The Use of HLA A2.1/p53 Peptide Tetramers to Visualize the Impact of Self Tolerance on the TCR Repertoire¹

To avoid autoimmunity, the immune system has developed multiple strategies by which T cells can achieve tolerance to self Ags. The issue of self tolerance has increasingly become of concern in tumor immunology, since many promising candidates for tumor Ags are derived from proteins that are also expressed on normal tissue (1, 2, 3, 4, 5). However, not much information is available concerning the extent to which self tolerance has shaped the repertoire specific for these tumor Ags.

The first and arguably most effective form of self tolerance is thymic deletion, by which thymocytes recognizing self peptides undergo apoptosis (6, 7, 8). This process, however, is highly dependent on the avidity of T cell recognition and the amount of epitope presented in the thymus (9, 10, 11, 12). As a consequence, many T cells that recognize self epitopes with an avidity below a certain threshold are permitted access to the periphery (13, 14, 15). In addition to thymic tolerance, antigenic encounter in the periphery, in the absence of the appropriate immunostimulatory environment, can lead to deletion or anergy (16, 17, 18, 19, 20, 21). However, the presence of the Ag in the periphery does not always affect T cells, and some may remain ignorant of Ag (22). Again, this is likely to be determined by T cell avidity and the amount of Ag presented in the periphery.

We and others have been studying the feasibility of targeting p53 as a tumor Ag since it is overexposed in more than half of all human tumors (23). However, p53 is expressed in normal tissues, including thymus, spleen, and lymphohemopoietic cells (24, 25, 26), and may act as a tolerogen during thymic development and in the periphery (5, 27). Previously, we compared the HLA-A2.1-restricted response to p53 epitopes in p53 deficient (p53⁻) and sufficient (p53⁺) HLA-A2.1/K^b transgenic mice (27). The effect of tolerance varied for different peptides. CTL specific for the epitope spanning residues 187–197 were completely eliminated in p53⁺ A2.1/K^b mice (27). In contrast, a CTL response specific for the 261–269 epitope was detected in p53⁺ mice; however, it was of low avidity as compared with CTL from p53⁻ mice and required more than 10-fold more peptide to achieve the same levels of lysis (27). This low avidity repertoire is likely to represent the only CTL precursors available for activation by a putative vaccine or immunotherapeutic agent targeting p53 and, as such, merits further examination.

In this report we have used A2.1/peptide tetramers for further characterization of the residual low avidity A2.1/K^b-restricted p53 261–269-specific CTL by comparing the ability of CTL from p53⁺ and p53⁻ A2.1/K^b mice to bind A2.1/p53 261–269 tetramers (28, 29). The results demonstrate that this self tumor Ag promotes functional deletion of the CD8⁺ T cells carrying TCRs with the highest affinity for p53, resulting in CTL incapable of stable binding to the tetramer complexes.

All mice used in these studies are on a C57BL/6 background. p53⁺ A2.1/K^b and p53⁻ A2.1/K^b transgenic mice have already been described (27, 30). p53⁻ mice were obtained from Tyler Jacks (Massachusetts Institute of Technology, Cambridge, MA) and mated with A2.1/K^b transgenics. Progeny were interbred, and offspring were screened for mice that were p53^−/− and expressed A2.1/K^b. C57BL/6 mice were purchased from the breading colony of The Scripps Research Institute. Mice were propagated and maintained under specific pathogen-free conditions in our vivarium at The Scripps Research Institute. All experimental procedures were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

The T2 cell line, which is deficient in TAP, was propagated in RPMI 1640 media containing 10% FBS, 2 mM glutamine, 5 × 10⁻⁵ M 2-ME and 50 μg/ml gentamicin sulfate (complete media). EL4-A2.1/K^b and Jurkat-A2.1 transfectants have already been described (27, 30). They were propagated in complete media supplemented with 0.4 mg/ml G418 (Gemini Bio-Products, Calabasas, CA).

