Peripheral Immune Tolerance Blocks Clonal Expansion but Fails to Prevent the Differentiation of Th1 Cells¹

The immune system has the remarkable ability to mount a highly specific response against invading pathogens while ignoring self molecules. This specificity is determined in part by the T lymphocyte, which expresses a randomly generated and unique TCR that recognizes a peptide Ag bound to an MHC molecule (1). Because MHC molecules can bind both self as well as foreign peptides, the specificities of the peripheral TCR repertoire or function of self-reactive T cells must be regulated such that the immune system ignores the self peptides. The physical elimination of autoreactive T cells during thymocyte development is the primary mechanism used by the immune system to establish such self-tolerance (2, 3). However, not all self peptides are present in the thymus. Therefore, the immune system must either ignore a tissue-specific self peptide (4), or develop an active self-tolerance that relies on the suppression (5, 6), physical elimination (7, 8), or functional inactivation (9, 10) of mature autoreactive T cells.

Inactivation of T cells by the tolerance mechanism called clonal anergy was originally described using a tissue culture system of cloned T cells (11, 12). Clonal anergy has since been defined as a reversible, induced tolerance state in which the T lymphocyte cannot produce its autocrine growth factor IL-2 or proliferate in response to the Ag it recognizes (13). In vitro, this unresponsive state is induced by stimulation of the T cell through its TCR in the absence of costimulatory signals, such as those occurring as a result of the interaction of B7 molecules on the APC with CD28 receptors on the T cell (14, 15). In the absence of such costimulatory signals, T cells fail to proliferate, and TCR occupancy unaccompanied by proliferation down-regulates the T cell’s responsiveness (16). Biochemical analyses have suggested that anergic T cells lose the capacity to synthesize IL-2 because of a defect in the coupling of the Ag receptor to downstream p21^Ras-dependent signaling cascades (17, 18, 19) necessary for the induction of activating protein-1-dependent IL-2 gene transcription (20, 21, 22). Importantly, the clonal anergy observed in vitro has not been associated with a global defect in T cell activation events. Specifically, the secretion of IFN-γ by anergic CD4⁺ Th cells is not blocked after the induction of clonal anergy (23). Furthermore, anergic CD8⁺CTL are capable of lysing target cells (24). Finally, anergy induction cannot effectively interfere with the delivery of helper signals for the stimulation of B cell polyclonal IgG secretion in vitro (25). Consistent with these functional results, anergic T cells can be shown to develop increases in intracellular calcium free ion concentrations that result in the dephosphorylation and nuclear translocation of the nuclear factor of activated T cells (NFAT) upon stimulation (22). Thus, the clonal anergy mechanism appears to selectively interrupt signal transduction to the nucleus, resulting in a defective proliferative capacity, yet it may not eliminate the potential of a T cell to participate in an effector cell response.

These findings of selectivity in the inhibitory effects of clonal anergy raise uncertainty as to how this mechanism might contribute to the development of immune tolerance in vivo. Proliferative T cell clonal anergy has previously been reported in animals made tolerant of foreign Ag by the systemic administration of aqueous peptides in the absence of infection or adjuvant (26). Clonal anergy has also been observed after the systemic exposure of animals to superantigens such as staphylococcal enterotoxin B (SEB)³ (27). Interestingly, TCR-Vβ8 transgenic mice immunized against tetanus toxoid showed no evidence of tolerance within the tetanus toxoid-specific Th cell population following exposure to SEB, despite the ability of the SEB to induce proliferative unresponsiveness in the naive T cell population (28). Likewise, systemic exposure of naive mice to the lymphocytic choriomeningitis virus (LCMV) glycoprotein-derived peptide GP33 induced tolerance within CD8⁺CTL precursors, whereas mice that had first been infected with LCMV instead developed destructive spleen immunopathology in response to systemic administration of the aqueous GP33 peptide (29). These findings suggest either that memory T cells are insensitive to clonal anergy induction, or that the anergy mechanism cannot regulate all effector functions by an expanded T cell clone in the peripheral immune system.

Previously, in vivo models of T cell activation have not had the power to quantitate the effector cell capabilities of lymphocytes following the induction of an Ag-specific tolerance. Specifically, differences in the frequencies of Ag-reactive T cells that develop as a consequence of clonal expansion and/or activation-induced cell death have complicated the analysis. Therefore, we have developed an experimental system that both allows for the induction of T cell tolerance in mice using an aqueous peptide Ag, and that permits the subsequent recovery and identification of Ag-specific cells from these tolerant animals. Using this system, the functional capacities of equal numbers of normal and tolerant T cells can be directly compared following their adoptive transfer into normal recipient mice. In this study, we describe the results of experiments designed to assess the functional effects of clonal anergy induction on both naive and Ag-experienced Th cells. Our findings indicate that exposure to an aqueous peptide Ag in vivo results in the induction of a clonal anergy that limits the expansion of T cells responding to that Ag, and yet fails to interrupt the generation and delivery of Th1-like effector activities by these lymphocytes.

