The Journal of Immunology
Preferential Escape of Subdominant CD8ⴙ T Cells During Negative Selection Results in an Altered Antiviral T Cell Hierarchy1 Mark K. Slifka,† Joseph N. Blattman,§ David J. D. Sourdive,§ Fei Liu,‡ Donald L. Huffman,* Tom Wolfe,* Anna Hughes,* Michael B. A. Oldstone,‡ Rafi Ahmed,§ and Matthias G. von Herrath2* Negative selection is designed to purge the immune system of high-avidity, self-reactive T cells and thereby protect the host from overt autoimmunity. In this in vivo viral infection model, we show that there is a previously unappreciated dichotomy involved in negative selection in which high-avidity CD8ⴙ T cells specific for a dominant epitope are eliminated, whereas T cells specific for a subdominant epitope on the same protein preferentially escape deletion. Although this resulted in significant skewing of immunodominance and a substantial depletion of the most promiscuous T cells, thymic and/or peripheral deletion of high-avidity CD8ⴙ T cells was not accompanied by any major change in the TCR V gene family usage or an absolute deletion of a single preferred complementarity-determining region 3 length polymorphism. This suggests that negative selection allows high-avidity CD8ⴙ T cells specific for subdominant or cryptic epitopes to persist while effectively deleting high-avidity T cells specific for dominant epitopes. By allowing the escape of subdominant T cells, this process still preserves a relatively broad peripheral TCR repertoire that can actively participate in antiviral and/or autoreactive immune responses. The Journal of Immunology, 2003, 170: 1231–1239.
N
egative selection is a unique process that determines which T cells will be deleted in the thymus and which ones will differentiate into mature cells that are then released into the periphery (1–5). Positive selection occurs first and requires that T cells demonstrate at least a weak interaction with self-peptides presented by the appropriate MHC molecules (6, 7). If there is no structural interaction with peptide and MHC, then the cells are considered irrelevant and essentially die of neglect. In contrast, if the structural avidity for self-peptide and MHC is too high, then the T cells are deleted by negative selection (5, 8). This system of checks and balances is intended to ensure the host of a rich and diverse population of useful T cells that are not inherently autoaggressive. To better understand the process of negative selection and how this influences the peripheral T cell repertoire generated in response to a viral infection, we have developed a transgenic model
*Division of Immune Regulation, La Jolla Institute for Allergy and Immunology, San Diego, CA 92121; †Oregon Health and Science University Vaccine and Gene Therapy Institute, Beaverton, OR 97006; ‡Division of Virology, Department of Neuropharmacology, The Scripps Research Institute, La Jolla, CA 92037; and §Emory Vaccine Center and Department of Microbiology and Immunology, Emory University School of Medicine, Atlanta, GA 30322 Received for publication September 5, 2002. Accepted for publication November 22, 2002. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 This work was supported by National Institutes of Health Grants AI09484 (to M.B.A.O.), U-19 AI51973 (to M.G.v.H.), AI44415 (to M.G.v.H.), National Institute of Diabetes and Digestive and Kidney Diseases Grant DK51091 (to M.G.v.H.), and the Oregon National Primate Research Center Grant RR00163 (to M.K.S.). 2 Address correspondence and reprint requests to Dr. Matthias von Herrath, Division of Immune Regulation, La Jolla Institute for Allergy and Immunology, 10355 Science Center Drive, San Diego, CA 92121. E-mail address:
[email protected]
Copyright © 2003 by The American Association of Immunologists, Inc.
(RIP-NP) in which H-2d mice express the nucleoprotein (NP)3 of lymphocytic choriomeningitis virus (LCMV) under the rat insulin promoter (RIP). This results in directed expression of the LCMV NP gene in the islet cells of the pancreas, and serendipitously, also in the thymus but not in any other organs (9). Thymic grafting experiments indicate that negative selection against LCMV NP occurs in the thymus before T cells reach the periphery (9, 10), and expression of the transgene does not lead to any signs of autoimmunity in naive, uninfected animals (9, 10). Mice that express the LCMV NP gene are able to mount a sufficiently strong antiviral T cell response to effectively clear an LCMV infection with kinetics that are similar to those observed in nontransgenic BALB/c mice (11). However, this also leads to a state of anti-islet autoimmunity in ⬎90% of the mice on the H-2d background within 3 wk, due to the activation of self-reactive T cells that target LCMV NP expressed in the  cells of the pancreas (9, 10). LCMV-induced activation and expansion of self-specific CD8⫹ T cells thus provides a useful in vivo model for examining the structural, functional, and phenotypic characteristics of autoreactive T cells that have been stimulated by a peripheral viral infection functioning as the initiator of a bona fide autoaggressive immune response. In this study, we have examined the consequences of negative selection on a peripheral CD8⫹ T cell pool capable of responding to a viral Ag. Remarkably, self-reactive CD8⫹ T cells specific for a subdominant epitope were largely unaffected by thymic and/or peripheral selection, whereas the T cell response against a dominant epitope was significantly compromised in terms of both functional avidity as well as structural avidity. Functional responsiveness was assessed by stimulation with subsaturating doses of peptides or activation by various peptide analogs, whereas TCR
3 Abbreviations used in this paper: NP, nucleoprotein; LCMV, lymphocytic choriomeningitis virus; RIP, rat insulin promoter; CDR3, complementarity-determining region 3; ICCS, intracellular cytokine staining; MFI, mean fluorescence intensity.
0022-1767/03/$02.00
1232 avidity of the dominant T cell populations was determined by peptidetetramer binding assays. In terms of negative selection events that occur in vivo, our results demonstrate three important points: 1) CD8⫹ T cells specific for subdominant epitopes can escape negative selection in the thymus and peripheral organs virtually unscathed; 2) the most promiscuous CD8⫹ T cells specific for an immunodominant self-epitope are also the most likely to be deleted during negative selection; and 3) CD8⫹ T cells specific for dominant epitopes are deleted as a function of their structural avidity, but these changes are not associated with any major changes in TCR expression levels or the family of V genes that are used. Together, this indicates that the process of negative selection narrows the structural diversity of the peripheral T cell pool, but without necessarily leading to unacceptable defects within the TCR repertoire.
Materials and Methods Mice and virus BALB/c mice were bred at the Scripps Research Institute breeding facility, and all animal experiments were performed in compliance with our institutional guidelines. RIP-NP transgenic mice expressing the LCMV NP gene in the thymus and pancreas were generated in our laboratory and backcrossed ⬎12 generations on the BALB/c background; they have been previously characterized (9, 10). Mice were infected by the i.p. route with 2 ⫻ 105 PFU of LCMV Armstrong and analyzed at 8 days postinfection.
