Introduction
Harald von Boehmer in 1990, with Jean Tinguely's version of the 
-helix that became the icon of the Basel Institute for Immunology.
Some 35 years ago, I finished my PhD thesis at the Walter and Eliza Hall Institute in Melbourne, Australia. Before my return to Europe, my mentor, Ken Shortman, had concluded from DNA-labeling studies that very few, if any, of the constantly renewing cortical thymocytes would ever become mature T cells1. He mocked the "heroic disregard" for his studies by his contemporaries at high-profile institutions and reminded me that whatever I was going to do with the thymus, it was probably of no importance if it did not explain his data. I liked the challenge, which he repeated whenever he visited the Basel Institute for Immunology in Switzerland, my new working place. There I shared an office and laboratory with another 'Aussie', Jonathan Sprent. Initially, we both became interested in studying the acquisition of specific tolerance by thymocytes and T cells developing from T cell–depleted bone marrow transplanted into major histocompatibility complex (MHC)-mismatched recipients. T cell depletion permitted immunological reconstitution in the absence of graft-versus-host disease2, a result that contributed to the cure of immunodeficiency patients by transplantation of T cell–depleted parental bone marrow3.
Selection or modulation of cells?
These bone marrow chimeras next became useful in establishing the 'plasticity' of MHC-restricted antigen recognition, which allowed us to exclude the contribution of the T cells' MHC genotype to the antigen-recognition process and to document the genetic potential of each organism to generate a very diverse repertoire of T cell receptors for antigen (TCRs) capable of recognizing diverse antigens in the context of the various highly polymorphic MHC molecules of the species4, 5. These discoveries raised the issue of whether the TCR repertoire underwent selection in each organism to most efficiently recognize antigens in the context of self MHC molecules at a time when it was not yet known that MHC-restricted antigen recognition involved peptide-MHC complexes.
Niels K. Jerne, then the director of Basel Institute for Immunology, insisted that the bone marrow chimeras be used to test his hypothesis that thymocytes expressing germline-encoded TCRs that bind to self thymic MHC molecules are selected to proliferate and that somatic mutation of TCR genes occurring during this proliferation diversifies the T cell repertoire6. T cell repertoire selection was thus studied in various bone marrow chimeras in Basel and elsewhere. Although some of these initial studies of T cell repertoire selection may seem conclusive in hindsight, they were in fact not so at the time. The crux of the issue was that the bone marrow chimeras had to be immunized to 'read out' the MHC-restricted antigen specificity of effector cells, and it was not clear whether any observed bias in MHC-restricted responsiveness was due to selection of cells rather than to properties of the cells, such as hyper-responsiveness or anergy. Too little was known then about the regulation of immune responses in general and the consequences of presentation and cross-presentation of TCR ligands by hematopoietic versus stromal cells in particular. A thymus transplantation experiment by Rolf Zinkernagel stood out because it seemed to have separated events that occur during thymocyte development from events imposed by immunization7. Enthusiasm for the conclusions of such studies vanished, however, when transplantation of thymi into nude, thymus-deficient mice yielded results that seemed at odds with the idea of thymic selection of the TCR repertoire8.
Cellular selection and lineage commitment
The then-beginning era of molecular immunology raised hope that the controversial issue of thymic TCR repertoire selection could be revisited with better tools once the TCR was actually identified. Eventually we succeeded in transferring MHC-restricted specificity for antigen with Tcra and Tcrb genes from one T cell clone to another9, which provided indisputable evidence of MHC-restricted antigen recognition by a single receptor. This opened the possibility of creating TCR-transgenic mice with the aim of studying TCR repertoire selection conclusively by monitoring the fate of antigen-specific T cells without needing to produce effector cells by immunization. Thus, this approach allowed studies to address cellular selection rather than responsiveness. The generation of TCR-transgenic mice succeeded through a collaborative effort among several investigators within and outside the Basel Institute for Immunology that introduced into the germline of mice rearranged Tcra and Tcrb genes from a T cell clone that recognized a male chromosome–encoded HY peptide presented by H-2Db MHC class I molecules10. This T cell clone was chosen because the response to HY had been extensively exploited before by my lab to study TCR repertoire selection in bone marrow chimeras.
With these TCR-transgenic mice, I planned to address the following questions derived from an independent hypothesis of TCR repertoire selection and T cell lineage commitment that did not involve the mutation of TCR genes11. Did TCR-dependent positive selection of developing T cells by thymic MHC molecules exist in the absence of immunization?12 If so, did it represent an essential step13 in 'rescuing' immature cortical thymocytes from programmed cell death14 rather than the then-favored 'repertoire bending' proposed to result from MHC-induced population expansion of some but not other thymocytes? This 'rescue' hypothesis was developed to combine T cell repertoire selection with Ken Shortman's results indicating intrathymic cell death of most cortical thymocytes1. Did positive selection by thymic MHC class I or class II molecules determine whether the selected cell would become a CD8+ killer cell or a CD4+ helper cell, respectively15? This model of CD4-versus-CD8 lineage commitment was proposed to achieve alignment of effector function (kill or help) and TCR specificity for MHC class I and class II molecules that acquire their peptides from the cytosol (endogenously produced viral proteins) and lysomes (exogenously produced bacterial proteins), respectively. Did the high-affinity binding of the TCR to 'real' antigens (conventional peptide-MHC complexes as opposed to superantigens) in the thymus result in cell death (negative selection)10, which would explain, at least in part, tolerance to self? This last question could be easily studied by comparison of female and male TCR-transgenic mice and needed to be addressed because studies of negative selection in bone marrow chimeras suffered from the same deficit as experiments studying positive selection: it was not clear whether nonresponsivness was due to elimination of responsive cells or to some form of anergy.
