The γδ T cell clone LBK5 recognizes the MHC molecule IEk. Here, we demonstrate that the affinity of this interaction is weaker than those typically reported for αβ TCRs that recognize peptide/MHC complexes. Consistent with our previous finding that peptide bound to the IE molecule does not confer specificity, we show that the entire epitope for LBK5 is contained within the polypeptide chains of the molecule, centered around the polymorphic residues 67 and 70 of the IE β-chain. However, LBK5 recognition is profoundly influenced by the N-linked glycosylation at residue 82 of the IE α-chain. Since infected, stressed, or transformed cells often change the posttranslational modifications of their surface glycoproteins, this finding suggests a new way in which γδ T cell Ag recognition can be regulated.
It has been shown that γδ T lymphocytes contribute to the host’s immune competence, but the precise role of γδ T cells in the immune system remains to be elucidated. This is in part due to the difficulty of identifying the specific targets and effects of γδ T cells in pathological situations (reviewed in Refs. 1–3). In a few infectious disease models, γδ T cells accumulate before the appearance of αβ T cells (4, 5, 6), suggesting that they are, in fact, responding to pathogens. In other situations, however, γδ T cells accumulate within inflammatory lesions late in the infection after the pathogens have been cleared (7, 8, 9). In these cases, attempts to demonstrate pathogen-specificity for the accumulated γδ T cells have failed. Similary, some γδ T cells can kill virus-infected cells in vitro, although recognition is not virus specific (10). In addition, γδ T cells are found to respond to testicular inflammation caused by an autoimmune mechanism as well as a bacterial infection (11). Based on these observations, it has been suggested that γδ T cells can respond to Ags induced upon cells that have been damaged and/or stressed from infection. Thus, γδ T cells may recognize induced or modified self Ags.
Although γδ T cell clones that recognize classical and nonclassical MHC molecules, such as LBK5 (specific for IE) and G8 (specific for T10/T22) (12, 13) have been used as model systems to elucidate the recognition requirements of γδ TCRs (14, 15), the physiological role for these specificities, as well as to what degree the characterization of their epitopes is relevant for γδ T cells in general, has been questioned. This uncertainty is in part due to the observation that most γδ T cells are not MHC-restricted and, further, that the frequency of γδ T cells, responsive to MHC or MHC-like molecules in a mixed lymphocyte reaction, is lower than that of αβ T cells. However, MHC-like molecules, such as T10 and T22, have been found to be the natural ligand of γδ double-negative thymocytes (16). More recently, Groh et al. (17) have shown that γδ T cells in the intestinal epithelium recognize the MHC-like proteins MICA and MICB, expressed by enterocytes. Despite the fact that some of these molecules have the potential to bind peptides, recognition does not appear to be peptide-specific and responsive γδ T cells are not restricted in their recognition to a particular MIC allele. Thus, γδ T cell recognition of MIC resembles LBK5 and G8 recognition of IE and T10/T22, respectively (14, 18). In fact, the characterization of the LBK5 epitope, described here, may be a blueprint for all γδ T cells that recognize MHC-like proteins. Since peptides bound to these MHC molecules do not confer specificity, and since some of these MHC-like molecules do not even bind peptide (18), it is important to analyze how these MHC or MHC-like molecules are recognized. This analysis should provide a better understanding of how self Ags may be ignored or responded to by γδ T cells.
Here, we present evidence showing that the affinity of the interaction between the LBK5 TCR and IE appears to be considerably weaker than the affinities of TCR-αβs that recognize peptide in association with IE. Interestingly, although the functional epitope of LBK5 is encoded by the polypeptide chains of IE and centers around residues 67 and 70 of the IE β-chain, LBK5 recognition of IE is sensitive to changes in the N-linked glycosylation on the end of the α-helix of the IEα domain at position α82. This type of protein-protein interaction has been postulated to ensure the specificity of a recognition event without the need for a high affinity (19). Since cells that are stressed, infected, or transformed often change the posttranslational modifications of their surface proteins, our data suggest a way to regulate a γδ T cell response by qualitative changes of self Ags.