Intracellular cytokine staining (ICCS) and flow cytometry Ag-specific CD8⫹ T cells were stimulated directly ex vivo for 6 or 10 h in the presence of the indicated concentrations of peptide as previously described (12) with the exception that 200 M DTT was added as a reducing agent to increase the solubility of gp283 and NP313 peptides (13). Addition of DTT at this concentration had no effect on the NP118-specific IFN-␥ response but had a substantial enhancing effect on the dose-response curves following NP313- or gp283-specific stimulation. For functional fingerprinting analysis using NP118 peptide analogs, each peptide variant was used at a concentration of 1 ⫻ 10⫺7 M. mAbs used to stain for IFN-␥ and cell surface markers were obtained from BD PharMingen (San Diego, CA) and samples were blocked with anti-FcR (clone 2.4G) and mouse IgG (Sigma-Aldrich, St. Louis, MO) before staining.
Tetramer-binding analysis NP118-Ld tetramer reagents were prepared as described (14) or prepared by the National Institutes of Health Tetramer Core Facility (Atlanta, GA). For structural avidity analysis, samples were normalized for the number of tetramer⫹ T cells by adding naive spleen cells, and then the cells were incubated with NP118 tetramers at the indicated dilutions for 15 min at 37°C, washed, stained with anti-CD8 (clone 53-6.7), washed and fixed in PBS containing 2% formaldehyde before acquisition and analysis on a FACScan (BD Biosciences, Mountain View, CA) instrument, and analyzed using CellQuest software (BD Biosciences).
DIFFERENTIAL EFFECTS OF NEGATIVE SELECTION
Results Expression of LCMV NP skews the antiviral T cell epitope hierarchy The peak of the effector CD8⫹ T cell response against LCMV occurs at ⬃8 days postinfection (14), a time that also represents the peak in functional responsiveness by NP118-specific CD8⫹ T cells in BALB/c mice (12). At this time point, ⬃40% of the splenic CD8⫹ T cells in BALB/c mice were specific for the immunodominant epitope, NP118, whereas only 3–5% of the T cell response was directed toward the subdominant epitopes, gp283 or NP313 (Fig. 1A). In contrast, the frequency of NP118-specific CD8⫹ T cells in RIP-NP mice was significantly reduced (by nearly 50%, p ⬍ 0.05), whereas the frequency of T cells specific for gp283 and NP313 was increased by 2- to 3-fold (Fig. 1B). However, after normalizing for the total number of virus-specific CD8⫹ T cells per spleen, the NP118-specific T cell response in RIP-NP mice is ⬃4-fold lower than that observed in BALB/c mice, whereas the total number of CD8⫹ T cells specific for gp283 and NP313 is nearly the same in each strain of mice (data not shown). A strong T cell response to gp283 was expected (10), because this viral protein is not expressed in the thymus or pancreas of RIP-NP mice. However, it was striking that negative selection had virtually no inhibitory effect on the activation and expansion of the NP313-specific T cell response. Instead, the NP313-specific T cell response in RIP-NP mice remained surprisingly similar to the gp283-specific T cell response. The remarkable effects of negative selection on the immunodominance hierarchy of Ag-specific T cell populations are best exemplified by comparing the ratio of T cell frequencies specific for intramolecular epitopes (NP118:NP313) vs intermolecular epitopes (gp283:NP313) (Fig. 1C). In BALB/c mice, immunodominant NP118-specific T cells outnumber subdominant NP313-specific T cells by nearly 10:1. This ratio is significantly decreased in RIP-NP mice ( p ⫽ 0.002) due to the lower frequency of NP118-specific T cells and the substantially higher frequency of NP313-specific T cells that are triggered after viral infection (Fig. 1B). In sharp contrast, the ratio of NP313-specific T cells to gp283-specific T cells remains at about a 1:1 ratio, and is not significantly altered ( p ⫽ 0.3) by the expression of LCMV NP in RIP-NP mice. Together, these studies demonstrate that expression of a viral Ag (LCMV NP) in the thymus and the periphery (i.e., the pancreas) can have a substantial effect on the outcome of an antiviral immune response, resulting in a sharply decreased T cell response to an immunodominant epitope (NP118) while at the same time inducing a strong T cell response to a selfspecific, subdominant epitope (NP313) and a non-cross-reactive epitope (gp283) located on a viral protein that is not expressed as a self-Ag.
T cell receptor repertoire analysis
Negative selection preferentially deletes dominant T cells of high functional avidity
Repertoire analysis was performed using a modified protocol to that described by Pannetier et al. (15). NP118-specific CD8⫹ T cells from RIP-NP and BALB/c mice at 8 days post-LCMV infection were stained with NP118 tetramers and anti-CD8 and sorted to ⬎98% purity before RNA extraction and cDNA synthesis. Total RNA was isolated from cell suspensions of individual samples using an RNA isolation kit (RNeasy; Qiagen, Valencia, CA) as per the manufacturer’s instructions. cDNA synthesis was then performed using an oligo(dT)12–18 primer according to the manufacturer’s instructions (Life Technologies, Rockville, MD). From each cDNA, PCR were then performed using V primers (see Ref. 15) and a common C primer (CACTGATGTTCTGTGTGACA). Using 2 l of this PCR product as template, run-off reactions were performed with internal fluorescent primers for the different Js (see Ref. 15), with a 5-min 94°C denature and then 12 cycles of 94°C for 45 s, 60°C for 45 s, and 72°C for 45 s, followed by a 10-min extension at 72°C. These products were then denatured in deionized formamide and analyzed on an Applied Biosystems 310 Prism Genetic Analyzer (Foster City, CA). Data was analyzed using the GeneScan 2.0 software (Applied Biosystems).
