T cell receptor contact to restricting MHC molecules is a prerequisite for peripheral interclonal T cell competition

We initially compared the ability of different class I–restricted TCR transgenic CD8⁺ T cells to proliferate in a lymphopenic environment using cells from TCR transgenic Rag^−/− mice. As reported by others (9–11), most class I–restricted transgenic cells demonstrate homeostatic proliferation under lymphopenic condition (OT1, 2C, A18, and F5), whereas others (HY) do not (Fig. 1 A and not depicted). For TCRs demonstrating homeostatic proliferation, increasing the dose of transferred cells resulted in a decrease in proliferation (Fig. 1 A). Apparently, when numbers increase, T cells are signaled to slow or cease proliferation. This could be caused by the progressive consumption of a soluble factor present in limited supply, accumulation of an inhibitory factor, and/or by increasing competition for restricting MHC molecules/MHC-bearing cells.

To distinguish between these possibilities, we cotransferred CFSE-labeled OT1 and P14 cells into P14 recipients. OT1 and P14 TCRs differ in the isotype of the restricting MHC molecule. Initially, this experiment was prompted by the idea that a separate niche of peptide–MHC complexes might exist for each clone expressing a given TCR specificity. T cell proliferation within a given niche would be possible so long as the number of T cells occupying it remained low. T cells occupying different niches would not interfere with each other (15, 16). Indeed, K^b-restricted OT1 cells divided in mice that had normal numbers of D^b-restricted P14 cells, whereas cotransferred P14 cells did not divide (Fig. 1 B). Because the only genetic difference between these two T cell populations is their TCR, proliferation of OT1 cells and lack of proliferation of P14 cells must be caused by a difference in TCR signaling. This differential ability to proliferate excludes the possibilities that proliferation is regulated exclusively by a factor becoming limiting, or the accumulation of an inhibitory factor acting on a receptor expressed by T cells or the APCs, as this should affect both populations. If the model proposing TCR-specific niches was correct, the complementary experiment should result in a similar outcome: P14 cells should divide in OT1 hosts. Surprisingly, with a preformed population of OT1 cells, neither P14 nor OT1 donor cells divided (Fig. 1 C). Interestingly, both proliferated in an environment prefilled with F5 cells (Fig. 1 D).

We interpret these results as showing that homeostatic proliferation is not a simple issue of TCR-specific niches being filled. This conclusion relies on the implicit assumption that the OT1 TCR, selected in the thymus on K^b molecules (19), does not cross-react with D^b molecules. This will be addressed experimentally below. However, it is evident that there is a hierarchy in the ability of T cells to undergo homeostatic proliferation within the prefilled environment of a recipient differing in TCR specificity in the order OT1 > P14. Our finding is similar to the hierarchical competition described for OT1 and 2C; these TCRs are, however, both restricted to K^b (17). Our result is contrary to those reported by Troy and Shen despite using the same TCRs. However, these authors used TCR transgenic cells from Rag-proficient mice in which TCR-α chain rearrangement can occur (16).

We found a significant increase in the number of memory T cells (CD44^hi), or recently activated CD62L⁻ T cells (not depicted), in OT1 mice compared with P14 or F5 mice (Fig. 2 A). As overrepresentation of memory OT1 cells in the inoculum might favor the proliferation of OT1 cells in adoptive hosts, we also performed cotransfer experiments with purified naive OT1 cells (CD44⁻122⁻). However, this led to similar results; naive OT1 cells did not proliferate in OT1 recipients, but they proliferated in P14 mice (Fig. 2 B). We also determined whether hierarchical competition could be observed between T cell populations that have similar numbers of naive and memory T cells, i.e., from P14 and F5 mice. We found that naive P14 cells (CD44⁻) divide in F5 recipients (Fig. 2 C). However, as for the OT1 and P14 combination before, the reverse experiment, i.e., the transfer of F5 cells into P14 hosts, did not result in proliferation (Fig. 2 C). Thus, we found a second case of hierarchical competitiveness and extended the order to: OT1 > P14 > F5. The observed hierarchy is most likely based on the avidity of each TCR for its restricting MHC molecule loaded with self-/environmental peptides. Unfortunately, comparative affinity measurements that would confirm such a notion are not available. In fact, binding affinities have been determined for a limited number of TCR and MHC/peptide ligands, but the chosen peptides may not reflect the full complement of self-/environmental peptides presented by each MHC molecule as recognized by a given TCR (20, 21).

