TCR Bias of In Vivo Expanded T Cells in Pancreatic Islets and Spleen at the Onset in Human Type 1 Diabetes Free

Type 1 diabetes (T1D) is an autoimmune disease defined by the selective destruction of insulin-producing pancreatic β cells, mediated by CD4⁺ and CD8⁺ autoreactive T cells (1, 2). Antigenic specificities of intrapancreatic T cells have been mainly defined in the NOD mouse by studies showing that the initial challenge is driven by insulin-specific T cell responses (3, 4), later spreading to GAD and other islet-related autoantigens (5, 6). In contrast, the T cell targets of human T1D have been identified only from the analysis of peripheral T cells, mostly because of the limited availability of human pancreatic tissue. Autoreactive T cells were identified in vitro by expansion against insulin, GAD, I-A2, IGRP (7–10), and ex vivo by positive staining with self-peptide–MHC tetramers (11, 12). Although there are no data on the specificity of intrapancreatic T cells in human T1D, phenotypic analysis of infiltrating lymphocytes have shown that the human pancreas is infiltrated with both CD8⁺ and CD4⁺ T cells (13–18). In addition, several analyses of the TCR repertoire of pancreatic T cells demonstrate the dominance of some Vβ families, despite the lack of consensus imposed by the interindividual variability of MHC (15, 16, 19, 20). Furthermore, there are no reports of monoclonal TCR expansions in human islets. The hypothesis that, in human T1D, few cells were expanded in the target tissue was demonstrated using pancreatic biopsies. The authors found increased TCR α-chain variable (TRAV) transcripts corresponding to a few dominant CDR3 sequences from one or two Vα families in each patient analyzed (19). An exhaustive study comparing TCR β-chain variable (TRBV) usage of in vitro expanded T cells from a pancreas allograft and PBMCs of two patients after recurrent T1D (20) showed oligoclonality, a restricted TCR usage in both patients and, as expected, interindividual differences in junctional diversity (TRBJ). Similar results were obtained from T cells expanded from pancreatic lymph nodes (PLNs) (21). In this study, the isolation of an insulin-specific T cell clone confirmed the involvement of this autoantigen in human T1D.

As demonstrated in the NOD mouse, the activation of autoreactive T cells occurs in the PLNs, from where effector T cells migrate to the bloodstream and reach the pancreas (22). In this model, the spleen can harbor autoreactive T cells, because splenocytes from a diabetic animal can transfer the disease to healthy recipients (23). Moreover, the circulation of diabetogenic T cells through the spleen can take place before or after reaching the target tissue (23–25). Although published studies demonstrate that the spleen can store central memory CD8⁺ T cells and contribute to the expansion of memory effector T cells in a secondary infection, there are no data describing the expansion of autoreactive T cells in the spleen (26). In the earlier mentioned human T1D study, an insulin-specific, PLN-isolated T cell clone was not found in the autologous spleen (21), suggesting that no expansion or retention of insulin-specific T cells was taking place. In contrast, autoreactive B cells from mouse and human diabetic pancreas infiltrates have been identified in the spleen of NOD mice and blood of diabetic patients, respectively (27, 28). B cells have been shown to be efficient autoantigen-specific APCs in the PLNs and the spleen of NOD mice, as they express high levels of costimulatory molecules (29, 30). Thus, NOD mouse autoreactive B cells from the spleen could activate and expand autoreactive T cell clones in situ, but no data are available from human T1D.

