Antigen Specificity of γδ T Cells Depends Primarily on the Flanking Sequences of CDR3δ*

The structural basis that determines the specificity of γδ T cell receptor (TCR) recognition remains undefined. Our previous data show that the complementary determining region of human TCRδ (CDR3δ) is critical to ligand binding. Here we used linear and configurational approaches to examine the roles of V, N-D-N, or J regions in CDR3δ-mediated antigen recognition. Surprisingly, we found that the binding activities of CDR3δ from different γδ TCRs to their target tissues and ligands depend on the conserved flanking sequences (V and J) but not as much on the D region of CDR3δ fragment. We further defined the key residues in the V and J regions of CDR3δ fragments, including the cysteine residue in the V fragment and the leucine residue in the J fragment that determine their ligand binding specificity. Our results demonstrate that TCRδ primarily uses conserved flanking regions to bind ligands. This finding may provide an explanation for the limited number of γδ TCR ligands that have as yet been identified.

The structural basis that determines the specificity of ␥␦ T cell receptor (TCR) recognition remains undefined. Our previous data show that the complementary determining region of human TCR␦ (CDR3␦) is critical to ligand binding. Here we used linear and configurational approaches to examine the roles of V, N-D-N, or J regions in CDR3␦-mediated antigen recognition. Surprisingly, we found that the binding activities of CDR3␦ from different ␥␦ TCRs to their target tissues and ligands depend on the conserved flanking sequences (V and J) but not as much on the D region of CDR3␦ fragment. We further defined the key residues in the V and J regions of CDR3␦ fragments, including the cysteine residue in the V fragment and the leucine residue in the J fragment that determine their ligand binding specificity. Our results demonstrate that TCR␦ primarily uses conserved flanking regions to bind ligands. This finding may provide an explanation for the limited number of ␥␦ TCR ligands that have as yet been identified.
Extensive studies suggest that ␥␦ T cells play important roles in host defense against microbial infections, monitoring of tumorigenesis, immunoregulation, and development of autoimmunity (1)(2)(3). However, little is known about the structural basis of antigenic recognition by ␥␦ T cell receptor (TCR) 3 because of the limited identified specific ligands for ␥␦ TCR and the lack of structural information revealing how ␥␦ TCR might interact with such ligands.
The crystallographic structure of a murine ␥␦ TCR in complex with major histocompatibility complex class (MHC) Ib T22 (4,5) showed that the CDR loops f ␥␦ TCR, predomi-nantly germline-encoded residues of the complementary determining region of human TCR␦ (CDR3␦), are in direct contact with T22, suggesting that the primary sequence of CDR3 in ␥␦ TCR, especially CDR3␦, serves as a key determinant for the specificity of antigen recognition. Our recent finding that CDR3␦ peptide mimics human ␥␦ TCR binding to tumor cells and tissues is consistent with the role of CDR3␦ in ␥␦ TCR recognition (6).
Based on this finding, we used synthesized CDR3␦ peptide as a probe to screen putative protein ligands in tumor protein extracts by affinity chromatography analysis. With this novel strategy, we have successfully identified seven tumor-related epitopes, two hepatitis B virus (HBV) infection-related antigenic epitopes, and two self proteins including heat shock protein (HSP) 60 and human mutS homolog 2 (hMSH2) that are recognized by human ␥␦ TCR (7). These results further support that the primary sequence of CDR3␦ in ␥␦ TCR determine the specificity of antigen binding.
