It is generally accepted that the avidity of TCR for self Ag/MHC determines the fate of immature thymocytes. However, the contribution of the quantity of TCR signal to T cell selection has not been well established, particularly in vivo. To address this issue, we analyzed DO-TCR transgenic CD3ζ-deficient (DO-Tg/ζKO) mice in which T cells have a reduced TCR on the cell surface. In DO-Tg/ζKO mice, very few CD4 single positive (SP) thymocytes developed, indicating that the decrease in TCR signaling resulted in a failure of positive selection of DO-Tg thymocytes. Administration of the peptide Ag to DO-Tg/ζKO mice resulted in the generation of functional CD4 SP mature thymocytes in a dose-dependent manner, and, unexpectedly, DO-Tg CD8 SP cells emerged at lower doses of Ag. TCR signal-dependent, sequential commitment from CD8+ SP to CD4+ SP was also shown in a class I-restricted TCR-Tg system. These in vivo analyses demonstrate that the quantity of TCR signal directly determines positive and negative selection, and further suggest that weak signal directs positively selected T cells to CD8 lineage and stronger signal to CD4 lineage.
Mature T cells develop in the thymus through a series of regulated differentiation events. During this process, thymocytes are tested for reactivity with self-Ags to ensure that only those cells expressing appropriate TCRs will be selected. The critical parameter determining the fate of thymocytes is the avidity of the interaction between the TCRs expressed on thymocytes and the peptide/MHC complexes of the thymic stromal cells. T cells are subjected to this selection process at the CD4/CD8 double positive (DP)3 stage (1, 2, 3). DP thymocytes that express a TCR with high avidity for self peptides presented by MHC are eliminated by TCR-mediated apoptosis (negative selection). Only thymocytes bearing TCRs with moderate avidity for self peptide/MHC are rescued (positive selection) from the default apoptosis pathway (neglect) that have not successfully rearranged their TCR genes or that express a receptor with subthreshold avidity for self peptide/MHC. Most of the studies that support this avidity model were based on the analysis in fetal thymus organ culture (FTOC), in which different concentrations of peptides were administered in vitro (4, 5, 6, 7). However, these analyses were only focused on the quantity of available peptide/MHC complexes, but not on TCR signaling.
As a consequence of positive selection, two major lineages of T cells are generated: CD4 T cells that recognize peptide Ags complexed with class II MHC and possess helper function, and CD8 T cells that recognize peptides in the context of class I MHC and possess cytotoxic function. This lineage commitment correlates with the specificity of the TCR and the selecting MHC molecules. Two models had proposed to explain the development of T cells bearing matched TCR and coreceptor specificities. An instructive model proposes that MHC restriction of the TCR determines lineage commitment (8, 9). T cells expressing MHC class I- or class II-restricted TCR down-regulate CD4 or CD8, respectively. Alternatively, according to a stochastic model, coreceptor down-regulation may be an arbitrary event, and the initial TCR stimulation leads to random coreceptor down-regulation (10, 11, 12). Continued maturation is dependent on a properly matched MHC-restricted TCR and coreceptor. These studies were interpreted as suggesting that signaling by CD4 or CD8 coreceptor is indispensable for survival and complete differentiation of immature thymocytes via corecognition of the peptide/MHC. However, recently, this idea has been challenged by analyzing the coreceptor re-expression patterns of immature thymocyte subsets. These studies indicated that thymocyte commitment to the CD8+ lineage requires MHC class I-dependent instructive signals, whereas thymocyte commitment to the CD4+ lineage is MHC-independent and may occur by default (13). Furthermore, several new studies analyzing the development of the thymocytes of TCR-transgenic (Tg) coreceptor-deficient mice showed that coreceptors do not possess essential signaling function required for T cell survival and development (14, 15, 16). These studies suggest that signals through the TCR but not coreceptors influence the lineage decision.
At the DP stage, thymocytes express TCRαβ dimers associated with the CD3 complex, which is composed of three dimers of γε, δε, and ζ-ζ. Ag recognition by the TCRαβ dimer initiates intracellular activation signals mainly through immunoreceptor tyrosine-based activation motif (ITAM) within the cytoplasmic region of the CD3 chains. Particularly, the ζ-chain with three ITAMs initially has been suggested to play a unique and indispensable role in the signaling function of TCR. However, recent analysis revealed that ITAMs of ζ are dispensable for T cell activation (17). Particularly for thymocyte development, previous experiments using Tg mice expressing ζ lacking the cytoplasmic region have suggested that the ζ-chain plays a crucial role, not because of its signaling function during positive selection but rather due to its function to induce a high level of stable TCRs on the cell surface of immature thymocytes (18).