We have studied the TCR of intrapancreatic T cells from a diabetic donor at disease onset. In a previous study (16), the usage of TRBV families of the lymphocytes associated with the pancreatic tissue was described. The diversity of infiltrating T cells was analyzed, and the predominance of some Vβ families, particularly Vβ1, was observed. The spleen was also included in this study as a control, showing no dominance of any Vβ family. Based on these results, we have now performed an exhaustive study of the pancreatic T cells at the clonal level, to identify dominant clones and to compare their distribution in the target organ and the periphery, that is, the spleen and peripheral blood. To our knowledge, this is the first analysis of in situ expanded T cells from human T1D. The data showed a diverse TCR repertoire of the islet-infiltrating T cells, with a few monoclonal expansions dominating over other less frequent T cells. Moreover, the analysis of a large number of CDR3 sequences from the pancreatic TCRs demonstrated a restricted amino acid usage, suggesting a bias that can be applied to the intraislet expanded TCRs. The expanded CDR3 sequences were then used to trace autoreactive T cells in the spleen and peripheral blood. Thus, one of the islet-associated monoclonal TCR expansions from Vβ22 was also found in the spleen. The same TCR was present in peripheral blood, indicating its homing through blood circulation.

This is the first study, to our knowledge, to describe the existence of monoclonal expansions in intraislet infiltrating T cells. Our data also suggest that the spleen may have an active role in retaining or maintaining some potential pancreas-reactive T cell clones in human T1D.

Pancreas, spleen, and a blood sample of a 19-y-old woman who died 5 d after the clinical onset of T1D were obtained with the family’s consent (16). Islet cell and insulin Abs were both positive. HLA typing was A*0201,*6801; B*3901,*4001; Cw*0304,*1203; DRB1*0404,*1301; DRw52,DR53, DQB1*0302,*0603, DQ2, DQ8. Tissue description and digestion protocol are detailed in Somoza et al. (16), which referred to the sample as “Case 1.” Samples used for this study were total digest (DM-TD); pancreatic islets, manually collected under a stereoscopic microscope with >90% purity (DM-ISL); three spleen samples obtained from individual frozen tissue blocks (DM-S1, DM-S2, and DM-S3); and PBMCs obtained by Ficoll-Hypaque centrifugation from a postmortem bleed (DM-PB).

The pancreases from two cadaver organ donors were collected and frozen. Islet cell Ab determinations were performed to discard possible prediabetic subjects. Control pancreas samples were named C-TD for total digest and C-ISL for purified islets (>90% purity). The spleen from an 18-y-old woman organ donor positive for DRB1*0405*0407; DQB1*0302 HLA class II typing was used as control (C-S). PBMCs were obtained from two healthy blood donors (C-PB1 and C-PB2). All protocols were approved by the Ethics Committee of the Germans Trias i Pujol University Hospital and followed the principles included in the Declaration of Helsinki.

Total mRNA was extracted with RNeasy Plus Mini Kit (Qiagen) and reverse transcripted using an oligo-dT primer. As a control of mRNA integrity, the GAPDH gene was amplified from all samples. Multiplex PCR was adapted from Chitnis and Pahwa (31). Amplification mixtures (MIX A to L), including 24 Vβ gene families, were carried out in 15-μl reactions containing 1.2 μl forward Vβ primer MIX and 0.75 μl reverse Cβ primer at 5 μM, 10× for each 2′-deoxynucleoside 5′-triphosphate (2.5 mM), 10× PCR buffer with 2 mM MgCl₂, 40 ng cDNA, and 0.6 unit Taq polymerase (Biotools, Madrid, Spain). PCR amplification was 3 min at 97°C, 35 cycles of 95°C for 30 s, 55°C for 30 s, and 72°C for 1 min, and 10 min at 72°C. Primer mixes were MIX-A (Vβ1/18/23), MIX-B (Vβ2/4/8), MIX-C (Vβ3/13.1), MIX-D (Vβ5.1/11), MIX-G (Vβ9/16), and MIX-H (Vβ5.2/12), 2.5 μM for each primer; MIX-E (Vβ6/24), MIX-F (Vβ7/21), MIX-I (Vβ13.2/15), and MIX-J (Vβ17/14), where primers were at 1.7 and 3.3 μM, respectively, and MIX-K (Vβ20) and MIX-L (Vβ22) at 2.5 μM. For each multiplex PCR, a negative control was run in parallel to exclude possible contaminations.

