TYK2, a Candidate Gene for Type 1 Diabetes, Modulates Apoptosis and the Innate Immune Response in Human Pancreatic β-Cells

Type 1 diabetes is a chronic autoimmune disease characterized by islet inflammation (insulitis) and specific destruction of pancreatic β-cells. Insulitis occurs in the context of a “dialog” between invading immune cells and the target pancreatic β-cells (1), which includes upregulation of islet human leukocyte antigen (HLA) class I expression in β-cells (2) and production of chemokines such as CXCL10 by the islet cells (3–5).

Susceptibility to type 1 diabetes is strongly linked to the genetic background. Recent linkage and genome-wide association studies have identified >50 genetic variants with association to type 1 diabetes, explaining ∼80% of the heritability (6). It has been generally assumed that candidate genes for type 1 diabetes modify risk for the disease by acting at the immune system level (7). Recent data (8,9 and present findings), however, indicate that human pancreatic β-cells express >80% of type 1 diabetes candidate genes. Furthermore, another study comparing single nucleotide polymorphism (SNP) locations against chromatin maps for different cell types indicates a primary signature of T1D SNPs in T-cell enhancers but also a highly significant (P < 10⁻⁷) enrichment in pancreatic islet enhancers (10). These genes may contribute to type 1 diabetes by regulating important pathways in the β-cells, such as antiviral responses, innate immunity, and activation of apoptosis (6,11–13).

Tyrosine kinase 2 (TYK2) is a member of the Janus kinase (JAK) family of tyrosine kinases. These kinases play a critical role in the intracellular signaling of several cytokines and type I interferons (IFNs) through phosphorylation and activation of signal transducers and activators of transcription (STATs) (14). TYK2 has been associated with several autoimmune diseases, such as systemic lupus erythematosus, multiple sclerosis, rheumatoid arthritis, and type 1 diabetes (15,16). Six SNPs (rs34536443, rs2304256, rs280523, rs280519, rs12720270, and rs12720356) of the TYK2 gene have been studied in relation to autoimmunity and inflammation, and two of these, rs34536443 and rs2304256, are significantly associated with multiple autoimmune and inflammatory diseases, with the minor allele decreasing susceptibility to these diseases (15). Importantly, the SNP rs2304256:C>A (odds ratio for A vs. C = 0.86), located in exon 8 at chromosome 19p13.2, is associated with protection against type 1 diabetes. rs2304256 is a nonsynonymous SNP causing a missense mutation in TYK2, where the A allele leads to a substitution of valine for phenylalanine at position 362 in the JAK-homology 4 (JH4) region (15,16). This region is critical for both the interaction between TYK2 and IFNAR1 (17) and preserving expression of the IFNAR1 at the cell membranes (18). Thus, it has been suggested that this SNP reduces TYK2 function, resulting in a decreased susceptibility to IFN-related autoimmune diseases (15).

We presently analyzed the role of TYK2 in immune-mediated human pancreatic β-cell apoptosis and local inflammation. The data obtained indicate that TYK2 inhibition prevents double-stranded RNA (dsRNA) (a by-product of viral proliferation, tested here as polyinosinic-polycitidilic acid [PIC]) induced apoptosis in human pancreatic β-cells through the reduction of the type I IFN–STAT signaling and consequent prevention of the increase of major histocompatibility complex (MHC) class I protein expression.

Human islets were isolated from 16 organ donors without diabetes (Supplementary Table 1) with approval from the local ethical committee in Pisa, Italy. Isolation of human islets was done by collagenase digestion and density-gradient purification (19). Subsequently, isolated islets were cultured in M199 medium containing 5.5 mmol/L glucose (19). Within 1–5 days of isolation, the human islets were shipped to Brussels. After arrival in Brussels and overnight recovery, the human islets were dispersed and cultured in Ham’s F-10 medium containing 6.1 mmol/L glucose, 2 mmol/L GlutaMAX, 50 μmol/L 3-isobutyl-1-methylxanthine, 1% charcoal-absorbed BSA, 10% FBS, 50 mg/mL streptomycin, and 50 units/mL penicillin. The proportion of β-cells in the preparations was determined by immunocytochemistry for insulin (9).

