Duration and Strength of Extracellular Signal-Regulated Kinase Signals Are Altered During Positive Versus Negative Thymocyte Selection¹

During differentiation in the thymus, double-positive (DP)³ thymocytes undergo two important selection steps en route to differentiating into mature CD4⁺ or CD8⁺ single-positive T cells. Positive selection directs survival and differentiation of thymocytes that recognize ligands presented by self-MHC molecules, whereas negative selection removes overtly self-reactive thymocytes (1). Paradoxically, these opposing thymocyte fates are primarily directed by signals triggered through the same TCR. Several models have been put forward to describe how these contrasting events of positive and negative selection may occur. The majority of evidence thus far supports the affinity/avidity model of selection: positive selection and differentiation are mediated via lower affinity-avidity interactions, whereas negative selection occurs after engagement of higher affinity/avidity ligands (2, 3, 4). However, the intracellular signals that discriminate between thymocyte-positive and -negative selection remain poorly understood.

Many studies have addressed the role of various mitogen-activated protein (MAP) kinase pathways during thymocyte develop-ment. Studies have shown that mice overexpressing dominant neg-ative components of the Ras (MAP) kinase signaling pathway (Ras/raf-1/MAP kinase kinase (MEK)) exhibit partially blocked positive selection (5, 6, 7, 8) and extracellular signal-regulated kinase (ERK) 1-deficient thymocytes show impaired positive selection (9). A selective role for ERK in positive selection was also shown in mice expressing a mutation in the TCR α-chain-connecting peptide (αCPM). This mutation results in abrogation of positive selection but not negative selection and prevents CD3δ from being recruited to the TCR-CD3 complex (10). Accordingly, positive selection is also attenuated in mice deficient in CD3δ and biochemical analysis revealed that ERK activation is severely impaired in these mice (11). Recent evidence has also shown that a novel guanine nucleotide exchange factor for Ras, RasGRP, is involved in thymocyte-positive selection (12). These studies along with studies using pharmacological inhibitors of the ERK pathway (13, 14, 15) clearly support the role for ERK during positive selection.

The role for ERK in negative selection remains controversial. Studies using transgenic mice expressing dominant negative effectors in the Ras/raf/MEK pathway suggest that Ras/raf/MEK do not play a role in negative selection (5, 6, 7). However, experiments have shown that modifying the ERK pathway by using pharmacological inhibitors can shift negative selection to positive selection (14, 15). In addition, studies have also shown that the absence of CD3δ not only impairs positive selection and ERK activation (11), but also impairs negative selection (16).

Other studies have suggested that p38 and c-Jun N-terminal kinase (JNK)/stress-activated protein kinase MAP kinases are important in mediating negative selection (17, 18, 19). Based upon the existing data, the currently favored model predicts that positive and negative selection results from the stimulation of unique signals involving discrete molecules. The induction of ERK leads to positive selection, while the induction of p38 and JNK results in negative selection. However, it is not clear how a strong signal that is associated with negative selection could induce p38 and JNK, and not ERK. Accordingly, recent studies suggest that ERK is activated during negative selection (10, 14, 15).

To gain further insights into TCR-mediated signals that determine cell fate, we developed a culture system in which a population of naive DP thymocytes were triggered with a defined set of ligands that induced either positive or negative selection. Biochemical analysis revealed that while positively selecting ligands triggered sustained low-level ERK activation, negatively selecting ligands induced strong but transient ERK activation. The ability to induce sustained vs transient ERK activation correlated with surface TCR levels. Furthermore, the transient ERK activation induced by negatively selecting ligands triggered important downstream events, since inhibition of such initial ERK activation abrogated clonal deletion. Together, our data suggest that the extent and duration of ERK activation could influence both positive and negative thymocyte selection.

Generation of P14 TCR recombination-activating gene (RAG) 2-deficient H-2^d/d mice has been previously described (15). C57BL/6 mice, β₂-microglobulin (β₂m)-deficient mice, and RAG-1-deficient mice were obtained from The Jackson Laboratory (Bar Harbor, ME). Mice were bred and maintained according to institutional guidelines.