Multiplex PCR for TRAV was designed in our laboratory using described primer panels (32). Primer mixes of 28 Vα families were MIX-I (Vα1/12/18), MIX-III (Vα2/17/23), MIX-VIII (Vα5/7/10) and MIX-XI (Vα14) at 1.7 μM for each primer; MIX-II (Vα8/24), MIX-VII (Vα3/4), and MIX-X (Vα13/27) at 2.5 μM; MIX-XII (Vα6/25) at 3.75 μM; MIX-IV (Vα9/20/26) at 1.25, 1.25, and 2.5 μM; MIX-V (Vα21/22) at 4 and 3 μM; MIX-VI (Vα15/16) at 2 and 3μM; and MIX-IX (Vα9/11/28) at 1.9, 1.9, and 3.75 μM, respectively. PCRs were performed using 2.25 μl forward Vα primer MIX and 1.5 μl reverse Cα primer (5 μM) at the PCR conditions: 97°C for 3 min, 35 cycles at 94°C for 1 min, 60°C for 1 min and 72°C for 1 min, and 5 min at 72°C.

Runoff extension was performed using multiplex PCR products as the template and the reverse primer Cβ 5′ labeled with FAM, consisting of 3× 2 min cycles at 95°C, 2 min at 55°C, and 20 min at 72°C. Fluorescent PCR products were run with a GS400HD size marker (Applied Biosystems) and were length separated by the ABI 3130XL analyzer (Applied Biosystems). Data were analyzed with the PeakScanner software (Applied Biosystems). Peak areas were quantified and normalized with the size marker to determine the total normalized area corresponding to each Vβ family. To assess the presence of monoclonal T cell expansions, the relative fluorescence intensity (RI) was calculated: RI (%) = (peak area/total Vβ area) (33, 34). A CDR3 peak with an RI > 50% was considered a monoclonal expansion. The same procedure was applied for Vα families. This method is summarized in Supplemental Fig. 1.

First-round multiplex PCR products were amplified with identical forward Vβ and reverse Cβ PCR primers with the polymerase Phusion (New England Biolabs) to blunt both 5′ and 3′ ends, and to facilitate intramolecular ligation. PCR amplicons were cloned into the PBE vector (pBSKII variant; kindly provided by Dr. J. Piñol). After transformation, plasmid DNA was purified using Wizard Plus SV Minipreps kit (Promega) and sequenced with the PT7 primer by BigDye kit and the genetic analyzer ABI 3130XL (Applied Biosystems). Sequences for Vβ, Jβ, and CDR3 were assigned with IMGT-V-Quest software (ImMunoGeneTics).

CDR3 amino acid preferences were analyzed by alignment of CDR3 sequences of identical size (11–14 aa) and comparison with expected frequencies in the human proteome. A binomial statistical test was applied for each amino acid found at a given sequence position. The cumulative binomial distribution with parameters n (number of sequences in the alignment) and p (probability of amino acid X in the proteome) was evaluated to obtain the probability that X is found in position L in m − 1 or less of the n sequences, P(X ≤ m − 1), with m being the actual number of sequences that display amino acid X in position L in the alignment and the positions in the sequence being assumed to be independent. Amino acid X was then considered to be significantly overrepresented in position L if 1 − P(X ≤ m − 1) < 0.05; that is, if the expectation to find this amino acid m times or more in the same position in n sequences, given its probability in the proteome, was <5%. The analysis was conducted by taking two different approaches: considering all sequences as equiprobable, or weighing each sequence by the number of times it was found in the experimental set (Supplemental Fig. 5A–D).

