Local Activation of Dendritic Cells Leads to Insulitis and Development of Insulin-Dependent Diabetes in Transgenic Mice Expressing CD154 on the Pancreatic β-Cells

Murine CD154 cDNA was cloned by PCR using oligo-dT primed cDNA from C57BL/6 mouse spleen that was prepared as described (18), using the primers CD154, 5′-GCCACCATGATAGAAACATA-3′, and CD154 stop, 5′-TCAGAGTTTGAGTAAGCCAA-3′. PCR products were cloned into the pCRII vector by TA cloning (Invitrogen). Ligated products were sequenced and confirmed to be the CD154 cDNA (according to GenBank no. NP035746). The CD154 cDNA was subcloned by EcoRI digestion into the pBlueSKII–β-globin (19) vector (pBG), containing the exon 2-intron 2-exon 3 cassette from the β-globin gene for efficient posttranscriptional processing in vivo. The obtained promoter-less construct was termed pBG-CD154. The RIP was cloned from the pJNL1-RIP-Tag (17) by digesting with KpnI and HindIII and was digested similarly into the pBG-CD154. This construct was termed pBG-RIP-CD154. For generation of transgenic mice, pBG-RIP-CD154 was digested by FspI and NotI, yielding the transgenic construct depicted in Fig. 1A. This fragment was purified and microinjected into (C57BL/6xCBA) F1 oocytes to generate RIP-CD154 transgenic mice (MouseCamp; Karolinska Institute, Stockholm, Sweden). Among 96 offspring, 14 carriers were identified on the basis of PCR screening. Among the 14 offspring, all carriers that expressed the transgene also had insulitis. Two lines (L60 and L90) were selected for further study and backcrossed to C57BL/6 mice.

RIP-CD154 mice were maintained by backcrossing to C57BL/6 or RAG12M (RAG KO on C57BL/6 background) and were bred under specific pathogen-free conditions at Taconic M&B (Ry, Denmark), and animals selected for further study were housed at our Animal Unit (Novo Nordisk, Gentofte, Denmark). DNA from tail biopsies was analyzed by PCR using the following primers: β, 5′-ATACTCTGAGTCCAAACCGG-3′, and γ, 5′-CTCCTCACAGTTCAGCAAGG-3′, and as an internal control the primers for the Hes1 gene 5′-AGCCAGTGTCAACACGACACC-3′ and 5′-TGTTAA GTGCATCCAAAATCAGTG-3′. To test for expression of the transgene, cDNA was prepared from RNA from various tissues from transgene-positive RIP-CD154 L60 and L90 as described (18). For detection of the transgenic CD154 cDNA, the primers α, 5′-GAGGAGGCTTTTTTGGAGGC-3′, and γ (see above) were used with the primers for the β-tubulin cDNA and 5′-ATCCTGGTACTGCTGGTACT-3′ and 5′-GAGCTGTTCAAGCGCATCTC-3′ as an internal control. Expected band sizes were 386 bp for the spliced transgenic cDNA, 959 bp for the unspliced transgenic cDNA (plasmid control), and 156 bp for the β-tubulin cDNA.

Presented data were obtained from generation N3 or later. No change in phenotype has been observed during backcrossing to C57BL/6 or at the different sites (Ry versus Gentofte). All animal experiments were conducted according to Danish legislation and approved by the Danish Animal Inspectorate. Blood glucose was measured by tail vein blood sampling and analyzed using the Medisense Precision Xtra Plus system. Maximum measurable blood glucose level was 34 mmol/l, and mice were considered hyperglycemic when blood glucose levels were >11 mmol/l (which was the maximum measurement for nontransgenic controls) and diabetic when blood glucose was >16.6 mmol/l for ≥2 consecutive weeks. Unless otherwise stated, samples were taken from nonfasting mice.

