Cross-reactivity between an autoantigen and unknown microbial epitopes has been proposed as a molecular mechanism involved in the development of insulin-dependent diabetes (type 1 diabetes). Type 1 diabetes is an autoimmune disease that occurs in humans and the nonobese diabetic (NOD) mouse. BDC2.5 is an islet-specific CD4+ T-cell clone derived from the NOD mouse whose natural target antigen is unknown. A biometrical analysis of screening data from BDC2.5 T-cells and a positional scanning synthetic combinatorial library (PS-SCL) was used to analyze and rank all peptides in public viral and bacterial protein databases and identify potential molecular mimic sequences with predicted reactivity. Selected sequences were synthesized and tested for stimulatory activity with BDC2.5 T-cells. Active peptides were identified, and some of them were also able to stimulate spontaneously activated T-cells derived from young, pre-diabetic NOD mice, indicating that the reactivity of the BDC2.5 T-cell is directed at numerous mouse peptides. Our results provide evidence for their possible role as T-cell ligands involved in the activation of diabetogenic T-cells.
Type 1 diabetes (insulin dependent) is an autoimmune disease that occurs spontaneously in humans and in the experimental nonobese diabetic (NOD) mouse model (1). In both humans and the NOD mouse, there is an inflammatory lymphocytic infiltrate (insulitis) within pancreatic islets accompanied by circulating anti-islet antibodies. The ultimate destruction of the insulin-producing β-cells is mediated by T-cells, with both CD4+ and CD8+ cells involved in disease development (2,3). There is also a strong genetic component for susceptibility to the disease associated with the major histocompatibility complex (MHC) in humans. Similarly, in the NOD model, it is known that the H-2g7 MHC allele influences the development of type 1 diabetes (4).
Cross-reactivity is thought to be a normal feature of the T-cell receptor (TCR) recognition (5,6). This facet of TCR recognition may arise from the overall nature of the repertoire available in an individual. Because only 5% of thymocytes are positively selected in the thymus (7) and T-cell numbers are limited, cross-reactivity may indeed represent economy of scale (8). On the other hand, T-cell cross-reactivity between self and microbial antigens (molecular mimicry) can trigger autoimmunity (9,10). Interestingly, it has been proposed (10–12) that T-cell cross-reactivity between β-cell autoantigens and microbial antigens plays a role in the development of type 1 diabetes. Molecular mimicry between the GAD65 protein and the Coxsackie viral protein P2C (11,13) has been suggested in the pathogenesis of type 1 diabetes. A different Coxsackie viral protein, VP-1, has also been shown to induce T-cells that are cross-reactive for tyrosine phosphatase (insulinoma-associated protein 2/IAR) (14). In the search to identify other potential molecular mimics, it can be difficult to predict which antigenic determinants will be cross-reactive. This is due to the fact that the relevant peptide/MHC conformation can be mimicked by distantly related peptides with no overt sequence homology (15). Positional scanning synthetic combinatorial libraries (PS-SCLs) have been successfully used to identify antigenic determinants cross-recognized by antigen-specific T-cells, as recently reviewed (16). Data generated utilizing clonal T-cells to screen PS-SCL have recently been used (6,17) in a biometric analysis to analyze and rank all peptides in public databases to identify sequences with predicted reactivity.
Following in vitro activation, the CD4+ T-cell clone (TCC), BDC2.5, is diabetogenic in NOD/scid mice (18). Previous studies have shown that BDC2.5 T-cells are specific for an as yet undefined β-islet granule membrane antigen when presented by the NOD class II MHC H-2g7. The transgenic strain of the NOD mouse that expresses the TCR α- and β-chain genes of the BDC2.5 TCC has been used in our previous studies (19) to identify MHC class II–restricted peptide ligands that are stimulatory for T-cells from NOD mice. The sequences of these superagonist peptides were derived from the combinations of the most active mixtures of a biased PS-SCL and do not correspond to any currently known natural sequences.
This report presents the integration of a biometrical analysis that uses the screening data from the biased positional scanning library (19) to analyze and rank peptides from public viral and bacterial protein databases to identify potential molecular mimic sequences with predicted reactivity (17). Groups of these peptides were synthesized and tested for their stimulatory activity with BDC2.5 T-cells. Interestingly, several of the peptides found in viral and bacterial proteins had significant stimulatory activity at 1 μg/ml. These peptides were also able to stimulate proliferative responses in T-cells derived from young, pre-diabetic NOD mice, indicating that their specificities were contained within the NOD T-cell repertoire. Our results suggest that T-cells from NOD mice can respond to a number of microbial antigens, signifying new instigators for the study of autoimmune pathogenesis.
RESEARCH DESIGN AND METHODS
The screening of the decapeptide positional scanning library, together with the biased library that was used for the identification of the BDC2.5 superagonist peptides (19), was used to generate a scoring matrix. A value for each of the 20 amino acids of the decapeptide is derived from mixtures defined with a given amino acid. Based on the assumption of independent and additive contribution of the individual amino acids at each position of a peptide to the peptide’s activity, the score of each individual peptide of all of the proteins in the public databases was calculated by adding individual stimulatory values of the composing amino acids. In this manner, the biometrical analysis scores all overlapping decapeptides contained in the public protein database Genpept version 117 and identifies those sequences with the highest predicted stimulatory value.