We examined the functional responsiveness of virus-specific CD8⫹ T cells from LCMV-infected RIP-NP mice and compared this to the T cell responses elicited by BALB/c mice (Fig. 2). Functional avidity was determined by stimulating T cells directly ex vivo with graded doses of specific peptide followed by quantitation of IFN-␥ production by ICCS. NP118-specific CD8⫹ T cells that escaped negative selection in RIP-NP mice required 30to 100-fold more peptide Ag than NP-118-specific T cells from BALB/c mice to elicit a 50% maximal IFN-␥ response (Fig. 2D). In contrast, the antiviral T cell responses to gp283 and NP313 were relatively similar between both groups, with RIP-NP mice exhibiting just slightly lower functional avidity than that of BALB/c mice (Fig. 2, E and F, respectively). The mean fluorescence intensity (MFI) of IFN-␥ production by Ag-specific CD8⫹ T cells was equivalent between BALB/c and RIP-NP mice (NP118 (10⫺4
The Journal of Immunology
1233
FIGURE 1. LCMV-specific CD8⫹ T cell epitope hierarchy is skewed in RIP-NP transgenic mice. Adult BALB/c and RIP-NP transgenic mice were infected with LCMV Armstrong, and the frequency of virus-specific T cells was determined directly ex vivo by ICCS analysis. A, At 8 days postinfection, LCMVspecific CD8⫹ T cells were cultured with no peptide or with saturating amounts of peptide Ag (10⫺4 M NP118, 10⫺6 M gp283, or 10⫺4 M NP313) and the frequency of IFN-␥⫹CD8⫹ T cells is shown in the upper right quadrant of each dot plot. B, Expression of the LCMV NP gene as a self-Ag resulted in a sharply lower NP118-specific T cell response, whereas there was a substantially higher frequency of gp283-specific and NP313-specific T cells in RIP-NP mice compared with that of BALB/c mice analyzed in parallel. Data are expressed as the percentage of CD8⫹ T cells that are specific for the indicated peptides. C, Skewing of the T cell epitope hierarchy in RIP-NP mice resulted in a significantly reduced ratio between dominant (NP118-specific) and subdominant (NP313-specific) T cell populations (ratio ⫽ number of NP118-specific T cells/number of NP313-specific T cells). In contrast, there was no significant difference in the ratio between NP313-specific and gp283specific T cell numbers in either strain of mice (ratio ⫽ number of gp283-specific T cells/number of NP313-specific T cells). The results show the average and SD of three mice per group and represent one of three experiments, each with similar results. Statistical significance was calculated by Student’s t test.
M) ⫽ 339 ⫾ 56 vs 376 ⫾ 25; gp283 (10⫺6 M) ⫽ 561 ⫾ 114 vs 507 ⫾ 36; NP313 (10⫺4 M) ⫽ 426 ⫾ 27 vs 481 ⫾ 42), indicating that there were no obvious defects in the activation and inflammatory cytokine production by any of these T cell populations after an appropriate threshold of activation was reached. The most striking conclusion to be drawn from these studies is that thymic selection resulted in a massive reduction in the functional avidity of immunodominant T cells while having little to no effect on the responsiveness of T cells specific for a subdominant T cell epitope. Of the three T cell populations studied, thymic selection had its greatest impact on the immunodominant T cell responses to NP118. To further analyze the differences between NP118-specific CD8⫹ T cell functions elicited by RIP-NP and BALB/c mice, we examined their relative T cell responsiveness to several NP118 peptide analogs, each consisting of a single amino acid substitution at sites believed to be TCR contact residues (Fig. 3). Similar to previous work (16), NP118-specific CD8⫹ T cells from BALB/c mice are capable of responding to several variants of the original NP118 peptide sequence, but with varying degrees of efficiency. This results in a functional fingerprint of TCR recognition (16 –20). Compared with NP118, peptide variant 122A stimulated NP118-specific T cells relatively well, whereas the other peptide variants shown in Fig. 3 are either intermediate or weak inducers of IFN-␥ production. Nonetheless, at least
a subpopulation of NP118-specific T cells from LCMV-infected BALB/c mice was able to respond to all 10 peptide analogs. NP118specific T cells from RIP-NP mice were similar to NP118-specific T cells from BALB/c mice in terms of their responsiveness to peptide variant 122A, but were substantially less capable of responding to many of the other peptide analogs and were unable to mount a detectable IFN-␥ response against 5 of 10 of the peptides examined in this study. The results of the functional avidity (Fig. 2) and functional fingerprinting (Fig. 3) analyses clearly demonstrate that, at the population level, the NP118-specific CD8⫹ T cells that escaped negative selection in RIP-NP mice are less functionally responsive to peptide stimulation than are those of their wild-type BALB/c counterparts. Expression of activation molecules on virus-specific T cells Upon antigenic stimulation, T cells will typically up-regulate several types of accessory molecules involved with T cell activation and/or T cell adhesion to APCs (12, 21–24). We performed phenotypic analysis of NP118-specific CD8⫹ T cells from LCMV-infected RIP-NP and BALB/c mice to see whether there were any discernable differences in their states of activation (Fig. 4). In these studies, we examined the expression of a representative panel of these activation/adhesion molecules including CD2, CD11a (LFA-1), CD28, CD43
1234
DIFFERENTIAL EFFECTS OF NEGATIVE SELECTION
FIGURE 2. Negative selection depletes dominant self-reactive CD8⫹ T cells of high functional avidity. At 8 days post-LCMV infection, the functional responsiveness of CD8⫹ T cells from RIP-NP (F) and BALB/c mice (E) was determined by IFN-␥ production following stimulation with the indicated doses of peptide Ag. The frequency of Ag-specific IFN-␥⫹CD8⫹ T cells that responded to a given dose of peptide was determined as a percentage of the total number of CD8⫹ T cells in the spleen following stimulation with NP118 (A), gp283 (B), or NP313 (C). To visualize functional avidity regardless of the frequency of the responding T cell populations, the number of IFN-␥⫹CD8⫹ T cells induced by peptide stimulation was expressed as a percentage of the maximum number of IFN-␥⫹CD8⫹ T cells observed after stimulation with 10⫺4 M NP118 (D), 10⫺6 M gp283 (E), or 10⫺4 M NP313 (F). The results show the average and SD of three mice per group and represent one of three experiments, each with similar results.
(mAb clone 1B11), and CD49d (VLA-4). At 8 days post-LCMV infection, the NP118-specific CD8⫹ T cells from RIP-NP mice expressed this group of markers at levels that were equivalent to those observed on NP118-specific T cells isolated from LCMV-infected
BALB/c mice. High expression levels of these activation/adhesion molecules on NP118-specific CD8⫹ T cells from RIP-NP mice indicated that these cells had reached an equivalent stage of T cell activation to that observed in NP118-specific T cells from BALB/c mice.
FIGURE 3. Low-avidity CD8⫹ T cells are less functionally responsive to altered peptide ligands. NP118-specific CD8⫹ T cells from LCMVinfected RIP-NP or BALB/c mice were stimulated directly ex vivo with either wild-type NP118 –126 peptide (WT ⫽ RPQASGVYM) or variations of this peptide in which a single amino acid substitution was performed at the indicated sites. The frequency of IFN-␥⫹CD8⫹ T cells observed following stimulation with the wild-type NP118 peptide was used as the maximum response (100%), and the frequency of IFN-␥⫹ T cells observed after stimulation with the altered peptide ligands is expressed as a percentage of the maximum NP118-specific T cell response. The results show the average and SD of two BALB/c mice and three RIP-NP mice.
