Abstract
The T cell αβ receptor (TCR) recognizes foreign peptide antigens bound to proteins encoded in the MHC. The MHC portion of this complex contributes much to the footprint of the TCR on the ligand, yet T cells are usually very specific for individual foreign peptides. Here, we show that the development of peptide-specific T cells is not intrinsic to thymocytes that undergo thymic-positive selection but is an outcome of eliminating, through negative selection, thymocytes bearing TCRs with extensive peptide cross-reactivity. Hence, thymic-negative selection imposes peptide specificity on the mature T cell repertoire.
Mature T cells bearing αβ T cell receptors (TCRs) react with cells expressing host MHC proteins bound to foreign antigens, usually in the form of peptides. In healthy animals, these same T cells do not react productively with host MHC bound to self-peptides (1, 2) although such T cells may detect the host combinations weakly (3). Individual T cells are usually very specific for particular foreign peptides, often reacting only with the immunogen or its close relatives (4, 5). However, occasional T cells have been identified that react with host MHC bound to a variety of peptides that are not closely related to each other (6, 7). It has been argued that this degeneracy for peptide is essential to the ability of T cells to anticipate all conceivable foreign peptides (8). TCRs that can react with two or more unrelated peptides bound to allogeneic MHC proteins have also been described (9, 10). In fact, peptide promiscuity may be more common for TCRs reacting with allogeneic rather than syngeneic MHC (11).
X-ray crystallography has revealed the footprint of TCRs on their MHC/peptide ligands. In most cases, contact between the amino acids of the TCR and the MHC contributes extensively to the binding (12). Thus, the fact that the TCR concerned is usually very specific for the peptide is perhaps surprising. Such specificity could arise for three reasons. First, by their very nature, the genomically encoded segments of TCR α and β chains might not allow sufficiently strong reaction with any MHC protein to drive the response of mature T cells. Additional binding energy may always have to be supplied by interactions between the TCR and the peptide. Second, peptide specificity may be the consequence of positive selection. Positive selection involves low avidity/affinity reactions between the TCRs on thymocytes and MHC/self-peptides in the thymus, and thus gives rise to mature T cells bearing TCRs with only low affinity for MHC (13). Mature T cells, however, need a strong reaction to be triggered to full activation (14–16). Because the MHC protein involved in positive selection of the thymocyte and activation of the mature T cell is usually one and the same, the additional binding energy required to activate the mature T cell must come from the TCR/peptide interaction. Such additional strength of binding may usually demand increased peptide specificity from the T cell. Finally, peptide specificity may be imposed on the TCR repertoire by negative selection. Many different self-peptides can bind to any given MHC protein (17, 18). Developing thymocytes are exposed to most of these combinations (19). Thus, any thymocyte that can recognize both self-peptides and foreign peptides will be eliminated. This phenomenon will lead to the removal of cells bearing both marginally and markedly promiscuous TCRs.
If the germ line-encoded TCR α and β segments confer peptide specificity, then all T cells regardless of the selecting MHC haplotype should be peptide specific. Similarly, if the ability to undergo positive selection requires peptide specificity, then T cells derived from mice expressing self-MHC, regardless of the diversity of peptides displayed, should be reactive to specific peptides in bound to the selecting MHC molecule. However, if negative selection imparts peptide specificity, then the degree of allowable peptide promiscuity of a given TCR repertoire should correlate with the level of negative selection.
The idea that self-tolerance controls the repertoire of TCRs on mature T cells has been examined in the past. We and others showed that a large percentage of thymocytes that can be positively selected are in fact deleted in the thymuses of normal mice (20–23), leading to a considerable reduction in the mature T cell repertoire. Moreover, others showed that the repertoire of T cells for allogeneic MHC could be pared down to a set of cells that is very peptide specific, once tolerance to the foreign MHC/mouse peptides had been achieved (24, 25). However, neither of these types of experiments investigated the specificities of the negatively selected thymocytes. Because thymocytes detect MHC/peptide combinations with lower avidity than mature T cells do (14–16), these cells might have been deleted by reactions with quite low avidity ligands, avidities that would be too low to drive reaction of the cells once they had matured. Thus, the deleted thymocytes may have been quite specific for MHC bound to just a few peptides, self or foreign. On the other hand, the thymocytes deleted in these situations may have included a subset that is very peptide promiscuous.
To find out how peptide specificity is conferred to the mature T cell repertoire, we analyzed T cells from mice with differing levels of MHC-specific tolerance. T cells from animals that were fully tolerant to a particular MHC protein/self-peptide were very specific for peptides against which they were immunized. In contrast, T cells that had not been tolerized for recognition of a particular MHC protein/self-peptide were often very promiscuous for peptide recognition. This result was not affected by positive selection on different H2 haplotypes. Hence it is negative selection that focuses the attention of the mature T cell repertoire on specific peptides. The existence of the extremely peptide cross-reactive TCRs in inadequately tolerized animals suggests that elements of TCRs may indeed be evolutionarily selected to react with MHC proteins with some affinity (26). Searches for the secrets governing the obsession of T cells for MHC proteins might do well to focus on these TCRs.
