Evidence for successive peptide binding and quality control stages during MHC class I assembly.

Intracellular antigens are continually presented to cytotoxic T lymphocytes by major histocompatibility complex (MHC) class I molecules, which consist of a polymorphic 43 kDa heavy chain and a 12 kDa soluble subunit beta 2-microglobulin (beta 2m), and which bind an 8-10 amino-acid antigenic peptide. The assembly of this trimolecular complex takes place in the lumen of the endoplasmic reticulum (ER) and almost certainly requires cofactors. Most MHC class I molecules in the ER that have not yet acquired peptide are simultaneously bound to the transporter associated with antigen processing (TAP), to the 48 kDa glycoprotein tapasin and to the lectin-like chaperone calreticulin, in a multicomponent 'loading complex'. Previous studies have shown that a mutant MHC class I molecule T134K (in which Thr134 was changed to Lys) fails to bind to TAP. Here, we show that this point mutation also disrupted, directly or indirectly, the interaction between MHC class I molecules and calreticulin. T134K molecules did not present viral antigens to T cells even though they bound peptide and beta 2m normally in vitro. They exited the ER rapidly as 'empty' MHC class I complexes, unlike empty wild-type molecules which are retained in the ER and degraded. We show here that, paradoxically, the rapid exit of empty T134K molecules from the ER was dependent on a TAP-derived supply of peptides. This implies that MHC class I assembly is a two-stage process: initial binding of suboptimal peptides is followed by peptide optimisation that depends on temporary ER retention.