Therefore, we addressed the question of how the level of TCR influences to the selection of T cells by analyzing in vivo T cell development in mice expressing a low level of surface TCR complexes. We took advantage of our CD3ζ knockout (ζ−η+, ζKO) mice in which T cells have a reduced surface TCR complex and impaired signal transduction through the TCR complex (19). We crossed ζKO mice with DO11.10-TCR transgenic (DO-Tg) mice and analyzed T cell development in DO-Tg/ζKO mice by administering various amounts of OVA peptide in vivo. Using this system, we were able to analyze the contribution of both the level of TCR and the dose of Ag to T cell development simultaneously without altering the expression of MHC and CD4/CD8 coreceptors. The data demonstrate that the magnitude of TCR signaling directly determines both positive and negative selection. Furthermore, unexpectedly, DO-Tg+ CD8+ single positive (SP) cells emerged by injection of low doses of OVA peptide, whereas CD4 SP mature thymocytes were generated by injection of higher doses. These results demonstrate that the quantity of TCR signal directly regulates not only positive/negative selection but also lineage commitment.
Materials and Methods
Mice transgenic for DO11.10 TCRαβ (DO-Tg) on a BALB/c background (20) were kindly provided by D. Y. Loh, and were maintained by interbreeding. CD3ζ-deficient (ζKO) mice have been previously characterized (19). These mice were deficient specifically in the CD3ζ gene but remained normal splicing and expression of the CD3η gene. Whereas mice deficient for both ζ and η do not develop SP thymocytes expressing surface TCR, our ζKO mice possess small numbers of SP thymocytes and express a low level of the TCR complex on the cell surface. DO-Tg mice were bred with ζKO mice that had been back-crossed at least seven times with BALB/c mice purchased from the Shizuoka Laboratory Animal Corporation (Hamamatsu, Japan). The offspring were intercrossed to obtain CD3ζ+/−, CD3ζ−/− mice, which were DO-Tg. The transgene was analyzed with tail DNA by PCR using primers (5′-CAGGAGGGATCCAG TGCCAGC-3′, and 5′-TGGCTCTACAGTGAGTTTGGT-3′). The CD3ζ alleles were analyzed with tail DNA by PCR using primers (5′-TCCTGTCAGCATTCTCAGGCAAG-3′, and 5′-GTGTCACCTTGAATCTCGAGCACCG-3′). Lymphocytic choriomeningitis virus (LCMV)-Tg/β2-microglobulin (β2m)-KO mice with C57BL/6 background were crossed with ζKO mice, and the offspring were intercrossed to obtain CD3ζ+/−, CD3ζ−/− mice, which were LCMV-Tg/β2m-KO mice. Mice were bred and maintained in our own animal facility.
Abs and peptides
The following Abs were purchased from PharMingen (San Diego, CA): anti-CD3ε (145-2C11)-biotin, anti-TCRβ (H57-597)-biotin, anti-heat-stable Ag (HSA) (J11d)-biotin, anti-CD69 (H1.2F3)-biotin, anti-CD8 (53-6.7)-PE, and anti-CD4 (RM4-5)-FITC. Biotinylated KJ1-26 (anticlonotypic mAb that recognizes the TCRαβ of DO11.10) were kindly provided by P. Marrack (National Jewish Center, Denver, CO). Biotinylated B20.1 (PharMingen) and F23.1 that recognize Vα2 and Vβ8.1, respectively, were used for analysis of FTOC of LCMV-Tg/β2m-KO mice.
Two different OVA peptides (OVA323–339 and OVA324–334; synthesized by Sawady Technology, Tokyo, Japan) were used for stimulation of DO-Tg thymocytes. DO-Tg T cells respond to OVA323–339 when presented in the context of I-Ad, whereas OVA324–334, which also binds to I-Ad, cannot stimulate the Tg TCR. Peptide sequences are as follows: OVA323–339, ISQAVHAAHAEINEAGR; and OVA324–334, SQAVHAAHAEI. A LCMV peptide (p33, KAVYNFATM) (16) was used for cocultivation with LCMV-Tg/β2m-KO thymic lobes in FTOC.
Induction of positive selection of adult thymocytes in in vitro culture
A total of 1 × 105 nonadherent APCs (irradiated (30 gray) BALB/c splenocytes) and 1 × 106 thymocytes were cultured in a 96-well round-bottom tissue culture plate (Becton Dickinson, Mountain View, CA) in 0.2 ml of RPMI 1640 medium/10% FCS containing Ag peptide for 72 h. Cells were harvested by pipetting, and harvested cells were counted, stained, and analyzed by FACScan as below.
Induction of positive and negative selection in vivo
OVA323–339 or PBS (no peptide) was administrated to DO-Tg and DO-Tg/ζKO mice as 250 μl of sterile solution at various concentrations by i.p. injection every 4 days. Seven days after the first administration, mice were sacrificed and thymocytes were counted, stained, and analyzed by FACScan as below.
