Summary of experimental conditions for the seven rounds of directed evolution.

1Random mutagenesis was preformed with either an error-prone polymerase (Mutazyme), or with an ordinary TAQ polymerase that lacks a proof-reading domain. In all cases, 60 cycles of PCR were conducted as part of the protocol for preventing mutations at position 219 (see Text S1). Mutational rates were calculated by implementing the same mutagenesis conditions on wild type pgk and sequencing of 10 randomly picked clones. The application of Mutazyme resulted in average of 2.5 mutations per gene, whereas 60 cycles with the ordinary TAQ polymerase incorporated, on average, about one mutation per gene. In rounds 1 and 7, Mutazyme was used in the first PCR, and TAQ polymerase was used in the second. In rounds 2 to 7, TAQ polymerase was used for both PCR reactions.
2In the second round, an in vitro homologous recombination protocol was applied (DNA shuffling). Overall, 90 cycles of PCR with TAQ polymerase were conducted since the shuffling protocol includes an additional PCR (for a detailed protocol, see supplemental methods).
3The total number of transformants was determined by plating a fraction of the transformation on glycerol-succinate plates (where no selection for PGK activity occurs) with kanamycin (the Δpgk selection marker) and ampicillin (the plasmid selection marker). The number of viable transformants on selection plates relates to colonies that appeared on the glucose-contacting selection plates.