Altered positive selection due to corecognition of floppy peptide/MHC II conformers supports an integrative model of thymic selection

Abstract

Thymocytes bearing the Eα52-68/I-A^b complex-specific 1H3.1 αβ T cell antigen receptor are positively selected in Ab-Ep [Ab-Ep transgenic, invariant chain (Ii)^−/−, I-Aβ^b−/−] mice, where I-A^b molecules present only Eα52-68. Although Ii reintroduction led to deletion, I-Aβ^b reintroduction disrupted positive selection. T cell antigen receptor transgenic Ab-Ep I-Aβ^b+ mice had a large thymus with an increased absolute number of CD4⁺CD8⁺ cells and no overt signs of deletion. Unlike Ab-Ep Ii⁺ antigen-presenting cells, Ab-Ep I-Aβ^b+ antigen-presenting cells did not activate 1H3.1 T cells. However, their capacity to present Eα52-68 was intact. Thus, positive selection of 1H3.1 thymocytes on the tight compact Eα52-68/I-A^b complex is neutralized by the corecognition of loose compact self-peptide/I-A^b conformers that do not interfere with the cognate activation of mature 1H3.1 T cells. The data support the notion that the integration of distinct signals generated by the simultaneous recognition of multiple self-peptide/MHC complexes directs intrathymic selection of T cells.

Thymocyte deletion (1) ensures that self epitopes expressed on professional antigen-presenting cells (APCs) will not activate autoreactive T cells when presentation of nonself antigens occurs under conditions of microbial product-induced high costimulation. Deletion of autoreactive T cells relies on recognition of self-peptide/self-MHC complexes. Whether self recognition is involved in positive selection of thymocytes (2, 3) or whether this process relies on a peptide-independent T cell antigen receptor (TCR)/MHC interaction has been a matter of debate (4–6). Studies based on natural MHC class I mutations affecting residues critical for peptide binding but not for TCR/MHC contact (7, 8) and the culture of fetal thymic lobes from MHC class I-deficient mice (9–13) have shown a role for peptide recognition in CD8⁺ T cell-positive selection. Evidence for peptide specificity in maturation of MHC class II-restricted thymocytes came from the analysis of mice with genetically manipulated antigen processing/presentation pathways. For instance, mice expressing only the Eα52-68/I-A^b complex (14, 15) or H-2M^−/− mice that express mainly the CLIP/I-A^b complex (16–18) and few other complexes (19) have a reduced number of CD4⁺ T cells with many of them being reactive to syngeneic APCs. In addition, multiple CD4⁺ T cell specificities identified by immunization of normal mice could not develop in H-2M^−/− thymi (19–23). The altered maturation of CD4⁺ T cells in mice with a reduced self complexity only on the thymic epithelium (cathepsin L^−/− mice) (24) or in mice with 95% of their MHC II molecules occupied by one peptide (25), the ability of peptide/MHC class II complex-specific monoclonal antibodies to block maturation of relevant T cells in vivo (23, 26, 27), the selection of different T cell specificities on intrathymic delivery of invariant chain (Ii)/peptide fusion proteins (28) and the restricted TCR α chain repertoire in mature T cells from single β chain TCR transgenic (Tg) mice (29) constituted independent demonstrations of self-peptide recognition during positive selection of CD4⁺ T cells. These studies and others (30–32) established that intrathymic positive selection of conventional αβ T cells and therefore, the generation of the mature αβ T cell repertoire is, in essence, self-referential (33).

We have observed that αβ T cells bearing two distinct antigen receptors can respond to nonself antigens by engaging a TCR intrinsically unable to support their intrathymic maturation. This is shown by the fact that KB TCR Tg T cells can mature intrathymically and populate the periphery only if they can express an alternative TCR (34). This dependency could reflect the inability of the KB TCR to interact with self-peptide/MHC II complexes on thymic stromal cells. Alternatively, the KB TCR could engage in interactions with self-peptide/MHC complexes, but such interactions may translate into a weak intensity signaling not permissive for positive selection. Thus, summing the signals generated on engagement of two TCRs could allow immature KB T cells to reach the signaling threshold required for maturation. The latter hypothesis, synergistic in nature, points to an integrative signaling model of intrathymic T cell development where the summation of distinct TCR-derived signals determines the fate of maturing T cells.

In line with this idea, we report here that the corecognition of a self-peptide/MHC II complex able to support positive selection of a given T cell specificity and of additional complexes characterized by a markedly distinct conformational signature appears to antagonize positive selection of thymocytes bearing the same specificity in vivo.

