Role of antigen persistence and dose for CD4⁺ T-cell exhaustion and recovery

Abstract

It is currently not understood how some chronic infections exhaust antigen-specific T cells over time and which pathogen components contribute to exhaustion. Here, we dissected the behavior of primed CD4⁺ T cells exposed to persistent antigen using an inducible transgenic mouse system that allowed us to control antigen presentation as the only experimental variable, independent of the persistent inflammation and disease progression that complicate infectious models. Moreover, this system restricted antigen presentation to dendritic cells (DCs) and avoided confounding B, CD8⁺ T, or innate cell responses. When antigen presentation was extended beyond the expansion phase, primed CD4⁺ T cells survived, but exhibited reduced memory functionality in terms of their proliferative capacity and cytokine expression potential. The effect was antigen dose and time dependent, not associated with increased PD-1 expression or reduced calcium influx, but impaired Jun phosphorylation in response to TCR engagement. Upon antigen removal, the cells regained the ability to proliferate, but remained unable to produce high levels of IL-2 and TNF-α. These data show that persistent antigen by itself rapidly induces a dysfunctional state in CD4⁺ T cells that is only partially reversible upon antigen removal. These findings have implications for vaccine optimization and for the possible reinvigoration of CD4⁺ T cells during chronic infection.

The high prevalence of chronic infections raises the issue of how T cells respond to persistent antigen presentation without causing immunopathology. Because chronic infections differ widely in terms of tropism, cytopathy, immune evasion strategies, replicative productivity, innate triggers, and antigenic load, it is likely that multiple strategies of lymphocyte inactivation exist (1–3). In persistent infections, such as those by hepatitis C virus (HCV) and HIV in humans, as well as chronic lymphocytic choriomeningitis virus (LCMV) variants in mice, T-cell functions become impaired over time. Such “exhaustion” has been correlated with specific gene expression profiles, negatively signaling receptors and cytokines, raising the promise of intervention to reinvigorate T-cell effector functions (4–8). Because exhausted T cells have been described in mice, nonhuman primates, and humans, it is likely that evolutionarily conserved mechanisms are in place to avoid T-cell-induced tissue damage and pathology. However, it is not known why some persisting infections, such as HIV and HCV, lead to T-cell exhaustion, whereas in others, such as EBV and CMV, T cells control viral spread throughout life despite repeated antigen encounters (1–3).

Although T-cell exhaustion has been mostly described for CD8⁺ T cells, a dysfunctional CD4⁺ T-cell compartment has been also observed in chronic infections. Because CD4⁺ T-cell help is required for raising and maintaining an efficient CD8⁺ T-cell response, CD8⁺ T cell controlled infections by chronic LCMV variants persist much longer once CD4⁺ T cells are deleted by mAb treatment or by genetic means (9–11). The Th1 functionality of CD4⁺ T cells is reduced by persisting LCMV (11, 12), although the cells are still able to transmit help to CTL via IL-21, indicating functional plasticity of Th cells faced with persistent antigen (13). Similar Th-cell inactivation has been described for chronic infections by gammaherpesvirus-68 and schistosomes (14, 15). Chronic infections in man also induce dysfunctionality followed by reduction and loss of specific CD4⁺ T cells. Numbers and functions of specific CD4⁺ T cells deteriorate in patients chronically infected by HCV, whereas in those that stably resolve the infection, T-cell percentages remain sufficiently high for detection in blood (16). Persisting HIV causes the CD4⁺ T-cell compartment to become nonproliferative to antigen and unable to produce IL-2 before its final decay (17).

