Using a mouse model system, we demonstrate that anergic CD8+ T cells can persist and retain some functional capabilities in vivo, even after the induction of tolerance. In TCR Vβ5 transgenic mice, mature CD8+Vβ5+ T cells transit through a CD8lowVβ5low deletional intermediate during tolerance induction. CD8low cells are characterized by an activated phenotype, are functionally compromised in vitro, and are slated for deletion in vivo. We now demonstrate that CD8low cells derive from a proliferative compartment, but do not divide in vivo. CD8low cells persist in vivo with a t1/2 of 3–5 days, in contrast to their in vitro t1/2 of 0.5–1 day. During this unexpectedly long in vivo life span, CD8low cells are capable of producing IFN-γ in vivo despite their inability to proliferate or to kill target cells in vitro. CD8low cells also accumulate at sites of inflammation, where they produce IFN-γ. Therefore, rather than withdrawing from the pool of functional CD8+ T cells, anergic CD8low cells retain a potential regulatory role despite losing their capacity to proliferate. The ability of anergic cells to persist and function in vivo adds another level of complexity to the process of tolerance induction in the lymphoid periphery.

One of the principal challenges faced by the immune system is to maintain tolerance to self Ags. Among T cells, self tolerance is maintained through both thymic and peripheral events. In the thymus, self-reactive T cells are deleted during maturation through the process of negative selection (reviewed in 1). However, some functional self-reactive T cells escape negative selection and reach the lymphoid periphery, particularly those that recognize tissue-restricted self Ags or Ags expressed in an age-dependent manner. Many mechanisms serve to maintain peripheral tolerance, including clonal ignorance (reviewed in 2), deletion (3, 4), diversion (5, 6), exhaustion (7, 8), and anergy with or without an associated down-regulation of TCR and/or accessory molecule expression on the autoaggressive cells (9, 10, 11, 12, 13, 14, 15, 16, 17, 18).

Whether and how deeply an autoreactive T cell will enter the anergic state or become deleted seems to be determined by the interplay of signals sensing the strength of the TCR/ligand interaction, the frequency of such interactions, the exposure to cytokines, and the delivery of costimulatory signals (16, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28). Because peripheral deletion of autoreactive T cells is often incomplete (15, 17, 18, 29), the remaining cells must be rendered anergic to protect the host, indicating that multiple tolerance pathways can converge to form a continuum in vivo. A variety of in vitro and in vivo systems have allowed a characterization of the functional defects that prevent anergic cells from responding to Ag. These studies have indicated that anergic CD4+ T cells demonstrate defects in only a subset of intracellular signaling pathways, primarily involving IL-2 transcription (23, 30, 31, 32, 33, 34, 35, 36). Therefore, anergic CD4+ T cells can retain some ability to respond to stimuli, for example by the secretion of IL-4 or IFN-γ (37, 38, 39, 40, 41). Anergy among CD8+ T cells is less well-characterized, although anergic CD8+ T cells have demonstrated impaired calcium mobilization and tyrosine phosphorylation after CD3 ligation (15, 42).

As a model system, our studies of CD8+ T cell anergy and deletion use C57BL/6 (B6)4 mice (H-2b, I-E) transgenic (Tg) for a gene encoding a rearranged TCR Vβ5.2 chain. These mice provide a unique setting, allowing the isolation and characterization of tolerant cells during a polyclonal response to endogenous self Ags in vivo (14, 15, 43, 44). In MHC class II I-E B6 mice, mature peripheral CD4+Vβ5+ T cells are activated and rendered anergic before their deletion by the viral superantigens vSAG8 and vSAG9 (14, 44). In contrast, mature peripheral CD8+Vβ5+ T cells transit through a CD8lowVβ5low compartment of deletional intermediates during their tolerance induction (15). Although the tolerogen that drives the formation of CD8lowVβ5low cells is unknown, it is not a mouse mammary tumor virus-encoded superantigen, nor is its expression dependent upon MHC class II molecules (44). Although CD8lowVβ5low cells are small, they bear an activated/memory phenotype (CD44high, CD45RBlow, CD62 ligandlow) and are Thy-1low, B220, and NK1.1 (15). Despite this surface phenotype, CD8low cells proliferate very poorly upon TCR cross-linking in the presence of IL-2 and are unable to kill target cells in vitro. Finally, their tendency to undergo apoptosis rapidly in vitro, their in vivo cortisone sensitivity, and their reduced expression of Bcl-2 indicate that CD8low cells are poised to die (15). The discrete phenotype of CD8low cells allows us to characterize these anergic cells during the induction of tolerance to endogenously expressed self Ags. Here, we measure the in vivo t1/2 of these distinct deletional intermediates and reveal their capacity to produce IFN-γ and accumulate at sites of inflammation. Our results suggest that CD8low cells retain some functions, pointing to the potential of anergic T cells to perform regulatory roles in vivo.

B6 TCR β-chain Tg mice (H-2Kb, I-E) were constructed by injection of a rearranged genomic Vβ5.2+ β-chain gene from a CD8+ CTL clone specific for a chicken OVA peptide bound by H-2Kb, and have been described previously (14, 15, 43). Tg mice were maintained as heterozygotes in a specific pathogen-free barrier facility at the University of Washington by breeding to B6 females purchased from The Jackson Laboratory (Bar Harbor, ME); nonTg mice were offspring from these same matings. For adoptive transfer experiments, B6 Tg mice were used. All other experiments used mice from F1 matings between B6 Tg and BXD-15 mice (an H-2b recombinant inbred line purchased from The Jackson Laboratory) as described previously (15). All animals were used at 12–25 wk of age. B6.PL-Thy-1a/Cy mice (referred to as B6.Thy1.1 mice) were also bred in the specific pathogen-free barrier facility at the University of Washington.

Hydrocortisone 21-acetate (Sigma, St. Louis, MO) was dissolved in a small volume of 100% ethanol and subsequently brought to 15 mg/ml in HBSS for injection i.p. of 3 mg. PE-conjugated anti-CD8α (53–6.7) and anti-CD4 (RM4–5) mAbs; FITC-conjugated anti-CD8α (53–6.7), anti-murine IFN-γ (XMG1.2, rat IgG1), and rat IgG1 isotype control (R3-34) mAbs; and biotin-conjugated anti-Thy-1.2 (30-H12) and XMG1.2 mAbs were purchased from PharMingen (San Diego, CA). Tricolor-conjugated streptavidin and goat anti-hamster Ig were purchased from Caltag (San Francisco, CA), and FITC-anti-bromodeoxyuridine (BrdU, B44) mAb was purchased from Becton Dickinson (Mountain View, CA). MR9.4.4 (anti-Vβ5.1 + 5.2) mAb was purified from ascites over a protein A column and conjugated with FITC according to established protocols (45). Tissue culture supernatant of anti-CD3ε (2C11) was used for stimulation. For immunohistochemistry, tissue culture supernatant of anti-CD8α (2.43) and column-purified digoxigenin-modified polyclonal goat anti-rat IgG were made. HRP-conjugated anti-digoxigenin mAbs and avidin were purchased from Boehringer Mannheim (Indianapolis, IN). For depletion, anti-CD4 (RL172.4R6) from unpurified ascites fluid was used and guinea pig complement was purchased from Bethesda Research Laboratories (Bethesda, MD). We obtained 5-carboxyfluorescein diacetate-succinimidyl ester (CFSE) and 7-amino-actinomycin D (7-AAD) from Molecular Probes (Eugene, OR). Propidium iodide (PI) and 3,3-diaminobenzidine (DAB) were purchased from Sigma.

Thymuses were removed by suction from young adult mice anesthetized with tribromoethanol. Where indicated, mice were given drinking water containing BrdU (Sigma) at 0.8 mg/ml, which was made fresh and changed daily (15).

Nylon wool nonadherent splenocytes from thymectomized B6 Vβ5 Tg mice were used as donor cells for adoptive transfer. CD8+ T cells were enriched by pretreatment with anti-CD4 Ab plus guinea pig complement (43). Cells were labeled with 5 μM CFSE at 37°C for 10 min, a procedure that does not induce apoptosis of CD8low cells (data not shown). CFSE labeling does not alter the trafficking or function of lymphocytes in adoptive hosts (46). The labeled cells were analyzed by flow cytometry, and a total of 5–10 × 106 CD8+ T cells were adoptively transferred by tail vein injection into unirradiated congenic B6.Thy1.1 hosts.