Methods

Animals.

C57BL/6 (B6) and B10.A(5R) (5R) mice were from the The Jackson Laboratory. 1H3.1 TCR Tg mice were described previously (23, 35). Ab-Ep mice (14) were a gift of P. Marrack (Howard Hughes Medical Institute, National Jewish Medical and Research Center, Denver) and were Ab-Ep Tg^± Mtv7⁻ in this study. Mice used were 6–10 weeks old and were genotyped by using the following primers (all 5′ to 3′): 1H3.1 TCR; (35), I-Aβ^b; ACGGACGTGGGCGAGCAC and GAGAACGAGGGAAATGACAG, I-Aβ^b null allele; AGCGACGTGGGCGAGCAC and AGGTGAGATGACAGGAGATC, Ii; GTGCAGCCGTGGAGCTCTGTA and CAAGGAGCATGTTATCCATGG, Ii null allele; CGTGCTTTACGGTATCGCCGC and CAAGGAGCATGTTATCCATGG, Ab-Ep Tg; CTAGCTTTGAGGCTCAGG and GTAGTTGTGTCTGCACAC.

Immunostainings.

For multicolor staining, cells were incubated with labeled mAb for 30 min on ice, washed, and analyzed immediately. For two-step staining, cells were incubated first with purified mAbs followed by a F(ab′)₂ fragment of goat anti-mouse Ig-FITC conjugate from Sigma. The mAbs used were anti-Vβ6-FITC (RR4-7), anti-Vα2,3.2,8,11-FITC (B20.1, RR3-16, B21.14, RR8-1), anti-CD4-CyChrome (H129.19), anti-CD24-FITC (M1/69), anti-CD69-phycoerythrin (PE)/FITC (H1.2F3), anti-CD45/B220-PE (RA3-6B2) from BD-PharMingen (San Diego, CA), and anti-CD8aPE/FITC (53-6.7) from Invitrogen Life Technologies. The Y3JP (mouse IgG_2a, anti-I-A^b) and Y17 (mouse IgG_2b, anti-I–E) (36), 25.9.17 (mouse IgG_2a, anti-I-A^b) and 14.4.4S (mouse IgG_2a, anti-I-E) (37), Y-Ae (mouse IgG_2b, anti-I-A^b+Eα) (38), GK1.5 (rat IgG_2b, anti-CD4), TIB 105 and TIB 210 (both rat IgG_2b, anti-CD8) mAbs were affinity purified. A FACScan flow cytometer and cellquest software (Becton Dickinson) were used to collect and analyze the data. Nonviable cells were excluded by electronic gating by using propidium iodide.

Functional Assays.

For proliferation assay, lymph node-derived T cells were cultured in U-bottom 96-well plates (Becton Dickinson) 3–4 days at 37°C in Click's EHAA medium (Irvine Scientific) supplemented with 5% heat-inactivated FCS, 5 × 10⁻⁵ M 2-mercaptoethanol, and antibiotics. T cells (20–50 × 10³ per well) were stimulated with irradiated B6 APCs (3 × 10⁵ per well, 2,000 rad) plus serial dilutions of synthetic Eα52-68 peptide (ASFEAQGALANIAVDKA). The I-A^b binding Integrin β1.778–794 peptide (39) was occasionally used as control. One microcurie (1 Ci = 37 GBq) of [³H]thymidine per well was added during the last 12 h. The plates were harvested, and cpm were determined. For inhibition experiments, purified mAbs were added to microcultures. Ltk− cells transfected with I-A^bα and I-Aβ^b or I-A^bα and Ab-Ep cDNAs (MARS cells) were mitomycin C-treated before stimulation assay. IL-2 production by T cell hybrids was assessed by using the IL-2-dependent CTLL cells.

mAb Treatment of Newborn Mice.

Newborn mice were typed by PCR on day 1 by using genomic DNA. Starting on day 2, mice were i.p. injected with 50 μg of purified mAb diluted in 50 μl saline every 2 days. On days 13–15, mice were killed for analysis.

Apoptosis Detection.

Thymocytes were triple-stained by using anti-CD4/CD8 mAbs and annexin V-FITC (BD PharMingen). The FSC/SSC gate used was designed to exclude dead cells after running the respective samples incubated with propidium iodide only.

Nonhematopoietic Thymic Stromal Cell Analysis.