Collectively, these data emphasize the need to understand the lack of CD4⁺ T-cell help in chronic infections and how it develops. However, experiments with infectious agents make it difficult to establish the causality of exhaustion for two reasons: first, T-cell exhaustion and pathogen persistence depend mutually on each other so that changing one necessarily affects the other (3). Second, pathogens trigger both innate and adaptive immune compartments with each of their own tolerance mechanisms (18, 19). While avoiding these problems, we show in this paper how persistent antigen by itself interferes with Th memory cell differentiation after expansion, without involving TCR-triggered calcium fluxes or PD-1 expression. Some memory cell functions cannot be fully regained even after extended periods of antigen absence, suggesting irreversible changes. We also show that the minimal antigen dose required for full exhaustion is higher than for priming, and that exhaustion by a low antigenic load deepens over time, but remains incomplete. These data demonstrate the tuning of T-cell activation thresholds over time and supply a minimal framework for understanding Th-cell exhaustion.

Results

Reliable Antigen Clearance by Molecular Switch.

In the double-transgenic (dtg) mouse model used here, the H-2E^k-restricted presentation of the moth cytochrome c (MCC) epitope MCC_93–103 can be induced in dendritic cells (DCs), but not B cells, by feeding the animals with doxycycline (dox) and rapidly turned back off upon dox withdrawal (Fig. S1A) (20). To use this model for chronic antigen presentation, we tested whether E^k/MCC complexes can be chronically presented by and removed from activated DCs. Dtg mice were treated with a stimulatory anti-CD40 mAb for DC activation when dox feeding was initiated. Four days of dox treatment were followed by a period 0–3 wk during which the animals were fed with normal (Fig. 1 Left) or dox-containing water (Fig. 1 Right). AND TCR-transgenic T cells labeled with carboxyfluorescein succinimidyl ester (CFSE) were then transferred as antigen indicators and analyzed 3 d later. The data indicated that 1 wk after turn-off, the MCC peptide is not presented any longer (Fig. 1 Left). However, the prolonged dox treatment regimes up to 4 wk do not impair antigen display by DCs in this model (Fig. 1 Right) and make it suitable for exploring the effects of long-term antigen presentation.

Fig. 1.

Reliable antigen clearance after innocuous long-term dox treatment of dtg mice. Antigen becomes undetectable 1 wk after TIM turnoff, whereas long-term dox treatment does not affect antigen presentation. Dtg mice were treated with anti-CD40 and dox for 4 d as indicated and fed with normal (Left) or dox-containing drinking water (Right) for 0, 1, 2, and 3 additional weeks (indicated by hatched boxes). Then CFSE-labeled AND indicator T cells were transferred and splenocytes analyzed 3 d later. The AND T cells shown were identified as CD4⁺ CD45.1⁺. The data are representative of two independent experiments.

Primed CD4⁺ T Cells Survive Despite Persistent Antigen Presentation by DCs.

Previously, we have established a treatment regime of dtg animals that converts adoptively transferred naive AND T cells efficiently into memory cells (21). This priming treatment consisted of an i.p. injection of the anti-CD40 mAb on day −1 for DC activation, the transfer of congenically marked AND T cells on day 0, and dox feeding from day −1 to day 3. The cells expanded in an antigen-dependent manner and were readily detectable 31 d after transfer (Fig. 2A). The effector/memory cell marker CD44 was up-regulated by day 3 (Fig. 1A), whereas Ly6C, a molecule more truthfully associated with CD4⁺ memory T-cell features (22), was expressed only later (Fig. 2B). To see how antigen persistence beyond the effector phase affects T-cell differentiation, we continued the dox treatment throughout the observation period and found that the transferred cells survived for this period with a CD44⁺ Ly6C^– phenotype (Fig. 2 A–C). Unlike CD44, Ly6C expression passes through a period of incomplete expression between days 6 and 10 and is then reduced by antigen persistence (Fig. 2 B and C). AND cells from lymph nodes, bone marrow, and liver (where fewer cells were found) also expressed lower levels of Ly6C with persisting antigen (Fig. S2). When the cells were acutely restimulated in cultures with phorbol myristate acetate (PMA) and ionomycin (IM), the CD40 ligand and early activation marker CD154 was not up-regulated by chronically stimulated cells on day 10 and 31, indicating that T-cell activation and potential interaction with CD40-expressing DCs and B cells becomes impaired by antigen persistence (Fig. 2C). In addition to these phenotypic changes, the chronically stimulated cells are slowly lost from the recipients by day 31 (Fig. S3). These data suggest that CD4⁺ T-cell differentiation toward a memory phenotype and their survival depend on antigen disappearance after the expansion phase and may be blocked or deviated by continued antigen presentation.