Materials and Methods
Mice and Generation of Bone Marrow Chimeras. C57BL/6, B10.BR, B10.D2, and C3H × C57BL/6F1 mice were purchased from The Jackson Laboratory. MHC class II locus-deficient mice were a gift from C. Benoist and D. Mathis (27). IEk-Hb P5A, IEk-MCC P5A, and IAb-Eα single peptide mice have been described and crossed onto the invariant chain (Ii) and MHC class II locus-deficient background (28, 29). IAb-2W1S single peptide mice were generated in the same way as the IAb-Eα mice and crossed onto the IAb and Ii-deficient background (30). Parent and F1 into parent bone marrow (BM) chimeras were generated by reconstituting lethally irradiated C57BL/6 mice with 5 × 106 BM cells isolated from either C57BL/6 or C57BL/6 × C3H F1 mice. After 8 to 12 wk, mice were immunized and tested. All mice were maintained in a pathogen-free environment in accordance with institutional guidelines in the Animal Care Facility at the National Jewish Medical and Research Center.
Generation of IAβb–/– Ii–/– C57BL/6 Fibroblast Cell Lines Expressing IEk and IAb Linked Peptides. Kidneys from IAβb–/– Ii–/– C57BL/6 mice were diced and digested in 0.07% collagenase/0.001% DNase for 4 h at 37°C, followed by 48 h adherence to plastic. Nonadherent cells were then washed away. After 1 wk in culture, cells were immortalized by infection with a retrovirus expressing human papilloma virus (HPV) E6 and E7 genes and subcloned twice (31). Intercellular adhesion molecule-1 and CD80 were introduced to the fibroblast cell line by retroviral transduction (32). The transformed cells were then transfected with a series of mouse stem cell virus (MSCV)-based retroviruses (33) expressing the cDNAs for IEαk and peptide-linked IEβk or IAαb and peptide-linked IAβb. Cells were subcloned and selected for equivalent expression of IEk or IAb.
Immunization of Mice and Generation of T Cell Hybridomas. In vitro-activated IEk allogeneic T cell hybridomas were derived from CD4+ T cells purified from the spleens and lymph nodes of C57BL/6 (H-2b) or B10.D2 (H-2d) mice by passage over Cellect columns (Cytovax Biotechnologies, Alberta, Canada) and incubated for 4 days with mitomycin C-treated B10.BR spleen cells in the presence of 20 μg/ml anti-IAk antibody 11-5-2 to prevent activation of T cells reacting with IAk. T cell blasts were then purified on Ficoll step gradients, incubated for 3 days with saturating amounts of IL-2, and fused to BW cells (34). Hybridomas were screened for reactivity with IEk by using fibroblasts expressing wild-type IEk MHC molecules. T cell hybridomas prepared from IEk-expressing mice immunized with various foreign peptides were collected from the laboratories that made them: 2B4, AD10 (35, 36); ST4 (37); KMAC 126, 5KC-73.8 (38); and KH8.3 (39). KC-99A 4.18, 2KC-99A-12.3, KH10.1, and KALA-7.13 were generated from peptide-immunized mice by using standard techniques (34).
IEk-Hb P5A and IEk-MCC P5A-specific T cell hybridomas were derived from the various mice immunized with dendritic cells (DC) derived from IEk-Hb P5A and IEk-MCC P5A mice (39). DC were grown from BM in medium containing granulocyte–macrophage colony-stimulating factor, 1% normal mouse serum for 7 days (40). Mice were immunized by i.v. injection of 107 DC. After 7 days, CD4+ T cells were purified from the immunized spleens and restimulated in vitro, again in 1% normal mouse serum, with spleen cells from IEk-MCC P5A or IEk-Hb P5A single peptide mice, and converted to T cell hybridomas (34).
IAb-3K-specific T cell hybridomas were derived from mice immunized with IAβb–/– Ii–/– BM-derived DC transduced with an IAb-3K-expressing retrovirus (41, 42). After 7 days, CD4+ T cells were purified from immunized spleen and restimulated in vitro with DC transfected with an IAb-3K-expressing retrovirus and converted to T cell hybridomas (34).
IEk-Linked Peptide Monomer Production. Soluble IEk bound to different peptides was prepared in insect cells and purified as described (43). The proteins at 200 μM were coated onto individual wells of Immulon 1B 96-well plates (Thermo Lab-systems, Franklin, MA) overnight at 4°C and then washed to remove unbound protein. Wells were then blocked with media containing serum for 2 hr at room temperature, followed by the addition of hybridomas for analysis.