Inflammation was induced in Vβ5 Tg mice by a variety of methods. For oxazolone challenge on the ear, mice were sensitized with 2.5 mg of oxazolone dissolved in 4:1 (v/v) acetone:olive oil in the groin or axial fold. After 5 days, the inner pinna of the right ear was challenged with 0.25 mg of oxazolone dissolved in 4:1 (v/v) acetone:olive oil; the mice were sacrificed 2 days later, the draining lymph nodes (LNs), spleen, and uninvolved LNs were harvested, and the ears were removed for sectioning. Gelfoam sponges (The Upjohn Company, Kalamazoo, MI) were implanted s.c. into the flanks of tribromoethanol anesthetized mice. At 3 days postimplantation, the sponges and spleens were harvested for analysis by flow cytometry. The sponges were minced finely and digested with 400 U/ml collagenase D (Boehringer Mannheim) at 37°C for 20 min in RPMI 1640 (BioWhittaker, Walkersville, MD) with 10% FCS. Debris was removed by filtration through nytex, and isolated cells were stained for flow cytometric analysis. Oxazolone challenge was performed by shaving the lower back of Vβ5 Tg mice and applying 750 μg of oxazolone dissolved in 3:1 (v/v) acetone:olive oil. After 5 days, the mice were challenged on the shaved lower back with 30 μg of oxazolone dissolved in 3:1 (v/v) acetone:olive oil; the mice were sacrificed after 2 days, and the draining LNs and spleen were harvested for analysis. Peritoneal inflammation was induced by an i.p. injection of 2.5 ml of 3% thioglycollate. After 4 days, peritoneal exudate cells were harvested and analyzed by flow cytometry. To induce footpad inflammation, mice were injected in one footpad with 100 μl of CFA (Sigma) emulsified 1:1 in PBS. The spleen and popliteal LNs were harvested at 2 days postinjection for analysis of lymphocytes by flow cytometry.

Unless otherwise noted, PBLs were isolated by water lysis of whole heparinized blood; LN cells were derived from pooled axillary, brachial, inguinal, cervical, and mesenteric nodes. Cells were stained as described previously (14) and analyzed on a FACScan using CellQuest software (Becton Dickinson). Logarithmic detectors were used for all three fluorescence channels, except for the linear FL-2 or FL-3 scale employed during cell cycle analyses with PI and 7-AAD, respectively. Unless otherwise noted, dead cells were excluded on the basis of forward and side scatter profiles, and a minimum of 1 × 105 events were collected. Detection of incorporated BrdU was performed by surface staining cells with PE-anti-CD8α and subsequently fixing, permeabilizing, and counterstaining with FITC-anti-BrdU mAb as described previously (15, 47). Ex vivo cell cycle analyses were performed by staining splenocytes from 12-wk-old Vβ5 Tg mice with FITC-anti-CD8α, fixing overnight in ethanol, and counterstaining with 50 μg/ml PI in the presence of 100 U/ml RNase A. Cell cycle analyses of stimulated cells were performed by culturing 3 × 106 splenocytes in 24-well plates precoated with goat anti-hamster Ig plus or minus anti-CD3ε (2C11) in the presence or absence of 50 U/ml rIL-2 (48). Cells were harvested at various times thereafter, surface stained with FITC-anti-CD8α, and subsequently fixed in 70% ethanol, washed, stained with 15 μg/ml 7-AAD in PBS (BioWhittaker), and analyzed without washing. Intracellular staining for IFN-γ was performed by stimulating splenocytes from Vβ5 Tg mice for various lengths of time with 20 ng/ml PMA and 500 ng/ml ionomycin (purchased from Calbiochem-Novabiochem, La Jolla, CA) in the presence of 3 μM of monensin (Sigma) during the latter two-thirds of the stimulation period to retain the expressed proteins intracellularly. Stimulated cells were harvested, washed, and surface stained with PE-anti-CD8α, and subsequently fixed, permeabilized with saponin (0.1% in PBS plus 1% BSA), and stained with FITC-anti-IFN-γ or FITC-R3-34 as an isotype control (49). Splenocytes stimulated for the indicated times with plate-bound anti-CD3ε Abs were also stained for intracellular IFN-γ. CD8low, CD8high, and CD4+ T cells from Vβ5 Tg and nonTg mice were purified by flow cytometric sorting on a FACStarPlus (Becton Dickinson) from nylon wool nonadherent splenocytes. Sorted populations were consistently >96% pure.

Inflamed and control ears were snap frozen after immersion in optimal cutting temperature compound (Sakura-Finetek, Torrance, CA). Frozen sections of 6–8 μm thickness were mounted on aminoalkylsilane-coated slides and allowed to air dry for ≥2 h before fixation in cold acetone. Endogenous peroxidase activity was blocked with a mixture of glucose, glucose oxidase, and sodium azide (Sigma). CD8 surface Ag was detected by three-step enzyme immunohistochemistry using mAb 2.43 or normal rat IgG as a primary reagent, followed by digoxigenin-modified polyclonal goat anti-rat IgG and HRP-conjugated goat anti-digoxigenin mAb. Enzyme activity was detected with a mixture of DAB and H2O2. Sections stained for CD8 alone were dehydrated at this stage through a series of ethanol and toluene washes; coverslips were mounted with Permount (Fisher Scientific, Pittsburgh, PA). Sections also stained for IFN-γ were treated with a solution of cobalt chloride/DAB/H2O2 to stabilize the brown/black color of CD8 localization. Sections were then blocked with avidin and biotin before sequential application of biotin-anti-IFN-γ mAbs and HRP-conjugated avidin. The second-step enzyme activity was detected with True Blue peroxidase substrate (Kirkegaard and Perry Laboratories, Gaithersburg, MD). Slides were then dehydrated through a series of ethanol and toluene washes before mounting coverslips with Permount. Preliminary experiments confirmed the specificity of the staining patterns observed. Color or gray scale images were captured with a Sony DXC970 MD (Meridian Instrument Company, Kent, WA) or a DAGE MTI CCD72 digital camera (Michigan City, IN).

Total RNA was extracted from purified cell populations with guanidinium thiocyanate/phenol (50) and reverse transcribed to cDNA with avian myeloblastosis virus reverse transcriptase (Life Technologies, Rockville, MD) and random hexamer primers (Pharmacia, Piscataway, NJ). To quantitate cDNAs, 3-fold serial dilutions of the cDNA reactions were subjected to PCR using primers specific for the housekeeping gene hypoxanthine phosphoribosyltransferase (HPRT, 51) for 30–35 cycles of 94°C for 1 min, 60°C for 1 min, and 72°C for 1 min on a DNA ThermalCycler 480 (Perkin-Elmer Cetus, Emeryville, CA). Beginning with similar amounts of cDNA, 3-fold serial dilutions of cDNA were then subjected to PCR for IFN-γ (51). PCR products were electrophoresed on a 2% agarose gel, Southern blotted under alkaline conditions to a zeta probe GT membrane (Bio-Rad, Hercules, CA), and detected with 32P end-labeled oligonucleotides specific for an internal sequence of the PCR product (HPRT, 5′-CGAGGAGTCCTGTTGATGTTGCCAGTAAAA; IFN-γ, 5′-ATCTGGAGGAACTGGCAAAAGGATGGTGAC). Bands were quantitated on a PhosphorImager 425 using ImageQuant software (Molecular Dynamics, Sunnyvale, CA). To normalize IFN-γ expression levels to HPRT levels, the integrated volume of the IFN-γ product was divided by the integrated volume of the HPRT product for each dilution.

Curve fits were performed on semilog plots using the exponential curve fit function in Cricket Graph (Cricket Software, Malvern, PA). Fits were converted to natural log scale, and t1/2 were calculated using the λ value from the following curve fit equation: n = Noet, where n is the number of cells at time t, No is the number of cells at time 0, and λ is the decay constant and is equal to 0.693/t1/2.

Our previous studies established that CD8low cells from Vβ5 Tg mice are stable intermediates in extrathymic T cell deletion (15). The extensive labeling of CD8low cells with BrdU demonstrates that CD8low cells either proliferate or derive from a proliferative compartment (Fig. 1,A). Cell cycle analysis indicates that <1% of CD8low cells but 12% of CD8high cells fall within the G2 or M phases of the cell cycle directly ex vivo (Fig. 1,B). To confirm a defect in CD8low cell proliferation, splenocytes from Vβ5 Tg mice were stimulated for 24 h with plate-bound anti-CD3 Abs. Although CD8high cells responded to this stimulation by entering the cell cycle and beginning division, no significant increase in the percentage of dividing cells was observed within the CD8low population, even in the presence of exogenous IL-2 (Fig. 1,C, left). Instead of driving cell division, anti-CD3 stimulation drives CD8low cells into apoptosis (Fig. 1,C, right). Furthermore, cultures of CD8low cells pulsed with tritiated thymidine between days 2 and 3 of anti-CD3 stimulation revealed a significant proliferative defect, even in the presence of exogenous IL-2 (15). The derivation of CD8low cells from a proliferating compartment (Fig. 1,A), the absence of cell division among CD8low cells directly ex vivo (Fig. 1,B), and the very poor proliferation of CD8low cells upon TCR ligation in vitro (Fig. 1 C and 15) indicate that CD8low cells are anergic (52).

FIGURE 1.