Fig. 2.

Comparison of antigen-specific CD4⁺ T cells in recipients with transient and continuous antigen expression. (A) CD45.1⁺ AND TCR-transgenic T cells were transferred into dtg animals that had been injected with anti-CD40 1 d before and fed with dox. The recipients were then treated with dox for 3 more days (Upper) or continuously (Lower) and analyzed at the time points indicated. Shown are CD4⁺ splenocytes. (B) As in A, cells shown have been gated on adoptively transferred CD4⁺ CD45.1⁺ T cells stained for Ly6C. (C) Compilation of CD44 (Left) and Ly6C (Center) expression by AND T cells and their CD154 responsiveness to in vitro PMA/IM restimulation (Right). The percentage of AND cells positive for the indicated markers are depicted. Data from three separate experiments with one animal per condition are shown. Each point represents an individual recipient. *P < 0.05; **P < 0.005.

Impairment of Memory Functions by Antigen Persisting Beyond the Priming Phase.

To evaluate the functionality of T cells that were chronically stimulated, we compared adoptively transferred AND cells from dtg animals that had been dox-treated for 0, 7, or 21 additional days following full priming. For secondary stimulation, the animals were again treated with the anti-CD40 mAb and dox on day 28. When the cells were analyzed 3 d later (i.e., 31 d after transfer), we first noticed that the memory cells expanded less efficiently than naive cells (Fig. 3A, treatments 1 and 2). This finding is in agreement with results that indicate reduced proliferation capacity of CD4⁺ memory cells compared with naive ones (23). Furthermore, expansion was increasingly hampered by antigen persisting for 10 and 24 d (Fig. 3A, treatments 3 and 4). More directly, >90% of naive and memory cells incorporated BrdU over the 3 d-period of rechallenge, and the uptake was reduced in mice with antigen presentation sustained for 10 and 24 d (Fig. 3B, treatments 3 and 4). These data indicate that the proliferative capacity of CD4⁺ memory T cells is suppressed by antigen presentation extending beyond the priming phase.

Fig. 3.

Secondary expansion of CD4⁺ T cells after extended periods of antigen presentation. (A) CD45.1⁺ AND TCR-transgenic T cells were transferred into dtg animals that were treated with extended dox treatments for 1, 3, or 4 additional weeks (Left). During the last 3 d, animals were treated with BrdU (rechallenge) or anti-CD40, dox, and BrdU for secondary expansion 3 d (+rechallenge) before analysis. A total of 31 d after transfer, splenocytes were analyzed. The cells shown in the cytometry panels are CD4⁺. (Right) Results of five separate experiments (one recipient per condition), with the difference between means ± SEM indicated in the upper left. (B) DNA synthesis of AND T cells from dtg recipients treated with BrdU during the last 3 d before analysis. Intracellular stainings with anti-BrdU (black line) and control (gray fill) mAbs of CD4⁺ CD45.1⁺ splenocytes are shown. Results are representative of three separate experiments.

