Ablation of CD8 and CD4 T Cell Responses by High Viral Loads¹

Cells of the adaptive immune system, including B cells, CD4 T cells, and CD8 T cells, act cooperatively to combat viral infections (1, 2, 3, 4, 5). Virus clearance represents an immunological success story and ideally renders the host immune to reinfection by the same pathogen. However, certain infections are not completely eradicated, but are instead brought under control at a steady state level (1, 2, 3, 4, 6). This suggests that under certain conditions the immune response is ineffective and the biological consequence of this is virus persistence. A combination of factors can contribute to virus persistence, including the deployment of immune evasion mechanisms by the pathogen and the inability of the host to induce or sustain appropriate immunological effector mechanisms (7, 8, 9, 10, 11).

CD8 T cells are key mediators of the immune response against viruses (1, 3, 4, 10). During a typical acute viral infection, naive CD8 T cells undergo massive proliferation as they differentiate and acquire effector capabilities (3, 4, 12, 13, 14, 15, 16). As the infection is resolved, the pool of virus-specific CD8 T cells contracts, homeostasis is restored, and a stable population of memory cells is established (1, 3, 4). The virus-specific CD8 T cells that are elicited by an acute infection are not usually a monoclonal population of cells, but instead are comprised of oligoclonal subsets that recognize a variety of distinct viral epitopes (17, 18, 19, 20). In addition, CD8 T cells can use multiple effector functions to control viral infections (1, 3, 4). They can kill cells that have become infected with a virus, primarily by secreting perforin and serine esterases (granzymes) (3, 21, 22, 23, 24, 25). CD8 T cells can also control viral infections by inducing an anti-viral state through the release of cytokines such as IFN-γ and TNF-α (26, 27).

The failure to mount an adaptive immune response is not a prerequisite for the establishment of chronic infections, as a hallmark of the acute phase of many persistent infections is the induction of clearly discernable populations of virus-specific CD8 T cells (3, 4, 28, 29, 30). These responses can, however, fail to clear the virus for a variety of reasons, including epitope escape, the skewed maturation of memory CD8 T cells, the deletion of cells reactive against particular viral Ags, and selective defects in cytokine production and/or cytotoxicity (9, 28, 29, 31, 32, 33, 34, 35, 36, 37, 38, 39).

Viral infections also induce CD4 T cell responses, and because of their multifaceted roles, these lymphocytes are critical for the successful resolution of many viral infections. CD4 T cell-derived cytokines promote T cell proliferation and activation, help B cell responses, and induce Ig class switching (2, 4, 5, 40, 41). CD4 T cells can also activate dendritic cells, exhibit regulatory properties, and, under certain circumstances, elaborate cytotoxic effector functions (2). Our previous studies have shown that CD4 T cell-deficient mice cannot control infection with certain isolates of lymphocytic choriomeningitis virus (LCMV)³ (29, 42). Under these conditions virus-specific CD8 T cell responses are induced, but their effector functions diminish over time (42). This suggests that the combination of persistently high viral burdens and the lack of CD4 T cell help negatively regulates CD8 T cell responses. Nevertheless, the contribution of high viral load vs absence of CD4 T cell help in silencing CD8 T cell responses remains unclear.

The purpose of the current study was to dissect the interplay between viral load and CD4 T cell responses in maintaining virus-specific CD8 T cell activity. Perforin-deficient (PKO) mice have intact CD4 and CD8 T cell compartments, but because of their inability to express perforin they cannot control isolates of LCMV that cause only an acute infection in normal adult mice (6, 21, 23, 43). This provided us with an attractive system to skew an acute to a chronic infection in hosts that possess CD4 T cells and determine the impact of high viral loads on virus-specific T cell responses.

C57BL/6 (B6) mice and perforin-deficient C57BL/6-Pfp^tm1Sdz (PKO) mice (21) (both H-2^b) as well as DBA/2 mice (H-2^d) were purchased from The Jackson Laboratory (Bar Harbor, ME). Animals were bred and maintained in accredited facilities at University of Alabama, Birmingham, AL). Male and female mice between 6–10 wk of age were used.

The Armstrong isolate of LCMV was provided by Dr. R. Ahmed (Emory University, Atlanta, GA). Plaque-purified viral isolates were propagated in BHK-21 cells. Mice were infected by i.p. inoculation with 2 × 10⁵ PFU of LCMV-Armstrong in a volume of 0.5 ml and were sacrificed at the indicated time points. The titers of viral stocks and serum samples were determined by plaque assays using Vero cell monolayers (44).

