The HLA-class I Restricted CTL Response in HIV-1 Infection:
Identification of Optimal Epitopes

Christian Brander and Bruce D. Walker

Massachusetts General Hospital, AIDS Research Center, 149 13th Street, Room 5234, Charlestown, MA 02129, USA, and Harvard Medical School Boston, MA USA

I. Introduction
HIV-1 infection is associated with a vigorous HLA class I restricted cytotoxic T lymphocyte (CTL)response [reviewed in 1]. By analogy to a number of animal models of viral infection, these CTL are likely to play an important role as a host defense in infected persons and are likely to be an important component of an effective AIDS vaccine. A large number of laboratories have now contributed to the identification of peptides containing HIV-1 CTL epitopes. In terms of class I-restricted CTL epitopes, studies have shown that these peptides are typically 8M-P11 amino acids in length and conform to certain motifs for different class I molecules ². A comprehensive list of peptides reported to contain putative CTL epitopes has been included in the data base. Many of these have not been further defined as to the optimal peptides required for recognition, or the restricting class I molecule. In addition, a number of peptides have been reported which do not appear to contain regions corresponding to binding motifs for the restricting class I molecules. Therefore some of these peptides require further investigation to confirm their immunogenicity and their HLA restriction. We attempt here to identify those HIV derived epitopes for which the optimal peptide has been precisely identified and for which the restricting molecule is well defined.

II. Characteristics of viral CTL epitopes
MHC class I restricted CTL recognize processed viral peptides that are presented in an antigen binding site on the class I molecule formed between two alpha helices, with the floor of the groove formed by a beta sheet ^2,3. A number of methods have been used to determine the identity of the presented peptides. The most precise definition has come from elution of such peptides from class I molecules, revealing consistent motifs shared by peptides presented by the same class I molecule. The binding motifs are characterized by anchor residues which serve as contact sites between the peptide and specific pockets in the class I binding groove ⁴. The amino and carboxy termini of epitopic peptides fit into the A and F pockets of this groove, respectively, due to hydrogen bonding. The amino terminal anchors of the peptides are variable in position and number, whereas the carboxy anchor is always at the C-terminus, with the side chain pointing directly into the bottom of the F pocket. Although there is considerable heterogeneity among amino terminal anchor residues, the F pocket appears to place more restrictions on the amino acids it will accommodate, such that either leucine, isoleucine, arginine, tyrosine, valine or phenylalanine is at the C-terminus of over 95% of known epitopes ⁵. Motif predictions for antigenic peptides have now been generated for a large number of HLA class I molecules, although the majority of class I epitope motifs are yet to be defined.

III. HIV-1 CTL epitopes
HIV-derived, HLA class I restricted CTL epitopes have been defined by a number of different strategies, not only in terms of the effector cells and the target cells used, but also in terms of the method of antigen presentation. Effector cells have consisted of freshly isolated bulk PBMC, CD8-positive cell lines stimulated in vitro with either whole virus, recombinant antigen, or peptide containing a CTL epitope, and CTL clones obtained by limiting dilution [reviewed in 1]. Each of these methods has led to reliable definition of certain epitopes. In terms of antigen presentation, the majority of CTL epitopes have been defined using a two part strategy consisting of recombinant protein expression, usually with vaccinia expression systems, followed by the use of short synthetic HIV-1 peptides to sensitize autologous B cell lines for lysis. Optimal epitopes have usually been defined as that peptide which results in maximal lysis at the lowest peptide concentration.

With the description of HLA allele specific binding motifs, a new method to identify peptides binding to a certain class I molecule became possible. Based on the anchor residues described for various HLA alleles, viral protein sequences can be screened for the presence of such anchor residues ^4,6. These peptides are then synthesized and tested either in cellular or cell free assays for their capacity to bind to the selected class I molecule, and the ability of peptides thus defined to serve as a target for CTL in natural infection is then determined, typically by in vitro stimulation of PBMC with the peptide in question. This M-RpredictiveM-S approach can also facilitate the identification of optimal peptides when larger peptides containing CTL epitopes have been defined, by identifying regions in the larger peptide which conform to the proposed motif. A more precise method to define optimal CTL epitopes is represented by peptide elution studies, although only few such studies have been performed in the HIV-1 system⁷. Infected cells are expanded to large numbers and their HLA class I molecules are isolated. Peptides within the class I binding groove can then be released from the class I molecule with trifluoroacetic acid and analyzed by HPLC. The relevant peaks in the HPLC profile can be sequenced and represent epitopes produced within the infected cell^8,9. Because epitopes identified in this manner have been identical to optimal peptides identified by performing peptide titrations with truncated synthetic peptides, the latter system appears to be a reasonable alternative ⁷.

