It was recently suggested that the picornavirus 3C proteases are homologous to the chymotrypsin-like serine proteases. The two structural models proposed differ in one of the postulated active site residues, Glu/Asp71 or Asp85. We changed Glu71 of the poliovirus type 1 protease to Asp or Gln and Asp85 to Glu by oligonucleotide-directed site-specific mutagenesis of an infectious cDNA, and attempted to recover virus after transfection. Both Glu71 changes were lethal for the virus and proteolytic activity was abolished in vitro with the exception of the primary cleavage event at the P2/P3 junction. In contrast, the Asp85----Glu virus was viable. This mutant was temperature-sensitive for growth at 39 degrees and exhibited a minute plaque phenotype at permissive temperature. This defect correlated with low levels of viral-specific RNA and protein syntheses and slow virus growth. Proteolytic processing at the COOH-terminus of 3C was impaired, reducing the production of mature 3C and the viral replicase 3D. In addition, 3C-mediated cleavage events within the P2 region of the polyprotein seemed to occur rather inefficiently. 3C-specific processing within P1 and elsewhere within P3 was unaffected. We suggest that Asp85 does not form part of the active site of 3C, but could be important for the specific recognition of cleavage sites within P2.