For fluorescence histology on frozen sections, tissue was snap frozen in Tissue-Tek on dry ice immediately after dissection, cut into 5-μm sections, and blocked using 5% normal goat serum and 5% normal donkey serum for 20 min at 20°C. Staining with primary antibodies was performed overnight at 4°C, using the following antibodies: CD11c (HL3), CD4 (H129.19), CD8 (53–6.7), CD19 (1D3), CD154 (MR1), and MHC-II (M5/114.15.2), all from BD Pharmingen, and insulin was from Zymed. All antibodies were diluted 1:100–1:200 in PBS with 0.25% BSA. After washing in PBS, second-step reagents were applied for 45 min at 20°C and included the following antibodies: goat-anti–hamster-Cy3 (127-165-160), goat-anti–rat-Cy3 (112-165-167), donkey-anti–guinea pig-FITC (fluorescein isothiocyanate) (706-095-148), and donkey-anti–guinea pig-biotin (706-065-148), all from The Jackson Laboratory. If necessary, a third-step reagent was streptavidin-AMCA (7-amino-4-methylcoumarin-3-acetic acid) (016-150-084; The Jackson Laboratory), which was incubated for 30 min at 20°C. Controls included staining with secondary antibody only or with isotype controls and were always negative compared to experimental slides. For formalin-fixed sections, tissue was fixed overnight in 4% formalin and transferred to 70% ethanol. Tissue was embedded in paraffin, and 4-μm sections were cut and stained for insulin (HUI-18; Novo Nordisk) using Histostain-SP Bulk (Zymed). Staining was revealed using the AEC (3-amino-9-ethylcarbazole chromogen) substrate kit (Zymed), and all sections were counterstained with hematoxylin. Images were recorded by a Hamamatsu C5810CCD cooled camera and processed in Adobe Photoshop.

For C-peptide measurements (oral glucose tolerance test), mice were fasted for 6 h, briefly sedated with isoflurane, and 2 mg/g body wt glucose at t = 0 min was injected via a mouse feeding needle. Blood samples for C-peptide measurements were obtained in EDTA-coated tubes at t = 30 min, and serum was prepared by centrifugation. C-peptide levels were measured by Linco Diagnostics, using a radioimmunoassay. For an intraperitoneal glucose tolerance test, mice were fasted before an intraperitoneal injection of 2 mg/g body wt glucose at t = 0 min. Blood glucose was measured at the indicated times.

Islets were isolated from 6- to 10-week-old C57BL/6 mice by collagenase treatment of the pancreas, followed by handpicking of the islets under a microscope. The islets were cultured for 6–7 days at 37°C in RPMI with 10% FBS, 11 mmol/l glucose, 2 mmol/l glutamine, 20 mmol/l HEPES, and antibiotics. Before transplantation, viable islets were recounted and transferred to medium containing 0.5% NUSerum. Diabetic transgene-positive RIP-CD154 or nondiabetic transgene-negative littermates (from N4 and N6 generation of backcrossing to C57Bl/6 mice) were anesthetized using fentanyl/fluanison/midazolam (0.6, 19, and 9 μg/g s.c., respectively). For postoperative analgesia, buprenorfine and carprofen (0.15 and 5 μg/g s.c., respectively) were used. Each mouse received 300–350 islets, which were placed under the left kidney capsule.

To remove the transplanted tissue, the kidney with the transplant was exposed using the same approach as above and the renal artery and vein ligated in common. The kidney was removed in toto. Kidney histology was performed after formalin fixation, as described.

For flow cytometric analysis on pancreatic lymph node cells, pancreatic lymph nodes were dissected from three to four transgenic mice, and tissue was homogenized using a cell strainer (Falcon). Cells were labeled with biotin-conjugated CD11c and phycoerythrin-conjugated CD40 antibodies, followed by streptavidin-APC (BD Pharmingen). All cells were analyzed on a FACSCalibur, and data were analyzed using CellQuest software. Events (40,000–100,000) falling into a live gate based on forward and side scatter characteristics were acquired.

Spleen cells from recently diabetic transgene-positive RIP-CD154 (L90) mice were isolated by homogenizing the tissue through a cell strainer, followed by washing in ice-cold Hank’s balanced salt solution. Erythrocytes were lysed, followed by two washes in ice-cold PBS. Living cells (10 × 10⁶; based on propidium-iodide/annexinV staining) were injected intraperitoneally in 200 μl PBS into 6- to 7-week-old RAG KO mice (RAG12M; Taconic). Blood glucose was tested once weekly for at least 8 weeks.

To test the hypothesis that activation of pancreatic APCs in situ could lead to insulitis, we cloned the CD154 cDNA from mouse spleen cDNA and obtained β-cell specific expression by placing the CD154 cDNA under the control of the RIP (Fig. 1A). The specificity of the construct was tested by transient transfections into the insulinoma cell line Min6 and the fibroblast cell line NIH-3T3. As expected, only Min6 cells expressed detectable levels of the transgene, and this expression was dependent on the RIP because expression was lost when this sequence was omitted from the construct (not shown).