The procedure used to obtain peptide-specific CTL lines has been described (27). Briefly, mice were injected s.c. at the base of the tail with 100 μg of either murine p53 261–269 peptide or human p53 149–157 peptide along with 120 μg of the I-A^b T helper peptide 128–140 of the hepatitis B virus core protein in IFA. After 10 days, mice were sacrificed, and spleen cells were stimulated in vitro (using 24-well plates) with irradiated LPS-activated syngeneic spleen cells pulsed with the priming peptide at 5 μg/ml in complete media. On day 6, effector cells were assayed for their lytic activity. The resultant effector CTL lines were restimulated weekly with irradiated EL4-A2.1/K^b cells (0.5 × 10⁶ cells/well) pulsed with peptide and C57BL/6 spleen cells (6 × 10⁶ cells/well) as fillers in complete media supplemented with 2% supernatant from Con A-stimulated rat spleen cells (27). On day 4 after restimulation, cells were tested in a standard 4-h ⁵¹Cr release assay using T2 cells pulsed with different amounts of peptide as targets at the indicated E:T ratio.

p53⁺ A2.1/K^b 261 clone 12 and 13 were derived from p53⁺ A2.1/K^b 261 CTL line 3 by limiting dilution methods previously described (31). p53⁻ A2.1/K^b 261 clone 6 and 7 were derived by limiting dilution from the p53⁻ A2.1/K^b CTL line 2, previously sorted in sterile conditions into a CD8⁺ tetramer⁺ subpopulation. They were propagated as the parental CTL lines, with 5% instead of 2% rat Con A supernatant. A2.1 149 clone 5 and A2.1 261 clone 45 were derived by limiting dilution from human p53 149–157 and murine p53 261–269, respectively, specific A2.1 CTL lines described previously (27). They were propagated as mentioned above using peptide-pulsed Jurkat-A2.1 cells as APCs instead of EL4-A2.1/K^b.

Production of MHC/peptide tetramers was described in detail elsewhere (28). Briefly, a 15-amino acid substrate peptide for BirA-dependent biotinylation (BSP)³ has been engineered onto the COOH terminus of HLA-A2.1. The A2.1-BSP fusion protein and human β₂-microglobulin were expressed in Escherichia coli and were folded in vitro with the specific peptide ligand. The properly folded MHC-peptide complexes were extensively purified using fast performance liquid chromatography and anion exchange, and biotinylated on a single lysine within the BSP using the BirA enzyme (Avidity, Denver, CO). Tetramers were produced by mixing the biotinylated MHC-peptide complexes with PE-conjugated avidin (PharMingen, San Diego, CA) at a molar ratio of 4:1.

HLA A2.1/peptide tetramers used in this study contained either murine p53 261–269 (A2.1/p53 261–269 tetramers), or human p53 149–157 (A2.1/p53 149–157 tetramers). To test their specificity of binding, CTL clones specific for each of these A2.1 binding peptides (27) were stained with both A2.1/p53 261–269 tetramers-PE and A2.1/p53 149–157 tetramers-PE. A2.1 261 clone 45 demonstrated binding of A2.1 tetramers containing its nominal Ag, the murine 261–269 peptide, but not A2.1 tetramers containing the human p53 149–157. Reciprocally, the A2.1 149 clone 5 can be stained specifically by A2.1/p53 149–157 tetramers, but not by A2.1/261–269 tetramers.

On day 5 after stimulation, cells were partially purified through a Ficoll-Paque (Pharmacia Biotechnology, Uppsala, Sweden) cushion and then washed in HBSS. Cells (0.5 × 10⁶) were incubated with different combinations of the following staining reagents for 30 min at room temperature in HBSS containing 0.1% BSA and 0.05% sodium azide: A2.1 tetramers-PE described above at 40 μg/ml; anti-murine CD8α-FITC mAb 53-6.7, at 2 μg/ml; anti-murine pan-TCRβ-PE mAb H57-597, at 2 μg/ml; anti-murine CD11a-biotin mAb 2D7, at 1 μg/ml; and streptavidin-FITC, at 5 μg/ml. All reagents were supplied by PharMingen. Propidium iodide was added after the final wash at 1 μg/ml to exclude dead cells in all experiments. Samples were analyzed on a Becton Dickinson (San Jose, CA) FACSort apparatus at The Scripps Research Institute FACS facility. Twenty thousand events were collected and analyzed using CellQuest software (Becton Dickinson).