CDR3␦ is composed of fragments derived from V, N-D-N, and J gene segments. The flanking sequences composed of V and J fragment is conserved while N-D-N region is diverse. The diversity of ␥␦ TCRs is supposedly higher than that of TCR␣␤ due to the link of D gene fragment and the insertion of nucleotide acids (8). However, the number of identified antigenic ligands recognized by ␥␦ TCR remains very limited. It has been demonstrated that ␥␦ TCR recognizes some protein antigens and small phosphate or amine-containing compounds, including nonclassical MHC class I molecule T22 and T10 in mice (9), UL-16-binding protein (ULBP) (10) and mitochondrial F1-ATPase in humans (11). Nevertheless, important questions regarding ␥␦ TCR recognition remain to be addressed. For example, given the seemingly high diversity of ␥␦ TCR, why have only limited antigenic ligands been identified? What are the contributions of individual fragments of CDR3␦ to antigen recognition? In ␣␤ TCR, a single mutation in D gene fragment (12) abolishes its antigenic recognition, whereas the contribution of the different fragments in ␥␦ TCR recognition remain unknown. Answers to these questions will shed important insights to antigen recognition of ␥␦ T cells.
In this study, we investigated the contribution of individual fragments of CDR3␦ in antigen recognition. We mutated V, N-D-N, or J fragments of a V␦2 TCR CDR3 sequence (OT3) in peptide and engineered ␥␦ TCR. We found that the conserved flanking regions of CDR3␦ play a critical role in antigenic binding to OEC cells/tissues or hMSH2 protein, a new ligand for ␥␦TCR we found recently (7). Furthermore, we have identified the cysteine residue in V fragment and the leucine residue in J fragment as critical residues in the binding activity of ␥␦ TCR. These results demonstrate that TCR␦ chain uses the conserved flanking regions to recognize their antigens, suggesting that ligands for ␥␦ ⌻CR may also be conserved and limited in number.

EXPERIMENTAL PROCEDURES
Cell Lines and Human Tissue Specimens-Various tumor cell lines including HO8910, 803, Hela, HepG2, K562, and J.RT3-T3.5 cell were obtained from the American Type Culture Collection (ATCC). The human ovarian tumor cell line SKOV3 was a gift from Dr. Keng Shen (Department of Gynecology, The Peking Union Medical College Hospital, China). PBMCs were obtained from peripheral blood of healthy donors by density gradient centrifugation on Ficoll-Hypaque (Amersham Biosciences). Fresh tumors and normal tissue specimens were obtained from the Peking Union Medical College Hospital. All of the tissue specimens from patients diagnosed by standard histopathological and immunohistochemical assay were collected prior to treatment with chemotherapy, radiotherapy, or Chinese traditional medical therapy. All human studies were carried out according to proven guidelines by PUMC.
RT-PCR and Sequence Analysis-Total RNA was isolated from the tumor infiltrating lymphocytes (TILs) of the ovarian epithelial carcinoma (OEC) tissues with TRIzol reagent (Promega). cDNA was synthesized using oligo-dT (Promega) as primer and moloney murine leukemia virus reverse transcriptase (Promega) in the reverse transcription reaction and amplified by PCR using V␦and C␦-specific primers as follows. The primers wereTCR␦V2: 5Ј-GCA CCA TCA GAG AGA GAT GAA GGG-3Ј; TCR␦C 5Ј-AAA CGG ATG GTT TGG TAT GAG GC-3Ј. Amplified cDNA was cloned into the pGEM-T easy vector (Promega) and sequenced with the ABI automatic sequencer 3770.
Construction of Cell Lines Expressing ␥␦ TCR-A full-length ␥ or ␦ chain was amplified from PBMC cDNA using primers containing Kpn-I and Xho-I restriction sites. Primers encoding the entire ␥9 and ␦2 CDR3 sequence were used to construct the first half and the second half of the ␥9 and ␦2 chain. The two overlapping PCR products were then used in a third PCR reaction to recreate the full-length ␥9 and ␦2 chain whose CDR3 region was the OT3 nucleic acid sequence. The full-length TCR chain was digested with Kpn-I and XhoI and cloned into pREP7 and pREP9 expression vectors containing either hygromycin or neomycin resistance. The J.RT3-T3.5 cells (1.2 ϫ 10 7 ) were transfected with 20 g each of pREP7-␥9 plasmid and pREP9-␦2 plasmid by electroporation at 260 V and 975 F using the Bio-Rad Gene-Pulser. After 48 h, the cells transfected with both vectors were selected by hygromycin and neomycin for 4 weeks. The resulting cell lines were then validated for expression of ␥␦ TCR using reverse transcription polymerase chain reaction and FACS analysis.