BALB/c and BALB/c nu/nu (nu/nu) mice, 6 to 8 wk old, were purchased from Charles River (Wilmington, MA) through a contract with National Cancer Institute at National Institutes of Health (Frederick, MD), and housed under specific pathogen-free conditions. The DO11.10 TCR-transgenic (TCR-Tg) mouse line has been described previously (30) and was bred to homozygosity in our animal facility in accordance with the guidelines of National Institutes of Health. CD4⁺ T cells in these mice have been engineered to specifically recognize the chicken OVA peptide 323–339. All mice used for any given experiment were sex and age matched.

Normal BALB/c mice were immunized with 100 μg OVA peptide 323–339 (OVAp) (synthesized and purified within Microchemical Facilities at University of Minnesota, Minneapolis, MN) emulsified in incomplete Freund’s adjuvant (IFA) (Difco, Detroit, MI) by s.c. injection in a volume of 100 μl at the base of the tail. DNP-specific Abs were subsequently elicited by secondary immunization at the same site either with 10 or 100 μg DNP-conjugated OVA (Pierce, Rockford, IL) (DNP-OVA) emulsified in CFA (Difco), or with CFA alone as a negative control. Alternatively, anti-DNP Ab production was determined in some experiments after only primary s.c. immunization with DNP-OVA in CFA. BALB/c mice that received an adoptive T cell transfer were immunized 1 or 2 days later s.c. at the base of the tail with either 10 or 100 μg DNP-OVA emulsified in CFA, or with CFA alone as a negative control, to induce a T cell clonal expansion and the production of anti-DNP Ab. For the measurement of ex vivo intracellular IL-2 levels in Ag-stimulated T cells, 250 μg OVAp was injected i.v. into adoptive transfer recipient BALB/c mice 1 day after cell transfer. Delayed-type hypersensitivity (DTH) was also assessed 1 day after adoptive T cell transfer into normal BALB/c mice by intradermal (i.d.) injection of 10 μg OVA in 10 μl PBS into the pinna of one ear (with PBS alone injected into the other ear as a negative control). Finally, a regimen of three i.p. injections of 100 μg OVAp in PBS at 5-day intervals was utilized in some experiments to induce immune tolerance within the BALB/c mice.

DO11.10 lymph node and spleen cells were harvested and treated with ACK lysis buffer (Biofluids, Rockville, MD), followed by anti-CD8 mAb 3.155 plus rabbit complement (Cedarlane, Westbury, NY) to deplete RBC and CD8⁺ T cells. Subsequently, ∼20 × 10⁶ OVA-specific KJ1-26⁺CD4⁺TCR-Tg T cells were injected i.p. into γ-irradiated (200 rad) nu/nu mice. The number of KJ1-26⁺ cells injected was calculated based on multiplying the percentage of KJ1-26⁺CD4⁺ cells measured by flow cytometry times the number of live cells purified. At days 15, 20, and 25 after T cell transfer, some nu/nu mice were injected i.p. with 100 μg OVAp to induce tolerance in the TCR-Tg cells. Negative control mice received PBS injections without Ag. At day 30 after transfer, the nu/nu mice were sacrificed, lymph node and spleen cells were isolated, and the KJ1-26⁺CD4⁺ cells were enumerated by flow cytometry. In some experiments, day 30 nu/nu mice were then primed s.c. with 100 μg OVAp in CFA at the base of the tail and sacrificed at day 40. Alternatively, nu/nu mice were primed first with OVAp in CFA at day 15, and then tolerized at days 25, 30, and 35 with i.p. OVAp injections before sacrifice at day 40. T cells isolated from these mice were subsequently transferred i.v. into unirradiated syngeneic BALB/c recipients, such that each host animal received 0.25 to 2 × 10⁶ KJ1-26⁺CD4⁺ donor cells in 0.5 ml PBS. Some of the donor cells were also used immediately for in vitro proliferation and lymphokine production assays.

Peripheral blood leukocytes, lymph node cells, or splenocytes were harvested, and 10⁶ cells were incubated on ice with biotinylated anti-clonotypic mAb KJ1-26 (31), followed by streptavidin Cy-Chrome (PharMingen, San Diego, CA) together with anti-CD4-phycoerythrin (Caltag, Burlingame, CA). After several washes, cells were either immediately analyzed or were fixed in 0.5% paraformaldehyde before the acquisition of 10,000 lymphocyte-gated events using forward and side scatter. Isotype-matched irrelevant mAb dye conjugates served as negative controls in all experiments. Flow cytometry was performed using a FACScan and CellQuest Software (Becton Dickinson, Mountain View, CA).

For intracellular IL-2 detection, peripheral lymph node (inguinal, axillary, brachial, mesenteric, and periaortic) and spleen cells were harvested 3 h after OVAp injection i.v., and 5 × 10⁶ cells were immediately incubated on ice with anti-FcRγ mAb 2.4G2 (American Type Culture Collection, Manassas, VA) plus 10% rat serum (Sigma, St. Louis, MO), followed by anti-CD4-FITC (Caltag) and the biotinylated KJ1-26 mAb, and then streptavidin Cy-Chrome. Cells were fixed in 2% formaldehyde (Fisher, Pittsburgh, PA) and permeabilized in 0.5% saponin (Sigma) before staining with phycoerythrin-conjugated rat anti-murine IL-2 mAb (PharMingen), as described (32). Following three washes, 10,000 lymphocyte-gated events in addition to 1,000 KJ1-26⁺CD4⁺-gated events were individually collected. Results indicate the percentage of T cells with increased IL-2-phycoerythrin staining in 1,000 cells from the KJ1-26⁺CD4⁺ subpopulation. Data are expressed as the mean ± SEM for replicate animals.