Materials and Methods
cDNA and transfectants
Site-directed mutagenesis of full-length IEk α- and β-chain cDNA was conducted as described (20). Oligonucleotides for mutagenesis are as follows: α84V, T→V 5′-CAACGTTCCAGATGCCAACG-3′; α79D, E→D 5′-GATGTCATGAAAGATAGATCTAACAACACTCCAGAT-3′; α79L, E→L 5′-GATGTCATGAAATTGCGGTCCAACAACACTCCAGAT-3′; β67I, F→I 5′-TGGAACAGCCAGCCGGAGATCCTCGAGCAAAAGCGG-3′; β70D, Q→D 5′-CCGGAGTTCCTCGAGGACAAGCGGGCCGAG -3′; β71A, K→A 5′-CCGGAGTTCCTCGAGCAAGCACGGGCCGAGGTG -3′.
Cells expressing mutant IE molecules were generated by either cotransfecting mutated IEk α-chain in pBJ1neo with native IEk β-chain in pBJ1 or by cotransfecting mutated IEk β-chain in pBJ1neo with native IEk α-chain in pBJ1. Transfectants were selected with the following concentrations of G418 (Life Technologies, Rockville, MD): 0.8 mg/ml for Chinese hamster ovary (CHO)8 cells and 0.75 mg/ml for 721.134. Resistant cells were stained with the anti-IE mAb 14.4.4 (21) and the brightest staining 5% sorted by FACS (FACStar; Becton Dickinson, Mountain View, CA) into microtiter wells, expanded, and rescreened by FACS. All transfectants expressed similar levels of IE.
T cell activation assay
T cell activation assays were conducted with 5 × 104 T cells and the indicated number of stimulator cells in 0.2 ml supplemented RPMI medium at 37°C. Temperature-sensitive CHO cell mutants were incubated at the indicated temperature for 6 h to induce the defect before the addition of Ag and T cell hybridomas. Ag solutions and cell suspensions were prewarmed to 34°C or 39°C before the addition to the CHO transfectants. IL-3 was assayed with R6X cells (22). In some experiments, T cell activation assays were conducted with LBK5 hybridoma transfected with the plasmid NFAT-SX (which contains the alkaline phosphatase gene driven by the IL-2 gene NFAT promotor enhancer) by assaying for secreted alkaline phosphatase (14). In other cases, LBK5 cells were transfected with the plasmid NFATZ, which contains the β-galactosidase (β-gal) gene driven by the IL-2 gene NFAT promotor enhancer (23). T cell activation was visualized by measuring the amount of NFAT-specific β-gal using the fluorescent substrate 4-methylumberiferyl-galactoside (Sigma, St. Louis, MO), as described (24). Each dose-response curve was reproduced in at least three independent experiments.
A total of 5 × 104 LBK5 cells/well were stimulated in 0.2 ml at 37°C with either detergent-solubilized and affinity-purified native IEk from the B cell lymphoma CH27 (14) or Escherichia coli-produced, in vitro-folded moth cytochrome c (MCC)/IEk complexes (25). Purified E. coli IEk heterodimers were site-specific biotinylated in vitro with the BirA enzyme at the engineered biotinylation site at the carboxyl terminus of the E α-chain (26). Various amounts of MCC/IEk complexes in PBS were added to each well of a microtiter plate (Immulon IV; Dynatech Laboratories, Chantilly, VA), which, in some cases, had been preincubated with 50 μg/ml streptavidin in PBS at 4°C for 18–24 h. The amount of oriented IE in the wells was determined by an ELISA using 10 μg/ml 14.4.4 Ab followed by a goat anti-mouse IgG alkaline phosphatase conjugate (Sigma), diluted 1/2000. Each incubation step of the ELISA was performed in 100 μl/well PBS + 2% BSA + 5 mM EDTA for 2 h at 4°C.
MCC/IEk tetramer staining reagent is made by conjugation of the biotinylated MCC/IEk complexes with PE-labeled streptavidin (Molecular Probes, Eugene, OR) as described (27). For analysis, 5 × 105 T cells per sample were first incubated for 30 min on ice with FITC-conjugated anti-CD3 (PharMingen, San Diego, CA) followed by ∼0.5 mg/ml tetramer for 60 min on ice. Propidium iodide (PI) at 1 μg/ml was included in the final wash. Flow cytometry was immediately performed on a FACStar (Becton Dickinson). A total of 10,000 events was collected and analyzed using the DESK software (Stanford University, Stanford, CA). Cells were gated according to forward/side scatter characteristics and PI staining.
