Infiltration of pancreatic islets by immune cells, termed insulitis, increases progressively once it begins and leads to clinical type 1 diabetes. But even after diagnosis some islets remain unaffected and infiltration is patchy rather than uniform. Traffic of autoreactive T cells into the pancreas is likely to contribute to insulitis progression but it could also depend on T-cell proliferation within islets. This study utilizes transgenic NOD mice to assess the relative contributions of these two mechanisms. Progression of insulitis in NOD8.3 TCR transgenic mice was mildly reduced by inhibition of T-cell migration with the drug FTY720. In FTY720-treated mice, reduced beta cell MHC class I expression prevented progression of insulitis both within affected islets and to previously unaffected islets. CTL proliferation was significantly reduced in islets with reduced or absent beta cell expression of MHC class I protein. This indicates that intra-islet proliferation, apparently dependent on beta cell antigen presentation, in addition to recruitment, is a significant factor in progression of insulitis.
Type 1 diabetes (T1D) is an autoimmune disease in which the insulin-producing beta cells of the pancreas are destroyed by T cells. One of the striking pathological features of T1D in humans is that infiltration and destruction of islets is not uniform or synchronous . Some islets can have florid infiltration while others are unaffected even up to 3 years after diagnosis (P. Campbell and T. Kay, unpublished data). Insulitis begins years before diagnosis, which suggests this variation can persist for long periods of time . The patchiness of insulitis is also observed in diabetes-prone NOD mice [2, 3]. Once insulitis begins in NOD mice it is generally progressive with increasing numbers of T cells and increasing beta cell destruction, especially in female mice. One explanation for variable development of insulitis could be that newly activated lymphocytes are continuously recruited to each islet in turn over time or alternatively, a small number of lymphocytes could randomly home to islets and then proliferate extensively. The aim of the current study was to determine the relative contributions of recruitment to the islet and proliferation within the islet in the development of insulitis.
Results and discussion
Continuous recruitment and local proliferation contribute to insulitis
Progression of insulitis may require lymphocyte trafficking from outside the pancreas. FTY720 the sphingosine-1 phosphate (S1P) receptor agonist prevents lymphocyte egress from lymph nodes by inducing the internalization of the S1P receptor on lymphocytes, preventing lymphocytes from responding to S1P gradient in the efferent lymphatics, trapping the lymphocytes within the node [3-7]. We treated NOD8.3 mice with FTY720, NOD8.3 mice are a CD8+ TCR transgenic strain specific for islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP206–214) presented by H-2Kd . Treatment began at 35 days when the mice have very mild peri-insulitis and was continued for 2 weeks. We analyzed peripheral blood for lymphocyte populations to ensure FTY720 was successful at preventing lymphocyte trafficking and found that the FTY720-treated mice were lymphopenic (Fig. 1A). Insulitis in untreated mice progressed to invasive insulitis, scores of 3 (4.5% at 35 days and 36.08% at 52 days) and 4 (0% at 35 days and 31.64% at 52 days) (Fig. 1B). FTY720 treatment resulted in a modest reduction in insulitis (p < 0.01) indicating that lymphocyte trafficking contributes to insulitis development. However, insulitis clearly progressed from day 35 in the FTY720-treated mice as we detected an increase in the number of islets scoring 3 (22.2%) and 4 (21.0%) after treatment. The total number of islets with insulitis after treatment also increased from 25% at 35 days of age to 59.16% at 52 days of age. This suggests intrapancreatic migration from affected to unaffected islets also contributes to insulitis progression. These observations suggest that in addition to T-cell trafficking from the outside the pancreas, local proliferation within the islet contributes to insulitis progression.
We also treated NOD8.3/RIP-SOCS1 mice with FTY720. These mice overexpress SOCS1 in the beta cells resulting in reduced MHC class I expression, antigen presentation, and cytokine secretion. NOD8.3/RIP-SOCS1 mice develop massive insulitis but diabetes does not occur. Our previous data suggest this is due to impaired beta cell recognition by autoreactive CTLs [9, 10]. Insulitis progression in untreated mice was impaired compared to NOD8.3 mice. At 52 days, only 47.3% of NOD8.3/RIP-SOCS1 islets had developed invasive insulitis (31.96% score 3 and 15.35% score 4) compared to 67.72% of NOD8.3 islets (36.08% score 3 and 31.64% score 4). This suggests that local islet factors including beta cell MHC class I expression, which is reduced by SOCS1 overexpression, stimulate local CTL proliferation. FTY720 treatment further stalled the progression of insulitis in NOD8.3/RIP-SOCS1 mice. Similar number of islets scored 0 (69.25% at 35 days and 74.52% at 52 days) and 1 (16% at 35 days and 9.94% at 52 days), before and after treatment (Fig. 1B). This confirms that both continuous trafficking and local proliferation are required for the progression of insulitis.