Hybridoma Stimulation Assay. For all T cell hybridomas/single-peptide antigen-presenting cell (SP-APC) stimulation experiments, 105 T cell hybridomas per well were incubated with 3 × 104 of each of the SP-APCs for 24 h. For hybridomas/plate-bound IEk monomer, 105 T cell hybridomas were incubated on plate-bound IEk monomer for 24 h. After overnight incubation at 37°C, supernatants were analyzed for IL-2 production by using a standard MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide) assay (34). All T cell hybridoma data were generated from a minimum of three independent experiments.
Results
Characterization of SP-APC Library. In these studies, we have concentrated on T cells specific for the MHC protein IEk bound to various peptides. To avoid contributions from unknown self-peptides bound to IEk in our APCs, we constructed a collection of APCs, each of which express IEk bound to a single, covalently linked peptide (SP-APC). The peptide sequences involved are listed in Fig. 1A. Because the linker in MHC molecules covalently bound to peptides is sensitive to proteases in an Ii-dependent manner, and many non-APCs express Ii, we prepared a fibroblast cell line from mice deficient in Ii as well as MHC class II to use as the parental cell line (described in Materials and Methods). SP-APC clones were selected for similar expression of IEk (Fig. 5, which is published as supporting information on the PNAS web site, www.pnas.org). All of the SP-APC we could test were able to present IEk/their linked peptide to T cells specific for that combination (data not shown). To determine the extent to which the linked peptide prevented IEk from presenting other peptides, we measured the ability of this collection of SP-APC to present a peptide from mouse Hb or moth cytochrome c (MCC) to six T cell hybridomas specific for IEk bound to either of these peptides. The SP-APC were unable to present exogenously added peptides at any concentration to four of the T cell hybridomas we tested (data not shown) and were able to present exogenously added peptide only very inefficiently (<10–4 to 10–5 of the activity of wild-type IEk) to the two other T cell hybridomas tested (Fig. 1 B and C). We concluded that almost all of the IEk on the SP-APC was occupied irreversibly with its covalently bound peptide.
Fig. 1.
Generation and analysis of APC-expressing IEk/linked peptides. The protein sources and sequences of peptides bound to IEk used in these studies are shown. Colored residues indicate the predicted TCR contact residues of each peptide; underlined residues indicate the MHC anchor positions. The sequences of peptides derived from MCC or mouse Hb but with substitutions at the P3, P5, or P6 positions of the peptides are not shown but can be inferred from their names, which indicate the position and nature of the change in amino acid sequence from the parent peptide. Nearly all of the IEk on SP-APC is occupied by the covalently linked peptide. SP-APC were incubated with various concentrations of Hb peptide and the IEk/Hb-specific T cell hybridoma KH-8.3 (B), or MCC peptide and the IEk/MCC-specific T cell hybridoma AD10 (C). Responses were measured by production of IL-2 by the T cell hybridomas.
Allogeneic T Cells Can Be Highly Peptide-Degenerate. To isolate IEk-restricted T cells from mice that have not undergone negative selection on IEk/self-peptides, CD4+ T cells from IAb-expressing C57BL/6 or IAd/IEd-expressing B10.D2 mice were stimulated with IEk/self-expressing B10.BR splenocytes in vitro in a one-way mixed lymphocyte reaction. Responses to the other potential target in this reaction, IAk, were blocked with an anti-IAk antibody. The IEk-reactive T cells were then converted to hybridomas for ease of analysis. These T cell hybridomas were, therefore, picked for their ability to recognize IEk/self-peptides normally occupying IEk in B10.BR mice. However, when probed for peptide specificity by using the SP-APC, three distinct groups of cross-reactivity were found. The most degenerate T cells (11/36 T cells) reacted with IEk bound to many different peptides, whereas T cells in the second group (5/36 T cells) were able to react with IEk bound to some structurally related peptides as well as the immunizing IEk MHC molecules presenting wild-type self-peptides (Fig. 2 A and C), and summarized for many (Fig. 3 A and B). The third group (20/36) responded only to the immunizing IEk MHC molecules presenting wild-type self-peptides (Fig. 3 A and B).
Fig. 2.
IEk-reactive T cells from mice lacking IEk can be highly peptide-degenerate. Individual T cell hybridomas were tested for their ability to react with IEk bound to different peptides. Data shown are the titers of IL-2 produced by two such hybridomas in assays in which the hybridomas were challenged with IEk bound to various single peptides on SP-APC fibroblasts (A and C) or as purified protein prepared in insect cells (B and D). The 17-5 T cell hybridoma from H2b mice reacts with IEk bound to many different peptides on SP-APC (A), as well as with purified, plate-bound IEk linked to many different peptides (B). The 17-37 T cell hybridoma from H-2b mice is differentially activated by all SP-APC tested (C) and responds similarly to purified, plate-bound IEk-linked peptide monomers and SP-APC (D).
Fig. 3.