CD8low cells arise from a proliferative compartment but do not divide. A, NonTg and Vβ5 Tg mice were thymectomized at 10 wk of age, allowed to recover for 2 wk, placed on BrdU water for 8 days, and subsequently transferred to normal water for the indicated number of days. Splenocytes stained with PE-anti-CD8α mAb and FITC-anti-BrdU were analyzed by flow cytometry using forward and side scatter gates that include both small cells and blasts. The level of BrdU staining in the CD8low and CD8high subsets was determined by data gating. Data are representative of three independent experiments with one to two mice per timepoint. B, Splenocytes from three 12-wk-old Vβ5 Tg mice were stained with FITC-anti-CD8α and PI before flow cytometric analysis. Linear histograms of PI fluorescence in CD8low and CD8high cells are represented. Population 1 includes cells in the G0 and G1 phases, population 2 includes cells in the S phase, and population 3 includes cells in the G2 and M phases of the cell cycle. The percentages of cells within each marker are indicated in the boxes within each panel. Data are representative of three independent experiments. C, Splenocytes from Vβ5 Tg mice were stimulated for 24 h on plates precoated with goat anti-hamster Ig plus or minus anti-CD3ε in the presence or absence of rIL-2 and subsequently subjected to cell cycle analyses as described in Materials and Methods. “Primary only” refers to cells stimulated on plates coated with goat anti-hamster Ig only. The percentage of dividing cells was determined by adding the percentage of cells in the S phase to those in the G2 or M phases. Apoptotic cells were defined as those with subdiploid DNA content. Error bars represent the SD of the data from four individual mice. A Student’s t test revealed no significant difference in the percentage of dividing CD8low cells before and after stimulation with anti-CD3 Abs either in the absence (p = 0.27) or the presence (p = 0.14) of exogenous IL-2. The percentage of apoptotic CD8low and CD8high cells increased significantly following anti-CD3 stimulation (p < 0.008 in all cases), as did the percentage of dividing CD8high cells (p < 0.0006).

FIGURE 1.

CD8low cells arise from a proliferative compartment but do not divide. A, NonTg and Vβ5 Tg mice were thymectomized at 10 wk of age, allowed to recover for 2 wk, placed on BrdU water for 8 days, and subsequently transferred to normal water for the indicated number of days. Splenocytes stained with PE-anti-CD8α mAb and FITC-anti-BrdU were analyzed by flow cytometry using forward and side scatter gates that include both small cells and blasts. The level of BrdU staining in the CD8low and CD8high subsets was determined by data gating. Data are representative of three independent experiments with one to two mice per timepoint. B, Splenocytes from three 12-wk-old Vβ5 Tg mice were stained with FITC-anti-CD8α and PI before flow cytometric analysis. Linear histograms of PI fluorescence in CD8low and CD8high cells are represented. Population 1 includes cells in the G0 and G1 phases, population 2 includes cells in the S phase, and population 3 includes cells in the G2 and M phases of the cell cycle. The percentages of cells within each marker are indicated in the boxes within each panel. Data are representative of three independent experiments. C, Splenocytes from Vβ5 Tg mice were stimulated for 24 h on plates precoated with goat anti-hamster Ig plus or minus anti-CD3ε in the presence or absence of rIL-2 and subsequently subjected to cell cycle analyses as described in Materials and Methods. “Primary only” refers to cells stimulated on plates coated with goat anti-hamster Ig only. The percentage of dividing cells was determined by adding the percentage of cells in the S phase to those in the G2 or M phases. Apoptotic cells were defined as those with subdiploid DNA content. Error bars represent the SD of the data from four individual mice. A Student’s t test revealed no significant difference in the percentage of dividing CD8low cells before and after stimulation with anti-CD3 Abs either in the absence (p = 0.27) or the presence (p = 0.14) of exogenous IL-2. The percentage of apoptotic CD8low and CD8high cells increased significantly following anti-CD3 stimulation (p < 0.008 in all cases), as did the percentage of dividing CD8high cells (p < 0.0006).

Close modal

Chronic interaction with tolerogen(s) leads to more rapid turnover of CD8high cells in Vβ5 Tg than in nonTg mice; however, 21 days are required to eliminate the BrdU-labeled CD8low cells that are generated during an 8-day BrdU pulse (Fig. 1,A). To evaluate directly the life span of CD8low cells in vivo, we chose two independent experimental approaches. First, we measured the loss of CFSE-labeled CD8low cells upon transfer from Vβ5 Tg donors to congenic B6.Thy-1.1 host mice (Fig. 2,A). To avoid disturbing any aspect of the tolerance process, particularly expression of the tolerogen, the adoptive hosts were not irradiated. Donor cells were clearly detected in the PBLs of host mice by their Thy-1.2 expression and CFSE staining, and the percentage of CD8low cells declined from 26% at 15 h after adoptive transfer (data not shown) to 16% at 4 days after adoptive transfer (Fig. 2,B). Consistent with the absence of proliferating CD8low cells directly ex vivo, CFSE fluorescence levels among CD8low cells did not diminish in a manner consistent with division during this timeframe (data not shown). Analysis of the numbers of donor CD8low and CD8high cells demonstrated the simultaneous proliferation of CD8high cells and the disappearance of CD8low cells in the host mouse (Fig. 2,C). From these data, the t1/2 of transferred CD8low cells was calculated to be 3 days (Fig. 2 C).

FIGURE 2.

The t1/2 of adoptively transferred CD8low cells can be measured in vivo. A, CFSE-labeled CD8+ T cell-enriched splenocytes from thymectomized Vβ5 Tg B6 mice were adoptively transferred into congenic B6.Thy1.1 recipients. PBLs, splenocytes, and LN cells were counted and analyzed by flow cytometry at 15 h after adoptive transfer to determine the number of surviving donor CD8low and CD8high cells. B, After the transfer of 5 × 106 CD8+ T cells, the donor CFSE+Thy-1.2+ population within the PBLs was analyzed for the percentage of Thy-1.2lowCD8low and Thy-1.2highCD8high cells. Donor CFSE+Thy-1.2+ cells comprised 0.58%, 0.55%, and 0.66% of PBLs at days 2, 3, and 4 posttransfer, respectively. C, The numbers of donor CD8low and CD8high cells were determined among recipient PBLs by multiplying the number of cells by the percentage of CFSE+Thy-1.2+ cells and subsequently by the percentage of those cells that were either CD8low or CD8high, as shown in B. The t1/2 of the CD8low cells was calculated as described in Materials and Methods.

FIGURE 2.

The t1/2 of adoptively transferred CD8low cells can be measured in vivo. A, CFSE-labeled CD8+ T cell-enriched splenocytes from thymectomized Vβ5 Tg B6 mice were adoptively transferred into congenic B6.Thy1.1 recipients. PBLs, splenocytes, and LN cells were counted and analyzed by flow cytometry at 15 h after adoptive transfer to determine the number of surviving donor CD8low and CD8high cells. B, After the transfer of 5 × 106 CD8+ T cells, the donor CFSE+Thy-1.2+ population within the PBLs was analyzed for the percentage of Thy-1.2lowCD8low and Thy-1.2highCD8high cells. Donor CFSE+Thy-1.2+ cells comprised 0.58%, 0.55%, and 0.66% of PBLs at days 2, 3, and 4 posttransfer, respectively. C, The numbers of donor CD8low and CD8high cells were determined among recipient PBLs by multiplying the number of cells by the percentage of CFSE+Thy-1.2+ cells and subsequently by the percentage of those cells that were either CD8low or CD8high, as shown in B. The t1/2 of the CD8low cells was calculated as described in Materials and Methods.

Close modal

As a second means to measure the in vivo t1/2 of CD8low cells, we labeled the dividing cells of Vβ5 Tg mice with BrdU and ablated existing CD8low cells with cortisone (Fig. 3,A). This protocol was designed to offset the asynchronous generation of CD8low cells in Tg mice (15). During the sampling period, BrdU levels were analyzed among CD8low cells, small CD8high cells, and CD8high blasts (Fig. 3,B). The very small numbers of CD8low cells relative to CD8high cells after cortisone-mediated ablation made it more accurate to track the percentage of BrdU+ cells within each compartment rather than their absolute numbers (data not shown). The partial cortisone sensitivity of CD8high blasts led to an initial decline in BrdU levels among total CD8high cells; however, this initial decline was followed by a period of relative stability lasting throughout the chase period (Fig. 3,C and data not shown). Cortisone-mediated depletion led to a decline in the percentage of CD8low cells that were BrdU+ between days 0 and 4. The percentage of BrdU+ CD8low cells increased at 6 days, consistent with depletion followed by regeneration of CD8low cells from labeled precursors (Fig. 3,B). The gradual depletion of the CD8low population between days 6 and 10 allowed us to calculate the t1/2 of CD8low cells in situ, without relying on adoptive transfer (Fig. 3 C). This t1/2, measured at ∼5 days, is in concordance with data from three experiments, each using multiple mice per timepoint. Thus, CD8low cells have an in vivo t1/2 of 3–5 days, which is considerably longer than the 0.5–1 day t1/2 we observed during in vitro experiments with CD8low cells (data not shown and 15).

