The first definitive experiments outlining the phenomenon of MHC restriction were published by Shevach and Rosenthal, 1973 (MHC class II) and then Zinkernagel and D oherty, 1974 (MHC class I) who made the fundamental observation that T cells recognise foreign material only when presented by MHC molecules common to both the accessory and target cells. The former study demonstrated that in order to inhibit the proliferative response of T cells to purified protein derivative (PPD), presented in the context of macrophage MHC class II molecules, blocking antisera had to be directed towards immune response genes common to both the T lymphocyte and the antigen presenting cell. Using monolayers of mouse fibroblasts infected with LCMV assayed with spleen cells from LCMV infected mice Zinkernagel demonstrated that only spleen preparations from mice sharing at least one set of H-2 antigenic specificities with the target monolayer caused high levels of specific lysis. The antigenic structures recognised by T cells were actively studied by many investigators and Chesnut etal., (1980) reported that, unlike ; B cells ^ T cells are stimulated by APC pulsed with native as well as denatured forms of protein. They later demonstrated the necessity for a trimolecular interaction between TCR/MHC/peptide (Shimonkevitz et al., 1984). On the basis of their early findings they proposed that protein antigens are degraded into peptides by APC before recognition by TCR. Subsequently, Grey et al., (1982) and Ziegler and Unanue (1981 and 1982) showed that after antigen binding in the cold, a period of incubation at 37°C for 1 hour was required before antigen presentation could be detected. Protease treatment immediately after binding in the cold abrogated the capacity to present antigen, but if this treatment was preceeded by a 2 hour incubation at 37°C no effect was apparent. They therefore concluded that the lysomotrophic agents chloroquine and ammonium chloride could inhibit APCs in presenting peptides to T cells indicating the need for a pH dependant proteolytic degradation of protein and a lag period before they could be presented on the cell surface and recognised by the TCR - events collectively called antigen processing and presentation.
Antigen processing and presentation
Figure 1.8 summarises MHC class II antigen processing and presentation. (Reviewed by Cresswell, 1994). Briefly, MHC class II molecules are synthesised in the endoplasmic reticulum (ER) and complex with a chaperone molecule named the invariant chain (li). Cross-linking studies suggest that the MHC molecules are released from the ER as complex of three MHC class II dimers and three li chain units. The formation of this nonamer correlates with the silencing of the li chain ER retention signal as a result of which the newly synthesised MHC class Il/Ii complex is translocated to the Golgi complex. MHC class II dimers formed in li chain negative mice are poorly expressed at the cell surface and accumulate in the ER. Unlike most cell surface proteins which are translocated directly to the cell surface the MHC/Ii complex is diverted via the endosomal/lysosomal pathway. It is postulated that a targeting sequence in the cytoplasmic domain of the li chain is responsible for sorting. Once inside the endocytic ('MHC class II containing compartment' - MIIC) compartment, li is proteolytically cleaved. The MHC class II molecules are then free to bind the newly generated peptides. In addition to providing the targeting signals which direct MHC class II from the ER to the endosomal pathway in vitro studies utilising isolated class II /li chain trimers have also demonstrated that the li chain serves to prevent peptides (from the MHC class I pathway) binding to MHC class II molecules. As discussed (section 1.5, ‘Characterisation of naturally bound peptides eluted from human class II molecules’) as many as 50% of purified MHC class II molecules are found to be associated with 'CLIP' peptide (Class II associated invariant chain peptide). CLIP consists of a nested set of peptides derived from amino acids 80-104 of the li chain which bind with high avidity to MHC class II molecules and compete with antigenic peptide for binding (Bangia and Watts, 1995).
Processed peptides bind to newly formed MHC class II molecules which have been sorted to an endosomal/lysosomal system. Immunocytochemical evidence originally suggested that the MHC class II molecules were targeted to the early endosome. More recently.
Harding and Geuze (1993b) identified a late endocytic vesicle containing HLA-DR molecules. This 'MHC class II containing compartment' (MIIC) had all the characteristics of a lysosome. A similar localisation of MHC class II molecules has been identified in mouse macrophages where it has also been shown that upregulation of the rate of transcription of MHC class II leads to an increase in the number of MHC class II molecules in these lysosomes (Harding and Geuze (1993a,b). MllC-like compartments have since been described in human Langerhans cells (Kleijmeer etal,, 1994) and mouse spleen dendritic cells (Kleijmeer et a l, 1995). The findings of Rudensky et a l, (1994) would suggest a specific role for MHC in the processing of peptide as well as the assembly of peptide-MHC class II complexes.
