CD40‐mediated signalling influences trafficking, T‐cell receptor expression, and T‐cell pathogenesis, in the NOD model of type 1 diabetes

Summary

CD40 plays a critical role in the pathogenesis of type 1 diabetes (T1D). The mechanism of action, however, is undetermined, probably because CD40 expression has been grossly underestimated. CD40 is expressed on numerous cell types that now include T cells and pancreatic β cells. CD40⁺ CD4⁺ cells [T helper type 40 (TH40)] prove highly pathogenic in NOD mice and in translational human T1D studies. We generated BDC2.5.CD40^−/− and re‐derived NOD.CD154^−/− mice to better understand the CD40 mechanism of action. Fully functional CD40 expression is required not only for T1D development but also for insulitis. In NOD mice, TH40 cell expansion in pancreatic lymph nodes occurs before insulitis and demonstrates an activated phenotype compared with conventional CD4⁺ cells, apparently regardless of antigen specificity. TH40 T‐cell receptor (TCR) usage demonstrates increases in several Vα and Vβ species, particularly Vα3.2⁺ that arise early and are sustained throughout disease development. TH40 cells isolated from diabetic pancreas demonstrate a relatively broad TCR repertoire rather than restricted clonal expansions. The expansion of the Vα/Vβ species associated with diabetes depends upon CD40 signalling; NOD.CD154^−/− mice do not expand the same TCR species. Finally, CD40‐mediated signals significantly increase pro‐inflammatory Th1‐ and Th17‐associated cytokines whereas CD28 co‐stimulus alternatively promotes regulatory cytokines.

Introduction

CD40 plays a pivotal role in type 1 diabetes (T1D), and yet its role in pathogenesis continues to emerge. CD40 was defined as a co‐stimulus for B cells, promoting cell survival1 and antibody class switch.2 CD40 is expressed on other professional (MHC‐II⁺) antigen‐presenting cells, i.e. macrophages and dendritic cells (DC). A more surprising discovery was that CD40 is induced on T cells, where it functions as a biomarker for pathogenic T cells.3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21 It became clear that CD40‐expressing CD4⁺ cells are pathogenic in the mouse model of T1D as well as in clinical studies of patients with T1D.5, 6, 8, 9, 10, 11, 12, 13, 17, 18, 19, 21, 22, 23 CD40 expression on CD8⁺ cells was described.24 Further complicating this story, CD40 expression has been discovered on a wide array of non‐immune cells including endothelial cells,25 neuronal cells,26 adipocytes,27 epithelial cells28 and, of particular interest to T1D, pancreatic islet beta cells.29

Pathogenic B cells producing auto‐antigen reactive antibodies (auto‐antibodies) are involved in T1D development, but an association of CD40 with the progression towards auto‐antibody production has not been defined. In the case of macrophages and DC, CD40 induces macrophage production of pro‐inflammatory cytokines including interleukin‐1α (IL‐1α), IL‐1β, tumour necrosis factor‐α (TNF‐α) and IL‐6.30, 31 CD40 triggers iNOS and nitric oxide production leading to tissue damage.32 CD40 engagement on DC creates licensing,33, 34, 35, 36 impacting the way in which DC interact with T cells.37 DC that are high CD40 expressers promote T helper type 1 (Th1) cell development whereas CD40^low or CD40^– DCs favour regulatory T cell development.38 CD40 expression on T cells was unexpected and we devoted much attention to defining CD40⁺ CD4⁺ T cells.5, 6, 8, 9, 10, 11, 12, 13, 17, 18, 19, 21, 22, 23, 39 We referred to these cells as Th40, because the cells are CD4⁺ helper cells that transcend classical T helper definitions. Th40 cells produce interferon‐γ (IFN‐γ) and TNF‐α typically associated with Th1 phenotype, but also produce IL‐17 and IL‐21 associated with Th17 cells.5 In addition, both RORγt, the defining transcription factor for Th17 cells, and t‐bet, the defining transcription factor for Th1 cells, are produced by Th40 cells simultaneously.13

T cells are critical to T1D development.40 Islet antigen‐specific T‐cell clones were generated advancing the autoimmune aspects of T1D.41, 42 One of the first diabetogenic T‐cell clones described was BDC2.5;42 the antigen specificity for BDC2.5 recently was defined as a chromogranin‐A–insulin peptide hybrid.43 BDC2.5 T cells rapidly and efficiently transfer T1D symptoms, including the hallmark hyperglycaemia, to NOD.scid recipient mice. An insulin‐responsive CD8⁺ T‐cell clone was also identified.44 CD8⁺ cells are detected in pancreatic islets before and during diabetes development. Transfer of T1D using primary T cells has historically been complicated. Early studies required an entire splenic complement of cells for disease transfer. In contrast, purified Th40 cells from diabetic mice or from pre‐diabetic mice (9 weeks of age, which equates to early/mid‐level insulitis) used at a relatively small number, transferred T1D with very rapid kinetics.21 The Th40 cells do not contain the regulatory T cell population (defined as CD4⁺ CD25^hiFoxP3⁺);11 and an observation was that Th40 cells are responsive to classic regulatory T cells, but much less so when they derive from autoimmune sources; interestingly the mechanism involved transforming growth factor‐β.11

