| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
,¶
* Division of Immunogenetics, Department of Neuroscience and Immunology, Kumamoto University Graduate School of Medical Sciences and
First Department of Internal Medicine and
Department of Pediatrics, Kumamoto University School of Medicine, Kumamoto, Japan;
Structural Biology Center, National Institute of Genetics, Mishima, Shizuoka, Japan;
¶ Biosystem Science, School of Advanced Science, Graduate University for Advanced Studies, Hayama, Kanagawa, Japan; and
|| Laboratory of Immunology, Heart Institute (Incor), University of Sao Paulo Medical School, Sao Paulo, Brazil
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In functional analysis using altered peptide ligands, it was reported that TCRs can discriminate not only between two peptides differing at a TCR contact residue but also between those differing at only a single MHC anchor residue, which does not significantly alter the binding affinity of a peptide to MHC (12, 13). In addition, it has been reported that certain TCR recognition is affected by amino acids adjacent or not adjacent to TCR contact residues or by amino acid combinations in antigenic peptides (14, 15, 16, 17). These observations indicate that a substitution at a certain residue would induce conformational changes of peptides and may affect other residues. In contrast, a quantitative strategy using combinatorial peptide libraries with a positional scanning format (PS-SCLs)3 and biometric score matrices dissected and predicted peptide mimicry ligands of a given cognate T cells (18, 19). However, the precise effect of successive combinations of residues in the antigenic peptide on recognition of certain TCR has heretofore not been clarified. In addition, there is no available technology that allows for systematic separation and identification of diverse T cell epitopes from a mixture of randomized peptide ligands.
Several groups, including ours, reported invariant chain (Ii)-based epitope-presenting vectors, in which class II-associated invariant chain peptide (CLIP, Ii89–101) was replaced by antigenic peptides (20, 21, 22, 23). Using this vector system, we reported a method to identify epitopes cross-recognized by autoreactive T cell clones from a library of randomized peptides (24).
In the current study, we used this vector to make libraries of the glutamic acid decarboxylase 65 epitope (GAD65115–127)-based degenerate peptides, where three successive residues within the epitope were randomized. GAD65 is one of the important islet Ags implicated in autoimmunity of the NOD mouse and Type I diabetes in humans. We used two T cell clones established from Japanese patients with Type I diabetes and restricted by disease susceptible HLA-DR53 and they responded to GAD65 protein (25). The epitope (GAD65115–127) used in this study was also reported to be immunodominant in studies using HLA-DR4 transgenic mice (26, 27). Epitopes stimulating GAD65-reactive CD4+ T cell clones were isolated from the series of epitope libraries. Recognition properties of these TCRs were intensively analyzed, and combinatorial effects of amino acid residues within antigenic peptide on recognition by TCRs were investigated. The information obtained by screening of this epitope expression library included the combinations of amino acid residues with TCR agonism that could not be predicted using panels of single amino acid substituted peptide analogs. Data acquisition of stimulatory TCR ligands combined with a pattern match search allowed for identification of self or microbe-derived peptides cross-recognized by CD4+ T cell clones autoreactive to GAD65.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Dodecamer peptide analogs with single-amino acid substitutions derived from GAD65116–127 were purchased from Chiron Mimotopes (Clayton South, Victoria, Australia), and 13-mer peptides were synthesized as described (28). Two human CD4+ T cell clones, SA32.5 and MK20.2, recognizing GAD65116–127 (NILLQYVVKSFD) in the context of HLA-DR53 molecules (DRA*0101 + DRB4*0103) susceptible to type I diabetes were used throughout the study (25). T cells were fed weekly with 50 U/ml human rIL-2 and the irradiated DR53-matched allogenic PBMCs prepulsed with the GAD65111–131 (LQDVMNILLQYVVKSFDRSTK) in RPMI 1640 supplemented with 10% heat-inactivated human plasma, 100 U/ml penicillin, 100 µg/ml streptomycin, and 2 mM L-glutamine. T cell proliferation assays were set up in 96-well flat-bottom culture plates (Falcon; BD Biosciences, San Jose, CA) with 3 x 104 T cells and irradiated (6000 cGy) DR53-positive 1.5 x 105 PBMCs or 1.0 x 104 dendritic cells (DCs) per well in the presence of peptides or recombinant proteins at various concentrations. After 48 h of culture, [3H]thymidine was added (1 µCi/well); and after an additional 16 h, cells were harvested onto glass fiber filters, and radioactivity was counted on a beta scintillation counter (Wallac, Gaithersburg, MD).
In vitro generation of DCs
DCs were generated from CD14+ monocytes
purified by positive immunoselection from HLA-DR53-positive allogeneic
PBMCs, using an anti-CD14 mAb coupled onto magnetic microbeads
(CD14 microbeads; Miltenyi Biotec, Auburn, CA). The
CD14+ monocytes were cultured at 1 x
106 cells/ml in the presence of 100 ng/ml GM-CSF
and 100 U/ml IL-4 (Ono Pharmaceutical, Osaka, Japan) in RPMI 1640
supplemented with 10% human plasma, 2 mM L-glutamine, and
100 µg/ml streptomycin. Cultures were fed on days 3 and 5 with fresh
medium containing GM-CSF and IL-4. On day 5, DCs were treated with
TNF-
(20 ng/ml). On day 7, the nonadherent cells were harvested and
served as mature DCs. For the assessment of HLA class II-restricted
presentation of recombinant microbial proteins, mature DCs were
cultured with recombinant proteins for 14 h before use as APCs for
proliferation assay.
