In Vivo CD4⁺ T Cell Tolerance Induction Versus Priming Is Independent of the Rate and Number of Cell Divisions¹

While the immune system must be able to respond to pathogens, an equally great challenge is to remain nonresponsive to self-tissues that contribute the vast majority of Ags that the immune system encounters. For T lymphocytes, cells expressing Ag receptors (TCRs) recognizing ubiquitously expressed Ags are eliminated during negative selection in the thymus (1, 2). However, additional mechanisms are needed to tolerize T cells that recognize nonthymic, peripherally expressed Ags (3).

Peripheral T cell tolerance can be mediated by a variety of mechanisms that range from physical elimination of the self-reactive T cell (4, 5, 6) to ignorance of the cognate Ags (7, 8). Between these two extremes, T cells can be desensitized to antigenic stimulation by down-regulation of the TCR (9) or accessory molecules (i.e., CD8) (10). Additionally, T cells can be functionally inactivated or anergized (11, 12, 13, 14). However, it is not clear which mechanism of peripheral tolerance will be used for a given self-Ag, although a number of factors such as cellular localization (compare Refs. 7 and 8 to 12 and 15), tissue site (16, 17), and level of expression (18) have been shown to be important variables affecting tolerance induction.

The pathways leading to the various forms of tolerance also remain to be completely elucidated. In the case of physical elimination, an initial phase of activation and cell proliferation appears to precede cell death (19, 20, 21, 22, 23). This process of clonal exhaustion appears to be mediated via the Fas-Fas ligand pathway (24, 25). While T cell proliferation appears to be involved in the tolerization pathway leading to elimination, it is not clear whether proliferation plays a role in the induction of anergy. In vitro studies using CD4⁺ T cell clones have indicated that anergy can be induced by TCR occupancy in the absence of costimulation (26). In response to this tolerizing stimulus, the T cells either proliferate weakly (27) or not at all (28, 29), suggesting that anergy induction is not linked to mitosis. In support of this notion, the observation that T cell clones could be rendered anergic when stimulated with APCs expressing costimulatory activity when IL-2-dependent proliferation is blocked suggested that the absence of a strong proliferative response immediately following Ag recognition results in anergy rather than activation (30). One hypothesis to explain this finding was that an intracellular inhibitor of lymphokine gene activation was induced upon TCR occupancy and a requisite number of cell divisions was required to dilute the inhibitor and allow for lymphokine production characteristic of an activated CD4 cell. An insufficient number of divisions maintained high enough concentrations of the inhibitor to produce a state of anergy (30). In vivo, the relationship between mitosis and anergy induction has not been as well established. As a population of anergic T cells often remain following clonal exhaustion (19, 20, 21), anergy might represent an intermediate step in a tolerization pathway in which proliferation precedes deletion (31). Alternatively, a distinct subpopulation of Ag-specific T cells might undergo proliferation followed by deletion, while another might be induced into an anergic state without undergoing a significant proliferative response. While it has been documented that CD4 cell division in vivo can precede the induction of a tolerant phenotype that resembles anergy (32, 33), it is not clear whether the rate of mitosis induced under these conditions might fall below a critical threshold required to induce effector/memory function rather than tolerance.

We have recently described a transgenic mouse system in which the expression of low levels of the model Ag hemagluttinin (HA)⁵ in a variety of nonlymphoid organs results in the induction of tolerance in a clonotypic population of adoptively transferred HA-specific CD4⁺ T cells. The mechanism of tolerance is consistent with anergy in that a significant number of nonresponsive T cells persist that express normal levels of TCR and CD4 molecules. These T cells are not tolerized through direct interaction with HA-expressing parenchymal cells, but rather by bone marrow-derived APCs that have acquired and presented the class II-restricted HA epitope in a toleragenic manner (34). In the present study, we first compared the tolerance induction pathway in the same CD4⁺ T cell population when adoptively transferred into transgenic mice expressing different levels of peripheral HA. High levels of transgenic HA expression induce an expansion in the clonotypic CD4 cell population that subsequently become functionally tolerant, while low levels of transgenic HA expression induce tolerance without clonal expansion. The fluorescent marker 5,6-carboxyfluorescein diacetate succinimidyl ester (CFSE) (35) was used to analyze the proliferative responses of the clonotypic CD4 cells during tolerance induction. Consistent with the observed clonal expansion, the clonotypic HA-specific CD4 cells proliferated vigorously upon transfer into transgenic mice expressing high levels of HA. Interestingly, the clonotypic cells also proliferated, albeit at a less rapid rate, in mice in which clonal expansion was not apparent (i.e., expressing low HA levels). In contrast, when clonotypic T cells were transferred into nontransgenic (NT) recipients and stimulated with a recombinant vaccinia virus-expressing HA, they underwent a similar proliferative response as T cells that had encountered HA as a self-Ag, but did not become tolerant. These results indicate that the context of Ag expression determines the functional outcome of Ag encounter (i.e., priming vs tolerance) by a mechanism that is independent of the mitotic rate induced upon TCR engagement.