Here, we considered mechanistic roles for CD40 during diabetogenesis, focused on T cells including how CD40 overall expression influences trafficking, T‐cell receptor usage, and CD40‐mediated T‐cell co‐stimulation. CD40 acts as a T‐cell co‐stimulus;20, 45 but its mechanistic influence on diabetes is poorly understood. We show now that CD40 expression or CD40‐mediated signals are required for diabetes development and for insulitis. The T‐cell receptor (TCR) repertoire of Th40 cells remains varied throughout diabetogenesis, with specific, yet unique TCR molecules becoming prominent at different stages of disease. This suggests that multiple antigens are required throughout disease development rather than disease being dependent upon few antigens and restricted clonal expansions. One prominent TCR constituent throughout disease development is Vα3.2. This TCR sequence from mice is highly homologous with a TCR, Vα8‐3*02, sequence that was identified out of human pancreatic lymph nodes (PLN).46 CD40 engagement of Th40 cells induces pro‐inflammatory cytokines, including Th1‐ and Th17‐associated cytokines. These data suggest that CD40 is a prime target for therapeutic approach to induce sustained tolerance. As suggested here, the mechanisms of tolerance would include TCR usage, influence on cell trafficking and cytokine re‐programming; each of these would have discernable impact on the auto‐inflammatory state.

Materials and methods

Mice

NOD mice were purchased from Taconic Laboratories (Hudson, NY). NOD.BDC2.5 and NOR mice were purchased from Jackson Laboratories (Bar Harbor, ME). NOD.CD154^−/− and BDC2.5.CD40^−/− were derived for us by Charles River Laboratories (Wilmington, MA). Transgenic and knockout status were confirmed by the vendors involving up to 131 microsatellite DNA sequences. After derivation, NOD.CD154^−/− and BDC2.5.CD40^−/− mice were bred in‐house and genetic status was tested regularly. Mice were maintained and treated under a University of Colorado Anschutz Medical Campus Institutional Animal Care and Use Committee (IACUC) approved protocol. All mice were housed in an Association for Assessment and Accreditation of Laboratory Animal Care approved vivarium with regular veterinary service.

Diabetes incidence and adoptive transfers

Mice were monitored weekly, beginning at 6 weeks of age through to 50 weeks of age, for hyperglycaemia measured by serum glucose. A blood glucose of 250 mg/dl, for three consecutive readings over a 1‐week period was considered to indicate diabetes. Mice that became diabetic were killed using an IACUC‐approved protocol. NOD mice at 4 weeks of age were injected intravenously with 50 μg 1C10, an agonistic anti‐CD40 monoclonal antibody, once weekly for 3 weeks.

Adoptive transfers were performed using 1 × 10⁷ BDC2.5 T‐cell clones that were maintained as described elsewhere.47 Primary Th40 cells were isolated from spleens of NOR mice. Spleens were collected and disrupted to generate single‐cell suspensions. A red blood cell lysing buffer was used. Cells were passed through lympholyte‐M to collect lymphocytes. Cells were washed and treated with Miltenyi anti‐MHC‐II, anti‐CD11b and anti‐CD19 followed by Miltenyi AutoMacs on cell‐delete setting (Miltenyi Biotec, Bergisch Gladbach, Germany). CD40⁺ T cells were isolated by treating the remaining cells with anti‐CD40 biotinylated antibodies #bib1C10, 4F11 and FGK45 followed by avidin‐conjugated beads, then passed through the AutoMacs on positive‐select setting. Cell purity was 94·6% CD4⁺ CD40⁺. Cells at 2·5 × 10⁶ were injected intraperitoneally into NOD.scid mice at 10–15 days old. NOD thymocytes were purified for CD4⁺ CD40⁺ cells as above. BDC2.5.CD40^−/− were purified CD4⁺ that had been CD25⁺ depleted.

Pancreatic histology

Pancreata were excised from mice, paraformaldehyde‐fixed, embedded in paraffin and sectioned. Sections were stained with haematoxylin & eosin and examined by microscopy at 20 × resolution. Insulitis scores were: 0 = no infiltrates; 1 = one pole infiltrate, < 25% of the islet; 2 = two polar infiltrate, c.50% of the islet; 3 = islet infiltrate > 75% of the islet.

