Expression of Interferon-Stimulated Genes in Insulitic Pancreatic Islets of Patients Recently Diagnosed With Type 1 Diabetes

Activation of the innate immune response in the islets of Langerhans has been suggested to be an important step in the development of type 1 diabetes (T1D) (1–3). In particular, viral infection of the β-cells leading to the production of type I interferon (IFN) and induction of several hundred IFN-stimulated genes (ISGs) has been suggested (4–6). ISGs constitute a group of genes that are upregulated in response to IFN and put the surrounding cells in an antiviral state, protecting them from being infected. Being induced by virus infection (7), signs of their expression have been interpreted as “viral footprints,” but, importantly, many of these genes are also overexpressed under other conditions, in the absence of viral infection (8,9).

Studies of human pancreatic islets at the onset of T1D are very rare. Although immunohistochemical studies have suggested the expression of type I IFN (10,11) and some ISGs (12,13) in T1D islets, other studies have failed to detect type I IFN in pancreata from subjects with diabetes (14). End point PCR, with no or limited quantitative value, has suggested IFN-α expression to be more prevalent in T1D than in control pancreata (15,16), but this has not been confirmed by quantitative methods. Whole-transcriptome analysis of pancreatic tissue demonstrated the upregulation of some ISGs in a pancreas soon after T1D onset, but this upregulation was not found in isolated islets from the same donor (17).

In this study, freshly frozen and cultured pancreatic tissue from living patients with recent-onset T1D collected within the Diabetes Virus Detection (DiViD) study (18) were examined. We have published previously that islets isolated from these patients mainly showed background levels of cytokine/chemokine release (19). However, the enzymatic digestion of the pancreas and the isolation of islets may also lead to the induction of inflammatory markers in the control subjects. In this present study, we used laser capture microdissection (LCM) to extract RNA from islets with insulitis directly from the frozen pancreatic tissue and compared the expression of 84 ISGs with that in noninsulitic islets from organ donors without diabetes. By separately microdissecting the islet core and the peri-insulitis leukocytes in insulitic T1D islets, we aimed to determine the origin of the detected overexpression. In addition, RT-PCR was used to search for the presence of enterovirus in the microdissected samples.

Pancreatic biopsy samples from a total of five patients recruited to the DiViD study and five multiorgan donors procured within the Nordic Network for Islet Transplantation were included in the study. The DiViD samples were collected by pancreatic tail resections performed 3–9 weeks after the diagnosis of T1D, and the patient characteristics have been described previously in detail (18–21). Good-quality RNA, suitable for use in this present study, was possible to extract from laser-captured islets from DiViD cases 2–6. The multiorgan donors were previously healthy, without known pancreatic disease, and age matched to the DiViD cases (mean age 25.8 years, age range 20–32). The DiViD study was approved by the Regional Ethics Committee of the Norwegian Government, and informed consent was obtained from the patients after oral and written information from the diabetologist and surgeon were separately obtained. The use of pancreatic tissue from deceased organ donors for research was obtained verbally from the deceased person’s next of kin by the attending physician and documented in the medical records in accordance with Swedish law and as approved by the Regional Ethics Committee in Uppsala (Dnr 2015/444).

Frozen tissue samples from the pancreatic tail region were sectioned and mounted on Superfrost Plus glass coverslips (Menxel-Gläser, Braunschweig, Germany) or Arcturus PEN Membrane Glass Slides (Life Technologies, Carlsbad, CA) for immunohistochemistry (IHC) and LCM, respectively. Consecutive sections were stained for CD3 or were used for LCM to microdissect islets with insulitis (≥15 CD3⁺ cells) from the diabetic samples and islets without insulitis from the nondiabetic samples. The procedure was as described previously (19).

In order to microdissect peri-insulitis and noninsulitic islet cores separately, slides for IHC were double stained for CD45 (mouse monoclonal antibodies 2B11 plus PD7/26, dilution 1:75; Dako) and insulin (polyclonal guinea pig anti-insulin, dilution 1:140; Dako). CD45 was chosen as a marker instead of CD3 in this case to allow the exclusion of any islet core affected by non–T-cell insulitis. Insulin-containing islets with peri-insulitis, but without insulitis in the islet core, were localized, first on the IHC slides, and then on the unstained consecutive membrane slides by their autofluorescence. In DiViD subject 3, several islets fulfilled these criteria in multiple sections per islet; and, from these, the noninsulitic islet core, the peri-insulitis, and a noninfiltrated peri-islet region could be microdissected and pooled in separate tubes for RNA extraction and expression analysis. The procedure is described in detail in Fig. 1.

