Introduction
Type 1 diabetes (T1D) is a T cell mediated disease in which both CD4 and CD8 lymphocytes infiltrate the islets of Langerhans, causing destruction of insulin-producing beta cells and consequently, hyperglycemia. Many characteristics of human T1D are shared with the spontaneous onset of disease in inbred Non Obese Diabetic (NOD) mice, which is commonly used as a model of human pathology. In NOD mice, T cell islet infiltration starts within 3-4 weeks of life, ultimately producing overt diabetes in 80% of female mice beyond 30 weeks of age. Interestingly, NOD.Igμ
null mice (which are B cell deficient) do not become diabetic [
1], but develop disease if reconstituted with B cells [
2]. B cell reconstitution performed early, at 4 weeks of age by a chimera approach (to bypass the MHC class I-mediated rejection), precipitates disease in 65% of the animals starting at 20 weeks of age.
Prior studies have indicated the role of B cells is to stimulate the auto-reactive T cell repertoire by providing enhanced antigen presentation and costimulatory capacities that compensate for natural defects in dendritic cells and macrophage antigen presenting cell populations in NOD mice [
3,
4]. It is known that to cause disease, the B cells are required to possess the I-A
g7 MHC class II molecule [
5] and that the specificity of the B cells is also important, as reconstitution of HEL-specific transgenic B cells in NOD.Igμ
null mice did not cause diabetes [
6]. B cell reconstitution has been shown to restore an autoimmune T cell response to GAD65, an autoantigen in diabetes, we and others have found to be important in disease etiology [
2,
7]. Importantly, NOD.Igμ
null mice have been shown to contain a functional autoimmune T cell repertoire (in the absence of B cells) capable of causing diabetes if transferred into NOD.
scid mice [
8].
CDR3 spectratyping or immunoscope analysis is a highly sensitive technique allowing a non-biased identification of the T cell receptor (TCR) repertoire ex-vivo in target organs, spleen and lymph nodes. Diversity in the TCR repertoire is the result of random combinations of V, D and J segments and nucleotide insertions during recombination. This process results in CDR3 lengths being generated that are between four and 14 amino acid residues long. If no T cell expansion is induced within a particular BV family, a Gaussian distribution of CDR3 length is observed, typical of background and polyclonal responses.
In this study, we performed TCR spectratype analysis of V beta (BV) gene expansions at the BV-C beta level on NOD.Igμnull mice in comparison to B cell-reconstituted NOD.Igμnull animals, at different time points post-reconstitution. This allowed us to identify the expanding TCR repertoire infiltrating the islets of NOD.Igμnull mice that are dependent on B cells. We observed that without B cell reconstitution, NOD.Igμnull mice had no pancreatic T cell expansion. No T cell receptor PCR product across the entire BV family repertoire was detected, despite Gaussian BV distributions (non-expanded T cells) being observed in pancreatic lymph nodes and splenocytes of these animals. However, upon B cell reconstitution, a progressive infiltration and increase in diversity of the T cell repertoire was detected in the pancreases, with most of the BV families present at pre-diabetic and diabetic stages. A similar expansion profile of the BV TCR repertoire was also observed in the pancreas of B cell-reconstituted animals treated with cyclophosphamide (CYP). CYP treatment produced accelerated diabetes onset, but no disease in age-matched unreconstituted NOD.Igμnull mice. These results demonstrate that B cells are required for the generation of a pathogenic repertoire of T cells infiltrating the pancreas that promote diabetes.
Discussion
Previous studies examining T cell responses and repertoire analysis involved in the autoimmune response of diabetes have produced conflicting results related to the identification of the pathogenic T cell repertoire. Some groups have described polyclonal T cell expansions arising very early in the pancreas being responsible for islet destruction, [
15,
16], while others have claimed that only particular clonal expansions are the driving force behind autoimmune responses in diabetes [
17,
18].
These variable findings likely reflect the different techniques employed to characterize T cell responses in the pancreas during the course of spontaneous disease. Here we have employed spectratyping analysis to detect T cell expansions
ex-vivo, in a non-biased attempt at examining the T cell responses in the pancreas following B cell reconstitution in NOD.Igμ
null mice. We found that by 9-10 weeks post-B cell reconstitution, the majority of the animals present pancreatic TCR expansions (at 13 weeks of life). Of note, these animals do not have clonotypic expansions in their pancreatic lymph nodes or spleens, suggesting that clonotypic TCR expansions in lymphoid organs are not involved in disease induction (Figure
2). The initial pancreatic T cell infiltration consisted of several clonotypes, including BV2, BV10, and BV12, clonotypes already described as reactive to insulin or GAD65 [
7,
14]. BV12 has been found to be enriched in islets of NOD mice when compared to thymus and spleens [
15,
19]. We also found a BV15 expansion that is a possible candidate for BDC-10.1, a chromogranin A-reactive BV15 T cell [
20], which had been previously characterized with a high diabetogenic capacity [
14]. As disease progressed, an even larger TCR repertoire infiltrating the organ was observed. This finding is consistent with spreading of the T cell response [
21]. Considering the ever-growing list of islet antigens described as being targets of autoimmune response in T1D this polyclonality is expected [
17,
22]. We found that during the pre-diabetic and diabetic stages, additional expansions in BV16, 17, 18, 19 and 20 families were commonly detected, but these were not predominant in early infiltrates (Figure
3A). It is possible some of these expansions could comprise already described pathogenic clones. A BV16 GAD65-reactive clone (11H11) has been found in the islets of pre-diabetic NOD, with the distinct promiscuous capacity of recognizing different GAD65 peptides using a single TCR [
23].
