Simultaneous recognition of allogeneic MHC and cognate autoantigen by autoreactive T cells in transplant rejection

Abstract

The autoimmune condition is a primary obstacle to inducing tolerance in type 1 diabetes patients receiving allogeneic pancreas transplants. It is unknown how autoreactive T cells that recognize self MHC molecules contribute to MHC-disparate allograft rejection. In this report, we show the presence and accumulation of dual-reactive, i.e. auto- and allo-reactive, T cells in C3H islet allografts that were transplanted into autoimmune diabetic NOD mice. Using high-throughput sequencing, we discovered that T cells prevalent in allografts share identical T cell receptors (TCRs) with autoreactive T cells present in pancreatic islets. T cells expressing TCRs that are enriched in allograft lesions recognized C3H MHC molecules, and five of six cell lines expressing these TCRs were also reactive to NOD islet cells. These results reveal the presence of autoreactive T cells that mediate cross-reactive alloreactivity, and indicate a requirement for regulating such dual-reactive T cells in tissue replacement therapies given to autoimmune individuals.

Introduction

Whole pancreas or isolated pancreatic islet transplantation is an effective strategy to extricate type 1 diabetes patients from daily insulin injection. In particular, islet transplantation is clinically indicated for patients having hypoglycemia unawareness. Transplantation of islets or pancreas is associated with improvements in overall metabolic management as measured by glycosylated hemoglobin as well as by decreased frequency and severity of hypoglycemia (1). In addition, ameliorations in multiple diabetic complications including cardiovascular, renal, neurologic, and ocular disorders have been observed following islet transplantation (1). Despite these benefits, graft rejection mediated by T cells limits wider application of beta cell replacement therapies, and consequently a significant number of patients revert to exogenous insulin administration within 3–5 years due to immune-mediated transplant destruction (1–5). There is accumulating evidence that active autoimmunity against pancreatic islets is correlated with negative outcomes of pancreas and islet transplantation (4, 6). Over half of patients positive for at least one type 1 diabetes-associated autoantibody (i.e., insulin autoantibody, glutamic acid decarboxylase (GAD) antibody, and/or islet antigen-2 (IA-2) antibody) became insulin-dependent within one year post pancreas transplant, whereas the majority of those not producing autoantibodies retained sufficient graft function (4). In addition, islet recipients with T cells reactive to GAD or IA-2 had lower C-peptide levels compared with those without autoreactivity (6). These studies suggest that islet autoimmunity contributes to the rejection of islet and pancreas allografts. To support this notion, Pugliese and colleagues demonstrated that there was migration of autoantigen-specific T cells into islet allografts following T cell transfer into immunocompromised mice (7).

It is poorly understood how autoreactive T cells could contribute to rejection of islet allografts. In the majority of cases in the clinic, at least one MHC gene is shared between the donor and the recipient. Thus, autoreactive T cells restricted to shared MHC molecules may participate in the rejection via recognition of self antigens presented by the shared MHC in the islet allograft. Even when no MHC genes are shared, autoreactive T cells conceivably cause allograft rejection via self APCs presenting a cognate self antigen. These activated APCs may induce recruitment of T cells recognizing peptides derived from donor MHC or minor antigens, leading to the rejection of allografts despite the absence of shared MHC. Alternatively, one potential explanation for why MHC-disparate islet allografts are targeted and rapidly rejected by self MHC-restricted autoreactive T cells in autoimmune recipients (8–10) is the concept of “heterologous alloimmunity.” Heterologous alloimmunity refers to memory/effector phenotype T cells that are specific for one antigen presented by a self MHC molecule, yet also mediate productive immune responses against structurally unrelated peptides presented by non-self MHC (11–14). Specifically, the contribution of anti-viral memory/effector T cells to allograft rejection through heterologous alloimmunity has been extensively studied. Welsh and colleagues demonstrated the presence and expansion of cross-reactive T cells that targeted both allografts and viruses (15–17). Similarly, anti-viral memory led to T cell expansion and participation in rejection of skin transplants as well as resistance to tolerance induction (18). Recently, Fairchild and colleagues showed that pre-existing endogenous memory CD8 T cells mediate heart allograft rejection in a mouse model (19), confirming the relevance of MHC cross-reactive memory T cells in solid organ transplant rejection. Thus, these studies provide conceptual proof-of-principle that pre-existing memory/effector T cells that react to virus-derived peptides are able to cross-react with allografts and facilitate rejection; however, it is unknown whether and how autoreactive T cells contribute to rejection of transplanted allogeneic tissues.

