Progress has recently been made in the use of synthetic peptide libraries for the identification of T cell-stimulating ligands. T cell epitopes identified from synthetic libraries are mimics of natural epitopes. Here we show how the mimicry epitopes obtained from synthetic peptide libraries enable unambiguous identification of natural T cell Ags. Synthetic peptide libraries were screened with Mycobacterium tuberculosis-reactive and -autoreactive T cell clones. In two cases, database homology searches with mimicry epitopes isolated from a dedicated synthetic peptide library allowed immediate identification of the natural antigenic protein. In two other cases, an amino acid pattern that reflected the epitope requirements of the T cell was determined by substitution and omission mixture analysis. Subsequently, the natural Ag was identified from databases using this refined pattern. This approach opens new perspectives for rapid and reliable Ag definition, representing a feasible alternative to the biochemical and genetic approaches described thus far.

Alarge number of autoimmunity and infection-related immunopathologies, as well as protective immune responses, are mediated via T cells. Therefore, knowledge about Ags that are recognized by disease or protection-related or tissue- or pathogen-specific T cells is essential for understanding the disease process. Three different methodologies for the identification of T cell-stimulating Ags have been reported. The first approach identifies T cell-stimulating peptide epitopes that are presented on APCs in the context of an HLA molecule by peptide elution (1, 2). This enables database searches to identify the Ag involved. The second procedure makes use of biochemical techniques to isolate and identify the Ag out of complex Ag mixtures that stimulate the T cell of interest (3, 4). A third possibility for Ag identification is based on screening expression libraries and subsequent database searches (5, 6, 7).

Alternatively, progress has been made recently in the use of synthetic peptide libraries to analyze the peptide specificity of T cell clones (8, 9, 10, 11). The synthetic epitopes arising from synthetic peptide library screening are not natural epitopes, but mimics of natural epitopes. These mimicry epitopes may subsequently be used for the identification of natural Ags.

For Ag definition by database searching with mimicry epitope sequence information, a certain degree of sequence similarity between the mimicry epitope and the natural epitope is required. It has been reported that T cell clones can be activated upon stimulation by ligands that hardly share any sequence homology (12, 13, 14). This might impair database searches based on mimicry epitopes. Here we show that the natural Ag can be unambiguously identified using a synthetic library approach for three unrelated CD4+ T cell clones.

We describe a complete protocol for library screening, search pattern definition, and database searches that led to the identification of T cell Ags. DR3-restricted CD4+ T cell clones were used to screen two sublibraries with slightly different DR3-binding submotifs (15, 16). Two clones, MT1 (17) and MT2 (18), are Mycobacterium tuberculosis reactive. The third clone, HG (N. C. Schloot, O, M. C. Batstra, G. Duinkerken, R. R. de Vries, T. Dyrberg, A. Chaudhuri, P. O. Behan, and B. O. Roep, unpublished observation), recognizes human 65-kDa glutamic acid decarboxylase (GAD65),3 a major autoantigen in insulin-dependent diabetes mellitus (19).

Synthetic peptides were synthesized on an Abimed 422 multiple peptide synthesizer (Abimed Analyes-Technik, Langenfeld, Germany) using fluorenylmethoxycarbonyl (F-moc)-protected amino acids and TentagelS-AC resins (Rapp, Tübingen, Germany) as described (10). The purity of the peptides was determined by reversed-phase HPLC, and the integrity of the peptides was determined by matrix-assisted laser desorption ionization time of flight (MALDI-TOF) mass spectrometry on a Lasermat mass spectrometer (Finnigan-MAT; Hemel Hempstead, U.K.).

Hybrid TentaGelH-AM resin (particle size, 90 μm; loading, 100 pmol/bead, 16 pmol acid stabile attached, 84 pmol acid labile attached) (Rapp) (20) was used to synthesize two random one-bead/one-peptide 14-mer peptide libraries containing two different DR3-binding motifs (15, 16). The hybrid resin allows for a convergent library screening using the acid cleavable part of the peptide material attached to the resin (20), combined with highly efficient peptide identification by Edman sequencing, using the noncleavable part of the peptide material attached to the resin. The rationale of the library design is the same as described before (10). The design of the two libraries with two different DR3-binding motifs is summarized by the following synthesis schemes: library 1, XXX(L,I,M,V,A,Y,F)XXDXXXXXXX-GABA; and library 2, XXX(L,I,M,V,A,Y,F)XX(N,E,Q,S,T)X(K,R,H)XXXXX-GABA, with X being one of 19 L-amino acids (all natural amino acids except C, which was omitted for synthetic reasons), and GABA being γ-aminobutyric acid. Each library was synthesized using chemistry as described above, following a mix and split protocol (21, 22), yielding two one-bead/one-peptide libraries, each with a complexity of 4 × 106.

