EndoC-βH1 cells, obtained from Dr. Raphael Scharfmann (Paris Descartes University, France) (20), were maintained as described previously (21). ER stress was induced 100nM Thapsigargin (TG; Adipogen Life sciences, Fuellinsdorf, Switzerland) for 24h.

For stimulation of nucleic acid-sensing pathway in EndoC-βH1, we transfected the cells with either polyinosinic-polycytidylic acid (poly(I:C); 5μg/ml) to mimic double stranded RNA (dsRNA) or fluorescein amidite (FAM)-labelled double stranded DNA (dsDNA) (derived from the HSV-1 genome; 1μg/ml) (22) to mimic exogenous DNA using lipofectamine 2000 (Invitrogen, Germany) as per manufacturer’s instructions. Cells were harvested 24h post-transfection and transfection efficiency was measured by either flow cytometry or gene expression determined by RT-qPCR. PolyI:C transfection efficiency was determined after staining with an anti-dsRNA antibody J2 (dilution 1:100) (Jena Bioscience) and Alexa fluro-647 labeled GAM secondary antibody. dsDNA transfection efficiency was determined using FAM fluorescence of the transfected dsDNA.

MDA5 downregulation was achieved by siRNA-smartpool (M-013041-00-0005; Horizon Discovery, Cambridge, UK) transfection using DharmaFECT 1 Transfection Reagent (Horizon Discovery) as per manufacturer’s instructions for 48h, followed by exposure to TG (0.1µM; 24h). The knockdown was analyzed RT-qPCR using gene specific primers.

cGAS knockdown EndoC-βH1 line was generated by lentiviral transduction of shRNA against cGAS (Sigma; TRCN0000146282) or negative control (Sigma; SHC002). Lentiviruses were produced as described before (23). EndoC-βH1 cells were cultured a day before transduction. Cells were incubated with virus supernatant and polybrene (8μg/ml) for 24h. After incubation, the virus medium was changed with fresh medium and the cells were incubated for 3 days, followed by puromycin (3μg/ml) selection. cGAS knockdown was analyzed RT-qPCR.

The cells were collected after treatment and fixed with 4% Paraformaldehyde solution containing 0.1% Saponin for 20 min at 4°C; followed by three washings with wash buffer (phosphate-buffered saline (PBS), 0.1% Saponin), blocking with PBS, 0.2% BSA, 0.1% Saponin; 10min at 4°C and incubation with anti-dsRNA monoclonal J2 antibody (1:500; diluted in blocking buffer; 30 min at 4°C)(Sigma-Aldrich, Amsterdam, Netherlands). The cells were washed three times with wash buffer and incubated with Alexa-fluro-647 labeled Goat anti-mouse antibody (1:500; diluted in blocking buffer)(Invitrogen, Karlsruhe, Germany) for 30 min at 4°C; followed by washing with PBS and acquisition on LSRII flow cytometer (BD Biosciences, San Jose, CA, USA). Analysis of flow cytometry data was performed using Flowjo software (BD, Ashland, USA).

RNA was isolated by using an RNA isolation kit (Bioké, Leiden, Netherlands), according to the manufacturer’s protocol. cDNA was synthesized from 500-1000ng of RNA, using Oligo-dT primer and Superscript II Reverse Transcriptase (Invitrogen); 5ng of cDNA was used per 10μl reaction of Reverse transcription polymerase chain reaction (RT-qPCR), containing 5 pmol of gene-specific primers (see Table 1 ) and iQ SYBR^® Green Supermix (Bio-Rad, Veenendaal, Netherlands). Assays were run on CFX Connect Real-Time PCR Detection System (Bio-Rad). Data are presented as 2 ^-ΔΔCT (24).

