MHC class II molecules are critical in the immune response for the presentation of peptide to mature T cells in the periphery, as well as centrally determining the positive and negative selection of the developing T cell repertoire in the thymus, see figure 1.7.
Cell types that mediate positive and negative selection
Early experiments with bone marrow and thymic chimeras (Bevan 1977; Zinkernagel et al., 1978a, 1978b; reviewed in Zinkernagel 1978) have shown MHC restriction to be dependant on MHC molecules expressed in the thymus. Several experiments were undertaken to determine whether the acquisition of MHC restriction was dependant on expression by radio resistant thymic stromal cells, radio sensitive bone marrow derived cells or both. For example, Bevan (1977) demonstrated that H-2^/H-2t> cytotoxic cells which mature in an irradiated H-2^ host respond preferentially to antigens plus H-2^, whereas H-2<^/H-2t> cells which mature in an irradiated K-2^ host respond to the same antigens in conjunction with Yi-2lP gene products. This was interpreted as evidence for the involvement of the radio-resistant host thymic cells in positively selecting the Fi T cell precursors for parental restriction. However, controversy exists over the effects of differing degrees of irradiation used in these early experiments (Elliott 1993) and the observation made by Longo and Schwartz (1980) that if chimeras were left for 8 months and the mature T cells removed, 3 months later they were shown to be able to mount responses restricted by host and donor MHC. Since this work the original findings of Bevan et al
have been supported by the findings of experiments undertaken by Lo and Sprent (1986) in which T cells from bone marrow chimeras were found to be restricted to the parental/host MHC type.
Is the cell specific distribution of MHC class II in the thymus an important factor in determining the fate of developing thymocytes? In order to answer this question it is
important to know which MHC class II positive cell types are capable of driving positive and negative selection.
An attempt to gain a more precise insight into the issue of the influence of MHC distribution on positive selection involved the use of transgenic mice that selectively express MHC antigens in defined thymic micro-environments. Using site-directed mutagenesis of the X and Y elements in the H-2E promoter van Ewijk et al., (1988) generated lines where expression of the H-2-E transgene was restricted to distinct anatomical sites in the thymus. Those mice carrying the AY mutation only expressed H-2E on cortical epithelium; whilst the recipients of the AX transgene were H-2E positive on medullary epithelium. On examining the various transgenic lines it was noted that unlike wild type H-2E+ controls that delete V pl7a+ cells, mice in which H-2E expression was limited to thymic epithelial cells failed to completely delete V pl7a+ cells (Marrack et al
1988). The authors concluded that H-2E on thymic epithelium was responsible for positively selecting the Vpi7a"^ cells. Flow cytometric analysis of the Vp6+ T cells in the peripheral lymphoid organs of Fi (AX or AY x SJL) (Benoist and Mathis 1989) mice provided further evidence that for positive selection of peripheral CD4+ Vp6+ T cells to occur H-2E has to be expressed intrathymically on cortical epithelial cells. (These studies were carried out before the effects of Mtvs on the T cell repertoire were understood, which would complicate interpretation of these results).
One of the problems in the study of the mechanisms involved in T cell positive selection is that the interaction of individual T cells with their selecting molecules is extremely difficult to follow due to low precursor frequency. This problem has been addressed by the generation of TCR transgenic mice. TCR genes from a T cell clone of known MHC and antigen specificity can be introduced into the germline, and by allelic exclusion inhibit the recombination of endogenous TCRs so that the large percentage of thymocytes express the transgenic TCR. As monoclonal antibodies against the TCR chains are often available cells expressing the transgenes can be identified and followed. In these experiments positive
selection could only be observed when the appropriate selecting MHC was expressed on thymic epithelium. For example, Berg et al., (1989) confirmed the data of van Ewijk by crossing the AX and AY animals onto mice transgenic for a TCR specific for a fragment of pigeon cytochrome C (PCC) in the context of H-2E^. Using the Vp3 clonotypic TCR to PCC Berg was able to demonstrate that, in correspondence with the data from van Ewijk, the (TCR X AY) mice were capable of positively selecting the PCC specific T cells,
whereas the AX mice were not. Hence the differential distribution of MHC in the thymus would seem to have major influence on the process of positive selection. In an attempt to generalise these observations to a broad repertoire of CD4+ T cells Cosgrove etal., (1992) crossed the previously established lines described above and MHC class II" mice and demonstrated that those mice which displayed MHC class II molecules in the cortex had normal numbers of CD4+ T cells in the thymus and periphery, while mice with an MHC class II" cortex were almost devoid of mature CD4+ cells.