Epitope expression library
The procedure for construction of an epitope-presenting library
is described in our previous report (24). Briefly,
oligonucleotide fragments encoding degenerate
GAD65115–127 were synthesized and purified using
polyacrylamide gel (Genemed Synthesis, South San Francisco, CA). These
oligonucleotide fragments were amplified by PCR with 5'-biotinylated
primers, 5'-TCC CTC CTG GTG ACT CTG CTC CTC-3' and 5'-ATT GTT ATC TGC
TGT TCC GAC TTG-3'. The purified PCR products were digested with
DraI and SacI, purified with
streptavidin-agarose, and ligated to
SmaI-SacI-digested pCI, the CLIP-substituted
epitope expression vector. The construct encodes Ii protein inserted
with partially degenerate fusion peptides based on
GAD65115–127 (MNILLQYVVKSFD) instead of
Ii89–101 (SKMRMATPLLMQA) within the CLIP
sequence. Escherichia coli (DH5) was transformed with the
ligation mixture, and the transformants were divided into 96-well
culture blocks (Qiagen, Studio City, CA) to generate transformant pools
of 30–50 clones and grown overnight in Luria-Bertani medium containing
ampicillin (100 µg/ml). The amplified plasmid DNA was purified using
a QIAprep 96 Turbo Miniprep system (Qiagen). The complexity of each
degenerate-GAD65115–127 expression sublibraries
was 
1.5 x 104–2.0 x
104.
Screening of epitope expression library by detecting IFN-
production
Library DNA pools and expression vectors for HLA-DRA*0101 and
DRB4*0103 were mixed with Transfectam reagent (Promega, Madison, WI) in
serum-free DMEM. The DNA-Transfectam mixtures were then added to the
COS-7 cells (1 x 104 cells/well) in 96-well
flat-bottom culture plates and incubated for 90 min at 37°C. After
removal of the transfection medium, the COS-7 cells were incubated
overnight in DMEM supplemented with 10% FCS. After 24 h, cells
were washed twice with DMEM, and T cells were added at a concentration
of 3 x 104/well in RPMI 1640 supplemented
with 10% heat-inactivated human plasma, 100 U/ml penicillin, 100
µg/ml streptomycin, and 2 mM L-glutamine. After 48 h
of incubation, the supernatant was collected, and IFN- concentration
was measured using a standard ELISA (Endogen, Woburn, MA). The library
of DNA pools, for which a significant production of IFN- was
detected was used to transform the bacteria to prepare a sublibrary of
DNA pools consisting of 10 clones. Secondary screening was done as
described above, using sublibraries. Single plasmid clones were
obtained after three rounds of screening. DNA sequences of the purified
plasmid clones were analyzed using BigDye Terminator Cycle Sequencing
Ready Reaction Kits and the ABI PRISM 310 Genetic Analyzer (Applied
Biosystems, Foster City, CA). For construction of expression vectors
encoding mimicry peptides or analogs of GAD65 epitope, both strands of
oligonucleotide fragments encoding these peptides were synthesized
(Espec Oligo Service, Tsukuba, Japan), annealed, and ligated to
SmaI-SacI-digested pCI. The amplified plasmid
DNAs were purified for transfection. The agonistic activity was
assessed in an IFN- secretion assay at various dilutions with pCI
(wild Ii: irrelevant DNA).
Analysis of TCR V-(D)-J junctional regions of GAD65-autoreactive T cell clones
Total RNA was extracted from T cells using the TRIzol reagent
(Life Technologies, Gaithersburg, MD), and first-strand cDNA was
synthesized using Superscript RNase H- reverse
transcriptase (Life Technologies) and random hexamers. cDNA was
subjected to PCR amplification for rearranged TCR- with 29 5'-TCRAV
family-specific oligonucleotides (V1–32) and a
3'-TCRAC (C) constant primer, and for rearranged TCR-
with 27
5'-TCRBV family-specific oligonucleotides (V1–25) and a
3'-TCRBC (C) constant primer (29). The amplified PCR
products of the -chain and of the -chain were cloned into a
plasmid vector, pGEM-T (Promega), and sequenced. The resulting
sequences were analyzed using IMGT, the international ImMunoGeneTics
database (http://imgt.cnusc.fr: 8104/).
Northern blot analysis
Northern blot analysis was conducted as described
(30). To prepare the probes, the MK20.2 cDNA was subjected
to PCR amplification for 5'-TCRV15- or 5'-TCRV16-specific
oligonucleotide and 3'-TCRJ region oligonucleotide primers, and each
PCR product was TA cloned, digested, and gel purified. The two cDNA
fragments of the TCRAV gene were labeled with
[-32P]dCTP. After hybridization and
quantitative analysis of signal intensities, probes were stripped, and
a second hybridization was conducted using an -actin probe. To
assess cross-reactivity of the V-specific probes, TCRAV cDNA
fragments were arrayed onto two copies of nylon membrane filters and
hybridized using TCRAV cDNA probes.