Timed breedings were established between LCMV-Tg/β2m−/− ζ−/− females with LCMV-Tg/β2m−/− ζ−/− males or LCMV-Tg/β2m−/− ζ+/+ males to obtain LCMV-Tg/β2m−/− ζ−/− or LCMV-Tg/β2m−/− ζ+/− fetuses, respectively. At day 15 of gestation, females were sacrificed and thymic lobes were removed from the fetuses. DNA was extracted from embryonic tails so that Tg fetuses could be determined using primers specific for the Vα2 TCR transgene (21). The fetal thymic lobes were submerged in 200 μl of IMDM, 1× Nutridoma SP (Boehringer Mannheim, Indianapolis, IN), 50 μM 2-ME, penicillin, streptomycin, 4 mM glutamine, 5 μg/ml β2-microglobulin (β2m) and with or without LCMV peptide. The plates were put into a plastic bag (Ohmi Odor Air Service, Hikone, Japan), and the air was exchanged with a mixed gas containing 80% O2, 5% CO2, and 15% N2 (22). These lobes were then cultured for 7 days at 37°C, during which time the medium and the mixed gas was exchanged every other day. After this incubation period, the thymic lobes were teased apart and viable cells were enumerated by trypan blue exclusion. Thymocytes were stained and analyzed by FACScan as follows.
Flow cytometric analysis
T cells were stained biotinylated mAb for 40 min, followed by PE-anti-CD4, FITC-anti-CD8, streptavidin-Quantum Red (Sigma, St. Louis, MO) for 30 min. After washing, cells were analyzed on a FACScan (Becton Dickinson) using CellQuest software. Dead cells were gated out by staining with propidium iodide at 4 μg/ml (Sigma) in the FL-3 channel.
Thymocytes from OVA-treated DO-Tg/ζKO mice or nontreated DO-Tg, BALB/c mice were incubated with anti-HSA mAb (J11D) for 30 min on ice, followed by addition of complement (normal rabbit serum) at 37°C for 30 min to enrich for CD4+ or CD8+ mature SP thymocytes. Thereafter, the T cells were stained with anti-CD4-PE and anti-CD8-FITC mAb, and CD4+ CD8− SP or CD4− CD8+ SP T cells were sorted by FACStarPlus (Becton Dickinson). The purity of the sorted cells was always >99%. SP T cells were stimulated at a concentration of 2 × 104 cells per well in a 96-well round-bottom tissue culture plate (FALCON) in 0.2 ml culture medium with 2 × 105 irradiated (30 gray) BALB/c splenocytes and 3 μM OVA323–339. A total of 2 × 104 SP T cells were stimulated with immobilized anti-CD3ε (20 μg/ml) in a 96-well flat-bottom tissue culture plate (FALCON) in 0.2 ml of medium containing10% mouse IL-2 culture supernatant with 2 × 105 irradiated (30 gray) BALB/c splenocytes as a feeder cells. After 3 days of incubation at 37°C, cells were pulsed with 37 kBq of [3H]thymidine (37 Mbq/ml; Amersham Pharmacia International, Aylesbury, U.K.). At the end of another 8-h incubation, cells were harvested, and incorporated radioactivity was measured with MicroBeta (Amersham Pharmacia Biotech, Uppsala, Sweden).
A shift from positive selection of T cells to neglect by decreasing TCR expression
Thymocyte maturation in our CD3ζ−η+ (ζKO) mice has been previously characterized by a reduction of CD4+CD8+ DP cells as well as CD4+CD8− and CD4−CD8+ SP thymocytes, whereas the CD4−CD8− double negative (DN) population was not affected (19) (Fig. 1, and Table I). This impairment of T cell development is due to the low expression of TCR on DN thymocytes and the signal to induce the transition to DP was insufficient. To assess whether the magnitude of TCR signals affects the positive selection of DO-Tg thymocytes, we crossed DO-Tg mice with ζKO mice (DO-Tg/ζKO mice) and analyzed T cell development in the thymus (Fig. 1). DO-Tg mice bear a TCR that recognizes an OVA peptide bound to I-Ad (20). As reported previously, the majority of thymocytes bearing this receptor develop into CD4+ SP cells (Fig. 1,c). In contrast, very few CD4+ SP thymocytes were detected in DO-Tg/ζKO mice (Fig. 1,d). However, the expression of the DO TCR transgene restored the cellularity of DP thymocytes in DO-Tg/ζKO mice to normal levels (Fig. 2, and Table I). These results indicate that positive selection of OVA-specific T cells was abrogated by the deficiency of CD3ζ.