Which thymic antigen presenting cells are responsible for clonal deletion (negative selection)? Early experiments demonstrated that thymic epithelial cells could not induce tolerance, von Boehmer and Schubiger (1984) showed that thymus grafts required cells sensitive to deoxyguanosine (dGuo) to generate tolerance to self MHC in athymic nude mice. Additionally, experiments conducted by Jenkinson et al., (1985) using foetal thymic organ culture (FTOC), also demonstrated that the dGuo resistant epithelium did not induce MHC tolerance to developing thymocytes. An early study by Marrack et ah, (1988) also suggested that thymic epithelial cells were inefficient at causing clonal deletion. Bone marrow from V pl7a+ H-2E^+ animals was grafted into H-2Q/H-2E", or H-2^/ H-2Ek+ irradiated recipients. This procedure generated mice able to produce V pi7a+ T cells and expressing H-2E only on bone marrow derived cells, or on all appropriate cells in H-2E" mice with an H-2E+ thymic graft or an H-2Ea transgene that was expressed exclusively on thymic epithelium. Clonal deletion was found to be inefficient in these animals, since the level of V pl7a+ cells was only slightly lower than in H-2E" animals. Subsequently Mls^" induced Vp6+ T cell deletion was studied in bone marrow chimeras (Speiser et ah, 1989a
and 1989b) in which the donor and host differed in the expression of Mtv and H-2E. Since the deletion of V(36+ T cells is dependant upon expression of both Mtv and MHC, it was possible to determine which cell type, bone marrow-derived or stromal, was HLA class II positive. Deletion of the Vpb"*" TCR occurred only when H-2E was present on the bone marrow derived cells. Similar findings have been reported by others.
However, evidence now exists that epithelial cells can also induce negative selection. In mice expressing a transgenic T cell receptor specific for lymphocytic choriomeningitis virus (LCMV) plus H-2D^> T cell development was analysed in irradiation bone marrow chimeras expressing the restriction element (H-2D^) on host tissue. The radioresistant (epithelial cells) in the host thymus were shown to be capable of deleting developing Va2+ and V|38+ T cells (Speiser et al., 1992). In addition, recent data indicates that positive selection in vivo may not be totally dependant on the epithelium. Using chimeras made with p2m" mice (Bix and Raulet, 1992) which are deficient for MHC class I, some positive selection of CD8+ T cells (20% of normal) was observed in irradiated mice reconstituted with bone marrow. The lack of requirement for specialised presenting cells in thymic selection was reinforced by the demonstration that even fibroblasts are able to promote positive selection. By inoculating p2m“ mice (deficient in MHC class I expression) intra thymically with thymic epithelial cells or cultured fibroblasts Pawlowski et al., (1993) were able to demonstrate that fibroblasts induced the maturation of CD8+ thymocytes in numbers comparable to heterozygous (-/wild type) litter mates.
In summary, it is apparent that for certain antigens and in certain conditions both thymic epithelial cells and bone marrow derived cells are capable of mediating both positive and negative selection.
Altered ligand model of selection
The notion that thymic epithelium is exclusively responsible for positive selection whereas the bone marrow derived cells mediate negative selection would suggest a resolution to the paradox that both positive and negative selection involve recognition of the same MHC molecule. If thymic epithelial cells express a distinct set of peptide/MHC complexes, then positive selection could be mediated by qualitatively different class E/antigen combinations. According to this 'altered ligand' model, the MHC of thymic epithelium presents a unique range of self-peptides allowing for positive selection (Marrack and Kappler, 1988). Negative selection of these thymocytes would not occur because these peptides are not found on cells of haemopoeitic origin. However, the lack of requirement for specific cells in positive and negative selection, as outlined above, would make the altered ligand hypothesis very unlikely. Furthermore, direct analysis of peptides eluted from thymic MHC molecules has not supported the idea that the cortical epithelium presents a unique set of peptides (Marrack et al.y 1993).