The TRBV repertoire of the infiltrating T lymphocytes from the T1D pancreas was analyzed in two types of samples belonging to the same area of the pancreas. The DM-TD was the total digest from a given tissue sample, containing both the endocrine and the exocrine parts of the tissue, whereas the second sample (DM-ISL) consisted of the islets of Langerhans, manually isolated under a stereoscopic microscope, from the TD sample. Control samples C-TD and C-ISL were obtained from two different donors. The protocol used is described in Supplemental Fig. 1. In brief, a multiplex PCR was performed for each sample, and the normalized spectratyping data for each amplified TRBV family were analyzed. The amplification of almost all TRBV families (Fig. 1A, Supplemental Fig. 2A) suggested that, at disease onset, the T cell infiltration of the pancreas and, especially, of the islets was very diverse. Besides, the Gaussian distribution pattern of the CDR3 sequence length displayed by most TRBV families proved their polyclonality, although some of them showed high peaks suggestive of possible expansions (Fig. 1A). In contrast, control C-TD and C-ISL samples showed scarce infiltration and low TRBV diversity (Fig. 1B). As expected, T cell content, based on the CD3γ/GAPDH ratio, was greater in diabetic samples than in controls, as determined by comparative quantitative PCR (data not shown). Thus, at disease onset, an active and complex T cell response was taking place inside the islets of a human diabetes patient.

Vβ monoclonal expansions were defined by calculating the RI of each spectratyping peak for the DM-TD and DM-ISL samples. The RI was calculated for all Vβ families to generate a matrix (see Supplemental Fig. 3). Vβ families with a total area lower than 10 were excluded from the analysis. Peaks with RI > 50% of the total area were considered monoclonal expansions. Following these criteria, six monoclonal expansions were detected in the DM-ISL sample, corresponding to Vβ1(Peak-2), Vβ7(+1), Vβ11(−1), Vβ17(0), Vβ18(−1), and Vβ22(+2) (Fig. 2). The peaks were numbered according to their CDR3 length, compared with the major peak of each family in healthy PBMCs, which was given a value of zero. Sample DM-TD contained four of the same expansions, that is, Vβ1(Peak-2), Vβ7(+1), Vβ17(0), and Vβ22(+2), whereas the remaining two, Vβ11(−1) and Vβ18(−1), did not fulfill the criteria (Fig. 2). Some monoclonal peaks were observed in control samples, but total area values for control Vβ families were very low, consistent with a scarce T cell presence (Supplemental Fig. 3). In summary, the spectratyping analysis showed six monoclonal expansions in the T1D samples, despite a large and diverse intraislet T cell infiltrate.

To study the possible participation of the periphery in the maintenance of the autoreactive T cell response, we studied the T cell repertoire from the T1D donor’s spleen by analyzing three separate tissue blocks (DM-S1, DM-S2, and DM-S3) from separate areas of the organ. The spleen T cell repertoire was polyclonal for most TRBV families, similar to that of an HLA-DR4 healthy spleen, except for three that contained monoclonal expansions: Vβ1(−1), Vβ16(0), and Vβ22(+2) (Fig. 3A). RI values for peaks Vβ16(0) and Vβ22(+2) were >50% in the three spleen samples analyzed, indicating that they corresponded to predominant T cell responses throughout the spleen, as shown in Fig. 3B. A connection between the pancreas and the spleen was established when comparing the Vβ repertoire of both organs: Vβ22(+2) TCR was also expanded in the pancreas, and the Vβ16(0) peak had a 45% RI in the islets, almost reaching the criteria of monoclonal expansion (see Supplemental Fig. 3). Interestingly, spleen monoclonal expansions were very large, skewing the CDR3 distribution of the Vβ16 and Vβ22 families with RI values >80% of the total area, whereas they represented only 45 and 60%, respectively, in the islet sample. These peaks did not increase in the control sample (Fig. 3B).

To further confirm this connection, we analyzed the Vα (TRAV) family repertoire (Fig. 4A, Supplemental Fig. 4). Six TRAV monoclonal expansions—Vα6(−3), Vα7(−1), Vα12(1), Vα14(−1), Vα20(−1), and Vα27(−2)—were found in the intraislet lymphocytes (detailed in Supplemental Fig. 4), and the two spleen samples (DM-S2 and DM-S3) contained two monoclonal expansions: Vα6(−3) with RI = 75%, present in both samples; and Vα16(−1) with RI = 55%, detected only in DM-S2 (Fig. 4B). Thus, the expanded peak Vα6(−3) was present in both spleen and intraislet lymphocytes, confirming their relationship (Fig. 4B).

