Peptide Presentation 9 i991 S. Karger AG, Basel 0257-277X/91/0101-005452.75/0
Immunol Res 1991;10:54-65
Peptide Presentation by Class-I Major Histocompatibility Complex Molecules Janko Nikoli6-Zugi~, Francis R. Carbone
Department of Immunology,Research Institute of Scripps Clinic, La Jolla, Calif., USA
Introduction Class-I-restricted T-cell recognition involves the presentation of fragmented endogenous antigens [1-3]. Peptides derived from cytosolic processing make their way into the endoplasmic reticulum where they bind to newly synthesized class I [4-8]. The probable configuration of the class-I-binding site for peptide has recently been described with the resolution of the three-dimensional structure of the human class-I proteins, HLA-A2 and HLA-Aw68 [9-t 1]. This crystallographic analysis defines the existence of a long groove flanked by two a-helical walls and supported on a 'floor' of [3 strands (fig. 1). The binding cleft is formed by the folding of the highly polymorphic cq and a2 domains from the class-I heavy chain. The description of the crystal structure provides an essential framework for designing the experiments to address how this polymorphism controls the T-cell response. In this article we review some recent experiments that shed light on how polymorphic changes in the binding cleft affect T-cell recognition and T-cell development.
Peptide: Major Histocompatibility Complex Contact and Its Influence on the Conformation of Antigenic Determinants One of the major puzzles in the field of major histocompatibility complex (MHC) class-I presentation concerns the contact of antigenic petide and MHC. How many residues does it involve from both molecules? How flexible is it? What is the position and conformation of the bound peptide? Although definitive answers to the above questions are still not available, results from several laboratories have yielded important information on these issues. Shortly after the initial demonstration that class-I restricted cytotoxic T lymphocyte (CTL) recognition of foreign antigens could be reproduced using short synthetic peptides, Maryanski et al. [3] established an experimental sytem to study this presentation in detail. They used an H-2 d mastocytoma line, P815 transfected with either HLA-A24 or HLA-CW3, to generate CTLs that recognized synthetic peptides corresponding to residues 170-182 from the respective HLA proteins. CTL recognition was
MHC Class-I Presentation of Self and Foreign Peptides
55
Fig. 1. The location of amino acid residue changes occurring in Kbrn molecules superimposed on a three-dimensional structure of class-I molecule.
Kd-restricted and the peptide presentation could be inhibited by various Kd-binding fragments [t2, 13]. For example, CW3-specific CTL recognition of CW3170_182 could be efficiently inhibited using the non-crossreactive A24170_~82 or an unrelated Kd-restricted peptide from the region 147-158 of the influenza nucleoprotein. This experimental system, using functional competition, was subsequently exploited by Maryanski et al. [14] to define the peptide residues important for binding to K d. By using a series of truncated and substituted A24170_182 peptide analogs, it was shown that three residues played a decisive role in Kd-binding. These were the Tyr at position 171 and the Thr-Leu at positions 178 and 179. The three residues could promote K d association regardless of the remainder of the peptide sequence. Namely, it was shown that a synthetic analogue containing only the Tyr and Thr-Leu linked by an appropriate polyproline spacer, was better at
binding to K d than the parental A24170-182. The constraints imposed by this particular spacer strongly suggested that the only requirement for class-I binding in addition to the presence of the critical amino acids was likely to be the relative orientation of the individual residue side chains. These results and the observation that the Tyr was likely to be a c o m m o n feature of the Kd-binding peptide prompted Maryanski et al. [14] to propose that class-I-binding peptide may share c o m m o n structural motifs and indicated that a single antigenic peptide binds to M H C in one predominant conformation. Similar motif-like sequences have been suggested to mediate the contact o f antigenic fragments with M H C class-II molecules [15, 16]. Different motifs are likely to reflect allelic class-I preferences for certain amino acid arrangements. This idea is consistant with the studies of Garrett et al. [ 1 1] describing the three-dimensional structure of a second human class-I protein, HLA-Aw68.
56
Comparison of the HLA-A2 and HLA-Aw68 structures revealed that MHC polymorphism apparently alters subsites or pockets within the binding cleft while leaving the overall architecture of the molecule unchanged. Such changes clearly alter peptide binding. A unique set of peptides was bound to the HLA-Aw68 molecule since the uninterpreted electron density in the cleft differed from that found in HLA-A2 [I t]. It is the specific requirements of the individual subsites within the cleft that give rise to the observed differences in peptide binding and thus form the basis for allelic antigen specificity. Moreover, since particular amino acid side chains are also likely to be required for binding to these pockets, peptides specific for one class-I product or pocket should exhibit a unique motif arrangement such as that described by Maryanski et al. [14] for K d binding. Changes within the peptide-binding site that dramatically affect the structure of a given subsite can be expected to completely abolish the binding of certain peptides. It is also evident that other more subtle changes within the peptide-binding groove can dramatically influence T-cell recognition without eliminating the peptide-MHC association. Hogan et al. [ 17] investigated the effect of substitutions at positions 152 and 156 of the HLA-A2 class-I molecule on the presentation of the influenza-A matrix protein fragment M57-68 to CTL lines. These changes located on the a2 helix of class I abrogated peptide recognition of most but not all CTLs tested. Given that CTLs from at least one individual could recognize the M57-68 peptide presented by the HLA-A2 variant, it was clear that alterations at residues 152 and 156 did not prevent peptide association. Consequently, Hogan et al. [17] proposed that
Nikolid-Zugid/Carbone
changes at these two residues were affecting T-cell recognition by altering the conformation of the bound peptide. Unfortunately it is impossible to rule out that such changes in the a2 helix do not alter the local regions of the class-I molecule that interface directly with the T-cell receptor (TCR). However, by turning their attention to changes localized to the floor of the classI-binding site, McMichael et al. [ 18] and Shimojo et al. [19] showed that certain mutation in this region, which could not directly interact with the TCR, had the same effect. For example, it was shown that changes at residues 9 and 99 on separate 13 strands of HLA-A2, abolished peptide presentation to some, but not all, human CTLs specific for M57-68. In this case it was absolutely clear that the changes were affecting peptide presentation by modifying the conformation of the bound antigenic fragment. Similar results have been found for the murine Kb-restricted recognition of ovalbumin (OVA) [20]. We have previously shown that in C57BL/6 mice the class-I-restricted CTL response to cytosolic forms of the protein OVA is specific for the Kb-binding fragment OVA253-276. The availability of the large panel of in vivo derived mutant K b molecules (Kbin) allowed us to assess how changes in the binding cleft could affect this peptide's presentation. The naturally occurring K bm molecules differ from the parental molecule by one to five amino acids in the a1 and ct2 domains [21 ] (table 1; fig. 1). By using K bm expressing target cells, we examined the fine specificity of eight OVA + Kb-specific CTL clones [20]. All K bm molecules, except K bml and K binS, could present OVA peptide to at least some of the clones, implying that, apart from K bm~ and K brnS, all other substitutions did not pre-
M H C Class-I P r e s e n t a t i o n o f Self a n d Foreign Peptides
-'ent antigen binding. Individual amino acid changes at positions 77. 80 and 116 were found to selectively affect recognition by certain clones. All three residues are predicted to contact the bound peptide. The residues 77 and 80 of K b molecule could also potentially interact indirectly (at residue 77) or directly (at residue 80) with TCR as well as with the peptide, yet their effect on OVA presentation is very selective (individual clones, rather than lines, are affected) and the changes are not detectable on the surface of the helices by serologic analysis [20]. Thus we feel the effect o f these two substitutions is probably not due to their direct influence on TCR binding. In contrast to the changes at positions on the ct helices, the K bin5 substitution at residue 116, on the bottom o f the peptide-binding site, is incapable o f direct interaction with TCR, and therefore must exert its influence on the T-cell recognition via the bound peptide. We conclude that K bmS, and most likely K bm3 and K bm11, induce a conformational change in the OVA peptide, making the complex no longer recognizable to certain TCR. Therefore, the changes in the structure of the peptide-binding site of classI MHC molecules can shape the immune response at the level of antigen presentation in two ways. The mechanism of action of these changes could be due to the abrogation of peptide binding (for example, in the case of OVA peptide and K bmS, see below) or alternatively, due to the change in peptide conformation (for example, K bm3,5, 11). Collectively, these results underline the functional importance of M H C polymorphism at the peptide contact residues. Knowing this, it is of particular interest to turn our attention to the role of M H C polymorphism during T-cell development.