Freshly explanted spleens were disrupted into single-cell suspensions using wire mesh screens. Erythrocytes were subsequently removed by lysis using 0.83% (w/v) NH₄Cl. After washing, splenocytes were finally resuspended in RPMI 1640 medium supplemented with 10% FCS, 50 μM 2-ME, 100 U/ml penicillin, and 100 μg/ml streptomycin.

The methodology for preparation of MHC class I tetramers was modified from the protocol described by Altman and coworkers (45). Recombinant MHC class I molecules fused to a BirA substrate peptide were produced in Escherichia coli BL21 (DE3) cells, and expression was induced with isopropyl β-d-thiogalactopyranoside. Recombinant β₂-microglobulin was similarly produced. After 4 h of induction, bacteria were harvested by centrifugation at 2800 × g, and cell pellets were stored at −70°C. Thawed bacterial pellets were resuspended in 100 mM NaCl, 100 mM NaH₂PO₄, and 5% (v/v) glycerol and were disrupted using a French press. Suspensions were then centrifuged at 9800 × g for 1 h at 4°C. Inclusion body pellets were resuspended in 50 mM Tris, 100 mM NaCl, 1 mM EDTA, 1 mM DTT, 0.1% (w/v) NaN₃, and 0.5% (v/v) Triton X-100 (wash buffer) and subsequently centrifuged at 4000 × g for 20 min. Pellets were washed twice more in wash buffer containing 0.5% (v/v) Triton X-100, twice in wash buffer without Triton X-100, and once in water. Recombinant proteins were solubilized in 8 M urea, 10 mM EDTA, 0.1 mM DTT, and 25 mM MES (pH 6.0), aliquoted, and snap-frozen. MHC-peptide complexes were generated by folding class I H chains in vitro with β₂-microglobulin and defined LCMV-derived peptide epitopes as previously described. Reaction mixtures were concentrated using a stirred ultrafiltration cell (Amicon, Bedford, MA) with a 30-kDa exclusion limit membrane. Buffer exchange into 200 mM NaCl, 5 mM MgCl₂, and 100 mM Tris (pH 7.6) was accomplished using PD-10 desalting columns (Amersham Pharmacia Biotech, Piscataway, NJ). Enzymatic biotinylation reactions using BirA were allowed to proceed for 18 h at 23°C. Reaction mixtures were then applied to a Sepharose G-75 column (Amersham Pharmacia Biotech), and fractions containing biotinylated class I-β₂-microglobulin-peptide complexes were pooled and exchanged into 20 mM Tris (pH 8.0) using centrifugal filtration devices (Millipore, Bedford, MA). The complexes were further purified by ion exchange chromatography using a QM-Sepharose column (Amersham Pharmacia Biotech). Appropriate fractions were concentrated and dialyzed against PBS containing 2 mM EDTA, 1 μg/ml of leupeptin, 1 μM pepstatin, and 200 μM PMSF (Sigma-Aldrich, St. Louis, MO). The complexes were tetramerized by the stepwise addition of allophycocyanin-conjugated streptavidin (Molecular Probes, Eugene, OR).

Splenocyte preparations were stained in PBS containing 2% (w/v) BSA and 0.2% (w/v) NaN₃ (FACS buffer). Samples were costained with anti-CD43-FITC Abs (clone 1B11; BD PharMingen, San Diego, CA) together with allophycocyanin-conjugated MHC class I tetramers complexed with either the H-2D^b-restricted LCMV epitopes gp33–41, nucleoprotein (NP)_396–404, and gp276–286 or the H-2K^b-restricted epitope NP_205–212. For costains with H-2D^b tetramers, the anti-CD8-PE Ab clone 53-6.7 (BD PharMingen or eBioscience (San Diego, CA)) was included; however, for costains with the H-2K^b (NP_205–212) tetramer the anti-CD8-PE Ab clone CT-CD8a (Caltag, Burlingame, CA) was used. After staining, cells were washed three times in FACS buffer and fixed in PBS containing 2% (w/v) paraformaldehyde. Flow cytometry was performed using a FACSCalibur instrument (BD Biosciences, San Jose, CA), and data were analyzed using the computer program CellQuest.

