Antigen-processing and presentation pathways select antigenic HIV peptides in the fight against viral evolution

The evolution of immunodominant epitopes in HIV-1 Gag proteins correlates with quantitative measures of several antigen processing events. Thus, peptides recognized by CD8⁺ cytolytic T cells are selected by their ability to pass through the antigen processing pathway, as well as by their binding to HLA molecules.

The vertebrate immune system uses different 'arms' that act together to combat viral infection. The cells and receptors of innate and adaptive immunity, including natural killer cells, B lymphocytes (and their protein derivatives, antibodies), T cells and specialized antigen-presenting cells work together to identify viruses before the viruses can even adsorb to cells, to mark infected cells for destruction and/or to limit reactivation of persistently or latently infected cells. Each of these activities waxes and wanes throughout the course of viral exposure and infection. Adaptive cells of the immune system like T cells and B cells repeatedly identify specific antigens; these antigens, called 'immunodominant antigens', have become one focus of efforts to develop better vaccination strategies¹. In this issue of Nature Immunology, Tenzer et al.² identify parameters that influence the generation and presentation of immunodominant human immunodeficiency virus type 1 (HIV-1) peptides to cytolytic CD8⁺ T lymphocytes.

Resistance to chronic viral infections such as HIV can be correlated to some degree with ongoing activity of cytolytic CD8⁺ T lymphocytes that express αβ T cell antigen receptors that recognize complexes of peptide and major histocompatibility complex (MHC) class I in which the peptides are derived from viral proteins. A major challenge in understanding the variety of immune responses to viral infection has been to identify the parameters, both molecular and cellular, that determine which peptides from which viral proteins are most commonly selected for presentation at the cell surface and to understand how those peptides are generated and loaded onto MHC class I molecules. However, the scientific challenge is not only to understand the rules that govern the generation of immunodominant peptides but also to elucidate the options available to a rapidly evolving virus able to modify the protein from which such peptides derive. Tenzer et al. build on previous work showing that wild-type sequences encoding an HIV-1 group-associated antigen (Gag) T cell epitope undergo natural escape mutations that then can ultimately revert back to the wild-type sequence³. Here, the authors study the wild-type sequence of the Gag protein p17 as well as seven common natural variants of this sequence to evaluate the influence of several crucial parameters of the MHC class I antigen-processing and antigen-presentation pathway on the detection of each variant peptide by CD8⁺ T cells. For the sake of completeness, they also analyze epitopes from the p24 Gag protein, but the discussion here is limited to the results obtained with p17.

Underlying their approach is the view that the final product of this complex antigen-processing and antigen-presentation pathway—that is, the cell surface peptide–MHC class I complex that is subsequently recognized by CD8⁺ T cells—is generated as a result of the quantitative efficiency of each step of the pathway and that these efficiencies can be examined by reconstitution of the relevant steps in vitro. The main steps of the pathway include cytosolic production of peptide precursors, transport of intermediate peptides from the cytosol to the peptide-loading complex of the endoplasmic reticulum by the transporter associated with antigen processing (TAP), and maturation and folding of peptide–MHC class I complexes expressed on the cell surface (Fig. 1). Studying all steps of the pathway completely for any given virus (or protein, for that matter) would be a next-to-impossible task, so the authors confine themselves to the merely overwhelming job of studying several steps of the pathway.

Figure 1: Steps in antigen processing and presentation in the MHC class I pathway.

Native, partially unfolded, or denatured protein in the cytosol (1) is further denatured concomitant with proteolytic processing by constitutive or immunoproteasomes (2). In experiments reported in this issue², 25–amino acid peptides are used as substrates for isolated proteasomes. Intermediate peptides that result from proteasome degradation (3) then bind TAP (4) and are transported to the peptide-loading complex (5), where the sequential and concerted effects of tapasin, ERp57, calreticulin, MHC class I heavy chain, β₂-microglobulin and ERAAP1,2 lead to the formation of a stable MHC class I–β₂-microglobulin–peptide complex (pMHCI; 6). This complex is transported via the Golgi to the cell surface, where it is available for binding to αβ T cell antigen receptor (αβTCR) and CD8αβ molecules on the surface of CD8⁺ T cells. ER, endoplasmic reticulum.

