In our previous study, we produced phenylpyruvic acid (PPA) in one step from L-phenylalanine by using an Escherichia coli whole-cell biocatalyst expressing an L-amino acid deaminase (L-AAD) from Proteus mirabilis KCTC2566. However, the PPA titer was low due to the degradation of PPA and low substrate specificity of L-AAD. In this study, metabolic engineering of the L-phenylalanine degradation pathway in E. coli and protein engineering of L-AAD from P. mirabilis were performed to improve the PPA titer. First, three aminotransferase genes were knocked out to block PPA degradation, which increased the PPA titer from 3.3 +/- 0.2 to 3.9 +/- 0.1 g/L and the substrate conversion ratio to 97.5 %. Next, L-AAD was engineered via error-prone polymerase chain reaction, followed by site-saturation mutation to improve its catalytic performance. The triple mutant D165K/F263M/L336M produced the highest PPA titer of 10.0 +/- 0.4 g/L, with a substrate conversion ratio of 100 %, which was 3.0 times that of wildtype L-AAD. Comparative kinetics analysis showed that compared with wild-type L-AAD, the triple mutant had higher substrate-binding affinity and catalytic efficiency. Finally, an optimal fed-batch biotransformation process was developed to achieve a maximal PPA titer of 21 +/- 1.8 g/L within 8 h. This study developed a robust whole-cell E. coli biocatalyst for PPA production by integrating metabolic and protein engineering, strategies that may be useful for the construction of other biotransformation biocatalysts.