57
Table I. Characteristics o f K b m u t a t i o n s Mutant
Position o f altered a m i n o acid
Classification ~ of residue
K bml
152 Glu --+ Ala 155 Arg --+ T y r 156 Leu --+ T y r
ligand/TCR TCR/ligand ligand
K bm3
77 Asp --+ Ser 89 Lys --+ Ala
ligand silent
K bin5
I 16 T y r ~ P h e
ligand
K bins
22 T y r ~ P h e 23 M e t --+ Ile 24 Glu --~ Ser
ligand silent ligand
K brat0
163 165 167 173 174
T h r - - ~ Ala Val --~ M e t T r p --~ Ser Lys ~ G l u Asn --~ Leu
TCR/ligand TCR/ligand TCR/ligand silent TCR
K brnJs
77 Asp --+ Ser 80 T h r ~ A s n
ligand ligand
K bin23
75 Arg--+ His 77 Asp ~ Ser
TCR ligand
a
F r o m B j o r k m a n et al. [10] a n d N a t h e n s o n et al.
[211.
Peptides and the Positive Selection of T-Cell Repertoire in the Thymus As discussed earlier, class-I MHC-restricted T cells recognize antigenic peptides bound to the peptide-binding groove of class-I molecules. The same class-I MHC molecules are used by thymic cortical epithelial cells to positively select the T-cell repertoire during T-cell ontogeny [22-25]. Are the thymic M H C molecules alone mediating the positive selection (fig. 2a), or do they bind self peptides, and then imprint this self recognition on the developing T cells (fig. 2b)?
58
Nikolid-7.ugid/Carbone
a
MHC
--1
[--
I
I
Fig. 2. Two models of intrathymic positive selection. a The first model postulates the exclusive importance of the a-helical portions of MHC molecules in the selection of TCR repertoire, b According to the second model, the helical parts play less importance and the bound self peptides are the major driving force in the repertoire selection. This allows the peptide-binding groove to shape the T-cell repertoire.
Recently several theoretical reports entertained the idea that thymus-specific self peptides [26] or erroneous peptides [27] might participate in positive selection. Experimentally, there has been some evidence that the background genes can influence positive selection. Using hapten-modified H-2 k cells as antigen, Ogata et al. [28] were the first to describe such an influence. Thereafter, Singer et al. [29] published a preliminary report in which they concluded that self class-I K bderived peptide is necessary for the positive
selection of alloantigen-specific C D 4 + T cells by a class-II I-A b molecule. Most recently, Fry et al. [30] described the effect of background genes on the positive selection of V~tl 1 TCR-bearing T cells. In an attempt to determine whether the self peptides are directly involved in positive selection, we examined the OVA-specific repertoire of C T L generated from ( H - 2 K b • H-2Kbm)Fl bone m a r r o w T-cell precursors which matured in H - 2 K b or H - 2 K bm thymus [31]. As described above, in H-2 b animals the OVA-specific CTLs are restricted solely by the K b class-I molecule and are specific for a determinant contained within the 0VA253-276 fragment [5]. In H-2 b thymus, the K b molecule positively selects this C T L repertoire. Importantly, s o m e K bm molecules (e.g. K bm5 and K binS) differ from K b at residues located deep on the floor of the peptide-binding groove (fig. i; table 1), and these residues are not accessible for contact with T C R [10]. Helical regions of both molecules are identical to the helices of K b by primary structure and serologic analysis [20, 21]. In addition, K bin8 and K bm5 substitutions are unlikely to affect folding of other parts of the molecule, since even extensive class-I p o l y m o r p h i s m has little effect on its overall three-dimensional structure [11]. Therefore, if helical regions alone can mediate positive selection, then all three molecules, K b, K brn5 and K bin8, should be the same for the selection of the OVA-specific C T L repertoire. Alternatively, if the floor of the peptide-binding groove can exert any influence on OVA-specific C T L repertoire, then it has to do so via a bound peptide, and we should see differences between repertoires selected on K b versus K bin5 and/or K broS. As can be seen from table 2, K bin8 could not select any OVA-specific T cells.
MHC Class-I Presentation of Self and Foreign Peptides
59
Table 2. The ability of Kb and Kbm molecules to This p r o v i d e s strong, albeit indirect, evimediate the positive selection and antigen presentadence that self peptides are critically intion to OVA and VSV-specific CTL volved in positive selection. F u r t h e r experiments shed light on the MHC OVA VSV Nc structural aspects o f positive selection. We class-I positive presenpositive presenf o u n d a tight correlation between the ability molecule selection tation selection tation of four K bm molecules to positively select O V A + Kb-specific T cells and their capacity Kb + + + + tO present 0VA253-276 to OVA-specific CTLs q_l _1 [31 ] (table 2). M o r e o v e r , experiments using a Kbml + + peptide c o m p e t i t i o n assay d e m o n s t r a t e d Kbm3 + + that K bm8 c a n n o t physically interact with the Kbrn5 + + + + O V A peptide (table 3). T h u s it appears that Kbm8 _ _ + + positively selecting self peptides and antigenic O V A peptide interact with various The positive selection was measured by using F~ --. P bone marrow irradiation chimeras, as described K bm molecules in the same way demonstrat[31]. The presentation was quantified by pulsing the ing that antigenic peptides are closely m i m Kb or K ~m expression targets with relevant synthetic icked by self peptides during intrathymic peptides of OVA and VSV NcP, and determining the positive selection. specific lysis of these target cells by appropriate O n careful consideration this should be CTL. Preliminary results. expected. T h e results o f Maryanski et al. [ 14] and the observations o f Garrett et al. [11] established the concept that class-I presentation involves the binding o f peptides via a c o m m o n m o t i f o f a m i n o acids to compleTable 3. The competition of OVA and VSV pepm e n t a r y pockets within the class-I-binding tides for presentation by Kb and K bins molecules cleft. F o r a selfpeptide to successfully m i m i c a foreign antigen, the two m u s t expose simi- Presenting Antigen Competitor Recognition by VSV Nclar c o n f o r m a t i o n a l determinants to the T C R molecule specific CTL when b o u n d to M H C molecules. In the case o f OVA, this is likely to be achieved by bind- Kb VSV Nc ing to the same class-I pockets. P o l y m o r v s v Nc OVA phisms that significantly change these pockvSV Nc Flu NP ets should u n i f o r m l y affect the binding o f both foreign peptide and the self analog. In Kbm8 VSV Nc VSV Nc OVA contrast to OVA, vesicular stomatitis virus VSV Nc Flu NP (VSV) nucleocapsid protein (NcP) can bind and be presented to T cells by K bmS. FurtherThe assay was performed as described by Mamore, K bin8 expressing t h y m u s can effi- ryanski et al. [12], using a high molar excess (about ciently select for this C T L repertoire (ta- 300 • of competing peptide. The NP-influenza virus nucleoprotein peptide, which binds to D b, but not K b ble 2). Consequently, a class-! molecule that [2], was used as a control. does not present a given antigenic fragment
60
Nikoli6-Zugi6/Carbone
Kb
sr
~
r
I
f--
OVA like self peptides. the OVA-specific TCR
VSV Nc like self pepEs selecting the VSV No.specific TCR
Fig. 3. The mechanism of repertoire selection by self peptides, a The schematic structure of the antigen binding pockets of class I MHC molecules K b and K broS. b Self peptides mimick the foreign antigens during the intrathymic positive selection. This mimicking involves binding to the same pocket of the MHC molecule.