Convergent peptide cleavage and screening was performed as described (20). Shortly, libraries were divided into pools of 20,000 beads. Part of the peptide material was released from the beads for testing in a T cell proliferation assay (first round of screening). Beads of active pools were subdivided to pools of 70 beads. Again, peptides were released partially for the second round of screening. For the third round of screening beads were divided in limiting dilution. The remaining acid labile-attached peptide was released and tested. Cleavage conditions for the hybrid resins (20) differ from those described for TentaGelS-AM (10). Peptide sequences were determined by manual application of single beads (still containing sequenceable amounts of peptide; 10–15 pmol) to a cartridge and subsequent sequencing using a Hewlett Packard (Palo Alto, CA) G1005A protein sequencer.

MT1 (=Rp151-1) is a DR3-restricted CD4+ T cell clone that recognizes the N-terminal part of a 65-kDa heat shock protein of mycobacteria (HSP65(2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15)) (17). MT2 (=CAAp151-3) (18) is a DR3-restricted CD4+ T cell clone that recognizes protein 85 (85B(55–68)) (23, 24) of M. tuberculosis (28). HG (=PM1#11) is a DR3-restricted CD4+ T cell clone that recognizes human 65-kDa glutamic acid decarboxylase (GAD65(339–352)) (N. C. Schloot et al., unpublished observations).

CD4+ T cell proliferation assays for testing library pools and synthetic peptides were performed using 1 × 104 T cells and 5 × 104 irradiated HLA-DR3-matched PBMCs per well in flat-bottom 96-well plates in complete Iscove’s modified Dulbecco’s medium (150 μl) (Life Technologies, Gaithersburg, MD) containing 10% pooled human serum. Phytohemagglutinin (10 μg/ml), purified protein derivative of M. tuberculosis (10 μg/ml), and IL-2 (T cell growth factor, 10% Lymphocult, Biotest Diagnostics, Danville, NJ) were used as positive controls for T cell proliferation. [3H]Thymidine (0.5 μCi in 50 μl RPMI 1640) was added after 72 h, cells were harvested (Micro Cell Harvester, Skatron, Lier, Norway), and activity of the T cell DNA was counted after another 18 h (Model 1205 Betaplate, Liquid Scintillation Counter, LKB Instruments, Gaithersburg, MD). Library pools were tested in a quantity of 7 μl per well, giving final test concentrations of 5 nM for each individual peptide and 0.1% DMSO (v/v).

The M. tuberculosis SHOTGUN database (TB_shotgun.dbs, December 1997) was retrieved from the Sanger Center (http://www.sanger.ac.uk/Projects/M_Tuberculosis/) by ftp (ftp.sanger.ac.uk). The SHOTGUN database, covering the complete genome of M. tuberculosis (strain H37Rv), was converted by translating all possible open reading frames on both the coding and the noncoding strand into hypothetical protein sequences. An incompletely defined codon was translated to an X (±1% of the database). Pattern searches were performed in the translated TB_shotgun.dbs using PeptideSearch (25).

Two distinct but related sets of peptides have been described to bind to HLA-DR3 molecules (15, 16). The first set uses a hydrophobic amino acid at relative position 1 and a D at relative position 4. The second set consists of peptides that need a third (basic) anchor at relative position 6 (K, R, or H), probably due to a weak anchor at position 4 (E, Q, N, S, T). Two 14-mer peptide libraries were synthesized that reflect these two binding motifs. Both libraries had a complexity of 4 × 106. The first library reflects 1912 × 7 × 1 theoretically possible 14-mers. For the second library, this theoretical complexity is 1911 × 7 × 5 × 3. This means that both libraries are highly incomplete. Identification of mimicry epitopes from these libraries implies that T cells can recognize a large number of different 14-mer peptides. Because a mix and split protocol is used for synthesis, random positions are not biased to certain synthetically preferable combinations. Pools of each library, containing 20,000 individual peptides, were screened with two DR3-restricted M. tuberculosis-reactive T cell clones (MT1 and MT2) and a DR3-restricted human GAD65-reactive T cell clone (HG) (Table I). Pools were considered positive when counts exceeded five times the background counts. For clone MT1, activity was observed in one pool of library 1. Convergent screening (10, 20) led to the identification of stimulatory peptide MT1-P1 (NSTVAYDEAMIFAQ) (Table II). For clone MT2, two active pools of library 1 were obtained, resulting in the identification of MT2-P1 (NSAIGIDIPVARRD) and MT2-P2 (SHFVGXDIPVSLKH) (Table II). For clone HG, activity was observed in one pool of library 1 as well as in one pool of library 2, demonstrating that this clone is able to recognize peptides bearing two distinct binding motifs. The stimulatory peptides were identified as SIAMAFDPQIPMAA (HG-P1) and TDSLAFEPKVPRRQ (HG-P2) (Table II). The stimulatory capacities of the identified peptides were compared with their natural counterparts in a T cell proliferation assay (Fig. 1). Retrospectively, it was shown that suboptimal concentrations of 5 nM (which equals the individual peptide concentrations during library screening) are indeed stimulatory for all mimicry epitopes.