We hypothesized that islet allografts in diabetic NOD mice would be uniquely enriched for autoreactive T cells that are cross-reactive with allogeneic MHC molecules via heterologous alloimmunity, and that these cross-reactive T cells would contribute to allograft rejection. To test this idea, we used high-throughput T cell receptor (TCR) sequencing to validate the presence of autoreactive T cells within rejected MHC-disparate islet allografts in NOD mice. We further evaluated heterologous reactivity (i.e., islet/allo dual-reactivity) of T cells that were enriched within the rejected islet allograft lesions in NOD mice. We demonstrate that autoreactive T cells are present and enriched in allograft lesions in autoimmune mice, and that these highly enriched TCRs show both alloreactive and autoreactive responses in vitro. These results suggest that universal donor beta cell replacement strategies in autoimmune recipients are vulnerable to rejection due to heterologous alloimmunity mediated by autoreactive memory/effector T cells. Thus, dual-reactive T cells represent a critical target for tolerance-inducing treatments for transplantation in individuals with autoimmune diseases.

Materials and Methods

Mice

NOD.Rag1^−/− (#003729), NOD.scid (#001303), NOD.Ca^−/− (#004444), C3H/HeJ (#000659), BALB/cJ (#000651), C57BL/6J (#000664) mice were purchased from Jackson Laboratory and bred in the Center for Comparative Medicine at the University of Colorado Anschutz Medical Campus. NOD/Bdc mice used as recipients for islet transplantation are the 99–116 generation of sibling breeding of the Barbara Davis Center subline. All animal experiments were approved by the Institutional Animal Care and Use Committee of the University of Colorado Denver.

Screening for diabetic recipients for islet transplantation

Blood glucose levels of female NOD/Bdc mice were monitored 2–3 times a week for diabetes onset. Mice between 10 and 26 weeks of age with at least three consecutive days of blood glucose readings >250 mg/dl became transplant candidates. Transplant recipients were grafted within 7 days of diabetes onset.

Islet isolation and transplantation

Islets were isolated from NOD.Rag1^−/− or C3H/HeJ mice by collagenase infusion through the bile duct followed by isolation of islets using Histopaque gradient centrifugation and further purification by hand-picking of islets as described previously (20, 21). Each transplant recipient received 450–500 islets transplanted underneath the kidney capsule. Following transplantation, blood glucose levels were monitored daily, and euglycemia (two consecutive days with a blood glucose reading <180 mg/dl) was established within 24–72 hours. Graft rejection was determined by recurrence of hyperglycemia (three consecutive days of a blood glucose reading >250 mg/dl).

TCR sequencing

Endogenous pancreatic islets as well as transplanted islet grafts were isolated for RNA extraction. Single-stranded cDNA was synthesized using the Clontech SMARTer™ RACE cDNA Amplification Kit (Clontech, Mountain View, CA) with oligo-dT primers according to the manufacturer’s instructions. To amplify TCR alpha and beta chains a two-step PCR was performed. Four to six PCR reactions were generated for individual samples using the Universal Primer A Mix supplied from the SMARTer RACE cDNA Amplification Kit, along with a primer to the constant region of alpha chain (GGGTGCTGTCCTGAGACCGAGGATC) and beta chain (AGCCCATGGAACTGCACTTGGCAGCG), respectively. The first PCR products were amplified with nested primers ligated with 454 adaptor sequences containing a multiple identifier sequence (CCTATCCCCTGTGTGCCTTGGCAGTCTCAGAAGCAGTGGTATCAACGCAGAGT along with CCATCTCATCCCTGCGTGTCTCCGACTCAG-multiple identifier sequence-GTACACAGCAGGTTCTGGGTTCTGG for alpha chain or CCATCTCATCCCTGCGTGTCTCCGACTCAG-multiple identifier sequence- CCTGGCCAAGCACACGAGGG for beta chain), agarose gel-purified, further purified with AMPure XP Beads (Beckman Coulter), subjected to emulsion PCR with 454 GSJR titanium chemistry, and sequenced on the 454 GSJR instrument. Sequence reads were processed using the 454 GS junior amplicon pipeline in the absence of the primer trimming filter, followed by analysis on the IMGT High V-quest (http://www.imgt.org/) to identify V(D)J sequences along with CDR3 junction sequences. TCR sequences which scored >1200 for the IMGT V-gene score (correlating with approximately 95% homology with reference sequences) were chosen for analyses, and mean frequencies of individual TCR alpha or beta clonotypes in the 4–6 PCR reactions were determined for individual samples.