Day 15 fetal thymic lobes were kept in organ culture with RPMI media supplemented with 10% FCS and 1.35 mM 2′-deoxyguanosine (Sigma) in transwell plates (Costar) for 6–7 days according to Coligan et al. (40). The remaining tissues were mechanically dissociated and sequentially digested with Liberase Blendzyme 1 (Roche Molecular Biochemicals) and trypsine/EDTA. The cells were processed for fluorescence-activated cell sorting analysis. Large cells (FSC^high) were analyzed after electronic gating.

Results

Positive-to-Negative Conversion of Intrathymic Selection in 1H3.1 TCR Tg Ab-Ep Mice on Ii Reintroduction.

Mice with I-A^b molecules occupied only by the Eα52-68 peptide (Ab-Ep mice) (14) are able to positively select immature T cells bearing the Eα52-68/I-A^b complex-specific 1H3.1 αβ TCR. Such selection could be inhibited in vivo by the Y-Ae mAb that also react to the Eα52-68/I-A^b complex (38, 42–44), indicating that it relies on direct recognition of the Eα52-68/I-A^b complex (41). In Ab-Ep mice, the conditions for the sole presentation of Eα52-68 are the attachment of the peptide to I-Aβ^b chain (Ab-Ep Tg) by a linker and the double deficiency in Ii and endogenous I-A^bβ molecules (14, 45). During the derivation of 1H3.1 TCR Tg Ab-Ep Tg Ii^−/− I-Aβ^b−/− mice (termed TCR Ab-Ep), we analyzed Ii⁺ TCR/Ab-Ep double Tg mice (TCR Ab-Ep Ii⁺). Ii reintroduction reverted the positive selection profile of TCR Ab-Ep mice to the intrathymic deletion profile typically seen in 1H3.1 Tg mice with a functional I-Eα gene [e.g., TCR Tg 5R or TCR/I-Eα double Tg mice (46)]. The thymic cellularity was very low (3–7 × 10⁶), there were few CD4⁺CD8⁻ β6^high cells, virtually no CD4⁺CD8⁺ cells, and Vβ6⁺ cells were mainly CD4⁻ D8⁻ (Fig. 1A). The few peripheral Vβ6⁺ cells were essentially CD4⁻ D8⁻ (Fig. 1B). Thus, Ii reintroduction, whose expression disrupts the linker attaching Eα52-68 to I-Aβ^b (45), imposes deletion in the TCR Ab-Ep thymus.

Figure 1.

Ii reintroduction in TCR Ab-Ep mice converts positive into negative intrathymic selection. Analysis of thymocytes (A) and splenocytes (B) from TCR Ab-Ep and TCR Ab-Ep Ii⁺ mice. (Center) The Vβ6 histograms. (Left and Right) The CD4/CD8 distribution with and without gating on Vβ6^high cells. Quadrant statistics are indicated. In this analysis, the cellularity was: TCR Ab-Ep, 75 × 10⁶; TCR Ab-Ep Ii⁺, 3.8 × 10⁶. The deletion pattern is representative of 12 TCR Tg Ab-Ep Tg Ii⁺ mice analyzed.

Reintroduction of I-Aβ^b into TCR Ab-Ep Mice Alters Intrathymic Positive Selection.

The phenotype of mice with a functional I-Aβ^b gene (TCR Ab-Ep I-Aβ^b+) markedly contrasted with those of both TCR Ab-Ep and TCR Ab-Ep Ii⁺ mice. TCR Ab-Ep I-Aβ^b+ thymi did not show signs of deletion because their cellularity was nearly identical to that of TCR Ab-Ep thymi, and CD4⁻ D8⁻ thymocytes did not accumulate. There were few CD4⁺CD8⁻ β6⁺ cells, and all thymocytes remained Vβ6^low. Most importantly, the absolute number of CD4⁺CD8⁺ cells was greatly increased (Fig. 2). In addition, although CD4⁺CD8⁺ cells undergoing induced apoptosis reduce their CD8 expression level (47), TCR Ab-Ep I-Aβ^b+ CD4⁺CD8⁺ thymocytes remained CD8^high (Fig. 2). In the periphery, Vβ6⁺ cells were scarce and many were CD8⁺ (Fig. 3A). These features suggested a deficiency in positive selection. For instance, 1H3.1 TCR Tg H-2Mα^−/− mice, which also show a deficient positive selection, display a rather large thymus and few CD4⁺Vβ6^high thymocytes (22). Finally, the analysis of TCR Ab-Ep Ii⁺ I-Aβ^b+ mice showed that the deletion imposed by Ii was dominant over the interference caused by I-Aβ^b (Fig. 2). The altered positive selection in TCR Ab-Ep I-Aβ^b+ mice implies that the corecognition of Eα52-68 independent self-peptide/I-A^b complex(es) expressed in these mice neutralizes the positive selection signal generated on recognition of the Eα52-68/I-A^b complex that is still expressed in its covalent configuration, because these mice remain Ii^−/−.