Poly(A)⁺ RNA from 4 × 10⁶ cells of the 261–269 peptide-specific CTL clones was extracted using MicroFastTrack Kit (Invitrogen, Carlsbad, CA) following the manufacturer’s instructions. cDNA was synthesized using SuperScript II Reverse Transcriptase (Life Technologies, Gaithersburg, MD) and oligo(dT)_12–18 primers (Life Technologies). cDNA (1/100) was used as a template for PCR amplification with 1 unit Taq polymerase (Life Technologies) in a Hybaid thermal cycler (Hybaid, Middlesex, UK). Methods and primer sequences for PCR-based screening of the TCR Vα usage have been previously described (32). PCR fragments obtained from CTL clones 7 and 13 were cloned into pCR2.1 vector using TA Cloning Kit (Invitrogen) and following the manufacturer’s instructions. M13 forward and reverse primers were used to determine sequences of three different molecular clones for each PCR fragment by the Nucleic Acids Core Facility at Scripps Research Institute using Applied Biosystems (Foster City, CA) automated sequencers.

Previous experiments suggested a lower avidity A2.1-restricted p53 261–269-specific CTL response in A2.1/K^b transgenics that express p53 than in p53⁻ A2.1/K^b mice, since the first needed a 10-fold higher concentration of peptide to achieve similar levels of lysis (27). As defined, avidity is a multifactorial property in T cell recognition, in which TCR and CD8 are main contributors to variability. Although murine CD8 can interact with the chimeric A2.1/K^b molecule that contains the α3 domain from the K^b molecule, it does not effectively interact with the intact HLA-A2.1 molecule (30). Accordingly, it is possible to evaluate the contribution of TCR affinity to the overall T cell avidity (in the absence of participation of CD8) by examining its interaction with the intact A2.1 molecule, rather than A2.1/K^b. To do so, we used two different approaches: 1) analysis of the lytic activity of targets expressing A2.1; and 2) analysis of the binding to A2.1/peptide tetramers.

We analyzed the lytic activity of 261–269-specific CTL lines from p53⁻ and p53⁺ A2.1/K^b using T2 cells loaded with peptide as targets (Fig. 1,A). The results obtained in the absence of CD8 contribution were similar to those obtained previously when the coreceptor was operative. CTL lines from p53⁻ mice required far less peptide than those from p53⁺ mice to achieve similar levels of lysis. There is approximately a 10-fold difference between lines 2 and 3 in the amount of peptide required to obtain 50% lysis (Fig. 1,A). The very same 261–269-specific CTL lines were tested for their ability to bind A2.1/p53 261–269 tetramers (Fig. 1,B). The p53⁻ CTL lines contained a subset of CD8⁺ cells capable of binding these tetramers, whereas p53⁺ lines did not (Fig. 1 B). An average of 21% of the CD8⁺ T cells from six different high avidity p53⁻ A2.1/K^b CTL lines analyzed could bind tetramers. In contrast, only an average of 1.2% of the CD8⁺ T cells from six different low avidity CTL lines from p53⁺ A2.1/K^b mice could bind tetramer. The level of staining observed using allospecific CTL, an average of 1% of the CD8⁺ T cells (data not shown), suggested this was the background level in our conditions. Also, no evidence for specific binding could be found among freshly isolated splenocytes from p53 261–269-immunized mice.