Table I.

Proliferative responses of T cell clones to active library fractions in the various screenings

T Cell CloneScreen 1 (cpm)aScreen 2 (cpm)Screen 3 (cpm)
MT1 785 (109) 1,502 (76) 5,392 (148) 
MT2 6,444 (153) 23,196 (533) 25,738 (159) 
 5,632 (153) 17,611 (533) 25,060 (159) 
HG 4,257 (121) 20,380 (75) 51,743 (115) 
 18,493 (121) 18,201 (99) 11,608 (251) 
T Cell CloneScreen 1 (cpm)aScreen 2 (cpm)Screen 3 (cpm)
MT1 785 (109) 1,502 (76) 5,392 (148) 
MT2 6,444 (153) 23,196 (533) 25,738 (159) 
 5,632 (153) 17,611 (533) 25,060 (159) 
HG 4,257 (121) 20,380 (75) 51,743 (115) 
 18,493 (121) 18,201 (99) 11,608 (251) 
a

Shown are counts of single library screening tests. In all library screenings, individual peptide concentrations are approximately 5 nM. Background counts are indicated between parentheses (duplicate tests for MT1 and MT2, triplicate tests for HG).

Table II.

Alignment of natural epitopes with synthetic library mimicry epitopes

Epitope/MimitopeaSequence
M. tuberculosis HSP65(2–15) AKTIAYDEEARRGL 
MT1-P1 NSTVAYDEAMIFAQ 
M. tuberculosis 85B(55–68) SPSMGRDIKVQFQS 
MT2-P1 NSAIGIDIPVARRD 
MT2-P2 SHFVGXDIPVSLKH 
Human GAD65(339–352) TVYGAFDPLLAVAD 
HG-P1 SIAMAFDPQIPMAA 
HG-P2 TDSLAFEPKVPRRQ 
Epitope/MimitopeaSequence
M. tuberculosis HSP65(2–15) AKTIAYDEEARRGL 
MT1-P1 NSTVAYDEAMIFAQ 
M. tuberculosis 85B(55–68) SPSMGRDIKVQFQS 
MT2-P1 NSAIGIDIPVARRD 
MT2-P2 SHFVGXDIPVSLKH 
Human GAD65(339–352) TVYGAFDPLLAVAD 
HG-P1 SIAMAFDPQIPMAA 
HG-P2 TDSLAFEPKVPRRQ 
a

Epitopes and mimitopes or clones MT1, MT2, and HG. Shown are 14-mer sequences with N termini starting three amino acids before the first anchor position (relative position 1). Due to problems with Edman degradation, relative position three of MT2-P2 remained unclear (X in sequence of MT2-P2).

FIGURE 1.

Proliferative responses versus peptide concentrations for clones MT1 (A), MT2 (B), and HG (C). Dose-response curves are shown for both natural epitopes and synthetic peptide library mimicry epitopes. Responses are the means of duplicate tests for MT1 and MT2, responses are means of triplicate tests for HG. Background cpm were 113 (MT1), 88 (MT2), and 340 (HG).

FIGURE 1.

Proliferative responses versus peptide concentrations for clones MT1 (A), MT2 (B), and HG (C). Dose-response curves are shown for both natural epitopes and synthetic peptide library mimicry epitopes. Responses are the means of duplicate tests for MT1 and MT2, responses are means of triplicate tests for HG. Background cpm were 113 (MT1), 88 (MT2), and 340 (HG).

Close modal

For mimicry epitopes MT1-P1 and HG-P1, the corresponding natural epitopes could be identified using a homology search. Searching with MT1-P1 (relative positions 1–7) in the M. tuberculosis database yielded HSP65(2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15) as hit number 8. Searching with HG-P1 (relative positions 1–7) in a human abstract of the Swiss Prot database yielded human GAD65(339–352) as hit number 30. Relative postions 1–7 were used for homology searching to maximize the chance of only including positions specifically involved either in DR3 binding or TCR interaction (16).