|.||DO−ζ+/− .||DO−ζ−/− .||DO+ζ+/− .||DO+ζ−/− .|
|CD4−CD8−||4.5 ± 1.1||6.4 ± 1.3||14.9 ± 3.3||13.8 ± 6.5|
|CD4+CD8+||142.6 ± 42.9||19.4 ± 5.7||195.0 ± 54.7||182.1 ± 61.1|
|CD4+CD8−||24.3 ± 6.3||3.0 ± 0.8||25.1 ± 11.1||2.7 ± 0.9|
|CD4−CD8+||16.1 ± 4.8||4.5 ± 1.2||3.4 ± 1.2||3.2 ± 1.6|
|Total||187.5 ± 52.2||33.2 ± 7.3||238.5 ± 53.5||201.8 ± 72.9|
|.||DO−ζ+/− .||DO−ζ−/− .||DO+ζ+/− .||DO+ζ−/− .|
|CD4−CD8−||4.5 ± 1.1||6.4 ± 1.3||14.9 ± 3.3||13.8 ± 6.5|
|CD4+CD8+||142.6 ± 42.9||19.4 ± 5.7||195.0 ± 54.7||182.1 ± 61.1|
|CD4+CD8−||24.3 ± 6.3||3.0 ± 0.8||25.1 ± 11.1||2.7 ± 0.9|
|CD4−CD8+||16.1 ± 4.8||4.5 ± 1.2||3.4 ± 1.2||3.2 ± 1.6|
|Total||187.5 ± 52.2||33.2 ± 7.3||238.5 ± 53.5||201.8 ± 72.9|
The numbers of each population (×106) are calculated from the data of CD4/CD8 staining on thymocytes from individual mice by flow cytometric analysis (see Fig. 1). The data are presented as the mean ± SD of six (8 week old) mice.
To investigate the relationship between TCR expression and thymocyte development, we measured the TCR level on thymocyte subsets from the different types of mice (Fig. 2, and Table II). In DO-Tg mice, the expression level of the DO-TCR decreases during thymocyte maturation from the DN to DP stage, whereas mature CD4+ thymocytes up-regulate the TCR density. The levels of the DO TCR on DN thymocytes from DO-Tg/ζKO mice were lower than those of DO-Tg mice but higher than the endogenous TCR expression on thymocytes from ζKO mice. This increased expression of the Tg TCR may explain the normal development of DP thymocytes in DO-Tg/ζKO mice. The low level expression of the TCRαβ-chains on DN cells instead of the pre-TCR may deliver sufficient signals to differentiate into DP cells. In contrast, TCR levels on DP thymocytes of DO-Tg/ζKO mice, which were lower than those of DO-Tg mice, were insufficient to induce positive selection.
|.||DO-Tg mice .||DO-Tg/ζKO mice .||ζKO miceb .|
|.||DO-Tg mice .||DO-Tg/ζKO mice .||ζKO miceb .|
Thymocytes were stained with anti-CD8-FITC mAb, anti-CD4-PE mAb, and anti-clonotypic Ab (KJ1-26)-biotin plus streptavidin-Quantum Red. Mean fluorescence intensity in each population was quantitated by CellQuest software. Representative data from three independent analysis are shown.
Since ζKO mice do not express transgenic TCR, the numbers for ζKO mice indicate the background level of negative control staining.
Induction of positive/negative selection by in vivo administration of Ag peptide in DO-Tg/ζKO mice
Because the DO-Tg/ζKO thymocytes expressed lower levels of TCRs, it was possible that the reduced avidity was responsible for the lack of positive selection. We reasoned that the addition of high affinity peptides for the TCR might be able to increase the avidity between TCR and MHC/peptide and restore the sufficient TCR signaling required for positive selection. To test this idea, we administered various concentrations of OVA peptide (OVA323–339) into DO-Tg and DO-Tg/ζKO mice (Fig. 3). In DO-Tg mice, negative selection was observed with high doses of peptide in a dose-dependent manner as previously described (23), whereas the lower doses did not have any effect on the selection. In contrast to DO-Tg mice, as expected, CD4+ SP thymocytes in DO-Tg/ζKO mice were increased in a dose-dependent manner at the concentration between 75 and 300 μM of peptide and then were deleted at 1000 μM. It was noted that negative selection took place at the Ag dose for maximum selection of CD4+ SP. This suggests that positive selection of CD4+ cells and negative selection occur at very close Ag doses. These results demonstrate that in vivo administration of OVA peptide compensates the avidity to provide sufficient signals for positive and negative selection in DO-Tg/ζKO mice. In this system, we were able to induce three different fates of thymocytes, neglect, positive selection, and negative selection, by changing the concentration of a single peptide without manipulating clonotypic TCRαβ-chains, MHC, and CD4/CD8 coreceptors.