BALB/c mice were bled by retroorbital puncture, and sera collected both before immunization (preimmune) and 10 to 12 days after primary or secondary Ag challenge. Sera were stored at −20°C, and Ab titers were determined by ELISA using DNP-BSA as the capture Ag. DNP-specific IgG1 and IgG2a were detected using isotype-specific, horseradish peroxidase-labeled, goat anti-mouse IgG1 and anti-mouse IgG2a Abs, respectively (Southern Biotechnology, Birmingham, AL). The Ab titer is calculated based on that dilution of a serum that has a concentration of anti-DNP capable of eliciting a one-fourth maximal OD within the peroxidase assay. Titers shown typically represent the geometric mean ± SEM of replicate animals.

Pinna thickness was determined for both ears using an engineer’s micrometer both before and 24 h after i.d. injection of 10 μg OVA into one ear and PBS into the opposite ear. The DTH response was determined as the difference between the pre- and postinjection measurements. Data shown are the mean ± SEM of replicate animals.

Five days after immunization with OVAp in CFA, animals were sacrificed, and draining lymph nodes (inguinal and periaortic) were harvested. The total viable lymph node cell number was determined by trypan blue exclusion microscopy. Lymph node cells were also analyzed by flow cytometry to detect the presence of KJ1-26⁺CD4⁺TCR-Tg T cells, as described above. The total number of KJ1-26⁺CD4⁺ T cells recovered was calculated by multiplying the percentage of KJ1-26⁺CD4⁺ cells times the total viable lymph node cell count. Data shown represent the mean ± SEM of replicate animals.

Harvested lymph node cells from nu/nu mice were cultured with 5 × 10⁵ irradiated (3000 rad) normal BALB/c splenocytes and 1 μM OVAp for 48 h in 200 μl complete medium (1:1 Eagle’s Hanks’ amino acids (EHAA) medium (Biofluids) and RPMI 1640 (Celox, Hopkins, MN) containing 10% FCS (HyClone, Logan, UT), 2 mM l-glutamine, penicillin, gentamicin, and 5 × 10⁻⁵ M 2-ME). T cells were then pulsed with 0.25 μCi of [³H]thymidine (Dupont-NEN, Boston, MA) for 12 to 16 h, followed by harvesting of the cells on a PHD cell harvester (Cambridge Technology, Cambridge, MA) and liquid scintillation counting of the samples to determine the level of thymidine incorporation.

IL-2 secretion into the supernatant was monitored by bioassay at 48 and 72 h of stimulation using CTLL-2 cells (33) that express a rBcl-x molecule (kindly provided by Dr. C. Thompson, University of Chicago, Chicago, IL), as previously described (34). IFN-γ secretion during the experiment was determined by ELISA using PharMingen capture and detection mAbs. Lymphokine production results are expressed as the amount secreted per OVA-specific KJ1-26⁺CD4⁺ T cell present in the culture.

Data are expressed as the mean ± SEM of replicate animals. In some cases, experimental treatment groups were compared using Student’s paired or unpaired t tests.

Previously, it had been observed that an immune tolerance to the chicken OVA peptide 323–339 (OVAp) develops in animals following i.v. exposure to the Ag in PBS because of the induction of T cell clonal anergy (26, 35). To prove this, some of these experiments have relied on an adoptive transfer of OVA-specific CD4⁺TCR-Tg DO11.10 T cells into normal BALB/c mice before the Ag infusion, and the subsequent detection of these OVA-reactive T cells with the KJ1-26 anti-clonotypic mAb (26, 30, 31). In the experiments described in this work, adoptive transfer of DO11.10-pooled lymph node and spleen cells into lightly irradiated T cell-deficient BALB/c nu/nu mutant mice was performed to allow for a detailed analysis of the effects of tolerance induction on the effector functions of a relatively large population of Ag-specific T cells in vivo (see illustrations, Fig. 1). Within 15 days of i.p. transfer, significant numbers of clonotype-positive (KJ1-26⁺) CD4⁺ T cells spontaneously left the peritoneum and appeared in the peripheral blood (Fig. 2, A and B). Recipient nu/nu mice were then challenged i.p. with OVAp in PBS three times at 5-day intervals beginning on day 15 after transfer (3×OVAp pretreatment) to induce tolerance within this TCR-Tg OVA-responder population. Exposure to aqueous 3×OVAp i.p. did not significantly alter the percentage of KJ1-26⁺CD4⁺ T cells observed in either lymph nodes (Fig. 2 C) or spleen (data not shown) on day 30 after transfer, and the total number of KJ1-26⁺CD4⁺ cells recovered from the 3×OVAp-treated mice averaged 8.5 ± 1 × 10⁶ vs 8.7 ± 1.1 × 10⁶ in 3×PBS-treated control mice (n = 10). Thus, an overall decrease in OVA-responder cell frequency as a result of peripheral deletion (36, 37, 38) did not accompany the exposure of T cells to aqueous peptide administration in the nu/nu mice.