Peptide Synthesis and Protein Expression-Peptides including wild-type CDR3␦2 (OT3) and its mutants were synthesized in the peptide synthesis facility of the Academy of Military Medical Sciences, China. Peptide OT3Vm, OT3Dm, and OT3Jm were respectively synthesized with random arrangement amino acid sequence in V, N-D-N, and J fragment of OT3. MAB peptide derived from melanoma Ag recognized by T cells-1 (MART)-specific ␣␤ T cells, as a negative peptide control, was synthesized (see Table 2). The purity of synthesized peptides was more than 90% in HPLC analysis. The synthesized peptides were all labeled with a biotin at their N terminus. The engineered chimeric protein and its mutants, containing the extracellular domains of the human ␥9 and ␦2 TCR chains fused to the hinge region, CH2 and CH3 domains of human IgG1, heavy chain, were expressed by the Sino Biological Inc. The CDR3␦ sequence of the chimeric protein is concordant with the OT3 peptide and its V/D/J mutants. The proteins were named as ␥9/␦2(OT3)-Fc, ␥9/␦2(OT3)Vm-Fc, ␥9/␦2(OT3)-Dm-Fc, and ␥9/␦2(OT3)Jm-Fc, respectively.
Sequence Mutagenesis of V/D/J and Site-directed Mutation in CDR3␦2(OT3)-We constructed the full-length ␦2 chain containing different mutations of OT3 according to the previous amino acid sequence. For the mutant D and J fragment, overlapping PCR was used to construct the full-length ␦2 chain. For the mutant V fragment and site-directed mutation, the fulllength ␦2 chain whose CDR3 region was the OT3 nucleic acid sequence was subcloned into pcDNA3.1. The plasmid of pcDNA3.1-␦2 chain was amplified with mutant-specific primer by pfu DNA polymerase. The PCR products were visualized on 0.8% TAE agarose gels and then purified using silica columns according to the manufacturer's instructions. The phosphate base was added to the 5Ј-end of the product using the T4 polynucleotide kinase under the action of ATP. After looping, the plasmid was transformed into DH5␣ Escherichia coli. Once the sequence was confirmed, the constructs were digested with Kpn-I and XhoI and cloned into pREP9 expression vectors. We co-transfected each of them together with an identical pREP7-␥9 chain into J.RT-T3.5 cells, and various transfected cells with stable expression of ␥␦ TCR were evaluated by function assay.
Stimulation of J.RT3-T3.5 Transfectants-The transfected cells expressing different ␥␦ TCR were preincubated with 10 ng/ml phorbol myristate acetate (PMA) at room temperature for 30 min. PMA treatment was essential for TCR-mediated activation in this system. After extensive washing with RPMI 1640, the cells were plated into 24-well plates at 1 ϫ 10 6 in the presence of tumor cell protein extracts, heat shock protein (HSP)70, or iso-butylamine (IBA). After 24 h, the supernatants or cells were harvested. The level of IL-2 in supernatants was detected using human IL-2 ELISA kit (B.D. Company) according to the manufacturer's instructions. Meanwhile, the RNA of cells was extracted to analyze the level of IL-2 by real-time PCR. For experiments in which the responses for different mutagenized TCR constructs were compared, surface ␥␦ TCR expression levels on all transfected cells were measured by FACS analysis to ensure there was no significant difference. Cells were stained within 48 h of their use in the stimulation assay.
Enzyme-linked Immunosorbent Assay-96-well plates were coated with 0.5 g of the N-terminal fragment of hMSH2 pro-tein (MNS) in 0.1 M NaHCO 3 (pH 9.6). After blocking with 5% bovine serum albumin, the plates were incubated with biotinconjugated CDR3␦ (OT3) peptide or ␥9/␦2 (OT3)-Fc protein and its variants for 1 h at room temperature. The plates were developed using HRP-conjugated streptavidin (Pierce) or HRPconjugated goat anti-human IgG antibody (Sigma) and substrate (Sigma) and read on a microplate reader at 450 nm (Labsystem).
