Peptide binding to MHC class I-a molecules typically requires anchoring at free peptide termini. Since the class I-b molecule H-2M3 preferentially binds AMbrmylated peptides, the anchoring residues at the N-terminus of its binding groove may be very different than other class I molecules. X-ray analysis by Wang et. al., revealed several non-conserved residues interacting with the formyl moiety. These residues occlude the N-terminal pocket of the groove, suggesting a "frame-shift" model in which the peptide /V-terminus is shifted into the space normally occupied by position 2. To test if nonconserved residues are sufficient to generate JV-fonnyl specificity, we have transplanted the W-formyl binding pocket of M3 into the class I-a molecule, H-2Kb. We used site-directed mutagenesis to change six residues of K to those in M3 that occlude the yV-terminal binding pocket and interact with the formyl moiety. Our data show that the chimera prefers N-formylated peptides while Kb retains its preference for a free iV-terminus. To assess the "frameshift" model, the chimera was tested with Kb peptides of varying length. H2Kb has strong side chain anchors for residues 5 and 8 of the peptide. The occlusion of the W-terminus of the binding groove by the addition of the Nformyl binding pocket reduces the length of the binding groove while still retaining the K4 side chain anchors. Our data show that removal of one amino acid from the JV-terminus of Kb peptides to permit repositioning in the reduced binding groove, increases their ability to bind the chimera. These data support the "frame-shift" model.