The extent of IEk-restricted peptide specificity correlates with the level of IEk-specific T cell tolerance. The hybridomas were tested for their ability to react with IEk/various peptides presented on SP-APC. T cell hybridomas from H2d mice (A) or H2b (B), raised in one-way mixed lymphocyte reactions against IEk, are very peptide promiscuous. (C) IEk-restricted peptide degeneracy or self-reactivity is not observed in T cell hybridomas derived from mice expressing wild-type IEk. (D and G) T cell hybridomas derived from H2b animals immunized with DCs bearing IEk bound to a single peptide are very peptide promiscuous. Immunizations were with DCs bearing Hb P5-A (D) or MCC P5-A (G). (E and H) Self-reactivity and some identifiable peptide cross-reactivity is observed from T cell hybridomas derived from IEk-MCC P5-A (E) and IEk-Hb P5-A (H) single peptide mice immunized with DCs expressing IEk-Hb P5-A or IEk-MCC P5-A, respectively. (F and I) IEk-restricted peptide degeneracy or self-reactivity is not observed in T cell hybridomas derived from IEk-expressing wild-type mice immunized with DCs expressing IEk-Hb P5-A (F) or IEk-MCC P5-A (I).
To check that the cross-reactivities were not due to presentation, by the SP-APC, of contaminating peptides from the mouse proteins of the APC or the FBS of the culture medium, we tested the ability of the hybridomas to respond to a panel of soluble IEk-SP proteins that had been produced in insect cells, purified, and then coated on plastic (43). The results for two of the hybridomas are shown in Fig. 2 B and D. The spectrum of peptides recognized by the hybridomas on the insect cell-produced IEk was similar to that bound to the SP-APC although the response to the plate-bound IEk-SP was usually smaller. Some low-affinity responses were absent in the plate-bound soluble IEk experiments as compared with the SP-APC experiments, likely due to a lack of adhesion or costimulatory molecules. Therefore, the hybridoma responses to the SP-APC were indeed against IEk/covalently linked peptide. This ability of allogeneic T cells to react with IEk bound to many peptides was not dependent on positive or negative selection on a particular MHC class II protein because it was common to T cells derived from either IAb-expressing or IAd and IEd-expressing animals. Additionally, TCR sequencing has shown that this peptide cross-reactivity is not due to expression of multiple TCR α or β chains by the hybridomas concerned because four of the five hybridomas so analyzed express only one TCR α and β chain (data not shown).
Over the years, a large number of T cell hybridomas have been made from mice expressing IEk that have been primed with various antigens (35–39). We tested some of these hybridomas for their ability to react with our collection of SP-APC. As shown in Fig. 3C, they were each specific for their immunizing peptide, or its close relatives, and did not react with APC expressing other peptides. Thus, the peptide promiscuity we found for T cells responding to allogeneic IEk is in sharp contrast to the peptide specificity of T cells isolated from IEk wild-type mice immunized with peptide or protein antigen.
In contrast to some reports of peptide-independent MHC recognition (44, 45), none of the allogeneic IEk-reactive T cells isolated were completely ignorant of the bound peptide because in no case did a T cell react with all of the SP-APC equivalently. Furthermore, the extent of IEk reactivity did not correlate with minor differences in IEk expression level on the SP-APC (Fig. 5). The cross-reactivity of the T cells did not necessarily correlate with the nature of the potential TCR-contacting amino acids of the peptide, (compare cross-reactivities in Figs. 2 and 3 with the peptide sequences in Fig. 1). Additionally, the cross-reactivity, or lack of it, of the T cells was not necessarily due to bulky side chains of the peptide masking critical TCR binding residues on the MHC molecule because many of these allogeneic T cells failed to recognize IEk with a peptide expressing alanines at all TCR contact residues (Figs. 2 and 3). Therefore, although some of the cross-reactivities are easily interpretable [for example the cross-reaction by many of the hybridomas between alanine and leucine at position P5 of the MCC peptide (Fig. 3)], many of them are not. Thus, some of these cross-reactivities may be due to subtle conformational changes in amino acids of the TCR or MHC when bound to different peptides (46, 47). Further understanding of this problem awaits more extensive structural studies.
Hb and MCC-Specific T Cells Derived from C57BL/6 Mice Can Be Highly Peptide-Degenerate. We wondered whether the marked MHC/peptide cross-reactivity of some T cells generated in the mixed lymphocyte reactions was simply a consequence of the way the T cells had been generated. To find out whether such T cells would also appear if they were generated in response to MHC bound to a single peptide, C57BL/6 mice were primed with DCs bearing a single allogeneic MHC protein, IEk, bound to a single peptide, MCC P5A or Hb P5A (Fig. 3 D and G). As controls, C57BL/6 × C3H F1 animals were similarly immunized. Some of the T cells in these F1 animals have, like those in C57BL/6 mice, been positively selected on IAb. However, all of the T cells in the F1 mice, unlike those in C57BL/6 mice, have been tolerized to IEk/mouse peptides.