Figure 2.

I-Aβ^b reintroduction in TCR Ab-Ep mice disrupts intrathymic positive selection. Shown is analysis of TCR Ab-Ep (positive selection), TCR Ab-Ep I-Aβ^b+ (altered positive selection), TCR I-Aβ^b−/− (neglect), TCR I-Aβ^b+ Ii^−/− (deficient positive selection), and TCR Ab-Ep I-Aβ^b+ Ii⁺ (deletion) thymocytes. Data are representative of 6, 10, 5, 7, and 3 mice, respectively. The cell counts were: TCR Ab-Ep, 65 × 10⁶; TCR Ab-Ep I-Aβ^b+, 62.5 × 10⁶; TCR I-Aβ^b−/− Ii^−/−, 180 × 10⁶; TCR I-Aβ^b+ Ii^−/−, 145 × 10⁶; TCR Ab-Ep I-Aβ^b+ Ii⁺, 5.2 × 10⁶. Note that those numbers translate into an increased absolute number of CD4⁺CD8⁺ cells (≈50%) in TCR Ab-Ep I-Aβ^b+ mice relative to TCR Ab-Ep mice.

Figure 3.

Arrest of thymocyte maturation at the CD4⁺CD8⁺ stage in TCR Ab-Ep I-Aβ^b+ mice. (A) Analysis of TCR Ab-Ep and TCR Ab-Ep I-Aβ^b+ splenocytes. (B) CD69 expression by and FSC of CD4⁺CD8⁺ thymocytes from TCR Ab-Ep (positive selection), TCR Ab-Ep I-Aβ^b+ (altered positive selection), TCR I-Aβ^b−/− (neglect), and TCR I-Aβ^b+ Ii^−/− (deficient positive selection) mice. Cell counts were: 74.5 × 10⁶, 158 × 10⁶, 76 × 10⁶, and 112 × 10⁶, respectively. In this analysis, the increase in the absolute number of CD4⁺CD8⁺ in Ab-Ep I-Aβ^b+ mice relative to TCR Ab-Ep mice was ≈40%. Note the lack of CD4^highCD8^high cells in the TCR Ab-Ep I-Aβ^b+ thymus. Mean fluorescence values for CD69 expression and percent of enlarged cells are indicated. Data are representative of three experiments. (C) Apoptosis among CD4⁺CD8⁺ thymocytes from TCR Ab-Ep (positive selection), TCR I-Aβ^b−/− (neglect), TCR Ab-Ep I-Aβ^b+ (altered positive selection), and TCR Ab-Ep Ii⁺ (deletion) mice. Annexin V log fluorescence was plotted as histograms after gating on CD4⁺CD8⁺ cells.

Immature Thymocytes Accumulating in TCR Ab-Ep I-Aβ^b+ Mice Are CD4^lowCD8^highTCR^lowCD69^low.

Unlike TCR Tg I-A^bβ^−/− mice where no intrathymic maturation takes place due to neglect, there were no true CD4CD8 double-positive (CD4^highCD8^high) thymocytes in TCR Ab-Ep I-Aβ^b+ mice (Figs. 2 and 3B). Rather, virtually all cells were CD4^lowCD8^highTCR^low, displayed an increased expression level of CD69 relative to true CD4⁺CD8⁺TCR^low cells from TCR I-Aβ^b−/− mice, and a sizable fraction of them were enlarged (Fig. 3B). Because CD69 induction on immature T cells is associated TCR engagement (48–50), this result indicated that accumulating CD4^lowCD8^high cells have been establishing TCR/self-peptide/I-A^b complex interactions. Under conditions of arrested positive selection, one could expect the level of apoptotic activity among CD4⁺CD8⁺ thymocytes to be in the range of that induced by neglect. We found that apoptosis among CD4⁺CD8⁺ TCR Ab-Ep I-Aβ^b+ thymocytes was indeed comparable to that seen under conditions of neglect and clearly lower than that characterizing residual CD4⁺CD8⁺ thymocytes undergoing deletion (TCR Ab-Ep Ii⁺ mice) (Fig. 3C). Thus, the corecognition of self-peptide/MHC II complex(es) present in TCR Ab-Ep I-Aβ^b+ mice appears to inhibit the Eα52-68/I-A^b-driven positive selection of immature 1H3.1 T cells. This interference translated into an increased absolute number (usually close to 50%), that is an accumulation, of CD4⁺CD8⁺ cells.