Generation of retroviral vectors

TCR alpha or beta genes were cloned into the mouse stem cell virus (MSCV)-based retroviral vectors as described previously (22). The retroviral vector for TCR alpha chains contains the neomycin-resistant gene, and that for beta chains contains the puromycin-resistant gene. The mouse CD8 alpha and beta genes connected by the porcine teschovirus-1 2A (P2A) peptide gene were engineered into a MSCV-based retroviral vector carrying the hygromycin-resistant gene. The mouse H-2K^k or H-2D^k and beta 2-microglobulin genes connected by P2A were engineered into a MSCV-based retroviral vector carrying the puromycin-resistant gene.

Generation of transductant cells

Replication-incompetent retroviruses were produced as described previously (23). The 5KC hybridoma T cell line lacking endogenous TCR alpha and beta chain genes (19), which is derived from B10.BR CD4 T cells, were spin-infected with supernatant containing retroviruses carrying the mouse CD8 alpha and beta genes followed by sorting for CD8⁺/CD4⁻ cells on a MoFlo cell sorter (Beckman Coulter). Both the original CD4⁺ cells and the sorted CD8⁺/CD4⁻ cells were then transduced with retroviruses carrying various combinations of TCR alpha and beta chain genes, followed by culture in the presence of G418 and puromycin. M12C3 mouse B cells were spin-infected with replication-incompetent retroviruses carrying the H-2K^k or H-2D^k and beta 2-microglobulin genes.

T cell stimulation assays

T cell transductants (1 × 10⁵ cells) were cultured in a 96-well plate with various antigens in the presence or absence of APCs as described below for 16–20 hours, and IL-2 secretion in supernatants was measured by ELISA as described previously (22). Co-culture with anti-CD3 antibody (145-2C11, 5 µg/ml) was assessed for each T-cell transductant as a positive control. Spleen cells from various mouse strains (4 × 10⁴ cells unless stated otherwise designated in the result section) received 3000 rad of gamma irradiation and were used to test alloreactivity. To test autoreactivity, CD8⁺ T cell transductants were cultured with NOD-derived NIT-1 insulinoma cells (4 × 10⁵ cells for screening unless stated otherwise designated in the result section) (24) or islet cells harvested from NOD. Rag1^−/− or NOD.scid mice (6–9 × 10⁴ cells). NIT-1 cells and islets were cultured with interferon-gamma (1000 U/ml) for 24 hours and washed with culture media three times prior to adding T cell transductants. The CD4⁺ T cell transductant (9860-A3B3) was cultured with peptides (100 µg/ml, pS3: SRLGLWVRME, WE14: WSKMDQLAKELTAE, insulin B:9–23: SHLVEALYLVCGERG, synthesized by Genemed Synthesis Inc.) or islet cells (6–9 × 10⁴ cells) in the presence of irradiated NOD.Ca^−/− spleen cells (3000 rad, 1 × 10⁵ cells).

Identification of MHC molecules presenting antigens to T cells

T cell transductants were cultured with M12C3 cells expressing either H-2K^k or H-2D^k (4 × 10⁵ cells) to test recognition of MHC class I molecules. To test MHC class II recognition, T cell transductants were cultured with irradiated C3H/HeJ spleen cells in the absence or presence of anti-I-A^k antibody (10–3.6). Following overnight culture, IL-2 secretion in supernatants was measured by ELISA.

Statistical analysis

Statistical tests were performed using GraphPad Prism 6.0 software. The statistical tests used for each analysis are indicated in the results section or figure legends. A two-tailed P value of <0.05 was considered significant.