As expected, T cell hybridomas generated from the F1 mice expressing wild-type IEk MHC molecules and immunized with IEk/antigenic peptide were always highly specific for their immunizing antigen. None reacted with IEk bound to self-peptides, and none of the T cells raised in response to Hb P5A reacted with any of the MCC-related peptides or vice versa (Fig. 3 F and I). In contrast, many of the T cells isolated from C57BL/6 mice immunized with IEk bound to one of the two single peptides were again found to be degenerate in their recognition of other peptides (Fig. 3 D and G). Although some were highly specific for the immunizing peptide, others were reactive with many other peptides in addition to the immunogen.
Foreign Antigen-Specific T Cells Derived from MHC-SP Mice Are Often Self-Reactive. The fact that T cells derived from mice expressing wild-type IEk MHC molecules react with IEk in a very peptide-specific fashion may be a function of positive selection on IEk or negative selection on IEk/the many self-peptides to which it is bound in wild-type IEk-expressing animals. Alternatively, peptide specificity may be due to tolerance to the overall conformational property of IEk, which is common to all IEk molecules regardless of the peptide to which they are bound. To distinguish these possibilities, we checked the peptide cross-reactivity of T cells that had been positively and negatively selected on IEk bound almost entirely to a single peptide, by using mice expressing IEk covalently bound to one peptide (IEk-SP mice) (39). T cells isolated from immunized IEk-SP mice reacted with IEk/immunizing antigen (Fig. 3 E and H). They were also to some degree peptide-degenerate because many of them reacted with IEk bound to some unknown self-peptides in an Ii-dependent and/or -independent manner (12/19 T cells from IEk-MCC P5A SP mice, 18/23 T cells from IEk-Hb P5A SP mice). They were, however, less peptide-degenerate than their counterparts from IEk-negative animals because none of them showed extensive cross-reactivity on the SP-APC panel.
Foreign antigen-specific T cells derived from two different IAb-SP mice show a similar frequency of cross-reaction with wild-type IAb-expressing cells (10/11 T cells from IAb-Eα SP mice, 4/4 T cells from IAb-2W1S SP mice) (Fig. 6, which is published as supporting information on the PNAS web site). These results indicate that a single peptide bound to MHC is not sufficient to eliminate all broadly reactive T cells, but that the process may require the many conformations offered by the normal collection of self-peptides bound to MHC.
BM-Derived Cells Eliminate Peptide-Degenerate T Cells. Our results suggest that the presence of wild-type IEk/many peptides during negative selection is sufficient to remove broadly peptide-reactive T cells regardless of the thymic MHC alleles involved in positive selection. As a final test of this idea, we prepared BM chimeric mice in which the positively selecting thymic MHC was IAb, but the negatively selecting BM-derived cells did ([H2b × H2k]F1 → H2b) or did not (H2b → H2b) express IEk. T cells from these mice were primed with DCs bearing IEk/Hb P5A and analyzed for their cross-reactivity with other peptides. Many of the T cells derived from animals that had undergone positive and negative selection on only IAb MHC molecules were highly peptide-degenerate (Fig. 4A). In contrast, T cells that had been positively selected on IAb but negatively selected in the presence of IEk were very peptide specific (Fig. 4B). Thus, the phenotype of T cells derived from the allogeneic, single peptide, wild-type, and chimeric mice indicates that the extent of allowable peptide degeneracy is controlled by MHC-specific tolerance.
Fig. 4.
Negative selection eliminates peptide-degenerate T cells from the mature T cell repertoire. Irradiated BM-reconstituted mice, in which positive selection could occur on IAb but the T cells were or were not tolerant to IEk, were prepared as described in Materials and Methods. The mice were immunized with DCs expressing IEk-Hb P5-A. T cell hybridomas specific for IEk-Hb P5-A were prepared from the animals and tested for their ability to react with APC-expressing IEk/various peptides. IEk-Hb P5-A-reactive T cell hybridomas derived from IAb-expressing C57BL/6 mice reconstituted with BM from IAb-expressing C57BL/6 mice can be highly peptide-degenerate (A) whereas IEk-Hb P5-A-specific T cell hybridomas derived from C57BL/6 mice reconstituted with IAb/IEk-expressing (C57BL/6 × C3H) F1 BM are highly peptide specific (B).
Discussion
The individual contains a limited number of mature T cells, yet must be able to respond to a vast number of potential foreign antigens. This feat could be achieved efficiently if individual TCRs could recognize many different peptide antigens. However, the increase in the breadth of the immune response that could be so occasioned has to be balanced with the fact that broadly cross-reactive T cells are likely to react with self-MHC bound to the many self-peptides. Thus, the requirement that T cells cover the waterfront of potential foreign antigens has to be balanced with the need that they not react with self.