Standard homology database searches with either the complete sequences of MT2-P1 and HG-P2 or part of the sequences of these mimicry epitopes did not lead to identification of the natural epitopes for clone MT2 and HG within the 100 best-matching hits out of either the M. tuberculosis database or a human abstract of the Swiss Prot database. Therefore, a detailed analysis was performed on MT2-P1 and HG-P2 to define amino acid requirements for stimulation of clones MT2 and HG.

Although identification of the natural epitopes for clones MT2 and HG using mimicry epitopes MT2-P1 and HG-2 was performed similarly, only the procedure for clone MT2 is described in detail.

We chose to focus on relative positions 1–7 of MT2-P1 (IGIDIPV), because the similarity in this part of the peptide with MT2-P2 suggested an important role for HLA binding and TCR interaction. To determine the absolute importance for T cell stimulation of the amino acids in this part of the sequence, an omission mixture analysis was performed. In an omission mixture for relative position 1, a mixture of 18 peptides was synthesized and tested for T cell stimulation (consensus sequence NSAO1GIDIPVARRD, where O1 means all 20 natural L-amino acids except I, which was omitted because it is present at relative position 1 in MT2-P1, and C, which was omitted for synthetic reasons). The omission mixtures for relative position 2 and 5 induced no proliferative response (Table III). Therefore, it could be concluded that, in the context of the MT2-P1 sequence, a G at relative position 2 and an I at relative position 5 are essential for proliferation of clone MT2.

Table III.

Omission mixture analysis for positions 1–7 of MT2-P1

Omission Mixturea100 nM (cpm/1000)10 μM (cpm/1000)
NSAXGIDIPVARRD 26 19 
NSAIXIDIPVARRD 
NSAIGXDIPVARRD 33 20 
NSAIGIXIPVARRD 12 
NSAIGIDXPVARRD 
NSAIGIDIXVARRD 17 18 
NSAIGIDIPXARRD 
Omission Mixturea100 nM (cpm/1000)10 μM (cpm/1000)
NSAXGIDIPVARRD 26 19 
NSAIXIDIPVARRD 
NSAIGXDIPVARRD 33 20 
NSAIGIXIPVARRD 12 
NSAIGIDXPVARRD 
NSAIGIDIXVARRD 17 18 
NSAIGIDIPXARRD 
a

X (identified in bold) means that the mixture contains all natural L-amino acids except C and is the amino acid present at the position of interest in MT2-P1. The peptide concentrations that are tested for proliferation induction, 100 nM and 10 μM, are total concentrations of all peptides in a mixture. Responses are the means of duplicate tests.

Amino acids L, I, M, V, A, G, W, Y, and F have been described as anchors for relative position 1 for HLA-DR3 (16). To confirm the expectation that in MT2-P1 the I at relative position 1 can only be substituted by this set of amino acids without disturbing stimulatory capacity, an omission mixture was synthesized: NSAX1GIDIPVARRD (X1 = all natural L-amino acids, except C and the expected anchor residues). This mixture of 10 peptides was not able to induce a significant proliferation of clone MT2 (Table IV). The same omission analysis was performed for the other anchor positions, relative positions 4 and 6. Anchor amino acids for relative positions 4 and 6, which have been described in literature (15, 16), and the amino acids present at positions 4 and 6 of MT2-P1 were taken into account. As interdependence of those anchor positions is expected (15), relative position 4 was studied in combination with a random relative position 6, and vice versa: NSAIGIX2IX3VARRD (X2 = all natural L-amino acids, except D, N, S, T, and C; X3 = all natural L-amino acids, except C) and NSAIGIX4IX5VARRD (X4 = all natural L-amino acids, except C; X5 = all natural L-amino acids, except P, H, R, K, and C). Both the omission mixtures for relative position 4 and relative position 6 were inactive in proliferation assays (Table IV).

Table IV.

Omission mixture anchor analysis for positions 1, 4, and 6 of MT2-P1

Omission Mixturea100 nM (cpm/1000)10 μM (cpm/1000)
NSAX1GID IP VARRD 
NSAI GIX2IX3VARRD 
NSAI GIX4IX5VARRD 
Omission Mixturea100 nM (cpm/1000)10 μM (cpm/1000)
NSAX1GID IP VARRD 
NSAI GIX2IX3VARRD 
NSAI GIX4IX5VARRD 
a

X1–X5 (identified in bold) mean that the position includes all natural L-amino acids except: L,I,M,V,A,G,W,Y,F,C (X1); D,N,S,T,C (X2); C (X3); C (X4); K,R,H,P,C (X5). The peptide concentrations that are tested for proliferation induction, 100 nM and 10 μM, are total concentrations of all peptides in a mixture. Responses are the means of duplicate tests.