Generating pancreatic T cells

Pancreata were excised from diabetic NOD mice and placed in 0·2% collagenase solution for 30 min at 37°. Single‐cell suspensions were created by passing the slurry that contained course tissue through a sieve using a syringe plunger. The slurry was passed through lympholyte‐M to isolate T cells and other lymphocytes. The tissue that migrated to the bottom during lympholyte passage was re‐suspended in PBS and passed again through lympholyte‐M. The lymphocyte layer was collected, washed three times with PBS and stained for flow cytometry.

Cell staining and flow cytometry

Antibodies: Anti‐CD4 generated in house from clone GK1.5 and labelled with Alexafluor 770 or clone H129.19 with FITC; anti‐CD8 from eBioscience (San Diego, CA) or Miltenyi Biotec (Auburn, CA), anti‐CD3 generated in house, clone 145.2C11 with phycoerythrin (PE); anti‐CD40, generated in house, clone 1C10 with Alexafluor 405, or FITC, or purchased from eBioscience with PE‐Cy5; anti‐TCR, clone H57.597 purchased from eBioscience in PE‐Cy5; anti‐CD44 [PE‐Cy5, FITC, allophycocyanin (APC)], anti‐CD62L (PE, PE‐Cy5, APC) and anti‐CD69 (APC or FITC) are all from eBioscience. Specific TCR Vα and Vβ antibodies, all in FITC, were from eBioscience.

Cytokine production

Lymphocytes were purified from spleens of female NOD mice, aged 9–12 weeks, defined as ‘pre‐diabetes’ extensive insulitis and euglycaemic, using lympholyte‐M. Th40 cells were further purified by depleting MHC‐II⁺ cells followed by CD40 sorting on a Miltenyi AutoMacs. Purified Th40 cells were plated at 1 × 10⁶ cells per well in round‐bottom, 96‐well plates. Cells were treated with isotype antibodies (Controls); anti‐CD3 (1 μg/ml) + anti‐CD40 (5 μg/ml); or anti‐CD3 + anti‐CD28 (5 μg/ml) for 1 hr, then washed, returned to the plate and incubated. After 24 hr, cell supernatants were collected and assayed for cytokine production using a Flow Cytomix kit from Miltenyi Biotec.

Results

CD40 expression is required for diabetes and insulitis

To better dissect the role of CD40 during diabetes we re‐derived a CD154 knockout on the NOD background, and generated a CD40 knockout on the BDC2.5 TCR transgenic background. NOD.CD154^−/− mice have Th40 cells at low percentages in younger mice that become expanded in number by 45 weeks of age (see Supplementary material, Fig. S1). Thymic development in NOD.CD154^−/− mice is skewed with a smaller proportion of CD4⁺ mature cells and expanded percentage of mature CD8⁺ cells (see Supplementary material, Fig. S1). Th40 cells develop in the thymus under restricted CD40 signalling conditions, but unlike in NOD mice, they localized to the CD4⁺ CD8^lo population, with very few Th40 cells detected in mature CD4⁺ cells (see Supplementary material, Fig. S1). Diabetes development in NOD.CD154^−/− and BDC2.5.CD40^−/− mice did not occur through to 50 weeks of age (Fig. 1a). NOD mice develop T1D with normal kinetics, 80% being diabetic by 18 weeks of age (Fig. 1a), as we, and others have shown.6, 8, 9, 10, 11, 12, 13, 17, 18, 19, 21, 22, 23 NOD mice injected with anti‐CD40 between 4 and 6 weeks of age broke tolerance earlier with more pronounced incidence (Fig. 1a). NOR mice are NOD congenic, containing 85% of NOD genetics,48, 49 including the disease‐decisive MHC, I‐Ag7, yet NOR mice did not develop diabetes through to 50 weeks (Fig. 1a).

Figure 1.

CD40 expression is required for diabetes development. (a) Diabetes incidence: NOD mice (12 female); NOD mice (6 female) injected intravenously with anti‐CD40, 1C10 at 50 μg in 100 μl; NOD.CD154^−/− (15 female); BDC2.5.CD40^−/− (15 female) and NOR (6 female) mice were examined by weekly blood glucose levels for 50 weeks. Mice were considered diabetic when blood glucose was ≥ 250 mg/dl for three consecutive readings. Data are reported as per cent of the total cohort that are diabetic. There was a significant (P = 0·031) difference between NOD and NOD 1C10 Trx as determined by a paired t‐test. (b) Adoptive transfers: NOD.scid recipients were given BDC2.5 T‐cell clones (6 mice, 1 × 10⁷ cells), purified Th40 cells from NOR mice (6 mice, 2·5 × 10⁶ cells), purified CD40⁺ NOD thymocytes (6 mice, 2·5 × 10⁶ cells) or total T cells isolated from BDC2.5.CD40^−/− mice (6 mice, 1 × 10⁷ cells). Mice were monitored by blood glucose for diabetes. Experiments were terminated after 60 days.