Generation of recombinant proteins
Genomic DNA of Legionella pneumophila (strain Philadelphia-1) was kindly provided by Drs. T. Akaike and T. Akuta (Kumamoto University School of Medicine, Kumamoto, Japan). Lactococcus lactis (subsp. lactis) was provided by the Institute of Physical and Chemical Research (Wako, Japan). Bacterial genomic DNA were purified using DNeasy Tissue Kits (Qiagen). The O-succinylbenzoic acid-CoA ligase gene (coding for aa 101–201)/L. lactis, putative PTS system, lactose-specific component IIBC gene (coding for aa 426–546)/Streptococcus pyogenes (ATCC 19615), putative dihydrolipoamide dehydrogenase gene (coding for aa 101–205)/Neisseria meningitidis, glutamine amidotransferase, class I gene (coding for aa 1–104)/Streptococcus pneumoniae (ATCC 49619), and the pilus assembly protein PilB gene (coding for aa 28–131)/Legionella pneumophila were PCR amplified and cloned into the plasmid vector (pGEM-T-easy vector system). Fusion proteins containing relatively small fragment (100–120 aa) of microbial proteins were generated, because larger recombinant proteins tend to become insoluble in bacteria and are difficult to purify. The inserted fragments were digested and ligated directionally into the prokaryotic expression vector pGEX-4T (Pharmacia, Peapack, NJ) to produce GST-fusion protein. The integrity of the constructs was confirmed by DNA sequencing. The procedure for protein induction and purification were described in our previous report (31). The purity and integrity of the fusion protein were confirmed by SDS-PAGE. The recombinant proteins were concentrated and separated from small peptide fragments with Centricon-30 (Amicon, MA), and the buffer was replaced with culture medium.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The goal of this study was to develop new and comprehensive methods for analysis of the combinatorial effects of residues in antigenic peptides on recognition by HLA class II-restricted TCRs. We also tried to clarify the significance of the effect of amino acid combination within antigenic peptide in recognition by TCR. By adopting the strategy, we sought to identify candidate mimicry epitopes for GAD65-autoreactive T cell clones established from type I diabetes patients.
Specificity of GAD65-reactive T cell clones SA32.5 and MK20.2 analyzed using single-amino acid residue-substituted peptide analogs
To investigate structural features of peptides cross-recognized by
autoantigen-specific TCRs, we analyzed HLA-DR53-restricted SA32.5 and
MK20.2 CD4+ T cell clones reactive to GAD65
(peptide 116–127). These T cell clones were established from two
independent type I diabetes patients in our previous study
(25). At the beginning of this study, we verified that the
two T cell clones expressed single TCRs, because it has been reported
that a significant fraction of T cells in human peripheral blood
expresses dual TCRs (32). The complementarity-determining
region (CDR) sequences of the Ag contact sites were defined. As shown
in Table I, TCR- and TCR--chains of
SA32.5 T cell clone revealed functional TCRAV1S2 and
TCRBV9S1 gene rearrangements as well as an out of frame
rearrangement of AV27S1. TCR - and TCR -chains of
MK20.2 TCR revealed an in frame dual V rearrangement
(TCRAV15S1 and TCRAV16S1 transcripts) and
TCRBV3S1 rearrangement. In Northern blot analysis (Fig. 1A), T cell clone MK20.2
expressed the TCRAV16S1 but not the TCRAV15S1
gene transcript. Therefore, TCRAV16S1 combined with the
TCRBV3S1 gene product mediates Ag recognition in MK20.2 TCR.
These data indicated that reactivities of these T cell clones were
determined by each single TCR. These T cell clones use distinct CDR3
and CDR3 sequences at the amino acid level and a distinct CDR3
length, whereas both T cell clones use the same J rearrangement.
|
|
|
We constructed a set of T cell epitope expression libraries by
using an Ii-based epitope presenting plasmid vector, pCI, in which the
CLIP region of Ii was substituted with MHC class II-restricted epitopes
(Fig. 3A). In the libraries,
sequences of peptides are derived from the
GAD65115–127, and three successive residues
within the sequence were totally randomized, the theoretical maximum
complexity of each library being 203 (Fig. 3B). We defined the I117 of
GAD65115–127 as the relative position 1 in this
study. As shown in Fig. 3B, the library set (series CIR) is
composed of seven sublibraries: CIR-1–2; CIR2–4; CIR3–5; CIR4–6;
CIR5–7; CIR6–8; and CIR7–9. (CIR-1–2 represents a library in which
three successive residues from a relative position -1 to 2 in the T
cell epitope GAD65115–127 are replaced by
randomized amino acids.) The positions of inserted randomized amino
acids are serially overlapped between individual sublibraries covering
the core epitope GAD65116–125. Individual
libraries contained 14,400–19,200 DNA clones and were divided into
subpools composed of 30–50 clones. Summation of all the complexity of
peptides is estimated to be at least 120,000 species.
