Effects of Retroviral Protease Inhibitors on Proteasome Function and Processing of HIV-Derived MHC Class I-Restricted Cytotoxic T Lymphocyte Epitopes

Abstract

1063 THE HIV-1 PROTEASE INHIBITOR RITONAVIR (RTV) potently suppresses viral replication and is an effective antiretroviral agent in the clinical setting.1–3 Although originally designed rationally to limit cross-reactive inhibition of structurally similar human aspartic proteases, concern has been raised regarding the negative effects of RTV in murine systems on both the generation of CD8 cytotoxic T lymphocyte (CTL) responses and the processing of major histocompatibility (MHC) class Irestricted peptide epitopes recognized by these effector cells.4,5 We assessed the impact of RTV, and other antiretroviral protease inhibitors, on human proteasome function and the generation of HIV-derived CTL peptide epitopes with well-characterized processing requirements. The proteasome, a large nonlysosomal multisubunit protease complex that accomplishes many of the essential proteolytic functions of the cell, is critically involved in the generation of antigenic peptides destined for the MHC class I pathway. Structurally, the core 20S proteasome comprises four heptameric rings arranged such that the a and b subunits, which range in size from 20 to 35 kDa, form a 700-kDa complex with the configuration a7b7b7a7. Proteolytic activities are restricted to specific threonine residues on certain b subunits, a catalytic mechanism characteristic of N-terminal nucleophile hydrolases.7 At least five distinct peptidase activities, based on amino acid cleavage preferences, have been ascribed to the proteasome; RTV has been shown to inhibit the chymotryptic activity, characterized by cleavage after hydrophobic residues, of isolated murine 20S proteasomes in vitro.4 We confirmed and extended this result with 20S proteasomes purified from human erythrocytes (Fig. 1a4,8). The b subunits MB1 and LMP7 have been identified as prime target sites for the reversible inhibitory effect of RTV on proteasome activity.5 The LMP7 subunit is interferon g (IFN-g) inducible, and replaces the constitutive MB1 subunit in “immunoproteasomes.” We identified two HLA-A2-restricted epitopes from the HIV-1 reverse transcriptase (RT) protein, ILKEPVHGV (Pol, residues 476–484HIV-LAI) and VIYQYMDDL (Pol, residues 346–354HIV-LAI), which are processed through distinct pathways.9 The C-terminal ILKEPVHGV epitope is generated by the constitutive proteasome containing MB1 and is independent of LMP7-associated activity. In contrast, the Nterminal VIYQYMDDL epitope appears to be generated by an alternative antigen-processing pathway and destroyed by the constitutive proteasome; both the expression of LMP7, either in transfection systems or through induction with IFN-g, and inhibition of proteasomal threonine protease activity with lactacystin, substantially enhanced the production of this epitope.9 In this well-characterized system, we were unable to detect any effect of RTV on antigen presentation at therapeutically relevant concentrations (Fig. 1b9). This observation suggests either that the generation of these epitopes is regulated independently of the chymotryptic activity of the proteasome, or that the effects of RTV at concentrations within the therapeutic window are discrepant according to whether the proteasome is examined in purified or cell-associated form. The former possibility is currently difficult to test since the precise anatomical localization of specific enzymatic activities and the effects of LMP7 incorporation on proteasomal hydrolysis patterns remains controversial. Consistent with the latter possibility, RTV had no net quantitative effect on the assembly of HLA class I molecules in pulse–chase experiments, indicating that this process is not limited by impaired production of optimized peptides capable of stabilizing these proteins in nascent form (Fig. 1c10). However, this does not exclude the possibility that the pattern of epitopes produced differs qualitatively in the presence of RTV, and further work is required to resolve this issue. In addition, we could detect no consistent effect of RTV (concentration range, 0.1–10 mM) on the generation of two commonly immunodominant CTL epitopes derived from the HIV-1 Gag protein, SLYNTVATL (p17, residues 77–85HIV-LAI, HLA-A2