G at relative position 2 and I at relative position 5 were essential for proliferation of clone MT2. The remaining potential TCR contact residues within the 7-mer core sequence of MT2-P1 are at relative position 3 and 7. For these positions, all substitution peptides (except C substitutions for synthetic reasons) were synthesized and tested for T cell stimulation (Fig. 2). For relative position 3, L, R, S, T, N, D, W, and Y are allowed amino acids. For relative position 7, I, T, and K were shown to be allowed.

FIGURE 2.

Substitution analysis for relative positions 3 and 7 of synthetic peptide library mimicry epitope MT2-P1 of clone MT2. Proliferative responses are shown for peptide concentrations of 100 nM and are the means of duplicate tests.

FIGURE 2.

Substitution analysis for relative positions 3 and 7 of synthetic peptide library mimicry epitope MT2-P1 of clone MT2. Proliferative responses are shown for peptide concentrations of 100 nM and are the means of duplicate tests.

Close modal

Based on the omission and substitution studies as described (Tables III and IV and Fig. 2), a search pattern was constructed for PeptideSearch. This resulted in a pattern at (relative) position 1 = (L,I,M,V,A,G,W,Y,F), position 2 = G, position 3 = (I,L,R,S,T, N,D,W,Y), position 4 = (D,N,S,T), position 5 = I, position 6 = (P,K,R,H), and position 7 = (V,T,I,K). The translated M. tuberculosis SHOTGUN database was searched with this pattern, yielding 19 hits (Table V). The natural epitope M. tuberculosis 85B(55–68), SPSMGRDIKVQFQS (number 9), was indeed part of the list. In addition to peptide 9, peptides 2, 15, and 16 were also recognized by clone MT2. Peptide 2 is derived from M. tuberculosis 85C. Peptide 15, which contains an X (due to a DNA sequencing uncertainty), is derived from either M. tuberculosis 85A or 85B (X = P). 85A (32 kDa), 85B (30 kDa), and 85C (30 kDa) are highly homologous proteins, probably originated from gene duplication (24, 26). They all show mycolyltransferase activity. Therefore, it is likely that these three proteins are involved in cell wall assembly (26). Peptide 16 differs at only one amino acid position from the corresponding 14-mer peptide derived from M. tuberculosis major protein Ag MPT51 (27) (SPSMGRDKPVAFLA, difference underlined). This K to I substitution probably results from a DNA sequencing error present in the M. tuberculosis SHOTGUN database (a single substitution of a T for an A can change the codon for K (AAA) into the codon for I (ATA)). The 14-mer peptide of MPT51 (SPSMGRDKPVAFLA) appeared not to be able to stimulate clone MT2, which was expected from the search pattern that was used for searching in the M. tuberculosis SHOTGUN database (only I was included at relative position 5).

Table V.

Hits from SHOTGUN database

PeptideM. tuberculosis SHOTGUN Database Hitsa
1. LFPVGLSIPTGGPL 
2. SASMGRDIKVQFQG 
3. DEIVGLNIPTGIPL 
4. QDPIGNSIKIIIAF 
5. LEPGGSTIPKIPFI 
6. MASIGLTIPTIALA 
7. LRPAGTSIRILFAF 
8. ASIVGRTIPVAGAV 
9. SPSMGRDIKVQFQS 
10. MNWAGWTIPTPSSS 
11. WIMAGLTIPTTGAC 
12. SRGIGDTIRVSLSA 
13. YPLVGTTIPIAGYT 
14. HPLFGSSIPTPIPS 
15. SXSMGRDIKVQFQS 
16. SPSMGRDIPVAFXX 
17. HCRYGTSIKITPGQ 
18. QALVGRDIKKLPTL 
19. RAAMGLTIPVLLIR 
PeptideM. tuberculosis SHOTGUN Database Hitsa
1. LFPVGLSIPTGGPL 
2. SASMGRDIKVQFQG 
3. DEIVGLNIPTGIPL 
4. QDPIGNSIKIIIAF 
5. LEPGGSTIPKIPFI 
6. MASIGLTIPTIALA 
7. LRPAGTSIRILFAF 
8. ASIVGRTIPVAGAV 
9. SPSMGRDIKVQFQS 
10. MNWAGWTIPTPSSS 
11. WIMAGLTIPTTGAC 
12. SRGIGDTIRVSLSA 
13. YPLVGTTIPIAGYT 
14. HPLFGSSIPTPIPS 
15. SXSMGRDIKVQFQS 
16. SPSMGRDIPVAFXX 
17. HCRYGTSIKITPGQ 
18. QALVGRDIKKLPTL 
19. RAAMGLTIPVLLIR 
a