Stimulations and intracellular cytokine analysis were performed essentially as previously described (46). Briefly, splenocytes were either untreated or were stimulated with LCMV-derived peptide epitopes (1 μg/ml for H-2D^b or H-2K^b restricted epitopes and 10 μg/ml for I-A^b restricted epitopes). The intracellular accumulation of cytokines was facilitated by the addition of either brefeldin A for CD8 T cell responses (Golgiplug; BD PharMingen) or monensin for CD4 T cell responses (Golgistop; BD PharMingen). Cells were cultured for 5–6 h at 37°C, and surface and intracellular stainings were performed. The Abs used were anti-CD8-PE or -PerCP (clone 53-6.7), anti-CD4-PE (clone RM4-5), anti-IL-2-FITC or -PE (clone JES6-5H4), anti-IL-4-allophycocyanin (clone 11B11), anti-IL-10-allophycocyanin (clone JES5-16E3), anti-IFN-γ-FITC or -allophycocyanin (clone XMG1.2), and anti-TNF-α-allophycocyanin (clone MP6-XT22). Conjugated Abs were purchased from either BD PharMingen or eBioscience.

TCR Vβ gene usage was evaluated by flow cytometry using a panel of anti-TCR Vβ Abs. Splenocytes (5 × 10⁵–10⁶) were treated for 15 min with 1 μg/sample of anti-CD16/CD32 Abs (Fc Block; BD PharMingen). Subsequently, cocktails containing anti-CD8-PE Abs (clone 53-6.7), D^b (gp276–286) tetramers, and a panel of FITC-conjugated anti-TCR Vβ Abs were added. Anti-TCR Vβ2, -3, -4, -5.1/5.2, -6, -7, -8.1/8.2, -8.3, -9, -10b, -11, -12, -13, and -14 (clones B20.6, KJ25, KT4, MR9-4, RR4-7, TR310, MR5-2, 1B3.3, MR10-2, B21.5, RR3-15, MR11-1, MR12-3, and 14-2, respectively) were all obtained from BD PharMingen. After staining, cells were washed three times and fixed in PBS containing 2% (w/v) paraformaldehyde before flow cytometric analysis.

Granzyme A activity was determined by BLT esterase release assays, performed as previously described with modifications (47, 48). Splenocytes prepared from B6 and PKO mice 8 days after LCMV infection were seeded into 96-well plates. Cultures were either left untreated or stimulated with non-fluorochrome-conjugated H-2D^b tetramers complexed with either LCMV-gp33–41 or LCMV-NP_396–404 epitopes (10 μg/ml) or by the addition of 50 ng/ml PMA (Sigma-Aldrich) and 500 ng/ml ionomycin (Calbiochem, San Diego, CA). To determine total BLT esterase content, unstimulated cells were lysed with Triton X-100 (0.09% final concentration) and processed as described below. After 5 h, plates were centrifuged at 74 × g for 5 min. Then 50 μl of supernatant was collected from each well and mixed with 150 μl of PBS containing 0.26 mM BLT, 0.29 mM 5,5′-dithio-bis-(2-nitrobenzoic acid), and 0.013% Triton X-100. Colorimetric reactions were allowed to proceed for 1 h at 37°C, and BLT esterase activity was measured by reading the A₄₁₂ using a microplate reader (Molecular Devices, Sunnyvale, CA). The percentage of BLT esterase release was calculated using the following formula: % BLT esterase release = 100 × [(experimental release − spontaneous release)/(total release − spontaneous release)].

Mixed lymphocyte reactions were set up essentially as previously described (49). Briefly, splenocytes from DBA/2 mice, prepared as described above, were irradiated (2000 rad) and mixed at a 1:1 ratio with responder splenocytes from either B6 or PKO mice. Cells were cultured in 24-well cluster plates (8 × 10⁶ total cells/well) for 5 days at 37°C in 6% CO₂. After this time viable cells were collected by centrifugation over a layer of Histopaque-1083 (Sigma-Aldrich). BLT esterase release was then determined as described above following a 5-h stimulation with either PMA (50 ng/ml) and ionomycin (500 ng/ml) or plate-bound hamster anti-mouse CD3 Abs (clone 145-2C11).

Previous studies have demonstrated that because of their inability to express perforin, a crucial mediator of cell-mediated cytotoxicity, PKO mice cannot resolve infections with isolates of LCMV that cause only an acute infection in normal mice (21, 23). As expected, no viremia was detectable in B6 mice >8 days after infection with the Armstrong isolate of LCMV (data not shown). By contrast, PKO mice became persistently infected, and serum virus titers of >10⁴ PFU/ml were detectable by ∼1 mo after infection (Fig. 1). LCMV infection of PKO mice was, however, not asymptomatic. During the second week following inoculation PKO mice developed hunched posture, ruffled fur, and lethargic behavior, and approximately one-third of these mice died between 2 and 4 wk after injection.