NOD/shi, NOD/Lt-scid/scid, and NOD.BDC2.5 TCR-transgenic mice were obtained from the mouse colony maintained at The Scripps Research Institute and were housed in a specific-pathogen–free environment. The generation of the BDC2.5 mouse has been described by others (20).
Individual peptides were synthesized by the simultaneous multiple peptide synthesis method (21). Purity and identity of each peptide were characterized using an electrospray mass spectrometer interfaced with a liquid chromatography system.
Culture conditions.
Whole spleen cells from 5- to 7-week-old BDC2.5 TCR-transgenic mice were depleted of erythrocytes and cultured in microtiter plates in standard T-cell medium containing dilutions of individual peptides. In BDC2.5 assays, both male and female splenocytes were used as T-cells (200 × 103 cells/well) without the addition of antigen-presenting cells (APCs). Cultures were harvested after 72 h incubation plus overnight exposure to 0.5 μCi/well [3H]TdR (6.7 Ci · mmol−1 · l−1), and incorporated radioactivity was assessed by scintillation counting. For NOD experiments, whole spleen cells from female NOD/shi mice were used as T-cells (100 × 103 cells/well), and irradiated male NOD/shi splenocytes (250 × 103 cells/well) were used as APCs. Cells were processed and cultured as described above. The medium used for all assays consisted of RPMI 1640 (BioWhittaker, Walkersville, MD) supplemented with 8% FBS (Hyclone Laboratories, Logan, UT), HEPES buffer (10 mmol/l; BioWhittaker), 2-mercaptoethanol (50 μmol/l; BioRad, Richmond, CA), penicillin-streptomycin (5 units/ml and 50 μg/ml, respectively; BioWhittaker) and glutamine (2 mmol/l; BioWhittaker).
Secondary proliferation and establishment of antigen-specific NOD cell lines.
In the original primary proliferation, 9- to 10-week-old NOD female splenocytes were stimulated with 25 μg/ml of peptides in the presence of irradiated NOD male splenocytes. After a 72-h incubation, the plates were split, with one set pulsed for primary proliferation measurements and the second set maintained with 150 units/ml interleukin-2 (BD Pharmingen, San Diego, CA) at 3-day intervals. The secondary proliferation assay was run 14 days later, in which the cultured wells were split into a background (− peptide) set, a stimulated (+ peptide) set, and a maintenance set. The stimulation index (SI) for the secondary proliferation was determined as counts per million (stimulated well)/counts per million (background well).
Determination of half-maximal proliferative response values of peptides.
T-cell populations were cultured using conditions described above with varying dilutions of peptides. The concentration causing a half-maximal proliferative response (EC50) was determined by curve-fitting using a scientific graphics software program (GraphPad Prism; Graph Pad Software, San Diego, CA). The EC50 values were determined by two different methods: fixing the maximal proliferative response to the mean of the highest values obtained in each experiment or using the mean individual maximum of each sample.
Anti-MHC assay.
BDC2.5 splenocytes or antigen-specific NOD T-cells were cultured with active peptides in the presence and absence of anti–H-2 Ig7, a strain-specific anti–MHC class II antibody (kindly provided by Alex Ilic, from the laboratory of N.S.). Peptides were also run in the presence of mouse IgG2a (BD Pharmingen) as an isotype control. Both antibodies were run at a final concentration of 50 μg/ml. Cultures were harvested after 72 h incubation plus overnight exposure to 0.5 μCi/well [3H]TdR (6.7 Ci · mmol−1 · l−1), and incorporated radioactivity was assessed by scintillation counting.
Cross-reactivity of antigen-specific NOD T-cell lines.
Rested antigen-specific cells that had been kept in culture by stimulation every 2 weeks with their respective peptides were washed and stimulated with either 25 or 5 μg/ml of test peptides in the presence of irradiated NOD male splenocytes. Peptide 1040-31 was tested at 5 and 0.5 μg/ml, respectively.
Flow cytometry Vβ analysis.
Antigen-specific NOD cell lines p1308-84 (Sendai virus protein) and p1308-29 (Neisseria meningitidis phage protein) were stimulated with their respective peptides for 3 days in the conditions mentioned above. On day 3 after culture, cells were washed and incubated with phycoerythrin-labeled monoclonal antibodies specific for the β-chain of the TCR (BD Pharmingen) and appropriate isotype controls. Samples were analyzed by flow cytometry.
Adoptive transfer of activated BDC2.5 cells.
Whole spleen cells from 7- to 8-week-old BDC2.5 TCR-transgenic mice were cultured in mouse T-cell medium for 4 days with selected stimulatory peptides at the specified concentrations. On day 4, 5 × 106 cells were injected intravenously into groups of three NOD/scid mice at 6 weeks of age.
Immunohistochemistry and insulitis indexes.
Excised pancreata were fixed overnight in 10% neutral buffered formalin (3.6% formaldehyde) and embedded in paraffin. Sections (5 μm) were stained with hematoxylin and eosin. The degree of insulitis within the pancreas was analyzed using a light microscope. Pathology was graded as 1) no insulitis (normal islet morphology), 2) peri-insulitis (mononuclear cells in the peri-insular space), 3) insulitis (substantial mononuclear infiltration), or 4) remnant islet. For statistical comparisons, at least 33 islets from each group were counted.
Blood glucose measurements and diabetes monitoring.