RNA isolation, cDNA synthesis, preamplification of cDNA, and expression analysis was performed with kits from Qiagen (Sollentuna, Sweden), as described previously (19). A pathway-specific primer mix (Human Type I Interferon Response, catalog #PBH-016Z; Qiagen) was used for the preamplification and a PCR array (Human Type I Interferon Response, catalog #PAHS-016ZC; Qiagen) was used for the expression analysis of 84 known ISGs. β-Actin (ACTB), GAPDH, and 60S acidic ribosomal protein P0 (RPLP0) were chosen for normalization. Genes with a quantification cycle (Cq) values >35 were regarded as nondetected and assigned a Cq of 35 to calculate fold induction. RNA from the same samples was analyzed for the presence of enterovirus by a semi-nested RT-PCR amplifying a part of the 5′ untranslated region of the enterovirus genome, as described previously (21,22).

Total RNA extracted from islets isolated by collagenase digestion from the tip of the tail was subjected to whole transcriptome sequencing as reported (20). Normalized reads per kilobase per million mapped reads for the genes included in the RT2 Profiler PCR array for ISGs are presented in this study, and the full data set (reads) is openly available in the BILS DOI repository (http://doi.bils.se/) under DOI number 10.17044/BILS/G000002.

P values were calculated for each gene using a Mann-Whitney signed rank test, and a rank-based volcano plot was created based on the raw P values and the fold expression relative to the mean of the controls.

Expression of 41 of the 84 analyzed genes was detected in all samples of laser-captured islets from all donors. Four genes (SH2D1A, CD80, HLA-G, and TLR7) had detectable expression levels in insulitic T1D islets but were not detected in islets from control subjects without diabetes. Twenty-three of the 84 genes showed at least a fivefold overexpression in insulitic T1D islets compared with those from control subjects without diabetes, whereas only 1 gene showed overexpression in islets from control subjects without diabetes (Fig. 2A and B). Interestingly, type I IFN (IFNA2, IFNB1, IFNA4, IFNA1) was rarely detected and was not overexpressed in the T1D islets.

The overexpression of some ISGs (e.g., GBP1, TLR3, OAS1, CXCL10, CCL5, and CASP1) was also detected in islets from subjects with T1D that were isolated by collagenase digestion and handpicked (Fig. 3 and Supplementary Fig. 1). However, compared with islets isolated from subjects without diabetes, fewer ISGs were overexpressed in isolated islets than in laser-captured islets, suggesting that the overexpression is most evident in islets with insulitis and/or that it is affected by the islet isolation process itself.

To distinguish islet expression from expression in the infiltrating immune cells, in islets affected by peri-insulitis, but not intrainsulitis, the noninsulitic islet core, the surrounding peri-insulitis, and the noninfiltrated peri-islet areas were microdissected separately in sections from DiViD subject 3. Because of a lower number of microdissected cells compared with when all insulitic islets were captured, genes with a relatively low expression were not detected. However, 38 of the 84 ISGs were detected in the islet core, peri-islet area, or both (Fig. 4). Most of these, including GBP1, TLR3, OAS1, EIF2AK2, HLA-E, IFI6, and STAT1, showed higher expression in the islet core compared with in the peri-islet area containing the surrounding immune cells. In contrast, the T-cell attractant chemokine CXCL10 that was overexpressed in T1D islets showed an almost 10-fold higher expression in the peri-islet area than in the islet core. Most genes were also detected in noninfiltrated peri-islet areas, but their expression levels were generally lower than those within the islet core (Fig. 4).

No enterovirus RNA was detected in any of the laser-captured insulitic islets from the T1D subjects using a highly sensitive PCR.