Despite their inability to bind A2.1/p53 261–269 tetramers, these CTL specifically lysed A2.1-expressing cells loaded with the 261–269 peptide (Fig. 1 A). Taken together with the results of the cytolytic assay using A2.1 targets, there is good correlation between high avidity, as defined by the lytic assay, and tetramer binding.

The poor binding and lysis exhibited by p53⁺ A2.1/K^b lines could be explained by the presence in these CTL populations of a very few T cells that are actually specific for this complex and capable of binding and cytolysis. To determine whether cells that did not demonstrate tetramer binding could actually be responsible for specific lysis, CTL clones were derived from p53⁺ A2.1/K^b CTL line 3 by limiting dilution (Fig. 2,A). As a control, clones were also isolated from the tetramer binding, CTL line 2 derived from a p53⁻ mouse (Fig. 2,A). Resulting clones were tested for both tetramer binding (Fig. 2,B) and lytic activity (Fig. 2,C). Clones 12 and 13 were unable to bind stably A2.1/p53 261–269 tetramers, whereas clones 6 and 7 demonstrated staining, although at different levels (Fig. 2,B). Both tetramer⁺ and tetramer⁻ clones were able to kill specifically p53 261–269 peptide-loaded T2 targets; however, they displayed different lytic capabilities (Fig. 2,C). A 10-fold difference in avidity in the dose-response curve translated into either strong tetramer binding as exhibited by clone 7 or no binding at all, as observed for clone 13 (Fig. 2, B and C). Clones displaying p53 261–269-specific lytic activity yet unable to bind tetramer were also isolated from a CD8⁺ tetramer⁻ subpopulation from p53⁻ CTL line 2 (data not shown), indicating the presence of both high and low avidity CTL in this population.

It has been shown previously that one way to achieve tolerance and a low avidity phenotype is by decreasing the density of the specific TCR on the cell surface (33, 34, 35). We therefore examined the levels of total TCR on the surface of both tetramer⁺ high avidity and tetramer⁻ low avidity clones, and no significant differences were observed (Fig. 3). Surface TCR levels were also equivalent in the CTL lines described above (data not shown). Even when total levels of TCR were similar, the expression of a second α-chain, which can compete for the same β-chain to produce another TCR with a different specificity, may reduce the actual density of the TCR specific for nominal Ag (36, 37, 38). To test this possibility, a fragment containing the V-J-C junctions region of the TCR α-chain mRNA from clones 7 and 13 was amplified and sequenced (Fig. 4). Two different α-chain mRNA were found in each clone, but only one was productively rearranged: Vα3-Jα17-Cα, in clone 7; and Vα1-Jα8-Cα, in clone 13 (Fig. 4). Analysis of the TCR Vα usage by PCR for clones 6 and 12 revealed identical results to that obtained for clones 7 and 13, respectively (data not shown). This suggested that they could be clonally related.

Finally, the expression of CD8 and LFA-1 was analyzed, since interaction with its ligands plays an important role in T cell avidity (30, 39, 40, 41, 42). Again, no significant differences were observed (Fig. 3). Taken together, these results suggest a difference in TCR affinity is responsible for the difference in lytic ability and tetramer binding observed between CTL from p53⁺ and p53⁻ mice. Thus, A2.1 tetramers can discriminate the distribution of affinities present in a CTL population. Only the highest affinity TCRs can be detected by FACS. It is also likely that the difference in intensity of tetramer staining between clones 6 and 7 (mean fluorescence intensity of 52 for clone 6, and 202 for clone 7) reflects differences in their TCR affinities (Fig. 2). However, since cell surface TCR expression of clone 6 is 50% lower than that of clone 7 (Fig. 3), the lower level of tetramer binding may also reflect a difference in TCR density.