T‐cell pathogenicity is demonstrated through adoptive transfers; in this case using NOD.scid recipients. Th40 cells from diabetic NOD mice transfer disease readily.11, 17, 21 CD40⁺ thymocytes from NOD mice, representing early developmental T cells, transfer hyperglycaemia (Fig. 1b). Th40 purified cells isolated from NOR mice transfer hyperglycaemia (Fig. 1b). BDC2.5.CD40^−/− T cells do not transfer diabetes (Fig. 1b), even when greater numbers of cells were used, suggesting that CD40 expression directly on otherwise pathogenic (BDC2.5) T cells, if not required, at least strongly promotes disease development.

Pancreata from NOD.CD154^−/− and BDC.2.5.CD40^−/− mice were examined for insulitis. CD40–CD154 interactions are critical for diabetes development.50 Insulitis was not detected in NOD.CD154^−/− mice even through to 50 weeks (Fig. 2a,e). Unlike CD40^−/− mice, NOD.CD154^−/− mice develop Th40 cells but the cells occur at low frequency, eventually expanding in number over time (see Supplementary material, Fig. S1). Insulitis does not develop in BDC.2.5.CD40^−/− even through to 45 weeks (Fig. 2b,e). This finding is significant given that BDC2.5.CD40^−/− T cells carry an islet antigen‐specific TCR, a chromogranin‐A/insulin hybrid protein,43 but are unable to traffic to the pancreas. NOD mice from our colony at 6 weeks of age demonstrate only limited insulitis, level 1 infiltration that occurred in a few islets (Fig. 2c,e). As expected, extensive insulitis develops in wild‐type NOD mice (Fig. 2d,e). Examining Th40 cell numbers during diabetogenesis in comparison to percentage of insulitis over time showed that as insulitis percentages increase, so do Th40 cell numbers in spleen (Fig. 2f). Expansion of Th40 cells in spleen precedes insulitis scores.

Figure 2.

CD40 requirements for insulitis. Pancreata were excised from NOD mice at (a) 6 and (b) 16 weeks, (c) BDC2.5.CD40^−/− mice at 45 weeks and from (d) NOD.CD154^−/− mice at 45 weeks. In each case, at least six mice were examined. Pancreata were sectioned in the tail, body and head of the pancreas. Fixed sections were stained with haematoxylin & eosin to visualize lymphocyte infiltrates. Representative sections are shown. Panels are 20 × resolution. (e) Cumulative insulitis scores. At least 100 islets from three different sections from six mice in each cohort, were examined and scored: 0 = no infiltrate; 1 = one pole infiltration, < 25% of islet; 2 = two poles infiltration, at least 50% of islet; and 3 = > 75% of islet is infiltrated. (f) Th40 cell levels in spleen relative to insulitis. Insulitis was determined by histology. Th40 cell percentage was determined by taking spleens, creating single‐cell suspensions and staining for Th40 cells as described in the Materials and methods section. Data represent at least three NOD mice at each stage.

Th40 cells from pancreatic lymph nodes expand in number over time and have an activated phenotype

Th40 cells from PLN were examined before insulitis, at a defined ‘pre‐diabetic’ time, equating to moderate insulitis, and in diabetic mice. In PLN from young mice before insulitis, aged 4 weeks, Th40 cell percentage (and actual cell number, data not shown) were relatively low (Fig. 3a). At 9–12 weeks during moderate insulitis, Th40 cell numbers have effectively doubled (Fig. 3b). At diabetes onset, Th40 cell numbers in PLN are substantially increased (Fig. 3c), again effectively doubling in number from that seen in pre‐T1D. When T cells encounter cognate antigen one consequence is cell surface down‐regulation of the TCR–CD3 complex.51 Total T cells isolated from PLN of diabetic NOD mice demonstrate TCR^hi and TCR^low sub‐populations (Fig. 3d). In diabetic mice, Th40 cells from PLN are predominantly TCR^low, whereas the CD40⁻ CD4⁺ conventional population is predominantly TCR^hi (Fig. 3d). Both the Th40 and CD4⁺ CD40⁻ subsets of cells from PLN are predominantly CD69⁺ (Fig. 3e). CD69 expression classically has been associated with activation, however, studies now show that CD69 expression is involved in tissue retention.52, 53 We examined 12 diabetic mice and in all cases Th40 cells comprised a large portion of CD4 T cells in PLN and the majority of those cells were CD69⁺ (Fig. 3f).