Recovered 3×OVAp-pretreated KJ1-26⁺CD4⁺ T cells did, however, demonstrate decreased proliferative responsiveness to OVAp rechallenge in vitro (Fig. 3,A), as well as a markedly reduced capacity to secrete their autocrine growth factor IL-2 (Fig. 3, B and C). In nine separate experiments, IL-2 production (corrected for the number of KJ1-26⁺CD4⁺ T cells present) in cultures from 3×OVAp-treated nu/nu recipient animals was inhibited an average of 93 ± 2% compared with control 3×PBS-treated animals. This failure to detect the production of IL-2 by 3×OVAp-treated T cells was not simply the consequence of a change in the kinetics of secretion, as determined by time-course studies (data not shown); furthermore, T cells recovered from nu/nu animals immunized s.c. with OVAp in CFA retained the capacity to secrete IL-2 (Fig. 3 C). Thus, exposure of CD4⁺TCR-Tg T cells in nu/nu mice to repeated aqueous peptide-Ag infusions resulted in clonal anergy induction.

It previously has been observed that the s.c. immunization of DO11.10 TCR-Tg T cells in BALB/c adoptive transfer animals with OVAp in CFA results in the differentiation of these cells to an IFN-γ-producing Th1 phenotype (39, 40). Consistent with this, increased IFN-γ production was observed in cultures containing KJ1-26⁺CD4⁺ T cells recovered from nu/nu mice after OVAp/CFA immunization in response to OVAp rechallenge in vitro (Fig. 3 C). Cells from the 3×OVAp i.p.-pretreated group, on the other hand, failed to secrete a significant quantity of IFN-γ into the culture medium during reexposure to Ag. Therefore, clonal anergy was induced in naive T cells in the absence of a differentiation to a Th1-like phenotype. In addition, no evidence for clonal diversion to a Th2-like phenotype was obtained in these experiments, since 3×OVAp-treated T cells also failed to develop the capacity to secrete either IL-4 or IL-5 (data not shown).

To test whether systemic exposure to aqueous peptide-Ag precludes later differentiation of an anergic T cell to a helper phenotype, 3×OVAp-pretreated adoptive transfer nu/nu mice were challenged s.c. on day 30 with OVAp in CFA (Fig. 1,B). When recovered from these animals 10 days later (on day 40), KJ1-26⁺CD4⁺ cells demonstrated little improvement in their capacity to secrete IL-2 (13% of control 3×PBS-pretreated T cells; p = 0.025). On the other hand, IFN-γ production in these cultures in response to Ag restimulation was now greatly augmented and equal to that of a normal OVAp/CFA-primed helper cell population (Fig. 3,C). This result suggested that the development of clonal anergy in a T cell in vivo cannot prevent its differentiation to a Th1-like phenotype, and indicated that defective IL-2 production can coexist with intact IFN-γ gene inducibility. In support of this second hypothesis, the induction of clonal anergy with 3×OVAp infusions i.p. in animals previously primed s.c. with OVAp in CFA (Fig. 1,C) also failed to interfere with the secretion of IFN-γ in vitro by the helper cells, even though IL-2 gene inducibility in the KJ1-26⁺CD4⁺ T cells was reduced 97% relative to animals treated only with OVAp in CFA s.c. (p = 0.014) (Fig. 3 C).

Our experiments using TCR-Tg T cells and nu/nu adoptive transfer mice confirmed that the repeated exposure of naive CD4⁺ T cells to aqueous peptide Ag in vivo leads to a down-regulation of IL-2 gene inducibility and reduced proliferative responsiveness, as measured by in vitro thymidine incorporation. This predicted that the induction of clonal anergy would also result in T cells with reduced capacity for clonal expansion in vivo. We investigated this possibility by pretreating TCR-Tg T cells in the nu/nu mice either with 3×PBS i.p. or the aqueous 3×OVAp i.p. infusions, and then transferring these T cells into normal naive BALB/c recipients. These adoptively transferred BALB/c mice were subsequently immunized s.c. either with CFA alone (as a negative control) or with DNP-OVA in CFA to test the capacity of the OVA-reactive T cells to undergo a clonal expansion during the next 5 days. As illustrated in Figure 4 A, KJ1-26⁺CD4⁺ T cells exposed to the aqueous peptide-Ag pretreatment (3×OVAp) did indeed demonstrate defective clonal expansion and accumulation within the draining lymph nodes, with only about one-quarter the increase in cell number as that seen with normal T cells transferred from PBS-treated nu/nu mice.