The MEK1 inhibitor PD98059 was purchased from Calbiochem (La Jolla, CA). U0126 was kindly provided by M. Favata (DuPont Pharmaceuticals, Wilmington, DE). The peptides p33 (KAVYNFATM), S7A (KAVYNFSTM), L6F (KAVYNLATM), and AV (SGPSNTPPEI) have been characterized before (20).

C57BL/6 mice or β₂m-deficient mice were used for harvesting day 4 thioglycolate-activated peritoneal macrophages as APCs. Macrophages were plated at 3 × 10⁵ cells/well in 24-well plates in IMDM supplemented with 10% FCS, 2 mM glutamine, 5 × 10⁻⁵ M 2-ME, penicillin, and streptomycin. After overnight incubation, nonadherent cells were gently washed away. The APCs were then pulsed with appropriate concentrations of peptides for 2 h. Thymocytes were harvested from P14 RAG2^−/− H-2^d/d mice and kept on ice for 2 h as a single-cell suspension. For peptide/APC stimulation, thymocytes (2 × 10⁶ cells/well) were then complexed with the peptide-pulsed macrophages by quick-spinning the cells at 1300 rpm for 30 s. The cells were placed in the 37°C incubator for appropriate periods of time. Thymocytes were then harvested by gentle pipetting. Usually thymocytes from two to three wells were pooled to obtain a sufficient number of cells to perform biochemistry. Residual macrophages were removed by incubating these thymocytes with anti-F4/80 Abs conjugated with anti-rat-IgG Dynal magnetic beads for 30 min at 4°C and subjecting them to a magnetic field. For studies using PD98059, thymocytes were preincubated with the inhibitor for 1 h at 37°C before stimulation.

Day 15 fetal thymic lobes from C57BL/6 mice were cultured for 5 days in 1.35 mM deoxyguanosine (Sigma, St. Louis, MO). The lobes were washed three times in medium. Briefly, 2 × 10⁵ thymocytes that had been stimulated for 16 h in the monolayer signaling cultures were placed in Terasaki wells along with deoxyguanosine-treated lobes. These were cultured overnight as hanging drops in a humidified chamber in the 37°C incubator. Reconstituted lobes were then placed on 0.8-μm polycarbonate filters (Costar, Cambridge, MA), and further incubated for 40 h in medium containing 12.5% FCS. After this incubation, the thymic lobes were teased apart, the cells were enumerated by trypan blue exclusion and stained for flow cytometric analysis with anti-CD4-PE, CD8-FITC, and H-2K^d-biotin.

Thymocytes were stained with PE-conjugated anti-CD4, FITC-conjugated anti-CD8, and biotin-conjugated anti-CD69, anti-CD5, anti-Vα2, or H-2K^d. All Abs were purchased from BD PharMingen (San Diego, CA). Cell viability was determined by enumerating trypan blue-negative cells or by flow cytometry, staining thymocytes with FITC-conjugated annexin V and propidium iodide (BioSource International, Camarillo, CA) or vital chromogenic dye 7-amino actinomycin D.

Peptide-APC-stimulated thymocytes were then washed once in PBS and pelleted. Whole cell lysates were prepared by addition of lysis buffer containing 50 mM Tris (pH 7.5), 20 mM EDTA, and 1% Triton X-100 supplemented with 10 mg/ml aprotinin, 10 mg/ml leupeptin, and 0.2 mM PMSF on ice for 20 min. The lysates were cleared by centrifugation. Protein concentrations were determined using a commercially available kit (Bio-Rad, Hercules, CA). Cell lysates with equivalent protein content were electrophoresed in a 12% SDS-polyacrylamide gel, transferred to a nylon membrane, and detected for “activated” ERK1/2 using a PhosphoPlus p44/42 MAP kinase (Thr²⁰²/Tyr²⁰⁴) Ab kit (NEB, Beverly, MA). Signals were detected using the ECL system (Amersham, Arlington Heights, IL). Densities of phospho-ERK bands were quantitated using ImageQuant software. To account for loading variability, values were normalized based on the density of total ERK bands. Values are expressed as fold increase over baseline stimulation conditions. For studies using PD98059 or U0126, thymocytes were preincubated with the inhibitor for 2 h at 37°C before stimulation and washed.