Quantity of TCR signal determines CD4/CD8 lineage commitment
In the mice administered with OVA peptide, CD8+ SP cells emerged unexpectedly at low doses (75–150 μM) of peptide (Fig. 3). However, these CD8+ SP thymocytes were deleted at intermediate concentration (300 μM) in which positive selection of CD4+ SP cells was still induced. Both CD4+ and CD8+ SP cells that developed in OVA-administrated DO-Tg/ζKO mice expressed the DO Tg TCR (see below). These results suggest that the quantity of TCR signal directly regulates positive and negative selection, and further suggests that weaker signals direct positively selected T cells to the CD8 lineage and stronger signals to the CD4 lineage. To ensure that these CD4+ or CD8+ SP thymocytes had undergone positive selection, thymocytes from OVA-administered DO-Tg/ζKO mice (150 μM) were stained with the maturity markers, HSA and CD69 (24, 25). Onsets of HSA down-regulation and CD69 up-regulation are closely correlated with positive selection. CD4+ and CD8+ SP thymocytes from OVA-administered DO-Tg/ζKO mice expressed low level of HSA and high level of CD69 as compared with DP thymocytes (Fig. 4). Higher levels of expression on SP cells were also shown for Tg TCR (KJ1-26 staining) and CD3 expression (CD3ε staining) despite low level of TCR expression due to ζ-deficiency. These data suggest that both CD4+ and CD8+ SP thymocytes from OVA-administered DO-Tg/ζKO mice were positively selected mature thymocytes.
Ag-selected SP thymocytes are functional
Proliferation assays were conducted to determine whether the CD4+ and CD8+ SP DO-Tg thymocytes selected in the presence of OVA323–339 peptide were functional. CD4+ and CD8+ SP thymocytes from OVA-administered DO-Tg/ζKO, DO-Tg, and BALB/c mice were stimulated with irradiated BALB/c splenocytes and OVA peptide or with immobilized anti-CD3ε mAb (2C11) plus IL-2 (Fig. 5). Positively selected CD4+ and CD8+ SP thymocytes from DO-Tg/ζKO proliferated in response to immobilized 2C11 plus IL-2 at about one-third the level of those of the CD4+ and CD8+ DO-Tg thymocytes, consistent with the impaired function of T cells from ζKO mice due to the reduction of TCR expression (19). CD4+ DO-Tg/ζKO thymocytes proliferated upon stimulation with peptide plus APCs but not with APC alone, demonstrating that the proliferation was OVA specific. In contrast, CD8+ DO-Tg/ζKO thymocytes, although positively selected with this peptide, failed to respond to OVA323–339/APC. It is likely that CD8+ DO-Tg/ζKO thymocytes did not respond to OVA/APCs in the absence of CD4. These data demonstrated that CD8+ DO-Tg/ζKO thymocytes are not functional for specific Ags presented by class II molecules, but are capable of responding to a strong signal by TCR aggregation. We concluded that OVA323–339 induced functional SP thymocytes in DO-Tg/ζKO mice.
Peptide-induced positive selection of TCR-Tg/ζKO thymocytes in vitro
In vivo administration of OVA323–339 to DO-Tg/ζKO mice resulted in the generation of CD4+ and CD8+ SP cells expressing Tg-TCR. It is difficult to determine the exact local concentration of peptide in thymus in the in vivo system. Therefore, we also tested whether positive selection of OVA-specific T cells could be induced in vitro. Thymocytes from DO-Tg, DO-Tg/ζKO, and BALB/c mice were cultured in vitro with irradiated splenocytes from BALB/c (H-2d) mice with OVA323–339 for 72 h (Fig. 6). Although thymocyte development of DO-Tg mice was not affected at low doses of OVA323–339, addition of high dose of the peptide induced the reduction of DP population and relative increase of other populations, demonstrating that OVA323–339 mediates negative selection of DO-Tg thymocytes in a dose-dependent manner (23). In contrast, addition of OVA323–339 induced the appearance of CD4+ SP cells in DO-Tg/ζKO thymocytes in a dose-dependent manner. Furthermore, CD8+ SP cells also emerged with high doses of OVA323–339. Because OVA peptide did not modulate the development of BALB/C thymocytes (Fig. 6), and a control peptide (OVA324–334, SQAVHAAHAEI) did not affect at all on the thymocyte development of all of these mice (data not shown), the induction of positive and negative selection was OVA specific. The SP cells from DO-Tg/ζKO cultures treated with OVA323–339 were KJ1-26 positive and HSA-negative (data not shown), suggesting that these SP cells had undergone positive selection and became mature thymocytes.
Moreover, when DP cells were sorted from DO-Tg/ζKO thymocytes and cultured with OVA/APC, mature CD4 and CD8 SP thymocytes emerged (data not shown). Because both CD4 and CD8 SP cells were also KJ1-26-positive, these in vitro results confirmed in vivo results that the addition of OVA compensated for the decrease in TCR density and provided adequate stimulation to positively select DO-Tg/ζKO thymocytes. These data also confirm that lineage commitment was affected by quantity of TCR signals, namely, relatively weak signals induce CD8 fate, whereas strong signals induce CD4 fate.