is less likely to select for T cells specific for that antigenic peptide. A model of peptide interaction with M H C pockets during the thymic selection and peripheral antigen presentation is depicted in figure 3. For the sake of simplicity, we will assume that the K b molecule expresses two types of antigen-binding pockets, A and B. We will further consider that all peptides with the ability to bind K b must contact either pocket A or pocket B. One set of antigenic peptides, including OVA, binds to the unique pocket A, whereas other peptides, such as VSV NcP, bind to pocket B (fig. 3a). Among self peptides which bind to pocket A, a small fraction bears resemblance
to OVA when bound in the groove. These peptides mediate positive selection of the OVA-specific CTL repertoire. Certain mutations (for example, in K binS) may alter pocket A such that the class-I molecule is no longer permissive for binding of 0VA253-276 o r any of the OVA-selecting self peptides. It is likely that a different set of self-peptides can now bind to this new subsite (pocket A1) mediating the selection of a unique TCR repertoire in bm8 animals. However, since none of these peptides bears significant structural resemblance to OVA, there is no selection for anti-OVA CTLs. In contrast, VSV NcP and its selecting self peptides maintain their overall configuration within pocket B, en-
MHC Class-I Presentation of Self and Foreign Peptides
abling both K brn8 and K b to mediate intrathymic positive selection and presentation of this particular antigen. Our explanation of the above results would be invalid if the K bm8 mutation acts by altering the TCR contact area; for example, by allosterically changing the a~ (or ct2) helix. Although the only technique capable of unequivocally determining this is X-ray crystallography, at least three lines of evidence argue against it: (i) K bm8 and K b share the same serological profile, and the monoclonal antibodies used for this analysis recognize the helical parts of the molecule [20]; (ii) VSV NcPs can be efficiently presented to Kb-restricted CTLs by K bin8 (and vice versa) which should not occur if the TCR contact residues on the helices of K bin8 are perturbed (table 3) [J. Nikolid-Zugid, unpublished data]; and (iii) crystallographic data on HLA class-I molecules indicate that despite 13 amino acid differences between HLA-A2 and Aw68, the two molecules appeared remarkably identical, indicating that individual substitutions produce only local effects Jill. How general is the involvement of self peptides in positive selection? We do know of two cases in which class-I MHC K bml molecule could positively select the appropriate TCR repertoire specific for Sendai virus [32] and VSV NcP [J. Nikolid-Zugid, preliminary results], but was not able to present the same antigen in the periphery. Some of these results have been explained by extrathymic rather than intrathymic positive selection [32]. The K bml molecule, however, contains a very drastic structural mutation, and behaves as a maverick among K bm molecules, since it is incapable of presenting any antigen to Kb-restricted T cells [21]. It is therefore quite possible that the observed effect of
61
reflects an exceptional cross-reaction rather than a model for positive selection. The suggestion that self peptide-mediated position selection may also offer the best explanation for the results obtained by Shimojo et al. [33]. These authors have demonstrated that a change at position 156 of HLAA2 from a Leu to Trp may be responsible for the positive selection of a unique CTL repertoire specific for the influenza virus matrix peptide 55-73. The change at this residue has been shown to take part in the formation of a ridge on the floor of the HLA-Aw68binding cleft and is most likely inaccessible to direct contact with TCR [11]. While this human experimental model is not able to provide direct answers to questions regarding positive selection, the results are best explained by postulating the involvement of self peptides in the selection process. As described earlier, this change at position 156 in HLA-A2 is also thought to affect T-cell recognition by altering the conformation of the bound matrix peptide [17]. Consequently, changes in the groove that result in new conformations of presented antigenic peptide may be able to select for a unique T-cell repertoire as well. Finally, D.Y. Lob and his collegues (personal commun.) have obtained evidence to suggest a critical role for self peptides in the positive selection of the Tcell population in transgenic mice expressing the receptor from an alloreactive T-cell clone. In total, all these data indicate that self peptides universally participate in positive selection. This proposal immediately raises the issue as to whether class I is being recognized at all during the positive selection (and antigen presentation), or whether it merely serves to put peptides into adequate conformation for TCR recognition. A detailed disK bml
62
cussion of this issue has recently been published by Germain [34]. While it is likely that TCR does indeed contact class I directly, we favor the view that MHC behaves largely as a frame which shapes and is being shaped in part by bound peptide. That TCR recognition of bound peptide probably dominates over MHC recognition is stressed by the fact that polymorphism in class-I residues contacting peptide is significantly greater than in residues contacting TCR directly [10]. It is currently uncertain to what extent polymorphism at TCR-binding residues affects T-cell specificity during the process of positive selection. Given that self peptides are involved in the thymic selection of the class-I-restricted T-cell repertoire, it remains to be shown what the origin of these fragments is. This explanation must of course reconcile the fact that individuals use the same MHC molecules and probably the same self antigens for both positive and negative selection to generate a broad and non-autoaggressive T-cell repertoire. One possibility would have the affinity threshold for positive selection to be considerably lower than that required for negative selection or peripheral T-cell activation. Alternatively, it has been suggested that positive selection involves a unique set of self peptides antigens within the thymic cortex [26]. To accommodate the need for as broad a repertoire as possible, Kourilsky and Claverie [27] have further suggested that these thymic-specific antigens are actually short mutated peptides derived by some aberrant transcriptional and translational mechanism. All these preceding explanations rely largely on the premise that peptide presentation by cortical epithelium in the thymus involves the same processing mechanisms
Nikolid-Zugid/Carbone
which have been defined for peripheral class-I presentation, namely, the endogenous pathway of class-I-restricted antigen processing. The diversity of endogenously derived class-I-associated self peptide determinants would, by necessity, be quite limited, and may not be sufficient for the generation of a respectable TCR repertoire. Furthermore, the same determinants would be readily available to clonally delete, or functionally paralyze, the TCR for they select. In contrast, it may be possible that class-I-mediated positive selection might actually involve the presentation of exogenous self peptides derived by some extracellular, rather than intracellular, proteolytic event. Exogenous peptide presentation is capable of expanding a specific T-cell population both in vivo [35] and in vitro [36-38]. These T cells are considered nonphysiological since they fail to recognize class-I-associated peptides formed as a result of endogenous gene expression [36-38]. Such T cells are also not deleted intrathymically by the same self peptides derived from endogenous antigen expression. For example, Schild et al. [39] showed that they could generate in vitro derived CTLs to a number of self antigens, including ~2-microglobulin, by exogenous peptide presentation. The lack of clonal deletion of such CTLs is easily explained by insufficient density of self peptide determinants presented by hematopoietic thymic elements. This theory has to invoke a different determinant density to explain the differences between positive and negative selection. We favor this explanation because it elegantly accommodates the self peptide model of positive selection without going out of the way to explain differences between the peptides involved in positive selection, negative selection and peripheral activation. The
MHC Class-I Presentation of Self and Foreign Peptides
g e n e r a t i o n o f d i v e r s i t y is a c h i e v e d by the fact t h a t e x t r a c e l l u l a r proteolysis can p o t e n tially g e n e r a t e far m o r e class-I-binding frag-
63
repertoire diversity, avoiding both the massive clonal deletion of the selected repertoire and the autoreactivity of its T cells.
m e n t s t h a n the usual cytosolic p r o c e s s i n g m e c h a n i s m [5, 36]. T h e T cells selected on these p e p t i d e s w o u l d h a v e only l i m i t e d reactivity to class I / p e p t i d e c o m b i n a t i o n resulting f r o m e n d o g e n o u s processing (and a part
Acknowledgement The authors wish to thank Dr. M.J. Bevan for his support.
o f these c o u l d be u n i q u e to the cortical epit h e l i u m ) , l e a v i n g the bulk o f the T-cell repertoire intact. T h e s a m e principles can easily
References
be a p p l i e d to c l a s s - I I - m e d i a t e d selection.