A number of people have thought about this problem and concluded that T cells must be somewhat, but not overly, peptide cross-reactive (8, 11, 48, 49). Experimental evidence suggests that the peripheral TCR repertoire is such that the TCRs on individual mature T cells are able to recognize only a limited number of peptides bound to self-MHC molecules (4, 5). How this balance of peptide degeneracy is established and whether each TCR in the repertoire is similarly cross-reactive had yet to be determined.
To ensure self-tolerance, thymic negative selection eliminates self-reactive T cells. Estimates have suggested that this process eliminates 50–75% of the positively selected T cells (20–23). There are perhaps 1010 possible combinations of TCRs; of these, ≈1–5% are thought to allow positive selection on any given haplotype set of MHC proteins (50–52). Thus, if the estimates are right, MHC proteins bound to a couple of thousand self-peptides are able to delete in the thymus thymocytes bearing >108 different TCRs. If TCRs have only limited peptide cross-reactivity, how is this possible? In part the paradox is resolved by the fact that thymocytes are more sensitive to negative selection than mature T cells are to activation (14–16). Thus MHC/peptides that could not productively stimulate peripheral T cells can nevertheless participate in thymocyte deletion.
The experiments described here, however, provide an additional explanation. In the absence of negative selection, the repertoire of TCRs is a continuum of peptide degeneracy. During negative selection, a significant percentage of thymocytes that are deleted suffer this fate because they bear very cross-reactive TCRs, TCRs that can react with MHC/many peptides with affinities that are high enough to drive responses by mature T cells bearing them. Thus, negative selection refines the potential TCR repertoire into one focused on self-MHC bound to specific peptides.
The repertoire of T cells specific for IEk/peptide, isolated from mice expressing allogeneic class II protein, includes TCRs that are much more peptide promiscuous than those isolated from IEk-SP mice, reflecting, probably, more efficient removal of IEk-reactive T cells in the latter animals. Previous estimates of the percentages of T cells removed by negative selection were based on the numbers of T cells that escape negative selection in class II-SP mice. However, the results with allogeneic animals reported here indicate that a reasonable percentage of potentially MHC-reactive T cells (those that are extremely cross-reactive) are deleted in the class II-SP animals. This percentage amounts to about half of the IEk-reactive T cells in both C57BL/6 and B10.D2 mice (26/50 for all allogeneic T cells analyzed in this study). Thus, the actual percentage of T cells able to undergo positive selection, which are deleted by negative selection, could be as high as 90%.
Our data offer an answer to the long-standing quandary about the nature of the high frequency of allo-MHC-specific T cells. The two, not necessarily exclusive, models proposed suggest either that the high frequency of allo-reactivity derives from T cells recognizing unique features of the self-peptides presented by the allogeneic MHC molecule (53), or that T cells recognize amino acid residue polymorphisms of the allogeneic MHC molecules with little or no regard for the bound peptide (46, 54). The data presented here indicate that mice lacking IEk-specific negative selection have IEk peptide-dependent, as well as peptide-independent, T cells in relatively equivalent numbers. When the T cell repertoire undergoes negative selection on IEk/single peptide, the IEk peptide-independent T cells are eliminated. The remaining foreign antigen-specific T cells in these mice, however, have a high degree of reactivity toward IEk/wild-type self-peptides. Whether the nature of this foreign and self-reactivity is due to an increase in peptide promiscuity of the T cells or is due to the T cells recognizing shared antigenic features between the foreign and self-peptides requires further investigation. The finding that the majority of foreign antigen-specific T cells are also self-reactive suggests that the inherent MHC bias of the germ line TCR repertoire may be even stronger than previously proposed and best examined in T cells that have been positively, but not negatively, selected (52, 55).
Thus, we conclude that it is primarily negative selection that reduces an intrinsically peptide-degenerate T cell repertoire to a collection of TCRs whose specificity is highly focused on the MHC-bound peptide. This process allows the body to balance the benefits of T cell recognition efficiency due to peptide promiscuity with the requirements of self-tolerance.
Supplementary Material
Acknowledgments
We thank Drs. C. Benoist and D. Mathis for the MHC class II promoter transgene cassette and the MHC class II locus-deficient mice; Dean Becker for production of the MHC-transgenic mice; Tibor Vass for care and breeding of the animals; and members of the Kappler/Marrack lab for helpful discussions and review of the manuscript. This research has been funded by U.S. Public Health Service Grants AI-17134, AI-18785, AI-52225, and AI-22295.
Abbreviations: BM, bone marrow; DC, dendritic cell; TCR, T cell receptor; APC, antigen-presenting cell; SP-APC, single-peptide APC; MCC, moth cytochrome c; Ii, invariant chain.