To further assess the cells' functionality, we tested them for Ly6C and cytokine expression. We again found Ly6C expressed on memory cells, but down-regulated by continuous antigen presentation (Fig. 4A, treatments 1 and 5). Interestingly, the cells that had been exposed to antigen for 10 and 24 d down-regulated Ly6C partially, suggesting that persistent antigen might compromise their memory features irreversibly (Fig. 4A). For cytokine secretion, we found a differential shut-down: IL-2 and TNF-α were reduced by the 10- and 24-d treatments, whereas IFN-γ secretion was not (Fig. 4B, treatments 2–4). However, memory cells produced all three cytokines (Fig. 4B, treatment 1) as rapid secretion in response to brief stimulation with PMA and IM is their key feature (24, 25). Our data suggest differential susceptibilities of these three cytokines to suppression by persistent antigen, probably via stepwise impairment by, or differential recovery from, antigenic exhaustion. Fig. 4C visualizes the analyses on day 31 side-by-side. The induction of the cytokines IL-2 and TNF-α upon PMA/IM stimulation, as well as the expression of the memory marker Ly6C, were still reduced despite antigen turnoff 3 wk before. In contrast, the expression of IFN-γ was only insignificantly reduced even when antigen was presented for 24 d, which indicates a distinct mode of regulation. Proliferation appeared to be in a third category, nested in between the two above, with a slight change by 10 d of dox treatment, but a reduction from 90% to 50% of divided cells by the 24 d-treatment. In summary, these data show the differential regulation of CD4⁺ T memory characteristics by prolonged antigen exposure.

Fig. 4.

Phenotypic changes and functional impairment of CD4⁺ memory cells due to persistent antigen presentation, analyzed on day 31. (A) Ly6C surface expression of AND T cells adoptively transferred into dtg recipients treated. AND T cells were identified as CD4⁺ CD45.1⁺ splenocytes. (B) Cytokine production of AND T cells in dtg recipients that were treated. Following in vitro PMA/IM restimulation, splenocytes were stained intracellularly with mAbs for the cytokines indicated (black lines) and controls (gray fill). Shown are CD4⁺ CD45.1⁺ splenocytes. (C) Overview of AND memory T-cell impairment by antigen exposure for different periods of time after transfer. The percentage of AND cells positive for the indicated markers are depicted. Each panel shows the results of three separate experiments with one recipient per condition. The BrdU panel depicts incorporation during a 3-d rechallenge period as described for Fig. 3B.

Functional Impairment by Day 10.

To clarify when memory differentiation was impaired by antigen, we analyzed the transferred AND T cells on day 10 following transient and continuous dox treatments. When the recipients were treated with anti-CD40 and dox for one additional day, the activation marker CD69 was induced, unless the cells had been chronically stimulated previously (Fig. 5A Left). During treatment with dox and BrdU for additional 3 d, 78.6% (±6.0 SEM) of the cells incorporated BrdU in transiently treated recipients, but only 28.4% (± 9.6) did in continuously treated animals (Fig. 5A Right). This demonstrates that antigen persistence beyond the expansion phase severely limits CD4⁺ T-cell activation and proliferation by day 10 already. The capacity to express CD154, IL-2, TNF-α, and IFN-γ was suppressed as well, IFN-γ, however, less completely so (Fig. 5B). Overall, the data indicate that unimpeded antigen presentation by DCs for 10 d induces a profound functional impairment up to this time point. Because IL-2, TNF-α, and Ly6C expression were still reduced 21 d later despite the absence of antigen (Fig. 4), it is likely that these parameters were irreversibly silenced by day 10. However, IFN-γ production and proliferation were much harder to suppress and bounced back easily.

Fig. 5.