Using synthetic 25–amino acid peptides with substitutions representing the sequences commonly found in escape variants, the authors first evaluate the relative abundance of smaller, or intermediate, peptides generated by the proteolytic activity of purified proteasome preparations⁴. These peptides of intermediate length are potentially able to be transported by TAP into the endoplasmic reticulum lumen and to bind by their carboxyl termini to endoplasmic reticulum–resident MHC class I molecules. Using fragment intensity identified by mass spectrometry as an indicator of the abundance of the intermediate peptides produced by proteasomal digestion, the authors find that the wild-type 25–amino acid peptide generates the precursor for the immunodominant peptide presented by HLA-A2 (SLYNTVATL) rather poorly. In addition, they note that different variant 25–amino acid peptide precursors give rise to distinct peptides of intermediate length. Next they measure the affinity of synthetic peptides that represent each of the intermediate peptides for TAP. As TAP-mediated transport is dependent on TAP affinity, peptides with a higher affinity for TAP are expected to be proportionately better represented in the pool of intermediates that reach the peptide-loading complex in the endoplasmic reticulum⁵. After that, they measure the relative efficiency with which ERAAP1,2, an endoplasmic reticulum–lumenal enzyme, trims each intermediate peptide. Finally, they determine the affinity of the presenting HLA molecule for each of the trimmed peptides. They find that cleavage 'preference', TAP binding, ERAAP trimming and HLA affinity all collectively contribute to the relative hierarchy of T cell recognition of the resulting peptide antigen.

In general, the overall hierarchy of T cell recognition of p17 and p24 epitopes correlates with a calculated function of epitope 'abundance' rather than with peptide affinity for HLA molecules. In addition, the detection of differences in the responses of HIV patients to peptide variants with a carboxyl-terminal extension of the core epitope prompted structural analyses of this elongated peptide bound to HLA-A2. The crystal structure shows that this elongated p17 core peptide may bind to HLA-A2 in a way that results in the adoption of an unusual 'flipped' orientation by its carboxy-terminal tyrosine residue. Thus, although HLA affinity certainly contributes to the hierarchy of CD8⁺ T cell recognition, it is not the sole predictor of the cytotoxic T lymphocyte response. Peptide abundance and structure when bound to HLA molecules probably also influence cytotoxic T lymphocyte recognition.

Does the enormous amount of data provided by the analysis by Tenzer et al.² provide a better sense of the parameters that might control immunodominance? If so, can a sensible theoretical course that might guide the empirical approach to vaccine development be outlined? In terms of the first question, one general take-home message of this study is not completely unexpected: the spectrum of peptides generated from a rather short 25–amino acid peptide precursor after proteasomal processing is very large, and this spectrum can be widely influenced by one to four substitutions in the sequence of the peptide precursor. The finding that different peptide intermediates are transported with different efficiency is also predictable, and the finding that different sequences are trimmed differently by ERAAP1,2 is expected⁶. However, for the second question, the overall analysis emphasizes the naiveté of believing that researchers could look at the sequence of a single wild-type protein, predict—using bioinformatics—its capacity to bind one or several common HLA molecules, and know with assurance exactly which protein sequence to incorporate into a favorite immunogenic vector.

Of course, these findings do not provide a simple guideline for choosing the 'right' peptide every time; HIV-1 is a rapidly mutating virus that can tolerate sequence variation in an essential protein for many generations sufficient to facilitate evasion of an immune response and then 'bounce back' by reverting to the more 'fit' wild-type sequence. Thus, not unlike the solutions to global warming or dependence on foreign oil, the proper immunization for rapidly evolving viruses cannot be a single T cell vaccine but instead must combine many different modalities that, like the 'cooperative arms' of the immune system, work in a complementary way to counter several steps of a pathogen's life cycle.