MHC class I-independent CD8 lineage commitment
Several studies have indicated that CD8 lineage commitment is dependent on the interaction of the TCR with MHC class I molecules and may also involve CD8-dependent signals (8, 13). In β2m-deficient mice, the expression of MHC class I was extremely reduced and positive selection of CD8 T cells was impaired (26). To determine whether CD8+ SP cells selected in the presence of OVA are generated by the MHC class I-CD8 interaction, DO-Tg/ζKO thymocytes were cultured with OVA323–339 and APCs from β2m-KO mice (Fig. 7). In culture with β2m-KO APCs, DO-Tg/ζKO thymocytes exhibit positive selection and the appearance of both CD4+ and CD8+ SP cells similar to those with BALB/c APCs. The culture with OVA in the absence of APC did not induce SP cells (data not shown), indicating that positive selection did not occur by the interaction between TCR and MHC class I on thymocytes. Because β2m-KO APCs did not express MHC class I even after the culture with OVA323–339 (data not shown), it is unlikely that MHC class I on the β2m-KO APCs was induced by contaminating β2m. Collectively, these data indicate that positive selection and CD4/8 lineage commitment of thymocytes bearing DO-Tg TCR is independent of the interaction between MHC class I and CD8, but depends on TCR recognition of the OVA peptide in the context of MHC class II on APCs.
Quantity of TCR signal determines the fate of MHC class I-restricted TCR-Tg thymocytes
Experiments with DO-Tg/ζKO mice have hitherto demonstrated that quantity of TCR signal determines the fate of thymocytes bearing DO TCR on both selection and commitment. We intended to determine whether this rule can be applied to other lines of TCR Tg mice, especially the one expressing MHC class I-restricted TCR. Because class I-restricted TCR-Tg mice exhibit positive selection of CD8+ T cells, we could address the particular question whether increasing TCR signals induce the appearance of CD4+ T cells. Therefore, we examined the T cell development of Tg mice expressing TCR (Vα2/Vβ8.1) specific for the LCMV peptide presented by H-2Db. We crossed LCMV-Tg/β2m-KO mice with ζKO mice to obtain LCMV-Tg mice, which are deficient for β2m and express low TCR levels. Day 15 fetal thymic lobes from LCMV-Tg/β2m−/− ζ+/− or ζ−/− mice were cultured with β2m and various concentrations of LCMV peptide (p33, KAVYNFATM) in the submersion organ culture (22) and T cell development in the thymic lobes was analyzed (Fig. 8). In the thymus from LCMV-Tg/β2m−/− ζ+/− mice, a low dose of the LCMV peptide (1 nM) induced positive selection of CD8+ SP cells, and the highest dose of the LCMV peptide (1 μM) induced negative selection as previously described (7). Furthermore, the proportion of CD4+ cells increased in a dose-dependent manner at the concentration between 1 and 100 nM. In LCMV-Tg/β2m−/− ζ−/− mice, CD8+ SP cells also developed in the presence of the LCMV peptide. However, a 10- to 100-fold higher concentration of the peptide was required for positive selection of CD8+ SP cells. Similar to DO-Tg/ζKO mice, positive selection of CD4+ SP cells was observed at higher concentration of peptide in which CD8+ SP cells were already decreased (Fig. 8). Peptide-induced CD4+ and CD8+ SP cells from LCMV-Tg/ζ−/− (Fig. 9) as well as LCMV-Tg/ζ+/+ mice (Ref. 7 , and data not shown) were stained with clonotype-specific anti-Vα2 mAb, B20.1 (Fig. 9), and anti-Vβ8 mAb, F23.1 (data not shown). The staining and calculated results of Vα2+ thymocytes indicate that both SP lineages express the Tg TCR. Collectively, the data indicate that LCMV-Tg T cells were neglected without the LCMV peptide, positively selected to the CD8 lineage with a low dose, positively selected to CD4 lineage with intermediate dose, and deleted with a high dose of peptide. The fate of LCMV-Tg/ζKO T cells shifted to the one of the LCMV-Tg T cells cultured with a low dose of peptide.
It is generally accepted that the avidity of the interaction between thymocytes and APCs determines the outcome of selection processes (3). In contrast to most of the studies that manipulate the peptide concentration to alter the avidity between TCR and peptide/MHC, we analyzed the consequence of the quantity of TCR signals for T cell fate. We intended to determine whether TCR-Tg thymocytes could be positively selected by altering TCR expression/signaling and whether the thymocytes could be positively or negatively selected in a peptide-specific, dose-dependent manner. In our system, T cells from TCR-Tg/ζ-KO mice express a low level of the surface TCR and positive selection of SP thymocytes were diminished. In vivo administration of an appropriate dose of OVA restored efficient positive selection and an excessive dose promoted negative selection. Low level of HSA and high level of CD69 confirmed that these SP cells had been positively selected. In addition, we found that, within the range of positive selection, low dose of peptide induced commitment to the CD8 lineage and high dose resulted in commitment to the CD4 lineage. We showed that both positive selection and lineage commitment were dependent on the interaction of class II-restricted TCR with MHC class II but not class I. These results demonstrate that the quantity of TCR signal directly regulates positive and negative selection, and further indicate that weaker signals direct positively selected T cells to the CD8 lineage and stronger signals to the CD4 lineage.