Summary MHC class-I molecules express distinct peptidebinding pockets within their antigen-binding groove. These are critically involved in the binding of antigenic peptides. The amino acid composition of a pocket dictates the structure of a peptide which can be bound in it. This is evident as a consensus amino acid motif which has to exist within a peptide in order for it to bind to a particular MHC allele. Perturbation ofa MHC pocket by amino acid substitution can result in the abolition of peptide binding. Less drastic mutations of the peptide-binding groove, particularly the ones away from the critical pocket, can subtly alter the conformation of bound peptide. Both types of substitution exert an influence on the TCR recognition of antigenic peptide. Peptides are also critically involved in the positive selection of the class-l-restricted TCR repertoire in the thymus. These self peptides act by mimicking their foreign antigens. This mimicking involves the binding of selfpeptides and foreign antigenic peptides to the same pockets of the MHC class-I-antigen binding groove. Consequently, MHC class-I polymorphism in the antigen binding groove controls the intrathymic positive selection and peripheral antigen presentation by the same mechanisms. The majority of positively selecting self peptides could well originate from the extracellular processing of circulating self proteins. Using the diverse, extracellularly generated self peptides and the different determinant density requirements for positive versus negative selection, the immune system can ensure the
1 Townsend ARM, Gotch FM, Davey J: Cytotoxic T cells recognize fragments of the influenza nucleoprotein. Cell 1985;42:457-467. 2 Townsend ARM, Rothbard J, Gotch FM, Bahadur G, Wraith D, McMichael AJ: The epitopes of influenza nucleoprotein recognized by CTL can be defined with short synthetic peptides. Cell 1986;44:959-968. 3 Maryanski JL, Pala P, Corradin G, Jordan BR, Cerottini JC: H-2-restricted cytolytic T cells specific for HLA can recognize a synthetic HLA peptide. Nature t986;324:578-579. 4 Yewdell JW, Bennink JR, Hosaka Y: Cells process exogenous proteins for recognition by CTL. Science 1988;239:637-640. 5 Moore MW, Carbone FR, Bevan MJ: Introduction of soluble protein into the class I pathway of antigen processing and presentation. Ceil 1988: 54:777-785. 6 Yewdell JW, Bennink JR: Brefeldin A specifically inhibits presentation of proteins for recognition by CTL Science 1989;244:1072-1075. 7 Nuchtern JG, Bonifacino JS, Biddison WE, Klausner D: Brefeldin A implicates egress from endoplasmic reticulum in class I restricted antigen presentation. Nature 1989;339:223-226. 8 Cox JH, Ye~vdcll JW, Eisenlohr LC, Johnson PR, Bennink JR: Antigen presentation requires transport of MHC class I molecules from the endoplasmic reticulum. Science 1990;247:715-718. Bjorkman P J, Saper MA, Samraoui B, Bennett WS, Strominger JL, Wiley D: Structure of the human class I histocompatibility antigen, HLAA2. Nature 1987;329:506-512. 10 Bjorkman P J, Saper MA, Samraoui B. Bennett WS, Strominger JL, Wiley DC: The foreign anti-
64
11
12
13
14
15
16
17
18
19
20
21
Nikolid-Zugid/Carbone
gen binding site and T cell recognition regions of class I histocompatibility antigens. Nature 1987; 329:512-518. Garrett TPJ, Saper MA, Bjorkman P J, Strominger JL, Wiley DC: Specificity pockets for the side chains of peptide antigens in HLA-Aw68. Nature 1989;342:692-696. Maryanski JL, Pala P, Cerottini J-C, Corradin G: Synthetic peptides as antigens and competitors in recognition by H-2-restricted cytolytic T cells specific for HLA. J Exp Med 1988;167:1391-1405. Pala P, Bodmer HC, Pemberton RM, Cerottini J-C, Maryanski JL, Askonas BA: Competition between unrelated peptides recognized by H-2K d restricted T cells. J Immunol 1988;141:22892294. Maryanski JL, Verdini AS, Weber PC, Salemme FR, Corradin G: Competition analogs for defined T cell antigens: Peptides incorporating a putative binding motif and polyproline on polyglycine spaces. Cell 1990;60:63-72. Guillet JG, I_ai M-Z, Brinner T J, Buus S, Sette A, Grey HM, Smith JA, Gefter ML: Immunological sell nonself discrimination. Science 1987;235: 865-870. Sette A, Buus S, Colon S, Miles C, Grey HM: IAd-binding peptides derived from unrelated protein antigens share a common structural motif. J Immunol 1988; 141:45-81. Hogan KR, Shimojo H, Walk SF, Engelhard VH, Maloy WL, Coligan JE, Biddison WE: Mutations in the alpha two helix of HLA-A2 affect presentation but do not inhibit binding of influenza virus matrix peptide. J Exp Med 1988;168:725-736. McMichael A J, Gotch FM, Santos-Aguado J, Strominger JL: Effect of mutations and variations of HLA-A2 on recognition of a virus peptide epitope by cytotoxic T lymphocytes. Proc Natl Acad Sci USA 1988;85:9194-9t98. Shimojo N, Anderson RW, Mattson DH, Turner RV, Coligan JE, Biddison WE: The kinetics of peptide binding to HLA-A2 and the conformation of the peptide-A2 complex can be determined by amino acid side chains on the floor of the peptide binding groove. Int Immunol 1990;2:193-200. Nikolid-Zugid J, Carbone FR: CTL recognition of foreign peptide is modified by amino acid changes in the MHC class I peptide binding groove. Eur J Immunol 1990;20:2431-2437. Nathenson SG, Geliebter J, Pfaffenbach GM, Zeff
RA: Murine major histocompatibility complex class I mutants. Annu Rev Immunol 1986;4:471502. 22 Bevan M J: In a radiation chimera, host H-2 antigens determine immune response of donor cytotoxic cells. Nature 1977;269;417-418. 23 Zinkernagel RM, Callahan GN, Althage A, Cooper S, Klein PA, Klein J: On the thymus in the differentiation of H-2 'self-recognition' by T cells: Evidence for dual recognition. J Exp Med 1978; 147:882-897. 24 Kisielow P, Teh HS, Bluthmann H, von Boehmer H: Positive selection of antigen-specific T cells in thymus by restricting MHC molecules. Nature 1988;335:730-733. 25 Lo D, Sprent J: Identity of cells that imprint H-2 restricted T-cell specificity in the thymus. Nature 1986;319:672-678. 26 Marrack P, Kappler J: The T cell repertoire for antigen and MHC. Immunol Today 1988;9:308315. 27 Kourilsky P, Claverie JM: MHC restriction, alloreactivity and thymic education: A common link? Cell 1989;56:327-329. 28 Ogata M, Shimizu J, Tsuchida T, Takai Y, Fujiwara H, Hamaoka T: Non-H-2-1inked genetic regulation of cytotoxic responses to hapten-modified syngeneic cells. J Immunol 1986;136:1178-1185. 29 Singer A, Mizuochi T, Munitz TI, Gress RE: The role of self antigens in the selection of the developing T cell repertoire. Prog Immunol 1986:6:6066. 30 Fry AM, Cotterman MM, Matis EA: The influence of self-MHC and non-MHC antigens on the selection of an antigen-specific T cell receptor repertoire. J Immunol 1989; 143:2723-2729. 31 Nikoiid-Zugid J, Bevan M J: Role of setf-peptides in positively selecting the T-cell repertoire. Nature 1990;344:65-67. 32 Kast WM, deWaal LP, Meleif CJM: Thymus dictates major histocompatibility complex (MHC) specificity and immune response gene phenotype of class II MHC-restricted T ceils but not of class I MHC-restricted T cells. J Exp Med 1984;160: 1752-1766. 33 Shimojo N, Cowan EP, Engelhard VH, Maloy WL, Coligan JE, Biddison WE: A single amino acid substitution in HLA-A2 can alter the selection of the CTL repertoire that responds to influenza'virus matrix peptide 55-73. J Immunol 1989;143:558-564.
MHC Class-I Presentation of Self and Foreign Peptides
34 Germain RN: Making a molecular match. Nature 1990;344:19-21. 35 Ishioka GY, Colon S, Miles C, Grey HM, Chesnut RW: Induction of class I MHC-restricted, peptidespecific cytolytic T lymphocytes by peptide priming in vivo. J Immunol 1989;143:1094-1100. 36 Carbone FR, Moore MW, Sheil JM, Bevan MJ: Induction of cytotoxic T lymphocytes by primary in vitro stimulation with peptides. J Exp Med 1988; t 67:1767-1779. 37 Carbone FR, Hosken NA, Moore NW, Bevan MJ: Class I MHC-restricted cytotoxic responses to soluble protein antigen. Cold Spring Harbor Symp Quant Biol 1989;54:551-555.