In the current study, we demonstrate the overexpression of ISGs in the islets of patients with recent-onset T1D. Because the number of cases available were limited, it is difficult to draw firm conclusions for each individual gene, especially when compensating for multiple comparisons. However, the finding that 26 of the 84 genes were either overexpressed at least fivefold or were overexpressed and had rank-based P values <0.05 (Fig. 2B) indicates that a pathway leading to the induction of ISGs is indeed active in the islets of patients with recent-onset T1D. Whether this activation is due to IFN in these T1D patients remains to be investigated.

When comparing the PCR array data of insulitic laser-captured islets with RNA sequencing data from isolated islets, the overexpression of GBP1, TLR3, OAS1, STAT1, CXCL10, CCL5, and CASP1 was confirmed, increasing the likelihood that these genes are truly overexpressed in T1D islets. CXCL10 and CCL5 encode the T cell–recruiting chemokines also known as IFN-γ–induced protein 10 (IP-10) and regulated on activation normal T cell expressed and secreted (RANTES), respectively. The expression of chemokine (C-X-C motif) ligand 10 (CXCL10)/IP-10 is induced by enterovirus infection of human pancreatic islets in vitro (23), a scenario that has been suggested to occur in virus-induced T1D (14). However, CXCL10/IP-10 can also be induced by IFN-γ under inflammatory conditions and is secreted from islets obtained from subjects with type 2 diabetes (24). Also CCL5/RANTES has been demonstrated to be upregulated in islets upon enterovirus infection (7,25,26). GBP1 encodes a GTPase, guanylate-binding protein 1 (GBP1), that is induced by types I and II IFN, and has antiangiogenic effects (27). It has been shown to be increased in the sera of patients with rheumatic autoimmune diseases characterized by chronic inflammatory vessel activation (28), but, to our knowledge, it has never before been associated with T1D. The finding of increased GBP1 expression in T1D islets in this study may suggest the involvement of endothelial activation in T1D pathogenesis.

Importantly, the RNA-seq analysis of isolated islets has limitations compared with PCR array analysis of laser-captured islets; the islet isolation procedure and islet culture may alter gene expression levels, and we possess no knowledge concerning whether the islets used for analysis were insulin positive and/or affected by insulitis. Thus, it is likely that several of the genes found to be overexpressed in insulitic islets by PCR array analysis, but not by RNA-seq analysis, of isolated islets represent true positive findings, reflecting actual differences between islets from T1D subjects close to the onset of disease and matched control subjects without diabetes.

The pattern of ISG overexpression is in line with that demonstrated in isolated islets infected in vitro with an enterovirus, Coxsackievirus B5 (7), or exposed to type I (8) or type II (7,9) IFN (Supplementary Fig. 2). The lack of evident type I IFN expression, together with the failure to detect enteroviral genome in the analyzed islets, may argue against viral infection as a source of ISG induction in the islets of these subjects, and the presence of type II IFN (IFNG) expression may in itself explain the induction of many of these genes. However, virus infection in a nearby cell, not present in the laser-captured area, cannot be excluded. The ISG overexpression in the analyzed samples could be a sign of an antiviral state in these cells induced by infected nearby cells secreting type I IFN. In fact, the presence of enterovirus was suggested in the islets of the DiViD subjects by immunostaining for the capsid protein VP1 in several endocrine cells and by trace amounts of enteroviral RNA sequences in the culture medium of isolated islets (21). However, only in one subject (case 6) could this virus be found by PCR in well-preserved frozen pancreatic tissue.

The induction of innate immunity by viruses is a complicated interplay between the virus and the infected host cell. Viruses produce proteins that efficiently inhibit the production of antiviral genes and use the cellular machinery for its own replication. Different strains and serotypes of a virus vary in their capacity to induce an innate immune response (29), dramatically affecting the outcome of infection. A strong innate immune response may limit viral replication, but at the same time contribute to unwanted tissue damage. It has been debated whether a “diabetogenic” viral strain likely is a strong or a weak inducer of innate immunity (30). In mice that have been genetically modified to lose their ability to respond to IFN specifically in their β-cells, enterovirus infection rapidly causes diabetes (31) and a reduced expression of the IFN-α receptor–associated protein TYK2, compromising the induction of ISGs by type I IFN, was recently shown to increase the susceptibility to virus-induced diabetes (32). In the current study, the ISGs were clearly overexpressed to levels comparable to when isolated human islets were cultured with high levels of IFN in vitro. These data suggest that if T1D in these cases were induced by a virus, it was not associated with the weak induction of innate immunity.