Confocal Microscopy-Cells were fixed on slides by 2% cold paraformaldehyde and in turn incubated with biotin-conjugated synthesized CDR3␦2(OT3) peptide and their V/D/J mutant peptide, or the ␥9/␦2(OT3)-Fc protein and its V/D/J variants. FITC-conjugated streptavidin or FITC-conjugated goat anti-human IgG antibody (Pierce) was then added and incubated (30 min, 4°C). Controls included phosphate-buffered saline or wild-type Ig as the primary antibody. Slides were examined with a confocal laser microscope (LSM 510; Carl Zeiss).
Immunohistochemistry-Formalin-fixed paraffin-embedded sections of tumor tissues were deparaffinized and then boiled by microwave for antigen retrieval. After quenching with peroxide, the sections were blocked with 5% goat serum. Then, biotin-conjugated synthesized peptides (0.5 g) or V/D/J mutants and g9/d2(OT3)-Fc protein or V/D/J mutants were added to the slides. The sections were incubated with HRPconjugated streptavidin or HRP-conjugated goat anti-human IgG antibody. Binding was visualized using diaminobenzidine (Sigma) as the substrate and observed under microscope.
Surface Plasmon Resonance (SPR)-SPR studies were carried out with a BIAcore 3000 instrument at 25°C using HBS-EP running buffer (BIAcore, Sweden). Protein MNS was diluted to 30 g/ml in 10 mM sodium acetate, pH 4.0, and immobilized on a CM5 chip using EDC/NHS according to the manufacturer's instructions. The amount of immobilized protein was about 6000 resonance units. OT3 or its V/D/J mutant peptide and ␥9/␦2(OT3)-Fc protein or its V/D/J mutant protein were flowed as analyte at different concentrations. The recorded sensograms were analyzed using BIAevaluation software (Biacore Life Sciences). The data thus obtained were globally fit using a 1:1 binding model to calculate the dissociation constants (K D ).

RESULTS
Specific Binding of CDR3␦2 Peptide to Antigens for ␥␦ TCR Depends on Its Flanking Sequences-We previously identified a gene sequence of CDR3␦ in V␦2 from TILs in OEC (6) and named its synthesized peptide as OT3. CDR3␦2 is formed by somatic rearrangement of V, N-D-N, and J fragments. V and J genes encode the conserved flanking sequences of CDR3␦, while N-D-N genes encode inner diverse regions (Fig. 1A). To determine which fragment of CDR3␦ is the key determinant for antigen recognition, we generated OT3 peptide mutants by replacing V, N-D-N, or J fragments of OT3 with randomly arranged amino acid sequences of the same length termed peptide OT3Vm, OT3Dm, and OT3Jm, respectively. The binding of these synthesized peptide mutants to target cells and tissues in vitro was examined. As shown in Fig. 1, B and C, in contrast to the staining by wild-type OT3 peptide, the percentages of SKOV3, HO8910, or Hela tumor cell lines stained by peptides OT3Vm and OT3Jm were dramatically reduced whereas the percentage of positive cells stained by peptide OT3Dm was only modestly reduced. Similar binding activities of these OT3 peptides to OEC specimens were also observed (Fig. 1D).
We further examined the interaction of CDR3␦ peptides with an identified ligand for ␥␦ TCR, hMSH2 protein. As shown in Fig. 1E, the binding activity of peptides OT3Vm and OT3Jm to the N-terminal fragment of hMSH2 protein (MNS) was dramatically reduced while that of OT3Dm was not significantly affected. Furthermore, SPR analysis showed dramatic decreases in binding affinity of peptides OT3Vm and OT3Jm with MNS when compared with peptide OT3. In contrast, the peptides OT3Dm retained its strong binding to MNS (Fig. 1F). These data strongly suggest that V and J fragments are the primary determinants for the binding capacity of CDR3␦2 peptide to target tumor cells/tissues and ligand, while the D fragment only contributes minimally to such binding.