The EndoC-βH1 human β-cell line (provided by Dr. R. Scharfmann, Centre de Recherche de l’Institut du Cerveau et de la Moelle Épinière, Paris, France) was cultured in plates coated with Matrigel-fibronectin (respectively 100 and 2 μg/mL) in low-glucose DMEM as previously described (20).

Rat INS-1E cells were used for experiments to evaluate TYK2 stability. These cells were provided by Dr. C. Wollheim (University of Geneva, Geneva, Switzerland) and were cultured in RPMI 1640 GlutaMAX-I medium (Invitrogen) (21).

The cells were treated with human IFNα (specific activity 1.8 × 10⁸ units/mg; PeproTech Inc., Rocky Hill, NJ) at 2,000 units/mL or transfected with 1 μg/mL of the synthetic dsRNA analog PIC (InvivoGen, San Diego, CA) (22). For the exposure to IFNα, medium without FBS was used, whereas for small interfering RNA (siRNA) and PIC transfection, medium without BSA and antibiotics was used.

B lymphoblastoid cell lines (BLCLs) from 12 HapMap CEPH founders were obtained and cultured as described previously (12). Out of the 12 BLCLs, 7 had the CC genotype and 5 the AA genotype corresponding to the TYK2 SNP rs2304256. BLCLs were seeded in six-well culture plates (1.0 × 10⁶ cells/well), precultured for 1 day, and then left untreated or stimulated with 1,000 units/mL human recombinant IFNα (PeproTech, Rocky Hill, NJ) for 30 min. BLCLs were washed in Hanks’ balanced salt solution and lysed in M-PER Mammalian Protein Extraction Reagent supplemented with Halt Protease and Phosphatase Inhibitor Cocktail (all from Pierce, Rockford, IL). Protein concentrations were measured using the DC Protein Assay (Bio-Rad, Hercules, CA) and Western blot for STAT1 performed as described below.

Dispersed human islets or the EndoC-βH1 human β-cells were transfected with 30 nmol/L of two different siRNAs targeting TYK2 (TYK2#1, 5′-CCAUCUGGUAAUAAACUCATT-3′, and TYK2#2, 5′-GAUGCUAUAUUUCCGCAUATT-3′; Qiagen) or Allstars Negative Control siRNA (siCTRL; Qiagen) using the Lipofectamine RNAiMAX lipid reagent (Invitrogen) in a two-step transfection protocol. In this approach, cells were exposed for 16 h to 30 nmol/L siTYK2 or siCTRL, washed, and allowed to recover in culture for 24 h. The cells were then exposed again for 16 h to the same siRNAs, allowed to recover for 24 h in culture, and then used for the subsequent experiments. The siCTRL used does not affect β-cell gene expression, function, or viability in both human islets (23) and EndoC-1βH1 cells (data not shown). In additional experiments, human islet cells were transfected with a previously validated siRNA targeting the ubiquitin-specific peptidase 18 (USP18) (22).

Measurements of living, apoptotic, and necrotic cells were determined after incubation with the DNA-binding dyes Hoechst 33342 (HO) and propidium iodide (PI) as previously described (13). These measurements were performed by two different observers, one of them unaware of sample identity (the level of agreement between the two observers was always >90%). At least 500–600 cells were counted in each experimental condition. In some experiments, apoptosis was confirmed by a second method, namely, Western blot for cleaved (activated) caspase 3, as described below.

Poly(A)⁺ mRNA was extracted using Dynabeads mRNA DIRECT kit (Invitrogen) and reverse transcribed as previously described (24). Quantitative real-time PCR was carried out using SYBR Green and compared with a standard curve (25). The housekeeping gene β-actin was used to correct expression values. β-Actin mRNA expression was not modified by the different treatments presently used (data not shown). The primers used herein are listed in Supplementary Table 2.

Cells were washed with cold PBS and lysed in Laemmli buffer. Immunoblot analysis was performed by overnight incubation with the antibodies listed in Supplementary Table 3. Membranes were incubated with secondary peroxidase-conjugated antibody (anti-IgG (H+L)-HRP; Invitrogen) for 1 h at room temperature. Immunoreactive bands were visualized using the SuperSignal West Femto chemiluminescent substrate (Thermo Scientific), detected using ChemiDoc XRS+ (Bio-Rad), and quantified with the Image Laboratory software (Bio-Rad).