The amount of cytosolic mtDNA leakage was determined by quantification of mtDNA in the cytosolic extract (25). Briefly, the cells were collected after stimulation with TG for 24h, resuspended in 1% NP-40 buffer, incubated for 15min on ice and spin down at 13,000 x g for 15min at 4°C. The supernatant (cytosolic extract) was used to isolate cytosolic DNA using DNeasy Blood & Tissue Kit (Qiagen) and the pellet was used to isolate total DNA. Cytosolic mtDNA was quantified in DNA isolated from the cytosolic extract by RT-qPCR using mitochondrial DNA specific gene tRNA leucine 1 (tRNA-LeuUUR) and normalized by nuclear gene (β2 microglobulin) in total DNA (26). Ct values for cytosolic mtDNA was normalized with Ct values of total DNA and relative mtDNA were calculated and plotted.

Real-time analysis of the oxygen consumption rate (OCR) was measured in EndoC-βH1 cells using a XF-96 Extracellular Flux Analyzer (Seahorse Bioscience). In short, cells were seeded in an XF-96 cell culture plate (Seahorse Bioscience) (30,000 cells/well), grown to confluence and next treated with 0.1uM TG for 24h or left unstimulated. After stimulation, the cells were washed and analyzed in XF culture medium consisting of unbuffered RPMI (Sigma-Aldrich) supplemented with 5% FCS (Serana), 100U/ml penicillin/streptomycin (Sigma-Aldrich) and 2mM L-glutamine (Sigma-Aldrich). Real-time changes in OCR were measured as described by the manufacturer’s protocol in response to 10mM D-glucose (Sigma-Aldrich), 1.5µM oligomycin (Cayman Chemical), 1µM fluoro-carbonyl cyanide phenylhydrazone (FCCP, Sigma-Aldrich) and a combination of 1µM Rotenone and 1µM Antimycin A (Sigma-Aldrich).

To analyze the cytosolic accumulation of dsDNA, the EndoC-βH1 cells were incubated with an anti-dsDNA antibody (ab27156, Abcam, Cambridge, UK). Briefly, the cells were washed with PBS, fixed and permeabilized with a methanol/acetone (70%/30%) solution at -20°C for 10 min, blocked for 1h with 5% skim milk dissolved in PBS at room temperature and incubated with the anti-dsDNA antibody (1:1000; 5% milk in PBS) overnight at 4°C. Next, the cells were washed thrice with PBS and incubated with anti-mouse IgG2a-Alexa fluor-568 antibody (1:1000; 5% skim milk in PBS; Invitrogen) for 1h at room temperature; followed by washings with PBS, mounting with vectashield mounting medium (Vector Laboratories, Newark, USA) and analyzed under Leica SP8 WLL (Leica Microsystems, Amsterdam, Netherlands) confocal microscope with 40x oil immersion objective.

For colocalization of dsDNA with mitochondria, the cells were sequentially co-stained with anti-dsDNA and anti-mitochondria antibody (ab92824; Abcam). First, the cells were stained with primary anti-dsDNA and secondary anti-mouse IgG2a-alexa fluor-568 antibody (Invitrogen) as described above; followed by staining with primary anti-mitochondria antibody (1:1000 in 5% skim milk dissolved in PBS) and secondary antibody anti-mouse-IgG1-Alexa fluor 488. Staining with anti-mitochondria were performed with an incubation time of 1h at room temperature. The cells were analyzed in Leica SP8 WLL (Leica Microsystems) confocal microscope with a 40x oil immersion objective.

The cytosolic dsDNA quantification was performed as previously described (27) using FIJI software (28). Briefly, the total fluorescence of the cells was measured by tracing the outline of a cell and defined as “X”. Similarly, the fluorescence of the nucleus of the same cell was measured and defined as “Y”. The rate of cytosolic dsDNA was measured and plotted as (X-Y)/Y. The quantification was done on 20-25 cells per condition from different randomly selected images of the same condition.

The apoptosis on stimulation with chemicals was analyzed by caspase-3/7 activity using the Caspase-Glo^® 3/7 Assay (Promega, Leiden, Netherlands), as per the manufacturer’s instructions.

