The introduction of the key non-oxidative glycolytic (NOG) pathway enzyme, phosphoketolases (PKTs), into heterologous hosts can improve the yield of a variety of acetyl CoA-derived products of interest. However, the low specific activity of existing PKTs compared with that of 6-phosphofructokinase (PFK), the key EMP pathway enzyme, largely limits their potential applications. To improve PKT activity, previous attempts have focused on increasing intracellular PKT concentration via the use of strong promoters. Herein, we report the establishment of a growth-coupled evolution strategy for the enrichment and selection of PKT mutants with improved specific activity in Corynebacterium glutamicum hosts with defective PFK. Five mutants from 9 Bifidobacterium adolescentis-source PKT (BA-PKT) mutant libraries were obtained. Site-directed mutagenesis analysis revealed 11 mutant sites which contributed to improved BA-PKT specific activity. Further structural analysis revealed that the mutant sites were located far away from the enzyme active site, which makes them almost unpredictable using a rational design approach. Mutant site recombination led to the construction of a novel mutant, PKTT2A/I6T/H260Y, with V-max 29.77 +/- 1.58 U/mg and K-cat/K-m 0.32 +/- 0.01 s(-1)/mM, which corresponds to 73.27 +/- 3.25% and 80.16 +/- 3.38% improvements, respectively, compared with the wildtype (Vmax; 17.17 +/- 0.59 U/mg, K-cat/K-m; 0.17 +/- 0.01 s(-1)/mM). Expression of PKTT2A/I6T/H260 in C. glutamicum Z188 resulted in 16.67 +/- 2.24% and 18.19 +/- 0.53% improvement in l-glutamate titer and yield, respectively, compared with the wildtype BA-PKT. Our findings provide an efficient approach for improving the activity of PKTs. Furthermore, the novel mutants could serve as useful tools in improving the yield of l-glutamate and other acetyl CoA-associated products.