Isoelectric focusing (IEF) gel electrophoretic analysis
A total of 107 IEk-expressing cells was labeled overnight with 400 μCi of 35S-methionine in 10 ml methionine-free medium containing 2% dialyzed FCS. After labeling, cells were harvested with PBS containing 2 mM EDTA, washed twice in PBS, and incubated with the mAb 14.4.4 for 2 h on ice. The cells were then washed in PBS again and lysed in 0.5 ml PBS containing 1% Nonidet P-40, 10 μg/ml leupeptin, 1 μg/ml pepstatin A, and 1 mM PMSF (Sigma). Detergent lysates were spun for 30 min at 10,000 × g, and the supernatants were incubated with 50 μl of a 20% protein A-Sepharose suspension for 90 min at 4°C. The beads were washed once in PBS containing 10% sucrose, 0.1% Nonidet P-40, 10 μg/ml leupeptin, 1 μg/ml pepstatin, and 1 mM PMSF, and four times in PBS containing 0.1% Nonidet P-40, 10 μg/ml leupeptin, 1 μg/ml pepstatin A, and 1 mM PMSF. After washing the beads once in PBS alone, the immunoprecipitated MHC molecules were separated on a IEF gel using a pH gradient ranging from 3.5 to 7.0 (Pharmacia, Piscataway, NJ). Fixed gels were treated with Amplify (Amersham, Arlington Heights, IL), dried, and exposed to Kodak (Rochester, NY) XAR film.
The LBK5 γδ TCR-IE interaction is weaker than that of αβTCR-peptide/IEk complexes
Recently, tetrameric peptide/MHC complex has been successfully used to identify T cells of different Ag specificities. The feasibility of this procedure to analyze γδ T cell ligand interaction has also been demonstrated by staining the γδ T cell G8 with a tetramerized version of its ligand (M. P. Crowley and Y. Chien, manuscript in preparation). Because it has been reported that the binding level of the tetrameric peptide/IEk complex to αβ T cells can be correlated with the affinity of the TCR for the monomeric peptide/IEk (28, 29), we thought to estimate the binding affinity of the LBK5 γδ TCR with its ligand IE by staining the LBK5 hybridoma with a tetrameric staining reagent of MCC peptide (88–103)/IEk complexes. To our surprise, no significant binding of the IE-tetramer to LBK5 cells was observed. In the same experiment, the reagent stains the αβ T cell 5cc7 (specific for MCC/IEk). The TCR expression levels of these two cell populations are similar, as determined by an anti-CD3 Ab (Fig. 1). The coreceptor CD4, which is expressed on 5cc7 cells, has not been found to contribute in any significant way to the binding of MHC tetramers (Ref. 29 and 30; and J. Hampl, unpublished observations).
Since the tetrameric MCC/IEk staining reagent is made with E. coli-produced, in vitro-folded IEk, we tested whether this material could be recognized by LBK5. As shown previously, native IEk molecules isolated from the B cell lymphoma, CH27, when bound to microtiter plates, can stimulate LBK5 cells to the same degree as the CH27 cells, indicating that the IE molecule alone, and no other signal, is required for T cell activation (14). Therefore, we tested whether immobilized E. coli-produced MCC/IEk complex can be similarly recognized. Because the native form of the IE molecules was loaded on the plate in the presence of detergent (to prevent aggregation of the protein with its transmembrane region), the majority of the plate-bound molecules are oriented with the membrane distal portions of the molecule facing upwards. In an effort to orient the soluble, E. coli-produced form of MCC/IEk complex in the well, we first coated the plate with streptavidin and subsequently added MCC/IEk complexes, which had been specifically biotinylated at the IEα2 domain. E. coli-produced, in vitro-folded IEk can stimulate LBK5 to the same extent as native IEk isolated from CH27 cells when it is oriented properly (Fig. 2,A). However, LBK5 fails to respond to MCC/IEk complexes randomly bound to the plate (loaded into plain wells without streptavidin coating) (Fig. 2,B). The αβ T cell 5cc7 responds to MCC/IEk bound to wells without orientation (25, 31). An ELISA using mAb 14.4.4, which blocks LBK5 activation, demonstrates that there is a higher density of IEk in streptavidin-coated wells, as compared with uncoated wells (Fig. 2 C). These data indicate that the activation of LBK5 requires a high ligand density, which is only achieved when the molecules are properly oriented on the plate. Thus, the lack of staining of LBK5 cells by the tetrameric reagent is likely caused by a lack of sufficient avidity and not a lack of recognition. This finding suggests that the affinity of the LBK5 TCR for IE is considerably lower than those reported for most TCR-αβs and their ligands.