Ab-Ep Ii⁺ but Not Ab-Ep I-Aβ^b+ APCs Specifically Activate 1H3.1 T Cells.

Because 1H3.1 T cells undergo deletion in Ab-Ep Ii⁺ thymi, we reasoned that Ab-Ep Ii⁺ APCs may activate mature 1H3.1 T cells. Alternatively, this may not be the case, because the sensitivity to TCR engagement by mature and immature T cells is known to differ by one to two orders of magnitude (51–55). Unlike Ab-Ep APCs, Ab-Ep Ii⁺ APCs were able to induce proliferation of TCR Tg Rag-1^−/− cells (Fig. 4A). The response was comparable to that induced by 5R APCs and was observed both for 1H3.1 TCR Tg T cells and the original 1H3.1 hybrid (Fig. 4C). In contrast, Ab-Ep I-Aβ^b+ APCs did not induce any detectable response (Fig. 4A). The response to Ab-Ep Ii⁺ APCs was specific because blockable by Y-Ae and the I-A^b specific, Y3JP mAb but not by 25.9.17, which reacts to multiple peptide/I-A^b complexes but not to the Eα52-68/I-A^b complex (39) (Fig. 4B). The use of TCR Tg Rag-1^−/− T cells indicated that both the response and the lack of the response observed were independent of endogenous TCR α chains. Thus, Ii, but not I-Aβ^b, reexpression by Ab-Ep APCs leads to the specific activation of 1H3.1 T cells. Ii expression was a dominant factor, because Ab-Ep Ii⁺ I-Aβ^b+ APCs activated 1H3.1 T cells (not shown). Finally, we earlier observed that 1H3.1 T cells can specifically react to the fibroblast-like Ltk− cells cotransfected with the I-Aα^b and I-Aβ^b-Ep cDNAs (MARS cells) but not to Ltk− cells expressing wild-type I-Aαβ^b (not shown). Despite the fact that Ltk− cells are assumed not to have a hematopoietic origin, the data obtained with Ab-Ep Ii⁺ APCs and the detection of Ii in non-APCs such as T cells, NK cells, tumor cells (56), and colon epithelial cells (57) prompted us to search for Ii expression. RT-PCR analysis revealed that the Ii transcripts were detectable in MARS cells (not shown).

Figure 4.

Ab-Ep Ii⁺, but not Ab-Ep I-Aβ^b+, APCs specifically stimulate 1H3.1 T cells. (A) Proliferative response of 1H3.1 TCR Tg Rag-1^−/− T cells to Ab-Ep, Ab-Ep I-Aβ^b+ (Ab-Ep I-Ab+), and Ab-Ep Ii^± (Ab-Ep Ii+) irradiated splenocytes. 5R cells were used as positive control. (B) 1H3.1 T cell reactivity to Ab-Ep Ii⁺ APCs is inhibited by Y3JP and Y-Ae but not by Y17 or 25.9.17 mAbs. (C) The original 1H3.1 hybridoma recapitulates the response of 1H3.1 TCR Tg T cells to Ab-Ep, Ab-Ep I-Aβ^b+, and Ab-Ep Ii^± APCs. The values represent the mean ± SD of duplicate microcultures. The data are representative of four (A), two (B), and three (C) experiments.

Ab-Ep Ii⁺ and Ab-Ep I-Aβ^b+ APCs Coexpress Distinct Eα52-68-Independent, Self-Peptide/I-A^b Complexes.