The analysis of a mutant HLA-DR3+ B cell line ('16.23') which expresses MHC class II molecules but does not present exogenous antigen led to the search for a gene encoded in the MHC class II region essential for endososmal peptide loading of MHC class II molecules. Transfected HLA-DR undergoes a conformational change in these cells: two determinants are lost. Furthermore, HLA-DR dissociates into monomers in SDS-PAGE. T2 cells have a large homozygous deletion in the MHC class II region similarly affecting antigen presentation. The mutant phenotype can be rescued by somatic cell fusion implying a recessive mutation. Cell fusion and complementation analysis using the above cell lines in conjunction with another selected on the basis of loss of HLA-DR cell surface expression suggest a homozygous deletion within the class II region of the MHC. Transfection of the wild type HLA-DM genes is sufficient to restore the correct phenotype. This data confirms a role for the HLA-DM gene products in antigen presentation. Although the precise mechanisms remain to be determined it has been proposed that HLA- DM has a catalytic role in the removal of CLIP from newly synthesised MHC class II molecules (Mellins et a l, 1990, Morris et a l, 1994; Reviewed in Sanderson and Trowsdale, 1995).
Binding of peptides to MHC class II molecules
In the past 10 years considerable progress has been made in understanding the characteristics of naturally associated self-peptides, the requirements for presentation of foreign peptides and the molecular features of MHC class II molecules which govern determinant selection.
Initially the binding affinities and specificities of MHC class II molecules for peptides were extensively studied using synthetic peptides and either purified or cell surface expressed MHC class II molecules. In 1985 Babbitt et al., and, subsequently Buus et at., (1986a) conducted experiments which demonstrated that class II MHC molecules can specifically bind peptides. For example. Babbitt demonstrated that a fluorescently labelled peptide (46- 61 of HEL) bound H-2A^ but not H-2A^. In these early experiments binding was evaluated by quantitating labelled peptide associated with MHC class II preparations after dialysis. Soon afterwards, Buus et al., (1986b) formally showed that complexes between MHC class II and synthetic peptides were the ligands recognised by mouse T cell hybridomas using complexes of synthetic labelled ovalbumin 323-339 and mouse H- 2A^ inserted into planar membranes to stimulate IL-2 release by ovalbumin peptide specific H-2A^ restricted T cell hybridomas. He showed that ovalbumin peptide plus MHC class II was 20,000 times more potent for T cells than the same amount of soluble protein presented to H-2A^, -E^ or -E^ membranes. Moreover, in this study Buus utilised a convenient method to measure binding of peptides to class II MHC molecules based on the separation of free and class II MHC bound radio-labelled peptides by gel filtration. This assay became an important tool in defining the different features of synthetic peptides that bind to various mouse and human class II alleles.
Peptide elution experiments/techniques
In 1988 Buus etal., were the first to isolate and characterise self-peptides directly bound to MHC class II molecules. Circumstantial evidence suggested that class II MHC molecules might bind peptides derived from endogenous as well as exogenous proteins. Firstly, Babbitt et al., (1986) had demonstrated that H-2A^^ was capable of binding the autologous mouse lysozyme peptide 46-61 and the homologous xenogeneic HEL peptide equally well. Brown et ah, (1993) later found evidence of electron dense material on the putative peptide binding groove of their crystallised HLA class II molecule. Buus and co workers had found that only 5-10% of affinity purified HLA class II molecules were capable of binding peptide and they speculated that this was caused by the presence of self peptides constitutively presented in the peptide binding groove. By using acid elution they demonstrated that mouse H-2A and -2E molecules constitutively bind peptides of molecular weights ranging from 2-12 Kd (mostly 2-4Kd).