Glucose levels were determined from tail vein blood samples using Glucometer Elite test strips (Bayer, Elkhart, IN) on a standard glucometer with a range of 20–400 mg/dl. Mice with two consecutive values >300 mg/dl were considered diabetic.
Statistical analysis.
All P values were calculated using a two-tailed Student’s t test.
RESULTS
Viral and bacterial sequences trigger proliferative responses in diabetogenic BDC2.5 T-cells.
The first goal of this study was to identify potential epitopes from natural proteins that were capable of stimulating diabetogenic BDC2.5 CD4+ T-cells. To do this, the information previously obtained from the screening of the biased sublibrary was used to perform a PS-SCL–based biometric analysis as recently described (17). This analysis allows the results of a PS-SCL screening to be systematically compared with all of the peptides within a given protein database to identify peptide sequences from naturally occurring proteins with predicted stimulatory activity. Interestingly, a number of peptide sequences from bacterial and viral sources were identified. All of the possible overlapping decapeptide sequences in each protein of the database were scored based on a matrix derived from the screening results of a biased PS-SCL with BDC2.5 cells. About 100,000 proteins each of viral and bacterial origin were analyzed, resulting in 20–40 million decapeptide sequences being scored per database. The peptides with the highest scores (47 bacterial and 27 viral) were selected for synthesis. These peptides were tested for their stimulatory activity with BDC2.5 T-cells (Fig. 1).
The sequences of the BDC2.5 stimulatory peptides having an SI ≥3 at 1 μg/ml or an SI ≥10 at 10 μg/ml are listed in Table 1 with their predicted activity score, protein source, and microorganism of origin. With an SI of 79, a sequence from a hydratase/aldolase PhnE enzyme (TPI 1308-26) found in a Burkholderia species of bacteria was the most active at 1 μg/ml. Sequences from a putative phage virion protein of Neisseria meningitidis (TPI 1308-29), an exopolyphosphatase enzyme found in a Synechocystis species (TPI 1308-33), and a ferrodoxin reductase enzyme from Pseudomonas putida (TPI 1308-24) were the most active bacterial sequences at 1 μg/ml. Interestingly, a peptide sequence from the tegument protein of human herpes simplex virus type I (TPI 1136-7) was the most active viral sequence at 1 μg/ml.
To compare the activity of these compounds with previously identified superagonists, dose-response assays were performed. The most stimulatory peptides were tested by serial dilution in proliferation assays with BDC2.5 splenocytes. As seen in Fig. 2, the EC50 values of these naturally occurring peptide sequences are higher than those for the superagonists (19). The PS-SCL–based biometrical analysis scores and ranks all of the peptides of a given database using a matrix derived from the screening of a PS-SCL. Therefore, the peptides with the highest scores, which were the ones synthesized for this study, do not necessarily have the optimal amino acids at all positions, leading to the identification of naturally occurring sequences with agonist or weak agonist capability. In contrast, the superagonist sequences (19) were derived from the combinations of the defined amino acids of the most active mixtures at each position of the library and did not correspond to any naturally occurring protein sequence in the Genpept version 117 database. Therefore, their stimulation is expected to be optimized compared with naturally occurring peptides.
Stimulation of BDC2.5 T-cells with naturally occurring peptides is dependent on binding to a class II molecule.
Sequence and structural modifications can alter the capacity of peptides to bind to the class II complex. Because most of the naturally occurring peptides identified in this study have residues that are different from those present in the superagonists (19), it was important to evaluate whether the capacity of these peptides to trigger proliferative responses with BDC2.5 was also dependent on MHC-TCR binding. To test this, BDC2.5 splenocytes were stimulated in proliferation assays with these peptides in the presence or absence of an anti–MHC class II antibody, anti–H-2 IAg7 (Fig. 3). In the presence of the blocking antibody and not the isotype control, peptide-induced proliferation was reduced by a range of 64% (P = 0.01) to 96% (P = 0.01) for the viral peptide p1136-7 and bacterial peptide p1308-26, respectively, as well as for the tested superagonist, p1040-31, to 92% (P = 0.01). These results show that the selected identified natural peptides are not acting as superantigens or growth factors to stimulate proliferative responses in autoreactive BDC2.5 T-cells. These peptides are presented in an MHC-restricted fashion that is similar to the majority of cross-reactive peptides that have been found to stimulate TCCs (22–24).
Spontaneously activated T-cells from pre-diabetic NOD mice contain subsets that respond to the microorganism sequences.
The cross-reactive peptides identified in this study were able to stimulate proliferative responses in BDC2.5 T-cells. BDC2.5 T-cells were originally isolated from diabetic NOD mice (25) and are contained within the autoreactive T-cell pool (26). Therefore to determine whether these naturally occurring specificities were part of the T-cell repertoire from pre-diabetic NOD mice, we looked for the presence of T-cell subsets that respond to these identified ligands. Splenocytes from 9- to 10-week-old female, pre-diabetic mice were cultured with the most active bacterial and viral peptides from the BDC2.5 assays listed in Fig. 2. With the exception of the TPI 1040-31 superagonist, little or no proliferation was observed during the primary proliferation assay period, indicating that responding cells may be present at a low frequency. However, upon secondary stimulation of the same cell populations, proliferative activity was demonstrated in response to six of these sequences (Fig. 4).