Sample clonal identity was confirmed by sequencing the CDR3 region from all TRBV and TRAV expansions (Table I). The high frequency of those sequences, within the total sequenced clones, confirmed that they were dominant in their respective families (all sequences obtained within cloned families are shown in Supplemental Tables I–III for TRBV and Supplemental Tables IV and V for TRAV). Only Vβ18(−1) and Vα16(−1) were not confirmed as monoclonal expansions because no dominant CDR3 sequence was demonstrated (Table I).

Despite a close anatomic location, the pancreas and spleen are connected only through blood circulation. To trace spleen and pancreas expansions in peripheral blood, we chose to analyze three representative Vβ families, that is, Vβ7 (expanded in islets), Vβ16 (in spleen), and Vβ22 (in both). The RI values for each Vβ family in the patient’s PBMCs were compared with the mean RI values from two control PBMCs (Fig. 5A). After sequencing each clone (Supplemental Table VI), no dominant sequence was found for Vβ7 or Vβ16 families, and none of the sequences from the PBMCs was found in the pancreas or spleen (see Supplemental Tables I–III). In contrast, the Vβ22(+2) peak was detectable in peripheral blood (Fig. 5A), where it increased, as compared with controls. Sequencing data confirmed the presence of the spleen- and pancreas-expanded Vβ22(+2) peak in blood. A second peak, Vβ22(+1), contained a sequence also present in the pancreatic infiltrate, where it corresponded to 11% of the total Vβ22 area (Figs. 2, 5B, Supplemental Tables I, II, VI).

If islet entry and accumulation of T cells is driven by Ag specificity as proposed in the NOD mouse model (25, 35), then the presence of T cell clones in the pancreatic infiltrate would be dependent on Ag recognition in situ. Therefore, the peptide–MHC complexes present in the tissue would determine the T cell clones to be expanded, resulting in a possible TCR restriction. When comparing the CDR3 amino acid composition of the expanded sequences shown in Table I, we observed some common residues in the hypervariable NDN region: glutamine (Q) was shared by the Vβ16(0) and Vβ22(+2) expansions defined in the spleen but was absent in the other pancreatic expansions, and proline (P) was found in two of four clones exclusive of the pancreas.

To confirm this apparent bias in the NDN region of the intraislet TCRs, we analyzed all sequences available for Vβ1, Vβ7, Vβ11, Vβ16, Vβ17, Vβ18, and Vβ22 expanded families, together with highly represented Vβ families Vβ4, Vβ6, Vβ8, and Vβ9, from islet (DM-ISL) and total digest (DM-TD) samples. The analysis of 139 different sequences from 401 clones (listed in Supplemental Tables I, II) showed a canonical distribution of Jβ usage, according to a recently published analysis of random TRBV sequences and CDR3 size, with a predominance of sequences of 11–14 aa (36) (Fig. 6). Sequences grouped by their CDR3 size were aligned (11–14 aa) to define any preferential amino acid usage within the NDN region. Statistical analysis showed that, in addition to the expected presence of G in every alignment (36), a CDR3 pattern of the most favored amino acids in each position could be defined for every group (p < 0.05; see Fig. 7 and Supplemental Fig. 5 for analysis details). Some of the significant amino acid preferences were indeed confirmed by the pancreas- and spleen-expanded T cell clones as follows (bold text represents an amino acid close to its position in the pattern; text is underlined if in the exact position): T in 11-aa sequences (CASSVSTTDTQYF for Vβ1 expansion); P, T, and N in 12-aa sequences (CASSDPGTQETQYF and CATSPLGMNNEQFF for Vβ11 and Vβ17, respectively); R, N, and Y in 13-aa sequences (CASSHRQMNYNEQFF for Vβ16) and Q, S, and T in 14-aa sequences (CASSEAQQGYSGELFF for Vβ22 and CASSQVAGAGTGELFF for Vβ7). CDR3 is a loop without secondary structure and its interaction to the peptide residues anchored in the MHC groove allow for a certain shift of the motives involved. In this context, the sequence of the expanded TCRs could fit with CDR3 patterns that differ in one amino acid. For example, amino acids in bold of sequences CASSDPGTQETQYF (Vβ11) and CATSPLGMNNEQFF (Vβ17), of 12-aa length, matched with the favored motives in the 11-aa CDR3 pattern (Fig. 7). A more thorough analysis including all sequences obtained, that is, each sequence was weighed by the number of times it had been observed, confirmed the patterns (see Supplemental Fig. 5). Altogether, these data define a restriction in the amino acid usage by the TCRs of the islet-infiltrating T cells that could be the result of selective expansions driven by a few islet-specific peptides. What is noteworthy is that the CDR3 sequence of spleen-expanded TCRs, Vβ22 and Vβ16, fit the corresponding pattern defined by the pancreas sequences. Although we cannot exclude a promiscuous TCR, a double-reacting TCR, or even a preferential retention of some clones in the spleen, the simplest explanation of these data are that the display of identical MHC–peptide complexes in the spleen can drive the amplification of the autoimmune response by certain clones in situ.