The LBK5 functional epitope is located on the IE β-chain
Previously, we analyzed the interaction between the LBK5 TCR and IE by analyzing the response of LBK5 to a panel of 13 CHO cell lines, each expressing a mutant IEk molecule with a single amino acid substitution at a residue predicted to be located on the MHC α-helices and pointing up toward the TCR-αβ (20). Consistent with the observation that peptides bound to the IE molecule do not confer specificity, mutations centrally located on the helices that affect αβ T cell recognition have no effect on the LBK5 response (14). To further identify critical residues on the IE surface that contribute to the specificity of the LBK5 TCR, which recognizes IEk,b,s, but not IEd (12), we focused our attention on the IE β-chain, because the α-chain of IE molecules is practically monomorphic. Examining the residues that are different between IEk,b,s and IEd and solvent exposed, we generated three more mutations on the IE β-chain, substituting the amino acid residues of the b, k, and s haplotype to that of the d haplotype: β67(F→I), β70(Q→D), and β71(K→A). The location of these residues within the IE molecule, as well as Eα82 (see below), are indicated in Fig. 3,A. In addition, the positions of the previously tested 13 point mutations are also marked. IE molecules carrying the β70 mutation, expressed on CHO cells, stimulate LBK5 suboptimally, and the conversion of a F into an I at position β67 ablated recognition completely. IE β71A is recognized reasonably well (Fig. 3 B). All three mutant IE molecules can be recognized by anti-IE Abs 14.4.4 and 17.3.3 and by several other mAbs specific for IEk/MCC complexes (data not shown). They can also present superantigen SEA to αβ T cell hybridomas and stimulate T cells from 2B4 transgenic mice (data not shown), indicating that their overall structure is not altered. These results suggest that these closely spaced amino residues constitute the “functional epitope” for LBK5 recognition and explain why LBK5 recognizes IEk,b,s, but not IEd.
LBK5 does not recognize IEk with altered posttranslational modification
In analyzing the Ag specificity of LBK5 cells, we have found three lines of evidence that indicate that the posttranslational modifications of the IE molecule influences its recognition by LBK5. These experiments are discussed below.
Previously, we tested whether functional Ag processing pathways in the stimulator cells are required for the recognition of IE by LBK5 by expressing IEk on temperature-sensitive CHO mutant cell lines with endosomal acidification defects of the End1, -2, and -3 complementation groups grown at nonpermissive temperature (39°C). All three were found to be fully stimulatory to LBK5 (14). In contrast, the response of LBK5 to the IEk-transfected CHO mutant of the End4 complementation group, which has a temperature-sensitive endoplasmic reticulum to Golgi transport defect (32), is impaired (Fig. 4 A). The level of IEk expressed on this mutant cell line is indistinguishable from that of the wild-type CHO cells at either the permissive (34°C) or nonpermissive temperature (39°C). At 39°C, IEk-expressing End4 cells can be recognized by anti-IE Abs and present peptide as well as SEA to αβ T cells (data not shown). These results suggest that the reduced ability of IEk-End4 cells to stimulate LBK5 is not due to the lack of functional IEk molecules on the surface.
Since LBK5 recognizes IE regardless of the peptide bound to the IE molecules, we were surprised to discover that the human lymphoblastoid cell line 721.134, which has been transfected with and expresses wild-type IEk, is unable to stimulate LBK5 (Fig. 4 B). This cell line has been widely used because of its defect in a peptide transporter gene important in the class I Ag processing pathway. IE-expressing 721.134 cells can process proteins and present peptide and SEA to IE-restricted αβ T cells and can be recognized by all the anti-IE Abs (data not shown), indicating that the IE molecules expressed on the cell surface are functional. Since LBK5 can respond to IE-expressing RMAS and T2 cells (14), which also have defects in the class I processing pathway, this lack of recognition is unlikely to be due to the lack of a functional class I Ag processing pathway in 721.134.