Local proliferation of diabetogenic cytotoxic T cells requires beta cell MHC class I
To investigate if beta cell MHC class I expression does stimulate intra-islet CTL proliferation, we measured the proliferation of CTLs within islets of NOD8.3 and NOD8.3/RIP-SOCS1 and class I beta-bald/8.3 mice, by giving a short (1 h) pulse of BrdU. This limits the incorporation of BrdU to cells in S-phase . The class I beta-bald/8.3 mice do not express MHC class I on beta cells due to a beta cell specific conditional deletion of β2M . A total of 19.6 ± 4.2% of IGRP-specific CTLs within the islets of NOD8.3 mice incorporated BrdU during the 1 h pulse (Fig. 2A and B, gating strategy for flow cytometric analysis is shown in Supporting Information Fig. 1A). As the 1 h pulse confines BrdU incorporation to cells proliferating within tissues and limits detection of cells recruited from outside the islets, this shows that 8.3 CTLs proliferate within the islet. BrdU incorporation was significantly reduced, to 7.2 ± 0.89%, in the NOD8.3/RIP-SOCS1 islets (p = 0.04) and, to 6.7 ± 0.16%, in the class I beta-bald/8.3 islets (p = 0.04). We found a similar reduction in BrdU incorporation by CTLs from NOD8.3/RIP-SOCS1 islets when BrdU was administered for 24 h prior to analysis (Supporting Information Fig. 2A and B).
We also compared the BrdU uptake of CFSE-labeled 8.3 CD8+ T cells adoptively transferred into three recipient strains with beta cell MHC class I expression that was normal (NOD), reduced (RIP-SOCS1), or completely eliminated (beta-bald). Beta cell MHC class I expression was confirmed by flow cytometry for each strain (Supporting Information Fig. 3). Recipients received a 1 h pulse with BrdU 5 days after transfer and proliferation of 8.3 CD8+ CTLs was assessed by CFSE dilution and BrdU incorporation. The gating strategy for the flow cytometric analysis is shown in Supporting Information Fig. 1B. No differences in CFSE dilution or BrdU incorporation were seen in the pancreatic lymph nodes (PLNs) (Fig. 3A). Within the islets of the NOD recipients, 8.3 CTLs divided several times as shown by their extensive CFSE dilution, with the majority of cells detected in divisions four to seven. Of the divided CFSE+CD8+ T-cell population, 30.3 ± 4.47% incorporated BrdU (Fig. 3B and C) indicating some of this division had occurred within the 1 h since BrdU was given. Incorporation of BrdU was reduced to 16.68 ± 3.5% in the divided CFSE+CD8+ CTL population from RIP-SOCS1 islets (p < 0.04) and further reduced to 12.2 ± 0.41% in the CTLs from the class I beta-bald islets (Fig. 3B and C). Therefore, both proliferation over the 5 days since injection, measured by CFSE dilution, and proliferative activity over the previous hour, measured by BrdU incorporation, were significantly reduced when MHC class I expression by beta cells was decreased.
Finally, we compared the proliferation of all CD8+ T cells infiltrating the islets of NOD and RIP-SOCS1 mice, using the 1 h pulse of BrdU. A total of 16.63 ± 2.9% of CD8+ T cells present in NOD islets incorporated BrdU and this was significantly reduced to 6.65 ± 0.09% in RIP-SOCS1 islets (p = 0.009) (Fig. 3D and E; the gating strategy for the flow cytometric analysis is shown in Supporting Information Fig. 4). CD8+ T-cell proliferation in the PLNs was similar for NOD (4.59 ± 1.25%) and RIP-SOCS1 (6.46 ± 1.53%). These data suggest that antigen presented on the beta cells surface stimulates local proliferation of CTLs. Collectively our observations demonstrate that both recruitment and local proliferation are required for insulitis progression and beta cell MHC class I expression is required for intra-islet proliferation.