What is the evidence for the involvement of MHC/peptide complexes in selection?
Experiments with TCR transgenic mice have demonstrated that negative selection depends on the recognition of peptide/MHC complexes. Typically the addition of the antigen in vitro
or in vivo TCR-transgenic systems results in the clonal deletion of the antigen-specific clone. Using in vitro organ culture of thymi from mice carrying a transgenic TCR specific for cytochrome c peptide bound to H-2B^ Spain and Berg (1992) found that addition of cytochrome c peptide led to deletion at the double positive stage of T cell development. Similarly, Pircher et al., (1993) using thymocytes from mice expressing a transgenic TCR specific for LCMV and H-2D^ co-cultured with various H-2^ cell types as ARC in the presence of varying concentrations of LCMV peptide found antigen specific deletion of double positive T cells. In this study deletion was also demonstrated to be peptide (LCMV) dose-dependant. Murphy and co-workers (1990) generated transgenic mice expressing a
TCR against ovalbumin 323-339 in the context of H-2A^. Injection of mice with the OVA peptide lead to a rapid deletion of double positive thymocytes. More recently, Mamalaki et al., (1992) developed a transgenic mouse in which the TCR was specific for the influenza virus nucleoprotein (FLU) 366-379 in association with H-2Dt>. Again injection of the mice with a synthetic peptide corresponding to amino acids 366-374 of the FLU protein led to a peptide specific and dose-dependant depletion of the transgenic TCR. Using a transgenic mouse carrying a TCR from 2B4 T helper hybridoma specific for the C-terminal fragment of cytochrome c in association with H - 2 E a ^ / H - 2 E p t > Berg et at., 1990 also determined
that the efficiency of selection of the transgenic TCR was increased in animals with the correct MHC class II heterodimer, allelic variants being less efficient, or incapable, of mediating selection. The variation between the alleles of MHC class II was determined to reside in the peptide binding groove of the molecule and not to affect the interaction with the TCR, suggesting that MHC/peptide complexes are involved in positive selection.
Various groups have recently investigated the peptide/MHC/TCR interaction leading to positive or negative selection using models in which selection is examined in FTOC from TCR transgenic mice, exploiting P2H1 or TAP knock-outs in which MHC class I expression is rescued by exogenous cognate peptide. In the TAPI" mice addition of exogenous MHC class I binding peptide restores the stable expression of the peptide/MHC complex at the cell surface of thymic stromal cells, although only some of these peptides promoted positive selection of CDS"*" T cells indicating that only a distinct population of peptides were capable of selecting the CD8+ cells in this system (Ashton-Rickardt et at,,
1993).
Furthermore complex mixtures of peptides would appear to be more efficient than single peptides in promoting the positive selection of CDS'*" T cells. (Hogquist et aL, 1993). To determine the specificity of positive selection of a single TCR p2m knock-out mice were crossed with mice carrying a TCR to ovalbumin recognised in the context of H-2Kt>. Several single amino acid variants of ovalbumin could mediate positive selectiori, however.
they were ineffective at stim ulating a response in mature T cells. Furtherm ore, a peptide was
identified which could lead to positive selection on p2m'/" phenotype (low peptide density)
and to clonal deletion on cells. These results demonstrate that not only is selection
influenced by peptide density but also that events can be peptide specific.
One of the first theories put forward to account for the processes of positive and negative selection taking place from the interaction of a TCR with an MHC molecule was based on
receptor affinity (Reviewed in Sprent et al, 1988). In this model cells recognising self-
MHC with a low affinity would be positively selected, whilst those interacting at a higher affinity would be deleted. For example, using mice expressing a transgenic TCR specific for LCMV it was shown that deletion of TCR+ thymocytes occurred following neonatal infection with wild type virus. In addition, tolerance could also be induced by neonatal infection with m utant viral strains which bore point mutations in the T cell epitope recognised by the transgenic TCR. In com parison when adult transgenic mice were challenged with the mutant viruses the authors found either a poor or complete lack of an activation response, suggesting the deletion effects observed following neonatal infection had occurred at a level that was too low for peripheral activation. This lead the authors to speculate that differences in TCR affinities during thymic education may account for the
subsequent positive or negative selection of thymocytes (Pircher et al, 1991).