65
38 Staerz UD, Zepp F, Schmid R, Hill M, Rothbard J: Recruitment of alloreactive cytotoxic T lymphocytes by an antigenic peptide. Eur J Immunol 1989;19:2191-2196. 39 Schild H, Rotzchke O, Kalbacher H, Rammensee H-G: Limit of T cell tolerance to self-proteins by peptide presentation. Science 1990;247:15871589. Dr. Janko Nikotid-7.ugid Immunology Program Sloan-Kettering Institute 1275 York Avenue New York, NY 10021 (USA)
Immunol Res 1991;10:54-65
Peptide Presentation by Class-I Major Histocompatibility Complex Molecules Janko Nikoli6-Zugi~, Francis R. Carbone
Department of Immunology,Research Institute of Scripps Clinic, La Jolla, Calif., USA
Introduction Class-I-restricted T-cell recognition involves the presentation of fragmented endogenous antigens [1-3]. Peptides derived from cytosolic processing make their way into the endoplasmic reticulum where they bind to newly synthesized class I [4-8]. The probable configuration of the class-I-binding site for peptide has recently been described with the resolution of the three-dimensional structure of the human class-I proteins, HLA-A2 and HLA-Aw68 [9-t 1]. This crystallographic analysis defines the existence of a long groove flanked by two a-helical walls and supported on a 'floor' of [3 strands (fig. 1). The binding cleft is formed by the folding of the highly polymorphic cq and a2 domains from the class-I heavy chain. The description of the crystal structure provides an essential framework for designing the experiments to address how this polymorphism controls the T-cell response. In this article we review some recent experiments that shed light on how polymorphic changes in the binding cleft affect T-cell recognition and T-cell development.
Peptide: Major Histocompatibility Complex Contact and Its Influence on the Conformation of Antigenic Determinants One of the major puzzles in the field of major histocompatibility complex (MHC) class-I presentation concerns the contact of antigenic petide and MHC. How many residues does it involve from both molecules? How flexible is it? What is the position and conformation of the bound peptide? Although definitive answers to the above questions are still not available, results from several laboratories have yielded important information on these issues. Shortly after the initial demonstration that class-I restricted cytotoxic T lymphocyte (CTL) recognition of foreign antigens could be reproduced using short synthetic peptides, Maryanski et al. [3] established an experimental sytem to study this presentation in detail. They used an H-2 d mastocytoma line, P815 transfected with either HLA-A24 or HLA-CW3, to generate CTLs that recognized synthetic peptides corresponding to residues 170-182 from the respective HLA proteins. CTL recognition was
MHC Class-I Presentation of Self and Foreign Peptides
55
Fig. 1. The location of amino acid residue changes occurring in Kbrn molecules superimposed on a three-dimensional structure of class-I molecule.
Kd-restricted and the peptide presentation could be inhibited by various Kd-binding fragments [t2, 13]. For example, CW3-specific CTL recognition of CW3170_182 could be efficiently inhibited using the non-crossreactive A24170_~82 or an unrelated Kd-restricted peptide from the region 147-158 of the influenza nucleoprotein. This experimental system, using functional competition, was subsequently exploited by Maryanski et al. [14] to define the peptide residues important for binding to K d. By using a series of truncated and substituted A24170_182 peptide analogs, it was shown that three residues played a decisive role in Kd-binding. These were the Tyr at position 171 and the Thr-Leu at positions 178 and 179. The three residues could promote K d association regardless of the remainder of the peptide sequence. Namely, it was shown that a synthetic analogue containing only the Tyr and Thr-Leu linked by an appropriate polyproline spacer, was better at
binding to K d than the parental A24170-182. The constraints imposed by this particular spacer strongly suggested that the only requirement for class-I binding in addition to the presence of the critical amino acids was likely to be the relative orientation of the individual residue side chains. These results and the observation that the Tyr was likely to be a c o m m o n feature of the Kd-binding peptide prompted Maryanski et al. [14] to propose that class-I-binding peptide may share c o m m o n structural motifs and indicated that a single antigenic peptide binds to M H C in one predominant conformation. Similar motif-like sequences have been suggested to mediate the contact o f antigenic fragments with M H C class-II molecules [15, 16]. Different motifs are likely to reflect allelic class-I preferences for certain amino acid arrangements. This idea is consistant with the studies of Garrett et al. [ 1 1] describing the three-dimensional structure of a second human class-I protein, HLA-Aw68.
56
Comparison of the HLA-A2 and HLA-Aw68 structures revealed that MHC polymorphism apparently alters subsites or pockets within the binding cleft while leaving the overall architecture of the molecule unchanged. Such changes clearly alter peptide binding. A unique set of peptides was bound to the HLA-Aw68 molecule since the uninterpreted electron density in the cleft differed from that found in HLA-A2 [I t]. It is the specific requirements of the individual subsites within the cleft that give rise to the observed differences in peptide binding and thus form the basis for allelic antigen specificity. Moreover, since particular amino acid side chains are also likely to be required for binding to these pockets, peptides specific for one class-I product or pocket should exhibit a unique motif arrangement such as that described by Maryanski et al. [14] for K d binding. Changes within the peptide-binding site that dramatically affect the structure of a given subsite can be expected to completely abolish the binding of certain peptides. It is also evident that other more subtle changes within the peptide-binding groove can dramatically influence T-cell recognition without eliminating the peptide-MHC association. Hogan et al. [ 17] investigated the effect of substitutions at positions 152 and 156 of the HLA-A2 class-I molecule on the presentation of the influenza-A matrix protein fragment M57-68 to CTL lines. These changes located on the a2 helix of class I abrogated peptide recognition of most but not all CTLs tested. Given that CTLs from at least one individual could recognize the M57-68 peptide presented by the HLA-A2 variant, it was clear that alterations at residues 152 and 156 did not prevent peptide association. Consequently, Hogan et al. [17] proposed that
Nikolid-Zugid/Carbone
changes at these two residues were affecting T-cell recognition by altering the conformation of the bound peptide. Unfortunately it is impossible to rule out that such changes in the a2 helix do not alter the local regions of the class-I molecule that interface directly with the T-cell receptor (TCR). However, by turning their attention to changes localized to the floor of the classI-binding site, McMichael et al. [ 18] and Shimojo et al. [19] showed that certain mutation in this region, which could not directly interact with the TCR, had the same effect. For example, it was shown that changes at residues 9 and 99 on separate 13 strands of HLA-A2, abolished peptide presentation to some, but not all, human CTLs specific for M57-68. In this case it was absolutely clear that the changes were affecting peptide presentation by modifying the conformation of the bound antigenic fragment. Similar results have been found for the murine Kb-restricted recognition of ovalbumin (OVA) [20]. We have previously shown that in C57BL/6 mice the class-I-restricted CTL response to cytosolic forms of the protein OVA is specific for the Kb-binding fragment OVA253-276. The availability of the large panel of in vivo derived mutant K b molecules (Kbin) allowed us to assess how changes in the binding cleft could affect this peptide's presentation. The naturally occurring K bm molecules differ from the parental molecule by one to five amino acids in the a1 and ct2 domains [21 ] (table 1; fig. 1). By using K bm expressing target cells, we examined the fine specificity of eight OVA + Kb-specific CTL clones [20]. All K bm molecules, except K bml and K binS, could present OVA peptide to at least some of the clones, implying that, apart from K bm~ and K brnS, all other substitutions did not pre-
M H C Class-I P r e s e n t a t i o n o f Self a n d Foreign Peptides
-'ent antigen binding. Individual amino acid changes at positions 77. 80 and 116 were found to selectively affect recognition by certain clones. All three residues are predicted to contact the bound peptide. The residues 77 and 80 of K b molecule could also potentially interact indirectly (at residue 77) or directly (at residue 80) with TCR as well as with the peptide, yet their effect on OVA presentation is very selective (individual clones, rather than lines, are affected) and the changes are not detectable on the surface of the helices by serologic analysis [20]. Thus we feel the effect o f these two substitutions is probably not due to their direct influence on TCR binding. In contrast to the changes at positions on the ct helices, the K bin5 substitution at residue 116, on the bottom o f the peptide-binding site, is incapable o f direct interaction with TCR, and therefore must exert its influence on the T-cell recognition via the bound peptide. We conclude that K bmS, and most likely K bm3 and K bm11, induce a conformational change in the OVA peptide, making the complex no longer recognizable to certain TCR. Therefore, the changes in the structure of the peptide-binding site of classI MHC molecules can shape the immune response at the level of antigen presentation in two ways. The mechanism of action of these changes could be due to the abrogation of peptide binding (for example, in the case of OVA peptide and K bmS, see below) or alternatively, due to the change in peptide conformation (for example, K bm3,5, 11). Collectively, these results underline the functional importance of M H C polymorphism at the peptide contact residues. Knowing this, it is of particular interest to turn our attention to the role of M H C polymorphism during T-cell development.