Results

Estimated frequency of autoreactive T cells retaining allogeneic reactivity

For any given allogeneic target MHC, the precursor frequency of alloreactivity has been estimated at 0.1–10% of T cells (25, 26). Therefore, we hypothesized that a proportion of autoreactive T cells recognize allogeneic MHC molecules. To test this hypothesis, we evaluated the frequency of autoreactive T cell clones that show alloreactivity. We tested T cell transductants expressing various islet-reactive TCRs derived from NOD mice (expressing H-2^g7) for in vitro reactivity against C3H (H-2^k), BALB/c (H-2^d), or C57BL/6 (H-2^b), spleen cells (Supplemental Table I). As a positive control, we confirmed responses against anti-CD3 stimulation for all T cell transductant lines studied (data not shown). As shown in Fig. 1, a number of autoreactive CD4 T cell transductants spontaneously reacted with allogeneic spleen cells, including three TCRs (12-1.19, 3–4, and BDC10.1) that reacted with multiple disparate allogeneic MHC molecules. Among 66 potential alloreactive responses, we detected 11 responses at a rate of 16.7%, which is moderately more frequent than previously proposed rates of alloreactivity (25, 26). Notably, seven TCRs restricted by I-A^g7 responded to BALB/c spleen cells, whereas only two responded to C57BL/6 or C3H (P < 0.02 by Chi-square test). I-A^g7 and I-A^d (expressed by BALB/c) share the identical alpha and 93% homologous beta chains (GenBank M15848 and AH002014). These data suggest that structural similarity of MHC may augment alloreactivity of TCRs derived from NOD mice. Overall, these results support the premise that autoreactive T cells retain the potential to respond to allogeneic MHC molecules.

Figure 1. Autoreactive T cells stochastically cross-react with allogeneic MHC.

A panel of known autoreactive T cell lines were co-cultured with irradiated spleen cells isolated from three mouse strains, C57BL/6 (H-2^b), C3H (H-2^k), and BALB/c (H-2^d). Following overnight culture with irradiated splenocytes, IL-2 production was quantified by ELISA. Results are stratified by degree of IL-2 production, with darker bars indicating production of greater levels of IL-2 in response to allogeneic spleen cells. Alloreactivity against BALB/c spleen cells was observed most frequently among the three mouse strains examined (P=0.02 by Chi-Square test). Data shown are representative of three independent experiments.

Presence of islet-autoreactive T cells in rejected allograft lesions

To address whether autoreactive T cells participate in islet allograft rejection, we first determined TCR repertoires in endogenous pancreatic islets as well as in islet grafts transplanted into NOD mice, and analyzed the frequency of TCR sequences shared between the two tissues. New-onset diabetic NOD mice were grafted with either syngeneic NOD.Rag1^−/− islets (isograft) or allogeneic C3H islets (allograft) which express completely disparate MHC class I and class II alleles (Supplemental Table I). At the time of transplant rejection, we collected both endogenous pancreatic islets and transplanted grafts. Total RNA isolated from these samples was analyzed by high-throughput sequencing to define and enumerate TCR sequences associated with autoimmune diabetes onset (detected in the pancreas) and those associated with transplant rejection (detected in the graft). Using this platform, we obtained 650,927 TCR alpha or beta chain sequence reads containing variable gene (V-gene), junction, and joining gene (J-gene) regions from 12 tissue samples isolated from 6 mice (3 mice transplanted with isograft, and 3 with allograft). The proportions of in-frame versus out-of-frame TCR sequences were similar across samples, with no significant differences between pancreatic islets and graft or between isograft and allograft (Supplemental Table II). Using all in-frame TCR alpha and beta chain sequences, we first assessed the presence and prevalence of TCRs detected in the endogenous pancreatic islets, which are assumed to be autoreactive, in transplanted graft lesions. As expected, over half of the TCRs detected in NOD.Rag1^−/− isografts were also found in the pancreas (average 60%, Fig. 2A), providing the baseline for detecting TCRs within the pancreas and transplant lesions by our TCR sequencing system. In mice receiving allogeneic C3H islet transplants, TCRs detected in the pancreas (denoted “pancreas-derived TCRs” hereafter) were also found in allograft lesions (Fig. 2B and 2C). The frequency of pancreas-derived TCRs in the allograft of two of three mice (9860 and 9812) was as high as that in the isograft (no significant difference by unpaired t test), whereas the remaining mouse (9805) had relatively low pancreas-derived TCRs in the graft; however approximately 14% of TCRs were still pancreas-derived. These results provide direct evidence of autoreactive T cells infiltrating allograft islet lesions in NOD mice with variable prevalence.