To determine how persistent exposure to high viral loads alters CD8 T cell responses, we compared the induction, activation, and function of virus-specific T cells in either B6 or PKO mice undergoing acute or chronic LCMV infection, respectively. Fig. 2, a and b, show the flow cytometric profiles of virus-specific CD8 T cells identified by staining with MHC class I tetramers complexed with four defined, H-2^b-restricted, LCMV-specific epitopes. Virus-specific CD8 T cell activation was determined by examining cell size (forward light scatter) and the expression of the activation-associated isoform of CD43 (1B11) (50), which has been reported to become up-regulated on T cells as they acquire effector activity, but decreases as the cells shift from an effector to a memory phenotype (51). By 9 days after infection, blasting virus-specific CD8 T cells expressing elevated levels of CD43 (1B11) were readily detectable in B6 mice (Fig. 2 a). By 14 days after infection, as replicating virus was cleared and the response contracted, the virus-specific CD8 T cells discontinued to blast, and CD43 (1B11) expression was down-regulated. Subsequently, as previously reported, a pool of resting memory CD8 T cells became established that was maintained over time (13).

A different pattern of virus-specific CD8 T cell responses was detected in chronically infected PKO mice (Fig. 2 b). There was marked activation of CD8 T cells specific for all epitopes examined by 9 days after infection of PKO mice. This CD8 T cell response, however, failed to resolve the infection, and strikingly, on day 14, virus-specific CD8 T cells in PKO mice retained a blasting phenotype and continued to express high levels of CD43 (1B11). Increased expression of CD43 on virus-specific PKO CD8 T cells was still apparent as late as 110 days after infection, although by this time point the lymphocytes were somewhat smaller in size. These data show that PKO mice respond to LCMV infection by mounting multiple epitope-specific responses. However, unlike virus-specific CD8 T cells in acutely infected hosts, persistent exposure to viral Ag in PKO mice was associated with pronounced and prolonged CD8 T cell activation. Thus, virus-specific CD8 T cells are not necessarily deleted in chronically infected hosts, but, as discussed below, there are changes in the hierarchy of viral epitope recognition and in the effector functions of these cells.

Since virus-specific CD8 T cells remained detectable over time in LCMV-infected PKO mice, we examined whether chronic exposure to high viral loads altered the capacity of these cells to produce effector cytokines. As expected, LCMV-specific IFN-γ- and TNF-α-producing CD8 T cells were readily detectable in acutely infected B6 mice (Fig. 2,e) (52, 53). By 9 days after infection significant responses to the five MHC class I-restricted epitopes tested were apparent. Many of these CD8 T cells produced both IFN-γ and TNF-α; however, a fraction produced IFN-γ only. By 14 days after infection, as replicating virus was cleared, the virus-specific CD8 T cells retained the potential to produce IFN-γ and TNF-α and remained fully functional even after returning to a resting state. This is corroborated by comparing the numbers of splenic virus-specific CD8 T cells identified by tetramer staining with the numbers of epitope-specific, IFN-γ-producing CD8 T cells (Fig. 3 a).

In PKO mice, virus-specific CD8 T cells were identifiable by both tetramer staining and intracellular cytokine analysis by 9 days postinfection (Figs. 2,f and 3b). However, not all the virus-specific CD8 T cells produced IFN-γ even though they were highly activated, as assessed by forward light scatter measurements and CD43 (1B11) expression (Fig. 2, b and c). Quite strikingly, the majority of these virus-specific CD8 T cells did not produce detectable levels of TNF-α. By 14 days after infection IFN-γ production was further reduced, and <10% of the CD8 T cells that retained the capacity to produce IFN-γ also produced TNF-α. The production of IFN-γ and TNF-α by virus-specific CD8 T cells continued to decline over time in chronically infected PKO mice, and by 60 days after infection only minimal functional activity was ascribable to these T cells. Overall these data suggest that chronic exposure to high levels of viral Ag results in a progressive loss of TNF-α production, followed by loss of IFN-γ production by virus-specific CD8 T cells.