Adenine nucleotide concentrations were determined in cell extracts by high-performance liquid chromatography, as described previously (29). Briefly, EndoC-βH1 cells (1 x 10⁶ cells) were treated as described in the figure legend, then medium was removed and cells were scrapped into 6% (v/v) ice-cold HClO₄. After centrifugation (10 000 x g for 2 minutes at 4°C), the supernatant was collected, neutralized and stored at -80c until subsequent determination of ATP and ADP.

Neutrophils were isolated from venous blood/buffy coats purchased from Sanquin (The Netherlands) using the Ficoll-Paque density gradient. Briefly, transverse EDTA blood was mixed with PBS (1:1), added to Ficoll-Paque density gradient media (GE Healthcare, IL,USA), and centrifuge at 600 x g for 30min at room temperature with lowest acceleration and brake. The bottom layer of red blood cells (RBCs) and granulocytes were collected, and RBCs were subsequently lysed with sterile MilliQ water for 20sec, washed with PBS and resuspended in RPMI medium (Invitrogen) supplemented with 10% FBS (Invitrogen) and pen/strep (Invitrogen). The neutrophils were counted and subsequently used in a migration assay. The migration assay was performed using 6.5 mm Transwell with 3 μm PVP-free polycarbonate filter membrane (Corning, Amsterdam, Netherlands) in a 24-well micro chemotaxis chamber. Neutrophils were labelled with CMTMR-red (C34552; Invitrogen) dye as per the manufacturer’s instructions. Labelled neutrophils (1x 10⁵/100µl) were added in the upper chamber and incubated with 250μl conditioned medium from stimulated EndoC-βH1 cells in the lower chamber for 2h at 37°C. After incubation, the cells were collected from the lower chamber and counted by flow cytometry. Prior to the acquisition by flow cytometry, an equal number of unlabeled beads (BD biosciences) were added per sample which were used to normalize the events measured.

Pancreatic islets were obtained from human organ donor pancreata. Human islets were isolated from organ donors. Islets were only studied if they could not be used for clinical purposes and if research consent had been obtained. According to the national law ethics approval is not required for research on donor tissues that cannot be used for clinical transplantation. The isolations were performed in the GMP-facility of LUMC. For experimental use, human islets were maintained in ultra-low attachment plates (Corning, NY 14831) in low glucose DMEM supplemented with 10% FBS, 100 units/ml Penicillin and 100 μg/ml streptomycin and exposed to 1µM thapsigargin for 5h. After treatment, culture medium was replaced and analyzed for cytokine profile 2 days after using Luminex 9-plex kit (BioRad) according to the manufacturer’s protocol. Islet cells were pelleted and RNA isolated to correct Luminex values. Human IL8 secretion from EndoC-βH1 cells exposed to TG was determined by ELISA MAX™ Deluxe Set Human IL-8 (Biolegend) according to the protocol from the manufacturer.

All graphs and statistical analysis were generated using GraphPad Prism 9.01 (GraphPad Software). Data are presented as the mean ± SEM from at least 3 independent experiments. Two-tailed Student’s t-test was performed for mean comparison and statistical significances estimation of two groups. One-way ANOVA was used for multiple comparisons, with Holm-Sidak’s multiple comparisons for significances estimation. Statistical significance was set at p < 0.05.

In order to determine whether human beta-cells have functional machinery for sensing foreign nucleic acids, EndoC-βH1 cells were transfected with fam-labelled dsDNA ( Figure 1A ) or Poly-IC (to mimic dsRNA), an experimental approach that has been previously reported and validated (30, 31) ( Figure 1B ). Forty eight hours post transfection, the expression of genes involved in DNA/RNA sensing machinery and in type-I interferon response was assessed by qPCR. As depicted on Figures 1C, D , the presence of exogenous DNA and RNA triggered the expression of the cytosolic nucleic acid sensors cGAS and MDA5, respectively, leading to upregulation of IFN-I related genes such as interferon-beta (IFNβ) ( Figure 1E ) and interleukin-8 (IL8) ( Figure 1F ).