Interestingly, draining lymph nodes in mice transferred with control (3×PBS i.p.-pretreated) T cells and then primed s.c. with DNP-OVA/CFA contained more than twice as many leukocytes (almost entirely KJ1-26⁻) as lymph nodes from similar animals immunized with CFA alone 5 days earlier. This DNP-OVA-dependent increase in total cell number, however, was not observed when the transferred responder T cells came from tolerant (3×OVAp i.p.-pretreated) nu/nu mice (Fig. 4 B). Thus, the induction of clonal anergy in vivo in response to aqueous peptide-Ag exposure led to a stable (for at least 5 days) impairment in clonal expansion by Ag-specific responder T cells, as well as to a defective capacity to recruit other (Ag-nonspecific) leukocytes into the draining lymph node.

Defective clonal expansion in this system may reflect either decreased production of or responsiveness to a growth factor such as IL-2, or increased death in the Ag responders. Tolerant T cells survived well in BALB/c adoptive transfer recipients for 5 days in the absence of Ag stimulation (Fig. 4,A). Likewise, reexposure of 3×OVAp-treated nu/nu mice to s.c. OVAp/CFA stimulation failed to result in a significant reduction in the number of anergic KJ1-26⁺CD4⁺ T cells recovered (data not shown). IL-2 secretion is likely to contribute to the development of the autocrine growth response, and may also act as a cell-survival factor by regulating the death-repressor proteins Bcl-2 and Bcl-x_L (34, 41). On the other hand, IL-2 production during an Ag response also appears to prime T cells for activation-induced cell death (37); therefore, reduced IL-2 production by anergic T cells may actually protect them from apoptosis. To address these issues, we examined intracytoplasmic IL-2 levels by flow cytometry in KJ1-26⁺CD4⁺ T cells ex vivo following i.v. infusion of OVAp. No differences in cell survival after Ag stimulation were noted between the 3×PBS and 3×OVAp pretreatment groups (data not shown). Instead, it was observed that pretreatment of the TCR-Tg T cells in nu/nu mice with 3×OVAp i.p. virtually eliminated their subsequent capacity to synthesize IL-2 upon i.v. OVAp rechallenge in the BALB/c adoptive transfer recipient mice (Fig. 5, A and B). Therefore, poor clonal expansion by anergic T cells in vivo correlated best with defective IL-2 production.

As predicted by the in vitro IL-2 production results, tolerant 3×OVAp-pretreated T cells subsequently challenged s.c. with OVAp in CFA within the nu/nu mice also failed to regain the capacity to accumulate intracytoplasmic IL-2 after adoptive transfer into BALB/c recipients (Fig. 5, C and D). Likewise, T cells initially primed s.c. with OVAp/CFA and subsequently treated with 3×OVAp infusions i.p. to induce tolerance within the nu/nu mice, failed to accumulate intracytoplasmic IL-2 when adoptively transferred to normal BALB/c mice and rechallenged with OVAp (data not shown). Taken together with the observed deficiency in clonal expansion, these IL-2 data suggested that peptide-induced clonal anergy leaves both naive and effector T cells defective for Ag-induced autocrine growth in vivo. Furthermore, they indicated that exposure of tolerant T cells to peptide Ag under immunogenic conditions neither reverses the anergy nor allows for a significant outgrowth of normal cells that may have escaped tolerance.

In the next series of experiments, this nu/nu mouse adoptive transfer system was used to investigate the effects of clonal anergy induction on the capacity of a population of TCR-Tg T cells to perform Th1-like helper functions in vivo. Normal BALB/c mice were again used as final adoptive transfer recipents of TCR-Tg cells initially parked in nu/nu mice, and these animals were then tested for Th cell activity in the production of DNP-specific anti-hapten IgG2a Abs. An anti-DNP response was utilized because DNP-specific B cells are present in naive BALB/c animals at relatively high frequency; therefore, no priming of the B cell repertoire was necessary to elicit anti-hapten IgG2a responses when adequate T cell help was available (42). The s.c. challenge of normal BALB/c animals with the hapten-carrier conjugate DNP-OVA in CFA was observed to elicit significant serum titers of anti-DNP IgG2a only when these mice had first been primed s.c. with OVAp in IFA to increase the frequency of OVA-reactive T cells (Fig. 6 A). Thus, anti-DNP IgG2a production in these BALB/c mice demonstrated the expected requirement for a primed and clonally expanded Th cell population, presumably of a Th1-like phenotype (43, 44).