The C3-HA transgene expression cassette has been previously described (34). In short, the HA gene derived from the influenza virus A/PR/8/34 (Mount Sinai strain) has been placed under the control of the rat C3(1) promoter. Upon microinjection of the DNA into B10.D2 embryos (performed at the University of Michigan Biomedical Research Core Facilities, Ann Arbor, MI), two founder lines were established. The first, line 142 (C3-HA^low), contains approximately three intact copies of the transgene and expressed the C3-HA hybrid mRNA in many nonlymphoid tissues including the prostate and lung (34). The second line, 137 (C3-HA^high), contains ∼30–50 transgene copies and expresses HA mRNA in the same subset of tissues as the 142 line (data not shown).

The TCR transgenic mouse line 6.5 (36) (generously provided by H. von Boehmer, Institut Necker, Paris, France) that expresses a TCR recognizing an I-E^d-restricted HA epitope (¹¹⁰SFERFEIFPKE¹²⁰) was back-crossed nine generations onto the B10.D2 genetic background.

A total of 1 × 10⁶ cells were preincubated with the Fc-γ receptor blocking Ab 2.4G2 (HB-197; American Type Culture Collection, Manassas, VA). mAbs used for staining were: biotinylated anti-clonotypic TCR (6.5, provided by H. von Boehmer), avidin-PE, and FITC-conjugated anti-CD4 (CT-CD4, Caltag, Burlingame, CA). CyChrome-conjugated anti-CD4 (RM4-5) and FITC-conjugated anti-CD44 (IM7) were purchased from PharMingen (San Diego, CA). A total of 20,000 lymphocyte-gated events were collected on a FACScan (Becton Dickinson, San Jose, CA) for double stains (i.e., CD4 vs 6.5). For three-color stains, ∼1000 clonotype-positive events were collected.

Adoptive transfers were performed as previously described (34). In short, 2.5 × 10⁶ clonotypic CD4 cells, prepared from pooled lymph nodes (LN) of 6.5 transgenic mice, were resuspended in 0.2–0.5 ml sterile Hanks’ buffer (Life Technologies, Rockville, MD) and injected i.p. into unirradiated male recipient animals and retrieved from the spleen and LN 5–14 days posttransfer for functional analysis. For CFSE-labeling experiments, clonotypic CD4 were labeled using the previously reported protocol (35) with the following modifications. LN suspensions from 6.5 mice were resuspended in CTL media (RPMI medium (Life Technologies) with 10% FBS (HyClone, Logan, UT), 0.1 mM 2-ME, 2 mM l-glutamine, 1 mM sodium pyruvate, and 1× nonessential amino acids and penicillin/streptomycin solutions (Sigma, St. Louis, MO)) at 1 × 10⁷ cells/ml and incubated with 1 μM CFSE ((Molecular Probes, Eugene, Oregon) stored as a 5-mM stock in DMSO desiccated at −20°C) at 37°C for 10 min and then washed three times in Hanks’ buffer before adoptive transfer of 2.5 × 10⁶ clonotypic CD4 cells. NT recipients receiving recombinant vaccinia virus-expressing HA (vacc-HA) (37) were vaccinated with 1 × 10⁵ pfu virus i.p. 18 h after adoptive T cell transfer. CFSE-labeled T cells were retrieved from the spleen and LN 5 and 14 days posttransfer for FACS analysis.