To characterize phenotypic features of NOD T-cells stimulated with several of these microbial antigens, a portion of the cells from each well of the secondary proliferation assay protocol was retained for the establishment of antigen-specific T-cell lines. Two antigen-specific cell lines have been established. One is specific for TPI 1308-29, a Neisseria meningitidis phage sequence, and the other is specific for TPI 1308-84, a Sendai virus C′ protein sequence. To evaluate the antigen specificity of these lines compared with BDC2.5 cells, they were stimulated with the panel of active sequences listed in Fig. 1, including a 17-mer ovalbumin peptide sequence that is known to bind H-2 Ig7 as a negative control (27). Interestingly, although the 1308-84 cell line (Fig. 5A) recognizes several of the other active peptide sequences, the 1308-29 cell line (Fig. 5B) appears to recognize only the peptide that was used for expansion as well as the superagonist. Neither cell line recognized the negative control. When the TCR used by these two cell lines was characterized, it was found that the Vβ8.1/8.2 TCR chain was almost exclusively selected in this response (Fig. 6), although a smaller percentage of the 1308-29 cells contain that particular Vβ chain. Interestingly, in experimental allergic encephalomyelitis, an animal model for the autoimmune disease multiple sclerosis, the Vβ8.2 TCR chain is also known to be the primary TCR V gene used by both the B10.PL and PL/J mouse T-cell repertoires directed against the immunodominant myelin basic protein epitope involved in the disease (28). These results show that even though both cross-reactive peptides are able to stimulate BDC2.5 cells, each of them expands a unique set of NOD T-cells.
Adoptive transfer of activated BDC2.5 T-cells with cross-reactive peptide 1308-84 triggers diabetes in NOD/scid recipients.
NOD/scid recipients become diabetic when they receive BDC2.5 T-cells stimulated with a superagonist peptide (19). To determine whether stimulation with the naturally occurring peptides identified in this study results in the activation of pathogenic capacity, NOD/scid mice were transferred with BDC2.5 cells that had been cultured for 3 days in the presence of the p1308-84 peptide derived from the murine Sendai virus. On the sixth day after transfer, homing of p1308-84–stimulated BDC2.5 T-cells to the pancreas of NOD/scid mice was significantly accelerated compared with NOD/scid mice transferred with nonstimulated cells. One hundred percent of islets were destroyed and blood glucose levels were >300 mg/dl in the NOD/scid recipients of peptide-stimulated cells, whereas only 24% of the islets in control mice were destroyed during this time (P = 0.0013), and none of these mice developed diabetes (Table 2). It is evident from these results that stimulation of cells with this natural viral mimotope was able to transfer disease in a rapid and aggressive fashion.
DISCUSSION
Type 1 diabetes results from a combination of genetic, immunologic, and environmental factors. T-cell cross-reactivity to β-cell autoantigens and unknown microbial epitopes has been proposed (29,30) as a molecular mechanism for the breakdown of immunologic tolerance and the subsequent appearance of T-cell autoreactivity. However, there are only limited examples of microbial agents linked to autoimmune pathogenicity (24,31–36). Our study demonstrates that antigens from common human and mice pathogens can activate T-cells arising within a spontaneous autoimmune environment. The use of the PS-SCL–based biometrical analysis to identify ligands of clones of unknown specificity has been validated in previously published reports (24,37).
It is worth noting that despite sharing the same P4W6M9 motif present in the superagonist peptides (19), these ligands are less stimulatory for BDC2.5 T-cells, indicating that other positions in the decapeptide may play a role in T-cell activation. Of the 72 peptides synthesized with the P4W6M9 motif, only 10 had an SI >3 at 1 μg/ml, indicating that the presence of this particular motif does not guarantee superagonist levels of activity. Some of these active peptides are derived from proteins of pathogenic bacteria, Neisseria meningitidis, a known human pathogen, and Alcaligenes faecalis, a potential human pathogen. The most active peptides of viral origin are derived from a protein of the human herpes simplex type 1 virus (p1136-7) and one derived from the murine Sendai virus (p1308-84). The Sendai virus is a major respiratory pathogen of mice. This paramyxovirus shares sequence homology with the human pathogen parainfluenza type 1 virus, which is responsible for respiratory disease in infants and children (38). To determine whether the identified peptides could be naturally processed epitopes, the source proteins for these two peptides as well as the other NOD stimulatory peptides (Fig. 4) were analyzed using asparaginyl endopeptidase, a cysteine protease known to be prevalent in APCs (39). This enzyme is now thought to be the dominant enzyme capable of initiating MHC class II–restricted antigen processing and has been extensively studied (40) in the tetanus toxin antigen model system. Cleavage site maps were generated for the corresponding proteins to the identified NOD stimulatory peptides, and in all cases the identified peptides were among the predicted naturally processed fragments (data not shown). Finally, while the biological relevance of these active sequences with regard to type 1 diabetes is as yet unknown, BDC2.5 T-cells, when stimulated by the peptide derived from the Sendai virus (p1308–84), have been able to adoptively transfer disease into NOD/scid mice (Table 2). However, we have not been able to transfer disease into NOD/scid mice using a peptide-specific NOD cell line stimulated in vitro with p1308-84. Our results are consistent with the recent report by Stratmann et al. (41), in which an NOD T-cell line, selected and expanded with an Ag7/BDC2.5 mimotope (AHHPIWARMDA) tetramer, was not able to generate consistent insulitis or diabetes when transferred to RAG0/0/NOD mice. That mimotope, and the microbial peptides reported here, contain the motif P4W6M9, which has been found to be highly effective in stimulating BDC2.5 cells. One possible explanation for the inability of NOD cells that are expanded in vitro to cause diabetes is the lack of avidity modulation by tissue APCs, the absence of CD8+ cells for optimal disease transfer (42), or the in vitro generation of regulatory T-cells able to counterregulate the pathogenic activity of the autoimmune antigen-specific T-cell population. Thus, the in vivo behavior of in vitro–derived CD4+ T-cell lines cannot be predicted based solely on their TCR specificity. Indeed, to our knowledge, there are no examples of diabetes induced by T-cell lines derived from spontaneously arising autoreactive cells that have been stimulated in vitro with peptide antigens.