This study describes the TCR repertoire of cells infiltrating the pancreatic islets of a diabetic donor at disease onset. The data showed a diverse TCR repertoire of the islet-infiltrating T cells, where only a few monoclonal expansions were dominant, although other less frequent T cells could also be contributing to the islet inflammation or its regulation. Some of these pancreatic clonotypes were also identified in the patient’s spleen and peripheral blood by using their CDR3 sequence as a cell tracer.

The pancreatic islets were infiltrated by T cells expressing most TRBV families, five of which, Vβ1, Vβ7, Vβ11, Vβ17, and Vβ22, contained monoclonal expansions as demonstrated by spectratyping and CDR3 sequencing. A TCR comparison between human samples is not feasible because the repertoire is randomly formed in each individual and shaped by the individual’s MHC polymorphisms. Luppi et al. (37) had reported an increased frequency of circulating Vβ7, Vβ1, and Vβ17 T cells in PBMCs from recently diagnosed T1D patients, as compared with control subjects. In contrast, the frequency of Vβ families containing monoclonal expansions in PBMCs from the T1D patient studied was not different from healthy control subjects (data not shown). Similarly, an increase in Vβ7 expression in PBMCs from T1D patients was associated with a superantigen-mediated T cell expansion (38), but in our study case, this was ruled out because CDR3 sequencing revealed a single clonotype for the Vβ7 family. Interestingly, Vβ20 was not amplified in any of the T1D patient samples. In a previous study, a similar absence of Vβ20 was related to the presence of a disease-independent null allele for the Vβ20 gene (39).

We could trace some of the intraislet T cell expansions to the periphery by analyzing the TRBV repertoire from spleen and peripheral blood samples without any in vitro manipulation other than enzymatic digestion. Studies published so far have described the repertoire of human T cells isolated from patients’ PBMCs that have been expanded in vitro in front of cognate autoantigens (7–10), but this does not reflect their in vivo distribution or frequency. One such study reported the repertoire of T cells from the PLN of diabetic donors expanded in vitro with insulin, but none of the clones could be found in the spleen of the same donor (21). Ex vivo experiments published by Li et al. (40) using the NOD model revealed that the TCR of intrapancreatic T cells that were sorted against the mimotope of the diabetogenic clone BDC2.5 were not found in peripheral blood or were present at a very low frequency. In a different study in NOD mice, Sarukhan et al. (41) could not find any of the most abundant islet-specific T cell clones in the spleen. Thus, the expansion against a given autoantigen or the analysis of very abundant T cells may not be enough to reveal all relevant infiltrating T cell clones. In our case, expanded clones within the whole repertoire were searched for. Thus, we found that only one of five monoclonal expansions infiltrating the islets, Vβ22, was present in the spleen and detected in PBMCs. Similarly, only one monoclonal expansion expressing Vα6 was present in both spleen and intraislet T cells, from six islet-expanded TRAV families. Therefore, our results demonstrate that there is a low frequency of islet-specific T cells in the periphery of human T1D samples, so the peripheral repertoire is not representative of what is going on in the target organ. However, the presence of the same clones in the spleen and islets hints at a role for the spleen in the expansion of some autoreactive T cells.