Obviously, the TCR-transgenic mice represented a wonderful tool for obtaining affirmative answers to all those questions. However, these mice also had drawbacks because of premature expression of the transgenic TCR
and the fact that the Tcra locus, unlike the Tcrb locus, is not allelically excluded16 and hence the transgenic TCR is not the only receptor expressed in TCR-transgenic mice. Although the last inconvenience could be fixed by crossing TCR-transgenic mice on the rearrangement-deficient severe combined immunodeficiency background13 or the recombination-activating gene–deficient background, the too-early expression of the TCR actually had some rather notable consequences on early lineage commitment beyond the scope of this essay17. There had also been some questions about whether negative selection, demonstrated by the substantial depletion of cortical thymocytes in male TCR-transgenic mice9, was dependent on a developmental block caused by the premature expression of the transgenic TCR rather than on the death of cortical thymocytes. Follow-up experiments, however, established a developmental 'window' for TCR-induced negative selection in the form of apoptotic cell death extending from very immature thymocytes in the thymic cortex to still-immature thymocytes in the medulla18.
Since the time of those experiments, it has become apparent that the efficacy of thymic negative selection in purging the T cell repertoire of potentially self-reactive cells is enhanced by the presentation of 'peripheral' antigens by the immigration of extrathymic antigen-presenting cells or the 'promiscuous' expression of 'peripheral' antigens in the thymus. Apparently, thymic negative selection is impaired in nonobese diabetic mice that spontaneously develop type 1 autoimmune diabetes19. In fact, we found that the incidence of diabetes in nonobese diabetic mice could be diminished considerably by increasing the negative selection of insulin-specific thymocytes through overexpression of insulin in the nonobese diabetic thymus20.
As for positive selection and TCR-dependent lineage commitment, we noted that positive selection determines the lineage fate not only of CD4+ helper and CD8+ killer cells but also of regulatory T cells, whose generation is likewise dependent on encounter with TCR ligands inside as well as outside the thymus21. Thus, tolerance to self is not only due to the elimination of self-reactive thymocytes but also to the lineage diversion of self-reactive lymphocytes into regulatory T cell lineages.
The TCR repertoire that remains after positive and negative selection not only efficiently detects and discriminates between foreign peptides presented by self MHC class I and class II molecules but also shows considerable cross-reactivity to allogeneic peptide-MHC complexes; the latter phenomenon represents a considerable barrier to tissue transplantation between unrelated people. At present there is no broad consensus on whether the apparent 'preoccupation' of the TCR repertoire with MHC molecules is due to the coligation of CD4 or CD8 coreceptors and the TCR by MHC molecules during positive selection and/or to an intrinsic germline-encoded MHC bias in the TCR repertoire because of the coevolution of TCR and MHC molecules6.
Many of the more straightforward conclusions about positive and negative selection have been confirmed and extended in HY and other TCR-transgenic mice as well as in knockout mice, such as those lacking MHC class I or class II. As for TCR-dependent lineage commitment resulting in the alignment of TCR specificity and T cell function15, considerable progress is being made in elucidating the molecular mechanisms involved in TCR-dependent lineage fate determination.
TCR transgenes in T cell biology
At the time when TCR-transgenic mice first became available, it was difficult to justify going into greater cellular and molecular details of thymic selection and lineage fate determination when so many fundamental questions about the adaptive immune system remained unanswered. TCR-transgenic mice aided the identification of the pre-TCR that selects cells with productive TCR rearrangements for further maturation22, provided initial insight into the cellular basis of immunological memory by defining the special properties of antigen-experienced cells that persisted in the absence of antigen23, and established deletion and reversible anergy24 as well as conversion into regulatory T cells25 as mechanisms of peripheral tolerance.
Overall, it was great fun to have such a tool and to identify the basic mechanisms governing the adaptive immune system rather than 'nibbling on' one particular issue until the very last detail was clarified. Admittedly, there was a strong element of 'play' and naive curiosity in my studies described here. I am glad, however, that despite all the playing, I did not entirely disappoint my Australian mentor and could provide a biologically meaningful explanation for his "heroically disregarded" but nevertheless crucial observations on the turnover of thymocytes. The fact that cortical thymocytes with either no TCR or TCRs that cannot bind to anything in the thymus have no future but die from 'neglect' represents the appropriate biological context for Ken Shortman's early conclusions that most cortical thymocytes are doomed to die1.
After spending much time exploiting TCR-transgenic mice to identify the fundamental principles that apply to the adaptive immune system, such as TCR-dependent cellular selection and lineage commitment, as well as tolerance and memory, I find it encouraging to witness how insights gained with these somewhat artificial mice are valid in wild-type mice and even contribute to the understanding of human diseases such as immunodeficiency and autoimmunity.
receptor in the development of 