Supernatants from dispersed human islets and EndoC-βH1 human β-cells were collected after treatments to determine the levels of CXCL10, IFNα, and IFNβ using commercially available ELISA kits (R&D Systems, Abingdon, U.K.).

Immunofluorescence was performed as previously described (26). In brief, cells were plated on polylysine-coated coverslips, treated with intracellular PIC or IFNα during 24 h, and fixed with 4% paraformaldehyde. After permeabilization with 0.3% Triton X-100, cells were incubated overnight with the primary antibody rabbit anti–MHC class I (W6/32) (1:1,000) or rabbit anti–cleaved caspase 3 (1:100), or for 1 h with mouse monoclonal anti-insulin (1:1,000). Alexa Fluor 568 goat anti-rabbit IgG or rabbit anti-mouse IgG and Alexa Fluor 488 goat anti-mouse IgG were respectively applied for 1 h (the antibodies used are listed in Supplementary Table 3). After nuclear staining with Hoechst, coverslips were mounted with fluorescent mounting medium (DAKO, Carpintera, CA) and immunofluorescence was visualized on a Zeiss microscope equipped with a camera (Zeiss-Vision, Munich, Germany). Images were acquired at ×40 magnification and analyzed using AxiVision software.

The same protocol used for immunofluorescence, but without permeabilization, was used for flow cytometry. Cells were detached by a mild trypsin treatment, suspended in 2% paraformaldehyde-containing PBS, and then analyzed by flow cytometry (FacsCalibur; BD Biosciences, San Jose, CA). Analysis was performed using CellQuest Pro software version 6.0 (BD Biosciences, San Jose, CA). The cellular populations were selected based on size and cell granularity and analyzed by red fluorescence.

Data are presented as means ± SEM. Comparisons were performed by two-tailed paired Student t test or by ANOVA followed by Student t test with Bonferroni correction, as indicated. A P value <0.05 was considered as statistically significant.

To better understand how candidate genes for type 1 diabetes affect human pancreatic β-cells, we analyzed 51 previously described candidate genes (Supplementary Table 4) using Ingenuity Pathway Analysis (Ingenuity Systems, http://www.ingenuity.com) (Supplementary Fig. 1). The expression of these genes was compared against our previous RNA-seq data of five human islet preparations (9). Forty-two out of 51 genes (82%) were found expressed (i.e., reads per kilobase of transcript per million mapped reads [RPKM] >0.5) in human β-cells. Ingenuity Pathway Analysis of the type 1 diabetes candidate genes expressed in human pancreatic β-cells identified as the three top canonical pathways “interferon signaling,” “role of JAK1, JAK2 and TYK2 in IFN signaling,” and “role of pattern recognition receptors in recognition of bacteria and virus.”

TYK2, a key regulator of IFN signaling (14), was listed in three out of the four top canonical pathways identified, namely, “interferon signaling,” “role of JAK1, JAK2 and TYK2 in IFN signaling,” and “Tec kinase signaling.” Furthermore, and in line with a previous bioinformatics prediction (15), BLCLs obtained from patients with genotype AA of the TYK2 SNP rs2304256 (the variant protective against type 1 diabetes) showed a trend for less marked IFNα-induced STAT1 phosphorylation, as compared with patients with genotype CC (Fig. 1). Thus, whereas AA patients increased by 3.5-fold STAT1 phosphorylation as compared with basal levels, CC patients showed a 5.7-fold increase.