57
Table I. Characteristics o f K b m u t a t i o n s Mutant
Position o f altered a m i n o acid
Classification ~ of residue
K bml
152 Glu --+ Ala 155 Arg --+ T y r 156 Leu --+ T y r
ligand/TCR TCR/ligand ligand
K bm3
77 Asp --+ Ser 89 Lys --+ Ala
ligand silent
K bin5
I 16 T y r ~ P h e
ligand
K bins
22 T y r ~ P h e 23 M e t --+ Ile 24 Glu --~ Ser
ligand silent ligand
K brat0
163 165 167 173 174
T h r - - ~ Ala Val --~ M e t T r p --~ Ser Lys ~ G l u Asn --~ Leu
TCR/ligand TCR/ligand TCR/ligand silent TCR
K brnJs
77 Asp --+ Ser 80 T h r ~ A s n
ligand ligand
K bin23
75 Arg--+ His 77 Asp ~ Ser
TCR ligand
a
F r o m B j o r k m a n et al. [10] a n d N a t h e n s o n et al.
[211.
Peptides and the Positive Selection of T-Cell Repertoire in the Thymus As discussed earlier, class-I MHC-restricted T cells recognize antigenic peptides bound to the peptide-binding groove of class-I molecules. The same class-I MHC molecules are used by thymic cortical epithelial cells to positively select the T-cell repertoire during T-cell ontogeny [22-25]. Are the thymic M H C molecules alone mediating the positive selection (fig. 2a), or do they bind self peptides, and then imprint this self recognition on the developing T cells (fig. 2b)?
58
Nikolid-7.ugid/Carbone
a
MHC
--1
[--
I
I
Fig. 2. Two models of intrathymic positive selection. a The first model postulates the exclusive importance of the a-helical portions of MHC molecules in the selection of TCR repertoire, b According to the second model, the helical parts play less importance and the bound self peptides are the major driving force in the repertoire selection. This allows the peptide-binding groove to shape the T-cell repertoire.
Recently several theoretical reports entertained the idea that thymus-specific self peptides [26] or erroneous peptides [27] might participate in positive selection. Experimentally, there has been some evidence that the background genes can influence positive selection. Using hapten-modified H-2 k cells as antigen, Ogata et al. [28] were the first to describe such an influence. Thereafter, Singer et al. [29] published a preliminary report in which they concluded that self class-I K bderived peptide is necessary for the positive
selection of alloantigen-specific C D 4 + T cells by a class-II I-A b molecule. Most recently, Fry et al. [30] described the effect of background genes on the positive selection of V~tl 1 TCR-bearing T cells. In an attempt to determine whether the self peptides are directly involved in positive selection, we examined the OVA-specific repertoire of C T L generated from ( H - 2 K b • H-2Kbm)Fl bone m a r r o w T-cell precursors which matured in H - 2 K b or H - 2 K bm thymus [31]. As described above, in H-2 b animals the OVA-specific CTLs are restricted solely by the K b class-I molecule and are specific for a determinant contained within the 0VA253-276 fragment [5]. In H-2 b thymus, the K b molecule positively selects this C T L repertoire. Importantly, s o m e K bm molecules (e.g. K bm5 and K binS) differ from K b at residues located deep on the floor of the peptide-binding groove (fig. i; table 1), and these residues are not accessible for contact with T C R [10]. Helical regions of both molecules are identical to the helices of K b by primary structure and serologic analysis [20, 21]. In addition, K bin8 and K bm5 substitutions are unlikely to affect folding of other parts of the molecule, since even extensive class-I p o l y m o r p h i s m has little effect on its overall three-dimensional structure [11]. Therefore, if helical regions alone can mediate positive selection, then all three molecules, K b, K brn5 and K bin8, should be the same for the selection of the OVA-specific C T L repertoire. Alternatively, if the floor of the peptide-binding groove can exert any influence on OVA-specific C T L repertoire, then it has to do so via a bound peptide, and we should see differences between repertoires selected on K b versus K bin5 and/or K broS. As can be seen from table 2, K bin8 could not select any OVA-specific T cells.
MHC Class-I Presentation of Self and Foreign Peptides
59
Table 2. The ability of Kb and Kbm molecules to This p r o v i d e s strong, albeit indirect, evimediate the positive selection and antigen presentadence that self peptides are critically intion to OVA and VSV-specific CTL volved in positive selection. F u r t h e r experiments shed light on the MHC OVA VSV Nc structural aspects o f positive selection. We class-I positive presenpositive presenf o u n d a tight correlation between the ability molecule selection tation selection tation of four K bm molecules to positively select O V A + Kb-specific T cells and their capacity Kb + + + + tO present 0VA253-276 to OVA-specific CTLs q_l _1 [31 ] (table 2). M o r e o v e r , experiments using a Kbml + + peptide c o m p e t i t i o n assay d e m o n s t r a t e d Kbm3 + + that K bm8 c a n n o t physically interact with the Kbrn5 + + + + O V A peptide (table 3). T h u s it appears that Kbm8 _ _ + + positively selecting self peptides and antigenic O V A peptide interact with various The positive selection was measured by using F~ --. P bone marrow irradiation chimeras, as described K bm molecules in the same way demonstrat[31]. The presentation was quantified by pulsing the ing that antigenic peptides are closely m i m Kb or K ~m expression targets with relevant synthetic icked by self peptides during intrathymic peptides of OVA and VSV NcP, and determining the positive selection. specific lysis of these target cells by appropriate O n careful consideration this should be CTL. Preliminary results. expected. T h e results o f Maryanski et al. [ 14] and the observations o f Garrett et al. [11] established the concept that class-I presentation involves the binding o f peptides via a c o m m o n m o t i f o f a m i n o acids to compleTable 3. The competition of OVA and VSV pepm e n t a r y pockets within the class-I-binding tides for presentation by Kb and K bins molecules cleft. F o r a selfpeptide to successfully m i m i c a foreign antigen, the two m u s t expose simi- Presenting Antigen Competitor Recognition by VSV Nclar c o n f o r m a t i o n a l determinants to the T C R molecule specific CTL when b o u n d to M H C molecules. In the case o f OVA, this is likely to be achieved by bind- Kb VSV Nc ing to the same class-I pockets. P o l y m o r v s v Nc OVA phisms that significantly change these pockvSV Nc Flu NP ets should u n i f o r m l y affect the binding o f both foreign peptide and the self analog. In Kbm8 VSV Nc VSV Nc OVA contrast to OVA, vesicular stomatitis virus VSV Nc Flu NP (VSV) nucleocapsid protein (NcP) can bind and be presented to T cells by K bmS. FurtherThe assay was performed as described by Mamore, K bin8 expressing t h y m u s can effi- ryanski et al. [12], using a high molar excess (about ciently select for this C T L repertoire (ta- 300 • of competing peptide. The NP-influenza virus nucleoprotein peptide, which binds to D b, but not K b ble 2). Consequently, a class-! molecule that [2], was used as a control. does not present a given antigenic fragment
60
Nikoli6-Zugi6/Carbone
Kb
sr
~
r
I
f--
OVA like self peptides. the OVA-specific TCR
VSV Nc like self pepEs selecting the VSV No.specific TCR
Fig. 3. The mechanism of repertoire selection by self peptides, a The schematic structure of the antigen binding pockets of class I MHC molecules K b and K broS. b Self peptides mimick the foreign antigens during the intrathymic positive selection. This mimicking involves binding to the same pocket of the MHC molecule.