|
The epitope recognition by the two
GAD65116–127-reactive TCRs was further
investigated using the T cell epitope expression system to identify the
diverse peptide ligands of these T cell clones. The two T cell clones
were examined in parallel with their production of IFN- in response
to a set of T cell epitope expression libraries. Fig. 4 shows the frequency of pools
stimulating T cell clones in the epitope libraries. The vector DNA
encoding native GAD65115–127 mixed with a 100x
excess amount of the wild-type Ii gene was assigned an arbitrary
activity of 1 U, leading to determination of the relative activity of
each separated pool. Thus, these TCRs responded to pools of each
epitope library at different frequencies. MK20.2 responded to many of
the pools of CIR-1–2 (99.7%) and CIR2–4 (66.0%) libraries with
>0.5 U of response. In contrast, this TCR responded to a few of the
pools in relatively C terminus-randomized libraries (CIR3–5 to CIR7–9
libraries) (<7%). These results suggest that the highest specificity
of MK20.2 TCR exists at the relatively C-terminal side of
GAD65115–127. In contrast, SA32.5 showed a
broader response profile against several libraries spanning the epitope
functional core. SA32.5 TCR showed a number of strong responses with
library CIR6–8 (73.4%). These results collectively indicate that the
spectrum of fine specificity in TCR recognition was clearly different
between these two TCRs.
|
The T cell epitope expression library was used to identify a
series of agonistic peptide ligands for these TCRs. We screened 2,796
pools (total clone size was estimated to be 120,000); and after 3
rounds of screening, possible stimulatory peptides with higher
antigenicity for SA32.5 (85 ligands) and MK20.2 (63 ligands) TCRs,
respectively, were identified. The wild-type
GAD65115–127 sequence was isolated from several
libraries but not from library CIR-1–2, CIR3–5, and CIR6–8. For
some peptide sequences identified, the same sequences were isolated
more than twice; and in 70% of such cases, peptides were encoded by
distinct nucleotide sequences. The sequences of stimulatory peptides
and their stimulatory activity are shown in Fig. 5. MK20.2 TCR responded to most of the
pools (99.7%) of library CIR-1–2 in the first round of screening
(Fig. 4). Therefore, agonistic peptide ligands for MK20.2 TCR were not
isolated from this library, and sequences isolated from library
CIR-1–2 as stimulatory ligands for SA32.5 TCR were tested for their
capacity to stimulate MK20.2 TCR (Fig. 5A). With regard to
library CIR2–4 for SA32.5 TCR (14.8%) and MK20.2 TCR (66.0%) and the
library CIR6–8 for SA32.5 TCR (73.4%), stimulatory ligands were not
exhaustively isolated because of the high frequency of positive
pools.
|
The panel of stimulatory ligands isolated from library CIR-1–2
(Fig. 5A) shows that both SA32.5 and MK20.2 tolerated
phenylalanine, methionine, leucine, isoleucine, valine, and cysteine at
relative position 1 (isoleucine in the native ligand). These amino
acids were compatible with the HLA-DR primary anchor residue that was
restricted by the Val/Gly dimorphism at DR86 where valine is used in
the HLA-DR53 molecule (33, 34). For SA32.5 TCR, relative
position 2 showed a strong preference for asparagine and leucine.
MK20.2 TCR tolerated most of the agonistic sequences identified with
SA32.5 TCR with different stimulatory capacities. However, MK20.2 TCR
did not respond to peptides that incorporated asparagine at relative
position 2.
Analysis of stimulatory ligands for SA32.5 TCR: CIR2–4 to CIR7–9 libraries
SA32.5 TCR had a preference for leucine and asparagine at relative position 2 in screening of CIR-1–2 (Fig. 5A) and CIR2–4 (Fig. 5B), and methionine and histidine were also tolerated in CIR2–4. This TCR showed a strong preference for proline with an imino ring at relative position 3 in the screening of two separate libraries, CIR2–4 and CIR3–5. At this position, isoleucine sharing a chemically conservative side chain with native leucine was also tolerated. At relative position 4, leucine, methionine, valine, and glutamine were tolerated almost equally in CIR2–4 and CIR3–5 libraries. In CIR3–5, isoleucine was also tolerated; however, glutamine at relative position 4, the same amino acid as in the native ligand, was the most frequently observed (7 of 14) in CIR4–6. At this position, histidine, glutamic acid, and serine were also tolerated in CIR4–6. At relative position 5, located at the center of the epitope, this TCR did not tolerate chemically conservative single amino acid substitutions except for tryptophan in the analysis peptide analogs with single-amino acid residue substitution (Fig. 2). However, ligands containing tyrosine, phenylalanine, and tryptophan, which share an aromatic ring, and histidine with an imidazole ring at this position were isolated from the two libraries, CIR3–5 and CIR4–6. Neighboring residues may compensate for conformational mimicking.