The requirements for the initiation of insulitis in an individual islet are unknown. Trafficking from outside the pancreas, via the draining lymph node must play a part in insulitis, so it was expected that FTY720 would diminish insulitis. However, we expected this reduction to be much greater and were surprised to find that suppression of intra-islet proliferation was also required to convincingly reduce insulitis progression. Our data also shed light on a related question; can immune cells move between islets, or is direct recruitment from the circulation required for infiltration of each individual islet? By examining the percentage of uninfiltrated islets (score = 0) in NOD8.3 mice, we found that FTY720 treatment preserved some islets without insulitis, as did the inhibition of local expansion in NOD8.3/RIP-SOCS1 mice. This suggests that CTL migration from outside the pancreas and from one islet to another occurs during insulitis progression.
FTY720 treatment has been used previously to prevent the development of diabetes in NOD mice [3, 6]. Penaranda et al.  found that FTY720 treatment of 10- or 14-week-old NOD mice, after insulitis is established, stabilized insulitis progression and prevented diabetes, provided treatment was given continuously. Within the insulitis of the treated mice a two- to threefold reduction in CD4+ T cells was observed. Interestingly, no change in infiltrating CD8+ T cells was detected. In light of our data, it is likely that the CD8+ T cells underwent local proliferation resulting in a consistent population within the islets. Collectively these data suggest CD4+ and CD8+ T cells might use continuous recruitment and local proliferation differently during insulitis development.
Sarukhan et al.  performed sequence analysis of the CDR3 regions of Vβ6 and Vβ8.2 TCRs of T cells infiltrating individual islets isolated from NOD mice. They observed nonoverlapping sets of CDR3 sequences, suggesting each islet lesion develops auto-nomously by local proliferation. Our observation of local proliferation is consistent with this previous study, however, our data also suggest movement between islets. It is likely the different experimental approaches can account for these discrepancies and further work is required to understand if T cells are capable of moving from islet to islet.
The migration of CD8+ T cells into islets is dependent on the expression of their cognate antigen within the islet therefore the local proliferation is antigen specific . We have interpreted this as being a consequence of direct CTL recognition of antigen on the beta cell surface, and reducing or eliminating MHC class I impairs the local proliferation. However, eliminating beta cell MHC class I could indirectly reduce the amount of antigen available for intra-islet DCs to cross-present to the CTLs and stimulate proliferation. Eliminating class I would prevent beta cell destruction and lower the amount of antigen released for cross-presentation by DCs. We did not observe a difference in proliferation of CTLs in PLNs (Fig. 2 and 3A), which suggests sufficient antigen is present for DC cross-presentation. Alternatively, DCs could acquire beta cell peptides via gap junctions or gain MHC class I and antigen through membrane sharing with the beta cell, a process termed cross dressing and stimulate local proliferation [15, 16]. These processes do not occur without expression of class I and could account for reduced proliferation in beta-bald mice, but a reduction in proliferation was also seen in RIP-SOCS1 mice where class I is still present arguing against cross dressed DCs stimulating local proliferation. We did detect some proliferation within islets in all our models, which we attribute to intra-islet DCs, and further work is necessary to clarify the role of intra-islet DCs in antigen presentation to CTLs within islets.
We have recently investigated how islet-reactive CTLs acquire cytotoxic effector function. We identified that CTLs are restimulated within the islets and that this increases the expression of cytotoxic effector molecules . These increases occurred independently of beta cell antigen presentation. Although beta cell antigen presentation is not required for a CTL to acquire cytotoxic effector function, our current data show it is a significant requirement for local proliferation and the progression of insulitis.
In conclusion, progression of insulitis can be due to T-cell proliferation within islets as well as traffic from outside the pancreas. To our knowledge, this is the first direct evidence that the target cell, the beta cell in this case, stimulates proliferation of activated effector CTLs. Understanding insulitis development and the factors that drive its progression is clinically important. This is particularly relevant to CTLs as CD8+ T cells are the main component of insulitis in diabetic patients . Clinical trials aimed at diabetes prevention usually intervene after the onset of insulitis. Our data suggest that even if peripheral pathogenic T cells are altered by an intervention, local proliferation within the islet could still contribute to disease progression. Therefore, successful treatment of autoimmunity will need to eliminate activated autoreactive T cells already in the target tissue as well as preventing further migration. This study highlights the interaction between beta cells and CTLs as more complex then a simple bridge required for the delivery of cytotoxic molecules, and suggests beta cells contribute to their own destruction by stimulating local proliferation.