is less likely to select for T cells specific for that antigenic peptide. A model of peptide interaction with M H C pockets during the thymic selection and peripheral antigen presentation is depicted in figure 3. For the sake of simplicity, we will assume that the K b molecule expresses two types of antigen-binding pockets, A and B. We will further consider that all peptides with the ability to bind K b must contact either pocket A or pocket B. One set of antigenic peptides, including OVA, binds to the unique pocket A, whereas other peptides, such as VSV NcP, bind to pocket B (fig. 3a). Among self peptides which bind to pocket A, a small fraction bears resemblance
to OVA when bound in the groove. These peptides mediate positive selection of the OVA-specific CTL repertoire. Certain mutations (for example, in K binS) may alter pocket A such that the class-I molecule is no longer permissive for binding of 0VA253-276 o r any of the OVA-selecting self peptides. It is likely that a different set of self-peptides can now bind to this new subsite (pocket A1) mediating the selection of a unique TCR repertoire in bm8 animals. However, since none of these peptides bears significant structural resemblance to OVA, there is no selection for anti-OVA CTLs. In contrast, VSV NcP and its selecting self peptides maintain their overall configuration within pocket B, en-
MHC Class-I Presentation of Self and Foreign Peptides
abling both K brn8 and K b to mediate intrathymic positive selection and presentation of this particular antigen. Our explanation of the above results would be invalid if the K bm8 mutation acts by altering the TCR contact area; for example, by allosterically changing the a~ (or ct2) helix. Although the only technique capable of unequivocally determining this is X-ray crystallography, at least three lines of evidence argue against it: (i) K bm8 and K b share the same serological profile, and the monoclonal antibodies used for this analysis recognize the helical parts of the molecule [20]; (ii) VSV NcPs can be efficiently presented to Kb-restricted CTLs by K bin8 (and vice versa) which should not occur if the TCR contact residues on the helices of K bin8 are perturbed (table 3) [J. Nikolid-Zugid, unpublished data]; and (iii) crystallographic data on HLA class-I molecules indicate that despite 13 amino acid differences between HLA-A2 and Aw68, the two molecules appeared remarkably identical, indicating that individual substitutions produce only local effects Jill. How general is the involvement of self peptides in positive selection? We do know of two cases in which class-I MHC K bml molecule could positively select the appropriate TCR repertoire specific for Sendai virus [32] and VSV NcP [J. Nikolid-Zugid, preliminary results], but was not able to present the same antigen in the periphery. Some of these results have been explained by extrathymic rather than intrathymic positive selection [32]. The K bml molecule, however, contains a very drastic structural mutation, and behaves as a maverick among K bm molecules, since it is incapable of presenting any antigen to Kb-restricted T cells [21]. It is therefore quite possible that the observed effect of
61
reflects an exceptional cross-reaction rather than a model for positive selection. The suggestion that self peptide-mediated position selection may also offer the best explanation for the results obtained by Shimojo et al. [33]. These authors have demonstrated that a change at position 156 of HLAA2 from a Leu to Trp may be responsible for the positive selection of a unique CTL repertoire specific for the influenza virus matrix peptide 55-73. The change at this residue has been shown to take part in the formation of a ridge on the floor of the HLA-Aw68binding cleft and is most likely inaccessible to direct contact with TCR [11]. While this human experimental model is not able to provide direct answers to questions regarding positive selection, the results are best explained by postulating the involvement of self peptides in the selection process. As described earlier, this change at position 156 in HLA-A2 is also thought to affect T-cell recognition by altering the conformation of the bound matrix peptide [17]. Consequently, changes in the groove that result in new conformations of presented antigenic peptide may be able to select for a unique T-cell repertoire as well. Finally, D.Y. Lob and his collegues (personal commun.) have obtained evidence to suggest a critical role for self peptides in the positive selection of the Tcell population in transgenic mice expressing the receptor from an alloreactive T-cell clone. In total, all these data indicate that self peptides universally participate in positive selection. This proposal immediately raises the issue as to whether class I is being recognized at all during the positive selection (and antigen presentation), or whether it merely serves to put peptides into adequate conformation for TCR recognition. A detailed disK bml
62
cussion of this issue has recently been published by Germain [34]. While it is likely that TCR does indeed contact class I directly, we favor the view that MHC behaves largely as a frame which shapes and is being shaped in part by bound peptide. That TCR recognition of bound peptide probably dominates over MHC recognition is stressed by the fact that polymorphism in class-I residues contacting peptide is significantly greater than in residues contacting TCR directly [10]. It is currently uncertain to what extent polymorphism at TCR-binding residues affects T-cell specificity during the process of positive selection. Given that self peptides are involved in the thymic selection of the class-I-restricted T-cell repertoire, it remains to be shown what the origin of these fragments is. This explanation must of course reconcile the fact that individuals use the same MHC molecules and probably the same self antigens for both positive and negative selection to generate a broad and non-autoaggressive T-cell repertoire. One possibility would have the affinity threshold for positive selection to be considerably lower than that required for negative selection or peripheral T-cell activation. Alternatively, it has been suggested that positive selection involves a unique set of self peptides antigens within the thymic cortex [26]. To accommodate the need for as broad a repertoire as possible, Kourilsky and Claverie [27] have further suggested that these thymic-specific antigens are actually short mutated peptides derived by some aberrant transcriptional and translational mechanism. All these preceding explanations rely largely on the premise that peptide presentation by cortical epithelium in the thymus involves the same processing mechanisms
Nikolid-Zugid/Carbone
which have been defined for peripheral class-I presentation, namely, the endogenous pathway of class-I-restricted antigen processing. The diversity of endogenously derived class-I-associated self peptide determinants would, by necessity, be quite limited, and may not be sufficient for the generation of a respectable TCR repertoire. Furthermore, the same determinants would be readily available to clonally delete, or functionally paralyze, the TCR for they select. In contrast, it may be possible that class-I-mediated positive selection might actually involve the presentation of exogenous self peptides derived by some extracellular, rather than intracellular, proteolytic event. Exogenous peptide presentation is capable of expanding a specific T-cell population both in vivo [35] and in vitro [36-38]. These T cells are considered nonphysiological since they fail to recognize class-I-associated peptides formed as a result of endogenous gene expression [36-38]. Such T cells are also not deleted intrathymically by the same self peptides derived from endogenous antigen expression. For example, Schild et al. [39] showed that they could generate in vitro derived CTLs to a number of self antigens, including ~2-microglobulin, by exogenous peptide presentation. The lack of clonal deletion of such CTLs is easily explained by insufficient density of self peptide determinants presented by hematopoietic thymic elements. This theory has to invoke a different determinant density to explain the differences between positive and negative selection. We favor this explanation because it elegantly accommodates the self peptide model of positive selection without going out of the way to explain differences between the peptides involved in positive selection, negative selection and peripheral activation. The
MHC Class-I Presentation of Self and Foreign Peptides
g e n e r a t i o n o f d i v e r s i t y is a c h i e v e d by the fact t h a t e x t r a c e l l u l a r proteolysis can p o t e n tially g e n e r a t e far m o r e class-I-binding frag-
63
repertoire diversity, avoiding both the massive clonal deletion of the selected repertoire and the autoreactivity of its T cells.
m e n t s t h a n the usual cytosolic p r o c e s s i n g m e c h a n i s m [5, 36]. T h e T cells selected on these p e p t i d e s w o u l d h a v e only l i m i t e d reactivity to class I / p e p t i d e c o m b i n a t i o n resulting f r o m e n d o g e n o u s processing (and a part
Acknowledgement The authors wish to thank Dr. M.J. Bevan for his support.
o f these c o u l d be u n i q u e to the cortical epit h e l i u m ) , l e a v i n g the bulk o f the T-cell repertoire intact. T h e s a m e principles can easily
References
be a p p l i e d to c l a s s - I I - m e d i a t e d selection.