To obtain transgenic mice, the purified pBG-RIP-CD154 construct was injected into (C57BL6/JxCBA)F1 hybrid oocytes, and founder mice were selected by PCR. Immunohistochemical staining of frozen pancreatic sections for insulin and CD154 demonstrated that the transgene was correctly expressed in vivo (Fig. 1C). Of 14 positive transgenic founders, 10 mice expressed the transgene in varying levels, and all of these mice also exhibited a mononuclear infiltration of the islets of Langerhans (not shown). For more detailed analysis, two lines designated L60 and L90 having a medium (L60) or strong (L90) expression of the transgene in the pancreatic β-cells were selected (Fig. 1C). No staining for CD154 was seen in nontransgenic littermates (not shown). To verify that the transgene was not expressed in any other tissue, we performed RT-PCR on RNA extracted from various organs, using primers specific for the transgenically encoded CD154 (Fig. 1A). As expected, the transgene was only expressed in the pancreas of transgene-positive RIP-CD154 mice (Fig. 1B). We therefore conclude that RIP-CD154 mice express CD154 selectively in the pancreatic β-cells.

Immunohistochemical staining of pancreatic tissue for insulin revealed that both RIP-CD154 L60 and L90 had a massive infiltration of the islets of Langerhans. In L60, the infiltration was visible as early as 10 days after birth and was sustained throughout the life of the mouse (Fig. 2A–D). Although the islet morphology was severely compromised early in life, insulin-positive β-cells were easily detectable as late as 18 weeks of age and such remaining insulin-positive cells also expressed CD154 (not shown). In L90, insulitis also started early after birth (Fig. 2E), although intact islets could also be found at this age (Fig. 2F). The insulitis persisted and increased in severity (Fig. 2G–I), and at 18 weeks the morphology of most islets was completely destroyed (Fig. 2J). However, also in this line, insulin-positive islets could be found as late as 18 weeks of age (i.e., after diabetes development). Interestingly, during disease progression, areas with a very high density of lymphocytes could be detected (Fig. 2H and J), probably reflecting the generation of lymphoid structures in the pancreas.

To determine the nature of the infiltrates, we performed immunohistochemical staining of frozen pancreatic sections. This revealed that the cellular infiltrate of the islets of transgene-positive RIP-CD154 mice contained large numbers of CD11c⁺, CD19⁺, CD4⁺, and CD8⁺ cells, demonstrating that both DCs, B-cells, and CD4⁺ and CD8⁺ T-cells were present (Fig. 3A–D). Furthermore, both T-cells and DCs were seen in close proximity with the insulin-producing β-cells (Fig. 3A and B, inserts), and many DCs were visible in the islets, especially around the β-cells (Fig. 3E). It is possible that this accumulation of DCs is due to CD154-CD40 signaling since this has been shown (20) to promote DC survival in vitro. We did not detect any differences with respect to the nature of the infiltrating cells when comparing L60 and L90 (not shown).

Blood glucose regulation was analyzed in both L60 and L90. We found that mice from both lines experienced several hyperglycemic incidents (Table 1 and Fig. 4A and B), and some also developed diabetes. In L60, the diabetes incidence was relatively low (10%, n = 30) and developed relatively late (earliest at 20 weeks of age), whereas over one-half of the transgene-positive mice (17 of 30) experienced at least one hyperglycemic incident (blood glucose >11 mmol/l). In L90, all transgene-positive mice (100%, n = 22) experienced several hyperglycemic incidents, and all mice developed diabetes with an onset time between 8 and 18 weeks (mean 11.8 ± 2.7 weeks, n = 22) (Fig. 4C). Thus, both lines exhibited a heavy infiltration of the islets, and diabetes developed in both lines.

Because L90 had the highest incidence of diabetes, we chose to further characterize this line. When challenged with an intraperitoneal injection of glucose after fasting, pre-diabetic transgene-positive mice (age 6–7 weeks) had a significantly impaired glucose tolerance when compared with their transgene-negative littermates (Fig. 4D). Furthermore, the mice quickly became clinically affected, as seen by an impaired growth rate starting at 10 weeks of age (Fig. 4E). Finally, residual β-cell function, measured as plasma C-peptide levels 30 min after oral glucose challenge in transgene-positive mice (n = 13) after diabetes onset, was significantly reduced (P < 0.02, Mann-Whitney test) compared with transgene-negative controls (n = 10) (data not shown). Together, these results suggested that β-cell function was severely impaired in RIP-CD154 mice and that the increase in blood glucose content was due to an inability of the pancreatic β-cells to produce sufficient amounts of insulin.

Expression of CD154 on β-cells was expected to lead to activation of tissue-resident APCs, followed by an increase of activated DCs in the pancreatic draining lymph nodes.