A total of 1.5 × 10⁵ splenocytes or LN cells extracted from transfer recipients were incubated in round-bottom 96-well tissue culture plates with the indicated concentration of synthetic HA peptide. Seventy-two hours later, cultures were pulsed with 1 μCi [³H]thymidine and incubated an additional 24 h before harvest and determination of the amount of incorporated radioactive counts.

Tissues taken from transgenic mice were homogenized in PBS using a hand-held tissue homogenizer (Omni International, Gainesville, VA). Tissue extracts were then incubated in a 96-well plate in hybridoma media (IMDM (Life Technologies), 10% FCS, 1× essential and nonessential amino acids, 1 mM sodium pyruvate, 3 mM dextrose, 8 mM sodium bicarbonate, 0.1 mM 2-ME, and penicillin/streptomycin solutions (Sigma)) with 5 × 10⁵ B10.D2 splenocytes and 7 × 10⁴ hybridoma cells (derived from the 6.5 transgenic mice) that secrete IL-2 when stimulated with APCs presenting the 110-120 I-E^d-restricted HA epitope. IL-2 levels in media taken from overnight cultures were measured using an ELIZA kit for murine IL-2 as per the manufacturer’s instructions (Endogen, Woburn, MA). Data is expressed as OD units (read with a 450-nm filter). An OD value of 2.5 represents saturating IL-2 levels, while a value of 0.1 represents the background.

RT-PCR analysis indicated that the C3-HA transgenic founder lines 137 and 142 mice express HA mRNA in the same subset of organs that include the ventral prostate, dorsal lateral prostate, testis, vas deferens, penis, seminal vesicles, bulbourethral gland, lung, salivary gland, kidney, skeletal muscle, and heart (142, Ref. 34 and 137, data not shown). To quantitate the relative levels of HA protein expression in these mice, protein extracts prepared from positively expressing organs were incubated with NT splenocytes (APC source) and an HA-specific CD4⁺ hybridoma that secretes IL-2 when stimulated with the I-E^d HA epitope. Protein extracts from several organs derived from the 137 founder line induced significant IL-2 secretion by the hybridoma (Fig. 1). HA expression was highest in the lung and ventral prostate where 1000-fold dilutions of extract induced IL-2 levels above the detection limit of the assay. Intermediate and lower HA levels were observed in the dorsal lateral prostate, penis, testis, seminal vesicles, vas deferens, and bulbourethral gland. Other tissues that showed positive expression by RT-PCR, such as the heart, salivary gland, kidney, and skeletal muscle as well as serum did not produce detectable IL-2 levels, nor did any of the extracts prepared from the 142 founder line (data not shown). While the difference in total HA protein expression between the 137 (C3-HA^high) and 142 (C3-HA^low) founder lines could not be directly measured, in the lung and prostate the difference is at least 1000-fold.

Previously, we had shown that naive HA-specific CD4 cells (prepared from the TCR transgenic line 6.5 (36)) become functionally tolerant after adoptive transfer into unirradiated C3-HA^low mice. Although the clonotypic cells had neither expanded or contracted in number, they did exhibit alterations in the expression of the cell-surface markers CD44 and CD45RB, indicative of TCR engagement. Furthermore, they were functionally tolerant as demonstrated by their hyporesponsiveness to in vitro restimulation with APCs pulsed with HA peptide (34). To ascertain whether the level of peripheral HA expression affects CD4 cell tolerance induction, adoptive transfer experiments were performed in both the C3-HA^low and C3-HA^high animals. Consistent with our previous studies, clonotypic cells retrieved from the spleen of C3-HA^low mice did not exhibit either a significant expansion or contraction in number (relative to NT control recipients) 9 days posttransfer (Fig. 2,A). Nonetheless, they did exhibit increased expression of surface CD44, indicating that they had encountered their cognate epitope and lost their naive phenotype (Fig. 2,B). In contrast, clonotypic CD4 cells reextracted from the spleens of C3-HA^high recipients underwent a marked clonal expansion (Fig. 2,A). On average, the number of clonotypic cells was 11- and 2-fold greater than in NT control recipients in the spleen and LN, respectively (Fig. 2 C).