Because we observed that P14 cells, restricted to D^b, failed to divide in hosts that are prefilled with OT1 cells, nominally restricted to K^b, one might deduce that the OT1-TCR has cross-reactivity to D^b. This has recently been concluded using a different experimental setting (18). We decided to determine whether any presumed cross-reactivity of the OT1-TCR could account for the observed hierarchical competitiveness.

We first determined whether the OT1-TCR could be selected on a class I molecule other than K^b by breeding OT1 Rag-2^−/− mice onto the K^b-deficient allele (6, 22). In the absence of K^b, the OT1-TCR was not positively selected in the thymus and no CD8⁺ cells were present in the periphery (Fig. S1, A and B, available) (19). Thus, in the absence of K^b there is no positive selection of the OT1-TCR by other MHC isotypes present in H-2^b mice. Also, K^b is essential for OT1 proliferation in response to Ova_257-264 (Fig. S1 C) (19).

Second, we determined whether division of OT1 cells in P14 recipients was mediated by K^b. To this end, we bred P14 K^b−/− Rag-2^−/− mice. These were then used as recipients for CFSE-labeled OT1 and P14 donor cells. 2 wk after transfer, OT1 cells showed only limited division in K^b-deficient P14 recipients compared with K^b-sufficient P14 recipients (Fig. 3 A).

Third, we wanted to determine whether the competence of OT1 cells to block division of P14 cells resides in the ability of the OT1-TCR to bind to K^b. P14 cells proliferated in OT1 Rag-2^−/− K^b-deficient recipients when these recipients had previously been transplanted with thymi from K^b-sufficient B6.Rag-2^−/− embryos to allow for the reconstitution of CD8⁺ cells expressing the OT1-TCR (Fig. 3 B, right). We conclude from these experiments that the ability of OT1 cells to proliferate in P14 recipients and the ability of OT1 cells to block homeostatic proliferation of P14 cells in OT1 recipients are both mediated by OT1-TCR binding to K^b molecules. Thus, there is no functional cross-reactivity of the OT1-TCR to other MHC isotypes that could account for its observed hierarchical competitiveness.

Two hypotheses could explain the data so far. T cells could compete for the MHC-bearing cell, either by competition for physical access or by modulation of the ability of this cell to stimulate homeostatic proliferation. Alternatively, T cells could derive a graded ability to compete for some (diffusible) ligands, whereas they transit through the peripheral lymphoid organs (23). These ligands would be available in limited supply and would not be bound to the APCs; the level of competitiveness for these ligands would be determined by the affinity of the TCR toward restricting MHC molecules. It would be possible to differentiate these possibilities by creating a situation where T cells of “higher” hierarchy could not interact with a proportion of the MHC-bearing cells. To generate a balanced situation in which only some APCs express K^b, we performed parabiosis between K^b-mutant or control K^b-sufficient mice and OT1 mice. After several weeks of parabiosis, CFSE-labeled OT1 and P14 cells were transferred into the animals. Interestingly, P14 donor cells did not divide more in K^b-mutant and OT1 parabiosis pairs compared with control K^b-sufficient and OT1 parabiosis pairs (Fig. 4). Thus, P14 cells are not able to obtain sufficient signals to allow for their division, even in the presence of K^b-deficient APCs that OT1 cells cannot interact with (Fig. S2, available). Similar parabiosis experiments with K^b-sufficient, K^bm1-mutant, and OT1 mice were also performed, resulting in identical outcomes (unpublished data).

To determine which factors may rule the level of competitiveness downstream of the TCR, we analyzed cytokine receptor expression before and after transfer. No differences were detected before transfer (Fig. S3, available). Upon transfer into a lymphopenic environment, IL-7Rα and -2Rβ receptor expression was similar for OT1 and P14 cells (Fig. 5, A and B). However, we observed differences between OT1 cells proliferating in T cell–proficient P14 recipients and OT1 or P14 cells undergoing lymphopenia-induced proliferation in immunodeficient recipients in respect to IL-7Rα downmodulation and gradual increases in IL-2Rβ expression (Fig. 5, A and B). Thus, the mechanism of OT1 cell dominance over P14 cells is different to the lymphopenia induced proliferation of both TCR specificities in lymphopenic Rag-2^−/− recipients. Proliferation in the latter may be lymphokine-only driven, whereas in P14 recipients, lymphokines may not be as abundant and TCR signaling is, as shown above, essential for OT1 cells to dominate over P14 cells. Indeed, we also observed reduced CD5 expression on OT1 cells in P14 Rag-2^−/−K^b−/− versus P14 Rag-2^−/− recipients (unpublished data).