We therefore analyzed the phenotype of DCs in the pancreas as well as in the pancreatic lymph nodes by immunohistochemistry and by flow cytometry. In the pancreas of L90 transgene-positive RIP-CD154 mice, we found both mature (CD11c⁺MHC-II⁺) as well as more immature DCs (CD11c⁺MHC-II^−/low) (Fig. 5A, panels a–d). The absolute number of cells in the pancreatic draining lymph node of transgene-positive mice was two- to threefold increased (not shown), and the fraction of CD40⁺ cells was doubled (from 9 to 18%) (Fig. 5B, top), possibly due to increased signaling through the receptor. Among these cells, an increase of activated DCs could be detected (CD11c⁺CD40⁺) (Fig. 5B, bottom), suggesting that activated DCs migrate to the pancreatic lymphoid tissue after activation in the pancreas.

To investigate whether T- and B-cells were necessary for development of diabetes, we crossed L90 onto a RAG KO background. As seen in Fig. 6A, transgene-positive RAG^−/− mice did not develop diabetes, whereas transgene-positive RAG^+/− littermates developed diabetes with similar kinetics as their wild-type transgene-positive counterparts (compare Fig. 6A and 4D). Furthermore, histological examination of pancreatic tissue showed less destructive cellular infiltrates in the islets of transgene-positive RAG^−/− mice than in transgene-positive RAG^+/− littermates (Fig. 6B). Immunohistochemical staining showed that the infiltrates in the transgene-positive RAG^−/− mice consisted mainly of CD11c⁺ cells (data not shown), suggesting that the infiltrating cells were DCs.

In order to investigate whether activation of the immune system was confined to the pancreas or if extra-pancreatic β-cells would also be destroyed, we engrafted wild-type C57BL/6 islets under the kidney capsule of established diabetic RIP-CD154 mice and monitored the development of their blood glucose. In all RIP-CD154 mice (6 of 6 transgene positive), the transplanted islets could rescue diabetes, and graft function was sustained for up to 100 days after transplantation (Fig. 6C). When the kidney with the transplant was removed, the mice quickly became diabetic, showing that the normalization of blood glucose was due to the transplanted islets (Fig. 6C, arrows). Immunohistochemical analysis confirmed the functionality of the graft since islet integrity was not compromised (Fig. 6D).

Finally, to test whether diabetes could be readily transferred by spleen cells, we adoptively transferred 10⁷ total spleen cells from diabetic RIP-CD154 mice to RAG KO mice (n = 12). The blood glucose of these mice was monitored for 8 weeks, and during this period none of the recipients developed either insulitis or diabetes (not shown). Together, these data suggest that the autoimmune process was confined to the organ where the DC activation originally occurred and did not result in the generation of β-cell–specific, autoreactive T-cells in the spleen capable of transferring disease.

In this study we show that transgenic expression of CD154 on pancreatic β-cells leads to a rapid development of mononuclear insulitis, including DCs, T- and B-cells, and a subsequent loss of β-cell function, resulting in the development of diabetes. The data are in agreement with the hypothesis that CD154 can activate islet-associated APCs (including DCs) in situ and demonstrate that transgenic expression of CD154 on pancreatic β-cells can lead to insulitis and diabetes.

The infiltration of the islets in the RIP-CD154 mouse was massive, and the islet morphology in these mice was severely disrupted. Both insulin- and glucagon-positive cells were visible at the time of diabetes onset; however, β-cells were fewer, insulin staining was faint, and plasma C-peptide was reduced in many of the diabetic mice (Fig. 2 and data not shown). Thus, the functional β-cell mass was too small to keep the mice normoglycemic at the time of diabetes onset.

Although diabetes developed in both transgenic lines presented here, only a few mice from L60 became diabetic despite massive insulitis. The reasons for this might be that the expression of the transgene in L90 was stronger and more uniform than in L60 (data not shown). Also, although we did not observe any change in phenotype, both lines were backcrossed to C57Bl/6 for three to five generations, and we cannot completely exclude the possibility that remaining CBA alleles may contribute to this discrepancy. Regardless, the data demonstrate that diabetes could develop in two independent lines, although the incidence and kinetics differed.

DCs were initially thought to be important primarily for the development of antigen-specific immunity, but lately an important role for DCs in the maintenance of peripheral tolerance in the steady state has emerged (21,22). Thus, many studies, including tolerance to β-cell–derived antigens (23), have shown that immature DCs induce tolerance (24–28) by taking up antigens in situ and traveling to the draining lymph nodes (29), where they regulate peripheral tolerance, e.g., by deletion of autoreactive T-cells (30–32). DCs are therefore important for sustaining tolerance in the steady state and for induction of immunity during inflammation. We believe that expression of CD154 on the β-cells could trigger the conversion of immature islet-associated DCs into mature DCs, thereby activating the immune system.