Cell Surface-expressed ␥␦ TCR Binding to Tumor Cells Depends on the V and J Fragment of CDR3␦2-The binding of various OT3 peptides to different target cell tissues or antigen ligand needs to be further validated in the context of intact ␥␦ TCR recognition. Therefore, we generated T cell lines expressing different ␥␦ TCRs containing OT3 or its mutants. We selected a ␥9 chain that paired with the ␦2 chain. We first analyzed the CDR3 sequence of the V␥9 chain in TIL of OEC (Table 1). No predominant motif was observed among these sequences. Nevertheless, the sequence of clone 9 appeared three times among 20 sequenced V␥9 chains. Therefore, we used the full-length ␥9 chain containing the CDR3 region from clone 9 to pair with various ␦2 chains containing different CDR3. These results from RT-PCR and FACS demonstrated the successful generation of J.RT3-T3.5 transfectants expressing ␥␦ TCR with OT3 or its mutants.
To test the antigen recognition capability of these cells expressing ␥␦ TCRs, we stimulated the transfected cells with tumor antigens. After stimulation with protein extracts from tumor cells, the production of IL-2 by the cell lines was detected by ELISA. As shown in Fig. 2A, compared with cells expressing ␥␦ TCR with wild-type OT3, those expressing ␥␦TCR with OT3Vm or OT3Jm produced much less IL-2 after the cells were stimulated with protein extracts from different tumor cells and recombinant HSP70 protein. Importantly, cells expressing ␥␦ TCR with OT3Dm secreted similar amounts of IL-2 to that secreted by cells expressing ␥␦ TCR with wild-type OT3. These IL-2 secretion results were further confirmed by quantitative analysis of IL-2 mRNA in the stimulated cells (Fig. 2B). These results suggest that mutations in the V and J but not D regions affect ␥␦ TCR antigen recognition capability.
Cell Surface-expressed ␥␦ TCRs Bind to Tumor Cells through Cysteine in the V or Leucine in the J Fragment of CDR3␦2-To determine the key positions of antigen recognition in the V and J fragment, we mutated each amino acid of the V and J fragment of OT3 (Table 2). After the transfected cells were stimulated with whole protein extracts of SKOV3 cells (OEC cell line), the production of IL-2 was compared. Mutations of cysteine in the V fragment or leucine in the J fragment dramatically decreased IL-2 production of the tested cells (Fig. 3A). We further investigated whether these residues are required for binding to nonpeptide antigens by ␥␦ T cells. As show in Fig. 3B, the cysteine in the V or leucine in the J fragment of CDR3␦2 did not affect the binding of ␥␦ TCR to non-peptides, suggesting that both residues are indispensable in binding activity of ␥␦ TCR to peptide antigen.