To ascertain that the differences between the CTL obtained from p53 expressing and p53 deficient mice were actually related to the fact the epitope under examination induced tolerance in the p53-expressing mice, we examined the CTL responses by these same mice against a peptide epitope that is not expressed in either strain of mouse. The human p53 149–157 epitope is also A2.1 restricted, yet its sequence differs from the murine molecule by 3 amino acids (27, 43). The avidity of the CTL response is almost identical in both p53⁻ A2.1/K^b and p53⁺ A2.1/K^b mice, showing that A2.1/K^b mice are not tolerant of this peptide (27). 149–157-specific CTL lines from both types of mice were obtained. Their lytic activity on T2 targets and their ability to bind specific A2.1/p53 149–157 tetramers were analyzed (Fig. 5). When using T2 targets, the avidity of p53⁻ A2.1/K^b and p53⁺ A2.1/K^b CTL was identical (Fig. 5,A), and the tetramer staining patterns were highly comparable (Fig. 5 B). All CTL lines tested from both p53⁻ A2.1/K^b and p53⁺ A2.1/K^b mice contained CD8⁺ T cells capable of binding A2.1/p53 149–157 tetramers.

Understanding the mechanisms of tolerance and its consequences to the T cell repertoire specific for self epitopes is of particular importance in autoimmunity and tumor immunity. Many potential tumor Ags are known to be expressed in normal as well as transformed cells (1, 2, 3, 4, 5). Our studies with p53 have indicated that tolerance is achieved by functionally eliminating T cells with high avidity (27). It is these residual low avidity CTL that constitute the repertoire available to a potential vaccine or immunotherapeutic agent. The goal of this study was to evaluate the contribution of each of several different variables, known to alter T cell avidity, in effecting tolerance to p53. Included among these was density of TCR specific for the tolerogen, and alterations in nonclonotypic molecules that alter T cell avidity, such as the CD8 coreceptor and the LFA-1 adhesion molecule (30, 33, 39, 40, 42). Although no evidence was obtained for alteration in expression of nonclonotypic molecules, by using A2.1/p53 261–269 tetramers, it was immediately apparent that there was a significant difference in CTL from p53⁺ and p53⁻ mice in their TCR binding of this self Ag.

Whereas p53 261–269-specific CTL obtained from p53⁻ mice demonstrated a population of CD8⁺ T cells capable of binding A2.1/p53 261–269 tetramers, CTL of similar specificity obtained from p53⁺ mice did not demonstrate stable binding of these tetramers. The small numbers of CD8⁺ tetramer⁺ cells that were detected in p53⁺ A2.1/K^b CTL lines (1%) most likely represent background staining, since similar numbers of tetramer-stained cells were found in alloreactive CTL. Also, whereas p53⁻ A2.1/K^b CTL lines maintained in vitro for prolonged periods of time demonstrated an enrichment of the tetramer-binding subpopulation, no such increase in the numbers of tetramer-binding CD8⁺ T cells was observed in the CTL populations from p53⁺ mice (data not shown). One possible explanation for the lack of high avidity (tetramer⁺) CD8⁺ T cells in the 261-specific CTL lines from p53⁺ mice is related to the relatively high concentration of peptide used for their propagation in vitro. It has been shown that high concentrations of peptide may induce apoptosis of high avidity CTL (44). However, since identical concentrations of peptide were used to propagate 261-specific CTL from p53⁻ mice, which were capable of strong tetramer binding, it is unlikely this can explain the lack of tetramer binding among 261-specific CTL from p53⁺ mice. Also, 149-specific CTL from both p53⁺ and p53⁻ mice stimulated under the same conditions contained high avidity CD8⁺ T cells capable of tetramer binding.