The kinetics of virus-specific CD8 T cell responses in acutely infected B6 mice and chronically infected PKO mice are compared in Fig. 3. In B6 mice, expansion, contraction, and memory phases of the anti-viral response are apparent, and there is an excellent correlation between the number of CD8 T cells visualized by tetramer staining and those that have the capacity to produce IFN-γ (Fig. 3,a). Enumeration of virus-specific CD8 T cell responses in LCMV-infected PKO mice shows several differences between the responses in acutely and chronically infected hosts (Fig. 3 b). Although virus-specific CD8 T cells are initially detectable in PKO mice, not all these cells are capable of producing IFN-γ. Between days 9 and 40 after infection the absolute number of virus-specific CD8 T cells declines in both acutely infected B6 mice and chronically infected PKO mice. In PKO mice, however, there is a disproportionate reduction in the number of IFN-γ-producing cells as these virus-specific CD8 T cells lose their capacity to elicit anti-viral effector activity.

Although responses to multiple viral epitopes are detectable in both acutely infected B6 mice and chronically infected PKO mice, the magnitude of the response to each individual epitope is not equal. Instead, a hierarchy of immunodominant and subdominant responses becomes established (Figs. 2 and 3). In acutely infected B6 mice the gp33–41 and NP_396–404 responses are codominant, followed by the somewhat weaker gp276–286 response and the subdominant NP_205–212 response. During the course of chronic LCMV infection this epitope hierarchy changes. Although the initial burst sizes of the responses are similar in both B6 and PKO mice, the failure to control the infection is associated with differential contractions of certain specificities of CD8 T cells. H-2D^b-restricted NP_396–404-specific CD8 T cells are more prone to deletion in chronically infected hosts (Fig. 3). By contrast, in PKO mice, despite a severe functional impairment, the H-2D^b-restricted gp276–286-specific response contracts less and becomes immunodominant (Fig. 3 b). Taken together these data show that chronicity results in sustained CD8 T cell activation and altered immunodominance; however, there is a progressive breakdown of effector functions as first TNF-α and then IFN-γ production is lost.

Given the marked loss of IFN-γ and TNF-α production by virus-specific CD8 T cells in chronically infected hosts, we checked for impairment of other CD8 T cell effector functions. Triple intracellular cytokine staining was used to determine IL-2, IFN-γ, and TNF-α production by virus-specific CD8 T cells in both acutely infected B6 mice and chronically infected PKO mice (Fig. 4). Three major populations of cytokine-producing cells were identified. In B6 mice, at 8 days postinfection, CD8 T cells produced either only IFN-γ, IFN-γ and TNF-α, or all three cytokines, IFN-γ, TNF-α, and IL-2. By 85 days after infection the proportion of cells that produced only IFN-γ declined, and the proportion of CD8 T cells that produced all three cytokines, including IL-2, increased (Fig. 4,a). In PKO mice, few, if any, CD8 T cells produced IL-2 at 8 days postinfection, but CD8 T cells that produced only IFN-γ or IFN-γ and TNF-α were detectable. Analysis of virus-specific CD8 T cells in PKO mice on day 85 postinfection confirmed that these cells become unable to produce any of the cytokines tested for (Fig. 4,b). Taken together with the results in Figs. 2 and 3, these data show that the failure to rapidly resolve the infection results in lost production of first IL-2, followed by TNF-α, and finally IFN-γ.

We also measured the release of granzyme A by virus-specific CD8 T cells prepared from either acutely infected B6 mice or chronically infected PKO mice at 8 days postinfection. The percentages of BLT esterase (granzyme A) released following stimulation of gp33–41-specific effectors were 27.8 ± 9.5 and 24.1 ± 11.4% for cells from B6 and PKO mice, respectively. For NP_396–404-specific effectors, the percentage of BLT esterase released was 26.9 ± 10.0% for cells from B6 mice and 24.6 ± 8.6% for cells from PKO mice. Although the proportions of BLT esterase released by virus-specific CD8 T cells from B6 and PKO mice were similar, these values are deceptive, as they do not take into account differences in the total level of BLT esterase in these effector cell populations. The actual total BLT esterase content was 40–60% lower in PKO effector cells compared with B6 effector cells (data not shown).

It has previously been reported that alloreactive T cells from PKO mice contain and release similar levels of granzyme A as alloreactive effectors from B6 mice (49). We confirmed these observations using H-2^b anti-H-2^d effector cells generated in 5-day mixed lymphocyte reactions. Alloreactive CD8 T cells from both B6 and PKO mice released similar levels of BLT esterase. Moreover, total BLT esterase levels in alloreactive PKO T cells were at least as high as those in alloreactive effector cells from B6 mice (data not shown). Overall these data suggest that the lower total levels of BLT esterase in CD8 T cells from chronically infected PKO mice are consistent with a more global functional impairment of these cells resulting from prolonged exposure to high viral loads.