PeptideSearch hits for clone MT2. Peptides 2, 9, 15, and 16 induce proliferation of clone MT2. X means one of all natural L-amino acids. A mixture of all natural L-amino acids except C is synthesized at each X position. All hits completely match the search pattern: (relative) position 1 = L,I,M,V,A,G,W,Y,F; position 2 = G; position 3 = I,L,R,S,T,N,D,W,Y; position 4 = D,N,S,T; position 5 = I; position 6 = K,R,H,P; position 7 = V,T,I,K. Relative positions 1–7 are printed in bold type.

Molecular mimicry enables epitope identification from incomplete synthetic peptide libraries (10, 11). Mimicry epitopes for three unrelated HLA-DR3-restricted CD4+ T cell clones were identified from synthetic peptide libraries. For all three T cell clones studied here, the similarity (conservative substitutions or even identical amino acids) of synthetic mimicry epitopes and natural epitopes was high (Table II). This implies that all three clones have a strict recognition pattern. These data further suggest that recognition of ligands without any sequence similarity by one T cell clone may not be a very common phenomenon for epitopes that are recognized in low nanomolar concentration with a regular binding motif (14, 15, 16). This implies a probability for Ag definition by database searching using sequence information from synthetic mimicry epitopes, provided that the Ag is contained in the database.

In two cases, a simple homology search with the core of the sequence of a synthetic mimicry epitope obtained from the library screening led to immediate identification of the natural epitope as part of a large set of heterogeneous database hits. A homology search with the core sequences (relative positions 1–7) of mimicry epitopes (MT1-P1 and HG-P1) identified the corresponding natural Ags within the 100 best-matching database hits (M. tuberculosis database for clone MT1 and a human abstract of the Swiss Prot database for clone HG).

In two other cases (mimicry epitopes MT2-P1 and HG-P2), the size of the set of hits containing the natural epitope appeared to be too large to be investigated. Therefore we developed a generally applicable approach that increases the probability of successful identification of natural epitopes by pattern searching with a precisely defined pattern (Fig. 3). This approach is based on the observation that stimulating ligands can be predicted by studying individual amino acid positions (14). Using a precisely defined pattern for searching the M. tuberculosis SHOTGUN database, the natural epitope of clone MT2 (BCG85B(55–68)) was unambiguously identified. A comparable analysis and subsequently a pattern search in the nonredundant database was performed for clone HG using mimicry epitope HG-P2. GAD65 was indeed part of the human fraction of all pattern hits (data not shown). This indicates that similarity rather than homology between mimicry epitopes and natural epitopes is required for natural Ag identification.

FIGURE 3.

Schematic representation of the procedure to define T cell Ags with synthetic peptide library mimicry epitopes.

FIGURE 3.

Schematic representation of the procedure to define T cell Ags with synthetic peptide library mimicry epitopes.

Close modal

This pattern search approach is generally applicable for T cell clones with unknown Ag specificity. The interpretation of the output of a pattern search, compared with a homology search, is easier, because the match score is equal (100%) for all hits without any assumptions and statistical calculations involved. In addition, the use of pattern searches allows for approaches that are independent of the numer of amino acid positions analyzed. The number of database hits using a pattern search depends on the size of the database, the number of positions that are defined in the pattern, the number of permitted amino acids for the defined pattern positions, the nature of amino acids in the pattern, and the relative occurrence of each amino acid in databases. From a practical point of view, a balance must be found between the experimental effort needed to determine a very precise pattern, yielding only few database hits, and the effort that is needed to synthesize and test a large list of hits resulting from a less precise search pattern. In this study, the definition of the search pattern resulted in 19 hits, which is sufficiently limited for synthesis and testing.

One clone (HG) was able to recognize two similar mimicry epitopes containing two slightly different HLA-DR3-binding motifs (library 1 and library 2 motifs). In this example, our approach succeeded in unambiguous identification of human GAD65 using the sequence information of HG-P2, whereas a homology search with this peptide did not identify the natural Ag. Our approach might further contribute to the search for functional T cell-epitope mimicry in autoimmune disease that is not based on simple-sequence homology.