Functional impairment of CD4⁺ T cell by persistent antigen presentation on day 10. Dtg recipients of AND T cells were treated as indicated. (A) Responsiveness of AND T cells to rechallenge in vivo. (Left) Activation marker CD69 expression of AND T cells from recipients treated and after 1 d of additional dox treatment. (Right) Proliferative capacity. Dtg recipients were treated with dox and BrdU from day 10–13, and splenocytes were intracellularly stained with BrdU-specific (black line) and control (gray fill) mAbs on day 13. Panels below depict the percentage of AND cells positive for the indicated markers and show the results of three separate experiments. (B) CD154 and cytokine expression capacity. Intracellular expression was visualized with specific (black line) and control (gray fill) mAbs following PMA/IM restimulation in vitro. Panels below depict results of three (CD154) and six (cytokines) separate experiments. (C) Surface expression of TCR and PD-1. Expression by AND T cells in differently treated recipients are depicted, with recipient 4 being an anti-CD40-treated Ii-MCC-transgenic animal. For TCR expression, mean fluorescence intensity (MFI) values are given in the upper left, with the relative expression compared with recipient 1 in brackets. The panels below show the relative expression of TCR (as percentage of naive cells of recipient 1) and PD-1 expression (as fold change over the control mAb staining) from six experiments. (D) Calcium fluxes in response to CD3/CD4 cross-linking. Splenocytes from animals transferred with sorted CD4⁺ AND T cells were labeled with Fluo-4 und Fura-Red, coated with anti-CD3 and anti-CD4 mAbs, and activated with streptavidin (SA) at 60 s and IM after 240 s as indicated. The ratio of Fluo-4 and Fura-Red fluorescences of CD45.1⁺ cells over time are shown. Recipient 4 is an anti-CD40-treated Ii-MCC-transgenic animal. The panel below shows the peak hights triggered by SA compared with the ones caused by IM, from three separate experiments. (E) Jun phosphorylation in response to in vivo challenge with the MCC peptide. AND recipients were treated with peptide (Right) or not (Left) and splenocytes were fixed ex vivo 1 h later and stained intracellularly with phospho-Jun-specific and control mAbs. The panel below shows the MFI ratios of phospho-Jun-specific stainings with and without peptide treatment from three separate experiments. In all cases except D, AND T cells were identified as CD4⁺ CD45.1⁺, with the black lines showing the specific and the gray fills the control stains. n.d., not determined. *P < 0.05; **P < 0.005; ***P < 0.0005.

We tested whether conversion to regulatory T cells was induced by persistent antigen and found that foxP3 was not induced by the transferred AND TCR-transgenic T cells (Fig. S4A). TCR expression has been found down-regulated on autoreactive T cells in some systems (26, 27) but was not reduced here, despite continuous antigen presentation (Fig. 5C Left). PD-1 expression, frequently associated with chronic infections, was not induced (Fig. 5C Right; see Fig. S4B for day 31). This indicates that continuous TCR signals are not sufficient for the long-term up-regulation of this marker, in agreement with a recent report on PD-1 induction by bacterial products and inflammatory cytokines (28). For comparison, we also used transgenic recipients, named Ii-MCC, that constitutively express an Ii-MCC_88–103 fusion protein under the control of an MHC class II promoter (29). Though in dtg animals the MCC-carrying reporter gene is expressed in DCs (20), the MCC-modified Ii mRNA is equally well detectable in DCs and B cells of Ii-MCC transgenics (Table S1). Here the TCR was found down- and PD-1 up-regulated on adoptively transferred AND T cells (Fig. 5C). These data suggest that T-cell exhaustion does not necessarily involve TCR down-regulation nor PD-1 expression, and that several tolerance-inducing mechanisms of exhaustion are operative in T cells, probably depending on the antigen-presenting cell type or antigen dose.

Continuous Antigen Presentation by DCs Does Not Impair TCR-Induced Calcium Fluxes, but Jun Phosphorylation.