If expression of CD154 on pancreatic β-cells leads to activation of tissue-resident APCs, at least two conditions must be fulfilled. First, DCs and macrophages must be present in islets of normal nontransgenic mice. Indeed, both DCs and macrophages can be detected in pancreatic islets in wild-type mice (33), which is consistent with the perception of DCs as sentinel cells present in virtually all organs (5,21). Second, the receptor of CD40 must be present on the APC. Surely, the important role of CD40-CD154 interactions in the conversion of an immature to mature DC is supported by many studies (6–8), and the tolerogenic potential of DCs can be ablated by CD40 ligation (9–11,34). The data presented here are therefore consistent with the hypothesis that activation of DCs by CD40 ligation can change the DC from a tolerogenic to an immunogenic APC.

One other study (35) has investigated the effects of activating DCs in situ by transgenic expression of CD154. In this study, expression of CD154 by a keratin-specific promoter leads to activation of skin-associated Langerhans’ cells and to inflammation of the skin. Our study supports the finding that local expression of CD154 can lead to activation of the immune system and inflammation of the organ expressing the transgene. Furthermore, both studies support a pivotal role of DCs in the inflammatory process, where activation of the DCs in the tissue is most likely the initiation point. It is intriguing, however, that expression of CD154 in the epidermis results in the development of autoreactive T-cells capable of transferring the disease to nontransgenic recipients. In contrast, expression of CD154 in the islet of Langerhans generates an immune response that is confined to the pancreas, since transplanted islets are not rejected and adoptive transfer of spleen cells does not transfer the disease to immunodeficient RAG KO mice. This discrepancy likely reflects 1) the need for initial activation of DCs in the islets, leading to secretion of β-cell cytotoxic and/or T-cell chemotactic factors that are absent in both the transplantation and adoptive transfer model but are required for development of insulitis and diabetes, and 2) the fact that the size of the two target organs are quite different (epidermis versus β-cells). In any case, RIP-CD154⁺ transgenic mice on a RAG^−/− genetic background do not develop diabetes (Fig. 6A), which implies that T- and B-cells are necessary for disease progression from mild insulitis to insulin-dependent diabetes. Therefore, diabetes development in this model is clearly dependent on adaptive immunity. An alternative explanation could be that the transgene per se was disrupting the integrity of β-cell function, thereby leading to diabetes; however, the fact that RIP-CD154⁺ RAG^−/− animals fail to develop diabetes documents that this is not the case. Rather, these animals exhibited a milder cellular infiltration of the islets that only consisted of CD11c⁺ DCs and had no apparent loss of β-cells. This is consistent with our hypothesis that expression of CD154 on β-cells activates local APCs, which in turn are dependent on adaptive immunity for diabetes to develop. We expected to be able to adoptively transfer diabetes by injection of spleen cells from diabetic RIP-CD154 mice into RAG KO recipients. However, the fact that disease could not be transferred does not necessarily contradict this hypothesis since similar experiments in other transgenic models of type 1 diabetes showed that disease was not readily transferable (36).

Although artificial in nature, the RIP-CD154 transgenic mouse may well illustrate early events in the development of type 1 diabetes. It is highly likely that the first event toward development of disease is activation of islet-associated DCs by some environmental trigger (e.g., a virus). This trigger is probably transient in nature but could potentially initiate an autoimmune reaction, provided that a sufficient number of autoreactive T-cells and/or a deficiency in regulatory T-cells are present.

In conclusion, we have presented a transgenic mouse model where the consequences of a constitutive activation of DCs in the pancreas can be studied. The RIP-CD154 develops a form of insulin-dependent diabetes, which in many ways resembles type 1 diabetes in human patients. It is our hope that this model can serve to study the vast effects of inducing DC maturation in vivo. It will be especially important in protocols of organ transplantation and tolerance induction to understand the mechanisms in more detail.

This study was supported in part by the Danish Ministry of Science, Technology and Development (to C.H.), by a Freja research grant from the Danish Research Agency (to B.K.M, grant no. 5008-01-0003), and institutional funds from Novo Nordisk. Hagedorn Research Institute is an independent basic research component of Novo Nordisk.

The authors thank Trine Larsen for excellent technical assistance; the employees of the Animal Unit, Gentofte, for taking care of the animals and for assistance in animal experiments; Drs. Lars Hornum and Dorthe Lundsgaard for critically reading the manuscript; and Dr. Helle V. Petersen for help in testing the transgenic construct in Min6 transfections.