To determine if other molecules were regulating this clonal competition beyond IL-7 (24, 25), we treated recipient mice with anti–IL-7Rα antibodies: OT1 cell proliferation in P14 mice was not affected (Fig. 5 C). This does not exclude that OT1 cells act as a “sink” for IL-7, thereby hindering P14 cell proliferation in OT1 recipients. Unobstructed proliferation of OT1 cells despite IL-7R blockage appears to contradict studies involving IL-7–deficient recipient mice. In contrast to these results, however, our experiments lasted longer (day 19 vs. day 5 or 7) and another study, using similar IL-7R blockage, showed limited OT1 cell proliferation on day 6, which was further diminished when recipient mice were IL-15 deficient (26). Thus, other cytokines may have allowed for later homeostatic proliferation in our experiments.

Collectively, our results suggest that competition based on TCR triggering is mediated by ILs, such as IL-7, -2, and/or -15 as their receptor expression is modulated, yet other soluble factors (e.g., trophic factors) probably play a role and remain to be characterized (23).

Several important conclusions can be made on basis of our results. First, hierarchical competitiveness operates beyond competition for MHC molecules. Thus OT1 cells, whose TCR is restricted by K^b, hinder the homeostatic proliferation of P14 cells, restricted to D^b. A similar conclusion had been made based on the finding of hierarchical competitiveness among OT1 and 2C-TCRs (17). However, even though the antigenic peptides recognized by these K^b-restricted TCRs are distinct, one does not know whether they compete for overlapping niches of self-peptide–MHC molecules. Based on our results, we can formally exclude such a mechanism of competition for the OT1 and P14-TCRs because in the absence of K^b expression the hierarchical dominance of OT1 cells is no longer seen.

Second, we demonstrate that hierarchical competitiveness is not caused by the (simple) model of cellular competition for the MHC-bearing cell, or alternatively, by modulation of the stimulating ability of this cell. This conclusion is derived from our experiments using parabiotic pairs of K^b-deficient and OT1 mice as recipients for OT1 and P14 cells. Even though in such parabiotic pairs OT1 cells cannot interact with K^b-deficient APCs, representing half of the APCs in the system, these cells were not able to stimulate homeostatic proliferation of P14 cells. Thus, OT1 cells were still able to dominate P14 cells in trans. Therefore, TCR binding to restricting MHC molecules appears only as a primary requirement for homeostatic competitiveness. Here, the TCRs' affinity toward self-MHC molecules, as well as the kinetics of DC interactions, while T cells migrate through the lymphoid organs, would result in a TCR-characteristic level of signaling within the lymphocyte. The overall signal received in this way, however, would only define the cells' relative ability to secondarily compete for trophic ligands essential for homeostatic proliferation, which are available in limited supply and not bound to the stimulating APC.

B6.Rag2^tm1Alt (Rag-2^−/−) mice were a gift from A. Rolink (Basel Institute for Immunology, Basel, Switzerland). B6.Rag2^tm1Alt-ptprc^a (Rag-2^−/− Ly-5.1) mice were bred at the Basel Institute for Immunology. The line, restriction specificity, and source of the TCR transgenic mice is as follows: OT1 [H-2K^b] (27) provided by S. Degermann and E. Palmer (Basel Institute for Immunology, Basel, Switzerland); P14 [H-2D^b] (28) provided by M. Bachmann (Basel Institute for Immunology, Basel, Switzerland) from Taconic or by CDTA (Orleans, France); F5 [H-2D^b] Rag-1^−/− (29) mice were provided by A. Kruisbeek (The Netherlands Cancer Institute, Amsterdam, Netherlands). TCR transgenic mice were bred with Rag-2^−/− or Rag-2^−/− Ly-5.1 mice to obtain TCR transgenic Rag-2^−/− Ly-5.2 or TCR transgenic Rag-2^−/− Ly-5.1/2 mice. K^bm1 Rag-2^−/− and K^b−/− Rag-2^−/− mice were derived by breeding the MHC donor strain (B6.C-H2K^bm1 [Jax] or B6.129P2-H2K^btm1 [6]) to Rag-2^−/− mice. Likewise, K^b mutant OT1 or P14 Rag-2^−/− mice were obtained.