Hyporesponsive or anergic states of CD4⁺ T cells—induced by low-dose ionophore treatment, transfer into transgenic animals expressing antigen systemically, or injection of fixed APCs—have been correlated with reduced calcium fluxes in response to TCR cross-linking (30–32). This signal dampening correlates with the degradation of TCR-proximal signaling components by E3 ubiquitin ligases (33). The common denominator of these models is incomplete T-cell priming due to the lack of costimulation or preferential engagement of coinhibitory receptors. However, in our system, the T cells were first primed by activated DCs but then acquired differentially tolerized phenotypes over time, depending on antigen persistence. To find out whether antigen-exhausted CD4⁺ T cells are anergized in this sense, we measured the cells' ability to mobilize calcium in response to CD3 and CD4 stimulation 10 d after transfer. Regardless of dox treatment, the AND cells were able to flux calcium not only in response to IM, but to CD3/CD4 cross-linking as well (Fig. 5D, recipients 1–3). Even after 21 additional days of antigen exposure, calcium influx was only slightly reduced (Fig. S4C). For comparison, we also used the Ii-MCC recipients mentioned before, with antigen presentation in both DCs and B cells. Ten days after adoptive transfer into these animals, AND T cells were unable to mobilize calcium (Fig. 5D, animal 4). These data suggest a different signaling block involving the TCR or TCR-associated molecules. To examine another pathway of TCR signaling, we measured the phosphorylation of the transcription factor Jun in response to i.v. peptide injection (34). This pathway appeared much more susceptible to antigen persistence as Jun phosphorylation in chronically stimulated cells was reduced compared with naive and transiently stimulated ones (Fig. 5E; see Fig. S4D for data on day 31). In summary, these findings suggest that persistent antigen presentation by DCs induces a hyporesponsive state in CD4⁺ T cells that is based not on TCR-proximal signaling impairment or PD-1 expression, but rather on the compromised signal transduction between the TCR and Jun phosphorylation.

Time and Dose Affect T-Cell Exhaustion.

To mimic viral antigen persistence at a low level, we established a dox dose that led to low but reproducible antigen presentation in dtg mice. A total of 10 μg/mL dox in the drinking water, 1/10th of the dose used until now, led reproducibly to AND T-cell proliferation in vivo (Fig. S5A). When the recipient dtg animals were treated exactly as in the exhaustion setup with the high dose in the first 4 d followed by a week of low-dose treatment, almost all of the cells proliferated (Fig. S5B). In both cases the cells divided with a wider spread of the CFSE peaks, indicating less-synchronized antigen encounters and a swift equilibration of dox. Importantly, both experiments demonstrate a quantitative threshold of T-cell priming between 0 and 10 μg/mL of dox. However, the threshold of exhaustion was found higher, ranging between 10 and 100 μg/mL Actually, none of the parameters investigated, memory marker Ly6C expression, cytokine production, and proliferative capacity, was significantly affected by low-dose antigen presentation by day 10 (Fig. S6A). These data suggest a window of opportunity for Th cells to fight pathogens persisting at low levels.

To assess whether these functions would change by an extended exhaustion period, cells were analyzed following 28 d at the low dox dose. All of the features tested were reduced compared with the memory cells, albeit none of them completely (Fig. S6B). These findings indicate that the exhausting signals of antigen presented at a low dose do not lead to complete unresponsiveness, but can nevertheless be integrated over surprisingly long periods of time.

Discussion

We have shown here that mere antigen persistence beyond the priming phase is sufficient to impair memory Th-cell differentiation, and how dose and time of exhausting antigen presentation affect the process. The necessity of antigen persistence for CTL exhaustion has been suggested previously in viral systems using escape mutants, bone marrow chimeras, or repeated injections via a noninfectius route (35–38). However, chronic infections, especially by LCMV, can induce immunosuppression and lymphopenia for weeks (6, 39, 40) and trigger immune mechanisms that impact significantly on parenchymal MHC molecule expression, hematopoietic stem cell, and DC turnover (41–43), thereby hampering causal analysis. In the experiments reported here, CD4⁺ T-cell deviation from a memory to an exhausted phenotype is dissected involving no other immune components but antigen. We show that T-cell functionality is dampened gradually in both a time- and dose-dependent manner. Our experiments revealed that persisting antigen interferes with memory differentiation by day 10 if present at high doses. Because antigen persistence is necessary in the expansion phase of CD4⁺ T cells, a switch occurs later between days 3 and 10 that turns the stimulatory TCR signal into an inhibitory one. Because the cell cycle subsides concomitantly, it is tempting to speculate that a TCR-dependent inhibitory factor accumulates in the absence of proliferation and leads to the exhausted phenotype. Early work on anergy induction by TCR triggering with inhibition of proliferation already suggested such a mechanism (44).