Tryptophan at relative position 5 was the most frequently observed amino acid (8 of 10) in stimulatory ligands isolated from CIR5–7. At relative position 6, this TCR has a preference for positively charged arginine and histidine in CIR4–6, CIR5–7, and CIR6–8. However, glutamic acid, serine, threonine, and valine were also tolerated in CIR4–6. In addition, serine, glutamine, valine, and glycine were also tolerated in CIR5–7. Similarly, glutamine and valine were also tolerated in CIR6–8. At relative position 7, this TCR showed a preference for hydrophobic residues (methionine, leucine, valine, and proline) in CIR5–7 and (leucine, methionine, valine, and isoleucine) in CIR6–8. This preference in chemical character is not significantly changed in these two libraries. However, there is a drastic change in the preference in CIR7–9 at this position in which the 3-mer randomized portion was moved to the C-terminal side only by one amino acid from CIR6–8. This TCR has a preference for positively charged arginine at relative position 7. In addition, threonine and arginine in CIR6–8 and methionine and valine in CIR7–9 were also permitted at this position. At relative position 8, several amino acids with different chemical properties (arginine, proline, serine, isoleucine, methionine, leucine, valine, alanine, and glycine) were tolerated almost equally in CIR6–8. On the contrary, similar drastic changes in preference observed at position 7 were also observed at position 8 in CIR7–9, and this TCR has a strong preference for proline at this position. Several amino acids with different chemical properties were tolerated at relative position 9.
Epitope expression library revealed the importance of linear combinations of residues in recognition by SA32.5 TCR
By observing the preference of residues in peptides in recognition by SA32.5 TCR, preferable residues at each position in the N-terminal side were relatively similar in each separated library. However, in the peptide C-terminal side (Fig. 5B, CIR5–7, CIR6–8, and CIR7–9), there is a different preference at each position depending on which successive residues were randomized. It is conceivable that this phenomenon is affected by fixed residues flanking the randomized regions. Amino acids at relative position 6 in library CIR6–8 are not fixed, so that this position tolerates several amino acids with different chemical properties. In the isolated ligands from CIR6–8, if positively charged amino acids (arginine and histidine) located at relative position 6, hydrophobic amino acids (leucine, methionine, isoleucine, and valine) were followed at relative position 7. Subsequently, arginine, serine, or hydrophobic and aliphatic amino acids were followed at relative position 8 (see clones 6.8.S1–S11 in Fig. 5B). At the same time, it shapes the most dominant motif in CIR6–8, and 11 of 15 agonistic clones (73%) were positive for this motif. Also if glutamine located at relative position 6, neutral or hydrophobic amino acids were followed at relative position 7, and aliphatic amino acids were followed at relative position 8 successively (6.8.S12–S14). In the case of valine located at relative position 6, arginine at relative position 7 and proline at relative position 8 were followed successively (6.8.S15 and 7.9.S1–S8). Valine at relative position 6 was fixed in the CIR7–9 library, so that it preferentially made up the most dominant V-R-P (relative position 6–8) motif (62%) and excluded other motifs. If methionine was located at relative position 7, proline or arginine followed at relative position 8 (7.9.S10 and -S11). In case of valine at relative position 7, positively charged arginine or lysine followed at relative position 8 (7.9.S12 and -S13). These observations indicate that fixed residues sometimes limit adjacent residues and exclude other recognition motifs. In this way, peptide libraries with three successive randomized residues and their overlapping system yielded different information related to TCR recognition motifs depending on which successive ones are randomized.
In summary, SA32.5 exhibited a preference for proline at relative position 3 of peptides in screening of two separate libraries; CIR2–4 and CIR3–5. This residue does not correspond to the native sequence. In addition, this TCR tolerated phenylalanine, tryptophan, and histidine with similar structural side chains at relative position 5 when combined with some successive residues different from native ones. In the C-terminal side of peptide, three overlapping libraries, CIR5–7, CIR6–8, and CIR7–9, presented distinct recognition profiles. In this case, linear combinations of residues formed by the chemical properties of each were significantly influenced. With these promiscuous interactions, all of the TCR contacts may be altered by substitutions of other residues.
Analysis of stimulatory ligands for MK20.2 TCR: CIR2–4 to CIR7–9 libraries
MK20.2 did not respond to most of the agonistic sequences among many peptides identified with SA32.5 (Fig. 5B). At relative position 2, MK20.2 TCR has a preference for methionine, leucine, and tyrosine with hydrophobic side chains and for histidine and arginine with positively charged side chains in CIR2.4 (Fig. 5C). It seems that this position tolerates bulky side chains, because there is no relationship among chemical properties of the amino acid side chains identified. At relative position 3, this TCR indicates a strong preference for methionine and glutamine in two overlapped libraries, CIR2–4 and CIR3–5, that do not correspond to the native sequence. In addition, isoleucine, cysteine, and threonine in CIR2–4 and leucine and isoleucine in CIR3–5 are also tolerated. At relative position 4, this TCR has a preference for leucine, methionine, valine, glutamine, and serine in two overlapping libraries (CIR2–4 and CIR3–5). In addition, phenylalanine in CIR2–4 and alanine and threonine in CIR3–5 were also tolerated. In contrast, with the screening of CIR4–6, this TCR has a preference for glutamine and serine at this position, and leucine and alanine were also tolerated at this position. At relative position 5, this TCR showed almost exclusive specificity for tyrosine in three consecutive libraries, CIR3–5, CIR4–6, and CIR5–7, which is the same amino acid as the native sequence. In addition, when serine is at relative position 6, tryptophan was tolerated at position 5. At relative position 6, several amino acids with different chemical properties were tolerated. In CIR4–6, MK20.2 TCR tolerated serine, threonine, arginine, valine, and isoleucine. In CIR5–7 and CIR6–8, small amino acids such as alanine, valine, glycine, and serine were preferred. At relative position 7, valine, serine, threonine, arginine, and aspartic acid in CIR5–7 and valine, cysteine, serine, and alanine, in CIR6–8 were permitted. However, this TCR has a preference for alanine, serine, asparagine, and aspartic acid, at this position in CIR7–9. At relative position 8, MK20.2 TCR was highly specific to lysine in two overlapping libraries, CIR6–8 and CIR7–9, as in the native GAD65115–127 sequence. Several amino acids with different chemical properties were tolerated at relative position 9. SA32.5 responded to a limited fraction of the agonistic sequences identified with MK20.2.