Although the functional epitope of LBK5 is mapped on the β-chain, we have observed that a glutamic acid to lysine change at position 79 of the α-chain (α79(E→K)) abolishes the recognition of LBK5 (14). This mutant did not affect any of >40 αβ T cells tested. To test whether α79 is also part of the functional epitope of LBK5, two other mutations at this position were introduced to change the glutamic acid residue to either an aspartic acid or a leucine (conservative and semiconservative amino acid exchange). Each mutated α-chain was expressed together with the native β-chain in CHO cells. As shown in Fig. 4 C, neither of these mutations has any effect on LBK5 recognition. Therefore, the inability of the α79 Lys mutant to stimulate LBK5 is unlikely due to a loss of the epitope.
The endoplasmic reticulum to Golgi transport defect in End4 CHO cells is known to have an impact on the carbohydrate addition of glycoproteins. Glycoproteins isolated from these cells show defects in N-linked glycosylation and remain sensitive to Endo H digest (32). Interestingly, 721.134 is also known to express glycoproteins with altered carbohydrate structures. 721.134 which was generated by γ-ray irradiation, is defective in the expression of Glyoxalase I in addition to HLA (33). Furthermore, position 79 of the IE α-chain is located near the carbohydrate addition site α82. It has been observed that changes in the peptide structure, or microenviroment can influence the extent of sialation and branching of oligosaccharides at this glycosylation site (34). The Glu to Lys mutation at α79 of IE α-chain represents a charge reversal, and may therefore influence the carbohydrate addition at position 82. Hence, it is possible that the failure of all of these IE molecules to stimulate LBK5 is caused by an altered carbohydrate structure at this position.
To test this, we first compared the charge distributions from IEk isolated from CHO cells with IE molecules from α79 (Gly→Lys) IE/CHO cells, IE-721.134 cells, and IE-End 4 cells (grown at either 34°C or 39°C) on an IEF gel. As shown in Fig. 5, IEk immunoprecipitated from cells that failed to stimulate LBK5 lack the most acidic band(s). IE α-chain has two potential N-linked glycosylation sites, 82 and 122. The analysis of oligosaccharide structures of IEkα isolated from a B cell lymphoma AKTB showed that the carbohydrates at the position 122 are less branched and less siaylated as compared with those at the 82 site (34). Consistent with this, we find that surface immunoprecipitated IEk with the Thr residue at α84 changed to a Val (α84V mutant) to abolish the glycosylation site at the α82, lacks the three most negative charged spots when resolved on a two-dimensional gel (data not shown). Thus, the lack of the most acidic bands of the IE molecules, analyzed by IEF in Fig. 5, is likely due to an alteration in the carbohydrate at α82 site.
Recognition of IE expressed on 721.134 cells can be restored by removing the carbohydrate at position α82
Despite the striking correlation between the IE molecules that cannot stimulate LBK5 and an alteration of the glycosylation at IE α82 site, a particular carbohydrate structure is not required for LBK5 recognition. This is demonstrated by the observation that the E. coli-produced, in vitro-folded IE can stimulate LBK5 (Fig. 2) and that the IEα84V mutant expressed with wild-type IE β-chain in CHO cells is fully stimulatory to LBK5 (data not shown). This suggests that the failure of IE expressed on 721.134 cells to stimulate LBK5 may be due to a negative effect. Therefore, we expressed the α84V mutation in 721.134. As shown in Fig. 4,B, the transfectant gains the ability to stimulate LBK5 cells. This indicates that the 721.134-specific posttranslational modification of the IE molecule at position α82 interferes with its recognition by the γδ T cell LBK5 (Fig. 3A).