Thymocytes were loaded with Indo-1 (10 μM) for 1 h at 37°C in IMDM supplemented with 2% FCS. Indo-1⁺ cells exhibiting a large forward scatter corresponding to thymocyte-APC duplexes were analyzed with FACSVantage (BD Biosciences, Mountain View, CA) and CellQuest software as previously described (20). For stimulation of thymocytes, appropriate macrophages were pulsed with various peptides for 2 h. Macrophages (2 × 10⁶ cells) were mixed with thymocytes (1 × 10⁶ cells), centrifuged, and warmed to 37°C for 3 min. Cells were gently resuspended and immediately analyzed. The basal level of Ca²⁺ observed in thymocytes in the presence of unpulsed APCs was calibrated at 200, as arbitrary value. The Ca²⁺ flux induced by nonstimulatory AV peptide was superimposable on this basal Ca²⁺ flux and hence should be read as the baseline response.

One of the limitations in directly examining various signaling molecules during thymocyte development is that ∼2 × 10⁶ thymocytes are required for Western blot analysis. Another limitation is that thymocytes should synchronously receive a defined TCR signal that will induce positive or negative selection. To follow the induction of the ERK pathway during thymocyte selection, we established an in vitro system for thymocyte selection. P14-transgenic mice that express an H-2^b-restricted lymphocytic choriomeningitis virus-glycoprotein-specific TCR was bred onto a nonselecting H-2^d/d RAG2-deficient background (referred to as P14 RAG2^−/− H-2^d/d). Since thymocytes from these mice were arrested at the DP cell stage, they were cultured in conditions that promoted either positive or negative selection. For most experiments, macrophages were used as APCs; however, similar results were obtained with thymic stromal cell lines (data not shown). Previous studies using different models have shown that other cell types, including macrophages, could promote positive selection (21, 22, 23, 24).

To define conditions that promoted negative selection, P14 RAG2^−/− H-2^d/d thymocytes were cultured on H-2^b macrophages prepulsed with either the nominal peptide p33 (KAVYNFATM, 10⁻⁶ M) or a strong agonist variant S7A (KAVYNFSTM, 10⁻⁶ M). These peptides were previously shown to induce effective negative selection in FTOC (20, 25, 26). After various times, thymocytes were harvested and stained with Abs specific for CD4, CD8, CD69, CD5, and Vα2. In addition, cell viability was assessed using propidium iodide and annexin V. When thymocytes were cultured with p33 or S7A on H-2^b APCs for 20 h, CD4 and CD8 coreceptor down-regulation was observed in conjunction with an increase in apoptotic cells (Fig. 1, A and B). These events have been correlated with negative selection (27). Under these conditions, thymocytes also rapidly up-regulated the activation markers CD69 and CD5 (Fig. 1, C and D), reflecting the intensity of TCR triggering.

Positive selection was induced by culturing thymocytes on β₂m-deficient H-2^b macrophages prepulsed with either the strong agonist peptide p33 (10⁻⁷ M) or the weak agonist peptide variant L6F (KAVYNLATM, 10⁻⁶ M). We postulated that decreasing the level of available peptide-MHC complexes using β₂m-deficient macrophages would favor positive selection by reducing the avidity of the thymocyte-APC interactions. Both p33 and L6F have been shown to mediate positive selection of functional thymocytes in FTOC (15, 20). A nonstimulatory adenovirus peptide, AV (10⁻⁶ M), was used as a control peptide. Upon culture of thymocytes with p33 or L6F on β₂m^−/− APCs, initial events associated with positive selection were detected, such as up-regulation of CD69 (28) and CD5 (29) (Fig. 1, C and D). Moreover, interactions that favored positive selection did not dramatically affect cell viability (Fig. 1, A and B) and TCR levels remained high (Fig. 1 E). This analysis demonstrates that these culture conditions are able to induce early events associated with either positive or negative selection.

It should be noted that in defining the conditions that were most efficient in promoting either positive or negative selection, a full titration of all of the peptides (between micromolar to picomolar concentrations) was done on both β₂m^+/+ and β₂m^−/− APCs. A clear correlation was observed where maximal TCR down-regulation only correlated with the induction of negative selection, while minimal down-regulation directly correlated with positive selection.