While the clonotypic CD4 cells appeared to have undergone different proliferative responses upon transfer into mice expressing different levels of peripheral HA, in both cases they were unable to proliferate when cultured with APCs pulsed with synthetic HA peptide (Fig. 3), thus demonstrating that they had been rendered tolerant.

The clonotypic expansion observed in the C3-HA^high mice is consistent with the induction of tolerance being linked with cell division. However, in the case of the C3-HA^low mice, it was not clear whether cell division was occurring. Although these T cells did exhibit altered surface expression of TCR engagement molecules (Fig. 2,B and Ref. 34) consistent with an activation phase, they had not undergone clonal expansion (Fig. 2). To address this issue, clonotypic cells were labeled with the fluorescent marker CFSE before adoptive transfer. As the progeny of a CFSE-labeled cell each retain half of the initial fluorescence, the profile of a cell populations fluorescence intensity provides a quantitative measurement of the strength of the proliferative response. Each peak on a FACS histogram plot represents a cell division that is one greater than the one immediately to its right (35). The CFSE-labeled clonotypic CD4 cells were retrieved 5 days posttransfer for analysis. While tolerance is not manifested until day 7 (data not shown), this time point provides a good picture of the initial response to Ag encounter. As expected, the clonotypic cells transferred into the C3-HA^high recipients underwent extensive cell division (Fig. 4,A). More surprisingly, in the C3-HA^low recipients in which clonal expansion was not evident (Fig. 2, A and C), the clonotypic cells also underwent significant division, albeit the rate of mitosis was less than in the C3-HA^high animals. The C3-HA^low recipients contained a lower frequency of clonotypic cells that had undergone more than six rounds of cell division and more that were undivided compared with C3-HA^high recipients (Fig. 4, B and C). Clonotypic T cells that were transferred into NT recipients and then stimulated with a vacc-HA (37) exhibited CFSE profiles intermediate to the C3-HA^high and C3-HA^low groups (Fig. 4, A–C). While the relative differences in proliferative responses observed between the various recipient groups were consistent in both the spleen and LN, overall proliferation was slightly greater in the spleen (Fig. 4, A–C), consistent with the increased frequency of clonotypic cells observed in the spleen relative to the LN in both the C3-HA^low and C3-HA^high recipients (Fig. 2 C). This difference may result from more exposure to Ag due to the higher ratio of APCs to CD4 cells in the spleen (5:1) than in the LN (2:1) (data not shown).

Fourteen days posttransfer, the vast majority of clonotypic CD4 cells had undergone extensive cell division in both the C3-HA^low and C3-HA^high recipients (Fig. 5, A–C). As tolerance is well established at this time point (Figs. 3 and 6), these results are consistent with the notion that cell division is part of the differentiation process leading to tolerance induction in naive CD4 cells.

This data indicates that division of the HA-specific CD4 cells precedes the induction of tolerance, suggesting that an inability to proliferate upon Ag recognition was not the determining factor in the differentiation of these T cells into a nonresponsive state. If tolerance induction was the result of weak proliferative responses, then one might predict that the clonotypic cells that had proliferated at a more rapid rate when transferred into the C3-HA^high recipients would be in a less tolerant state than those transferred into the C3-HA^low recipients. To the contrary, HA-specific CD4 cells retrieved from both HA transgenic recipients were equivalently tolerant 9 days posttransfer (Fig. 6,A). In contrast, at both 9 (Fig. 6,A) and 14 days (Fig. 6,B) posttransfer, the clonotypic cells that were stimulated in vivo with vacc-HA were able to incorporate [³H]thymidine quite efficiently during in vitro restimulation. Given that the clonotypic CD4 cells stimulated with vacc-HA did not become tolerant, despite having undergone a proliferative response that was similar if not slightly less robust than those that had encountered HA as a peripheral self-Ag at both early and later time points (Figs. 4 and 5), indicates that the kinetics of the proliferative response elicited upon Ag encounter does not determine whether the functional outcome is tolerance or priming.