As protective Th1 memory has been correlated with multifunctionality (45), the down-regulation of select cytokines by persisting antigen is important for optimal vaccination regimes. Because IL-2 and TNF-α production by T cells disappear first in the course of chronic LCMV infections (12, 46, 47), we found the same cytokines most susceptible to chronic antigen persistence. The fact that the potential to produce IL-2 and TNF-α did not recover from exhaustion, whereas the proliferative capacity did, suggested programming and specific regulation more downstream of the TCR. Consequently, calcium fluxes were hardly changed in chronically stimulated AND T cells, in agreement with data on LCMV-exhausted CD8⁺ T cells (48). TCR signaling to Jun phosphorylation, however, was reduced. This pathway has been found compromised in anergic T cells (49, 50), and perhaps similar mechanisms dampening signal transmission from phospholipase c to the Fos/Jun complex via the mitogen-activated protein kinase pathway are in place (51, 52). This compromised pathway may also contribute to the slow cell death we observed, because homeostatic survival signals are not properly transduced. In support of this, a recently identified molecule required for homeostatic signaling in T and B cells is a member of the ras family (53).

It also became clear that IL-2 does not govern IFN-γ expression in this system, and it is likely that cytokines are suppressed independently by persisting antigen. It remains to be determined whether IL-2 and TNF-α expression are silenced by epigenetic modifications while cell-cycle progression and IFN-γ production are regulated by transcriptional or posttranslational means. Interestingly, AND cells transferred into Ii-MCC animals that express the E^k/MCC epitope on both DCs and B cells expressed PD-1, down-regulated their TCR and were unable to flux calcium upon cross-linking, reminiscent of classical work demonstrating CD4⁺ T-cell tolerance induction by naive B cells. Although the precise mechanisms of TCR down-regulation and PD-1 induction in this system remain to be elucidated, the recent finding that T cells can distinguish DCs from B cells by specific receptors (54) make this context dependence of adaptive tolerance plausible.

The finding that the low antigen dose impairs functionality significantly only weeks later, and even then not completely, suggests that late effector and memory T cells, unlike thymocytes, adapt dynamically to persistent antigenic loads and maintain partial functionality even after prolonged antigen exposure (55). This plasticity may explain that T cells can protect against chronic infections with low or partitioned viral loads like murine CMV or gammaherpesvirus-68 in mice, or HSV, CMV, and EBV in man.

In summary, our findings suggest that antigen-mediated exhaustion is quickly induced, established in a time- and dose-dependent manner, and partially irreversible. Pathogens probably use and amplify these basic tolerance mechanisms to escape specific T cells.

Materials and Methods

For additional details, see SI Materials and Methods.

All animals were kept on a B10.BR-H2^kH2-T18^a/SgSnJ background (The Jackson Laboratory). Dtg animals carry the Ii-rTA [Tg(Cd74-rtTA)#Doi according to the Mouse Genome Database] (56) and TIM [Tg(tetO-Cd74/MCC)#Doi] transgenes described previously (20). The CD45.1 or CD90.1 congenic markers for AND TCR transgenic T cells [Tg(TcrAND)53Hed] (57) were originally derived from B6.SJL-Ptprc^a Pepc^b/BoyJ and B6.PL-Thy1^a/CyJ animals (The Jackson Laboratory), respectively. Ii-MCC-transgenic animals [Tg(H2-Ea-Cd74/MCC)37GNnak] express an Ii-MCC_88–103 fusion protein under the control of an H-2Eα promoter (29). All animals were housed and bred, and experiments were conducted, in compliance with German federal guidelines and approved by the Government of Upper Bavaria.