References
- 1.Matzinger, P., Zamoyska, R. & Waldmann, H. (1984) Nature 308, 738–741. [DOI] [PubMed] [Google Scholar]
- 2.Rammensee, H. G. & Bevan, M. J. (1984) Nature 308, 741–744. [DOI] [PubMed] [Google Scholar]
- 3.Freitas, A. A. & Rocha, B. (2000) Annu. Rev. Immunol. 18, 83–111. [DOI] [PubMed] [Google Scholar]
- 4.Jameson, S. C., Carbone, F. R. & Bevan, M. J. (1993) J. Exp. Med. 177, 1541–1550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Reay, P. A., Kantor, R. M. & Davis, M. M. (1994) J. Immunol. 152, 3946–3957. [PubMed] [Google Scholar]
- 6.Finnegan, A., Smith, M. A., Smith, J. A., Berzofsky, J., Sachs, D. H. & Hodes, R. J. (1986) J. Exp. Med. 164, 897–910. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Brock, R., Wiesmuller, K. H., Jung, G. & Walden, P. (1996) Proc. Natl. Acad. Sci. USA 93, 13108–13113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Mason, D. (1998) Immunol. Today 19, 395–404. [DOI] [PubMed] [Google Scholar]
- 9.Gundlach, B. R., Wiesmuller, K. H., Junt, T., Kienle, S., Jung, G. & Walden, P. (1996) J. Immunol. 156, 3645–3651. [PubMed] [Google Scholar]
- 10.Tallquist, M. D., Yun, T. J. & Pease, L. R. (1996) J. Exp. Med. 184, 1017–1026. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Daniel, C., Horvath, S. & Allen, P. M. (1998) Immunity 8, 543–552. [DOI] [PubMed] [Google Scholar]
- 12.Bankovich, A. J. & Garcia, K. C. (2003) Immunity 18, 7–11. [DOI] [PubMed] [Google Scholar]
- 13.Starr, T. K., Jameson, S. C. & Hogquist, K. A. (2002) Annu. Rev. Immunol. [DOI] [PubMed]
- 14.Yagi, J. & Janeway, C. A., Jr. (1990) Int. Immunol. 2, 83–89. [DOI] [PubMed] [Google Scholar]
- 15.Davey, G. M., Schober, S. L., Endrizzi, B. T., Dutcher, A. K., Jameson, S. C. & Hogquist, K. A. (1998) J. Exp. Med. 188, 1867–1874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lucas, B., Stefanova, I., Yasutomo, K., Dautigny, N. & Germain, R. N. (1999) Immunity 10, 367–376. [DOI] [PubMed] [Google Scholar]
- 17.Rudensky, A., Preston-Hurlburt, P., Hong, S. C., Barlow, A. & Janeway, C. A., Jr. (1991) Nature 353, 622–627. [DOI] [PubMed] [Google Scholar]
- 18.Hunt, D. F., Michel, H., Dickinson, T. A., Shabanowitz, J., Cox, A. L., Sakaguchi, K., Appella, E., Grey, H. M. & Sette, A. (1992) Science 256, 1817–1820. [DOI] [PubMed] [Google Scholar]
- 19.Marrack, P., Ignatowicz, L., Kappler, J. W., Boymel, J. & Freed, J. H. (1993) J. Exp. Med. 178, 2173–2183. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Ignatowicz, L., Rees, W., Pacholczyk, R., Ignatowicz, H., Kushnir, E., Kappler, J. & Marrack, P. (1997) Immunity 7, 179–186. [DOI] [PubMed] [Google Scholar]
- 21.Surh, C. D., Lee, D. S., Fung-Leung, W. P., Karlsson, L. & Sprent, J. (1997) Immunity 7, 209–219. [DOI] [PubMed] [Google Scholar]
- 22.Tourne, S., Miyazaki, T., Oxenius, A., Klein, L., Fehr, T., Kyewski, B., Benoist, C. & Mathis, D. (1997) Immunity 7, 187–195. [DOI] [PubMed] [Google Scholar]
- 23.van Meerwijk, J. P., Marguerat, S., Lees, R. K., Germain, R. N., Fowlkes, B. J. & MacDonald, H. R. (1997) J. Exp. Med. 185, 377–383. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Thomas, D. W. & Shevach, E. M. (1977) Proc. Natl. Acad. Sci. USA 74, 2104–2108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Doherty, P. C. & Bennink, J. C. (1979) J. Exp. Med. 149, 150–157. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Jerne, N. K. (1971) Eur. J. Immunol. 1, 1–9. [DOI] [PubMed] [Google Scholar]
- 27.Madsen, L., Labrecque, N., Engberg, J., Dierich, A., Svejgaard, A., Benoist, C., Mathis, D. & Fugger, L. (1999) Proc. Natl. Acad. Sci. USA 96, 10338–10343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bikoff, E. K., Huang, L. Y., Episkopou, V., van Meerwijk, J., Germain, R. N. & Robertson, E. J. (1993) J. Exp. Med. 177, 1699–1712. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Grusby, M. J., Johnson, R. S., Papaioannou, V. E. & Glimcher, L. H. (1991) Science 253, 1417–1420. [DOI] [PubMed] [Google Scholar]