In humans, the self-antigens GAD65 and proinsulin are known to be stimulatory for T-cells from individuals considered at risk for type 1 diabetes (43). CD4+ GAD65-reactive TCCs have been used to screen synthetic mimics of natural epitopes to identify the natural antigen sequences (44). GAD65 TCCs have also been used to screen peptide libraries to identify microbial epitopes derived from sources such as the human cytomegalovirus (hCMV) (45) and Neisseria meningitidis (36), which are cross-reactive with these autoreactive cells. Enterovirus infection (Coxsackievirus) has been implicated in the pathogenesis of type 1 diabetes due to sequence similarities between the virus and tyrosine phosphatase pIAR and heat shock protein 60 (HSP60), and humoral immune responses were documented in a small percentage of patients after infection with the virus (14). A role for molecular mimicry between GAD65-reactive TCCs and a homologous Coxsackievirus protein sequence was proposed a decade ago (13). However, cross-reactivity between GAD65-specific TCCs and Coxsackievirus sequences could not be detected in several studies conducted since that time (46). T-cell cross-reactivity between GAD65-specific T-cells and a peptide of the hCMV has been demonstrated (35), and clinical onset of type 1 diabetes accompanied by acute hCMV infection has been reported (47). As further evidence that molecular mimicry may be involved in the pathogenesis of type 1 diabetes, rotavirus infection in high-risk children caused a statistically significant association with a rise in islet autoantibodies for tyrosine phosphatase IA protein, insulin, and GAD65 (48). Infections can abrogate as well as enhance autoimmune diabetes (49,50). For example, spontaneous diabetes in NOD mice can be corrected by exposure to microbial and viral agents. Conversely, it has been well documented that infections can enhance autoimmune diabetes via molecular mimicry (11,13,48) or by bystander mechanisms (51). In this study, we have uncovered the footprint of common pathogens in the activation of diabetogenic T-cells. Perhaps these seemingly disparate observations can be reconciled by viewing the activation of this T-cell as an initiating event in the disease process, which ultimately requires further untoward events. In the initial phase, mimicry could supply the force for expansion, which may, due to immune diversion mechanisms, be protective. However, subsequent events, such as infection by pancreas-specific viruses, may later lead to target tissue damage, further immune expansion, and diversification of the islet response, leading to clinical disease. Therefore, our work uncovers new relationships between disease-causing T-cells and their environmental stimulants.
Several of the peptides that were stimulatory for BDC2.5 TCCs were also able to stimulate T-cells from young, pre-diabetic NOD mice, demonstrating that the antigen specificities of this clone are present in many different cells of the NOD T-cell repertoire. These results confirm previous evidence that different TCRs can share some but not all specificities with another T-cell population (5). The results from this study also suggest that within the NOD T-cell repertoire there exists a group of cross-reactive T-cell subsets that recognize peptides from proteins of pathogen origin. Therefore, the accessibility of autoreactive T-cells to bind peptides of pathogen origin (degeneracy of antigen recognition) could directly affect the frequency of the autoreactive T-cell pool. In other words, it is possible that pathogen peptides shape the peripheral autoreactive T-cell pool.
The identification of natural peptide sequences that are cross-reactive with an islet cell–specific TCC is fundamental to the understanding of the mechanisms initiating type 1 diabetes and may have clinical relevance in the development of early prediction assays and antigen-based intervention strategies.
The proliferative response of splenic T-cells from transgenic BDC2.5 mice to a group of bacterial (A) and viral (B) peptide sequences with predicted reactivity. The y-axis represents the SI (sample cpm/background cpm) following 3 days of culture with either 10 μg/ml (▪) or 1 μg/ml (□) of peptide. Data represent the average of three assays with a mean background of 727 cpm.
The proliferative response of splenic T-cells from transgenic BDC2.5 mice to a group of bacterial (A) and viral (B) peptide sequences with predicted reactivity. The y-axis represents the SI (sample cpm/background cpm) following 3 days of culture with either 10 μg/ml (▪) or 1 μg/ml (□) of peptide. Data represent the average of three assays with a mean background of 727 cpm.
Comparison of the proliferative activities of the most active bacterial (TPI 1308-24 through -44) and viral (TPI 1308-84 and 1136-7) peptides with BDC2.5 splenocytes. A threefold serial dilution scheme for the peptides was used, and the EC50 values were determined by curve fitting using GraphPad Prism software. TPI 1040-31 is a previously identified superagonist that does not correspond to a natural protein sequence.