Using the nu/nu mouse T cell tolerance model and secondary adoptive transfer of TCR-Tg T cells into normal BALB/c mice, we next assessed the capacity of clonal anergy induction to block the development or delivery of effector cell activities by CD4⁺ Th cells. Naive KJ1-26⁺CD4⁺ T cells recovered from control 3×PBS i.p.-pretreated nu/nu mice were found to be competent to provide helper activity for the production of anti-DNP IgG2a in a dose-dependent fashion following their adoptive transfer into the normal BALB/c mice and subsequent s.c. immunization with DNP-OVA in CFA (Fig. 6,A). The ability of these naive T cells to help in the production of anti-DNP IgG2a in vivo apparently reflected their capacity to differentiate into potent Th1-like effector cells during the 10- to 12-day course of the DNP-OVA/CFA priming, since this population of cells displayed an undifferentiated lymphokine production profile in vitro and demonstrated no capacity to act as helpers in a 48-h DTH reaction in vivo (Fig. 3,C, and data not shown). Remarkably, KJ1-26⁺CD4⁺ T cells recovered from tolerant nu/nu mice (3×OVAp i.p. pretreated) were at least as potent as the control naive cells in providing help for IgG2a anti-DNP Ab (Fig. 6, A and B). In fact, immunization of BALB/c adoptive transfer recipients with a low (10 μg) dose of DNP-OVA in CFA revealed that the tolerant 3×OVAp-pretreated T cells could be a more potent helper population than either 3×PBS-pretreated naive T cells, or Ag-experienced TCR-Tg T cells that had initially been primed in the nu/nu mice with OVAp in CFA (Fig. 6 C). Therefore, under controlled conditions in which T cells were shown to be anergic at the level of the IL-2 gene, this peptide-induced tolerance mechanism failed to eliminate the delivery of Th signals for the production of IgG2a in vivo.

To determine whether this failure of in vivo tolerance induction to regulate T cell participation in the production of the anti-DNP IgG2a represents a general resistance of Th1 effector functions to the effects of clonal anergy, T cells were also examined for their ability to direct a DTH response after anergy induction. As expected, naive KJ1-26⁺CD4⁺ T cells recovered from 3×PBS i.p.-pretreated control nu/nu mice and transferred into normal BALB/c recipients were incapable of inducing DTH in response to i.d. ear injection with OVA (Fig. 7). Exposure of the T cells to 3×OVAp i.p. administration in the nu/nu mice to induce tolerance also failed to result in the development of a capacity to mediate DTH. This was consistent with the finding that these cells also demonstrated little ability to secrete IFN-γ (Fig. 3 C). In contrast, s.c. immunization with OVAp in CFA induced the generation of a potent Th1 effector cell population in normal untransferred BALB/c mice that was capable of eliciting a DTH reaction to OVA (data not shown).

These results indicated that the stimulation of KJ1-26⁺CD4⁺ T cells with OVAp under anergy-inducing conditions either was insufficient to induce differentiation to a DTH-effector phenotype, or resulted in a block of any newly acquired DTH-effector activity. To begin to distinguish between these two possibilities, TCR-Tg T cells in nu/nu animals were first made anergic with 3×OVAp injections i.p., or were injected i.p. with 3×PBS alone as a control. The nu/nu mice were then immunized s.c. on day 30 with OVAp in CFA to induce the differentiation of OVA-specific DTH Th cells (Fig. 1,B). As expected, positive control 3×PBS-pretreated KJ1-26⁺CD4⁺ T cells recovered on day 40 from these OVAp/CFA s.c. immunized nu/nu mice were found to be fully capable of performing a DTH response within 24 h after adoptive transfer into naive BALB/c mice and subsequent ear challenge with OVA (Fig. 7). T cells pretreated with 3×OVAp i.p. also demonstrated equally strong dose-dependent DTH reactivity if primed s.c. with OVAp/CFA before adoptive transfer into the BALB/c mice (Fig. 7), consistent with their acquired ability to secrete IFN-γ (Fig. 3,C), and despite a persistent inability to produce IL-2 (Fig. 5, C and D). These same T cells were also potent Th cells for the production of anti-DNP IgG2a (Fig. 6 C). The results indicate that exposure of naive T cells to aqueous peptide Ag in the absence of infection or adjuvant is not sufficient to cause their differentiation into DTH effectors, and instead induces clonal anergy at the level of the IL-2 gene. Furthermore, these experiments suggest that the development of peptide-induced tolerance in vivo cannot interfere with the capacity of anergic T cells to subsequently develop a Th1-like effector phenotype if exposed to Ag in adjuvant, even though they never regain the ability to secrete IL-2.

The finding that differentiation to an IgG2a- and DTH-helper phenotype could occur even in tolerant T cells implies that clonal anergy induction in vivo does not regulate Th1 effector function per se. To formally test this, KJ1-26⁺CD4⁺ T cells were first primed s.c. in nu/nu mice with OVAp in CFA to induce the differentiation of Th1 effector cells. Animals were then either treated beginning on day 25 with 3×PBS injections alone, or given 3×OVAp injections i.p. to induce T cell tolerance (Fig. 1,C). Indeed, the capacity of OVAp/CFA-primed KJ1-26⁺CD4⁺ T cells to promote the development of OVA-dependent DTH in BALB/c recipients was not found to be adversely affected by three infusions of OVAp i.p. before the final adoptive transfer (Fig. 7). Likewise, these same primed/tolerant T cells remained fully competent to mediate an anti-DNP IgG2a response (Fig. 6 C), despite the fact that they had lost the capacity to accumulate intracytoplasmic IL-2 (data not shown).