Parabiosis of 4–10-wk-old sex-matched mice was performed as previously described (30). For thymus transplantation, fetal B6.Rag-2^−/− thymus lobes were isolated at day 14.5–15.5 of gestation, and 5–10 lobes were transplanted under the kidney capsule of the indicated recipients. Studies on animals were approved by the Regierungspräsidium Freiburg (regional council of the federal state of Baden-Württemberg, Germany) and the Comité Local des Animaleries Institut National de la Santé et de la Recherche Médicale/CEA with the approval of the Direction Départementale des Services Vétérinaire, Grenoble (local committee and regional representative of the Ministry of Agriculture, France).

Single-cell suspensions of pooled lymph nodes and spleen (RBCs lysed) were each prepared in PBS with 2% FCS. Antibodies were purchased commercially (BD) or prepared by protein G affinity chromatography of tissue culture supernatants, followed by labeling with fluorochrome or biotin: 104.2.1 (Ly-5.2), A20-1.7 (Ly-5.1), F23.1 (TCR-Vβ₈), MR9-4 (TCR-Vβ₅), RR3-15 (TCR-Vβ₁₁), B20.1 (TCR-Vα₂), Mel-14 (CD62L), 53–6.7 (CD8), RM4-5 (CD4, BD), IM7 (CD44), AF6-88.5.3 (K^b specific, reactive to K^bm1), H57-597 (TCR-β), M1/69 (CD24), 1D3 (CD19), M1/70 (CD11b), A7R34 (CD127), PC61 (CD25), CD122 (BD), CD5 (BD), and CD69 (BD). 5F1 ascites (K^b specific, not-reactive to K^bm1) was a gift of Nathenson (Albert Einstein College of Medicine, Bronx. NY). Biotinylated mAbs were revealed with streptavidin-allophycocyanin (Invitrogen), -PE (SouthernBiotech) or -PerCP (BD); others with sheep F(ab)₂ anti-mouse Ig-FITC (Silenus) or R33-18 (mouse κ; R. Grützmann and K. Rajewsky, 1981, Cologne).

As applicable, cells were incubated with tissue culture supernatant of 2.4G2 (CD16/32) hybridoma cells to block unspecific staining. Cells were stained with mAbs or 2° reagents at predetermined optimal dilution. Flow cytometry was done on FACSCalibur and LSRII (BD) instruments, using CellQuest, Diva (BD), or FlowJo (FlowJo) software. Except where indicated, data shown are from lymph node cells. However, cells from spleen always gave similar results.

Cells were obtained from pooled lymph nodes, washed into PBS, and labeled with CFSE (5 μM CFDA[-SE] C-1157; Invitrogen) for 8 min at room temperature while mixing. FCS was added to 20% incubating for an additional 2 min, and cells were washed into PBS. Except where noted, 1–5 × 10⁶ cells were transferred into various 6–15-wk-old recipient mice i.v.

To block IL-7 cytokine signaling, mice were given multiple injections of blocking antibodies specific for the IL-7Rα chain (A7R34, CD127) or isotype-matched control antibodies i.p. (Fig. S4, available).

Fig. S1 shows lack of positive selection of OT1 cells in the absence of K^b. Fig. S2 shows unobstructed donor T cell exchange in OT1 Rag-2^−/− and K^b−/− Rag-2^−/− parabiotic couples, as well as DC chimaerism for one exemplary couple. Fig. S3 shows CD127, CD25, CD5, and CD69 expression on CD8⁺ lymph node cells from B6, P14 Rag-2^−/−, P14 Rag-2^−/− K^b−/−, and OT1 Rag-2^−/− mice. Fig. S4 shows effective blocking of IL-7 cytokine signaling by injecting antibodies specific for the IL-7Rα chain, resulting in >10-fold decreased thymocyte numbers and >30-fold reduction in CD19⁺ pre–/pro–B cell percentages in the bone marrow.

We thank Y. Joos for expert technical assistance; G. Besin and G. Turchinovich for breeding K^b−/− Rag-2^−/− mice and FACS; Dr. Nathenson for 5F1 ascites; M. Dessing, T. Kline, H. Kohler, S. Meyer, and A. Würch for help with flow cytometry; the animal care takers for animal husbandry; H.-P. Stahlberger, B. Pfeiffer, and H. Spalinger for art work/photography; F. McBlane, R. Ceredig, H. Jacobs, and D. van Essen for critical reading of the manuscript; and all people from BII for the pleasurable time there. F. A. is greatly indebted to Dr. R.-P. Sekaly for hosting him at Institut National de la Santé et de la Recherche Médicale U743. The Basel Institute for Immunology was founded and supported by F. Hoffmann-La Roche Ltd., Basel, Switzerland.

The authors declare that they have no conflicting financial interests.