We conclude that mimicry epitopes identified from synthetic peptide libraries are highly similar to their natural counterparts. Therefore, in some cases the natural epitope can be defined by a simple homology database search, depending on the degree of homology and the size of the database of interest. Because the success probability of homology searching is limited, we present a generally applicable protocol based on similarity, which represents a convenient approach for Ag definition using sequence information obtained from synthetic mimicry epitopes (Fig. 3). This approach opens new perspectives for rapid and reliable Ag definition and represents a feasible alternative for the biochemical and molecular biology approaches described thus far. Defined Ags can be used for vaccine design in infectious disease and for the development of immune intervention strategies in autoimmune disease.

We thank Dr. F. Koning for critically reading the manuscript and Kees L. M. C. Franken for cloning and purification of M. tuberculosis 85B.

1

This work was supported by the Leiden University Medical Center, the Dutch Leprosy Relief Association, the Commission European Community, the Royal Netherlands Academy of Arts and Sciences, the Netherlands Organization of Scientific Research, the Macropa Foundation, and the Diabetes Fonds Nederland.

3

Abbreviation used in this paper: GAD65, 65-kDa glutamic acid decarboxylase.

1
Hunt, D. F., R. A. Henderson, J. Shabnowitz, K. Sakaguchi, H. Michel, N. Selilir, A. Cox, E. Apella, V. H. Engelhard.
1992
. Characterization of peptides bound to the class I MHC molecule HLA-A2. 1 by mass spectrometry.
Science
255
:
1261
2
Udaka, K., T. J. Tsomides, H. N. Eisen.
1992
. A naturally occurring peptide recognized by alloreactive CD8+ cytotoxic T lymphocytes in association with a class I MHC protein.
Cell
69
:
989
3
Van de Wal, Y., Y. Kooy, P. van Veelen, A. S. Pena, M. L. Mearin, T. Mutis, W. Benckhuijsen, J. W. Drijfhout, F. Koning.
1998
. T cells from the small intestinal mucosa of a coeliac disease patient recognized a unique peptide sequence of gliadin.
Proc. Natl. Acad. Sci. USA.
95
:
10050
4
Lamb, J. R., D. B. Young.
1987
. A novel approach to the identification of T cell epitopes in Mycobacterium tuberculosis using human T lymphocyte clones.
Immunology
60
:
1
5
Boon, T., B. Cerottini, P. Van den Eynde, P. Van der Bruggen, A. Van Pel.
1994
. Tumor antigens recognized by T lymphocytes.
Annu. Rev. Immunol.
12
:
337
6
Neophytou, P. I., B. O. Roep, S. D. Arden, E. M. Muir, G. Duinkerken, A. Kallan, R. R. P. De Vries, J. C. Hutton.
1996
. T-cell epitope analysis using subtracted expression libraries (TEASEL): application to a 38-kDa autoantigen recognized by T cells from an insulin-dependent diabetic patient.
Proc. Natl. Acad. Sci. USA
93
:
2014
7
M. A., Gavin, B. Dere, A. G. Grandea, III, K. A. Hogquist, M. J. Bevan.
1994
. Major histocompatibility complex class I allele-specific peptide libraries: identification of peptides that mimic an H-Y T cell epitope.
Eur. J. Immunol.
24
:
2124
8
Gundlach, B. J., K.-H. Wiesmüller, T. Junt, S. Kienle, G. Jung, P. Walden.
1996
. Determination of T cell epitopes with random peptide libraries.
J. Immunol. Meth.
192
:
149
9
Blake, J., J. V. Johnston, K. E. Hellström, H. Marquardt, L. Chen.
1996
. Use of combinatorial peptide libraries to construct functional mimics of tumor epitopes recognized by MHC class I-restricted cytolytic T lymphocytes.
J. Exp. Med.
184
:
121
10
Hiemstra, H. S., G. Duinkerken, W. E. Benckhuijsen, R. Amons, R. R. P. De Vries, B. O. Roep, J. W. Drijfhout.
1997
. The identification of CD4+ T cell epitopes with dedicated synthetic peptide libraries.
Proc. Natl. Acad. Sci. USA
94
:
10313
11
Hemmer, B., B. T. Fleckenstein, M. Vergelli, G. Jung, H. McFarland, R. Martin, K.-H. Wiesmüller.