LCM is a powerful tool to selectively study the gene expression in different tissues and cells. In this study, we used the method to quantify gene expression specifically in islets affected with insulitis and successfully used it to distinguish the expression in the islet core from the expression in the surrounding immune cells. Our finding of the T cell–recruiting chemokine CXCL10 in the peri-islet area, but not within the islet core, is in contrast to the IHC-based finding of this chemokine in the β-cells of subjects with recent-onset T1D (13,14) and in conflict with the idea that a direct virus infection of the β-cells is the cause of its induction (14,24,33). However, our finding agrees well with the location of T cells mainly in the peri-islet area and more rarely infiltrating inside the islet core (19,34). Future studies are needed in order to find the underlying cause of CXCL10 induction and the recruitment of T cells to the peri-islet area. Also, this finding highlights the importance of characterizing different pancreatic regions separately, especially in a disease with a lobular pattern like T1D affecting a heterogeneous organ like the pancreas.

One weakness of this study is that it is mainly descriptive, due to the nature of the studied material. When we report “overexpression” of ISGs in T1D islets, we do not know whether these genes were upregulated as a part of the pathological processes leading to T1D are a consequence of a destructive process, or even whether baseline expression levels of these genes are higher in patients in whom T1D develops. Because of the complications associated with acquiring pancreatic tissue from live patients (18), novel imaging techniques will be required to follow ongoing pathogenic processes in the pancreas over time. Nevertheless, in this study, we used the pancreatic tissue taken from subjects with recent-onset T1D in the DiViD study, which is the, so far, largest collection of well-preserved pancreatic tissue from subjects with recent-onset T1D. This, together with a sophisticated LCM strategy, allowed us to perform a molecular characterization of T1D islets that was not previously possible to perform. Advancing the protocols for LCM, and using them on well-characterized biopsy materials, allows high-resolution characterization of the ongoing pathogenic processes in affected pancreata.

In conclusion, we demonstrate the overexpression of many ISGs in T1D islets affected with insulitis. Strategic use of LCM allowed us to separate islet expression from expression in the surrounding immune cells. The upstream inducer of this ISG overexpression, possibly playing an important role in the induction of T1D, remains to be defined.

Acknowledgments. The authors thank specialist nurse Trine Roald, Oslo University Hospital, Oslo, Norway, whose invaluable efforts were essential to the success of the DiViD study.

Funding. This study was supported by South-Eastern Norway Regional Health Authority (to K.D.-J.); Novo Nordisk Foundation (to K.D.-J.); and the Persistent Virus Infection in Diabetes Network (PEVNET) Study Group, funded by the European Union Seventh Framework Programme (grants FP7/2007-2013 and 261441 PEVNET), the Swedish Medical Research Council (grants VR K2011-65X-12219-15-6 and K2015-54X-12219-19-4), the Diabetes Wellness Foundation, the Stiftelsen Family Ernfors Foundation, the Novo Nordisk Foundation, the Åke Wiberg Foundation, the Tore Nilsson Foundation, the Swedish Diabetes Association, Gillbergska Stiftelsen, and Barndiabetesfonden. Work performed at the Uppsala Genome Center has been funded by Beredningshandbok forskningsinfrastruktur (RFI)/Vetenskapsrådet (VR) Swedish National Infrastructure for Large-Scale Sequencing and Science for Life Laboratory (SNISS), Uppsala, Sweden. Human pancreatic biopsy samples and isolated islet samples were obtained from The Nordic Network for Clinical Islet Transplantation, which was supported by the Swedish National Strategic Research Initiative EXODIAB (Excellence Of Diabetes Research in Sweden) and JDRF.

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

Author Contributions. M.L. and E.K. performed the experiments, interpreted the data, and contributed to the writing of the manuscript. L.K. was responsible for clinical coordination and recruitment of the patients and interpretation of the data and participated in the writing of the manuscript. K.D.-J. is the principal investigator of the DiViD study and, as such, was responsible for design of the study, funding, regulatory issues, and international collaboration. K.D.-J. also participated in data analysis and interpretation and in the writing of the manuscript. O.S. designed the experimental setup, researched and interpreted the data, and wrote the manuscript. O.S. 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.