PIC, but not IFNα, induced a mild increase in TYK2 expression in human islet cells. On the other hand, only IFNα increased TYK2 expression in human EndoC-βH1 cells (Supplementary Fig. 2A and B). Culture of dispersed human islets and EndoC-βH1 under different glucose concentrations (6 or 28 mmol/L) did not change TYK2 expression (Supplementary Fig. 2C and D). We next attempted to study the putative role of TYK2 in human β-cells by using specific siRNAs to inhibit TYK2 expression (knockdown [KD]) but faced major technical problems to obtain an adequate (i.e., >50%) inhibition of the protein. To evaluate if this could be explained by a prolonged stability of the TYK2 protein, we treated INS-1E cells with cycloheximide and followed TYK2 and α-tubulin expression over a 24-h period. Both proteins had a similar and rather long half-life (around 20 h) (Supplementary Fig. 3A and B), explaining the observed difficulties in inhibiting TYK2 by a single period of exposure to the siRNA. To overcome this problem, we designed a two-step transfection protocol (see 2research design and methods) with the siRNAs targeting TYK2. By this approach, TYK2 mRNA and protein expression were inhibited by >50% (Fig. 2A, B, E, and F).

Exposure to intracellular PIC increased apoptosis of dispersed human islet cells and EndoC-βH1 cells (Fig. 2C and D), and this was partially prevented by TYK2 KD. Whereas basal apoptosis was not modified by TYK2 KD, PIC-induced apoptosis was reduced by 30% (Fig. 2C and D). The protective effects of TYK2 KD against PIC-induced apoptosis were confirmed by a lower expression of cleaved caspase 3 (Fig. 2E and F). We confirmed the identity of the apoptotic cells in dispersed human islets by immunofluorescence using cleaved caspase 3 as an apoptotic marker. Thus, PIC treatment led to an increase of 7.0% in apoptosis of insulin-positive cells, whereas TYK inhibition prevented by 37% this increase in cell death (Supplementary Fig. 4).

TYK2 is a member of the JAK family of kinases that transduce type I IFN signals by phosphorylating STAT proteins, such as STAT1 and STAT2 (14). PIC treatment induced a two- and fourfold increase in phospho-STAT2 and phospho-STAT1, respectively, and a nonsignificant trend toward higher expression of total STAT1 and STAT2 in dispersed human islet cells and EndoC-βH1 cells (Fig. 3 and Supplementary Fig. 5). The PIC-stimulated STAT1 and STAT2 phosphorylation was abrogated by TYK2 KD (Fig. 3A–D and Supplementary Fig. 5). No significant changes were observed in total STAT1 and STAT2 expression after TYK2 KD in dispersed human islets (Fig. 3E and F). Similar results were observed for STAT1 phosphorylation in EndoC-βH1 cells treated with IFNα as a direct stimulus (Fig. 4A and B). Of note, IFNα treatment did not induce β-cell apoptosis (Fig. 4C).

We have previously shown that intracellular PIC upregulates type I IFNs and their downstream genes (IFN-stimulated genes [ISGs]) in rodent β-cells via NF-κB, STATs, and IFN regulatory factor 3 (IRF3) activation (27,28). To evaluate the role of TYK2 in PIC-stimulated upregulation of type I IFNs and chemokines in human β-cells, we measured mRNA expression and release into the medium of IFNα, IFNβ, and CXCL10 by dispersed human islet cells (Fig. 5A–F) and EndoC-βH1 cells (Fig. 5G–L). Both mRNA expression and release of IFNα, IFNβ, and CXCL10 were increased in human β-cells after PIC treatment, and this was partially prevented by TYK2 KD in the case of IFNα and CXCL10, but not IFNβ (Fig. 5).

The mechanisms by which islet cells overexpress MHC class I in early type 1 diabetes remains to be clarified, but it has been hypothesized that this is secondary to persistent viral infection and consequent type I IFN production (2,29). Since we observed that TYK2 regulates IFNα production and signaling in human β-cells, we next examined whether this type 1 diabetes candidate gene plays a role for the upregulation of MHC class I molecules in β-cells. In dispersed human islet cells, PIC and IFNα promoted a twofold increase in HLA A, B, and C (HLA-ABC) mRNA expression at 24 and 48 h, respectively (Fig. 6). Interestingly, TYK2 KD (Fig. 6A and B) prevented both PIC- and IFNα-induced HLA-ABC expression in these cells (Fig. 6C and D). PIC and IFNα also stimulated an increase in HLA-ABC mRNA expression in EndoC-βH1 cells, and this was prevented by TYK2 KD (Fig. 6E and F). These observations were confirmed at the protein level by two different approaches, namely, flow cytometry (Fig. 6G) and immunofluorescence (Fig. 6H). Thus, 30 and 70% of the cells were positive for MHC class I protein upon treatment with PIC and IFNα, respectively, and this was decreased by ∼50% in the case of TYK2 KD (Fig. 6G). Similar results were observed by immunofluorescence, where TYK2 KD clearly reduced the PIC- and IFNα-induced MHC class I protein expression in human EndoC-βH1 cells (Fig. 6H).