FIGURE 4. Low-avidity self-reactive T cells express the same activation/adhesion molecules as high-avidity T cells. The expression levels of several commonly studied activation/adhesion molecules on Ag-specific T cells were examined at 8 days post-LCMV infection. NP118-specific T cells were identified by IFN-␥ production following 6 h of direct ex vivo stimulation with 10⫺4 M peptide, and the expression of this panel of markers was compared with that observed on naive CD8⫹ T cells from uninfected BALB/c mice. Each histogram shows one representative example of six individual mice for each group. The PE control shows the level of nonspecific staining on NP118-specific and naive T cells following incubation with an irrelevant PE-conjugated rat IgG control Ab.
The Journal of Immunology
FIGURE 5. Negative selection depletes dominant T cells of high structural avidity. The TCR avidity of NP118-specific T cells for peptide in the context of MHC class I was determined by staining the cells with NP118 tetramers directly ex vivo. A, CD8⫹ T cells from naive mice or LCMVinfected RIP-NP and BALB/c mice were stained with saturating amounts of NP118 tetramer and examined by flow cytometry. The frequency of NP118 tetramer⫹CD8⫹ T cells in RIP-NP and BALB/c mice was similar to that obtained by ICCS following NP118 peptide stimulation (see Fig. 1). However, the MFI of NP118 tetramer staining was significantly lower in RIP-NP mice compared with that observed in BALB/c mice (p ⫽ 0.004). This appears to be due to a preferential loss of Ag-specific CD8⫹ T cells in RIP-NP mice that bind the highest levels of peptide tetramer, as indicated by the arrows and boxes in the upper right quadrant of the dot plots. B, To further quantitate these differences in structural avidity, Ag-specific CD8⫹ T cells from LCMV-infected RIP-NP and BALB/c mice were stained with graded concentrations of NP118 tetramer at 37°C for 15 min followed by washing and staining for CD8 and analysis by flow cytometry. The results show the average and SD of three mice per group with statistical significance between groups determined at each tetramer concentration using Student’s t test.
Together, these results indicate that the differences in functional avidity between NP118-specific T cells from RIP-NP and BALB/c mice are unlikely to be due to differences in their state of activation. Negative selection purges the system of T cells with high structural avidity The rules of negative selection stipulate that T cells with the highest structural avidity for self-Ags will be deleted from the peripheral T cell repertoire (5, 8). To determine how much this affected the structural avidity of NP118-specific CD8⫹ T cells that escaped thymic depletion in RIP-NP mice, we used MHC class I-peptide tetramer binding assays. CD8⫹ T cells of high structural avidity will typically bind higher levels of peptide tetramers than will
1235
FIGURE 6. Immunodominant CD8⫹ T cells of either low or high structural avidity express the same conserved V repertoire. The TCR usage of NP118-specific CD8⫹ T cells from LCMV-infected RIP-NP and BALB/c mice was determined by staining cells with NP118 tetramers, CD8, and Abs specific for the indicated V gene families. A, The values represent the average and SD of two BALB/c mice and five RIP-NP mice, and V usage is expressed as a percentage of the total NP118-specific T cell population. B, Low-avidity NP118-specific CD8⫹ T cells from RIP-NP mice and highavidity T cells from BALB/c mice both predominantly used the V8 (V8.1/8.2) gene family. Analysis of the relative MFI of V8 staining indicated that the levels of TCR on the surface of NP118 tetramer⫹ lowavidity and high-avidity T cells were essentially the same. The data represent the average and SD of three mice per group. C, Expression of the CD8 coreceptor on low-avidity NP118-specific T cells from RIP-NP mice was indistinguishable from the expression levels of CD8 expression on high-avidity NP118-specific T cells from BALB/c mice. The data represent the average and SD of three mice per group.
low-avidity T cells (25–30). Likewise, staining Ag-specific T cells with graded dilutions of peptide-tetramer reagents is a technique used to reveal differences in structural avidity (12, 26, 31–33) and corresponds well with tetramer off-rate kinetics (33). We examined the NP118 tetramer binding capacity of CD8⫹ T cells from RIP-NP and BALB/c mice at 8 days post-LCMV infection (Fig. 5). In our studies, we observed a significantly lower level of NP118 tetramer staining on virus-specific CD8⫹ T cells from LCMVinfected RIP-NP mice (MFI ⫽ 52 ⫾ 5) compared with that observed in BALB/c mice (MFI ⫽ 72 ⫾ 5) when saturating amounts of peptide tetramer were used ( p ⫽ 0.004). In particular, we noticed that the Ag-specific T cells that bound the highest levels of peptide tetramer were the most sharply depleted in RIP-NP mice (Fig. 5A). We next stained the CD8⫹ T cells with graded dilutions of NP118 tetramer to see whether there were differences in their ability to bind subsaturating amounts of peptide tetramer (Fig. 5B). We found statistically significant divergence in tetramer binding at
1236 five different dilutions, resulting in at least a 3-fold difference in the 50% maximum binding efficiency of NP118 tetramer binding. These results demonstrate that the NP118-specific CD8⫹ T cells that escaped negative selection in RIP-NP mice were indeed purged of their peripheral T cells bearing the highest structural avidity for self-peptide Ag. TCR V gene family usage by low- and high-avidity T cells We assumed that thymic deletion of self-specific T cells of high structural avidity would lead to a hole in the TCR repertoire, resulting in substantial changes in the TCR V usage of the crossreactive antiviral T cell pool. To examine this issue, we stained Ag-specific CD8⫹ T cells from LCMV-infected RIP-NP and BALB/c mice with NP118 tetramers and mAbs against several different TCR V families (Fig. 6). To our surprise, we found very few changes in the TCR V usage of T cells bearing receptors of low or high structural avidity (Fig. 6A). Although NP118-specific T cells from RIP-NP mice demonstrated a slight increase in their frequency of new V families such as V4 and V6, the dominant V families still consisted mainly of V8.1/8.2 and V10, similar to that observed in BALB/c mice (16, 34). TCR or CD8 down-regulation have been proposed as mechanisms for escape from thymic deletion (35, 36), so we next examined the expression levels of V8.1/8.2 and the CD8 coreceptor on NP118-specific T cells (Fig. 6, B and C). We found no discernable differences in the expression levels of TCR or CD8 on NP118 tetramer⫹ T cells of high or low structural avidity. This would further suggest that the differences in NP118 tetramer binding (Fig. 5) are not due to differences in TCR or CD8 coreceptor expression levels and instead are most likely due to basic differences in the structural interactions between the TCR and peptide-MHC. At 8 days post-LCMV infection, CD8 expression levels on NP118-spe-