Several evidences suggest that cellular stress could initiate an IFN-I response (32–34). To test this hypothesis, EndoC-βH1 cells were treated for 24h with thapsigargin (TG), a Ca²⁺ ATPase inhibitor known as a potent ER stress inducer. In these conditions, TG significantly increased the expression of ER stress markers, such as Binding immunoglobulin protein (BIP) and X-box-binding protein 1spliced variant (XBP-1s) ( Figures 2A, B ), and also led to upregulation of cGAS and MDA5 ( Figures 2C, D and Figure S1 ), as previously described (35), as well as IFN-I responsive genes, including IFNβ, (Interferon Induced Protein With Tetratricopeptide Repeats 1 (IFIT1), Interferon Alpha Inducible Protein 27 (IFI27), and IL8 ( Figures 2E–H ). Altogether, these results suggest that Ca²⁺-dependent ER stress could trigger the innate immune response secondary to cytosolic accumulation of nucleic acids.

To investigate whether ER stress is associated with elevated levels of cytosolic nucleic acids, dsDNA and dsRNA specific antibodies were used to stain EndoC-βH1 cells after TG exposure. While poly I:C transfection, used as positive control, led to strong accumulation of dsRNA in the cytosol ( Figure 3A ), TG treatment did not affect cytosolic dsRNA ( Figure 3B ). In addition, siRNA-mediated knock down of MDA5 did not affect IFNβ gene expression induced by TG, ruling out dsRNA as potential activator of the type I interferon response ( Figure S2 ). In contrast, an immunostaining performed with a dsDNA specific antibody revealed that induction of ER stress was accompanied by an increased dsDNA signal outside nuclei ( Figures 3C, D ), suggesting leakage of genomic material either from the nucleus or the mitochondria. The decreased TG induced IFNβ gene expression observed after cGAS inhibition by shRNA illustrate the role of dsDNA as activator of the type I interferon response ( Figure S2 ).

To characterize the source of nucleic acids and discriminate free cytosolic dsDNA from mitochondria dsDNA, we combined dsDNA staining with mitochondria labelling. Immunohistochemistry analyses presented in Figure 4A demonstrate the presence of free cytosolic dsDNA outside mitochondria in TG exposed EndoC-βH1 cells ( Figure 4A ). The absence of apoptosis, as assessed by caspase 3/7 activity ( Figure 4B ), and changes in cellular ADP/ATP ratio ( Figure 4C ) after TG treatment suggests that dsDNA does not originate from damaged nuclei of pro-apoptotic cells but rather from mitochondrial leakage. To validate the mitochondrial origin of the cytosolic dsDNA, we used a PCR-based approach combined with primers directed against mtDNA sequences in order to quantify mtDNA leakage. As depicted in Figure 4D , TG treatment led to increase expression of mtDNA in the cytosol of EndoC-βH1 cells whereas total mtDNA content remained unchanged ( Figure S3 ). Of note, the presence of cytosolic mtDNA also correlated with mitochondria stress, as shown by increased expression of Activating Transcription Factor 5 (ATF5), Heat Shock Protein Family A (Hsp70) Member 9 (HSPA9) and DnaJ homolog subfamily A member 3 (DNAJA3) ( Figures 4E–G ), and decrease in mitochondrial oxygen consumption rate (OCR), as assessed by Seahorse flux analyzer in both basal and uncoupled state ( Figure 4H ).

In order to determine the consequence of cellular stress and the induced type-I interferon response in the beta-cell crosstalk with innate immune cells, we first assessed the impact of ER stress on cytokine production by primary human islets. As shown on Figure S4 , TG exposure triggered the secretion of small amount of IL6 and high level of IL8, one of the most potent neutrophils chemoattractant. To test the neutrophils migration capacity towards stressed beta cells, we designed a trans-well migration assay ( Figure 5A ) and evaluated IL8 secretion by EndoC-βH1 upon TG treatment using IL8 specific ELISA. In these assays, we confirmed the increased IL8 secretion by beta cells in response to TG ( Figure 5B ), the IL8 dose-dependent neutrophils migration ( Figure 5C ) and demonstrated a role for the IL8 secreted upon TG treatment in neutrophils chemotaxis ( Figure 5D ).