However, it has since been demonstrated that a high affinity peptide can both negatively and positively select depending on the density of peptide/M HC com plexes on thymic stromal cells. This has lead to the recent refinem ent of the affinity model, making
allowances for the influence of the avidity of TCR/MHC complexes (Ashton-Rickardt et
a l , 1994). An LCM V specific H -ZD ^-restricted TCR transgene (called p i 4) was introduced into mice which were either TAP1+ or TA PI". Positive selection of the transgenic CDS""" TCR was impaired in TAP" mice. Addition of LCMV peptide to TAP" FTOC at low and high concentration induced positive and negative selection respectively. Addition of the same peptide to TAP"*" FTOC induced negative selection even at high
concentration. The authors argue that in TAP1+ mice the basal level of avidity necessary for positive selection has already been reached through the interaction of MHC/TCR/peptide, addition of LCMV leading to high avidity and negative selection. In contrast, in the TAP" mice the decreased MHC class I expression results in a very low basal avidity initially such that infection with LCMV only leads to a relatively small increase in the overall avidity and thus results in positive selection. Similar findings have been reported by Sebzda et al., (1994).
Peripheral tolerance
It is clear that both thymic positive and negative selection events shape the T cell repertoire. However, if the generation of tolerance relied solely on thymic negative selection events all self epitopes would have to be expressed on thymic tissue, on cells capable of entering the thymus or as soluble proteins. It has been proposed that thymic cortical epithelial cells are capable of presenting an endogenous haemoglobin peptide suggesting that some antigens could recirculate to the thymus (Matzinger and Guerder 1989) A more recent report also implied that during the neonatal period a limited number of CD4+ and CD8+ T cells recirculate to the thymus, allowing tolerance against mature T cell specific antigens.
However, other studies have shown that extrathymic mechanisms for the generation of tolerance can act upon mature T cells, thus removing the need for compulsory thymic expression. Mechanisms include peripheral deletion, anergy, the down regulation of TCRs and accessory molecules, T cell ignorance and/or the induction of specific suppressive T cell subsets. For example, there is mounting evidence that clonal deletion can occur in the peripheral immune system. Injection of Mlsa+ lymphocytes into Misa" mice resulted in the expansion of Vp6+ T cells within 4 days, followed by a specific depletion over the period of 2 weeks (Webb et a l , 1990). Similarly, within 4 days of injection of Staphylococcus enterotoxin B (SEE) into mice there was an increase in the proportion of VpS"*" T cells, but
by day 7 they had largely disappeared (Kawabe and Ochi, 1991). In this case apoptosis was detected demonstrating that exposure to SEE lead to T cell death.
Tolerance to extra-thymic antigens has been extensively studied using transgenic mice that express an immunogen or MHC molecule at an extrathymic site (Lo et al., 1988,1989). For example, expression of H-2E on pancreatic B cells in an H-2E" strain resulted in tolerance to H-2E as judged by a lack of response to H-2E+ splenocytes in vitro (Lo et at.,
1988). The same group constructed transgenic mice (on an H-2E“ background) in which the elastase promoter was used to drive H-2E expression on pancreatic acinar cells (Lo et at,, 1989). Again both peripheral and thymic T cells were shown to be unresponsive to H- 2E in vitro. Since both Vp5+ and Vpl7a+ cells were not deleted but could not be activated even by anti-Vp specific antibodies, the T cells were determined to be anergic. In most cases the anergised state of the T cell can be overcome by the addition of exogenous IL-2 (Essery et at., 1988). It is therefore proposed that anergy occurs when the T cell is stimulated in the absence of IL-2 ('second-signal hypothesis'). However, the use of MHC molecules as model antigens for extra-thymic expression makes interpretation difficult, as the presentation of tissue specific peptides may influence the response of T cells. Thus tolerance may be the result of the MHC molecule failing to present the correct peptide rather than the induction of tolerance by the MHC molecule itself. Furthermore, experiments on transgene expression driven by tissue specific extra-thymic promoters are subject to the caveat that a small amount of aberrant thymic transcription is often detectable (Jolicoeur et a l, 1994).
Thymie precursor