We next examined whether the thymocytes that had received positively selecting stimuli could further differentiate into CD8⁺ single-positive T cells. P14 RAG2^−/− H-2^d/d thymocytes that had received different signals in the monolayer culture system for 16 h were transferred to culture wells containing deoxyguanosine-treated C57BL/6 (H-2^b/b) fetal thymic lobes. After 60 h of culture, flow cytometric analysis was performed on thymocytes that repopulated the fetal thymic lobes. Thymocytes that received positively selecting stimuli in the initial in vitro culture system (10⁻⁷ M p33/β₂m^−/− or 10⁻⁶ M L6F/β₂m^−/−) were able to mature into CD8⁺ T cells, compared with thymocytes that received “null stimulation” with AV peptide (Fig. 2). On the other hand, thymocytes that initially received strong negatively selecting stimuli (10⁻⁶ M p33/β₂m^+/+ or 10⁻⁶ M S7A/β₂m^+/+) were unable to survive and differentiate in the fetal thymic lobes. This was demonstrated by the 8- to 10-fold reduction in thymic cellularity and the presence of only the double-negative thymocyte subset. This analysis demonstrates that this in vitro system is appropriate for examining early selection events that promote positive and negative selection.

To investigate the activation of the Ras/Raf/MEK/ERK pathway during selection events, Western blot analysis was done using an Ab specific for the phosphorylated activated form of ERK. As shown in Fig. 3,A, P14 RAG2^−/− H-2^d/d thymocytes cocultured in positively selecting conditions induced sustained ERK activation. During positive selection with either p33 or L6F, activated ERK was still detectable after 16 h compared with “null-selecting” AV peptide (Fig. 3, A and B). Notably, positive selection with low-avidity interactions (10⁻⁷ M p33/β₂m^−/− APCs) or low-affinity peptides (10⁻⁶ M L6F/β₂m^−/− APCs) were similar, consistent with the affinity/avidity model.

The duration of ERK activation was examined in thymocyte cultures that induced negative selection. P14 RAG2^−/− H-2^d/d thymocytes were cultured with p33 (10⁻⁶ M) and S7A (10⁻⁶ M) on β₂m^+/+ APCs. At various time points, cell lysates were prepared and Western blot analysis was done using the Ab specific for the phosphorylated forms of ERK. The activation of ERK reached maximum at 3–4 h and then declined 5–6 h after stimulation (Fig. 4,A). A similar transient kinetics was observed for negative selection that was induced by the strong agonist ligand S7A (Fig. 4 B). Therefore, negative selection is associated with a strong transient activation of ERK, while conditions that induce positive selection generate sustained ERK activation.

The degree of initial ERK activation was also compared between positively and negatively selecting conditions. Again, P14 RAG2^−/− H-2^d/d thymocytes were cocultured with β₂m^−/− and β₂m^+/+ APCs that were prepulsed with positive (10⁻⁷ M p33/β₂m^−/− or 10⁻⁶ M L6F/β₂m^−/−) or negative (10⁻⁶ M p33/β₂m^+/+ or 10⁻⁶ M S7A/β₂m^+/+) selecting ligands. Cultures were prepared in parallel and harvested after 3 h. Fig. 4,C shows that negatively selecting conditions induced strong ERK phosphorylation in contrast to weak phosphorylation triggered by the positively selecting interactions. It should be noted that in all cases a background level of ERK is seen (Figs. 3 and 4). Our data suggest that this level is not sufficient for the induction of positive or negative selection under these defined culture conditions. Activation of ERK above a certain threshold may help discriminate between death by neglect and positive vs negative selection.