Figure 2. Detection of autoreactive TCR clonotypes within allograft lesions.

NOD mice were transplanted with pancreatic islets isolated from NOD.Rag^−/− (panel A) or C3H mice (panels B and C), followed by high-throughput sequencing to determine TCR sequences in the graft and endogenous pancreatic islets at the time of transplant rejection. Individual TCR clonotypes (represented by individual single dots) are depicted as % TCR detected within the graft on the x-axis and % detected in the endogenous pancreas on the y-axis for each mouse. Percentage of TCR reads shared with the pancreas among total reads detected in the graft is shown at the upper right corner in each panel. Red symbols represent TCR clonotypes analyzed for functional auto- and allo-reactivities. Panels A and B represent analysis for alpha chain clonotypes, and panel C for beta chains.

Vigorous accumulation of autoreactive T cells in islet grafts

Memory T cells have been recognized as a critical population that accelerates transplant rejection. Recent studies demonstrated the presence of anti-viral memory CD8 T cells in rejected allografts in mice (18, 19). In addition, the pervasiveness of such memory CD8 T cells increases the risk of rejection in patients receiving lung transplantation (30). We hypothesized that autoreactive T cells, which have likely encountered antigens in the endogenous pancreatic islets and thus may be considered memory/effector T cells, accumulate and play a role in allograft rejection. To address enrichment of autoreactive T cells in graft lesions, we examined whether pancreas-derived autoreactive T cells preferentially accumulate in islet graft lesions as compared to T cells that are solely detected in the graft and likely “purely” alloreactive. Fig. 3 shows the association between the prevalence of TCR clonotypes in graft lesions and the presence of these clonotypes in endogenous pancreatic islets. Lower numbers (1st–20th) denote higher rankings (i.e., higher frequency) of TCR clonotypes than higher numbers (>1000th). For example, TCRs ranked 1st–20th are the most frequent 20 clonotypes in a TCR frequency list of each sample. In isograft recipient mice, the majority of high-ranked (i.e., frequent) TCR clonotypes were found in the pancreas, whereas low-ranked clonotypes were preferentially represented by those that are detected only in graft lesions (Fig. 3A). These results suggest that pancreas-derived T cells accumulate more vigorously than non-pancreas-derived T cells, which is consistent with the prediction that pancreas-derived memory/effector T cells rapidly propagate due to recognition of identical antigens expressed in syngeneic islet graft lesions. This observation has been previously demonstrated by Tisch and colleagues who showed that the prevalence of memory/effector T cells expressing a particular TCR V-gene (Vbeta 12) increased in syngeneic islet grafts transplanted in NOD mice (31). Our study confirms and extends these findings at the complete TCR sequence level containing the junction region.

Figure 3. Detection of autoreactive TCR clonotypes within allograft lesions.

TCR clonotypes detected in the graft were grouped by frequency (i.e. ranked at 1st–20th, 21st–100th, 101st–1000th, and greater than 1000th in a TCR frequency list) for each mouse. In each group of TCR clonotypes, percentages of those that are also detected in the pancreas are depicted as black bars, and percentages of those that are detected only in the graft are depicted as white bars. For example, 100% of the 20 most frequent TCR clonotypes in the graft of mouse 1687 were also detected in the pancreas, whereas only <10% of least frequent TCR clonotypes ranked greater than 1000th in the frequency list were found in the pancreas (A. left panel). Panels A represent analysis for alpha chain clonotypes in isograft recipients. Panels B and C represent analysis of allograft recipients for alpha and beta chains, respectively.

Despite the disparate MHC expression by the allograft, the same trend found in isograft recipients was observed in two allograft recipients that exhibited high prevalence of pancreas-derived T cells in allograft lesions (9860 and 9812, Fig. 3B and 3C). To corroborate these results, we performed an independent analysis of the most frequent 100 pancreas-derived TCR clonotypes in the graft, and found that the majority of clonotypes were more prevalent in the graft than the endogenous pancreas (Supplemental Fig. 1D, 1E, 1G, and 1H). In contrast, the remaining allograft recipient (9805) that had a relatively low frequency of pancreas-derived T cells had limited infiltration of robustly accumulating pancreas-derived T cells in the graft (Fig. 3B and 3C, Supplemental Fig. 1F and 1I). Taken together, pancreas-derived T cells preferentially accumulate in the islet grafts transplanted in two of three allograft recipient mice, suggesting that autoreactive T cells play a key role in allograft rejection in a proportion of animals.