Materials and methods
All mice were bred and maintained at the St. Vincent's Institute animal facility (Fitzroy, Victoria, Australia). NOD/Lt mice were purchased from the Walter and Eliza Hall Institute animal breeding facility, Melbourne, Australia. NOD8.3, NOD8.3/RIP-SOCS1, NODRIP-SOCS1, and class I MHC “beta-bald” mice have been described previously [8, 9, 12]. Beta-bald and 8.3 mice were bred together to generate class I beta-bald/8.3 mice. The institutional animal ethics committee approved all experiments.
The S1P receptor-1 agonist FTY720 was purchased from Cayman Chemicals, Ann Arbor, MI, USA. Mice were treated with 2 mg/kg FTY720 in water by i.p. injection.
Histology and immunohistochemistry
Pancreata were dissected and snap frozen in OCT (Sakura Finetek, Torrance, CA, USA). For histological scoring, 5 μm sections were acetone fixed and stained for insulin using a guinea pig anti-insulin polyclonal (Dako Cytomation, Carpenteria, CA, USA) Ab followed by a horseradish peroxidase-conjugated anti-guinea pig Ig (Dako Cytomation) as described previously [9, 18]. Insulitis was scored on three sections, 200 μm apart, using the following scale: 0 = no infiltrate, 1 = peri-islet infiltrate, 2 = extensive (>50%) peri-islet infiltrate, 3 = intra-islet infiltrate, and 4 = extensive intra-islet infiltrate (>80%) or total beta cell loss.
BrdU incorporation and adoptive transfer
Mice were injected with 3 μg BrdU (Sigma Aldrich) in PBS by i.p. injection 1 h prior to harvest. For adoptive transfer, CD8+ T cells from 8.3 mice were labeled with CFSE as previously described . Cells were resuspended at 2.5 × 107/mL in PBS and 200 μL injected i.v. into the tail vein of recipient mice.
Islets of Langerhans were isolated using collagenase P (Roche, Basel, Switzerland) and histopaque-1077 density gradients (Sigma Aldrich) as previously described [5, 9].
Lymph nodes harvested from recipient mice were prepared as single cell suspensions. Islets were dispersed to single cells as previously described . Anti-CD8 (Ly2(53–6.7) conjugated to PE or allophycocyanin were purchased from BioLegend, San Diego, CA, USA. The peptide IGRP206–214 (VYLKTNVFL) was purchased from Auspep (Melbourne, Australia) and the H2-Kd tetramers made by ImmunoID (Melbourne, Australia). For BrdU staining, the allophycocyanin-conjugated anti-BrdU was purchased with the BrdU Flow kit from BD Biosciences. Allophycocyanin-conjugated rat IgG1 (R3–34) (BD Pharmingen) was used as isotype control. Detection of BrdU incorporation was carried out according to the manufacturer's specifications using the BrdU Flow kit. All analysis was performed on a FACSCalibur (Becton Dickinson, Franklin Lakes, NJ, USA) and using FlowJo analysis software (Treestar, Ashland, OR, USA).
Analysis of data was performed using GraphPad Prism (GraphPad Prism Software, San Diego, CA, USA) and the Student's unpaired t test or two-way ANOVA used to assess statistical significance. Error bars on all graphs represent the standard error of the mean.
We thank Lorraine Elkerbout for geno-typing and Stuart Mannering for helpful discussions. K.L.G received a Postdoctoral Fellowship from the Juvenile Diabetes Research Foundation (JDRF) and a Skip Martin Early Career Postdoctoral Fellowship from the Australian Diabetes Society. B.K. received a Career Development Award from JDRF and a Clinical Research Excellence Fellowship from the National Health and Medical Research Council of Australia (NHMRC). P.S. is a Scientist of Alberta Innovates — Health Solutions and is supported by the Canadian Institutes of Health Research and the Natural Sciences and Engineering Research Council of Canada. H.E.T received a Career Development Award from the NHMRC. T.W.H.K. received a Millennium Research Grant from Diabetes Australia, a program grant from JDRF and a program grant form NHMRC. St Vincent's Institute receives support from the Operational Infrastructure Support Scheme of the Government of Victoria. The JMDRC is supported by the Canadian Diabetes Association.
The authors declare no financial or commercial conflicts of interest.
Islet-specific glucose-6-phosphatase catalytic subunit-related protein