DIFFERENTIAL EFFECTS OF NEGATIVE SELECTION cific T cells from both strains of mice were equally down-regulated in comparison to those of naive CD8⫹ T cells (data not shown), and are another indication that both groups of T cells were highly activated, effector cell populations (37, 38). These data are consistent with the phenotypic analysis shown in Fig. 4, which together argues against differences in T cell activation being a factor involved in the observed differences in structural avidity (31). In an attempt to determine whether there were major differences in the molecular structures of the NP118-specific CD8⫹ T cell populations within a specific V family, we performed complementarity-determining region 3 (CDR3) length “immunoscope” analysis, a technique also known as spectratyping (15) (Fig. 7). The length of the CDR3 region is an important factor in determining Ag recognition (29, 39 – 41), and the immunoscope approach allows one to identify shifts in CDR3 length profiles. Naive T cells typically exhibit a CDR3 profile representing a Gaussian curve centered on an arbitrary number of amino acids within the CDR3 region. In contrast, if there has been selective expansion of a subset of T cells expressing a specific CDR3 length, then the profile will be skewed in favor of the CDR3 length containing the preferred fit of the expanded T cell population (16, 34) and may differ between individual, yet genetically identical mice (34). In these experiments, we sorted NP118 tetramer⫹CD8⫹ T cells from RIP-NP and BALB/c mice at 8 days post-LCMV infection, and used the immunoscope technique to yield multiple RT-PCR bands representing the relative sizes and distribution of the various VDJ combinations in V8.1, V8.2, and V10 gene families (comprising 36 different combinations; data not shown). These different-sized bands are then represented as individual electropherogram peaks, as shown for three representative VDJ combinations in Fig. 7. Although the NP118-specific CD8⫹ T cells from LCMV-infected RIP-NP mice were consistently of low structural avidity (Fig. 5),
FIGURE 7. Immunoscope analysis of low- and high-avidity CD8⫹ T cells. The CDR3 length profiles of low-avidity (RIP-NP) or high-avidity (BALB/c) NP118-specific CD8⫹ T cells were determined using sorted NP118 tetramer⫹CD8⫹ T cells from mice at 8 days post-LCMV infection (y-axis in arbitrary units). Analysis of the V8.2-J1.1 TCR provides an example showing greater skewing of the CDR3 length polymorphisms in T cells from RIP-NP mice compared with that observed in NP118-specific T cells from BALB/c mice. However, the dominant CDR3 peaks in NP118-specific CD8⫹ T cells from two of the three RIP-NP mice still corresponded to at least one of the two dominant peaks observed in NP118-specific T cells from BALB/c mice. Examination of V8.2-J2.7 TCR revealed that NP118-specific T cells from RIP-NP mice expressed a narrower CDR3 length profile than did NP118specific T cells from BALB/c mice, whereas assessment of the V10-J1.6 TCR indicated that three of three of the samples from RIP-NP mice were identical with two of three samples obtained from BALB/c mice. Together, these results indicate that a definitive pattern of CDR3 length polymorphism could not be directly correlated with low structural avidity.
The Journal of Immunology we did not identify a defining feature/pattern in CDR3 length polymorphisms that could readily distinguish between TCR of low or high structural avidity. In some cases, such as V8.2-J1.1, we did indeed observe skewing of the low-avidity TCR CDR3 usage, which was more diverse than that observed in high-avidity T cells from BALB/c mice (Fig. 7). However, we also identified cases in which the V CDR3 regions of low-avidity T cells represented a more narrow CDR3 length usage (e.g., V8.2-J2.7), or nearly identical CDR3 length usage (e.g., V10-J1.6), as that observed in high-avidity T cell populations. Overall, these results indicate that low-avidity T cell populations are similar to high-avidity populations in that they lose the typical Gaussian curve of CDR3 length polymorphisms found in the naive T cell repertoire and are thus selected for populations representing the best fit for their stimulatory peptide Ag. However, within these polyclonal T cell populations, there appears to be no absolute loss of any particular V CDR3 length that would correspond directly with low structural avidity.
Discussion It is well established that a small proportion of lower affinity (2, 10), self-reactive T cells are capable of escaping thymic deletion and go on to survive in the periphery (1, 3–5). However, many of these studies have involved the use of TCR transgenic T cells and provide little direct information on the effects of negative selection on a polyclonal T cell repertoire or how this might affect the overall T cell response mounted against a viral pathogen. In our studies, we found that thymic and/or peripheral deletion in RIP-NP mice resulted in a significantly lower CD8⫹ T cell response against an immunodominant T cell epitope than that observed in BALB/c mice. However, expression of LCMV NP as a self-protein had virtually no effect on the peripheral T cell response against a subdominant T cell epitope, even though it is expressed on the same protein as the immunodominant epitope. These differences in negative selection caused significant skewing of the antiviral T cell epitope hierarchy. Further analysis revealed that the immunodominant NP118-specific CD8⫹ T cells from LCMV-infected RIP-NP mice were of lower functional avidity than that observed in BALB/c mice and required 30- to 100-fold more peptide to induce an effective IFN-␥ response. In contrast, the functional responsiveness of subdominant NP313-specific T cells was virtually unaltered by thymic/peripheral expression of LCMV NP. The lower functional responsiveness of NP118-specific T cells in RIP-NP mice was further verified by experiments showing that these cells were also substantially less capable of responding to various altered peptide ligands. Phenotypic analysis of several activation/ adhesion molecules indicated that low functional responsiveness was not due to insufficient T cell activation, but instead appeared to hinge on low structural avidity of the TCR, as determined by peptide-tetramer binding assays. Remarkably, although dominant T cells with the highest structural avidity for self-Ag were purged from the immune system, this did not greatly alter the TCR V gene families used by the responding T cell populations or result in the absolute deletion of any one specific CDR3 length within an individual V family. These results indicate that negative selection can be very biased, resulting in preferential deletion of high-avidity dominant T cell populations while leaving subdominant T cell responses virtually unaltered. It is unknown whether these differences are due to alterations in Ag processing or presentation (or perhaps mediated by immunoproteosomes (42)), but nevertheless, this has important implications for determining the nature and specificity of autoreactive T cell populations observed during an autoimmune response.