In this study, we demonstrated that the decrease in TCR signaling caused by the CD3ζ deficiency resulted in the failure of positive selection in DO-Tg/ζKO mice. Although T cells in DO-Tg mice use both CD3ζ and CD3η in the CD3 complex, those in DO-Tg/ζKO mice lack only CD3ζ and retain CD3η. It has been demonstrated that there is not any significant difference in quality between signals through CD3ζ and CD3η, especially on T cell selection and development, as revealed by studies using CD3ηKO mice (27, 28) and CD3η-transfected hybridomas (29, 30), although the functional efficiency of the η-chain for cell surface expression of TCR complex and signal transduction appears to be lower than ζ. Because it has been shown that the number of ITAMs of ζ affects thymocyte selection in quantitative fashion (48), the difference may depend on less number of ITAM in η. Therefore, it is unlikely that a distinct signal through CD3η induced positive selection in DO-Tg mice compared with neglect in DO-Tg/ζKO mice, although we cannot exclude the possibility that quantitative difference between ζ and η might induce some qualitative difference in signaling. Furthermore, previous studies have demonstrated that the crucial role of the ζ-chain on thymocyte development is to express the cell surface TCR at high levels, rather than its signaling property (18). This was shown by introduction of a ζ transgene with a truncated cytoplasmic tail into ζKO mice. The “tailless” ζ transgene restored normal TCR level and thymocyte maturation. Therefore, it is most likely that T cell selection in DO-Tg/ζKO mice reflects a quantitative change in TCR signaling by the reduced TCR expression but not a qualitative difference.
Previously, we have reported a similar shift in selection from neglect to positive selection in HY-specific TCR-Tg mice crossed with ζKO mice (HY-Tg/ζKO) (31). In female HY-Tg/ζKO mice, HY-specific T cells were neglected, whereas those of female HY-Tg mice were positively selected. In male HY-Tg/ζKO mice, HY-specific T cells, which are autoreactive, were positively selected. In this system, because the amount of endogenous HY-Ag and MHC were fixed, and only TCR signals were reduced, these results imply that T cell selection is controlled by the magnitude of TCR signals. The present studies using DO-Tg/ζKO mice extended the analysis to another TCR-Tg system in which Ag concentration and the surface TCR level could be manipulated simultaneously. We administered OVA peptide into DO-Tg/ζKO mice, which is a high affinity agonist for DO-TCR. In DO-Tg mice, OVA323–339 induced negative selection of DO-Tg thymocytes in vivo (20) and in vitro (23). Our finding that administration of OVA323–339 into DO-Tg/ζKO mice resulted in positive selection of CD4+ SP DO-Tg thymocytes in a dose-dependent manner, and in negative selection at the higher concentrations, implies that selection depends upon the quantity of TCR signal. Although TCR ligation by a very low number of peptide-MHC complexes results in neglect, the same peptide-MHC complexes result in positive selection and then negative selection as the peptide concentration increases. These results support the avidity model in the light of the quantity of TCR signals. Similar results obtained from LCMV-Tg/ζKO thymocytes suggest that, regardless of the MHC-restriction specificity of TCR, differentiation of mature CD8+ and CD4+ thymocytes can be induced by the appropriate avidity of stimulation.
Various experiments have demonstrated that CD4 and CD8 coreceptors play critical roles for lineage commitment through its functional association with p56lck (32, 33), which is involved in early thymocyte development (34, 35). Models have proposed that coengagement of TCR and coreceptor to MHC conveys lck to activate downstream signaling. The developmental link between MHC specificity and coreceptor expression as shown by the analysis of TCR-Tg mice has been explained by an instructional model (8, 9). However, the recent findings demonstrating the existence of coreceptor transitional cells, CD4+CD8low in class II-deficient (11) and CD4lowCD8+ in class I-deficient mice (12), is inconsistent with the instructional model and has been offered as an evidence for a stochastic model, and these populations are thought to be lineage-committed intermediates in positive selection.
Recently, studies have challenged the validity of the stochastic model. Analysis of coreceptor transitional cells in vivo (36) and in vitro (37) revealed that CD4+CD8low population is composed of a mixture of CD4-committed and CD8-committed thymocytes, whereas the CD4lowCD8+ population exclusively consists of CD8-committed cells. Similar conclusions were reached using a coreceptor re-expression assay (13). Appearance of CD8-committed cells among CD4+CD8low or CD4lowCD8+ thymocytes was completely dependent on prior instructional TCR-class I MHC interactions, whereas the appearance of CD4-committed cells among CD4+CD8low thymocytes depends on prior signals through the CD3 complex (13, 38). Furthermore, several studies using coreceptor-deficient mice showed that coreceptors do not possess essential signaling properties required for T cell survival and development (14, 15, 16). These studies implied that signals through TCR but not coreceptors influence the lineage decision.