In summary, MK20.2 TCR was specific for tyrosine at relative position 5 and lysine at relative position 8 of the antigenic peptide, as judged by screening data from most of the separate libraries. However, several combinations of other residues were observed depending on which successive residues were randomized.
Combinatorial effects of multiple residues in exhibition of antigenicity
Several investigators reported that single amino acid
modifications in a given peptide exert positive or negative effects
when combined in one peptide species containing multiple substitutions
(16, 17, 35). These phenomena were also observed in a
series of agonistic sequences we identified using the CLIP-substituted
Ii library, as shown in Fig. 6. As shown
in Fig. 6, A and B, SA32.5 shows decreased or
totally abrogated T cell reactivity against Y5W (standing for a peptide
analog having Y to W substitution at the relative position 5), Y5F,
Y5H, V7R, V7 M, K8P, and S9T. However, when these residues in each
position were combined with other adjacent residues, these peptides had
a completely restored T cell reactivity. For instance, when QWH (4.6.S6
clone in Fig. 5; substituted residues are underlined), QFR (4.6.S4),
MHH (4.6.S9), at relative positions 4–6, and RPT (7.9.S5), and MPG
(7.9.S10), at relative positions 7–9, were introduced into the peptide
in combination, SA32.5 exhibited marked IFN- production and
proliferation. Although the MK20.2 T cell clone exhibited strong
specificity for lysine at relative position 8 (Fig. 5C),
combinatorial effects were also observed for this TCR (Fig. 6D). For example, substitution of a single residue S9D
decreased T cell reactivity but in combination with asparagine (7.9.M9)
or serine (7.9.M5) at relative position 7 in combination exhibited an
increased response. These data indicate that this strategy allows one
to identify linear combinations of residues in antigenic peptide
triggering TCRs.
|
We asked whether the combinatorial effects we found in the analysis
using the CLIP-substituted vector system were reproducible when
epitopes were added as synthetic peptides. We synthesized 13-mer
peptides and tested then for proliferative response of T cell clones to
the peptides at various doses. As shown in Fig. 6, the results of
proliferation assay using synthetic peptides were almost in parallel
with those obtained in experiments using epitope presenting vector and
quantification of IFN- produced by the T cell clones (Fig. 6).
Therefore, we could verify the combinatorial effect also in the
experiments using synthetic peptides. Thereafter, 13-mer peptides
incorporating three successive amino acids which have been proven to
stimulate T cell response. MYFNIQYQTVGFD for SA32.5 and MPVRMLYVSKGFD
for MK20.2, were synthesized. As shown in the right panels of
Fig. 6, C and E, these peptides with no
resemblance to GAD65115–127 activated T cell
clones, respectively, at lower concentrations than did the native
sequence, thereby indicating that simultaneous multiple residue
modification deduced from the results obtained using epitope expression
cloning strategy could generate superagonists.
These data strongly suggest that linear combinations of residues on antigenic peptides affect recognition by TCRs. Therefore, amino acid combinatorial effects should be considered when searching for cross-reactive epitopes. We also suggest that T cell epitope expression cloning can provide the strategy for efficient identification of optimal sequences.
Identification of microbial and self mimics for GAD65115–127-autoreactive TCRs
The results obtained from T cell epitope expression library were used to establish recognition motifs for SA32.5 and MK20.2 TCRs. As shown in Fig. 7, amino acids with similarities in chemical characteristics and with successive stimulatory combination with residues obtained from library screening that may induce mimicry were considered to provide search criteria for peptide mimics of GAD65115–127. As for the SA32.5 TCR, motifs consisting of amino acids leucine, isoleucine, and proline at relative position 3, and tyrosine, tryptophan, and phenylalanine at relative position 5 were considered because these residues were critical for recognition by SA32.5 TCR (Fig. 5). With regard to the peptide C-terminal side (relative positions 6–8), three kinds of search criteria were considered (Fig. 7A), because a successive combination of residues in the antigenic peptide may significantly affect the recognition by SA32.5 TCR. As for MK20.2 TCR, amino acids methionine, glutamine, isoleucine, and leucine at relative position 3 with strong preference were considered. Amino acids tyrosine at relative position 5 and lysine at relative position 8 were fixed on the basis of exquisite specificity of MK20.2 TCR (Fig. 7B).