FIGURE 3.

The disappearance of an experimentally synchronized population of CD8low cells permits an independent t1/2 measurement. A, In three independent experiments, Vβ5 Tg mice, aged 12–17 wk, were placed on water containing BrdU for 8 days, injected i.p. with 3 mg of hydrocortisone to ablate CD8low cells, maintained on BrdU water for an additional 2 days to allow time for clearance of CD8low cells, and finally transferred to normal drinking water for analysis during the chase period. The reappearance and subsequent disappearance of newly arisen BrdU-labeled CD8low cells were detected by staining splenocytes with PE-anti-CD8α and FITC-anti-BrdU for flow cytometric analysis. B, The upper left panel shows a representative forward scatter/CD8 profile of splenocytes from unmanipulated Vβ5 Tg mice. Gate G1 represents CD8low cells, G2 represents CD8high cells, and G3 represents CD8high blasts. Histograms of BrdU expression in gated CD8low populations are indicated before (upper right), at 4 days after (lower left), and at 6 days after (lower right) cortisone injection. The y-axis represents the relative cell number, and the percentage of cells falling within the BrdU+ markers is indicated. Each point depicts a pool of cells from two to four mice per timepoint. C, The percentage of CD8low cells and total CD8high cells (including blasts) that are BrdU+ is plotted at each timepoint from a pool of two to four mice, allowing calculation of the t1/2 of CD8low cells between days 6 and 10.

FIGURE 3.

The disappearance of an experimentally synchronized population of CD8low cells permits an independent t1/2 measurement. A, In three independent experiments, Vβ5 Tg mice, aged 12–17 wk, were placed on water containing BrdU for 8 days, injected i.p. with 3 mg of hydrocortisone to ablate CD8low cells, maintained on BrdU water for an additional 2 days to allow time for clearance of CD8low cells, and finally transferred to normal drinking water for analysis during the chase period. The reappearance and subsequent disappearance of newly arisen BrdU-labeled CD8low cells were detected by staining splenocytes with PE-anti-CD8α and FITC-anti-BrdU for flow cytometric analysis. B, The upper left panel shows a representative forward scatter/CD8 profile of splenocytes from unmanipulated Vβ5 Tg mice. Gate G1 represents CD8low cells, G2 represents CD8high cells, and G3 represents CD8high blasts. Histograms of BrdU expression in gated CD8low populations are indicated before (upper right), at 4 days after (lower left), and at 6 days after (lower right) cortisone injection. The y-axis represents the relative cell number, and the percentage of cells falling within the BrdU+ markers is indicated. Each point depicts a pool of cells from two to four mice per timepoint. C, The percentage of CD8low cells and total CD8high cells (including blasts) that are BrdU+ is plotted at each timepoint from a pool of two to four mice, allowing calculation of the t1/2 of CD8low cells between days 6 and 10.

Close modal

This longer-than-expected in vivo t1/2 led us to investigate possible functions for CD8low cells in vivo. To assess the capacity of CD8low cells to produce cytokines, we analyzed cytokine mRNA levels by RT-PCR from freshly sorted, unstimulated cells from Vβ5 Tg and nonTg mice (51). We significantly increased the sensitivity of this RT-PCR assay by Southern blotting PCR products and probing with an internal 32P end-labeled oligonucleotide. Quantitation by phosphorimaging indicated that CD8low cells express ∼10-fold more IFN-γ mRNA than do CD8high or CD4+ T cells from Tg mice or CD4+ T cells from nonTg mice (Fig. 4). No IFN-γ mRNA was detected in CD8+ T cells from nonTg mice (Fig. 4). CD8low cells express detectable but much lower levels of TNF-α mRNA, and even lower levels of IL-4 mRNA (data not shown). TGF-β mRNA is not detectable in any subpopulation (data not shown).

FIGURE 4.

CD8low cells express high levels of IFN-γ mRNA directly ex vivo. A, cDNA was synthesized using RNA isolated from CD8low, CD8high, and CD4+ T cells sorted from spleens of Vβ5 Tg mice and CD8+ and CD4+ T cells sorted from spleens of nonTg mice. Serial 3-fold dilutions of the cDNA reactions were PCR amplified with primers specific for IFN-γ and HPRT. Reactions were Southern blotted and probed. B, Signal intensities were quantitated by phosphorimaging; the relative IFN-γ expression in each cell population was determined by dividing the IFN-γ intensity by the HPRT intensity for each dilution. Error bars represent the SDs of results from these three dilutions. Results are representative of three experiments using independently sorted populations.

FIGURE 4.

CD8low cells express high levels of IFN-γ mRNA directly ex vivo. A, cDNA was synthesized using RNA isolated from CD8low, CD8high, and CD4+ T cells sorted from spleens of Vβ5 Tg mice and CD8+ and CD4+ T cells sorted from spleens of nonTg mice. Serial 3-fold dilutions of the cDNA reactions were PCR amplified with primers specific for IFN-γ and HPRT. Reactions were Southern blotted and probed. B, Signal intensities were quantitated by phosphorimaging; the relative IFN-γ expression in each cell population was determined by dividing the IFN-γ intensity by the HPRT intensity for each dilution. Error bars represent the SDs of results from these three dilutions. Results are representative of three experiments using independently sorted populations.

Close modal

To determine whether CD8low cells can be driven to produce cytokines, IFN-γ protein was quantitated by intracellular cytokine staining of PMA/ionomycin-stimulated, unseparated splenocytes from Vβ5 Tg mice counterstained with anti-CD8. The CD8low and CD8high phenotypes were maintained without TCR or coreceptor down-regulation during stimulation with PMA/ionomycin for ≤5 h (data not shown). In the absence of stimulation, intracellular staining was insufficiently sensitive to detect IFN-γ protein in either CD8low or CD8high cells, even after culture in monensin for 2–5 h to retain proteins intracellularly (Fig. 5,A and data not shown). However, after 3 h of stimulation with PMA/ionomycin, CD8low cells were enriched for IFN-γ production relative to CD8high cells (Fig. 5,A), paralleling the trend in IFN-γ mRNA levels. Furthermore, only 24% of CD4+ T cells from Tg mice, 12% of CD4+ T cells from nonTg mice, and 22% of CD8+ T cells from nonTg mice were IFN-γ+, compared with 43% IFN-γ+ among CD8low cells after 3 h of stimulation (Fig. 5 and data not shown). CD8low cells were similarly enriched for IFN-γ+ cells relative to CD8high cells upon stimulation for 3–48 h with plate-bound anti-CD3ε Abs in the presence or absence of exogenous IL-2 (data not shown). Therefore, CD8low cells can respond to TCR-mediated signals by producing IFN-γ.

FIGURE 5.

CD8low cells produce IFN-γ protein after PMA/ionomycin stimulation in vitro. A, Splenocytes from Vβ5 Tg mice were stimulated with PMA/ionomycin for 5 h, including 3 μM of monensin during the final 3 h. Cells were stained with PE-anti-CD8α and subsequently permeabilized and stained with FITC-anti-murine IFN-γ. Histograms represent intracellular fluorescence intensity in gated populations of stimulated CD8low cells, stimulated CD8high cells, unstimulated CD8low cells, and unstimulated CD8high cells. The percentages of positive cells are shown for each gated population. B, Splenocytes were stimulated for various times with PMA/ionomycin in the presence of monensin and subsequently stained with FITC-anti-IFN-γ, PE-anti-CD8α, and biotin-anti-CD44 followed by streptavidin-tricolor. The percentage of IFN-γ+ cells is graphed at each timepoint for CD8lowCD44high, CD8highCD44high, and CD8highCD44low cells.

FIGURE 5.

CD8low cells produce IFN-γ protein after PMA/ionomycin stimulation in vitro. A, Splenocytes from Vβ5 Tg mice were stimulated with PMA/ionomycin for 5 h, including 3 μM of monensin during the final 3 h. Cells were stained with PE-anti-CD8α and subsequently permeabilized and stained with FITC-anti-murine IFN-γ. Histograms represent intracellular fluorescence intensity in gated populations of stimulated CD8low cells, stimulated CD8high cells, unstimulated CD8low cells, and unstimulated CD8high cells. The percentages of positive cells are shown for each gated population. B, Splenocytes were stimulated for various times with PMA/ionomycin in the presence of monensin and subsequently stained with FITC-anti-IFN-γ, PE-anti-CD8α, and biotin-anti-CD44 followed by streptavidin-tricolor. The percentage of IFN-γ+ cells is graphed at each timepoint for CD8lowCD44high, CD8highCD44high, and CD8highCD44low cells.