In this study, we confirm and extend the previous finding that peptides bound to IEk do not confer specificity for the IE recognition by LBK5 by showing that E. coli-produced IEk folded in vitro around the MCC peptide 88-103 stimulates LBK5 cells to the same degree as IEk isolated from mammalian cells. More strikingly, however, is the fact that, although LBK5 cells can be activated by E. coli-produced IE molecules bound to a solid surface, the same molecules, when assembled into a tetramer, fail to remain bound to the LBK5 cell surface for more then a few seconds (∼30 s). Only a small percentage (∼2.5%) of the cell population may exhibit some weak tetramer staining. This observation indicates that the LBK5/IE receptor ligand interaction has a fast off-rate and probably an affinity that is considerably lower than those of most TCR-αβs for their ligands, i.e., 10–100 μM. The low affinity may reflect a general mechanism for a tight control of γδ T cell Ag recognition to avoid activation by self Ags that are constititutivly expressed at high levels.
Although IE expression can be induced under certain physiological conditions and, therefore, regulate LBK5 recognition at the level of protein expression, our data suggest that changes in the posttranslational modification of the ligand can also regulate recognition. Thus, the quality of the ligand contributes significantly to the specificity of the recognition. A similar scenario may also exist for the recognition of the nonclassical class I MHC molecule T10/T22 by the γδ T cell G8. In this case, E. coli-produced and refolded T10/β2-microglobulin complex can stimulate G8 very efficiently (18). However, the same ligand expressed on Drosophila cells, which has the insect cell specific carbohydrate attached, is poorly recognized (15). It should be noted that it is formally possible that the alteration of carbohydrate structures might permit the engagement of a “negative receptor” on LBK5. However, we consider this possibility to be rather unlikely in that three different types of alteration of the N-linked glycosylation expressed on two different cell types lead to the nonstimulatory phenotype. In addition, all αβ T cell hybridomas tested (>40) can recognize peptides in association with IE molecules expressed on these transfectants (Refs. 14 and 20, and data not shown).
In this study, we have mapped the functional epitope of LBK5 to residues 67 and 70 on the helical region of IEβ. The observation that mutations at these positions reduce the recognition drastically or abolish it all together suggests that these amino acids contribute significantly to the binding affinity. This is in contrast to the carbohydrate at IE α82 and especially its negative charges, which are critical for specificity but must contribute very little to the binding per se, as removing the carbohydrate at this position does not affect recognition. Although the binding surface of γδ TCRs have yet to be determined, it is reasonable to assume that they will be similar to those reported for Abs and αβ TCRs (35, 36, 37). From the coordinates of the published crystal structure of HLA-DR1 (38), the distance between IEβ 67-70 (the LBK5 epitope) and the carbohydrate attachment site at position Eα82 is estimated to be around 28–33 Å. Hence, it is likely that the carbohydrate structure is peripheral to the LBK5 epitope of this molecule. This type of interaction may be similar to that described for human growth hormone and the extracellular domain of its receptor (39). X-ray crystal structure and mutational and binding analysis of this receptor ligand pair show that the protein-protein interaction consists of a central hydrophobic region at the contact site, dominated by the two tryptophan residues, which account for more than three quarters of the binding-free energy. Peripheral to this “functional epitope” are the electrostatic contacts that contribute substantially to the specificity of binding, but not to the net binding energy. This analysis lends support to the general contention that electrostatic interactions confer specificity to biological recognition without necessarily being energetically favorable (19).
Changes in the posttranslational modification of surface glycoprotein are an indication that a tissue has been infected, undergone neoplastic transformation, or has experienced some other type of cellular stress. For example, it has been shown that infection of mice with Listeria monocytogenes impairs sialic acid addition to host cell glycoproteins, including MHC molecules (40). In breast, ovarian, and pancreatic carcinomas, the peptide backbone of the surface glycoprotein mucin is unmasked by underglycosylation, while these molecules are heavily glycosylated in normal cells, making mucin a common cancer marker (41). Here, we report that the recognition of a glycoprotein by a γδ T cell is acutely sensitive to changes in the glycosylation of the ligand, suggesting a novel way in which Ag recognition by γδ T cells can be regulated.
We thank Dr. J. D. Altman for protocols and inclusion bodies of the E. coli IEk protein; Dr. Mark M. Davis for discussion and critical reading of the manuscript; and Loan Nguyen for technical help.
This work was supported by grants from the National Institutes of Health (to Y.C.).
Abbreviations used in this paper: CHO, Chinese hamster ovary; PI, propidium iodide; IEF, isoelectric focusing; MCC, moth cytochrome c; SEA, superantigen A; β-gal, β-galactosidase.