|
by T cell clones using
COS-7 cells expressing the CLIP-substituted Ii and HLA-DR53. Among the
47 candidates selected, SA32.5 responded to 7 candidates, L.
lactis, S. pyogenes, N. meningitidis,
Chlamydia pneumoniae, Homo
sapiens/claudin-17, tafazzin, and tafazzin-like protein-derived
peptides. T cell clone MK20.2 responded to five candidates,
Legionella pneumophila, S. pneumoniae,
Staphylococcus aureus, Rickettsia prowazekii, and
human herpesvirus 6-derived peptides.
|
|
To investigate whether these microbial peptide mimics identified by our library scanning can be naturally processed and presented, five recombinant proteins carrying mimicry epitopes were generated as GST-fusion proteins and tested for their capacity to induce proliferative responses of these T cell clones. As APCs, we used DCs generated from CD14+ monocytes. As shown in Fig. 9, T cell clone SA32.5 responded to fusion proteins of putative dihydrolipoamide dehydrogenase (101–205) derived from N. meningitidis and O-succinylbenzoic acid-CoA ligase (101–201) derived from L. lactis. T cell clone MK20.2 responded to a fusion protein of glutamine amidotransferase, class I (1–104) derived from S. pneumoniae. However, fusion proteins derived from S. pyogenes and Legionella pneumophila did not activate either T cell clone. Furthermore, these responses were markedly inhibited by HLA-DR mAb L243 (data not shown). Cross-reactivity of the SA32.5 T cell clone to fusion proteins carrying mimicry epitopes for MK20.2 were not observed, and vice versa. These data suggest that some of these microbial peptide mimics identified by our strategy can be naturally processed in DCs and presented to GAD65-specific CD4+ T cell clones.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because of the importance of CD4+ T cells in autoimmunity, much effort has been directed toward identifying cross-reactive epitopes of microbial Ags recognized by autoreactive CD4+ T cells. Thus far, cross-reactive epitopes have been predicted and identified by primary sequence homology, the data obtained from single-residue-substituted peptide analogs, or PS-SCLs (2, 7, 8, 36, 37). The majority of these approaches using synthetic peptides have been fundamentally based on the concept that Ag recognition surface of TCRs is relatively flat and that each amino acid on each position of the peptide independently contributes to recognition by TCR (7, 38). On the basis of this assumption, mimicry epitopes have been searched in protein databases using pattern match searching with information on suitable amino acids for each position of the peptide. The use of PS-SCLs for analyses of T cell recognition has yielded the many useful concepts. However, some T cell clones did not respond to PS-SCLs (39) or responded to PS-SCLs ambiguously (40). In addition, artificial peptides composed of the optimal residue for each position selected based on analysis with PS-SCLs do not necessarily show agonistic activity, as we previously described (39). The major drawback of this method is that it identifies only the relative importance of each amino acid for each position of the peptide, and one could not directly analyze the agonistic activity induced by combinations of residues in antigenic peptides.
To address these problems, we used a totally different approach. The novelty of this system is that it enables one to directly identify sequences of T cell epitopes from plasmid-based epitope libraries composed of thousands of randomized sequences. Based on the accumulated information on sequences of agonistic peptides, one can determine the combinations of residues with agonistic properties. We used two different GAD65-specific T cell clones, SA32.5 and MK20.2, expressing distinct TCR but recognizing the same epitope with the same restriction element (25). At first, we verified that two T cell clones expressed single TCR, then we identified epitopes agonistic to each of these two TCRs from the epitope libraries and compared their sequences.
It is clear that TCRs of SA32.5 and MK20.2 represent distinct cross-reactivity and different recognition profiles (Fig. 5, B and C). It was observed that SA32.5 TCR tolerates structurally related amino acids at position 5 only when combined with specific adjacent residues (Figs. 5B and 6B). The patterns of the recognition profiles significantly differ depending on where randomized residues had been inserted. These findings were observed mainly on the peptide C-terminal side (Figs. 5B and 6B). Also, overall combinations of residues in the antigenic peptide affect the recognition by SA32.5 TCR more significantly than that by MK20.2 TCR (Figs. 5B and 6, A–C). Notably SA32.5 TCR permits the exchange of residues at all of the positions, as reported (41). In contrast, MK20.2 TCR permits specifically tyrosine and lysine for relative positions 5 and 8, respectively, even when residues around them are randomly exchanged (Fig. 5C). These residues are also conserved in the original GAD65115–127. However, even in the case of this TCR, significant amino acid combinatorial effects were observed in some positions (Figs. 5B and 6D).
As shown in Table III, MK20.2 had the
same V3.1 usage as did HA1.7, and it was reported that V3.1 was
predominantly expressed in HA306–318-specific
TCRs which preferred P8K of HA306–318
(42). Molecular modeling (Fig. 10A) of the fit of MK20.2
TCR to the recent crystal structure of HA1.7 TCR-HA peptide/HLA-DR1
complex (42) was done. As shown in Fig. 10, the model
predicted that the D28 and E30 of the CDR1 loop of TCRV3.1
made a charged interaction with lysine at position 8 of HLA-DR53-bound
GAD65115–127 peptide (Fig. 10B), thus
implying the exclusive specificity for lysine at relative position 8.