Close modal

To ensure that enhanced IFN-γ production by CD8low cells is not solely a result of their previously activated phenotype, we analyzed the kinetics of IFN-γ production in splenocytes counterstained with CD8 and the memory marker CD44. Although all CD8low cells are CD44high directly ex vivo, only ∼60% of CD8high cells exhibit this phenotype in 15- to 18-wk-old Vβ5 Tg mice (data not shown and 15). CD44 levels remained stable throughout a 7-h PMA/ionomycin stimulation with or without monensin added to the culture, whereas a homogeneous induction of CD69 expression among CD8+ T cells confirmed that all of the cells within the cultures had been activated to a similar degree (data not shown). Although most IFN-γ-producing CD8high cells were CD44high, a greater percentage of CD8lowCD44high cells were IFN-γ+ than were CD8highCD44high cells at every timepoint (Fig. 5 B). Therefore, even when analyzing only previously activated cells, more CD8low cells produce IFN-γ than do CD8high cells. Additional evidence that CD8low cells produce functional IFN-γ was obtained by intracellular staining for PKR (the protein kinase associated with dsRNA), a protein whose expression is up-regulated by IFN-γ (53). The increase in levels of PKR after stimulation was more pronounced in CD8low than in CD8high cells (data not shown).

The activated phenotype of CD8low cells (15) and their ability to persist in vivo (Figs. 2 and 3) led us to ask whether CD8low cells resemble memory T cells in their ability to home to sites of inflammation. We used oxazolone to induce inflammation on the ears of Vβ5 Tg mice to determine whether CD8low cells accumulate in inflamed tissue. Sections from inflamed and control ears were stained with anti-CD8 mAbs, which allowed us to visualize both CD8low and CD8high cells within the sections (Fig. 6). Typically, CD8+ T cells appeared in clusters in tissue sections (Fig. 6 A and data not shown), and these infiltrates also contained CD4+ T cells (data not shown). We found an average of 5.7 CD8low and 4.7 CD8high cells per ×20 field after counting >200 cells in three independent experiments (data not shown). As expected, fewer CD8+ T cells were found within control ears, with 0.4 CD8low cells and 0.5 CD8high cells per ×20 field (data not shown).

FIGURE 6.

CD8low cells accumulate in inflamed ears. Mice were sensitized to oxazolone and challenged with oxazolone on the ear as described in Materials and Methods. Shown are representative ×20 sections from an inflamed ear (A and B) and a control ear (C and D) from the same mouse, stained with either anti-CD8 mAbs (A and C) or no primary Ab (B and D) as a control. Control sections stained with normal rat IgG as a primary stage revealed no staining. In CD8-stained sections, CD8low cells are indicated by asterisks and CD8high cells are indicated by arrowheads.

FIGURE 6.

CD8low cells accumulate in inflamed ears. Mice were sensitized to oxazolone and challenged with oxazolone on the ear as described in Materials and Methods. Shown are representative ×20 sections from an inflamed ear (A and B) and a control ear (C and D) from the same mouse, stained with either anti-CD8 mAbs (A and C) or no primary Ab (B and D) as a control. Control sections stained with normal rat IgG as a primary stage revealed no staining. In CD8-stained sections, CD8low cells are indicated by asterisks and CD8high cells are indicated by arrowheads.

Close modal

To confirm the finding that CD8low cells accumulate in inflamed sites, we used a variety of methods to induce inflammation in Vβ5 Tg mice. Because CD8low cells express little or no CD62 ligand and are normally absent in LNs (15), their accumulation in LNs reflects entry through afferent lymph draining inflamed areas. Therefore, we can detect CD8low cells both directly at sites of inflammation and indirectly by examining the LNs draining inflamed sites (Tables I and II). The immunohistochemical analysis of inflamed ears revealed that CD8low cells make up 55% of the total CD8+ T cells within the ear at 2 days after the induction of inflammation (Table I). CD8low cells also comprise 37% of the CD8+ T cells infiltrating gelfoam sponges implanted s.c. and 44% of the CD8+ T cells within peritoneal exudate after injection of thioglycollate (Table I). The spleens of the same mice are relatively depleted of CD8low cells (Table I). Therefore, we see a consistent enrichment for CD8low cells among CD8+ T cells within tissues inflamed by each of three protocols. We also analyzed the percentage of CD8low cells in LNs draining sites inflamed by oxazolone or CFA. In each case, we found a 2-fold increase in the %CD8low cells in the inflamed LNs relative to control LNs, although fewer CD8low cells were found within LNs than were found directly at the inflamed site (Table II).

Table I.

Homing of CD8low cells to sites of inflammation: CD8low cells detected directly at sites of inflammation

Method of Inflammatory Inductiona% CD8low/Total CD8+ T Cells in Inflamed SiteFold Increase in %CD8low Cells Within Inflamed Tissue Relative to Spleen
Oxazolone challenge on earb 55 ± 9d 6.5 
Sponge implanted s.c.c 37 2.3 
Thioglycollate i.p.c 44 ± 5e 2.0 
Method of Inflammatory Inductiona% CD8low/Total CD8+ T Cells in Inflamed SiteFold Increase in %CD8low Cells Within Inflamed Tissue Relative to Spleen
Oxazolone challenge on earb 55 ± 9d 6.5 
Sponge implanted s.c.c 37 2.3 
Thioglycollate i.p.c 44 ± 5e 2.0 
a

Inflammation was induced in Vβ5 Tg mice as outlined in Materials and Methods.

b

The representation of CD8low cells was determined by staining sections of inflamed ears of one to three Tg mice in each of three independent experiments for CD8 expression and counting the number of CD8low and CD8high cells to determine the percentage of CD8low cells/total CD8+ T cells. Control sections stained with normal rat IgG and without primary Ab were negative. In each experiment, 15–25 ×20 fields were counted for a total of >200 cells from each mouse.

c

The representation of CD8low cells was determined by staining cell suspensions with anti-CD8 Abs and determining the percentage of CD8low cells/total CD8+ T cells by flow cytometry. Isotype-matched controls were negative, and markers were set based on the CD8 profiles from PBLs of unmanipulated Tg mice.

d

Mean ± SE.

e

Mean ± range.

Table II.

Homing of CD8low cells to sites of inflammation: CD8low cells detected in LN draining inflamed sites

Method of Inflammatory Inductiona% CD8low/Total CD8+ T Cells in LN Draining Inflamed SiteFold Increase in %CD8low Cells Within Inflamed LNs Relative to Control LNs
Oxazolone challenge on earb 8.1 ± 1.3c 2.0 
Oxazolone challenge on backb 12 ± 4d 2.1 
CFA in footpadb 7.8 ± 2.4c 2.1 
Method of Inflammatory Inductiona% CD8low/Total CD8+ T Cells in LN Draining Inflamed SiteFold Increase in %CD8low Cells Within Inflamed LNs Relative to Control LNs
Oxazolone challenge on earb 8.1 ± 1.3c 2.0 
Oxazolone challenge on backb 12 ± 4d 2.1 
CFA in footpadb 7.8 ± 2.4c 2.1 
a

Inflammation was induced in Vβ5 Tg mice as outlined in Materials and Methods.

b

The representation of CD8low cells was determined by staining cell suspensions with anti-CD8 Abs and determining the percentage of CD8low cells/total CD8+ T cells by flow cytometry. Isotype-matched controls were negative, and markers were set based on the CD8 profiles from PBLs of unmanipulated Tg mice.

c

Mean ± SE.

d

Mean ± range.

To determine whether CD8low cells at sites of inflammation could produce IFN-γ, we stained sections of inflamed ears with both anti-CD8 and anti-IFN-γ mAbs (Table III). Within inflamed ears, 31% of CD8+ T cells were both IFN-γ+ and CD8low, whereas only 18% were IFN-γ+CD8high (Table III). This corresponds to 56% of CD8low cells and 40% of CD8high cells expressing IFN-γ in inflamed ears. Therefore, in inflamed tissues, CD8low cells both accumulate and produce IFN-γ.

Table III.

Homing of CD8low cells to sites of inflammation: IFN-γ production by CD8low cells in inflamed earsa

Phenotype of CD8+ T Cells% of Total CD8+ T Cellsb
CD8lowIFN-γ+ 31 
CD8lowIFN-γ 24 
CD8highIFN-γ+ 18 
CD8highIFN-γ 27 
Phenotype of CD8+ T Cells% of Total CD8+ T Cellsb
CD8lowIFN-γ+ 31 
CD8lowIFN-γ 24 
CD8highIFN-γ+ 18 
CD8highIFN-γ 27 
a

Sections of oxazolone-inflamed ears were prepared and stained for two-color immunohistochemistry as described in Materials and Methods. Control staining with normal rat IgG and without primary Abs demonstrated the specificity of the staining patterns and indicated that no cross-reactivity between color reactions developed.

b

Percentages represent the fraction of total CD8+ T cells that each subset comprises and were counted from >120 ×40 sections using one to two Tg mice in each of three independent experiments.