Since the first description of HIV specific CTL 8 years ago, an increasing number of CTL epitopes have been identified by using one of the above described methods. However, the reliability of the described optimal peptides depends strongly on the method(s) used. The complete list of HIV derived peptides containing CTL epitopes is included in the data base. Because the same rigid criteria have not been applied for all of the peptides containing HIV-1 CTL epitopes published to date, we have compiled here a list of those peptides for which there are reasonably good data regarding the HLA restriction and optimal peptide (Table 1). From almost 200 HIV peptide sequences reported to contain class I restricted CTL epitopes, only peptides were selected which 1) have been truncated to the minimal peptide required for sensitization, 2) where the optimal epitope has been identified based on motif prediction, dilution curves or peptide elution and 3) where clear definition of the HLA restriction element has been shown. The listed peptides do not necessarily fulfill the binding motifs given in Table 1, reflecting the limited efficiency of the binding motif based predictive approach. The peptides reported to contain CTL epitopes but which are not included in Table 1 lack in most cases data on the minimal epitope. Furthermore, there are about 20 additional peptides for which the HLA restriction has not been clearly defined thus far.

IV.Emerging characteristics of HIV-1 CTL epitopes
Although the list of HIV derived CTL epitopes is far from being complete, a few features already emerge. There are far fewer epitopes identified in the natural immune response to HIV than the prediction based on binding motifs would suggest, which may be due to a variety of factors. One is that the number of predicted binding candidates depends strongly on the accuracy of the binding motif. Thus, a well defined binding motif reduces the number of peptides that seem to have binding capacity and which therefore have to be tested for binding; and it also enhances the chances of the remaining peptides to actually bind to the class I molecule. It is also likely that not every binding peptide will be generated in the infected cell, or that there are other constraints on processing or binding to the newly synthesized class I molecule in the endoplasmatic reticulum such that the peptides are nor presented at the cell surface (i.e. flanking sequences, ref. 10). Furthermore, sequences of peptides very similar to self sequences may not be targeted by precursor cells within the self-tolerant CTL population, which may additionally reduce the number of possible CTL epitopes.

The expanding list of CTL epitopes also provides a perspective on regions which appear to be highly immunogenic for CTL responses. One example is a 10a.a. sequence in gp41(589M-P598), in which three different class I molecules (HLA A24, B8 and B14) present peptides containing CTL epitopes ^11,12. Another region of such overlapping CTL epitopes is the RT region 476M-P485, containing epitopes restricted by HLA-A2 and -Bw62, and the region 73M-P82 in the Nef protein, which contains epitopes restricted by HLA-A3, A11 and B35 ¹³. Recently, several epitopes were defined within a short region at the C-terminal end of the Nef protein (183M-P196, HLA-A2, A52, B35)¹⁴. It is not known how this accumulation of CTL epitopes within certain regions occurs. In non-immunogenic regions, it is possible that some parts of the proteins are not accessible for the processing machinery (proteasomes) of the infected cell, and thus no CTL epitopes are generated from those areas, perhaps due to sequence peculiarity or to interactions of the viral proteins with host structures or viral products. In contrast, highly immunogenic areas may have advantageous flanking regions that allow efficient processing of these regions ¹⁰.

Sequence variation has been observed in all the peptides in the table. Although there are differences in the degree of conservation, no epitope is derived from an absolutely conserved region, although some epitopes are localized in highly functional regions. For example, the gp120 peptide 424M-P432 (restricted by HLA-A32) contains sequences involved in CD4 binding and the peptide Gag p17/18M-P26 (HLA-3.1 restricted) is located within a region required for nuclear localization, which is important in infection of non-dividing cells [15]. Peptide RT 346M-P354 (HLA-A2 restricted) contains part of the active catalytic site of the HIV polymerase [16]. All these three peptides show only moderate intrinsic variability and may thus represent relatively conserved epitopes.

V.Conclusions
The investigation of the CTL response to HIV-1 has led to the most extensive characterization of the spectrum of class I restricted CTL epitopes of any viral infection studied this far. However, numerous questions remain that are of particularly acute interest. The reasons for the observed decline in CTL activity with disease progression remains undefined, and issues regarding immune escape and clonal exhaustion need further investigation ^{1, 17}. The relationship between CTL pressure and in vivo sequence variation needs to be defined, as does the potential for altered peptide ligands to influence disease progression ^{18, 19, 20}. The relative contributions of specificity and breadth of the response to in vivo efficacy remain controversial, with some studies suggesting that a broadly directed response is favorable whereas others suggest that a narrowly directed response to a dominant epitope is advantageous ^21,22. Especially with regard to the development of a CTL vaccine, this controversy and the demonstration of the protective role of CTL in HIV infection have to be resolved. To address these issues it is essential to characterize the CTL response in its epitope-diversity as well as in its T cell receptor usage, and to compare these findings with variations in the viral genome. To this end, all available information about CTL epitopes, their variability and the responding T cell populations has to be verified and collected to offer a reliable basis to understand the cellular immunologic events occurring after HIVinfection. We thus plan to update this list periodically and would appreciate further information from any investigator who can contribute new data to keep this list current.

Acknowledgments We thank the investigators who have provided unpublished data for inclusion in table 1. This work was supported by the Swiss National Foundation and a grant from the NIH (R37 AI 28568).

Table 1 Best Defined HIV CTL Epitopes

---

References