In these culture conditions, negative selection has been correlated with peptides that were able to induce rapid TCR internalization (Fig. 1,E) (20). To examine whether any remaining or re-expressed TCR were able to trigger other downstream pathways, we examined the ability of the thymocytes to mobilize calcium. P14 RAG2^−/− H-2^d/d thymocytes were complexed with negatively or positively selecting ligand/APCs. After 3 min, calcium mobilization was monitored. Initially, negatively selecting interactions induced a strong Ca²⁺ flux, whereas positively selecting interactions induced a weaker Ca²⁺ flux (Fig. 5,A). After 5 h of coculture, thymocytes were transferred to plates with fresh APCs pulsed with high concentration of p33 (10⁻⁶ M). Thymocytes that were cocultured with negatively selecting ligands for 5 h were refractive to further Ca²⁺ flux on re-exposure to class I⁺ APC pulsed with the strong agonist p33 (Fig. 5,B). On the contrary, thymocytes that were incubated with peptide/MHC ligands that induced positive selection (10⁻⁷ M p33/β₂m^−/− or 10⁻⁶ M L6F/β₂m^−/−) were able to generate a Ca²⁺ flux in response to positively selecting stimuli or a strong antigenic stimulus (Fig. 5 B; data not shown). These data provide direct evidence that negatively selecting TCR-peptide-MHC interactions decrease TCR expression to the extent where the remaining TCRs are unable to trigger other downstream signal transduction pathways.

The next question we wanted to address was whether the transient ERK activation transmits downstream signals, leading to negative selection. To address this issue, experiments were done to block the activation of ERK using the pharmacological inhibitor PD98059. P14 RAG2^−/− H-2^d/d thymocytes were preincubated with 25 μM PD98059, a selective compound that has been shown to inhibit the activation of MEK1/2 by upstream activators of the MAP kinase cascade (30, 31). These thymocytes were cultured for 1 h with β₂m^+/+ APCs that have been prepulsed with 10⁻⁶ M p33. The data show that 25 μM PD98059 markedly diminishes ERK activity in p33-induced thymocytes (Fig. 6 A).

To evaluate the consequence of blocking ERK activation in clonal deletion, P14 RAG2^−/− H-2^d/d thymocytes that were treated with medium alone (control) or pretreated with PD98059 were cultured with β₂m^+/+ APCs pulsed with either the nonstimulatory AV (10⁻⁶ M) peptide or the antigenic p33 (10⁻⁶ M) peptide. After 16 h of incubation, DP thymocytes cultured under conditions that promoted negative selection showed a substantial down-regulation of CD4 and CD8 coreceptors, along with enhanced apoptosis (Fig. 6, B and C). In contrast, PD98059 was able to inhibit clonal deletion by negatively selecting stimuli. Consistent with the block in ERK activation, flow cytometric analysis showed that up-regulation of Ras-dependent CD69 expression was also abrogated (Fig. 6,D). Examining the P14-transgenic TCR levels confirmed that the thymocytes received the p33 peptide/MHC signal and internalized the P14 TCR (Fig. 6,E). In addition to PD98059, we also used U0126, a second, noncompetitive inhibitor of MEK (32). As shown in Fig. 7, U0126 inhibition of MEK had the same consequence as inhibition of PD98059 in that deletion of DP thymocytes was significantly reduced. Together these data demonstrate that inhibition of MEK activity could inhibit p33-induced clonal deletion. These data suggest that the MEK/ERK pathway does generate downstream signaling events necessary to induce negative selection.

In the present study, we have developed a novel in vitro culture system that is suitable for evaluating early biochemical events associated with positive and negative selection of a synchronous population of naive DP thymocytes (Fig. 1). We have also shown that the early signals received during the initial 16-h culture, either negatively selecting signals or positively signals, were sufficient to lead to death or promote CD8⁺ selection after maturation in deoxyguanosine-treated thymic lobes, respectively (Fig. 2). Although several attempts were made to promote maturation in vitro in the initial monolayer APC culture system, successful maturation was only achieved in the thymic microenvironment. This is consistent with previous studies suggesting that this environment is essential for the maturation of thymocytes (24, 33).