In Ab-Ep APCs, Ii reexpression makes the CLIP region of Ii and the covalently linked Eα52-68 peptide compete for the I-A^b peptide groove in the endoplasmic reticulum. The linker is therefore exposed to proteases when the complexes reach the endocytic compartments due to the targeting motif of Ii (58). This causes surface expression of I-A^b molecules presenting noncovalently associated Eα52-68 peptide as well as endogenous peptides, because the peptide exchange factor H-2M is present. For instance, the frequency of Y-Ae⁺ complexes in Ab-Ep construct-transfected B cells is ≈35% of all MHC class II molecules (45). Apart from the expected Y3JP⁺ staining, both Ab-Ep Ii⁺ and Ab-Ep I-Aβ^b+ APCs revealed a 25.9.17⁺ profile indicative of the presentation of Eα52-68-independent peptides bound to I-A^b (Fig. 5A). The 25.9.17 signal on Ab-Ep I-Aβ^b+ APCs was comparable to that on regular B6 APCs, and repeatedly lower for Ab-Ep Ii⁺ APCs. The Y-Ae staining was significantly reduced on Ab-Ep Ii⁺ APCs but was comparable on Ab-Ep I-Aβ^b+ and Ab-Ep APCs. The Y-Ae signal was also comparable on 2′-deoxyguanosine-resistant (presumably epithelial) thymic stromal cells from Ab-Ep and Ab-Ep I-Aβ^b+ mice (Fig. 5B). Regarding the Y3JP staining, the hierarchy was as follows; B6 > Ab-Ep I-Aβ^b+ ≫ Ab-Ep >/= Ab-Ep Ii⁺. MARS cells were also 25.9.17⁺ indicating that, similar to Ab-Ep Ii⁺ APCs, they coexpress Eα52-68/I-A^b and other complexes.

Figure 5.

Surface expression of loose compact (25.9.17⁺) self-peptide/I-A^b complexes by MHC class II⁺ cells and their capacity to interfere with positive selection of 1H3.1 thymocytes under the Ab-EP Tg I-Aβ^b+ conditions in vivo. (A) Analysis of B220⁺ splenocytes from B6 (I-A^bu/I-Eα⁻), Ab-EP, Ab-EP Ii, and Ab-EP I-A^bβ⁺ mice for surface expression of the Y-Ae, Y3JP, and 25.9.17 epitopes. Control histograms (open) were obtained by using anti-I-E (Y17 or 14.4.4S) mAb. (B) Y-Ae epitope expression by 2′-deoxyguanosine-resistant Ab-Ep and Ab-Ep I-A^b+ thymic cells. (C) Thymocytes from 12- to 15-day-old TCR Ab-Ep I-Aβ^b+ mice treated with 25.9.17 versus the isotype-matched anti-I-E 14.4.4S mAb. The data are representative of three experiments. (D) CD4⁺CD8⁻ thymocytes emerging on 25.9.17 infusion display a mature phenotype. Vβ6 and CD24/HSA expression levels (mean fluorescence indicated) as well as the expression of Vα (2, 3.2, 8.3, 11.1/.2) (percent of positive cells indicated) were plotted after gating on CD4⁺CD8⁻ cells. For comparison, the CD24/HSA fluorescence intensity on CD4⁺CD8⁺ cells was 2,014 in TCR Ab-Ep mice and 2,102 in 25.9.17-treated TCR Ab-Ep I-Aβ^b+ mice.

In Ab-Ep I-Aβ^b+ APCs, the assembly of 25.9.17⁺ complexes results from competition between Ab-Ep and I-Aβ^b for pairing with I-Aα^b but due to lack of Ii, most complexes do not traffic to the endocytic compartments (59). Thus, Ab-Ep Ii⁺ and Ab-Ep I-Aβ^b+ APCs share a Y-Ae⁺ Y3JP⁺ 25.9.17⁺ phenotype that nevertheless corresponds to coexpression of distinct forms of the Eα52-68/I-A^b complex and multiple Eα52-68-independent self-peptide/I-A^b complexes. The Eα52-68-independent peptide/I-A^b complexes expressed on Ab-Ep I-Aβ^b+ APCs correspond to complexes expressed on Ii^−/− APCs. That is, I-A^b molecules displaying a reduced mobility in SDS/PAGE analysis. These were termed loose compact or “floppy” αβ dimers as opposed to the tight compact dimers adopting a high-mobility profile (60–62), of which Eα52-68/I-A^b is an example (39). Loose compact I-A^b dimers are well recognized by 25.9.17 (39). It is believed that MHC class II molecules adopting this conformation are essentially bound by low-affinity peptides (61). The injection of 25.9.17 to newborn TCR Ab-Ep I-Aβ^b+ mice allowed some CD4^highCD8⁻ T cells with a mature phenotype (TCR^high, HSA^low) to emerge intrathymically (Fig. 5 C and D), showing that loose compact self-peptide/I-A^b complexes effectively contribute to altered positive selection in TCR Ab-Ep I-Aβ^b+ mice. The rescued CD4^highCD8⁻ thymocytes were functional because they could react to Eα52-68 stimulation (not shown).

