Immunodominant epitopes defined by a yeast-expressed library of random fragments of the rabies virus glycoprotein map outside major antigenic sites.

Nineteen yeast colonies secreting rabies virus glycoprotein (G) peptides immunoreactive with polyclonal anti-rabies virus sera were selected from a random expression library. The peptides, around 80 amino acids long, spanned amino acids 54-494 of the G protein. These peptides, together with two constructions including, respectively, immunodominant sites II and III, were analysed for their immunoreactivity with 40 anti-G protein monoclonal antibodies (MAbs) composed of 12 MAbs that reacted with SDS-treated protein in Western blot under reducing conditions (WB+) and 28 representative MAbs that did not react after denaturation (WB-). This last category represents 98% of anti-rabies virus G MAbs. None of the WB- MAbs bound peptides. Of the 12 WB+ MAbs, one bound two peptides situated before the transmembrane domain of the protein and six bound peptides overlapping a region situated between amino acids 223 and 276. These six MAbs define a new antigenic region that would be considered 'immunodominant' if the peptide strategy had been used to study the antigenicity of the protein; however, this region is only recognized by about 1% of our MAbs. Three of these WB+ MAbs had significant neutralizing activity; two were used for the selection of antigenic mutants (MAR mutants). Some mutants had a substitution within the region delimited by the peptides, confirming the pertinence of both the peptide and escape mutant approaches. However, a few mutants had a substitution outside the peptide-delimited region, suggesting that remote mutation(s) could affect epitope accessibility in the native protein.