Although it is possible that maturation of CD8⁺ T cells may require extended signals from the class I molecules encountered in the fetal thymic lobes, the initial culture conditions definitely provided the appropriate signals to direct positive selection of these cells. We observed that the nonstimulatory peptide AV (10⁻⁶ M/β₂m^−/−) induced minimal maturation of CD8⁺ T cells (6%), whereas the weak agonist ligand (10⁻⁶ M L6F/β₂m^−/−) or low concentrations of strong agonist ligand (10⁻⁷ M p33/β₂m^−/−) triggered maturation of a significant proportion of CD8⁺ T cells (27 and 30%, respectively). The ability of weak agonists or low concentrations of strong ligands to promote positive selection is fully consistent with an affinity/avidity model for selection. The inefficient maturation of nonstimulatory peptide AV-induced thymocytes in C57BL/6 FTOC may appear peculiar at first, since P14 TCR-transgenic T cells are effectively selected in this genetic background. However, a 60-h culture in the FTOC may not be sufficient for maturation via interactions with natural ligands. Cumulative nuclear signals may be required for positive selection and, as such, thymocytes that were induced with positive-selecting ligands in the initial monolayer culture would have acquired a temporal advantage. Accordingly, our data demonstrate that thymocytes cultured in the positively selecting conditions acquire active ERK molecules over a sustained period compared with thymocytes cultured in null-selecting conditions.

This study demonstrates that exposure to either strong agonist peptide, p33 or S7A, induces intracellular events that trigger negative selection. Evidence for apoptosis was detected after 16–20 h in the initial culture, and further culture in the thymic microenvironment demonstrated that no detectable DP or single-positive cells repopulated the thymic lobes (Fig. 2). Interactions with relatively higher concentrations of strong agonist ligands led to extensive TCR internalization (Fig. 1,E), to a degree where remaining TCRs were unable to transmit further signals, as measured by the mobilization of intracellular calcium (Fig. 5). Biochemical analysis revealed that under conditions that induced negative selection, ERK phosphorylation was much stronger compared with positively selecting conditions (Fig. 4,C). In addition, the strong ERK activation is transient as it declines after 5 h (Fig. 4 A).

Does this strong ERK activity observed during the initial TCR triggering under negatively selecting conditions contribute to the induction of clonal deletion? This contention is supported by studies that have shown that reducing the intensity of ERK signals may shift negative selection to positive selection (14, 15). In addition, several studies have now shown that negative selection is correlated with a transient ERK signal (10, 13). However, the latter studies do not explicitly address whether the transient ERK activation has any biological significance for clonal deletion. Using our novel in vitro culture system, we investigated the requirement of ERK activation during negative selection by abrogating the initial ERK activity in thymocytes by preincubating them with either of two independent MEK inhibitors, PD98059 or U0126. These two MEK inhibitors used in this study do not block JNK or p38 activation (14). In addition, this protocol selectively inhibits MEK only in thymocytes and obviates the concerns surrounding the effects of the drugs on APC function during negative selection. Our studies demonstrate that abrogating the ERK activity in thymocytes blocked the rapid clonal deletion in response to strong negatively selecting stimuli (Figs. 6 and 7).

Previously, using the P14 TCR β₂m^−/− transgenic system in FTOC, we have shown that peptide-mediated negatively selecting stimuli could be converted into positively selecting signals by diminishing MEK-mediated ERK activity with the use of PD98059 (15). In these FTOC assays, the switch from negative to positive selection was observed at p33 peptide concentrations closer to the negative/positive selection thresholds. In this previous model, we showed that the MEK inhibitor could reduce, but not completely block, ERK activation. This is in contrast to this report where ERK activation is ablated. PD98059 may not effectively reduce ERK activity and block negative selection in FTOC for a variety of reasons. In FTOC, the lobes are not completely submersed in media. Therefore, the ability of the drug to permeabilize each cell at the same concentration is very unlikely. It is likely that a gradient is achieved where the levels of active ERK is reduced and not completely eliminated. Also in FTOC, it is unlikely that all thymocytes are synchronously receiving the negative selection signal. As the activity of PD98059 declines in culture, some thymocytes may receive a slightly reduced ERK signal and become positively selected. Nonetheless, these studies clearly show a role for ERK during negative selection and support a model where negative selection signals are stronger than positive selection signals.