Engineered Soluble ␥␦ TCRs Critically Depend on the V and J Fragment of CDR3␦2 for Binding to Tumor Cells/Tissues and Ligands-We used a third approach to examine the interaction between ␥␦ TCR and its target cells/tissues or protein ligands. We engineered chimeric proteins composed of the extracellular domains of human ␥9 and ␦2 chains containing OT3 or its V/D/J mutants fused to the hinge region, CH2 and CH3 domains of human IgG1 heavy chain and expressed the fusion protein in eukaryocytes. We examined the binding capacity of these soluble ␥␦ TCR to tumor cells by FACS and confocal microscopy analyses. In contrast to the high percentages of SKOV3 cells (90%) and other tumor cells (60 -65%) stained positive with ␥9/␦2(OT3)-Fc fusion protein, both fusion proteins with V or J mutation in CDR3␦ stained a significantly lower proportion of the tumor cells (from 9 to 58%) in FACS analysis (Fig. 4A). Furthermore, confocal microscopy analysis showed dramatic decreases in fluorescent intensities stained by Vm and Jm fusion proteins when compared with wild-type OT3 fusion protein (Fig. 4B). In both analyses, the Dm fusion protein retained its strong binding to these tumor cells (Fig. 4, A and B). Importantly, similar binding activities of these g9/d2(OT3)-Fc fusion proteins to OEC specimens Confocal images are one representative of two independent experiments. Bars, 50 M. D, immunohistochemistry analysis of the binding activities of wild-type OT3 peptide or its V/D/J mutants to specimens (OEC and normal ovarian specimens). Synthesized V␤ CDR3 peptide MAB was used as a negative control. Binding was visualized using diaminobenzidine as the substrate (brown) (ϫ200). E, ELISA of the binding by wild type OT3 peptide and its V/D/J mutants to the N-terminal end protein of hMSH2 (MNS, 1-329 amino acids). 96-well plates were coated with 0.5 g of MNS in 0.1 M NaHCO 3 (pH 9.6). After blocking with 5% bovine serum albumin, the plates were incubated with biotin-conjugated CDR3␦2(OT3) peptide or its V/D/J mutants for 1 h at room temperature. The plates were developed using HRP-conjugated streptavidin (Pierce) and substrate (Sigma) and read on a microplate reader at 450 nm. Data are shown as mean of OD Ϯ S.D. of three independent experiments. F, SPR analysis of the binding affinity by wild-type OT3 peptide and its V/D/J mutants to MNS. About 6000 RU of MNS protein were immobilized on a CM5 chip. OT3 peptide and its V/D/J mutants as analyte were injected at the flow rate of 30 l/min. At each cycle, the sensor chip was regenerated by passing 10 mM Gly-HCl. The concentration of peptide analyte in SPR studies was 1:1 serial solutions from 700 M to 43.75 M. Data are representative of two independent experiments. were also observed (Fig. 4C). In addition, these engineered soluble ␥␦ TCR proteins also interacted with an MNS in ways similar to those displayed in tumor cell binding (Fig. 4,  D and E). Taken together, these results further support that V and J but not D regions in CDR3␦2 (OT3) critically contribute to ligand recognition.

DISCUSSION
Although ␥␦ TCRs have been identified for more than 20 years, their known ligands are still very limited in number. It is not clear why this is the case. Our data here may provide a possible explanation: ␥␦ TCRs critically depend on their conserved regions to recognize ligands, and the limited diversity in these regions correspond to a limited number of ␥␦ T cell ligands.
We have applied three approaches to determine the functional importance of the individual fragment of CDR3␦. First, we used wild-type CDR3␦2and its V/D/J mutant peptides to examine their ligand binding capabilities. Mutation in the V or J fragment dramatically impaired their staining capability to tumor cells or tissues, as well as hMSH2, a ligand for ␥9/␦2 TCR. Mutation in N-D-N region did not affect its binding to the same target. Second, we generated cells which expressed ␥␦ TCRs containing wild-type CDR3␦2 or its V/D/J mutants and examined their antigen recognition capability. Mutation in the V or J fragment impaired the binding capability of transfected cells expressing ␥␦ TCR, while mutation in the N-D-N fragment did not have this effect. Third, we examined ␥␦ TCR recognition using engineered chimeric proteins containing the extracellular domains of the human ␥9 and ␦2 chains with the CDR3␦2 or its V/D/J mutants. Again, the results suggest that the V and J but not D region contribute critically to ␥␦ TCR recognition. We also identified cysteine in the V fragment or leucine in the J fragment is the critical determinant in ␥␦ TCR antigen recognition of antigen. Taken together, this evidence strongly suggests that the conserved flanking sequences of the CDR3␦ domain play a key role in determining antigen binding activity.