1997
. Identification of high potency microbial and self ligands for a human autoreactive class II-restricted T cell clone.
J. Exp. Med.
185
:
1651
12
Nanda, K. N., K. K. Arzoo, H. M. Geysen, A. Sette, E. E. Sercasz.
1995
. Recognition of multiple peptide cores by a single T cell receptor.
J. Exp. Med.
182
:
531
13
Quarantino, S., C. J. Thorpe, P. J. Travers, M. Londei.
1995
. Similar antigenic surfaces, rather than sequence homology, dictate T-cell eptiope molecular mimicry.
Proc. Natl. Acad. Sci. USA
92
:
10398
14
Hemmer, B., M. Vergell, B. Gran, N. Ling, P. Conlon, C. Pinilla, R. Houghton, H. F. McFarland, R. Martin.
1998
. Cutting edge: predictable TCR antigen recognition based on peptide scans leads to the identification of agonist ligands with no sequence homology.
J. Immunol.
160
:
3631
15
Geluk, A., K. E. Van Meijgaarden, S. Southwood, C. Oseroff, J. W. Drijfhout, R. R. P. De Vries, T. H. M. Ottenhoff, Sette A..
1994
. HLA-DR3 molecules can bind peptides carrying two alternative specific submotifs.
J. Immunol.
152
:
5742
16
Rammensee, H.-G., T. Friede, S. Stevanovic.
1995
. MHC ligands and peptide motifs: first listing.
Immunogenetics
41
:
178
17
Geluk, A., K. E. Van Meijgaarden, A. A. M. Janson, J. W. Drijfhout, R. H. Meloen, R. R. P. De Vries, T. H. M. Ottenhoff.
1992
. Functional analysis of DR17(DR3)-restricted mycobacterial T cell epitopes reveals DR17 binding motif and enables the design of allele-specific competitor peptides.
J. Immunol.
149
:
2864
18
Janson, A. M., P. R. Klatser, R. Van der Zee, Y. E. Cornelisse, R. R. P. De Vries, J. E. R. Thole, T. H. M. Ottenhoff.
1991
. A systematic molecular analysis of the T cell-stimulating antigens from Mycobacterium leprae with T cell clones of leprosy patients.
J. Immunol.
147
:
3530
19
Roep, B. O..
1996
. T-cell responses to autoantigens in IDDM: the search for the holy grail.
Diabetes
45
:
1147
20
Hiemstra, H. S., W. E. Benckhuijsen, R. Amons, W. Rapp, and J. W. Drijfhout. 1998. A new hybrid resin for stepwise screening of peptide libraries combined with single bead edman sequencing. J. Pept. Sci. In press.
21
Lam, K. S., S. E. Salmon, E. M. Hersh, V. J. Hruby, W. M. Kazmiersky, R. J. Knapp.
1991
. A new type of synthetic peptide library for identifying ligand-binding activity.
Nature
354
:
82
22
M., Lebl., V. Krchnak, N. F. Sepetov, B. Seligman, P. Strop, S. Felder.
1995
. One-bead-one-structure combinatorial libraries.
Biopolymers
37
:
177
23
Thole, J. E. R., R. Schöning, A. A. M. Janson, T. Garbe, Y. E. Cornelisse, J. E. Clark-Curtiss, A. H. J. Kolk, T. H. M. Ottenhoff, R. R. P. de Vries, C. Abou-Zeid.
1992
. Molecular and immunological analysis of a fibronectin-binding protein antigen secreted by Mycobacterium leprae.
Mol. Microbiol.
6
:
153
24
Young, D. B., S. H. Kaufmann, P. W. Herman, J. E. Thole.
1992
. Mycobacterial protein antigens: a compilation.
Mol. Microbiol.
6
:
133
25
Mann, M., M. S. Wilm.
1994
. Error-tolerant identification of peptides in sequence databases by peptide sequencing.
Anal. Chem.
66
:
4390
26
Belisle, J. T., V. D. Vissa, T. Sievert, K. Takayama, P. J. Brennan, G. S. Besra.
1997
. Role of the major antigen of Mycobacterium tuberculosis in cell wall biogenesis.
Science
276
:
1420
27
Harboe, M., T. Oettinger, H. G. Wiker, I. Rosenkrands, P. Andersen.
1996
. Evidence of occurrence of the ESAT-6 protein in Mycobacterium tuberculosis and virulent Mycobacterium bovis and for its absence in Mycobacterium bovis BCG.
Infect. Immun.
64
:
16
28
Geluk, A., V. Taneja, K. E. van Meijgaarden, E. Zanelli, C. Abou-Zeid, J. E. R. Thole, R. R. P. de Vries, C. S. David, and T. H. M. Ottenhof. 1998. Identification of HLA class II-restricted determinants of M. tuberculosis-derived proteins using HLA-transgenic class II-deficient mice. Proc. Natl. Acad. Sci. USA. In press.