Summary MHC class-I molecules express distinct peptidebinding pockets within their antigen-binding groove. These are critically involved in the binding of antigenic peptides. The amino acid composition of a pocket dictates the structure of a peptide which can be bound in it. This is evident as a consensus amino acid motif which has to exist within a peptide in order for it to bind to a particular MHC allele. Perturbation ofa MHC pocket by amino acid substitution can result in the abolition of peptide binding. Less drastic mutations of the peptide-binding groove, particularly the ones away from the critical pocket, can subtly alter the conformation of bound peptide. Both types of substitution exert an influence on the TCR recognition of antigenic peptide. Peptides are also critically involved in the positive selection of the class-l-restricted TCR repertoire in the thymus. These self peptides act by mimicking their foreign antigens. This mimicking involves the binding of selfpeptides and foreign antigenic peptides to the same pockets of the MHC class-I-antigen binding groove. Consequently, MHC class-I polymorphism in the antigen binding groove controls the intrathymic positive selection and peripheral antigen presentation by the same mechanisms. The majority of positively selecting self peptides could well originate from the extracellular processing of circulating self proteins. Using the diverse, extracellularly generated self peptides and the different determinant density requirements for positive versus negative selection, the immune system can ensure the
1 Townsend ARM, Gotch FM, Davey J: Cytotoxic T cells recognize fragments of the influenza nucleoprotein. Cell 1985;42:457-467. 2 Townsend ARM, Rothbard J, Gotch FM, Bahadur G, Wraith D, McMichael AJ: The epitopes of influenza nucleoprotein recognized by CTL can be defined with short synthetic peptides. Cell 1986;44:959-968. 3 Maryanski JL, Pala P, Corradin G, Jordan BR, Cerottini JC: H-2-restricted cytolytic T cells specific for HLA can recognize a synthetic HLA peptide. Nature t986;324:578-579. 4 Yewdell JW, Bennink JR, Hosaka Y: Cells process exogenous proteins for recognition by CTL. Science 1988;239:637-640. 5 Moore MW, Carbone FR, Bevan MJ: Introduction of soluble protein into the class I pathway of antigen processing and presentation. Ceil 1988: 54:777-785. 6 Yewdell JW, Bennink JR: Brefeldin A specifically inhibits presentation of proteins for recognition by CTL Science 1989;244:1072-1075. 7 Nuchtern JG, Bonifacino JS, Biddison WE, Klausner D: Brefeldin A implicates egress from endoplasmic reticulum in class I restricted antigen presentation. Nature 1989;339:223-226. 8 Cox JH, Ye~vdcll JW, Eisenlohr LC, Johnson PR, Bennink JR: Antigen presentation requires transport of MHC class I molecules from the endoplasmic reticulum. Science 1990;247:715-718. Bjorkman P J, Saper MA, Samraoui B, Bennett WS, Strominger JL, Wiley D: Structure of the human class I histocompatibility antigen, HLAA2. Nature 1987;329:506-512. 10 Bjorkman P J, Saper MA, Samraoui B. Bennett WS, Strominger JL, Wiley DC: The foreign anti-
64
11
12
13
14
15
16
17
18
19
20
21
Nikolid-Zugid/Carbone
gen binding site and T cell recognition regions of class I histocompatibility antigens. Nature 1987; 329:512-518. Garrett TPJ, Saper MA, Bjorkman P J, Strominger JL, Wiley DC: Specificity pockets for the side chains of peptide antigens in HLA-Aw68. Nature 1989;342:692-696. Maryanski JL, Pala P, Cerottini J-C, Corradin G: Synthetic peptides as antigens and competitors in recognition by H-2-restricted cytolytic T cells specific for HLA. J Exp Med 1988;167:1391-1405. Pala P, Bodmer HC, Pemberton RM, Cerottini J-C, Maryanski JL, Askonas BA: Competition between unrelated peptides recognized by H-2K d restricted T cells. J Immunol 1988;141:22892294. Maryanski JL, Verdini AS, Weber PC, Salemme FR, Corradin G: Competition analogs for defined T cell antigens: Peptides incorporating a putative binding motif and polyproline on polyglycine spaces. Cell 1990;60:63-72. Guillet JG, I_ai M-Z, Brinner T J, Buus S, Sette A, Grey HM, Smith JA, Gefter ML: Immunological sell nonself discrimination. Science 1987;235: 865-870. Sette A, Buus S, Colon S, Miles C, Grey HM: IAd-binding peptides derived from unrelated protein antigens share a common structural motif. J Immunol 1988; 141:45-81. Hogan KR, Shimojo H, Walk SF, Engelhard VH, Maloy WL, Coligan JE, Biddison WE: Mutations in the alpha two helix of HLA-A2 affect presentation but do not inhibit binding of influenza virus matrix peptide. J Exp Med 1988;168:725-736. McMichael A J, Gotch FM, Santos-Aguado J, Strominger JL: Effect of mutations and variations of HLA-A2 on recognition of a virus peptide epitope by cytotoxic T lymphocytes. Proc Natl Acad Sci USA 1988;85:9194-9t98. Shimojo N, Anderson RW, Mattson DH, Turner RV, Coligan JE, Biddison WE: The kinetics of peptide binding to HLA-A2 and the conformation of the peptide-A2 complex can be determined by amino acid side chains on the floor of the peptide binding groove. Int Immunol 1990;2:193-200. Nikolid-Zugid J, Carbone FR: CTL recognition of foreign peptide is modified by amino acid changes in the MHC class I peptide binding groove. Eur J Immunol 1990;20:2431-2437. Nathenson SG, Geliebter J, Pfaffenbach GM, Zeff
RA: Murine major histocompatibility complex class I mutants. Annu Rev Immunol 1986;4:471502. 22 Bevan M J: In a radiation chimera, host H-2 antigens determine immune response of donor cytotoxic cells. Nature 1977;269;417-418. 23 Zinkernagel RM, Callahan GN, Althage A, Cooper S, Klein PA, Klein J: On the thymus in the differentiation of H-2 'self-recognition' by T cells: Evidence for dual recognition. J Exp Med 1978; 147:882-897. 24 Kisielow P, Teh HS, Bluthmann H, von Boehmer H: Positive selection of antigen-specific T cells in thymus by restricting MHC molecules. Nature 1988;335:730-733. 25 Lo D, Sprent J: Identity of cells that imprint H-2 restricted T-cell specificity in the thymus. Nature 1986;319:672-678. 26 Marrack P, Kappler J: The T cell repertoire for antigen and MHC. Immunol Today 1988;9:308315. 27 Kourilsky P, Claverie JM: MHC restriction, alloreactivity and thymic education: A common link? Cell 1989;56:327-329. 28 Ogata M, Shimizu J, Tsuchida T, Takai Y, Fujiwara H, Hamaoka T: Non-H-2-1inked genetic regulation of cytotoxic responses to hapten-modified syngeneic cells. J Immunol 1986;136:1178-1185. 29 Singer A, Mizuochi T, Munitz TI, Gress RE: The role of self antigens in the selection of the developing T cell repertoire. Prog Immunol 1986:6:6066. 30 Fry AM, Cotterman MM, Matis EA: The influence of self-MHC and non-MHC antigens on the selection of an antigen-specific T cell receptor repertoire. J Immunol 1989; 143:2723-2729. 31 Nikoiid-Zugid J, Bevan M J: Role of setf-peptides in positively selecting the T-cell repertoire. Nature 1990;344:65-67. 32 Kast WM, deWaal LP, Meleif CJM: Thymus dictates major histocompatibility complex (MHC) specificity and immune response gene phenotype of class II MHC-restricted T ceils but not of class I MHC-restricted T cells. J Exp Med 1984;160: 1752-1766. 33 Shimojo N, Cowan EP, Engelhard VH, Maloy WL, Coligan JE, Biddison WE: A single amino acid substitution in HLA-A2 can alter the selection of the CTL repertoire that responds to influenza'virus matrix peptide 55-73. J Immunol 1989;143:558-564.
MHC Class-I Presentation of Self and Foreign Peptides
34 Germain RN: Making a molecular match. Nature 1990;344:19-21. 35 Ishioka GY, Colon S, Miles C, Grey HM, Chesnut RW: Induction of class I MHC-restricted, peptidespecific cytolytic T lymphocytes by peptide priming in vivo. J Immunol 1989;143:1094-1100. 36 Carbone FR, Moore MW, Sheil JM, Bevan MJ: Induction of cytotoxic T lymphocytes by primary in vitro stimulation with peptides. J Exp Med 1988; t 67:1767-1779. 37 Carbone FR, Hosken NA, Moore NW, Bevan MJ: Class I MHC-restricted cytotoxic responses to soluble protein antigen. Cold Spring Harbor Symp Quant Biol 1989;54:551-555.
65
38 Staerz UD, Zepp F, Schmid R, Hill M, Rothbard J: Recruitment of alloreactive cytotoxic T lymphocytes by an antigenic peptide. Eur J Immunol 1989;19:2191-2196. 39 Schild H, Rotzchke O, Kalbacher H, Rammensee H-G: Limit of T cell tolerance to self-proteins by peptide presentation. Science 1990;247:15871589. Dr. Janko Nikotid-7.ugid Immunology Program Sloan-Kettering Institute 1275 York Avenue New York, NY 10021 (USA)