Ab-Ep I-Aβ^b+ APCs Normally Activate 1H3.1 T Cells on Presentation of Eα52-68 Peptide.

To test whether the coexpression of loose compact self-peptide/I-A^b complexes could interfere with the specific activation of mature 1H3.1 T cells, we used Ab-Ep I-Aβ^b+ splenic APCs loaded with Eα52-68. Under such conditions, naïve TCR Tg Rag-1^−/− T cells readily proliferated (see Fig. 6, which is published as supporting information on the PNAS web site, www.pnas.org.) even at the lowest peptide doses. The response was Y-Ae blockable (not shown). In addition, despite their reduced MHC II expression level (60–62), Eα52-68 presentation to naïve 1H3.1 T cells by Ii^−/− APCs was slightly better than that by B6 APCs including at very low doses of Eα52-68 (Fig. 6). Thus, the corecognition of loose compact self-peptide/I-A^b complexes, which appears able to neutralize positive selection of immature 1H3.1 T cells had no detectable antagonistic effect on the activation of their mature counterparts.

Discussion

Mice with I-A^b molecules presenting only Eα52-68 (Ab-Ep) positively select immature T cells bearing the Eα52-68/I-A^b complex-specific 1H3.1 TCR (41). Ii reintroduction caused intrathymic deletion comparable to that seen in TCR Tg/I-Eα⁺ mice (46). In sharp contrast, I-Aβ^b reintroduction inhibited positive selection. In TCR Ab-Ep I-Aβ^b+ mice, the thymic cellularity and CD4⁻ D8⁻ fraction were comparable to those of TCR Ab-Ep mice, and the apoptotic activity among CD4⁺CD8⁺ cells resembled that of 1H3.1 thymocytes under conditions of neglect. These features do not support a significant role for deletion in the phenotype observed. The absolute number of CD4⁺CD8⁺ thymocytes was increased by ≈40–50%. Virtually all cells were CD4^lowCD8^highTCR^lowCD69⁺. The amount of Eα52-68/I-A^b complex expressed in Ab-Ep I-Aβ^b+ mice was nearly comparable to that of Ab-Ep mice. Because Ab-Ep I-Aβ^b+ mice remain Ii^−/−, the Eα52-68/I-A^b complex remains expressed in its covalent configuration. For instance, multiple Eα52-68/I-A^b complex-specific T hybrids unable to react to the covalent configuration of the Eα52-68/I-A^b complex (Ab-Ep APCs) remained totally unresponsive to Ab-Ep I-Aβ^b+ APCs despite the fact that they responded to 5R, Ab-Ep Ii⁺, or Eα52-68-loaded I-A^b+ APCs (not shown). Further, Ii deficiency itself does not preclude activation of 1H3.1 T cells by APCs presenting the noncovalent form of the Eα52-68/I-A^b complex because the response curve of the 1H3.1 T cells to I-A^b+I-E⁺Ii⁻ APCs closely resembles that of the response to I-A^b+I-E⁺Ii⁺ APCs with only a slightly reduced sensitivity despite the markedly reduced I-A^b level (19). Thus, in Ab-Ep I-Aβ^b+ mice, APCs coexpress the covalent configuration of the Eα52-68/I-A^b complex and self-peptide/I-A^b complexes, which are homologous to the self-peptide/MHC class II complexes expressed on Ii^−/− APCs (60–62). The corecognition of these self-peptide/MHC class II complexes alters positive selection of immature 1H3.1 thymocytes driven by recognition of the covalent configuration of the Eα52-68/I-A^b complex.

Certainly, “stealing” of the Eα52-68 peptide by I-Aαβ^b molecules cannot account for the arrested thymocyte maturation in TCR Ab-Ep I-Aβ^b+ mice, because in this instance, the peptide involved in positive selection is physically attached to the I-Aβ^b chain and the Ab-Ep complexes preserved from proteolytic cellular compartments (59). We also rule out antagonism by a low amount of chimeric I-Aβ/I-Eα MHC II molecules, because these mice lack a functional I-Eα gene. Altered positive selection due to a reduced amount of Ab-Ep complexes on stromal cells is not a valid explanation, because 2′-deoxyguanosine-resistant Ab-Ep and Ab-Ep I-Aβ^b+ thymic cells were equally stained by Y-Ae. Further, infusion of the 25.9.17 mAb allowed mature functional CD4⁺CD8⁻ β6⁺ thymocytes to emerge in the TCR Ab-Ep I-Aβ^b+ thymi. As in the case of TCR Ab-Ep mice, the emergence of these CD4⁺CD8⁻ β6⁺ cells was independent of alternate TCR α chain.