Highly prevalent pancreas-derived TCR clonotypes are functionally alloreactive

Autoreactive T cells, which spontaneously cross-react with allogeneic MHC molecules, could be one explanation for accelerated allograft rejection observed in autoimmune recipients. We next addressed whether highly enriched T cells in allograft lesions respond to both autoantigens and allogeneic MHC. To determine whether NOD pancreas-derived TCR sequences, which were highly prevalent in the graft, were functionally alloreactive, we recapitulated these TCRs in 5KC T cell hybridoma cells expressing either CD4 or CD8 co-receptors using retroviral vectors. We chose the most prevalent TCR alpha and beta clonotypes detected in the grafts of the two allograft recipients that had robust infiltration of pancreas-derived T cells, and examined nearly all possible alpha and beta combinations; the exception being the third most frequent TCR alpha and beta clonotypes in mouse 9860, 9860-A3 and 9860-B3 (Supplemental Table III). Because TCR pair 9860-A3B3 is known to be expressed by a chromogranin A-reactive CD4 T cell clone derived from pancreatic islets of another NOD mouse (M. Nakayama, unpublished data), 9860-A3B3 was expressed on 5KC cells along with the CD4 receptor, and no other combinations with either 9860-A3 or 9860-B3 were included in the screening analyses. We confirmed that all cell lines expressing the various TCR alpha and beta combinations produced IL-2 in the presence of antibodies against CD3 (data not shown). We tested these candidate cell lines for alloreactivity against irradiated C3H spleen cells in vitro, and found six cell lines showing allogeneic responsiveness including 9860-A3B3 (Fig. 4A). Dose-dependent responses to C3H cells confirmed alloreactivity of these TCR combinations (Fig. 4B). These data demonstrate that TCR clonotypes enriched within the rejected islet allograft lesions in NOD mice, all of which were simultaneously detected in the pancreatic islets, are functionally alloreactive.

Figure 4. Alloreactivity by prevalent TCR clonotypes in the graft.

(A) Various combinations of most frequent alpha and beta chain clonotypes detected within allograft lesions were expressed in T hybridoma cells along with either the CD4 or CD8 co-receptor. Heat maps show IL-2 production by individual cell lines in response to irradiated C3H spleen cells. For each TCR alpha and beta combination, the upper and lower squares represent IL-2 production by cells expressing the CD4 or CD8 co-receptor, respectively. Cell lines consecutively producing >50 pg/ml of IL-2 in three independent experiments were designated as positive. The CD8 co-receptor facilitated more robust IL-2 production than CD4 for TCR combinations other than 9860-A3B3; therefore, subsequent experiments with the other 5 TCRs were performed using CD8-expressing T cell lines. (B) Titration experiment showing dose-dependent IL-2 production in response to irradiated C3H or NOD.Ca^−/− (negative control) spleen cells. Each panel depicts one of the six cell lines that produced IL-2 in the screening test shown in panel A. (C) IL-2 production by six C3H-reactive cell lines in response to cell lines expressing H-2D^k or K^k. (D) IL-2 production by six C3H-reactive cell lines cultured with irradiated C3H spleen cells in the presence or absence of antibody against I-A^k. Statistical significance for experiments shown in panels C and D was analyzed by unpaired t test. Each data point shown in panels B–D is average ± SEM of duplicate samples. Data shown in A–D are representative of three independent experiments for each panel.