Figure 3.

Th40 cells in pancreatic lymph nodes of NOD mice. Pancreatic lymph nodes were excised from (a) NOD mice (six female mice) at 4 weeks of age, before demonstrable insulitis; (b), 9–12 weeks of age, (six female mice), and (c) ≥ 18 weeks, diabetic (12 female mice) mice. Single‐cell suspensions were analysed for Th40 cells. Black dots are isotype control overlay in the dot plot. Th40 cells from diabetic NOD mice were further analysed for (d) T‐cell receptor (TCR) expression or (e) CD69 expression levels. (f) Th40 cell percentages and CD69 percentages within the Th40 cell subset within each diabetic mouse examined. [Colour figure can be viewed at wileyonlinelibrary.com]

TCR usage in NOD mice through diabetogenesis

We examined potential changes in the TCR repertoire restricted to PLN over time as NOD mice progressed towards diabetes. Because they can rapidly transfer hyperglycaemia to NOD.scid recipient mice,21 we focused TCR expression analysis on Th40 cells. Mice were analysed at different developmental stages of diabetes described earlier. Vα2, Vα3.2, Vα8, Vα8.3 and Vα11, each were detected in young NOD mice (Fig. 4a). Vα11‐expressing cells were present only in very small numbers, however. TCR Vα3.2 and Vα8.3 were significantly (P = 0·0379) elevated in young mice. We performed a protein blast to discover if there was homology between murine Vα3.2 and human TCR sequences. Murine Vα3.2 and human Vα8‐3*02 have 78% identity, with 86% amino acid positive matches and 0/76 gaps (see Supplementary material, Fig. S2). Multiple Vβ molecules including Vβ4, Vβ5, Vβ8 and Vβ14 (Fig. 4b) were significantly (P = 0·0003) expanded in Th40 cells from young, pre‐insulitis NOD mice.

Figure 4.

T‐cell receptor (TCR) usage in Th40 cells as NOD mice develop diabetes: TCR usage from pancreatic lymph nodes was examined by flow cytometry. Th40 cells were characterized by antibody staining using the available Vα and Vβ antibodies. Pancreatic lymph nodes were taken from NOD mice at 4 weeks of age and examined for TCR Vα (a) and TCR Vβ (b) usage patterns, at 9–12 weeks, with moderate insulitis representative of pre‐type 1 diabetes for TCR Vα (c) and TCR Vβ (d) usage patterns; and at diabetes onset TCR Vα (e) and TCR Vβ (f) usage patterns. Data represent at least four mice at each stage. One‐way analysis of variance was performed and significant differences, P < 0·001, are indicated by asterisks.

Once NOD mice had progressed to pre‐diabetes, Th40 cells from PLN demonstrated additional alteration in the Vα and Vβ repertoires. Vα2⁺ cells within the Th40 population became expanded. Vα3.2⁺ cells increased further from pre‐insulitis stage and Vα8.3⁺ cells were reduced when compared with the pre‐insulitis stage (Fig. 4c). At pre‐diabetes stage, Vβ3⁺ and Vβ12⁺ Th40 cells were significantly (P < 0·003) increased (Fig. 4c). Vβ4⁺ and Vβ5⁺ Th40 cell percentages remained at the same percentages as in the younger mice. Vβ8⁺ cells became substantially reduced and Vβ14⁺ cells were somewhat reduced in percentage (Fig. 4b). There was no significant alteration in the other Vβ ⁺ cells examined.

TCR usage in Th40 cells from diabetic lymph nodes demonstrated alterations from pre‐diabetic mice. Vα2⁺ cell percentages were unchanged from pre‐diabetic values, although Vα2⁺ cells were now significantly lower than Vα8 or Vα11 numbers (Fig. 4e). Vα3.2⁺ Th40 cell percentages increased further in diabetic PLN, comprising a major portion of the Th40 cell subset. Vα8.3⁺ cells were significantly greater than Vα8 and Vα11 numbers, but when compared with pre‐diabetic mice, had substantially decreased (Fig. 4e compared with 4a). Vβ4, Vβ5, Vβ8, Vβ12 and Vβ14 were significantly (P = 0·0017) elevated (Fig. 4f). Vβ3⁺ and Vβ4⁺ percentages decreased compared with the pre‐diabetic stage whereas the Vβ5⁺ percentage increased (Fig. 4f). Vβ12 and Vβ14 were relatively unchanged from pre‐diabetic percentages and remained statistically increased relative to other Vβ percentages (P < 0·0001).