To confirm the participation of TYK2 and IFN-driven gene networks as regulators of MHC class I protein expression in pancreatic β-cells, we silenced USP18, an IFN-stimulated gene 15–specific protease, in dispersed human islets. We have previously shown that USP18 inhibition induces inflammation and apoptosis by increasing IFN-induced STAT1/2 signaling in β-cells (22). We have now observed that USP18-silenced cells presented increased STAT1 phosphorylation and higher levels of HLA-ABC mRNA expression upon IFNα treatment (Supplementary Fig. 6). IRF1, a transcription factor involved in type I IFN–stimulated gene expression, contributes to IFN-stimulated expression of the immunoproteasome, which promotes antigen processing for presentation by MHC class I molecules (30). As observed in Supplementary Fig. 7, PIC-induced IRF1 expression was 50 and 30% lower after TYK2 inhibition (>50% TYK inhibition, as in Fig. 2) in EndoC-βH1 cells and dispersed human islets, respectively.

Pathway analysis of type 1 diabetes candidate genes expressed in human pancreatic islets identified three top canonical pathways, namely, “interferon signaling,” “role of JAK1, JAK2 and TYK2 in IFN signaling,” and “role of pattern recognition receptors in recognition of bacteria and virus.” This suggests that type 1 diabetes candidate genes play a role in regulating β-cell “self-defense” or “cell autonomous immune responses” against infection. These self-defense mechanisms are present in many nonimmune cell types and are upregulated upon virus infection (31,32). In vertebrates, cellular self-defense synergizes with innate and adaptive immunity to combat infections (32). The cell susceptibility/resistance of differentiated and poorly proliferating cells, such as pancreatic β-cells and neurons, to viral infection is an important determinant of clinical outcome. For instance, higher basal expression and faster upregulation of IFN-induced genes improve the survival of neurons infected by West Nile virus (33), suggesting that candidate genes that affect these pathways may have a major impact on the outcome of viral infections (or exposure to other “danger signals”) in β-cells. This may determine the amplitude of the local inflammation, the degree of β-cell loss, and an eventual progression to full autoimmunity and type 1 diabetes (2,6,29,34).

The candidate gene TYK2 may play a key role in these self-defense pathways. TYK2 was identified in the pathway analysis described above, and it plays a critical role in the intracellular signaling of type I IFNs via phosphorylation and activation of STAT proteins (14). Of note, type I IFNs play an important role in several autoimmune diseases, including T1D (35), by, among other effects, activating dendritic cells to present self-antigens to potentially autoreactive T cells (36). Furthermore, very recent data indicate that severely reduced TYK2 expression in pancreatic β-cells, due to a natural mutation, is responsible for susceptibility to virus-induced diabetes in SJL and SWR mice (37).

It has been previously shown that mice on a high-fat diet present decreased TYK2 expression in brown adipose tissue and skeletal muscle, but not in white adipose tissue and liver, suggesting that modulation of TYK2 expression by diet is tissue specific (38). We did not observe, however, changes in TYK2 expression after exposure of dispersed human islets and EndoC-βH1 human β-cells to high glucose. These findings are in agreement with previous RNA-seq data where TYK2 was found similarly expressed in human pancreatic islets from individuals with different degrees of glucose tolerance (donors with normoglycemia, impaired glucose tolerance, and type 2 diabetes) (39) or in human islets exposed or not to palmitate (40).

One of the polymorphisms in the TYK2 gene, rs2304256, predicted to decrease its function (a prediction presently confirmed experimentally), is associated with protection against type 1 diabetes (15,16). This is in line with our present observations, indicating that a 50% inhibition in TYK2 activity by specific siRNAs decreases dsRNA-induced apoptosis and proinflammatory pathways in human pancreatic β-cells. This protection is mediated via inhibition of IFNα signaling, as indicated by decreased phosphorylation of STAT1 and STAT2. Of particular relevance, inhibition of TYK2 prevents dsRNA- or IFNα-induced upregulation of MHC class I protein expression and expression of the chemokine CXCL10 in human β-cells.