The present study examines several issues related to the induction of peripheral tolerance to self-Ags in CD4⁺ T cells. The first addresses the effect of Ag concentration on tolerance induction. Naive clonotypic CD4 cells specific for a parenchymal Ag (HA) were induced into a tolerant state upon adoptive transfer into transgenic mice that express HA in the same subset of tissues, but differ in the level of HA expression by several orders of magnitude in organs such as the lung and prostate. Given our previous study, which established that the naive clonotypic CD4 cells encounter their cognate epitope on cross-presenting bone marrow-derived APCs rather than on the HA-expressing parenchyma (34), the relevant HA protein levels might be those released from tissues. Interestingly, HA protein was not detectable in the serum of either HA^low or HA^high mice. That the HA^high mice expressed HA protein levels that were 1000-fold over the detection limit of our assay, but had undetectable HA levels in the serum, suggests that only a small fraction of the total HA protein is released from the tissues into the circulation (subsequently to be taken up by cross-tolerizing APCs (tAPCs)). Alternatively, HA might not be directly released into the circulation, but rather might be taken up by the tAPC in the tissues. Regardless of the mechanism by which HA is acquired by tAPCs, given that HA is expressed in the same tissues in both transgenic founder lines, it would seem reasonable that the relative levels of HA peptide-MHC complexes on the surface of tAPCs is proportional to the level of HA protein in the parenchyma. Nonetheless, irrespective of the exact quantitative difference in the “relevant” Ag levels between these two mice, tolerance induction did not differ significantly.

Several previous studies have examined the effect of self-Ag concentration on CD8⁺ T cell tolerance. Transgenic expression of an allogenic class I MHC on an inducible liver-specific promoter induced either partial or complete TCR down-regulation on the cognate CD8 cell population when the Ag was expressed at low and high levels, respectively (18). When H-Y Ag was expressed on a relatively low proportion of hemopoietic cells, the cognate CD8⁺ clonotypic population were first activated, but soon disappeared. When H-Y Ag was expressed on a greater proportion of hemopoietic cells, the T cells were rendered anergic (38). CD8 cells specific for a pancreatic β cell Ag are ignorant to low levels of Ag, but undergo proliferation followed by deletion when the Ag is expressed at much higher levels (39). When HA is expressed at a relatively high level in pancreatic β cells, the cognate CD8 cells undergo an initial proliferative response followed by the induction of tolerance. In mice expressing only 2-fold higher levels of β cell HA, the HA-specific CD8 cells undergo both a more rapid proliferative response and induction of tolerance (40). In our system looking at CD4 cell tolerance induction, the level of peripheral Ag expression had a more subtle effect. In animals that differ in their level of peripheral Ag expression by at least 3 orders of magnitude, higher Ag levels elicited a more rapid mitotic rate (compare CFSE profiles of C3-HA^low and C3-HA^high recipients at day 5). However, CD4 cells exposed to low Ag levels eventually developed a similar CFSE profile as those exposed to higher Ag levels (compare CFSE profiles at day 14). In either case, a significant proliferative response preceded the establishment of tolerance. Taken together, both naive CD4 and CD8 cells might undergo an initial proliferative response upon encountering their cognate peripherally expressed self-Ags before the establishment of tolerance; however, the threshold level of Ag expression required to drive this response might be higher for CD8 than for CD4 cells. Given that naive T cells encounter their cognate self-epitopes on cross-presenting APCs (34, 41), the cross-tolerance pathway might be more efficient at presenting class II than class I epitopes.