In this study, when OVA peptides were administered in DO-Tg/ζKO mice, low to mid-range concentrations of OVA induced the generation of SP mature thymocytes expressing the Tg-TCR in a dose-dependent manner. Within the range of Ag-mediating positive selection, lower doses induced predominantly CD8+T cells, whereas higher doses induced CD4+ T cell development. To understand the mechanism of our findings, we first considered whether our data could be explained in the context of the stochastic model. Because DO-Tg/ζKO mice expresses normal levels of coreceptor and MHC molecules, the stochastic loss of one coreceptor triggered by TCR engagement should only leads to CD4+ T cells survival and complete differentiation. This does not explain why a certain number of CD8+ T cells would be positively selected with only low doses of peptide. It is also difficult to explain the data with an instructional model. Because CD8 was not used in the recognition of MHC class II in DO-Tg/ζKO mice, it cannot instruct CD8 lineage commitment by coengagement with TCR complex. In addition, it is unlikely that processed OVA peptide presented on MHC class I interacted DO TCR and CD8 and then promote positive selection because the peptide could not be presented on MHC class I even when processed (39). If the engagement of CD8 and MHC class I could provide an instructive signal for positive selection, CD8+ T cells would not appear with OVA plus APCs from β2m-KO mice. Thus, our results are inconsistent with the original instructional model.
Therefore, we propose a revised instructional model in which the intensity of signals per se delivers to DP thymocytes through the TCR complex determines the lineage choice. In this “quantitative instructional model,” the net quantity of TCR signal directs the CD4/CD8 lineage choice. Moderate intensity of TCR signal leads to positive selection, and, in this process, signals of a stronger and weaker intensity promote CD4 and CD8 lineage commitment, respectively. In this case, CD4 and CD8 coreceptors do not play any obligatory role in signaling and generally stabilize TCR-peptide/MHC interactions (40, 41, 42, 43). For individual thymocytes, the quantity of TCR signal varies to a certain extent upon triggering with a particular concentration of Ag. Such signaling heterogeneity including duration of signals may explain the overlapping between positive selection of CD4+ SP cells and negative selection of DP cells at higher dose of Ag in ζ−/− mice. However, the overall fate of thymocytes changes from neglect to CD8 lineage commitment, CD4 lineage commitment and negative selection in this order. It is not clear that the signals generated by the TCR complex directly promote termination of one coreceptor synthesis or render thymocytes competent to respond to other lineage-specific signals such as Notch (44). These data are also consistent with the observation that the maturation of CD4 lineage requires higher signaling threshold than CD8 maturation (45) and that the maturation of CD4 SP cells is more sensitive to diminished TCR level (46).
A similar quantitative instructional model has been proposed to explain why a few but distinct proportion of DO-Tg CD8+ T cells exist in DO-Tg mice and increase when CD4 is absent (14). In some circumstances, the DO-Tg thymocytes received suboptimal TCR signals and developed into the CD8 lineage, and, when CD4 was not available, the weaker signals promoted CD8 lineage commitment. Furthermore, in CD8-deficient class I-restricted TCR Tg mice, an increase of the selecting ligand can overcome the need for coreceptor involvement and promote positive selection of CD8 T cells (15, 16). Similarly, in ζKO mice, CD4+ SP cells were predominantly reduced as compared with CD8+ SP cells (19) (Fig. 1 b). This is also explained by a quantitative instructional model where development of CD4+T cells requires a stronger signal and is mainly affected by a reduction in TCR expression. A similar mechanism has also been proposed from the analysis of in vitro commitment induced by PMA/ionophore stimulation, in which weak stimulation induced CD8 SP and stronger stimulus induced CD4 SP thymocytes (47).
Our results provide a strong evidence that the quantity of TCR signals directly regulates not only positive and negative selection but also lineage commitment. Molecular mechanisms to determine the relationship between signal intensity and distinct selection have to be solved.
Note added in proof.
A recent report by Yasutomo et al. (49) described that CD4/CD8 lineage commitment is determined by the duration of TCR signaling, which is consistent with our conclusion.
We thank Dr. D. Y. Loh for DO-Tg mice, Drs. Hozumi and S. Habu for helping FTOC, Drs. K. Takase and J. Suzuki for experimental help, and Dr. H. Ohno for discussion. We are also grateful to M. Sakuma and R. Shiina for technical assistance, and H. Yamaguchi and Y. Kurihara for secretarial assistance.
This work was supported by Grants-in-Aids for Scientific Research from the Ministry of Education, Science, and Culture and from the Agency for Science and Technology and a grant from the Human Frontier Scientific Program.
Abbreviations used in this paper: DP, double positive; β2m, β2-microglogulin; FTOC, fetal thymus organ culture; ITAM, immunoreceptor tyrosine-based activation motif; KO, knockout; LCMV, lymphocytic choriomeningitis virus; SP, single positive; DN, double negative; HSA, heat-stable Ag; Tg, transgenic.