1237 The underlying mechanisms that determine immunodominance between T cells of different peptide specificity are not well understood. Ag processing and presentation are believed to be the main factors involved in determining T cell hierarchy (43), but this process may be further complicated by the capacity of the available CD8⫹ T cell repertoire to recognize any given peptide epitope. Our results indicate that negative selection can have a profound impact on the available T cell repertoire that is specific for a crossreactive viral peptide and that this can result in dramatic changes in the outcome of the antiviral T cell epitope hierarchy. Negative selection is not absolute, and as shown here, the low-avidity NP118-specific CD8⫹ T cells that escaped deletion in RIP-NP mice were still capable of dominating the antiviral T cell response. This suggests that other factors besides TCR avidity for peptide/ MHC must also be involved in determining immunodominance. These factors may include differences in processing/presentation of different peptide Ags (43), as well as potential differences in T cell effector functions, such as IFN-␥ synthesis (Refs. 44 and 45; F. Liu, manuscript in preparation). If virus-specific T cells coincidentally cross-react with a self-Ag, then they will likely be selected from a more narrow range of lower TCR avidities, and these potential differences in the available T cell repertoire may represent an important piece of the immunodominance puzzle. We found significant differences in the functional responsiveness of NP118-specific T cells from LCMV-infected RIP-NP mice and NP118-specific cells from BALB/c mice (Figs. 2 and 3). There was nearly a 100-fold difference in their ability to produce IFN-␥ after stimulation with subsaturating levels of peptide Ag. This confirms our previous observations noting large differences in the peptide requirements for inducing CTL activity by these low-avidity and high-avidity T cell populations (9, 10), which is consistent with the avidity model for thymic negative selection (2). To further assess these differences, we performed functional fingerprinting analysis using altered peptide ligands for stimulation (Fig. 3). We found that 10 of 10 peptide variants were immunostimulatory for at least a small subpopulation of NP118-specific CD8⫹ T cells from BALB/c mice, whereas only 5 of 10 peptide variants were able to elicit a detectable IFN-␥ response from NP118-specific T cells from RIP-NP mice. This leads to the interesting conclusion that single amino acid changes in peptide specificity have a more profound effect on the recognition and functional responsiveness of low-avidity T cell populations than high-avidity T cell populations. These results indicate that negative selection in RIP-NP mice not only deletes high-avidity, self-reactive T cells, but also preferentially depletes the more promiscuous subpopulations of T cells that recognize variants of the original selecting peptide epitope. This suggests that the lower-avidity T cells that are present within a typical autoreactive T cell repertoire may not cause disease in most instances because they are less promiscuous in terms of Ag recognition. To determine the underlying factors involved with the low functional responsiveness of NP118-specific T cells that escaped negative selection in RIP-NP mice, we examined the phenotype and surface marker expression of these Ag-specific cells. Previous studies using TCR transgenic mice have indicated that T cells that escape negative selection may do so by becoming anergic (46) or by down-regulating surface receptors such as their TCR or CD8 (35, 36). Our results indicate that the low-avidity T cells from RIP-NP mice were not anergic but instead were highly activated and expressed a variety of activation/adhesion markers at levels that were identical with those observed on virus-specific T cells from BALB/c mice (Fig. 4). Furthermore, we found no differences in the levels of TCR (V8.1/8.2) or CD8 coreceptor expression on NP118-specific T cells of low or high avidity (Fig. 6, B and C). In
1238 addition, NP118-specific CD8⫹ T cells from RIP-NP mice expressed similar levels of IFN-␥ (Fig. 1A) and, if normalized for their differences in frequency (Fig. 1), they would likely show similar levels of direct ex vivo CTL activity (9, 10) when stimulated with saturating amounts of peptide. Together, this indicates that these low-avidity T cells do not possess any obvious functional defects, aside from a significantly higher threshold of activation required to initiate their effector functions. The selective loss of high-avidity T cells in LCMV-infected RIP-NP mice could be visualized by simply measuring the MFI of peptide-tetramer staining (Fig. 5A). This result is a common feature among many model systems that examine the effects of negative selection on Ag-specific T cells (28 –30). We further quantitated the differences between the structural avidity of NP118-specific T cells in RIP-NP and BALB/c mice by staining the cells with subsaturating amounts of NP118 tetramer and found significant differences in their ability to bind low levels of peptide tetramer (Fig. 5B). These differences in avidity are likely to be due to structural constraints imposed by thymic/peripheral selection (9, 10) and not due to potential differences in selective expansion of high-avidity T cells because our previous studies in BALB/c mice clearly show that the peptide-tetramer binding capacity of NP118-specific T cells at early (days 4.5 and 5 postinfection) and late time points (day 8 postinfection) are virtually superimposable (12). In BALB/c mice, ⬃60 –70% of the Ag-specific T cells reside within just three V families consisting of V8.1, V8.2, and V10 (16). The lowavidity NP118-specific CD8⫹ T cells recovered from LCMV-infected RIP-NP mice also maintained this relatively conservative TCR V repertoire (Fig. 6). This indicates that the range and/or frequency of lower-avidity T cell subsets within a privileged V family surpasses that which would be found in other V families for a given peptide epitope. This is consistent with the results of a previous study that examined the V␣ repertoire in an H-Y-specific TCR V transgenic model system (29). In this study, negative selection in male mice resulted in H-Y-specific T cells of lower TCR avidity, even though the TCR V-transgenic T cells still used the same V␣ gene family (V␣9) as did the H-Y-specific T cells of high structural avidity. In contrast to our study, this group noted that low-avidity H-Y-specific T cells in male mice demonstrated a preferential depletion of CDR3 sequences with an optimal length of amino acids, whereas we were unable to find an absolute depletion of any one specific CDR3 length polymorphism (Fig. 7). The differences between these two studies may be that we analyzed polyclonal T cells in which changes in the CDR3 polymorphisms of the V repertoire may have been due to changes in the amino acid composition of the retained CDR3 motif or they may have been compensated by changes/deletions in the CDR3 polymorphisms of the accompanying V␣ repertoire. Our study focused on the impact of negative selection on the V repertoire and biological functions of autoreactive CD8⫹ T cells activated in vivo by a cross-reactive viral pathogen. We found that thymic selection preferentially depleted high-avidity self-reactive CD8⫹ T cells specific for an immunodominant epitope while sparing antiviral and potentially self-reactive T cells specific for a subdominant epitope. A similar dichotomy regarding negative selection of dominant and subdominant CD4⫹ T cells has also been noted (47, 48). However, direct ex vivo quantitation and analysis of TCR diversification of the dominant and subdominant T cells were not performed in these earlier studies. In addition, a comparison of how negative selection affected the structural avidity, functional avidity (cytokine production or CTL activity), and epitope promiscuity of these autoreactive T cell populations was not determined. Nevertheless, our work, combined with these previous studies (47, 48), suggests that preferential escape of subdominant
DIFFERENTIAL EFFECTS OF NEGATIVE SELECTION T cells from negative selection is probably far more common than previously believed. Self-reactive T cells require activation by an exogenous trigger such as a cross-reactive viral infection before they can cause symptoms of autoimmunity (49), and our studies (Fig. 3) suggest that negative selection preferentially deletes the most promiscuous or cross-reactive T cell populations, which might help explain why autoimmunity is relatively infrequent in the general population. By increasing our understanding of the structural and functional attributes of self-reactive T cells that escape negative selection, we might be able to develop more effective therapies for treating a variety of T cell-mediated autoimmune diseases such as insulin-dependent diabetes and multiple sclerosis. In addition, knowledge of the preferential survival of self-specific subdominant T cell populations reveals that this might be an important new avenue to pursue in the search for effective tumor cell Ags for developing antitumor vaccines to aid in the fight against cancer.