Based on the early crystallographic analysis it was expected that the great majority of peptides associated with a particular MHC class II isoform would bind with a common configuration and share amino acid binding motifs. Based on these assumptions Falk et al., (1991) began to define such motifs for class I MHC molecules by Edman degradation of unfractionated mixtures of peptides extracted from MHC molecules, rather than of individual peptides. The key assumption is that the binding motif is fixed in relation to the NH2 terminus for all peptides binding to a given MHC molecule. Since the N-terminus of natural MHC class II ligands is 'ragged' ie., the distance from the N terminus to the peptide core can vary, it was assumed that pool sequencing of class II ligands would not be useful in defining motifs. However, in 1994 Verreck et al., used pool sequencing to produce the first sequence information available for naturally occurring HLA-DQ2 and HLA- DRw3/w4 associated peptides. The resulting HLA-DQ eluted peptides predominantly contained lysine, isoleucine and phenylalanine at the positions representing putative anchor residues for HLA-DQ. Two peptides were homologous to the muscarinic acetylcholine receptor 407-421 and the measles virus matrix protein 270-284 and bound to an HLA-
DQ2+ B lymphoblastoid cell line (BLCL). Analysis of the HLA-DRw3/w4 eluted peptide pool yielded sequences matching an epitope from the endogenous enzyme gluteraldehyde- 3-phosphate dehydrogenase. In addition Falk et aL, (1994) recently used pool sequencing in order to obtain information about the peptide motifs of HLA-DQ, -DR and -DP molecules (see section 1.5, 'MHC allele/isotype-specific binding).
Binding of peptides is degenerate in that one peptide may be capable of binding to several MHC class II alleles. In some cases peptides can also be demonstrated to be promiscuous in that a single T cell clone can recognise the same peptide bound to different MHC class II specificities. It could be predicted that degeneracy and promiscuity would be more common for H-2E and HLA-DR molecules due to the non-polymorphic nature of their a
chains. In support of this hypothesis no peptides have so far been found in common between the H-2A^, -A^ and -A^* alleles. In the study by Chicz degenerate invariant chain peptides were found associated with numerous alleles (HLA-DR 1, -DR2, -DR3 and -DRV). With the wisdom of hindsight it is now known that these li chain fragments represent CLIP peptides (amino acids 83-107 of the li chain). Papers by Malcherek et aL,
(1995) and Sette et aL, (1995) indicate that CLIP possesses the characteristics of a degenerate peptide. It uses the same peptide frame to bind the grooves of different class II molecules ie. HLA-DRB 1*0101 and -DRB 1*0301, methionine as a position 1 anchor and aliphatic rather than aromatic residues to bind to all -DR alleles independently of the polymorphism existing at p86 of the -DR|3 chain, which excludes large aromatic residues (Tyr and Trp) in some -DR alleles. CLIP also lacks particular allele-specific contact sites at the major anchor and inhibitory positions 4 and 6.
MHC class II binding motifs
A variety of approaches have been used to define the structural requirements for the interaction between antigenic peptides and MHC class II molecules. One approach consists of synthesising analogues of MHC class II binding peptides containing single amino acid
substitutions and testing them for their binding capacity to MHC class II molecules. Following this strategy, several groups have analysed peptide determinants and defined amino-acid residues crucial for the interaction with MHC class II molecules in both mouse and man. For example, 80-90% of the residues in HEL 107-116 could be replaced without appreciable effects on their ability to activate H-2B^ restricted T cell hybridomas, whereas three substitutions led to a marked decrease in T cell activation. The effect of the three residues at positions 112, p i 14 and pi 16 suggested that positively charged residues determined the capacity to interact with the MHC class II molecule. Hydrophobic amino acids had previously been shown to be important in the binding of ovalbumin 323-336 to H-2A^, but were relatively unimportant for the interaction of HEL with H-2E^. (Buus et aL, 1986a). O'Sullivan et al,, (1991) conducted similar experiments using HLA-DR molecules. Several singly substituted/truncated analogues of haemaglutinnin 307-319 and tetanus toxoid 830-843 were tested for their capacity to bind various HLA-DR alleles. The results suggested that efficient binders would contain an aromatic/hydrophobic residue such as Y, W, F, I, L or V at position 1, followed by S, T , V, I, L, P or C at position 6 and a hydrophobic residue at position 9. Based on this prediction they found that 65% of such peptides were effective at binding MHC class II molecules.