Using Vβ5 Tg mice as a model system, we have explored the ability of tolerant CD8+ T cells to persist and retain some functional capabilities. As Vβ5 Tg mice age, a population of CD8low T cells develops in response to endogenous self Ag(s) (15, 44). An unusual feature that characterizes this model of tolerance induction is the fact that the anergic CD8low T cell population is polyclonal, using a number of different TCR Vα-chains paired with the Tg TCR Vβ5 chain (44). Therefore, we are able to follow a polyclonal but phenotypically distinct population of cells responding to the self Ags expressed in aging mice in vivo.

Our analyses of this physiologically relevant model system have provided clues about how anergic CD8low cells are generated in vivo. The loss of CD8high blasts as CD8low cells recover following CD8low ablation with cortisone (Fig. 3 and data not shown) and the increased turnover of CD8high cells in Vβ5 Tg mice (Fig. 1) both support a model in which a CD8high cell encounters its tolerogen, becomes activated, undergoes blastogenesis, and subsequently down-regulates its TCR and coreceptor to transit into the CD8low compartment. The efficient development of anergic CD8low T cells in adult thymectomized mice (15) provides further evidence that CD8low cells arise from CD8high cells after encounter with tolerogen in the lymphoid periphery. Once they are within the CD8low compartment, these cells are compromised in their ability to proliferate in vivo (Fig. 1,B) and in vitro in the presence of exogenous IL-2 (Fig. 1,C and 15), indicating that they have become functionally anergic (17, 18, 41, 52, 54). Whereas anergy among CD4+ T cells can frequently be reversed by culture in the presence of IL-2 (23, 55), CD8low cells are CD25 (15) and remain unable to proliferate in the presence of IL-2 (Fig. 1 and 15), indicating that differences in the functional defects among anergic CD4+ and CD8+ T cells may exist.

Although CD8low cells exist as intermediates in a peripheral deletion pathway (15), this compartment is relatively stable in vivo, particularly in light of its brief in vitro t1/2 and potential for autoreactivity. We chose two independent methods to assess the in vivo survival of these T cells. The in vivo t1/2 estimate of 3 days in the adoptive transfer experiment may be an underestimate, because removing the CD8low cells from their environment may shorten their t1/2 (Fig. 2). Alternatively, imperfect synchronization of CD8low cells would lead to an overestimate of their 5 day t1/2 as calculated using cortisone-injected animals (Fig. 3). Preferential homing of CD8low cells to areas other than the spleen or blood is unlikely to cause skewing of the results, because we did not find CD8low cells in the LNs in these experiments. Furthermore, CD8low cells do not accumulate in the liver, among intraepithelial lymphocytes, or within Peyer’s patches of the gut (data not shown). Therefore, a consistent estimate of in vivo CD8lowt1/2 is ∼4 days, leading to two key observations. First, this in vivo t1/2 is significantly longer than the in vitro t1/2 of CD8low cells, measured at ∼1 day for unstimulated cells and at <0.5 day for anti-CD3 stimulated cells (Ref. 15 and data not shown). Second, an in vivo t1/2 of 4 days leaves these anergic cells with sufficient time to home to, and possibly function within, various tissues or sites.

The difference between the t1/2 of CD8low cells in vivo and in vitro (Figs. 2 and 3 and data not shown) and the fact that CD8low cells are not apoptotic directly ex vivo (Ref. 15 and Fig. 1, B and C) suggest that CD8low cells are not merely dying cells. In addition, CD8low cells appear to survive longer in vivo than cells undergoing activation-induced cell death (AICD) by either the Fas-mediated (56, 57, 58, 59, 60) or the TNF receptor (TNFR)-mediated (61, 62, 63, 64) pathways. In one report using bulk cultures of lymphocytes, T cell death within the first 24 h of anti-CD3 stimulation was primarily Fas-mediated, but both Fas- and TNFR-mediated death were involved at 48 h (61). Although data suggest that CD8+ T cells are more susceptible to the slower TNFR-mediated death than to Fas-mediated death (61, 63, 64), the 4-day in vivo t1/2 of CD8low cells is longer than would be predicted by a population of CD8+ T cells undergoing AICD. In fact, the kinetics of AICD correlate more closely with the kinetics of CD8low cell death in vitro rather than in vivo. Therefore, it is tempting to speculate that a survival signal maintains CD8low cells in vivo. Recent data suggest that this survival signal could be delivered by IL-15, which can maintain memory cells within the CD8+ but not the CD4+ T cell compartment (65).

Placing the in vivo t1/2 calculation of CD8low cells into the context of other estimates of T cell life span is complicated by the existence of both resting and rapidly dividing subpopulations of cells within naive and memory T cell compartments (47). Tracking the decay of dividing, labeled cells is prone to error when one labeled cell gives rise to two labeled daughters, a risk that probably does not complicate the analysis of largely nondividing CD8low cells. Most current estimates suggest that both naive and memory T cells are generally long-lived and can persist for months (47, 66), although other estimates suggest that T cells survive only a few days (67). Although this range of estimated T cell life spans is broad, a 4-day t1/2 of anergic CD8low T cells could represent a significant fraction of an average T cell life span. Our estimate of the CD8low cell life span is consistent with the 50% renewal time of 3–4 days described for anergic B cells when compared with the 4–5 wk renewal time of naive B cells (68). The relative stability of the CD8low compartment is also in line with recent data suggesting that subpopulations of anergic CD4+ and CD8+ T cells can persist in vivo or in vitro (17, 18, 29, 54).

Despite their inability to proliferate and mediate target cell lysis in vitro, the in vivo persistence of CD8low cells (Figs. 2 and 3) and their capacity to release calcium from internal stores upon TCR cross-linking (15) led us to question whether CD8low cells might serve some immune function. Therefore, we evaluated whether CD8low cells could secrete cytokines in a manner analogous to cytokine secretion by anergic CD4+ T cells (37, 38, 39, 40, 41) and CD8+ T cells (18). Using a sensitive RT-PCR assay, we found that freshly sorted, unstimulated CD8low cells express high levels of IFN-γ mRNA (Fig. 4). Although this observation confirms that anergic cells can produce cytokines, cytokine secretion by anergic CD8low cells is distinct from cytokine production by cells undergoing AICD (69, 70). CD8low cells are not apoptotic directly ex vivo (Ref. 15 and Fig. 1, B and C), and prior studies demonstrating cytokine production by dying cells used CD4+ T cells activated in vitro (69, 70) rather than CD8+ T cells responding to endogenous Ags in vivo. The absence of detectable IFN-γ mRNA in the largely naive CD8+ T cells from nonTg mice indicates that the RT-PCR assay is sensitive but also very specific. In addition, CD8low cells produce IFN-γ protein after a brief stimulation in vitro (Fig. 5 and data not shown). Although these data indicate that efficient IFN-γ production is primarily restricted to cells that have been activated previously (Figs. 4 and 5), the high IFN-γ production by CD8low cells does not solely reflect their state of prior activation. At 12–20 wk of age, when these Vβ5 Tg mice were analyzed, virtually all of the CD4+ T cells have been activated previously and are CD44high (Ref. 14 and data not shown); however, these CD4+CD44high cells still have 10-fold less IFN-γ mRNA (Fig. 4) and contain fewer IFN-γ+ cells after PMA/ionomycin stimulation (data not shown) compared with CD8low cells. In addition, CD8low cells produce IFN-γ more readily than do CD8highCD44high cells during in vitro restimulation with PMA/ionomycin (Fig. 5 B).

Why do these anergic cells, which are unable to kill, proliferate, or efficiently respond to IL-2, nevertheless produce IFN-γ more readily than their fully functional CD8high counterparts? One answer may be that CD8low cells are poised to secrete cytokines rapidly following their encounter with the tolerogen, but their inability to expand limits the danger they pose to the host. If this is the case, what situations would compel CD8low cells to secrete cytokines? Either TCR-mediated signals, to which CD8low cells can respond by fluxing calcium (15) and producing IFN-γ (data not shown), or exposure to an inflammatory environment could contribute to this cytokine secretion. Their high expression of adhesion molecules such as CD54, CD49d, and CD11a and their recirculation pattern (15) may allow CD8low cells to accumulate at sites of inflammation, where cytokine secretion might be expected to have an especially dramatic effect. In fact, we have consistently demonstrated an accumulation of CD8low cells at sites of inflammation and LNs draining these sites (Tables I and II and Fig. 6). The subtle 2-fold enrichment for CD8low cells that we observe is remarkably consistent, despite our induction of inflammation using a variety of methods. The tendency of CD8low cells to die may make their accumulation after infiltration transient, thus limiting this enrichment. Cell death, perhaps at the site of inflammation, may also explain why less enrichment for CD8low cells was seen at sites of inflammation induced by sponges and by thioglycollate (analyzed at 3 and 4 days posttreatment, respectively) than in oxazolone-inflamed ears (analyzed 2 days after induction of inflammation). CD8low cell death at the site of inflammation would also explain why fewer CD8low cells are observed in LNs draining inflamed sites than within the inflamed sites themselves.

