Antigen presentation by major histocompatibility complex class I (MHC-I) molecules is the cornerstone of adaptive immune surveillance and CD8+ T cell cytotoxic responses. In the principal pathway of this process, which operates in all nucleated cells, the peptides destined to bind MHC-I are generated through sequential proteolysis: cellular proteins are first degraded in the cytosol by the proteasome, and a fraction of the resulting fragments then reaches the lumen of the endoplasmic reticulum via the TAP transporters. However, the size of the fragments produced by the proteasome is not optimal for binding to MHC-I. Only peptides of 8 to approximately 16 amino acids are efficiently transported by TAP, and a large proportion of them still require trimming of their N-terminus to attain the appropriate length.
This final adjustment step is carried out by the endoplasmic reticulum aminopeptidases, the ERAP enzymes, identified in the early 2000s. The mouse genome encodes a single one, designated ERAP or ERAAP, whereas the human genome encodes two, ERAP1 and ERAP2, all belonging to the M1 family of aminopeptidases and to the oxytocinase subfamily. This review highlights the considerable impact of these enzymes on the repertoire of ligands presented by MHC-I: in their absence, the production of certain epitopes may be decreased, unchanged, or, conversely, increased in the case of epitopes that are normally "over-trimmed." Genetic deletion of ERAP also leads to an approximately 20% reduction in cell-surface MHC-I, attributed to reduced stability and probably to increased internalization and degradation of surface molecules, whereas the assembly and export of MHC-I through the secretory pathway retain normal kinetics.
While the critical role of ERAP in establishing an optimized ligandome is well established, how this outcome is achieved mechanistically remains uncertain. The authors therefore compare the two alternative models proposed to explain how ERAP activity adapts to the binding requirements of MHC-I: the "MHC template" model, according to which the MHC-I molecule guides the trimming, and the "molecular ruler" model, in which the enzyme itself measures the length of its substrate. Finally, the review examines the evidence supporting dimerization of the two human enzymes ERAP1 and ERAP2 and its possible functional relevance, this association having been linked to enhanced efficiency of peptide trimming.