In attempts to find commonalities in clonotype expansions in different pathological states of islet infiltration in the B cell reconstituted NOD.Igμ
null model, we also examined the cyclophosphamide-accelerated diabetes and the rejection of pancreatic implants in diabetic NOD.Igμ
null B cell reconstituted mice. Following cyclophosphamide treatment, known to eliminate regulatory T cells from the pancreas [
11], we found the pancreas-infiltrating repertoire to be quite distinct from that of age-matched non-diabetic NOD.Igμ
null B cell reconstituted mice, demonstrating that some regulatory component is also promoted by B cell reconstitution. Interestingly, BV1, BV8 and BV11 T cell expansions were greatly reduced or lost, while a new set of BV4, BV5S2, BV9 and BV16-20 expansions arose, suggesting their role in pathogenicity. Furthermore, BV8S1 and BV8S2 are absent in the cyclophosphamide treated group (a treatment known to destroy regulatory T cells), but present in 50% of the B cell-reconstituted animals. BV8S1 has been previously described as a predominant clonotype infiltrating the islets of partially diabetes-resistant male NOD mice [
24] and, interestingly, is also present in the blood of T1D patients [
25]. This may indicate that some regulatory component may still be present, although ineffective, at final diseased stages post-B cell reconstitution.
In another approach to address the identity of the pathogenic repertoire, we examined the infiltrating T cells rejecting new pancreatic implants (Figure
5). Pancreatic tissue from neonatal NOD.
scids transplanted under the kidney capsule of diabetic B cell reconstituted NOD.Igμ
null mice were rejected very fast (within 4 days), with the peak of T cell infiltration occurring within 2 days after implantation. The spectratype profile of the BV repertoire from day 2 implants (Figure
5, black bars) was very similar to that seen in the diabetic pancreas (Figure
5, grey bars) but with over 70% of the implants presenting BV1, BV2, BV4, BV5S2, BV8S3, BV10 and BV15 clonotypes. Interestingly, BV1 and BV8 clonotypes were decreased by cyclophosphamide treatment (Figure
4, black bars), therefore, with potential regulatory function. These findings indicate that in B cell reconstituted NOD.Igμ
null mice, highly invasive clonotypes predominantly infiltrating transplants are composed of particularly high pathogenic effectors, as well as regulatory T cells.
The breaking of T cell tolerance and passage through "Checkpoint 1-End of Ignorance" [
26] by B cell reconstitution, may result owing to two different possibilities. First, homeostatic proliferation of pathogenic T cells following sublethal irradiation, could awaken autoimmune responses. Homeostatic proliferation in an immunodeficient host due to sublethal irradiation or in NOD.
scid recipients, follows a pattern of expansion that takes circa 6 weeks for complete reconstitution [
14]. This mechanism has been shown in the past to generate autoimmune responses [
8,
27]. Second, autoimmunity could be mediated by the expansion of a T cell repertoire remodeled by the presence of B cells, through their unique antigenic display and enhanced proinflammatory and costimulatory capacities. We argue that homeostatic proliferation seems less likely, as T cells from control animals (reconstituted with bone marrow following irradiation) also go through homeostatic proliferation but do not develop diabetes! B cells from NOD mice are known to produce strong inflammatory responses, when compared to other non-autoimmune strains [
28] and present higher levels of costimulatory molecules [
29]. Therefore, our data describing a B:T cell ratio of 4 in the pancreases support the second mechanism, and the role for B cells as an important antigen presenting cells in NOD.Igμ
null mice. It is likely that B cells help in the induction/activation of the autoreactive TCR repertoire.
During diabetes promoted after B cell reconstitution in NOD.Igμ
null mice, B cells encompass over 64% of the lymphocyte population infiltrating the pancreas, despite equal numbers of B and T cells in the other lymphoid organs analyzed (spleens and pancreatic lymph nodes). Interestingly, recent study on the cellularity composition of individual pancreatic islets in female and male NOD mice at different time points of disease evolution do not report a high accumulation of B cells in the pancreas when compared to lymphoid organs [
30], while another study identify comparable B:T cell ratios in the spleen for NOD animals [
5]. The accumulation of B cells in the pancreas of NOD.Igμ
null reconstituted mice could bypass the requirement for T cell lymph node recruitment and may help explain why clonotypic T cell expansions detected in the pancreas are not likewise present in adjacent pancreatic lymph nodes by spectratyping studies. The autoimmune responses may be preferentially localized to the pancreas, induced by larger numbers of antigen presenting B cells (B:T ratio of 4) which could be promoting the effector T cell repertoires unbalancing the regulatory clonotypes. The number of CD19
+ B cells circulating in blood post-reconstitution in NOD.Igμ
null mice varied from 1 to 13%, with no correlation of higher blood B cells and diabetes onset (data not shown). B cell accumulation in the pancreases but not in lymphoid organs, suggest that the direct activation of effector T cells in the target organ by B cells may be the crucial trigger for disease induction.
B cell accumulation in the pancreas appears to maintain CD4 and CD8 lymphocytes in an activated state (CD44
high Figure
1) and IL-6 secreted by the mononuclear pancreatic infiltrate could modulate T cell activity. IL-6 is known to alter phagolysosomal processing, enhancing presentation of cryptic antigenic determinants [
31] and to provide survival signal for T cells [
32]. Thus, reintroduction of B cells appear to provide an ideal environment for pathogenic T cell activation and survival.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AV, ND and PN performed NOD.Igμnull bone marrow and B cell chimera reconstitutions, blood glucose measurements and spectratype experiments. FACS and cytokine studies were performed by AV, HS and CG. CG, ES and TB conceived and designed experiments. CG and TB wrote the manuscript. Authors have read and approved the manuscript.