Th40 cells comprise the majority of cells in the pancreas of diabetic NOD mice

The majority of CD4⁺ cells isolated from pancreata of diabetic mice were Th40 cells (Fig. 5a) that were predominantly CD69⁺ (Fig. 5b). Analysis of TCR usage in CD69⁺ Th40 cells resident in the pancreas revealed significant percentages of Vα3.2⁺ and Vα8.3⁺ cells (Fig. 5c). However, the other Vα molecules were represented albeit at much lower percentages. The predominant Vβ molecules represented on Th40 cells were Vβ3, Vβ5, Vβ12 and Vβ14 (Fig. 5d). We compared TCR usage on Th40 cells from NOD.CD154^−/− mice that do not develop insulitis to diabetic NOD Th40 cells. Th40 cells in NOD.CD154^−/− mice do not expand to numbers seen in NOD mice (see Supplementary material, Fig. S1). We focused on TCR constituents that expanded in NOD mice during diabetes. Most of the Vα and Vβ molecules that significantly expanded during diabetes occurred at very low numbers in NOD.CD154^−/− Th40 cells (Fig. 6). The notable difference was TCR Vβ14, which was statistically as elevated in NOD.CD154^−/− as in diabetic NOD Th40 cells. This suggests that Vβ14 does not promote diabetes given that NOD.CD154^−/− do not develop diabetes. Vα2, Vβ3 and Vβ9 were statistically similar between diabetic NOD and NOD.CD154^−/− Th40 cells, suggesting that those TCR‐bearing cells do not participate in diabetogenesis or insulitis.

Figure 5.

T‐cell receptor (TCR) usage in activated Th40 cells isolated from pancreata of diabetic mice. Pancreata were collected from at least four diabetic NOD mice. Tissue was digested with collagenase and lymphocytes were purified as described in the Materials and methods section. (a) Cells were stained for CD4 versus CD40 (Th40 cells); and for (b) CD69. Black dots are isotype control, overlay in the dot plot grey dots are CD4 and Th40 cells. CD69 is above isotype control. For TCR (c) Vα and (d) Vβ molecules. Vα and Vβ molecules are from within the defined CD4⁺ CD40⁺ CD69⁺ population. One‐way analysis of variance was performed and significant differences, P < 0·001, are indicated by asterisks. [Colour figure can be viewed at wileyonlinelibrary.com]

Figure 6.

Comparison of T‐cell receptor (TCR) usage between diabetic NOD mice and age‐matched NOD.CD154^−/− mice. TCR usage was compared using T cells isolated from spleens. There were very few cells present in pancreatic lymph nodes, consistent with CD154^−/− mice not developing diabetes. Only the TCR species that were associated with diabetes development in NOD mice were examined. Direct comparison is represented. Data represent at least four mice from each group. One‐way analysis of variance was performed and significant differences, P < 0·0001, are indicated by asterisks.

Th40 and cytokine production

CD40 engaged as an independent co‐stimulatory molecule induced significantly higher levels of IFN‐γ (P = 0·034 by analysis of variance, comparing both treatment conditions with isotype, Fig. 7) and TNF‐α (P = 0·0016, Fig. 7) relative to CD28. CD40 induced IL‐6, but was statistically no better than CD28 (Fig. 7), although both were significantly above isotype. CD40 was more effective than CD28 co‐stimulation for IL‐17 induction (P = 0·0034, Fig. 7) and IL‐22 (P = 0·024, Fig. 7). CD28 was more effective at inducing IL‐21 (P = 0·0021, Fig. 7) yet CD40 induced low‐level production. Both IL‐22 and IL‐21 have pro‐inflammatory properties, but the pleiotropic effects of IL‐21 include predominantly regulatory properties.54 CD40 and CD28 co‐stimulation independently were equivalent at inducing IL‐2 (Fig. 7). CD28 co‐stimulation was better at inducing IL‐10 (P = 0·0012, Fig. 7) and IL‐4 (P = 0·021, Fig 7). For IL‐10, CD40 stimulation was not different from isotype.

Figure 7.

Cytokine production in Th40 cells. Purified Th40 cells from spleens of 9‐ to 12‐week‐old female NOD mice were treated with anti‐CD3⁺ anti‐CD28 (classical T‐cell co‐stimulus), anti‐CD3⁺ anti‐CD40, or isotype controls. Cytokine production was measured after 24 hr. One‐way analysis of variance was performed and significant differences are indicated.