Acknowledgments We thank J. Lindsay Whitton for his continued support and scientific input. A portion of the tetramer reagents used in this study was generated through the National Institutes of Health National Institute of Allergy and Infectious Diseases core facility.
References 1. Ashton-Rickardt, P. G., A. Bandeira, J. R. Delaney, L. Van Kaer, H. P. Pircher, R. M. Zinkernagel, and S. Tonegawa. 1994. Evidence for a differential avidity model of T cell selection in the thymus. Cell 76:651. 2. Hoffmann, M. W., W. R. Heath, D. Ruschmeyer, and J. F. Miller. 1995. Deletion of high-avidity T cells by thymic epithelium. Proc. Natl. Acad. Sci. USA 92:9851. 3. Jameson, S. C., and M. J. Bevan. 1998. T-cell selection. Curr. Opin. Immunol. 10:214. 4. Sebzda, E., S. Mariathasan, T. Ohteki, R. Jones, M. F. Bachmann, and P. S. Ohashi. 1999. Selection of the T cell repertoire. Annu. Rev. Immunol. 17: 829. 5. Kishimoto, H., and J. Sprent. 2000. The thymus and negative selection. Immunol. Res. 21:315. 6. Nikolic-Zugic, J., and M. J. Bevan. 1990. Role of self-peptides in positively selecting the T-cell repertoire. Nature 344:65. 7. Surh, C. D., D. S. Lee, W. P. Fung-Leung, L. Karlsson, and J. Sprent. 1997. Thymic selection by a single MHC/peptide ligand produces a semidiverse repertoire of CD4⫹ T cells. Immunity 7:209. 8. Ashton-Rickardt, P. G., and S. Tonegawa. 1994. A differential-avidity model for T-cell selection. Immunol. Today 15:362. 9. von Herrath, M. G., J. Dockter, and M. B. Oldstone. 1994. How virus induces a rapid or slow onset insulin-dependent diabetes mellitus in a transgenic model. Immunity 1:231. 10. von Herrath, M. G., J. Dockter, M. Nerenberg, J. E. Gairin, and M. B. Oldstone. 1994. Thymic selection and adaptability of cytotoxic T lymphocyte responses in transgenic mice expressing a viral protein in the thymus. J. Exp. Med. 180:1901. 11. von Herrath, M. G., B. Coon, and M. B. Oldstone. 1997. Low-affinity cytotoxic T-lymphocytes require IFN-␥ to clear an acute viral infection. Virology 229:349. 12. Slifka, M. K., and J. L. Whitton. 2001. Functional avidity maturation of CD8⫹ T cells without selection of higher affinity TCR. Nat. Immunol. 2:711. 13. Chen, W., J. W. Yewdell, R. L. Levine, and J. R. Bennink. 1999. Modification of cysteine residues in vitro and in vivo affects the immunogenicity and antigenicity of major histocompatibility complex class I-restricted viral determinants. J. Exp. Med. 189:1757. 14. Murali-Krishna, K., J. D. Altman, M. Suresh, D. J. D. Sourdive, A. J. Zajac, J. D. Miller, J. Slansky, and R. Ahmed. 1998. Counting antigen-specific CD8 T cells: a reevaluation of bystander activation during viral infection. Immunity 8:177. 15. Pannetier, C., M. Cochet, S. Darche, A. Casrouge, M. Zoller, and P. Kourilsky. 1993. The sizes of the CDR3 hypervariable regions of the murine T-cell receptor -chains vary as a function of the recombined germ-line segments. Proc. Natl. Acad. Sci. USA 90:4319. 16. Sourdive, D. J. D., K. Murali-Krishna, J. D. Altman, A. J. Zajac, J. K. Whitmire, C. Pannetier, P. Kourilsky, B. Evavold, A. Sette, and R. Ahmed. 1998. Conserved T cell receptor repertoire in primary and memory CD8 T cell responses to an acute viral infection. J. Exp. Med. 188:71. 17. Vijh, S., and E. G. Pamer. 1997. Immunodominant and subdominant CTL responses to Listeria monocytogenes infection. J. Immunol. 158:3366. 18. Cunningham, B. C., and J. A. Wells. 1989. High-resolution epitope mapping of hGH-receptor interactions by alanine-scanning mutagenesis. Science 244:1081. 19. Bachmann, M. F., A. Oxenius, D. E. Speiser, S. Mariathasan, H. Hengartner, R. M. Zinkernagel, and P. S. Ohashi. 1997. Peptide-induced T cell receptor down-regulation on naive T cells predicts agonist/partial agonist properties and strictly correlates with T cell activation. Eur. J. Immunol. 27:2195.