Following tolerance induction via clonal exhaustion, there is often a population of residual T cells that are hyporesponsive to subsequent antigenic stimulation (i.e., anergic) (19, 20, 21). In our system, anergy induction as well as deletion might be occurring simultaneously. The presence of a population of clonotypic cells in HA-expressing mice, 9 to 14 days after adoptive transfer, that are hyporesponsive to restimulation is consistent with anergy. The evidence for deletion is more indirect. As the steady-state number of clonotypic T cells remained constant in the C3-HA^low recipients despite the vast majority of T cells having undergone multiple rounds of cell division, cell death might have been occurring at a rate that balanced cell division. Furthermore, the clonotypic cells expanded 11-fold in the spleen of C3-HA^high recipients, which could have occurred if each T cell had only divided three to four times. As virtually all of the clonotypic cells underwent at least six cell divisions, which could have produced a 64-fold or greater expansion, cell death might have limited the magnitude of the expansion in total cell number. Attempts to directly measure apoptosis using annexin V, which binds to the early cell-surface apoptosis marker phosphatidylserine (42), were unsuccessful (data not shown), possibly due to a combination of a short half life of apoptotic cells in vivo as well as the relatively small population size of the T cells being analyzed. Therefore, we cannot rule out the possibility that the less than expected clonal expansions observed in the lymphoid tissues was the result of a migration of a substantial fraction of the T cells into nonlymphoid organs. One way to resolve this issue might be to repeat this experiment using clonotypic CD4 cells prepared from TCR transgenic mice back-crossed onto a Fas-deficient (i.e., lpr) background. If Fas-dependent apoptosis does occur in this system, then Fas-deficient clonotypic T cells might undergo more robust clonal expansions that more closely resemble the hypothetical expansions predicted from the CFSE profiles. Thus, while future experiments will help to establish the relative contributions of anergy vs apoptosis in mediating tolerance induction in our system, the current study establishes that naive CD4 cells specific for a parenchymal Ag become functionally tolerant irrespective of the level of Ag expression.

This study also addressed the question of whether the rate of mitosis induced upon TCR engagement determines whether a T cell will become primed or tolerized. Previous in vitro studies of anergy induction using T cell clones have suggested that the strength of the proliferative response elicited upon TCR engagement is pivotal in determining functional outcome. Thus, conditions that hinder T cell proliferation induce anergy, while those promoting proliferation result in activation (30). More recently, it has been shown that toleragenic forms of Ag can induce an initial phase of T cell proliferation (32, 33); however, it was possible that a critical threshold in the rate of mitosis was not achieved for T cells to be rescued from tolerance induction. In the present study, the fluorescent marker CFSE was used to compare the rate of CD4 cell division elicited upon recognition of the cognate Ag expressed in either a toleragenic (i.e., self) or an immunogenic (i.e., viral) form. When the CD4 cells encounter HA expressed as a viral Ag through vaccination with a high titer of vacc-HA (i.e., 1 × 10⁷ pfu), they divide vigorously, resulting in ∼100% of the cells having undergone multiple rounds of cell division by day 5 (data not shown), and become primed (37). However, this strong proliferative response is not the critical determinant in inducing priming rather than tolerization as these T cells are not tolerized when stimulated with a lower titer of vacc-HA (i.e., 1 × 10⁵ pfu), which elicits a proliferative response that is similar if not slightly weaker than that elicited by self-HA expression. Thus, while cell division clearly precedes the establishment of tolerance to self-Ags, the context of Ag expression determines the functional outcome of TCR engagement (i.e., priming vs tolerance induction) by a mechanism that is independent of the induced rate of mitosis.

What then are the critical contextual factors that determine priming vs tolerance induction? While some in vivo studies have suggested that expression of the costimulatory ligand B7 on Ag-expressing cells results in T cell activation rather than tolerance (43, 44, 45), a more recent study provides evidence that B7-expressing cells can induce tolerance (46). In our system, tolerance is mediated by bone marrow-derived APCs that have acquired the HA epitope and presented it in a toleragenic manner (34). As all bone marrow-derived APCs are capable of expressing B7, this observation is consistent with the notion that the absence of B7 expression is not a cause of tolerance induction in vivo. However, we cannot rule out the possibility that the tolerizing APC either expresses low levels of B7 (i.e., resting B cells or macrophages) or a particular ratio of B7-1 to B7-2 that favors tolerance induction over priming as previously suggested (46). Alternatively, tolerance induction might be determined by the presence or absence of other factors on the tolerizing APC. The identification and characterization of the tolerizing APC should provide additional insight into this issue.

We thank Firouzeh Korangy and Denise Golgher for the 6.5 hybridoma, Frank Guarnieri for vacc-HA, and Bruce Lyons for providing helpful advice regarding CFSE-labeling experiments. We also thank Abdel Hamad, Hy Levitsky, and Marc Jenkins for helpful discussions and critical reading of the manuscript.