The development of an efficient peripheral immune system depends on the establishment of tolerance to tissue-specific self Ag, while retaining reactivity to foreign pathogens, a process that cannot rely solely on the clonal deletion of autoreactive T cell precursors within the thymus. It follows, then, that the onset of an autoimmune disease such as multiple sclerosis or rheumatoid arthritis in an individual may represent either a specific breakdown of such self-tolerance to a particular Ag, or may result from a more global defect in the development and/or maintenance of self-tolerance in the periphery. The experiments shown in this study aimed to establish the mechanism of T cell tolerance induced in the periphery of adult mice following aqueous peptide-Ag infusion, in an attempt to define those factors likely to be important in restoring self-tolerance to tissue-specific Ag in patients with autoimmunity.

Numerous reports have indicated previously that a systemic exposure of the mature peripheral immune system to deaggregated forms of a protein Ag or simple peptides, in the absence of adjuvant or infection, fails to productively engage the immune system and instead reduces the individual’s responsiveness to that Ag (45, 46, 47, 48, 49). Recently, several studies have taken advantage of TCR-Tg technology to probe the nature of this Ag unresponsiveness under conditions in which Ag responders can be carefully monitored. In one of these experiments, i.v. exposure of CD8⁺LCMV glycoprotein-specific T cells to a peptide derived from that Ag led to a tolerance as a result of the elimination of the majority of Ag responders and anergy induction within the survivors (36). Similarly, CD4⁺OVA-reactive T cells underwent an abortive clonal expansion following i.p. infusion of peptide Ag, and surviving cells were shown to have an IL-2 production defect in vitro, consistent with the induction of clonal anergy (26). This combination of activation-induced cell death and clonal anergy induction would be expected to hinder T cell clonal expansion in response to a later exposure to the Ag, and both most likely contribute to peripheral immune tolerance.

We speculate that naive T cells responding to aqueous peptide Ag undergo an abortive differentiation as a result of TCR stimulation in the presence of a relatively low level of B7 molecule expression. While insufficient to induce T cell proliferation or the development of DTH-helper activity, perhaps because of the counter-regulatory influences of simultaneous CTLA-4 ligation, this activation event does appear to leave the T cell susceptible to clonal anergy (39, 50). Upon further TCR occupancy in the absence of infection or adjuvant, the capacity of the T cells to produce IL-2 is progressively lost (Fig. 8).

We have examined the ability of repeated i.p. peptide-Ag administration to influence the capacity of CD4⁺ T cells to participate in the production of anti-hapten IgG2a Ab, and to mediate DTH, by utilizing a modification of the TCR-Tg adoptive transfer system originally pioneered by Kearney and Jenkins (26). In this new model, TCR-Tg T cells were first adoptively transferred into T cell-deficient nu/nu mice, and then were exposed to peptide Ag, either under immunogenic conditions or in a fashion capable of inducing T cell tolerance. This system of adoptive cell transfer afforded us the opportunity to recover peptide-affected T cells at high number for use in various functional assays. Using an anti-clonotypic mAb to enumerate Ag-specific TCR-Tg T cells after the peptide treatments, the study confirmed that systemic peptide Ag administration in the absence of adjuvant fails to induce either a clonal expansion of the Ag-reactive T cells or their deletion. Closer examination of the T cells recovered from the peptide-treated nu/nu mice also identified the development of a profound IL-2 production defect in surviving Ag responders, despite continued high level expression of both CD4 and the Ag-specific TCR. This loss of IL-2 inducibility was not accompanied by a diversion to the Th2 phenotype (51), since these peptide-treated T cells never gained the ability to produce IL-4 or IL-5 (E. Malvey, unpublished observation). Therefore, our results suggest that clonal anergy is induced in T cells after the series of i.p. injections with peptide Ag in PBS.

Peptide-induced anergy, if similar to in vitro models of Ag unresponsiveness, might be expected to inhibit the clonal expansion of Ag-reactive cells in vivo because of a block in signal transduction between the TCR and the IL-2 gene. Peptide-pretreated KJ1-26⁺CD4⁺ T cells did, in fact, demonstrate only poor clonal expansion in vivo in response to Ag rechallenge, following adoptive transfer into normal BALB/c recipients. Additionally, these tolerant T cells were incapable of recruiting other Ag-nonspecific leukocytes into the draining lymph node during the course of the Ag response. Consistent with these data, the in vivo induction of IL-2 synthesis within the tolerant T cells was found to be markedly defective. These results argue conclusively against a role for either peripheral elimination or suppression in this TCR-Tg immune tolerance model. While it is currently uncertain which factors are most responsible for the development of a strong clonal expansion response or recruitment of bystander cells in vivo, the results of these experiments suggest that tolerance develops because the elaboration of at least one critical factor by the responding T cells is blocked as a consequence of the induction of clonal anergy.