The Journal of Immunology 20. Bachmann, M. F., D. E. Speiser, and P. S. Ohashi. 1997. Functional management of an antiviral cytotoxic T-cell response. J. Virol. 71:5764. 21. Ahmed, R. 1994. Virus persistence and immune memory. Semin. Virol. 5:319. 22. Bachmann, M. F., K. McKall-Faienza, R. Schmits, D. Bouchard, J. Beach, D. E. Speiser, T. W. Mak, and P. S. Ohashi. 1997. Distinct roles for LFA-1 and CD28 during activation of naive T cells: adhesion versus costimulation. Immunity 7:549. 23. Bachmann, M. F., M. Barner, and M. Kopf. 1999. CD2 sets quantitative thresholds in T cell activation. J. Exp. Med. 190:1383. 24. Harrington, L. E., M. Galvan, L. G. Baum, J. D. Altman, and R. Ahmed. 2000. Differentiating between memory and effector CD8 T cells by altered expression of cell surface O-glycans. J. Exp. Med. 191:1241. 25. Crawford, F., H. Kozono, J. White, P. Marrack, and J. Kappler. 1998. Detection of antigen-specific T cells with multivalent soluble class II MHC covalent peptide complexes. Immunity 8:675. 26. Busch, D. H., and E. G. Pamer. 1999. T cell affinity maturation by selective expansion during infection. J. Exp. Med. 189:701. 27. Yee, C., P. A. Savage, P. P. Lee, M. M. Davis, and P. D. Greenberg. 1999. Isolation of high avidity melanoma-reactive CTL from heterogeneous populations using peptide-MHC tetramers. J. Immunol. 162:2227. 28. de Visser, K. E., T. A. Cordaro, D. Kioussis, J. B. Haanen, T. N. Schumacher, and A. M. Kruisbeek. 2000. Tracing and characterization of the low-avidity selfspecific T cell repertoire. Eur. J. Immunol. 30:1458. 29. Bouneaud, C., P. Kourilsky, and P. Bousso. 2000. Impact of negative selection on the T cell repertoire reactive to a self-peptide: a large fraction of T cell clones escapes clonal deletion. Immunity 13:829. 30. Nugent, C. T., D. J. Morgan, J. A. Biggs, A. Ko, I. M. Pilip, E. G. Pamer, and L. A. Sherman. 2000. Characterization of CD8⫹ T lymphocytes that persist after peripheral tolerance to a self antigen expressed in the pancreas. J. Immunol. 164:191. 31. Fahmy, T. M., J. G. Bieler, M. Edidin, and J. P. Schneck. 2001. Increased TCR avidity after T cell activation: a mechanism for sensing low-density antigen. Immunity 14:135. 32. Bullock, T. N., D. W. Mullins, T. A. Colella, and V. H. Engelhard. 2001. Manipulation of avidity to improve effectiveness of adoptively transferred CD8⫹ T cells for melanoma immunotherapy in human MHC class I-transgenic mice. J. Immunol. 167:5824. 33. Kalergis, A. M., N. Boucheron, M. A. Doucey, E. Palmieri, E. C. Goyarts, Z. Vegh, I. F. Luescher, and S. G. Nathenson. 2001. Efficient T cell activation requires an optimal dwell-time of interaction between the TCR and the pMHC complex. Nat. Immunol. 2:229. 34. Lin, M. Y., and R. M. Welsh. 1998. Stability and diversity of T cell receptor repertoire usage during lymphocytic choriomeningitis virus infection of mice. J. Exp. Med. 188:1993. 35. Teh, H. S., H. Kishi, B. Scott, and H. Von Boehmer. 1989. Deletion of autospecific T cells in T cell receptor (TCR) transgenic mice spares cells with normal TCR levels and low levels of CD8 molecules. J. Exp. Med. 169:795.
1239 36. Schonrich, G., U. Kalinke, F. Momburg, M. Malissen, A. M. Schmitt-Verhulst, B. Malissen, G. J. Hammerling, and B. Arnold. 1991. Down-regulation of T cell receptors on self-reactive T cells as a novel mechanism for extrathymic tolerance induction. Cell 65:293. 37. Busch, D. H., I. M. Pilip, S. Vijh, and E. G. Pamer. 1998. Coordinate regulation of complex T cell populations responding to bacterial infection. Immunity 8:353. 38. Slifka, M. K., and J. L. Whitton. 2000. Activated and memory CD8⫹ T cells can be distinguished by their cytokine profiles and phenotypic markers. J. Immunol. 164:208. 39. Pantaleo, G., J. F. Demarest, H. Soudeyns, C. Graziosi, F. Denis, J. W. Adelsberger, P. Borrow, M. S. Saag, G. M. Shaw, R. P. Sekaly, et al. 1994. Major expansion of CD8⫹ T cells with a predominant V usage during the primary immune response to HIV. Nature 370:463. 40. McHeyzer-Williams, M. G., and M. M. Davis. 1995. Antigen-specific development of primary and memory T cells in vivo. Science 268:106. 41. Maryanski, J. L., C. V. Jongeneel, P. Bucher, J. L. Casanova, and P. R. Walker. 1996. Single-cell PCR analysis of TCR repertoires selected by antigen in vivo: a high magnitude CD8 response is comprised of very few clones. Immunity 4:47. 42. Chen, W., C. C. Norbury, Y. Cho, J. W. Yewdell, and J. R. Bennink. 2001. Immunoproteasomes shape immunodominance hierarchies of antiviral CD8⫹ T cells at the levels of T cell repertoire and presentation of viral antigens. J. Exp. Med. 193:1319. 43. Yewdell, J. W., and J. R. Bennink. 1999. Immunodominance in major histocompatibility complex class I-restricted T lymphocyte responses. Annu. Rev. Immunol. 17:51. 44. Badovinac, V. P., A. R. Tvinnereim, and J. T. Harty. 2000. Regulation of antigenspecific CD8⫹ T cell homeostasis by perforin and interferon-␥. Science 290: 1354. 45. Rodriguez, F., S. Harkins, M. K. Slifka, and J. L. Whitton. 2002. Immunodominance in virus-induced CD8⫹ T-cell responses is dramatically modified by DNA immunization and is regulated by ␥-interferon. J. Virol. 76:4251. 46. Stockinger, B. 1999. T lymphocyte tolerance: from thymic deletion to peripheral control mechanisms. Adv. Immunol. 71:229. 47. Cibotti, R., J. M. Kanellopoulos, J. P. Cabaniols, O. Halle-Panenko, K. Kosmatopoulos, E. Sercarz, and P. Kourilsky. 1992. Tolerance to a self-protein involves its immunodominant but does not involve its subdominant determinants. Proc. Natl. Acad. Sci. USA 89:416. 48. Klein, L., T. Klein, U. Ruther, and B. Kyewski. 1998. CD4 T cell tolerance to human C-reactive protein, an inducible serum protein, is mediated by medullary thymic epithelium. J. Exp. Med. 188:5. 49. von Herrath, M. G., C. F. Evans, M. S. Horwitz, and M. B. Oldstone. 1996. Using transgenic mouse models to dissect the pathogenesis of virus-induced autoimmune disorders of the islets of Langerhans and the central nervous system. Immunol. Rev. 152:111.