Finally, two-color immunohistochemistry indicates that CD8low cells are not only able to accumulate at sites of inflammation but can also produce IFN-γ protein at these sites (Table III). Taken together, these data indicate that anergic CD8low cells persist for a sufficient time in the mouse to be capable of a broad range of functional responses. CD8low cells can home to sites of inflammation and produce IFN-γ. Because CD8low cells can be detected in nonTg mice (44), these processes are not limited to Tg mice and are likely to occur in a more subtle form in unmanipulated mice with diverse TCRs. Our data point to the potential of deletional intermediates to influence immune responses in vivo, revealing a new level of complexity to the process of tolerance induction in the lymphoid periphery. Continued characterization of this model system will increase our understanding of the pathways that T cells travel between full responsiveness and deletion.

We thank S. Tucker for sharing unpublished findings, E. Clark for the gift of CFSE, and our many colleagues for helpful comments on the manuscript. We also thank K. Kline, G. Turk, and M. Anderson for technical assistance; K. Allen for assistance with flow cytometry; D. Wilson for maintaining our mouse colony; L. Hood for use of the PhosphorImager; and A. Aderem for use of phosphorimaging analysis software.

1

This work was supported by Research Grant AG-13078 from the National Institutes of Health (to P.J.F.) and Basic Immunology Training Grant CA-90537 (to S.R.D.) C.A.B has been supported by Medical Scientist Training Program Grant GM-07226 from the National Institutes of Health, the Life and Health Insurance Medical Research Fund, and the Poncin Scholarship Fund.

4

Abbreviations used in this paper: B6, C57BL/6; Tg, transgenic; B6.Thy1.1, B6.PL-Thy-1a/Cy; BrdU, bromodeoxyuridine; CFSE, 5-carboxyfluorescein diacetate-succinimidyl ester; 7-AAD, 7-amino-actinomycin D; PI, propidium iodide; DAB, 3,3-diaminobenzidine; LN, lymph node; HPRT, hypoxanthine phosphoribosyltransferase; AICD, activation-induced cell death; PKR, protein kinase associated with dsRNA; TNFR, TNF receptor.