Our studies found no evidence for down-regulation of TCR levels or expression of a second TCR as a mechanism for decreasing T cell avidity in p53⁺ mice. The later was somewhat surprising since it is well documented that more than one TCR can normally be expressed by T cells, thereby decreasing the actual concentration of the TCR specific for the first epitope (36, 37, 38, 45, 46). It has been shown in TCR transgenic models how down-regulation of the total levels of TCR may allow T cells to escape tolerance (33). Also, it was shown using TCR transgenics that cells expressing an autoreactive TCR can escape thymic deletion if they express a second TCR (34, 35). Perhaps more extensive analysis of T cell clones from p53-expressing mice will reveal CD8⁺ cells that have escaped tolerance by expression of a second TCR. Alternatively, it is possible that, for this particular Ag, it was not possible to achieve sufficiently low avidity to avoid deletion by these alternative mechanism.

p53 261–269-specific effector CTL from p53⁻ mice and p53 149–157-specific CTL from either p53⁻ or p53⁺ mice contain a mixture of both specific tetramer binding and nonbinding CD8⁺ T cells. These results indicate that, in response to a foreign peptide Ag, a heterogeneous set of TCR affinities may arise, including CTL with a relative high affinity TCR, as well as very low avidity CTL. Furthermore, the use of MHC tetramers represents an important tool to separate CTL by avidity. Several reports have correlated differences in the intensity of tetramer staining with differences in TCR affinity for the Ag for both class II and class I MHC tetramers (47, 48, 49, 50). Interestingly, in two of these reports it has been demonstrated that a higher intensity of tetramer staining reflects a lower tetramer dissociation rate (49, 50). Probably, the dissociation rate of the A2.1/p53 peptide tetramers is too high to detect binding to the low affinity p53-specific TCR by FACS. However, the difference in avidity, as measured in a cytolysis assay, between the CTL that bind tetramer and those that cannot is quite narrow, no higher than 10-fold.

The existence of CD8⁺ T cells within p53 261–269 peptide-specific CTL lines, which are not able to bind tetramers yet display specific lytic activity, contrasts with results from several other studies using HLA A2.1, Mamu-A0.1, or K^b tetramers. It was reported that the tetramer⁺ but not tetramer⁻ cells from human, monkey, or mouse effector CTL were able to lyse peptide-pulsed targets (28, 51, 52, 53). An explanation for these different findings is that human CD8⁺ T cells bind A2.1 tetramers better than murine CTL do. In previous studies, some contribution to the avidity of tetramer binding may have come from CD8 (M. M. Davis, unpublished observation). In the current study, since murine CD8 does not bind the human A2.1 molecule, this interaction was not possible and therefore could not increase the avidity of tetramer binding to the CTL. This suggests that, in the absence of CD8 contribution, cytotoxicity assays are more sensitive than tetramer binding in detecting murine A2.1-restricted 261–269-specific CTL.

The experiments presented here suggest the occurrence of functional elimination of the CD8⁺ T cells carrying TCRs with relatively high affinity for the Ag, when it is expressed as a normal self Ag. Since there is expression of p53 in the thymus (24, 25, 26), it is likely that high avidity interaction with physiological amounts of the 261–269 peptide on thymic APCs results in negative selection. On the other hand, thymocytes expressing TCRs that have low affinity for this peptide may be positively selected in the same environment. Using TCR transgenic mice, several studies have demonstrated that weak agonist peptides are capable of positive selection (10, 54, 55, 56, 57). It is interesting to speculate that the 261–269 endogenous peptide may have contributed to the selection of some of the peptide-specific CTL obtained from p53⁺ mice.

In summary, this study shows the mechanism by which functional tolerance to a natural self epitope has been achieved. TCR affinity for a given MHC class I/peptide complex appears to be the main factor determining the overall avidity in CTL recognition for p53. In order for a potentially autoreactive T cell to be maintained in the repertoire, its TCR affinity must be kept under a certain threshold. The existence of a residual p53-specific, low avidity effector CTL could provide a window of opportunity for tumor rejection. Promisingly, two recent studies have demonstrated it is possible to activate CTL specific for self epitopes, such as p53, and get tumor rejection by the use of viral vectors or dendritic cells as vaccines (58, 59). Future studies will compare the ability of CTL-expressing high and low affinity TCRs to eliminate tumors in vivo.

We thank Judy Biggs, Karen Holst, and Alice Ko for excellent technical assistance, and Carol Wood for secretarial assistance.