As illustrated in Fig. 3, because of differential contraction, the gp276–286-specific CD8 T cell response becomes immunodominant in chronically infected PKO mice. To determine whether this altered epitope recognition profile resulted from the selection of particular clones or subsets of gp276–286-specific CD8 T cells, we compared TCR Vβ usage by these cells at different stages of the immune response.

Naive CD8 T cells from control B6 and PKO mice show a similar pattern of TCR Vβ gene usage, as determined using a panel of Abs specific for 14 different TCR Vβ segments (Fig. 5,a). By 8 days after infection, which corresponds to the effector phase of the immune response in normal mice, gp276–286-specific CD8 T cells are detectable, and the predominant TCR Vβ genes used by this population include Vβ10 and Vβ8.1/8.2 (Fig. 5,b). Similarly, an oligoclonal response was detectable during the contraction phase of the response, at 20 days after infection (Fig. 5,c). The TCR Vβ gene usage by gp276–286-specific CD8 T cells was also examined on days 110 and 121 postinfection. By this time a pool of fully functional memory CD8 T cells is established in B6 hosts, whereas in chronically infected PKO mice, gp276–286-specific cells are present, but have severely impaired effector capabilities. At this late time point, gp276–286-specific CD8 T cells in both B6 and PKO mice also predominantly used TCR Vβ10 and Vβ8.1/8.2 (Fig. 5,d). Thus, no striking differences in TCR Vβ gene usage are apparent between gp276–286-specific CD8 T cells with effector, memory, or unresponsive phenotypes (Fig. 5, b–d).

Fig. 6 shows the intracellular cytokine staining profiles and enumeration of virus-specific I-A^b-restricted CD4 T cell responses at various times after acute or chronic LCMV infection of B6 and PKO mice, respectively. Responses to the gp61–80 epitope and subdominant NP_309–328 epitope were analyzed. As expected, acute infection of B6 mice elicited a Th1-type response (2, 6, 19, 54). By day 9 postinfection virus-specific CD4 T cells were detectable that produced both IFN-γ and IL-2 or only IFN-γ. At this time point ∼50% of the LCMV-specific CD4 T cells could produce IL-2. Both IL-2- and IFN-γ-producing CD4 T cell responses remained detectable throughout the effector (day 9), contraction (day 14), and memory (day 110) phases of the response (Fig. 6, a and c).

In PKO mice, the failure to control LCMV infection was associated with decreased numbers of IL-2-producing CD4 T cells. By 9 days after infection, although marked IFN-γ production was detectable, only ∼18% of the total virus-specific CD4 T cell population were capable of producing IL-2. This loss of IL-2 production was more dramatic 14 days postinfection and declined even further as the infection progressed. By day 110, IL-2 production by gp61–80-specific CD4 T cells was slightly above background levels, whereas IL-2 production by NP_309–328-specific CD4 T cells was below the limit of detection (Fig. 6, b and d).

To determine whether virus persistence resulted in immune deviation and caused the responding CD4 T cells to acquire a Th2 phenotype, we assessed the ability of the virus-specific CD4 T cells to produce IL-4 and IL-10. No IL-4-producing gp61–80- or NP_309–328-specific CD4 T cells were detectable at any time following infection of either B6 or PKO mice (data not shown). A small fraction of gp61–80-specific IL-10-producing CD4 cells were visualized in both B6 mice (0.16 ± 0.06%) and PKO mice (0.59 ± 0.12%) at 9 days postinfection. However, this activity was not maintained over time, and by day 20 only background levels of IL-10 production were detectable in either host (data not shown). As a positive control we validated our intracellular cytokine staining using a bulk DO11.10 Th2 cell line and demonstrated that, as expected, these cells produced IL-4 and IL-10 (data not shown).

Taken together, these data show that acute LCMV infection induces a marked, virus-specific, Th1 CD4 T cell response. The failure of PKO mice to clear LCMV infection does not result in a shift of this response to a Th2 phenotype; however, prolonged exposure to high viral loads ablates IL-2 production by the responding CD4 T cells.

