus and LCMV infection (A) Left: contour plots are gated on live α, β, and “other” cell subsets from 96-h mock- or HCoV-OC43-infected islet cell cultures. Right: summary of HCoV-OC43 NP expression in live and dead islet cell subsets (n = 4 donors). (B) GCG and INS expression levels across mock-infected, NP⁻, and NP⁺ α or β cells. (C) MC analyses were conducted with 96-h mock- or HCoV-OC43-infected islet cells. HCoV-OC43 NP staining and tSNE visualization of live and dead mock-infected (left) and SARS-CoV-2-infected (right) non-hematopoietic cells; overlaid red dots (corresponding to the red demarcated regions in the respective left contour plots) indicate background HCoV-OC43 NP staining for mock cultures and HCoV-OC43 NP detection for infected cultures. (D and E) 96-h mock or HCoV-NL63 infection of islets; data are displayed as in (A) and (B) (n = 4 donors). (F and G) 72-h mock or LCMV infection of islets; contour plots gated on live cells show the viral burden in α, β, and “other” cells; dot plots indicate relative survival/death of β cells, and histograms compare GCG and INS levels in the respective uninfected, NP⁻, and NP⁺ fractions of α or β cells. All summary diagrams represent mean ± SD and/or scatter for the indicated numbers of donors; statistics were calculated by paired t test or repeated-measures ANOVA with Tukey’s multiple comparisons where applicable (∗, p < 0.05; ∗∗, p < 0.01).