How does ERK signaling contribute to thymocyte-negative selection? One possible downstream target gene that is regulated by ERK signaling is Id3. Id gene products lack DNA-binding domains, but exert their transcriptional influence by dimerizing with E proteins and thereby, disrupting E protein activity (34). In this manner, Id gene products enhance TCR-mediated responses by attenuating E protein activity. Recent data from Bain et al. (35) have positioned ERK MAP kinase signaling and Id3 gene regulation in a common pathway . They also show that PD98059 inhibits Id3 gene induction in a dose-dependent manner. Interestingly, previous studies have shown that Id3 deficiency can perturb both positive and negative selection (36). In addition to Id3, up-regulation of Nur77 has also been shown to be sensitive to PD98059-mediated ERK inhibition (13). Nur77 is an orphan member of the steroid nuclear superfamily and is rapidly induced by TCR-mediated signaling in immature thymocytes (37, 38). Transgenic mice that overexpress dominant-negative forms of Nur77 show protection from thymocyte apoptosis (39, 40), whereas overexpression of the wild-type Nur77 facilitated apoptosis (39, 41). These studies suggest that the ERK-MAP kinase module regulates genes that are involved in both thymocyte-positive and -negative selection. In this instance, the outcome of the thymocyte fate may be determined by the extent and/or timing of activation of ERK-sensitive nuclear genes.

Therefore, our data favor a model in which a strong but transient ERK activation does play a role in negative selection (Fig. 8). This ERK signal along with signals from other pathways may lead to the induction or inactivation of a subset of genes that lead to cell death. Conversely, an alternative scenario exists where the lack of sustained ERK activation during negative selection may not afford sufficient protection for thymocytes from JNK or p38-mediated apoptosis. Several studies have shown that JNK and p38 are involved in negative selection (16, 17, 18, 19), and there is evidence in other biological models where it has been shown that the dynamic balance between growth-activated ERK and stress-activated JNK-p38 pathways is important in determining whether a cell survives or undergoes apoptosis (42). On the other hand, positively selecting low-affinity/avidity ligands may achieve sustained signaling by their inability to induce maximal TCR internalization and the TCR complexes that remain on the surface are able to relay continuous signals for survival and differentiation. This is consistent with observations from other models that show that multiple interactions are required to promote thymocyte differentiation (43, 44). The current literature supports the role of ERK in positive selection (5, 6, 7, 8, 9, 11, 12) and other reports show that sustained ERK signals are observed during positive selection (10, 13).

The rat PC12 pheochromocytoma cells can be induced to differentiate or proliferate depending upon a given stimulus provided in vitro. Nerve growth factor-driven differentiation of PC12 cells into sympathetic neurons is induced by sustained ERK activation that is associated with “slow” receptor engagement (reviewed in Ref. 45). Similarly, thymocytes undergoing differentiation (positive selection) require slow TCR triggering which is accompanied by sustained ERK signaling (Fig. 3).

On the contrary, epidermal growth factor (EGF)-driven proliferation of PC12 cells is coupled with rapid receptor internalization and transient ERK activation. Similarly, thymocytes undergoing negative selection rapidly internalize their TCRs and become refractive to further stimulation. Intracellular signals such as ERK and calcium flux are transient (Figs. 1, 4, and 5). It should be noted that other mechanisms may also blunt any sustained ERK or Ca²⁺ activities in response to strong, negatively selecting stimulation in DP thymocytes. In addition, negatively selecting ligands can stimulate a robust proliferation in mature T cells in analogy with EGF-driven proliferation in the PC12 model (20). Interestingly, if PC12 cells are engineered to overexpress EGF receptors, there is prolonged ERK and, as a result, cells differentiate but do not proliferate in response to EGF stimulation (46).

Sustained ERK activation also promotes differentiation in other models. Protein kinase C-mediated sustained ERK activation has been shown to induce megakaryocyte differentiation in K562 cells. In this system,12-O-tetradecanoylphorbol-13-acetate (TPA) and bryostatin are known to activate protein kinase C but paradoxically have opposing effects on megakaryocyte differentiation. TPA, a differentiation inducer, caused sustained ERK activation (>24 h), whereas bryostatin, a differentiation blocker, only transiently activated ERK (∼6 h) and attenuated the activation of ERK by TPA (47, 48).

Intriguingly, in several models including thymocyte selection, cell fate determination is influenced by the dynamics of receptor engagement and signaling. Further experiments are necessary to elucidate the interplay of MAPK pathways and the induction or inhibition of genes that are necessary to induce positive or negative thymocyte selection.

We are indebted to Jim Woodgett for insightful discussions and Nancy Berg for critical reading of this manuscript.