Does the induction of clonal anergy in vivo result in a complete functional inactivation of the T cell? The evidence suggests otherwise. First, peptide-pretreated T cells appeared fully competent to participate in the production of anti-hapten IgG2a after challenge with the hapten-carrier protein in adjuvant. Cognate help for IgG2a isotype switching is thought to rely both on the delivery of the CD40 ligand as well as on the production of IFN-γ by Th cells responding to carrier-derived peptide Ag presented on the surface of the hapten-specific B cells (52, 53). Therefore, the production of high titer anti-DNP IgG2a in animals transferred with high numbers of peptide Ag-pretreated TCR-Tg cells implies that these T cells could be induced to differentiate to a Th1-like phenotype upon exposure to DNP-OVA in adjuvant, and provide help to DNP-specific B cells during the course of the response. This helper activity was not simply the result of an outgrowth of T cells that had escaped tolerance induction, nor did it result from a reversal of the anergy: after transfer into the BALB/c recipients, 3×OVAp i.p.-pretreated T cells maintained an anergic phenotype based on their weak ability to undergo a clonal expansion. Furthermore, aqueous peptide Ag-pretreated T cells intentionally primed s.c. with OVAp in CFA within the nu/nu mice never regained a normal ability to produce IL-2 in vivo.

Our own in vitro studies have also recently addressed this issue and determined that under conditions in which T cell numbers were held constant, the delivery of Th signals for the induction of B cell polyclonal Ab production was only partially sensitive to clonal anergy induction (25). The secretion of IL-4 and IL-5 by anergic T cells responding to Ag on B cells was found to be moderately depressed in these long-term tissue culture experiments. Nevertheless, CD40 ligand was found to be inducible and fully functional after activation of anergic T cells, and IFN-γ could still be secreted in a normal fashion, consistent with these in vivo results.

Additional evidence that anergic T cells remain functionally responsive to Ag stimulation includes the finding that aqueous peptide-Ag-pretreated TCR-Tg T cells also developed the capacity to direct a DTH response following priming s.c. with Ag in adjuvant within the nu/nu mice. Neither naive cells nor T cells exposed to the aqueous peptide Ag pretreatment alone had such a helper capacity. The failure of 3×OVAp i.p.-pretreated T cells to directly help in the development of a DTH response is consistent with their inability to secrete IFN-γ, and may reflect insufficient IL-12 production by APC or the effects of CTLA-4 ligation during the course of the clonal anergy induction (40). Regardless, our data provide no support for the notion that clonal anergy induction diverts differentiation away from the Th1 phenotype, as anergic T cells responded to Ag stimulation in the presence of adjuvant-induced costimulatory signals with the development of Th1 helper activity.

Finally, these experiments demonstrate that clonal anergy at the level of the IL-2 gene can coexist in vivo with intact Th1 effector function. Specifically, we assessed the capacity of primed Th1-like memory cells to be made tolerant to Ag upon repeated i.p. exposure to the OVAp. It was shown that these Th1 cells lost the capacity to synthesize IL-2 following the 3×OVAp i.p. injections, consistent with the development of clonal anergy. Nevertheless, such tolerant T cells retained their capacity to immediately induce a DTH reaction, as well as participate in the production of anti-hapten IgG2a. Thus, primed effector cells are indeed susceptible to anergy induction at the level of the IL-2 gene, but this clonal anergy tolerance mechanism is incapable of interfering with the activation of T cells for the delivery of Th1-dependent activities important to the development of DTH and IgG2a Ab responses.

These findings suggest that the most important role for clonal anergy in the development of peripheral T cell tolerance may be the regulation of autocrine growth-factor production and clonal expansion, with the resultant maintenance of potentially deleterious T cells at low frequency. In pathologic states in which autoreactive T cells presumably reach high frequency, this mechanism may no longer have the inherent capacity to control their actions. Similarly, the design of future therapies for autoimmune disease based on this tolerance mechanism should anticipate that clonal anergy may not inhibit the delivery of all potentially dangerous Th1 activities.

It should be noted that Gilbert et al. (54) previously determined that the delivery of help for IgG production in vitro is sensitive to the induction of T cell clonal anergy. Likewise Finck et al. (55) demonstrated a capacity of the B7/CD28 antagonist CTLA-4Ig to interrupt the production of pathogenic autoantibodies and improve survival in (NZB/NZW)F₁ lupus-prone mice, a therapy perhaps predicted to promote clonal anergy induction in the self Ag-specific Th cells. Both experiments, however, lacked the power to establish whether the development of tolerance stemmed from a reduced expansion of critical Ag-specific helper cells or from an ability of the tolerance mechanism to interrupt the delivery of Th activities.

Finally, Marusic and Tonegawa (56) recently showed that a single i.p. injection of peptide Ag in IFA induced clonal anergy within a large population of TCR-Tg T cells specific for myelin basic protein. In their experiments, anergy was associated with a defect in IFN-γ production, and was in fact sufficient to protect the mice from experimental autoimmune encephalomyelitis. This apparent inconsistency with the findings of our system may stem from their use of IFA in the single peptide Ag infusion, since systemic Ag persists considerably longer after i.p. injection with IFA than with PBS (M.K.J., unpublished observation). The continuous presence of nonimmunogenic peptide most likely desensitizes T cells to subsequent Ag stimulation, and may promote the appearance of a more profound functional anergy.

We are grateful to Matthew Mescher and Kristin Hogquist for their helpful discussions and reading of the manuscript. We also thank Elizabeth Kearney and Kathy Pape for their critical technical advice regarding the T cell receptor-transgenic adoptive transfer system.