Using the peptide library approach Hammer et aL, (1992) have shown that when oligonucleotides encoding peptides known to bind to HLA-DR 1 are inserted into a coat protein of phage M l3 they are expressed on the surface of the phage and can be bound by HLA-DR 1 molecules. A library of 20 million random nonamer peptides displayed by M13 was screened for HLA-DR 1, -DR4 or -D R ll binding and up to 60 of the peptide inserts were sequenced. Analysis of the binding region of the peptide led to the identification of two potential anchor positions. The first an aromatic residue (Tyr, Phe or Trp) at the NH2 terminus of the peptide and the second three residues downstream (Met/Leu). In addition they found that the negatively charged amino acids Asp and Glu were largely excluded from binders and that small residues such as Gly and Ala were enriched at position 6.
Tissue specific peptides
Different tissues might be predicted to present various self peptides either because of differences in antigen processing and/or because of tissue specific variation in transcribed genes. Demotz et al., (1989) demonstrated that peptides of HEL containing the immunodominant epitope (107-116) eluted with different profiles from H-2E^ transfected L cells and from A20 BLCL, indicating that the different APCs produced different peptides spanning the HEL determinant 107-116.
Bonomo and Matzinger (1993) proposed that because peptides complexed with MHC are derived from normal cellular proteins and that different tissues, because of their different functions, might express different arrays of proteins, each tissue should display it's own unique pattern of MHC/peptide complexes ('MAP'- MHC antigen profile). Their study revealed that thymic epithelium expressed a surface MAP that overlapped with, but was not identical to, the MAPs of other tissues.
However,when Marrack et ah, (1993) examined the representation of peptides associated with MHC class II molecules expressed on cells from spleen, thymus and thymic cortical epithelium of normal and transgenic animals they found extensive sharing of peptides from these various sources. Many, but not all, of the peptides bound by H-2A^ in the spleen were also bound by H-2A^ in the thymus. Peptides derived from the B-cell specific proteins were absent, reflecting the near absence of B cells in the thymus. Thus tissue- specific differences in peptide expression seem to account for a small percentage of the total peptide pool.
MHC allele/isotype-speciflc binding
As described previously the initial binding experiments analysing the ability of H-2E or H- 2A epitopes to bind to the respective molecules showed a degree of specificity since
binding only occurred to the presenting MHC class II molecule. The extensive study Buus
et al., (1987) provided evidence of isotype-specific binding. Purified H-2A<^ and H-2E^ peptides were acid eluted and separated by gel filtration; each fraction was then tested for it's ability to inhibit the binding of radiolabelled peptides to bind to H-2A^ or H-2E^. Material isolated from H-2A<^ was able to inhibit binding to H-2A^ but not H-2E^. However, material isolated from H-2E^ could inhibit binding to both isotypes, implying that the H-2E molecule could bind an overlapping set of peptides with H-2A, but not vice versa. R udensky et al., (1991) found that the sequence motifs identified for H-2E^ derived peptides are similar to those determined by synthetic peptide binding to HLA-DR, yet different from the binding motifs associated with the H-2A^ and H-2At>. Hence different MHC isotypes may allow the presentation of distinct peptides.
The two most recent and comprehensive studies comparing peptides bound to various alleles/isotypes of MHC class II molecules have been undertaken by Marrack et at.,
(1993) and Falk et at., (1994). Marrack and co-workers isolated and characterised, by sequence analysis, self-peptides from the spleen of normal and MHC class II transgenic mice. Initial observations on H-2A^ and H-2E^ molecules of C3H/HeJ mice produced peptide profiles for H-2A which were very different from those eluted from H-2E, see table 1.2. In a separate series of runs they compared peptides eluted from H-2E^ and H- 2A^ isolated from the spleens of AY transgenic mice (see section 1.4 ‘Cell types that mediate positive and negative selection’). Once again the peptide profiles were different. With the exception of a major peptide derived from P2M (p42-58) that is found in the H- 2E^ and -E^ eluted profiles, there were no sequenced peptides in common between H-2E^ and H-2A^^ or H-2E^. The situation with the peptides bound by H-2E molecules is different. Here a moderately strong motif was evident. All the self peptides bound by the H-2E molecules have isoleucine or valine at position 5, and most have a hydrogen bond at position 6. All but two have a hydrophobic residue (or alanine) at position 6 and a basic amino acid, usually lysine, at position 13 or 14. The self peptides bound by the H-2A^^ molecule did not show any obvious sequence motif. This is similar to peptides bound by