The results of Adams et al. (13) demonstrated that the D gene segment is responsible for binding to T22 antigen for G8 ␥␦ T cells. Our results showed that the V/J segments of TCR ␦ are responsible for binding to ␥␦ TCR. The contradiction may be explained by two differences: first, the antigen recognized by ␥␦ TCR derived from human and mice was different. T22 was the antigen originating from mouse ␥␦ TCR with the OT3 sequence recognizing the human antigen. Second, the length of ␥␦ TCR CDR3 from human and mice was different. The length of mice ␥␦ TCR CDR3 was shorter than that of human. So our conclusion was that for tumor peptide antigen, V/J segments of TCR ␦, but not the D gene segment, are responsible for binding to ␥␦ TCR.   Our results have revealed an important aspect of ␥␦ TCR antigenic recognition. All heterodimers of ␣␤ TCR, ␥␦ TCR, and BCR are products of rearranging genetic segments. At the first glance, these three receptors are similar in structure and genetic composition. However, the mechanisms of diversity formation are quite different. Because of a large number of gene segments, the diversity of antibodies relies to a large extent on V, D, and J gene rearrangement (14). By contrast, there are only a few V, (D), and J segments encoding the ␣␤ TCR, so the use of D segments and N region addition greatly increase the available diversity of ␣␤ TCR (15). There are even fewer V, D, and J segments that encode the ␥␦ TCR. The potential diversity of ␥␦ TCR comes primarily from the use of D␦ segments in all their reading frames, and N region addition at three different positions. Comparing to 10 15 for the ␣␤ TCR repertoire, the total potential repertoire of ␥␦ TCR is about 10 18 of junctional diversity; however, to date only a handful of ligands for ␥␦ TCR have been identified. It appears paradoxical that a huge potential repertoire of ␥␦ TCR only recognizes a few ligands unless the potential repertoire is not utilized for the generation of specificity. Alternatively, the N-D-N fragment in ␥␦TCR may not play a critical role in determining antigen specificity. Our data clearly support the second possibility. The flanking sequences encoded by V and J fragments are more important in antigen recognition, since the mutation of V and J portion of the CDR3␦ almost completely abolished their binding to tumor cells or tissues, while mutation of N-D-N residues minimally affected ␥␦ TCR antigen recognition. Based on our results, we postulate that the flanking sequences may act as "rough-adjusters" for antigen recognition, while the highly diverse fragment of N-D-N may serve to fine-tune the affinity of ␥␦ TCR to certain antigens. Our finding supports the notion that the conserved fragment within CDR3␦ and the flanking sequences form a broad specificity to "rough-see" the molecules induced by stressful stimuli such as tumorigenesis or infection by recognizing that limited epitopes existed in these molecules. With evolution, such a "rough-see" way connects with "fine-see" way of ␣␤ T cells and B cells by delivery of co-stimulator signals. Both ␣␤ T cells and B cells are capable of distinguishing precisely a tiny and unique difference (epitope) in a given antigen from those of numerous foreign antigens. This very narrow specificity is based on the high diversities of antigenreceptor repertoires.
␥␦ TCR recognition may behave more similar to pattern recognition by innate immune cells. Macrophages and DCs can recognize non-self structures such as pathogen-associated molecular patterns (PAMPs) by pattern recognition receptor (PRR) such as Toll-like receptors (TLRs) (16). More recent studies strongly emphasize the innate features and functions of ␥␦ T cells, including the participation in wound healing (17), tissue repair (18), and the ability to present antigen (19 -20). Previously, we have used CDR3 peptide as probe to pan twelve peptide libraries. Nine peptides as putative epitopes were identified, among which seven were tumor-related and two were HBV infection-related. BLAST searches revealed that most matched proteins were conserved proteins of prokaryotes (7). Results suggest the possibility that ␥␦ TCR recognizes some conserved molecules. During the recognition, the flanking sequences within CDR3␦, the conserved fragment of ␥␦ TCR, might recognize readily their ligands via a pattern recognition model. Our finding may provide an explanation for the limited number of ␥␦ TCR ligands that have as yet been identified.