1
Nossal, G. J. V..
1994
. Negative selection of lymphocytes.
Cell
76
:
229
2
Miller, J. F., W. R. Heath.
1993
. Self-ignorance in the peripheral T-cell pool.
Immunol. Rev.
133
:
131
3
Webb, S., C. Morris, J. Sprent.
1990
. Extrathymic tolerance of mature T cells: clonal elimination as a consequence of immunity.
Cell
63
:
1249
4
Huang, L., I. N. Crispe.
1993
. Superantigen-driven peripheral deletion of T cells: apoptosis occurs in cells that have lost the αβ T cell receptor.
J. Immunol.
151
:
1844
5
Scott, B., R. Liblau, S. Degermann, L. A. Marconi, L. Ogata, A. J. Caton, H. O. McDevitt, D. Lo.
1994
. A role for non-MHC genetic polymorphism in susceptibility to spontaneous autoimmunity.
Immunity
1
:
73
6
Qin, S., S. P. Cobbold, H. Pope, J. Elliot, D. Kioussis, J. Davies, H. Waldmann.
1993
. “Infectious” transplantation tolerance.
Science
259
:
974
7
Moskophidis, D., F. Lechner, H. Pircher, R. M. Zinkernagel.
1993
. Virus persistence in acutely infected immunocompetent mice by exhaustion of antiviral cytotoxic effector T cells.
Nature
362
:
758
8
Rocha, B. A., A. Grandien, A. A. Freitas.
1995
. Anergy and exhaustion are independent mechanisms of peripheral T cell tolerance.
J. Exp. Med.
181
:
933
9
Rammensee, H. G., R. Kroschewski, B. Frangoulis.
1989
. Clonal anergy induced in mature Vβ6+ T lymphocytes on immunizing Mls-1b mice with Mls-1a-expressing cells.
Nature
339
:
541
10
Burkly, L. C., D. Lo, O. Kanagawa, R. L. Brinster, R. A. Flavell.
1989
. T-cell tolerance by clonal anergy in transgenic mice with nonlymphoid expression of MHC class II I-E.
Nature
342
:
564
11
Blackman, M. A., H. Gerhard-Burgert, D. L. Woodland, E. Palmer, J. W. Kappler, P. Marrack.
1990
. A role for clonal inactivation in T cell tolerance to Mls-1a.
Nature
345
:
540
12
Rocha, B., H. von Boehmer.
1991
. Peripheral selection of the T cell repertoire.
Science
251
:
1225
13
Schonrich, G., U. Kalinke, F. Momburg, M. Malissen, M. Schmitt-Verhulst, B. Malissen, G. J. Hammerling, 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
14
Fink, P. J., C. A. Fang, G. L. Turk.
1994
. The induction of peripheral tolerance by the chronic activation and deletion of CD4+Vβ5+ T cells.
J. Immunol.
152
:
4270
15
Dillon, S. R., V. L. MacKay, P. J. Fink.
1995
. A functionally compromised intermediate in extrathymic CD8+ T cell deletion.
Immunity
3
:
321
16
Kundig, T. M., A. Shahinian, K. Kawai, H.-W. Mittrucker, E. Sebzda, M. F. Bachmann, T. W. Mak, P. S. Ohashi.
1996
. Duration of TCR stimulation determines costimulatory requirement of T cells.
Immunity
5
:
41
17
Pape, K. A., R. Merica, A. Mondino, A. Khoruts, M. K. Jenkins.
1998
. Direct evidence that functionally impaired CD4+ T cells persist in vivo following induction of peripheral tolerance.
J. Immunol.
160
:
4719
18
Tanchot, C., S. Guillaume, J. Delon, C. Bourgeois, A. Franzke, A. Sarukhan, A. Trautmann, B. Rocha.
1998
. Modifications of CD8+ T cell function during in vivo memory or tolerance induction.
Immunity
8
:
581
19
Sloan-Lancaster, J., B. D. Evavold, P. M. Allen.
1993
. Induction of T-cell anergy by altered T-cell-receptor ligands on live antigen-presenting cells.
Nature
363
:
156
20
Schonrich, G., J. Alferink, A. Klevenz, G. Kublbeck, N. Auphan, A.-M. Schmitt-Verhulst, G. J. Hammerling, B. Arnold.
1994
. Tolerance induction as a multi-step process.
Eur. J. Immunol.
24
:
285
21
Kawai, K., P. S. Ohashi.
1995
. Immunological function of a defined T-cell population tolerized to low-affinity self antigens.
Nature
374
:
68
22
Akkaraju, S., W. Y. Ho, D. Leong, K. Canaan, M. M. Davis, C. C. Goodnow.
1997
. A range of CD4 T cell tolerance: partial inactivation to organ-specific antigen allows nondestructive thyroiditis or insulinitis.
Immunity
7
:
255
23
Schwartz, R. H..
1996
. Models of T cell anergy: is there a common molecular mechanism?.
J. Exp. Med.
184
:
1
24
Heath, W. R., J. Allison, M. W. Hoffmann, G. Schonrich, G. Hammerling, B. Arnold, J. F. A. P. Miller.
1992
. Autoimmune diabetes as a consequence of locally produced interleukin-2.
Nature
359
:
547
25
Ferber, I., G. Schonrich, J. Schenkel, A. L. Mellor, G. J. Hammerling, B. Arnold.
1994
. Levels of peripheral T cell tolerance induced by different doses of tolerogen.
Science
263
:
674
26
Guerder, S., J. Meyerhoff, R. Flavell.
1994
. The role of the T cell costimulator B7-1 in autoimmunity and the induction and maintenance of tolerance to peripheral antigen.
Immunity
1
:
155
27
MacDonald, H. R., R. K. Lees, S. Baschieri, T. Herrmann, A. R. Lussow.
1993
. Peripheral T cell reactivity to bacterial superantigens in vivo: the response/anergy paradox.
Immunol. Rev.
133
:
105
28
Combadiere, B., C. Reis e Sousa, C. Trageser, L.-X. Zheng, C. R. Kim, M. J. Lenardo.
1998
. Differential TCR signaling regulates apoptosis and immunopathology during antigen responses in vivo.
Immunity
9
:
305
29
Tanchot, C., B. Rocha.
1997
. Peripheral selection of T cell repertoires: the role of continuous thymic output.
J. Exp. Med.
186
:
1099
30
Li, W., C. D. Whaley, A. Mondino, D. L. Mueller.
1996
. Blocked signal transduction to the ERK and JNK protein kinases in anergic CD4+ T cells.
Science
271
:
1272
31
Fields, P. E., T. F. Gajewski, F. W. Fitch.
1996
. Blocked Ras activation in anergic CD4+ T cells.
Science
271
:
1276
32
DeSilva, D. R., W. S. Feeser, E. J. Tancula, P. A. Scherlie.
1996
. Anergic T cells are defective in both jun NH2-terminal kinase and mitogen-activated protein kinase signaling pathways.
J. Exp. Med.
183
:
2017
33
La Face, D. M., C. Couture, K. Anderson, G. Shih, J. Alexander, A. Sette, T. Mustelin, A. Altman, H. M. Grey.
1997
. Differential T cell signaling induced by antagonist peptide-MHC complexes and the associated phenotypic responses.
J. Immunol.
158
:
2057
34
Madrenas, J., R. L. Wange, J. L. Wang, N. Isakov, L. E. Samelson, R. N. Germain.
1995
. ζ phosphorylation without ZAP-70 activation induced by TCR antagonists or partial agonists.
Science
267
:
515
35
Boussiotis, V. A., G. J. Freeman, A. Berezovskaya, D. L. Barber, L. M. Nadler.
1997
. Maintenance of human T cell anergy: blocking of IL-2 gene transcription by activated Rap1.
Science
278
:
124
36
Kersh, E. N., A. S. Shaw, P. M. Allen.
1998
. Fidelity of T cell activation through multistep T cell receptor ζ phosphorylation.
Science
281
:
572
37
Evavold, B. D., P. M. Allen.
1991
. Separation of IL-4 production from Th cell proliferation by an altered T cell receptor ligand.
Science
252
:
1308
38
Mueller, D. L., L. Chiodetti, P. A. Bacon, R. H. Schwartz.
1991
. Clonal anergy blocks the response to IL-4, as well as the production of IL-2, in dual-producing Th cell clones.
J. Immunol.
147
:
4118
39
Hargreaves, R. G., N. J. Borthwick, M. S. G. Montani, E. Piccolella, P. Carmichael, R. I. Lechler, A. N. Akbar, G. Lombardi.
1997
. Dissociation of T cell anergy from apoptosis by blockade of Fas/Apo-1 (CD95) signaling.
J. Immunol.
158
:
3099
40
Itoh, Y., R. N. Germain.
1997
. Single-cell analysis reveals regulated hierarchical T cell antigen receptor signaling thresholds and intraclonal heterogeneity for individual cytokine responses of CD4+ T cells.
J. Exp. Med.
186
:
757
41
Malvey, E.-N., M. K. Jenkins, D. L. Mueller.
1998
. Peripheral immune tolerance blocks clonal expansion but fails to prevent the differentiation of Th1 cells.
J. Immunol.
161
:
2168
42
Dubois, P. M., M. Pihlgren, M. Tomkowiak, M. Van Mechelen, J. Marvel.
1998
. Tolerant CD8 T cells induced by multiple injections of peptide antigen show impaired TCR signaling and altered proliferative responses in vitro and in vivo.
J. Immunol.
161
:
5260
43
Fink, P. J., K. Swan, G. Turk, M. W. Moore, F. R. Carbone.
1992
. Both intrathymic and peripheral selection modulate the differential expression of Vβ5 among CD4+ and CD8+ T cells.
J. Exp. Med.
176
:
1733
44
Blish, C. A., B. J. Gallay, G. L. Turk, K. M. Kline, W. Wheat, P. J. Fink.
1999
. Chronic modulation of the T cell receptor repertoire in the lymphoid periphery.
J. Immunol.
162
:
3131
45
Harlow, E., D. Lane.
1988
.
Antibodies: A Laboratory Manual
Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.
46
Fulcher, D. A., A. B. Lyons, S. L. Korn, M. C. Cook, C. Koleda, C. Parish, B. Fazekas de St. Groth, A. Basten.
1996
. The fate of self-reactive B cells depends primarily on the degree of antigen receptor engagement and availability of T cell help.
J. Exp. Med.
183
:
2313
47
Tough, D. F., J. Sprent.
1994
. Turnover of naive- and memory-phenotype T cells.
J. Exp. Med.
179
:
1127
48
Wang, A., S. D. Su, D. F. Mark.
1984
. Site-specific mutagenesis of the human interleukin-2 gene: structure-function analysis of the cysteine residues.
Science
224
:
1431
49
Openshaw, P., E. E. Murphy, N. A. Hosken, V. Maino, K. Davis, K. Murphy, A. O’Garra.
1995
. Heterogeneity of intracellular cytokine synthesis at the single-cell level in polarized T helper 1 and T helper 2 populations.
J. Exp. Med.
182
:
1357
50
Chomczynski, P., N. Sacchi.
1987
. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction.
Anal. Biochem.
162
:
156
51
Reiner, S. L., S. Zheng, D. B. Corry, R. M. Locksley.
1993
. Constructing polycompetitor cDNAs for quantitative PCR.
J. Immunol. Methods
165
:
37
52
Jenkins, M. K., R. H. Schwartz.
1987
. Antigen presentation by chemically modified splenocytes induces antigen-specific T cell unresponsiveness in vitro and in vivo.
J. Exp. Med.
2
:
302
53
Katze, M. G..
1995
. Regulation of the interferon-induced PKR: can viruses cope?.
Trends Microbiol.
3
:
75
54
Ryan, K. R., B. D. Evavold.
1998
. Persistence of peptide-induced CD4+ T cell anergy in vitro.
J. Exp. Med.
187
:
89
55
Beverly, B., S.-M. Kang, M. J. Lenardo, R. H. Schwartz.
1992
. Reversal of in vitro T cell clonal anergy by IL-2 stimulation.
Int. Immunol.
4
:
661
56
Dhein, J., H. Walczak, C. Baumler, K. M. Debatin, P. H. Krammer.
1995
. Autocrine T-cell suicide mediated by APO-1/(Fas/CD95).
Nature
373
:
438
57
Brunner, T., R. J. Mogil, N. J. LaFace, N. J. Yoo, A. Mahboubi, F. Echeverri, S. J. Martin, W. R. Force, D. H. Lynch, C. F. Ware, D. R. Green.
1995
. Cell-autonomous Fas (CD95)/Fas-ligand interaction mediated activation-induced apoptosis in T-cell hybridomas.
Nature
373
:
441
58
Ju, S. T., D. J. Panka, H. Cui, R. Ettinger, M. El-Khatib, D. H. Sherr, B. Z. Stanger, A. Marshak-Rothstein.
1995
. Fas (CD95)/Fas-L interactions required for programmed cell death after activation.
Nature
373
:
444
59
Van Parijs, L., A. Ibraghimov, A. K. Abbas.
1996
. The roles of costimulation and Fas in T cell apoptosis and peripheral tolerance.
Immunity
4
:
321
60
Rafaeli, Y., L. Van Parijs, C. A. London, J. Tschopp, A. K. Abbas.
1998
. Biochemical mechanisms of IL-2-regulated Fas-mediated T cell apoptosis.
Immunity
6
:
615
61
Zheng, L., G. Fisher, R. E. Miller, J. Peschon, D. H. Lynch, M. J. Lenardo.
1995
. Induction of apoptosis in mature T cells by tumor necrosis factor.
Nature
377
:
348
62
Sytwu, H.-K., R. S. Liblau, H. O. McDevitt.
1996
. The roles of Fas/APO-1 (CD95) and TNF in antigen-induced programmed cell death in T cell receptor transgenic mice.
Immunity
5
:
17
63
Speiser, D. E., E. Sebzda, T. Ohteki, M. F. Bachmann, K. Pfeffer, T. W. Mak, P. S. Ohashi.
1996
. Tumor necrosis factor receptor p55 mediates deletion of peripheral cytotoxic lymphocytes in vivo.
Eur. J. Immunol.
26
:
3055
64
Alexander-Miller, M. A., G. R. Leggatt, A. Sarin, J. A. Berzofsky.
1996
. Role of antigen, CD8, and cytotoxic T lymphocyte (CTL) avidity in high dose antigen induction of apoptosis of effector CTL.
J. Exp. Med.
184
:
485
65
Zhang, X., S. Sun, I. Hwang, D. F. Tough, J. Sprent.
1998
. Potent and selective stimulation of memory-phenotype CD8+ T cells in vivo by IL-15.
Immunity
8
:
591
66
von Boehmer, H., K. Hafen.
1993
. The life span of naive α/β T cells in secondary lymphoid organs.
J. Exp. Med.
177
:
891
67
Rocha, B., C. Penit, C. Baron, F. Vasseur, N. Dautigny, A. A. Freitas.
1990
. Accumulation of bromodeoxyuridine-labeled cells in central and peripheral lymphoid organs: minimal estimate of production and turnover rates of mature lymphocytes.
Eur. J. Immunol.
20
:
1697
68
Fulcher, D. A., A. Basten.
1994
. Reduced life span of anergic self-reactive B cells in a double-transgenic model.
J. Exp. Med.
179
:
125
69
Lenardo, M. J..
1991
. Interleukin-2 programs mouse αβ T lymphocytes for apoptosis.
Nature
353
:
858
70
Combadiere, B., C. Reis e Sousa, R. N. Germain, M. J. Lenardo.
1998
. Selective induction of apoptosis in mature T lymphocytes by variant T cell receptor ligands.
J. Exp. Med.
187
:
349