Comparison of the proliferative activities of the most active bacterial (TPI 1308-24 through -44) and viral (TPI 1308-84 and 1136-7) peptides with BDC2.5 splenocytes. A threefold serial dilution scheme for the peptides was used, and the EC50 values were determined by curve fitting using GraphPad Prism software. TPI 1040-31 is a previously identified superagonist that does not correspond to a natural protein sequence.
BDC2.5 T-cell activation by selected peptides in the presence (□) or absence (▪) of an anti–MHC class II antibody, anti–H-2 Ig7. Peptides were also tested in the presence of mouse IgG2a (&;) as an isotype control. Bacterial peptide 1308-26 (A) was tested at 1 μg/ml, viral peptide 1136-7 (B) at 10 μg/ml, and superagonist 1040-31 (C) at 0.1 μg/ml. All antibodies were added at a 50-μg/ml final concentration per well. Data represent the average of two assays with a mean background of 50 cpm.
BDC2.5 T-cell activation by selected peptides in the presence (□) or absence (▪) of an anti–MHC class II antibody, anti–H-2 Ig7. Peptides were also tested in the presence of mouse IgG2a (&;) as an isotype control. Bacterial peptide 1308-26 (A) was tested at 1 μg/ml, viral peptide 1136-7 (B) at 10 μg/ml, and superagonist 1040-31 (C) at 0.1 μg/ml. All antibodies were added at a 50-μg/ml final concentration per well. Data represent the average of two assays with a mean background of 50 cpm.
The percentages of positive wells of NOD T-cells that respond to BDC2.5 stimulatory peptides after secondary proliferation are shown. The combined results from three separate proliferation assays totaling 36 wells were used. ▪, SI ≥2.0;&;, SI ≥3.0; □, SI ≥5.0.
The percentages of positive wells of NOD T-cells that respond to BDC2.5 stimulatory peptides after secondary proliferation are shown. The combined results from three separate proliferation assays totaling 36 wells were used. ▪, SI ≥2.0;&;, SI ≥3.0; □, SI ≥5.0.
The cross-reactivity of an NOD T-cell line specific for peptide 1308-84 (A) or 1308-29 (B) with selected peptides having known stimulatory activity for both BDC2.5 and NOD splenocytes. A 17-mer ovalbumin peptide sequence was run as a negative control. The y-axis represents the SI (sample cpm/background cpm) following 3 days of culture with either 25 (▪) or 5 (□) μg/ml peptide. Data shown are the average of four values. Data represent the average of two assays with a mean background of 304 (A) and 431 (B) cpm.
The cross-reactivity of an NOD T-cell line specific for peptide 1308-84 (A) or 1308-29 (B) with selected peptides having known stimulatory activity for both BDC2.5 and NOD splenocytes. A 17-mer ovalbumin peptide sequence was run as a negative control. The y-axis represents the SI (sample cpm/background cpm) following 3 days of culture with either 25 (▪) or 5 (□) μg/ml peptide. Data shown are the average of four values. Data represent the average of two assays with a mean background of 304 (A) and 431 (B) cpm.
Spontaneously arising 1308-84–and 1308-29–specific T-cells preferentially use TCR Vβ8.1/8.2 chain. The T-cell lines were stained with anti–CD4-peridinin chlorophyll-α protein and a selection of anti-TCR Vβ antibodies (x-axis) before flow-activated cell sorter analysis. The expression of the respective TCRs on CD4+ T-cells was analyzed 16 days after the initial round of in vitro stimulation (3 days after second restimulation). The percentage of Vβ usage appears on the y-axis. ▪, cell line p1308-84; □, cell line p1308-29.
Spontaneously arising 1308-84–and 1308-29–specific T-cells preferentially use TCR Vβ8.1/8.2 chain. The T-cell lines were stained with anti–CD4-peridinin chlorophyll-α protein and a selection of anti-TCR Vβ antibodies (x-axis) before flow-activated cell sorter analysis. The expression of the respective TCRs on CD4+ T-cells was analyzed 16 days after the initial round of in vitro stimulation (3 days after second restimulation). The percentage of Vβ usage appears on the y-axis. ▪, cell line p1308-84; □, cell line p1308-29.