To confirm allogeneic reactivity at the molecular level, we determined MHC molecules that are recognized by these six TCRs. Responses by the five alloreactive T cell lines other than 9860-A3B3 were stronger when co-expressing CD8 rather than CD4 (Fig. 4A); therefore, we used CD8-positive cell lines for these five TCR pairs for subsequent analysis. All T cell lines except 9860-A3B3 responded to a mouse B cell line expressing D^k (MHC class I expressed by C3H mice), and one T cell line (9812-A5B1) recognized K^k as well, whereas a negative control cell line expressing CD8 receptor without TCR did not respond to either D^k or K^k-expressing cells (Fig. 4C). In parallel analyses, responses against C3H spleen cells by 9860-A3B3 were suppressed in the presence of anti-I-A^k antibodies (Fig. 4D), indicating that 9860-A3B3 reacts with I-A^k. In contrast, production of IL-2 by the other cell lines was not significantly changed in the presence of anti-I-A^k antibodies (Fig. 4D). These results suggest that all six T cell lines directly recognize donor C3H MHC molecules, but not a peptide derived from donor MHC loaded in recipient MHC. This observation of direct alloreactivity mediated by highly enriched TCRs within the rapidly rejected allograft is in agreement with rejection seven days following islet transplantation. In summary, we identified six functionally alloreactive TCR pairs, consisting of pancreas-derived TCR alpha and beta clonotypes, in the graft of diabetic NOD mice. Five of the TCR pairs recognized MHC-class I, and one pair recognized MHC-class II.

Functionally alloreactive TCR pairs react to syngeneic islets

To verify that the six functionally alloreactive TCR pairs are autoreactive against NOD islets, we tested cell lines expressing these TCRs for their ability to respond to islet cell lines or islet-derived peptides. Four of the five MHC class I-restricted T cell lines responded to NOD insulinoma beta cell-derived NIT-1 cells (24) in a dose-dependent manner (Fig. 5A). The 9860-A3B3 MHC class II-restricted T cell line responded to the pS3 mimotope of the chromogranin A peptide (WE14) as expected, but not to the negative control insulin peptide (Fig. 5B). The 9860-A3B3 line, as well as the NIT-1-reactive T cell lines, reacted to islet cells isolated from syngeneic NOD.scid or NOD.Rag1^−/− mice, further confirming the true autoreactivity of these five alloreactive TCR pairs (Fig. 5B and 5C). Combined with the in vitro alloreactive responses described above, these data provide evidence of T cells that mediate both autoreactivity and alloreactivity within islet allografts transplanted in autoimmune recipients.

Figure 5. Autoreactivity by prevalent TCR clonotypes in the graft.

(A) Five MHC class I-restricted alloreactive T cell lines were cultured with NIT-1 cells derived from NOD insulinoma cells. Four out of 5 cell lines produced IL-2 in response to NIT-1 cells in a dose-dependent manner. (B) MHC class II-restricted T cell line, 9860-A3B3, was cultured in the presence of various peptides or NOD islet cells. IL-2 production was detected in response to the pS3 mimotope of the chromogranin A peptide as well as islet cells. (C) IL-2 production by the five MHC class I-restricted alloreactive T cell lines or TCR-null negative control cell line in response to NOD islet cells. Four out of 5 cell lines that reacted with NIT-1 cells produced higher levels of IL-2 in response to NOD islet cells compared to the negative control cell line not expressing TCR. Data shown in A–C are representative of three independent experiments.

Discussion

In this report, we demonstrated the presence and accumulation of autoreactive T cells in rejected islet allografts in NOD mice using an unbiased TCR sequencing approach. In two thirds of the mice analyzed, approximately half of TCR sequences found in allograft lesions were also detected in endogenous islets. While it is unclear whether T cells expressing these overlapping TCRs migrate from endogenous islets to allograft lesions or vice versa, our sequencing data indicate that a large proportion of T cells in allografts are autoreactive. We found that these pancreas-derived autoreactive TCRs were more frequent than those that were solely detected in allografts, implying rapid accumulation of autoreactive T cells. These results suggest a key role for autoreactive T cells in islet allograft rejection in recipients with pre-existing autoimmunity. Furthermore, functional analysis of the most prevalent TCRs in allografts, which were also detected in endogenous islets, indicated that they were not only autoreactive to syngeneic islets, but also recognized allogeneic MHC molecules. Taken together, these results provide evidence that autoreactive T cells participate in allogeneic graft rejection by heterologous alloimmunity.