The goal of the current study was to determine how persistent exposure to high viral loads impacts the maintenance and function of virus-specific T cell responses. To address this we took advantage of PKO mice that contain both CD4 and CD8 T cell subsets, but become persistently infected with LCMV (21, 23). In this way we were able to initiate an acute or chronic infection using the same viral isolate and perform a comparative analysis of virus-specific CD8 and CD4 T cell responses. Our findings reveal several key differences between T cell responses that successfully resolve acute infections and those that are induced, but fail to control chronic infections. Notably, our data show that the failure to rapidly resolve LCMV infection in PKO mice results in marked defects in both virus-specific CD4 and CD8 T cell responses. In the case of CD4 T cell responses, virus persistence was associated with the loss of IL-2 production. Multiple virus-specific CD8 T cell responses were initially induced in PKO mice; however, unlike their counterparts in acutely infected hosts, these cells were maintained in an activated state, but progressively lost effector functions. The immunodominance of the virus-specific CD8 T cell population was also altered and could be attributed to differential contraction of certain specificities of T cells following the peak of the response. Taken together these data suggest that sustained exposure to high viral loads has a compounding effect, resulting in the progressive functional inactivation of virus-specific T cell responses, which, in turn, further promotes virus persistence.

Numerous viral infections, including LCMV infection, induce a marked virus-specific CD8 T cell response (3, 4, 12, 13, 14, 15, 16). The current study shows that pronounced expansion and activation of multiple virus-specific CD8 T cell responses occur during the first week of LCMV-Armstrong infection of both B6 and PKO mice. This mobilization of virus-specific CD8 T cells results in up-regulation of the activation-associated isoform of CD43 on the responding cells. More marked divergence between the responses in acutely and chronically infected hosts becomes noticeable during the second week following infection. Arguably, the most fundamental difference between B6 and PKO mice at this stage is the presence of replicating virus in PKO mice. As LCMV is cleared from B6 mice, the CD8 T cell response contracts, and the infectious event concludes with the emergence of resting virus-specific memory cells. The numbers of virus-specific CD8 T cells also decline in PKO mice; nevertheless, the virus-specific CD8 T cells that remain exhibit striking differences in their activation profiles, as they continue to express high levels of CD43 (1B11) and are larger in size (Fig. 2, a and b). These observations are consistent with the idea that a programmed contraction of CD8 T cell responses can occur during the initial phase of chronic infections, but that continued exposure to viral Ag sustains the activated state of the surviving cells (55, 56). In addition to its primary role in cell-mediated cytotoxicity, perforin has been reported to play a role in regulating CD8 T cell homeostasis (reviewed in Ref. 57). Therefore, it is possible that the alterations in the activation state and magnitude of LCMV-specific CD8 T cell responses that occur in PKO mice may be due to a compounding effect of high viral loads superimposed on a host with altered T cell homeostasis. Nevertheless, changes in viral epitope hierarchies and loss of T cell effector functions have also been shown to occur in persistently infected hosts that express perforin (29).

We have shown that virus persistence is associated with changes in the viral epitope hierarchy, and this appears to be a consequence of differential contraction of certain epitope-specific CD8 T cell responses. Our data also suggest that the preferred maintenance of gp276–286-specific CD8 T cells in chronically infected hosts does not result from selection or preferential outgrowth of particular clones of T cells. Both TCR Vβ10 and Vβ8.1/8.2 were predominantly used by gp276–286-specific CD8 T cells, regardless of whether they were effector, memory, or functionally unresponsive CD8 T cells (Fig. 5). Although the exact determinants of immunodominance remain unclear, it is likely that various parameters contribute to the size of each individual epitope-specific response. These factors may include the abundance of Ag, the type of cells that are presenting the Ag, the temporal order of Ag presentation, the stability of the MHC-peptide complex, and competition between dominant and subdominant clones as they encounter APC (58, 59, 60). It has been previously documented that changes in the hierarchy of LCMV-specific CD8 T cell responses can occur as a result of the cell type presenting the viral Ag (61). In the context of the current study, this implies that changes in viral tropism, which can occur in chronically infected hosts, may influence the magnitude of epitope-specific T cell responses (62). The shifts in immunodominance due to altered contraction of responses in chronically infected hosts may have implications for interpreting which specificities of CD8 T cells are most important for initially controlling the acute phase of chronic viral infections. Epitopes ascribed as immunodominant at later time points in chronically infected hosts may not necessarily represent the specificities of CD8 T cells that play a principal role in initially dampening the infection (7, 30, 63).