Discussion

Th40 cells become pathogenic during diabetes,21, 22, 23 transferring rapid insulitis and hyperglycaemia, the hallmarks of type 1 diabetes. Most NOD mice develop spontaneous onset diabetes #bib80% of female mice and 50% of male mice, by 20 weeks of age. Only 20% of the first functional TCR transgenic mice developed using the BDC2.5 TCR sequence develop diabetes by 22 weeks of age. However, 100% of BDC2.5 TCR transgenic mice have become diabetic by 45 weeks of age.9 Interestingly, Th40 cell numbers in BDC2.5 mice increase to numbers greater than in NOD mice. The delay in diabetes onset in BDC2.5 mice was because of high FoxP3 levels within the Th40 cells until late in the mouse's development. Th40 cells isolated from BDC2.5 TCR transgenic mice do transfer diabetes to NOD.scid recipients, but with slower kinetics than BDC2.5 T‐cell clones.9 The issue being that Th40 cells that developed naturally in BDC2.5 TCR transgenic mice express FoxP3, but lose FoxP3 expression over time.9 T cells from BDC2.5.CD40^−/− retained FoxP3 expression, even after adoptive transfer and do not transfer diabetes. When taken from any developmental stage BDC2.5 Th40‐FoxP3‐negative cells rapidly transferred insulitis and hyperglycaemia.9 The current work demonstrates that not only is CD40 expression required for diabetes onset, it is also required for insulitis. We derived BDC2.5.TCR.Tg‐CD40^−/− mice and re‐derived NOD.CD154^−/− mice, neither of which developed insulitis. NOD.CD154^−/− mice were generated previously and shown to play a crucial role in diabetes development.50 Interestingly, a full dose of CD154 is required for diabetes; NOD.CD154^+/− mice did not develop diabetes either.50 In a NOD, CD8⁺ TCR‐transgenic mouse that was deficient in CD154, diabetes developed equally to in CD8⁺ TCR‐transgenic mice that were CD154 competent.55 Our data, in addition to these previous reports, suggest that CD40 expression rather than CD154 expression is the molecular switch for lymphocyte trafficking. An interesting caveat is that pancreatic islet β cells express CD40,29, 56 which was described as inducing inflammatory cytokines and chemokines.

Islets are composed of five cell types. β cells produce insulin and comprise about 70% of the islet with α, δ, γ and ε cells comprising the remaining 30% of the islet. α cells (20% of islet) produce glucagon, δ cells produce somatostatin‐D, γ cells produce pancreatic polypeptide and ε cells produce ghrelin, the hunger hormone.40 The first cells to infiltrate islets are innate immune cells including macrophages and DC.41, 42 Adaptive immune cells including T cells, both CD4 and CD8, as well as B cells soon follow. Our previous adoptive transfer data suggest that Th40 cells are among the first cells of the adaptive response to infiltrate the islets,21, 22, 23 but why do any cells traffic to the islets? An important issue during diabetogenesis is that while the microenvironment of the entire islet experiences immune‐driven inflammation, insulin but no other islet‐generated protein is affected. Glucagon (α cells) and insulin secretion are asynchronous, but somatostatin (δ cells) and insulin secretion are synchronous.46 Glucagon and somatostatin production/secretion are not affected by insulitis. One difference is that β cells are the only cell type in the islet that expresses CD40.29, 56 One of the many CD40 functions is production of cytokines and chemokines; thus β cell CD40 engagement may recruit Th40 or other cell types to the islet. Th40 cells produce IFN‐γ and other inflammatory cytokines that induce further damage, so increasing the auto‐antigen pool. Because Th40 cells carry an expanded repertoire, even early in diabetogenesis as demonstrated here, they may be more likely to recognize an essential β cell antigen leading to β cell attacks and recruitment of additional cells. These interactions create a spiral of inflammation leading to further auto‐antigen generation.

Several auto‐antigens have been identified in T1D, including insulin peptides and glutamic acid decarboxylase (GAD) peptides among others.57 T cells that respond to these antigens have also been identified. In autoimmune diseases T‐cell clonal expansions are considered crucial drivers of disease development. TCR usage patterns have been examined loosely relative to T1D, but identifying the overall TCR repertoire will be more informative. For example, does a small cadre of clonally expanded T cells, i.e. with a very limited TCR repertoire, drive disease or does a wider ranging repertoire of T cells responding to a more diverse antigen base more influence disease? Further, when do any of these T cells arise relative to disease stage? TCR usage within the Th40 cell subset altered as NOD mice progressed towards diabetes; from no insulitis to moderate insulitis that equates to pre‐T1D, then to extensive insulitis followed by disease onset. Of note was the consistency of the appearance of Vα3.2⁺ cells. That TCR constituent was significantly elevated on Th40 cells even before insulitis had occurred and continued to expand in number and as a percentage of Th40 population in PLN throughout diabetogenesis. Vα3.2⁺ cells transfer diabetes to NOD.scid recipient mice and Vα2⁺ cells do not.21 The murine Vα3.2 sequence was highly homologous to the human Vα8‐3*02 TCR sequence. When PLN from human patients with diabetes were analysed for T‐cell clonality, the Vα8‐3*02 sequence was prominent.46 In young mice, before insulitis Vβ4, Vβ5, Vβ8 and Vβ14 were prominent. Vβ4 is associated with BDC2.5 TCR.58 Vβ5 has been reported to be prominent during diabetes.59 Over time, e.g. during pre‐T1D, Vβ12 became more prominent and Vβ12 has been reported to be involved in diabetes development.59 In NOD mice, Vβ14 remained prominent, but NOD.CD154^−/− Th40 cells also demonstrated elevated Vβ14⁺ cells. Because NOD.CD154^−/− do not develop insulitis and Th40 cells from those mice cannot transfer diabetes, Vβ14 expansion by default must be related to antigens other than those involved in diabetes development. In the CD154^−/− mice the other TCR constituents: Vα3.2, Vα8.3, Vβ4, Vβ5 and Vβ12 failed to expand, suggesting that CD40 plays a prominent role in development/expansion of these TCR molecules. Although NOD.CD154^−/− mice developed a T‐cell repertoire, it skewed significantly from wild‐type NOD, and those mice did not experience insulitis or diabetes. There arguably could be oligoclonal expansion independently of CD154 expression, based on heightened expression of some TCR constituents, Vβ14 for example, but those clones clearly are not diabetogenic.