MHC class I upregulation is one of the most consistent findings in islets from patients with type 1 diabetes (41,42). This may be one of the mechanisms by which β-cells become “visible” to the immune system in early type 1 diabetes (2). This MHC class I hyperexpression is observed both in inflamed and apparently noninflamed insulin-containing islets, suggesting that this upregulation precedes insulitis during diabetes development (42). The present data provide the first evidence that this phenomenon is regulated by a type 1 diabetes candidate gene, namely, TYK2. In line with these findings on the role of TYK2 in MHC class I regulation, granulocyte-macrophage colony-stimulating factor–induced downregulation of TYK2 and JAK1 tyrosine phosphorylation, as well as TYK2 protein expression, contributes to decreased STAT1 phosphorylation and subsequently diminished MHC class I antigen levels in hematopoietic cell lines (43). Furthermore, IFNβ-stimulated MHC class I expression is abrogated when JAK1, TYK2, or the IFNα/β receptor is suppressed by siRNAs in human Ewing sarcoma cell line (44).

The chemokine CXCL10 has been previously shown to be upregulated in human and rodent islets exposed in vitro to the proinflammatory cytokines IL-1β, TNFα, IL-17, and IFNγ (3,4). Importantly, CXCL10 expression is also upregulated in islets from patients with type 1 diabetes (5,45) and in NOD mice (3,46), and its neutralization prevents diabetes in NOD mice (47).

As discussed above, TYK2 may play an important role in the pathogenesis of diverse autoimmune diseases, and novel therapeutic strategies based on specific TYK2 inhibitors, such as Cmpd1, are being evaluated (48,49). The present data suggest that TYK2 inhibition could be tested as a novel therapeutic approach to prevent type 1 diabetes development.

In conclusion, we provide evidence that the candidate gene TYK2 plays a key role in the activation of cell-autonomous immune responses through the activation of STATs and consequent triggering of the IFN response, that may lead to hyperexpression of MHC class I proteins in human pancreatic β-cells (Fig. 7). Polymorphisms that decrease function of TYK2 (present data) and of MDA5 (encoded by IFIH1 and functioning as a detector of viral infection in β-cells and other cell types [11,50]) decrease the risk of type 1 diabetes (51,52). This suggests that a genetically determined excessive inflammatory response to viral infections may contribute to autoimmunity and eventual diabetes in susceptible individuals.

Acknowledgments. The authors are grateful to M. Pangerl, A.M. Musuaya, N. Pachera, and I. Millard for excellent technical support, Drs. M. Cnop and M. Igoillo-Esteve for providing data on human samples, and Drs. J. Juan, O. Villate, and J.-V. Turatsinze for experimental support and RNA-seq data analysis.

Funding. This work was supported by grants from Fonds National de la Recherche Scientifique (FNRS), Belgium, Communauté Française de Belgique-Actions de Recherche Concertées (ARC), and the European Union (projects Naimit and BetaBat, in the Framework Programme 7 of the European Community). L.M. is supported by an FNRS postdoctoral fellowship. R.S.D.S. is the recipient of a postdoctoral fellowship from Conselho Nacional de Desenvolvimento Cientifico e tecnológico (CNPq), Brazil. T.F. and F.P. were supported by a grant from the European Foundation for the Study of Diabetes (EFSD/JDRF/NN). I.S. was the recipient of a postdoctoral fellowship from the Education Department of the Basque Country.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. L.Marr. and R.S.D.S. contributed to the original idea and the design of the experiments; researched data; contributed to discussion; and wrote, revised, and edited the manuscript. T.F., F.A.G., A.O.d.b., L.Mars., P.M., and F.P. researched data and revised and edited the manuscript. I.S. researched data, contributed to discussion, and revised and edited the manuscript. D.L.E. contributed to the original idea and the design and interpretation of the experiments; contributed to discussion; and wrote, revised, and edited the manuscript. L.Marr. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.