Self-peptide/MHC II complexes expressed on Ii^−/− APCs are loose compact (floppy) complexes showing a reduced mobility in SDS/PAGE analysis (60, 61). The band of the tight compact Eα52-68/I-A^b complex corresponds to the very bottom part of the broad Y3JP band that includes all peptide/I-A^b conformers. In contrast, loose compact conformers, which are bound by 25.9.17, generate signals corresponding usually to the middle of the Y3JP band. The band corresponding to the floppy CLIP/I-A^b complex is even positioned at the very top of the Y3JP band (39). The dominance of loose compact self-peptide/self-MHC II complexes on Ii^−/− APCs is also reflected by the fact that, despite the overall reduction of MHC class II expression level, the response of T hybrids specific for the β2m 48–58/I-A^b and LDLr 486–501/I-A^b floppy (i.e., 25.9.17⁺) complexes was increased compared with control APCs. Conversely, the response of T hybrids specific for the tight compact (25.9.17⁻ AAT 394–410/I-A^b complex was reduced (39, 67). Corecognition of loose compact conformers does not affect activation of mature 1H3.1 T cells because Ab-Ep I-Aβ^b+ APCs efficiently present exogenous Eα52-68 peptide to naïve 1H3.1 T cells and I-A^b+I-E⁺Ii⁻ APCs, which constitutively assemble and present the Eα52-68/I-A^b complex efficiently stimulate the 1H3.1 T cell hydrid despite the reduced MHC II expression level imposed by Ii deficiency (19). Complexes able to interfere with positive selection of 1H3.1 thymocytes neither caused (TCR Ii^−/− mice) nor prevented (TCR Ab-Ep Ii⁺ I-Aβ^b+ mice) their intrathymic deletion. Clearly, self-peptide/self-MHC II complexes assembled in the absence of Ii are not deleterious for all CD4⁺ αβ T cells maturation, because thymocytes bearing the HEL46–61/I-A^k complex-specific 3A9 αβ TCR develop in Ii^−/− mice (64), and CD4⁺ T cells are detected in Ii^−/− mice (60–62). Thus, there are circumstances in which CD4⁺ T cells develop normally in vivo despite the abundance of floppy peptide/MHC II conformers.

The notion that the sum of agonistic and antagonistic signals may determine the outcome of intrathymic selection (65) comes from the observation that an antagonist peptide for the F5 TCR inhibits negative selection of F5 TCR Tg T cells in fetal thymic lobes and accelerate positive selection when unopposed by deletion (66, 67). Interference with positive selection of CD4⁺ T cells has also been observed (68, 69). However, most of these studies were conducted in vitro and, in all cases, interference was caused by synthetic mutants of the antigenic peptide engineered to gain antagonistic activity. Inhibition of positive selection by recognition of natural self-peptide(s) has been seen only once; 3A9 TCR Tg thymocytes are selected in H-2^k mice but fail to be so in H-2^k/b mice despite an unchanged cellularity. However, the exact identity of the MHC II molecules (I-Aαβ^b or I-Aβ^b/I-Aα^k) involved in the assembly of the inhibitory complexe(s) remained unknown. Such peptide/MHC II complexe(s) were able to inhibit the activation of mature 3A9 T cells (70).

Thus, TCR Ab-Ep I-Aβ^b+ mice illustrate an unrecognized situation where endogenous self-peptide/MHC II complexes with particular conformational features and no detectable inhibitory activity for mature T cell bearing a given TCR, neutralize positive selection of their immature counterparts in vivo. The data support an integrative model of T cell selection predicting that, during its contacts with stromal cells, a given thymocyte is exposed to multiple self-peptide/MHC II complexes such that its fate is determined by the sum of all signals received (65). Such a model may be useful to describe the permissive role alternate αβ TCRs appear able to exert on T cell-positive selection (34).

Abbreviations

TCR

T cell antigen receptor

Tg

transgenic

APC

antigen-presenting cell

Ii

invariant chain