- 30.Ignatowicz, L., Kappler, J. & Marrack, P. (1996) Cell 84, 521–529. [DOI] [PubMed] [Google Scholar]
- 31.Galloway, D. A., Demers, G. W., Foster, S. A., Halbert, C. L. & Russell, K. (1994) Cold Spring Harbor Symp. Quant. Biol. 59, 297–306. [DOI] [PubMed] [Google Scholar]
- 32.Schaefer, B. C., Mitchell, T. C., Kappler, J. W. & Marrack, P. (2001) Anal. Biochem. 297, 86–93. [DOI] [PubMed] [Google Scholar]
- 33.Mitchell, T. C., Hildeman, D., Kedl, R. M., Teague, T. K., Schaefer, B. C., White, J., Zhu, Y., Kappler, J. & Marrack, P. (2001) Nat. Immunol. 2, 397–402. [DOI] [PubMed] [Google Scholar]
- 34.White, J., Kappler, J. & Marrack, P. (2000) Methods Mol. Biol. 134, 185–193. [DOI] [PubMed] [Google Scholar]
- 35.Hedrick, S. M., Matis, L. A., Hecht, T. T., Samelson, L. E., Longo, D. L., Heber-Katz, E. & Schwartz, R. H. (1982) Cell 30, 141–152. [DOI] [PubMed] [Google Scholar]
- 36.Hedrick, S. M., Engel, I., McElligott, D. L., Fink, P. J., Hsu, M. L., Hansburg, D. & Matis, L. A. (1988) Science 239, 1541–1544. [DOI] [PubMed] [Google Scholar]
- 37.Evavold, B. D., Williams, S. G., Hsu, B. L., Buus, S. & Allen, P. M. (1992) J. Immunol. 148, 347–353. [PubMed] [Google Scholar]
- 38.White, J., Pullen, A., Choi, K., Marrack, P. & Kappler, J. W. (1993) J. Exp. Med. 177, 119–125. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Liu, C. P., Parker, D., Kappler, J. & Marrack, P. (1997) J. Exp. Med. 186, 1441–1450. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Kedl, R. M., Schaefer, B. C., Kappler, J. W. & Marrack, P. (2002) Nat. Immunol. 3, 27–32. [DOI] [PubMed] [Google Scholar]
- 41.Marrack, P., Bender, J., Jordan, M., Rees, W., Robertson, J., Schaefer, B. C. & Kappler, J. (2001) J. Immunol. 167, 617–621. [DOI] [PubMed] [Google Scholar]
- 42.Rees, W., Bender, J., Teague, T. K., Kedl, R. M., Crawford, F., Marrack, P. & Kappler, J. (1999) Proc. Natl. Acad. Sci. USA 96, 9781–9786. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Kozono, H., White, J., Clements, J., Marrack, P. & Kappler, J. (1994) Nature 369, 151–154. [DOI] [PubMed] [Google Scholar]
- 44.Elliott, T. J. & Eisen, H. N. (1990) Proc. Natl. Acad. Sci. USA 87, 5213–5217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Smith, P. A., Brunmark, A., Jackson, M. R. & Potter, T. A. (1997) J. Exp. Med. 185, 1023–1033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Sherman, L. A. & Chattopadhyay, S. (1993) Annu. Rev. Immunol. 11, 385–402. [DOI] [PubMed] [Google Scholar]
- 47.Reiser, J. B., Darnault, C., Gregoire, C., Mosser, T., Mazza, G., Kearney, A., Van Der Merwe, P. A., Fontecilla-Camps, J. C., Housset, D. & Malissen, B. (2003) Nat. Immunol. [DOI] [PubMed]
- 48.Ogasawara, K., Maloy, W. L. & Schwartz, R. H. (1987) Nature 325, 450–452. [DOI] [PubMed] [Google Scholar]
- 49.Nanda, N. K., Arzoo, K. K., Geysen, H. M., Sette, A. & Sercarz, E. E. (1995) J. Exp. Med. 182, 531–539. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Huesmann, M., Scott, B., Kisielow, P. & von Boehmer, H. (1991) Cell 66, 533–540. [DOI] [PubMed] [Google Scholar]
- 51.Merkenschlager, M., Graf, D., Lovatt, M., Bommhardt, U., Zamoyska, R. & Fisher, A. G. (1997) J. Exp. Med. 186, 1149–1158. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Zerrahn, J., Held, W. & Raulet, D. H. (1997) Cell 88, 627–636. [DOI] [PubMed] [Google Scholar]
- 53.Matzinger, P. & Bevan, M. J. (1977) Cell. Immunol. 29, 1–5. [DOI] [PubMed] [Google Scholar]
- 54.Bevan, M. J. (1984) Immunol. Today 5, 128–130. [DOI] [PubMed] [Google Scholar]
- 55.Blackman, M., Yague, J., Kubo, R., Gay, D., Coleclough, C., Palmer, E., Kappler, J. & Marrack, P. (1986) Cell 47, 349–357. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.




