Deep mutagenesis of HLA-A*02:01 reveals elements of folded structure necessary for MHC-I-specific chaperone interactions and plasma membrane trafficking

The loading of high affinity peptides onto nascent class I MHC (MHC-I) molecules is facilitated by chaperones, including the class I-specific chaperone TAP-binding protein-related (TAPBPR). NMR experiments reveal that conformational dynamics of specific MHC-I allotypes, especially around regions known to be in physical proximity to the TAPBPR interface, are correlated with TAPBPR binding affinity. HLA-A*02:01 residues in the alpha1 and alpha2 domains were deep mutationally scanned to determine the effects of nearly all single amino acid substitutions on HLA-A2 surface expression and TAPBPR interactions. Surface trafficking, which demands HLA-A2 be properly folded with bound peptide, imposes strict sequence constraints on the HLA-A2 core, surfaces contacting the beta2-microglobulin and alpha3 domains, and on the A, B and F pockets that ‘grip’ peptide anchor residues. However, TAPBPR binding is permissive to many mutations of HLA-A2, both in the core and on the surface, with the exceptions of two conserved clusters of hydrophobic residues that pack the alpha2-1 and alpha2-2 helices against the beta-sheet. TAPBPR binding is therefore dependent on a local conformation of the alpha2 domain, most likely with a widened peptide binding groove based on published crystal structures. Overall, the data are consistent with a conformational selection model for MHC-I/TAPBPR interactions, in which high MHC-I dynamics increases representation in the structural ensemble of appropriate conformers for TAPBPR recognition. A c-myc tag was fused to the extracellular N-terminus of HLA-A*02:01 for surface detection. Interactions between HLA-A2 and TAPBPR were investigated using bimolecular fluorescence complementation (BiFC), in which the N-terminus (VN) of split fluorescent Venus was fused to the cytosolic tail of HLA-A2, and the C-terminus (VC) of split Venus was fused to the cytosolic tail of TAPBPR. The native transmembrane helix and cytosolic tail of TAPBPR were replaced with a generic transmembrane domain to allow TAPBPR to escape intracellular compartments; this means the experiment reports on TAPBPR/HLA-A2 interactions regardless of where HLA-A2 mutants localize. A library encoding nearly all single amino acid substitutions in the alpha1 and alpha2 domains of HLA-A2 was generated by overlap extension PCR, and the plasmid library was transfected into human Expi293F cells (a suspension culture derivative of HEK 293) under conditions where each cell typically acquires no more than one coding variant. Using FACS, wild-type Expi293F cells expressing endogenous tapasin were collected that displayed high surface HLA-A2, based on fluorescent antibody staining for the c-myc tag. For FACS-based collection of cells showing high BiFC between TAPBPR and HLA-A2, the library was transfected in to cells stably expressing VC-fused TAPBPR. RNA transcripts in the collected cell populations were deep sequenced and compared to the naive plasmid library. The enrichment or depletion of all HLA-A2 variants was calculated and is used as a proxy for relative phenotypic fitness under the selection regime. All deep mutational scans were independently replicated (n = 2).