A central finding of the current study is that although virus-specific CD8 T cells can be induced in persistently infected hosts, their anti-viral effector activities become silenced. In acutely infected hosts, virus-specific CD8 T cells have multiple potential effector activities, and there is an excellent correlation between the numbers of virus-specific CD8 T cells identifiable by MHC tetramer staining and those that are revealed by intracellular cytokine analysis for IFN-γ. At the peak of the response to acute LCMV infection, a significant fraction of the virus-specific CD8 T cells also produce TNF-α, and this proportion increases as replicating virus is cleared and the response progresses through the contraction phase. The diminished potential to produce TNF-α during the effector phase of the response may reflect down-regulated TNF-α expression as a result of recent antigenic activation (53, 64). This Ag-driven cessation of TNF-α is further suggested in PKO mice, in which prolonged exposure to high viral loads manifests first as a loss of TNF-α production by virus-specific CD8 T cells and subsequently as an extinguishment of IFN-γ production. The constant exposure of Ag-specific CD8 T cells to high viral loads not only impacts their ability to produce IFN-γ and TNF-α, but also results in dysregulation of IL-2 production by these cells. IL-2 has been reported to up-regulate granzyme A synthesis, and thus lower IL-2 levels may be responsible for the reduced granzyme A content of effector cells from PKO mice (65, 66, 67). Taken together, these data suggest that a progressive and ultimately global misprogramming of the effector activities of CD8 T cells occurs in chronically infected hosts. This may directly result from repeated TCR triggering due to contact with presented Ag or may be because of indirect consequences, such as cytokine deprivation.

We have previously reported that infection of CD4-deficient mice with certain isolates of LCMV induces a life-long infection due to the loss of functional CD8 T cell responses (29). Other studies of persistent LCMV infections of adult mice have also documented defects in virus-specific CD8 T cell responses (38, 68, 69, 70, 71). In this report we have shown that high viral loads are not only associated with marked defects in virus-specific CD8 T cell responses, but also result in impaired virus-specific CD4 T cell functions. Our data show that both acute and chronic LCMV infections elicit Th1-type CD4 T cell responses. During persistent LCMV infection of PKO mice, virus-specific CD4 T cells are initially induced, but these cells are not fully functional by comparison with their counterparts in acutely infected hosts, as these cells fail to produce abundant levels of IL-2. Notably, loss of LCMV-specific CD4 T cell responses has also been reported in other situations where the infection is only slowly controlled or is never cleared (54, 71, 72). A possible scenario is that high viral loads impair virus-specific CD4 T cell responses, and consequently both chronic antigenic stimulation and ineffective CD4 T cell help ablate virus-specific CD8 T cell responses. Virus persistence in the absence of CD4 T cells can ultimately result in the complete loss of virus-specific CD8 T cells, whereas fluxes in viral loads in chronically infected B6 mice may result in the partial restoration of at least certain CD8 T cell effector functions (M. J. Fuller and A. J. Zajac, unpublished observations). This further supports the idea that both high Ag load and poor CD4 T cell responses contribute to the overall impairment of virus-specific CD8 T cells in chronically infected hosts.

Impairment of CD4 and CD8 T cell responses not only occurs during LCMV infection of mice. Lee et al. (73) reported that ∼2% of peripheral CD8 T cells were tumor-associated Ag-specific in a patient with metastatic melanoma, but were unable to ascribe any effector activity to this population. Varying defects in effector functions have also been reported for HIV-specific CD8 T cells, and in the case of hepatitis C virus infection, a transient “stunned” loss of effector functions and also more long term dysfunction of CD8 T cell responses have been reported (8, 9, 28, 36, 74, 75). In addition, robust CD4 T cell responses are associated with better control of chronic viral infections. In each of these situations impaired functional activity is associated with difficulties in clearing the Ag. Thus, strategies to boost CD4 T cell activity and decrease Ag loads may help to sustain the functions of the more effective specificities of CD8 T cells. CD4 T cell responses could be potentially enhanced by prophylactic or therapeutic vaccination strategies, and a variety of approaches could be employed to lower viral loads. For experimental studies in LCMV-infected mice, viral loads could potentially be reduced by passive transfer of neutralizing Abs, by adoptive immunotherapy using immune T cells, or by chemotherapeutic approaches. Using such methods may help to further dissect the roles of high viral loads and ineffective CD4 T cell responses in functionally inactivating CD8 T cells. Although the exact biological consequences of aberrant CD4 and CD8 T cell responses remain ill defined, it is likely that pathogenic infections as well as tumor outgrowth will be more successfully controlled by multi-epitope-specific and multifunctional T cell responses.

We thank Roger Harris for excellent technical assistance, and Aaruni Khanolkar, Daniel Quinn, and Laurie Harrington for helpful discussions. We also thank Rafi Ahmed for supplying viral seed stocks, and John Altman and the University of Alabama Center for AIDS Research Molecular Biology Core for advice and assistance with MHC-tetramer production.