BDC2.5 stimulatory activity of high-scoring decapeptide sequences
Peptide no. . | Sequence . | Score . | Species . | Protein . | SI . | . | |
---|---|---|---|---|---|---|---|
. | . | . | . | . | 10 μg/ml . | 1 μg/ml . | |
Bacterial | |||||||
1308-24 | Ac- L G V P M W V K M D -NH2 | 2522.0 | Pseudomonas species | Ferrodoxin reductase component | 42 | 10 | |
1308-25 | Ac- G D Q P L W L R M D -NH2 | 2465.8 | Escherichia coli K-12 | Bifunctional penicillin-binding protein 1C | 20 | 3 | |
1308-26 | Ac- L G V P M W S R M E -NH2 | 2425.1 | Burkholderia sp. Strain RP007 | Hydratase/aldolase PhnE | 72 | 79 | |
1308-29 | Ac- L V H P V W G R M H -NH2 | 2354.8 | Neisseria meningitidis | Putative phage virion protein | 56 | 24 | |
1308-32 | Ac- T I G P L W K G M K -NH2 | 2336.5 | Methanotroph species | Methane monooxygenase α-subunit | 37 | 2 | |
1308-33 | Ac- L S L P I W P E M E -NH2 | 2332.9 | Synechocystis species | Exopolyphosphatase gb | 57 | 20 | |
1308-34 | Ac- A I G P L W K G M K -NH2 | 2331.7 | Methanotroph species | Methane monooxygenase α-subunit | 34 | 2 | |
1308-36 | Ac- V G R P M W L A M D -NH2 | 2323.9 | Alcaligenes faecalis | Phenanthrene degradative gene cluster | 29 | 4 | |
1308-38 | Ac- K V P P L W I L M L -NH2 | 2320.6 | Synechocystis sp. PCC6803 | Polyphosphate kinase | 33 | 5 | |
1308-39 | Ac- E T R P L W K A M H -NH2 | 2319.8 | Campylobacter jejuni | Lipopolysaccharide biosynthesis protein wlaK | 18 | 1 | |
1308-43 | Ac- M I P P L W F K M M -NH2 | 2311.7 | Acinetobacter species | Terminal alkane hydroxylase | 10 | 1 | |
1308-44 | Ac- A I Q P A W V I M E -NH2 | 2301.3 | Herpetosiphon aurantiacus | Methyltransferase HgiDIM | 28 | 4 | |
1308-47 | Ac- S F V P L W A T M L -NH2 | 2296.3 | Mycobacterium tuberculosis | Hypothetical protein Rv0235c | 11 | 2 | |
1308-51 | Ac- R M S P F W Q R M F -NH2 | 2235.0 | Mycobacterium tuberculosis | Hypothetical protein Rv3629c | 26 | 1 | |
1308-55 | Ac- S L A P F W L R M Q -NH2 | 2174.4 | Streptomyces coelicolor A3(2) | Anthranilate synthase | 19 | 1 | |
1308-57 | Ac- L F S P F W G R M A -NH2 | 2165.6 | Bacillus firmus | msyB gene | 20 | 1 | |
1308-59 | Ac- L T G P D W L R M G -NH2 | 2141.8 | Escherichia coli | URF | 12 | 1 | |
1308-63 | Ac- S A T P N W M R M F -NH2 | 2121.0 | Synechocystis sp. PCC6803 | Cytochrome c oxidase subunit I | 15 | 1 | |
1308-64 | Ac- K T J P H W Y R M I -NH2 | 2119.2 | Escherichia coli | 49 kd protein | 29 | 1 | |
1308-66 | Ac- K P T P G W Q R M V -NH2 | 2065.5 | Mycobacterium tuberculosis | Probable oxidoreductase | 27 | 1 | |
Viral | |||||||
1308-72 | Ac- K E R P L W N E M V -NH2 | 2322.5 | Equine herpes virus type 1 | Glycoprotein 14 | 15 | 1 | |
1308-73 | Ac- M P K P L W D A M Q -NH2 | 2285.4 | Kaposi’s sarcoma-associated virus | Glycoprotein M | 32 | 1 | |
1136-7 | Ac- M T A P S W A R M E -NH2 | 2260.0 | Human herpes simplex virus type 1 | Tegument protein | 123 | 94 | |
1308-80 | Ac- M W L P V W V I M A -NH2 | 2233.9 | Mycobacteriophage 15 | Predicted 8.2Kd protein | 18 | 1 | |
1308-81 | Ac- D E H P L W R Q M L -NH2 | 2232.7 | Reovirus type 1 | Sigma-1 protein | 11 | 1 | |
1308-84 | Ac- A T L P A W I K M P -NH2 | 2195.0 | Sendai virus | C′ protein | 39 | 4 | |
1308-89 | Ac- Y K E P K W F V M E -NH2 | 2152.6 | Clover yellow vein virus | Polyprotein | 24 | 1 | |
1308-90 | Ac- S M M P Q W S Y M H -NH2 | 2151.3 | Porcine adenovirus type 3 | Hexon protein (virion component ii) | 13 | 1 | |
1308-91 | Ac- S M L P Q W S Y M H -NH2 | 2138.1 | Human adenovirus type 34 | Hexon protein | 28 | 1 |
Peptide no. . | Sequence . | Score . | Species . | Protein . | SI . | . | |
---|---|---|---|---|---|---|---|
. | . | . | . | . | 10 μg/ml . | 1 μg/ml . | |
Bacterial | |||||||
1308-24 | Ac- L G V P M W V K M D -NH2 | 2522.0 | Pseudomonas species | Ferrodoxin reductase component | 42 | 10 | |
1308-25 | Ac- G D Q P L W L R M D -NH2 | 2465.8 | Escherichia coli K-12 | Bifunctional penicillin-binding protein 1C | 20 | 3 | |
1308-26 | Ac- L G V P M W S R M E -NH2 | 2425.1 | Burkholderia sp. Strain RP007 | Hydratase/aldolase PhnE | 72 | 79 | |