The spleen is usually considered as a source of peripheral T cells in animal models. In our case, it appears that the autologous spleen not only stores some T cell clones from the pancreas, but appears to drive their expansion. How relevant these expansions are in the disease progression is unknown. The spleen and the tail of the pancreas drain into the pancreatic-splenic lymph nodes, which could favor a T cell response toward common autoantigens. Certain islet autoantigens must indeed be present in the spleen if putative islet-reactive T cells are expanded in this location. Islet autoantigens such as peripherin are expressed in the spleen (42) and anti-peripherin B cells have been isolated from the NOD spleen (43). We should take into account that one of the most diabetogenic clones from NOD, BDC2.5, generated from the spleen, recognizes a peptide from chromogranin A, a protein secreted both by β cells and by the spleen after nervous stimulation (44). For such autoantigens, the spleen could act as an amplifier of the autoreactive response. B cells are good candidate APCs in this scenario. Several groups have reported their role in disease progression and B cell deletion protects NOD mice from diabetes (45, 46). B cells are efficient presenters of peptides derived from the Ag recognized by the BCR, as proposed by GAD65 presentation in mouse and human T1D (47, 48). A recent report of the transcriptome of purified islets from the donor reported in this article showed a high level of IgG transcripts that correlated with the presence of infiltrating B cells measured by tissue immunostaining (18). Some autoreactive B cells could present epitopes from proteins shared between the pancreas and spleen and, thus, mediate the cross talk between the two tissues. To our knowledge, no other studies are available that describe a role for the spleen in human T1D.

In the NOD mouse, a restricted islet cell-reactive T cell repertoire has been proposed for early pancreatic-infiltrating T cells (49). Our analysis of 139 CDR3 sequences shows that infiltrating T cells have similar size distribution and Jβ usage as healthy PBMCs. Data were compared with a large study by Freeman et al. (36) describing CDR3 size distribution, sequence, and Jβ usage of 30,366 clonotypes from healthy donor PBMCs. However, the CDR3β sequences of the pancreas- and spleen-expanded clones showed a bias that could relate to the immunodominance of certain epitopes presented in the target organ. To facilitate alignment, we restricted the analysis by CDR3 size. A binomial statistics test was applied, resulting in four patterns (11–14 aa) that defined amino acid preferences in the NDN region. Conserved CDR3β features have been associated with Ag specificity among TCRs expressed by responding T cell clones (50). According to this, the defined CDR3 patterns, formed by the most abundant amino acids in certain positions, should be related to restricted specificities. These patterns were indeed shared by the CDR3 from the expanded T cells, presumably selected by peptide-specific recognition. In summary, our data defined a TCR bias in pancreas-infiltrating T cells, shared by the islet and the spleen-expanded T cell clones. Thus, even if the immune response against tissue-specific T cell epitopes is mainly sustained by the target organ and PLNs, the spleen may contribute to the maintenance of some T cell clones, thus participating in the perpetuation of the disease.

We thank Annabel Segura for technical support, Drs. Xavier Daura and Juan Cedano for contributions to the CDR3 alignment analysis, and the Genomic Sequencing Service (Institute of Biotechnology and Biomedicine Research B Unit, Autonomous University of Barcelona) for help in the spectratyping analysis. This manuscript has been proofread by Chuck Simmons, a native English-speaking University Instructor of English.

The authors have no financial conflicts of interest.