In contrast with BV3S1, BV9S1 of SA32.5 TCR uses the small amino acid
glycine in 28 and the negatively charged amino acid aspartic acid in
30 (Table III), which would weaken electrostatic interactions toward
lysine at position 8. This observation seems consistent with the
finding that SA32.5 TCR did not have a strong preference for lysine at
position 8 (Fig. 5B).
|
|
86, which commonly
accommodates aliphatic residues, it permits lysine and tyrosine
(45). These results are consistent with our finding that
relative position 1 permits several residues with different chemical
properties (Fig. 1). However, the peptide sequences isolated from the
epitope expression library indicate that relative position 1 preferred
aliphatic residues. A more comprehensive scan may yield the information
of this issue. In celiac disease, the disease-susceptible HLA-DR53
molecule selectively binds a large number of gliadin-derived peptides
that are extremely rich in glutamine and proline (46).
Accordingly, it was suggested that one or more peptide binding pockets
of HLA-DR53 molecules preferentially engaged the side chain of
glutamine or proline. In the TCR recognition motif of the T cell clone
SA32.5, proline at relative position 3 and glutamine at relative
positions 4 and 6 are the most important residues for TCR agonism (Fig. 5B). Although the two TCRs represented a distinct
specificity at relative positions 3, 5, 7, or 8 of the peptides, SA32.5
TCR and MK20.2 TCR represented similar preferences for certain amino
acids at some positions. For example, at relative position 1, both TCR
tolerate phenylalanine, methionine, leucine, isoleucine, valine, and
cysteine; at relative position 4 they tolerate leucine, methionine,
glutamine, and serine; and at the relative position 6 neutral or
positively charged residues and small aliphatic amino acids are allowed
(Fig. 5). These residues may be HLA-DR53 anchor residues; this remains
to be elucidated. It is conceivable that findings observed in this
study also reflect peptide-binding characteristics in which anchor
combinations are important for HLA binding, as described
(47). Not only requisition for direct TCR contact but also
constraints of HLA-DR53 binding could explain the combinatorial effect
observed on the peptide C-terminal side, i.e. a positively charged
amino acid is required at positions 122–124 in combination with a
bulky amino acid at position 121 for the recognition by SA32.5
TCR. Several groups of investigators have reported epidemiologic correlations between infection and type I diabetes (48, 49). Among the cross-reactive ligands identified in this study, epitopes derived from N. meningitidis, S. pneumoniae, and L. lactis can actually be produced from proteins by APCs to stimulate the T cells (Fig. 9). N. meningitidis is a Gram-negative and pathogenic bacterium. The outcome of meningococcal infection ranges from asymptomatic carriage to meningitis and fulminant meningococcemia in children and young people (50). S. pneumoniae is a Gram-positive and pathogenic bacterium, which causes bacterial sepsis, pneumonia, meningitis, and otitis media in young children (51). N. meningitidis and S. pneumoniae, which are also human respiratory commensals, in common can spread in the bloodstream under an immunocompromised state, etc. It is conceivable that these bacteria normally residing as harmless commensals may spread and participate in priming of autoreactive T cells and increase memory pools in the periphery. Also, asymptomatic carriages may predispose genetically susceptible children and initiate the autoimmune process. L. lactis is a nonpathogenic Gram-positive bacterium, used to produce fermented foods and dairy goods. When taken orally with food, this bacterium is metabolically active in all compartments of the intestinal tract (52). It is presumed to be taken up by M cells in intestinal lymphoid organs and to sensitize T cells. Although its relationship to autoimmune process has yet to be reported, it is intriguing in consideration of a recent report that short term breast-feeding and the early introduction of cow’s milk-based infant formula can predispose genetically susceptible children to type I diabetes (53).
In conclusion, we established a novel system to analyze the combinatorial effects of residues in the antigenic peptide on recognition by HLA class II-restricted TCR. The degree of combinatorial effects differs depending on part of the antigenic peptide and structure of TCRs, even when the epitope and restriction molecule are identical. Importantly, the defined TCR recognition motif incorporating combinatorial effects proved useful in identifying mimicry epitopes of autoreactive TCRs. The findings demonstrate the importance of combinatorial effects of each amino acid residue in the antigenic peptide on TCR recognition and propose a new direction for examining cross-reactive epitopes of TCR in investigating autoimmunity.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
2 Address correspondence and reprint requests to Dr. Yasuharu Nishimura, Division of Immunogenetics, Department of Neuroscience and Immunology, Kumamoto University Graduate School of Medical Sciences, Honjo 2-2-1, Kumamoto 860-0811, Japan. E-mail address: mxnishim{at}gpo.kumamoto-u.ac.jp
3 Abbreviations used in this paper: PS-SCLs, positional scanning synthetic combinatorial peptide libraries; CLIP, class II-associated invariant chain peptide; GAD, glutamic acid decarboxylase; Ii, invariant chain; DC, dendritic cell; CDR, complementarity-determining region.
Received for publication June 20, 2002. Accepted for publication November 13, 2002.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
chain to mature T cell apoptosis. J. Exp. Med. 183:2109.