1308-29 | Ac- L V H P V W G R M H -NH2 | 2354.8 | Neisseria meningitidis | Putative phage virion protein | 56 | 24 | |
1308-32 | Ac- T I G P L W K G M K -NH2 | 2336.5 | Methanotroph species | Methane monooxygenase α-subunit | 37 | 2 | |
1308-33 | Ac- L S L P I W P E M E -NH2 | 2332.9 | Synechocystis species | Exopolyphosphatase gb | 57 | 20 | |
1308-34 | Ac- A I G P L W K G M K -NH2 | 2331.7 | Methanotroph species | Methane monooxygenase α-subunit | 34 | 2 | |
1308-36 | Ac- V G R P M W L A M D -NH2 | 2323.9 | Alcaligenes faecalis | Phenanthrene degradative gene cluster | 29 | 4 | |
1308-38 | Ac- K V P P L W I L M L -NH2 | 2320.6 | Synechocystis sp. PCC6803 | Polyphosphate kinase | 33 | 5 | |
1308-39 | Ac- E T R P L W K A M H -NH2 | 2319.8 | Campylobacter jejuni | Lipopolysaccharide biosynthesis protein wlaK | 18 | 1 | |
1308-43 | Ac- M I P P L W F K M M -NH2 | 2311.7 | Acinetobacter species | Terminal alkane hydroxylase | 10 | 1 | |
1308-44 | Ac- A I Q P A W V I M E -NH2 | 2301.3 | Herpetosiphon aurantiacus | Methyltransferase HgiDIM | 28 | 4 | |
1308-47 | Ac- S F V P L W A T M L -NH2 | 2296.3 | Mycobacterium tuberculosis | Hypothetical protein Rv0235c | 11 | 2 | |
1308-51 | Ac- R M S P F W Q R M F -NH2 | 2235.0 | Mycobacterium tuberculosis | Hypothetical protein Rv3629c | 26 | 1 | |
1308-55 | Ac- S L A P F W L R M Q -NH2 | 2174.4 | Streptomyces coelicolor A3(2) | Anthranilate synthase | 19 | 1 | |
1308-57 | Ac- L F S P F W G R M A -NH2 | 2165.6 | Bacillus firmus | msyB gene | 20 | 1 | |
1308-59 | Ac- L T G P D W L R M G -NH2 | 2141.8 | Escherichia coli | URF | 12 | 1 | |
1308-63 | Ac- S A T P N W M R M F -NH2 | 2121.0 | Synechocystis sp. PCC6803 | Cytochrome c oxidase subunit I | 15 | 1 | |
1308-64 | Ac- K T J P H W Y R M I -NH2 | 2119.2 | Escherichia coli | 49 kd protein | 29 | 1 | |
1308-66 | Ac- K P T P G W Q R M V -NH2 | 2065.5 | Mycobacterium tuberculosis | Probable oxidoreductase | 27 | 1 | |
Viral | |||||||
1308-72 | Ac- K E R P L W N E M V -NH2 | 2322.5 | Equine herpes virus type 1 | Glycoprotein 14 | 15 | 1 | |
1308-73 | Ac- M P K P L W D A M Q -NH2 | 2285.4 | Kaposi’s sarcoma-associated virus | Glycoprotein M | 32 | 1 | |
1136-7 | Ac- M T A P S W A R M E -NH2 | 2260.0 | Human herpes simplex virus type 1 | Tegument protein | 123 | 94 | |
1308-80 | Ac- M W L P V W V I M A -NH2 | 2233.9 | Mycobacteriophage 15 | Predicted 8.2Kd protein | 18 | 1 | |
1308-81 | Ac- D E H P L W R Q M L -NH2 | 2232.7 | Reovirus type 1 | Sigma-1 protein | 11 | 1 | |
1308-84 | Ac- A T L P A W I K M P -NH2 | 2195.0 | Sendai virus | C′ protein | 39 | 4 | |
1308-89 | Ac- Y K E P K W F V M E -NH2 | 2152.6 | Clover yellow vein virus | Polyprotein | 24 | 1 | |
1308-90 | Ac- S M M P Q W S Y M H -NH2 | 2151.3 | Porcine adenovirus type 3 | Hexon protein (virion component ii) | 13 | 1 | |
1308-91 | Ac- S M L P Q W S Y M H -NH2 | 2138.1 | Human adenovirus type 34 | Hexon protein | 28 | 1 |
Accelerated disease in NOD/scid recipients of p1308-84–activated BDC2.5 T-cells
Transferred cells . | Islets . | . | . | . | Diabetes incidence . | |||
---|---|---|---|---|---|---|---|---|
. | No insulitis . | Peri-insulitis . | Insulitis . | Remnant* . | . | |||
Peptide activated | 0 | 0 | 0 | 100 | 100 | |||
Nonactivated | 36 | 25 | 24 | 24 | 0 |
Transferred cells . | Islets . | . | . | . | Diabetes incidence . | |||
---|---|---|---|---|---|---|---|---|
. | No insulitis . | Peri-insulitis . | Insulitis . | Remnant* . | . | |||
Peptide activated | 0 | 0 | 0 | 100 | 100 | |||
Nonactivated | 36 | 25 | 24 | 24 | 0 |
Data are percent.
Significantly different (two-tailed P = 0.0013).
V.A.J. and G.M.A. contributed equally to this work.
Article Information
This work was supported by Mixture Sciences, the Diabetes National Research Group (grant no. DNR0301Pini), and National Institutes of Health grants from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) and National Institute of Allergy and Infectious Diseases (NIAID) (to N.S.).
We thank J. Ostresh, A. Nefzi, and the chemistry group at Torrey Pines Institute for Molecular Studies for the preparation of the peptide compounds used in this study.