Studies in NOR mice provide an interesting perspective on diabetic tolerance. NOR mice carry 85% of NOD genetics, including the unique MHC class II, I‐Ag7,48 but tolerance remains intact. NOR mice develop, but do not expand, Th40 cells.21 One possibility for this involves activation‐induced cell death (AICD). T cells from NOR mice are more resistant to AICD than T cells from NOD.60 We showed that Th40 cells from NOD are defective in both Fas‐mediated AICD and in CTLA‐4‐mediated self‐regulation.11 Data here show that when transferred to a tolerance‐breaking environment, NOD.scid mice, Th40 cells drive diabetes development. Tolerance mechanisms, exogenous to T cells, must be lacking in the NOD background.

In addition to influencing cell trafficking and TCR usage as NOD mice develop T1D, CD40 acts as a T‐cell co‐stimulus independently of CD28, one of the first described T‐cell co‐stimulatory molecules.61 Interestingly, CD40 co‐stimulus favours pro‐inflammatory T‐cell outcomes, whereas CD28 co‐stimulation favours more regulatory outcomes. CD28 mice develop severe autoimmune disease, suggesting that CD28 plays a role in regulatory T cell development.62, 63 CD40^−/− mice on the NOD background, or as we showed here, on the BDC2.5.TCR transgenic background, do not develop autoimmunity. Part of the regulatory versus inflammatory mechanisms would include the source of ligands. CD28 interacts with CD80 and CD86, expressed on antigen‐presenting cells.64 CD40 interacts with CD154 expressed predominantly on activated T cells but also on platelets.65

These studies demonstrate that CD40 interaction with CD154 influences multiple phases during diabetes development. The earliest stage of disease development, insulitis, is dependent upon CD40‐mediated signals. The following stages, T‐cell interaction with auto‐antigens likewise is influenced by CD40 mediated signals, relative to TCR usage patterns. Finally, CD40 influences Th40 cells to a pro‐inflammatory phenotype, whereas CD28 influences a more regulatory phenotype. Targeting CD40 therefore would impact diabetogenesis at multiple stages, demonstrating how CD40 functions as a central nexus in autoimmune inflammation.

Conclusions

• CD40 expression is required not only for hyperglycaemia, the hallmark symptom in T1D, but also is required for cell trafficking leading to insulitis.

• TCR expression during diabetogenesis is CD40 driven. Individual TCR species become more prominent at different stages of diabetogenesis.

• TCR species Vα3.2, which has very high homology to a human TCR, Vα8‐3*02, was strongly represented in PLN from patients with T1D.

• CD40 acts as a strong pro‐inflammatory co‐stimulus in comparison with CD28, which acts as a regulatory co‐stimulus during diabetogenesis.

Disclosures

DHW was recipient of grants from the American Diabetes Association grant number 7‐13‐TS‐30 and the National Institutes of Health grant number 5R01DK075013 and R21AI096468 supporting this work. The funding agencies had no involvement in study design, collection, analysis and interpretation of data, or decision to submit. DHW is Chief Scientific Officer of Op‐T‐Mune, Inc. MGY is Chief Medical Officer of Op‐T‐Mune, Inc. Op‐T‐Mune had no involvement in study design, collection, analysis and interpretation of data; Op‐T‐Mune provided no financial assistance.

Supporting information

Figure S1. Th40 cells in NOD.CD154^−/− mice.

Figure S2. Direct comparison of murine Vα3.2 sequence and human Vα8.3 sequence.