Two distinct mechanisms by which autoreactive T cells recognize allogeneic graft tissues have been proposed by studies utilizing experimental rodent models with limited T cell repertoires: 1) autoreactive T cells react with antigens presented by self APCs of recipients (indirect pathway); or 2) T cells may directly interact with allogeneic MHC molecules expressed by donor cells through heterologous recognition (direct pathway) (32–34). Notably, these mechanisms have not been previously demonstrated in spontaneous autoimmune disease settings. Although we analyzed only three mice and variation between individual animals was detected presumably due to different autoimmune activity at the time of transplantation, our functional analyses suggest that heterologous alloreactivity is a substantial pathway mediated by autoreactive T cells in autoimmune transplant recipients. Interestingly, the relative strengths of autoreactive and alloreactive responses by each TCR varied. The CD4 T cell line (9860-A3B3) responded to syngeneic NOD islet cells with higher IL-2 production compared to allogeneic spleen cells. Of note, this TCR pair was prevalent in the endogenous islets, suggesting that T cells expressing 9860-A3B3 are the major contributor to autoimmune beta cell destruction mediated by recognition of an antigen presented by self MHC. These results intimate that recognition of allograft via the indirect pathway may be a prominent mechanism of the rejection. In contrast, IL-2 production by the four dual-reactive CD8 T cell lines in response to allogeneic antigens was enormous compared to the response to autoantigen, suggesting their contribution to the rejection by heterologous alloimmunity. An important question is whether the phenotype of these dual-reactive T cells changes upon migration from the endogenous islet where they react with autoantigens to the allograft where they encounter a non-self MHC molecule. While outside the scope of the current study, this question could be pursued using advanced technologies that allow in situ analysis of single cells.

Due to simultaneous reactivity against both autoantigens and transplant-derived allogeneic MHC, our data suggest that autoimmune individuals are pre-sensitized against transplanted tissues expressing allogeneic MHC haplotypes. Therefore, heterologous alloimmunity mediated by autoreactive T cells may represent a stringent barrier to tolerance-promoting therapies as previously shown for memory/effector T cells specific for viruses and other pathogens (15–17, 35–39). Current immunosuppressive regimens for pancreas or pancreas-kidney transplants include calcineurin inhibitors as well as T cell depletion (40, 41). Because these therapies non-specifically suppress both memory/effector and naïve T cell subsets (42), several trial medications exclusively targeting memory T cells have been tested for prevention of allogeneic graft rejection in nonhuman primate models, as well as a small number of patients receiving pancreas or pancreas-kidney transplants (43–47). While drug regimens specifically targeting memory T cells have shown similar or improved graft survival compared to standard immunosuppressive therapies, severe impairment of immune protection against pathogens remains a serious side-effect.

Our findings demonstrate that the presence of autoimmunity mediates heterologous allogeneic reactivity, and therefore pose the need for therapies that render specific suppression of autoreactive memory/effector T cells in allogeneic transplantation for individuals having autoimmune diseases such as type 1 diabetes. Several immunotherapies have shown promise in clinical trials to delay onset of type 1 diabetes, and such therapies may be useful to inhibit autoreactivity in the recipient of allogeneic islets (48). Therapeutic targets against memory/effector T cells, such as blockade of the IL-7 receptor (49), have been demonstrated to be effective in suppressing diabetes of NOD mice. In particular, anti-thymocyte globulin, which is widely used to suppress alloreactivity in organ transplantation including pancreas transplantation, has been shown to prevent diabetes and reverse hyperglycemia when combined with other immune modulators such as CTLA-4 immunoglobulin or G-CSF (50, 51). While caution should be applied to doses and timings since mechanisms for suppressing alloreactive and autoreactive T cells may differ, agents that are involved in pathways of both alloimmunity and autoimmunity may be able to efficiently protect allogeneic tissues transplanted in autoimmune recipients. Alternatively, autoantigen-specific immune modification may be worthwhile, especially combined with drugs targeting memory/effector T cells that are discussed above. A large number of clinical trials directing autoimmunity are ongoing to prevent and cure type 1 diabetes. Provision of such therapies to regulate autoreactive memory/effector T cells in conjunction with general immunosuppressive treatments is worth considering for type 1 diabetes patients receiving allogeneic beta cell replacement therapy.

In conclusion, we demonstrate that autoimmunity is an endogenous source of heterologous alloreactivity in the rejection of allogeneic transplantation. This study provides evidence that silencing of autoimmunity may be required for successful allogeneic tissue transplantation in autoimmune individuals.

Abbreviations used in this article

GAD

glutamic acid decarboxylase

IA-2

islet antigen-2

MSCV

mouse stem cell virus