Escherichia coli is an attractive candidate as a host for polyketide production and has been engineered to produce the erythromycin precursor polyketide 6-deoxyerythronolide B (6dEB). In order to identify and optimize parameters that affect polyketide production in engineered E. coli, we first investigated the supply of the extender unit ( 2S)-methylmalonyl-CoA via three independent pathways. Expression of the Streptomyces coelicolor malonyl/methylmalonyl-CoA ligase ( matB) pathway in E. coli together with methylmalonate feeding resulted in the accumulation of intracellular methylmalonyl-CoA to as much as 90% of the acyl-CoA pool. Surprisingly, the methylmalonyl-CoA generated from the matB pathway was not converted into 6dEB. In strains expressing either the S. coelicolor propionyl-CoA carboxylase (PCC) pathway or the Propionibacteria shermanii methylmalonyl-CoA mutase/epimerase pathway, methylmalonyl-CoA accumulated up to 30% of the total acyl-CoA pools, and 6dEB was produced; titers were fivefold higher when strains contained the PCC pathway rather than the mutase pathway. When the PCC and mutase pathways were expressed simultaneously, the PCC pathway predominated, as indicated by greater flux of (13)C-propionate into 6dEB through the PCC pathway. To further optimize the E. coli production strain, we improved 6dEB titers by integrating the PCC and mutase pathways into the E. coli chromosome and by expressing the 6-deoxyerythronolide B synthase (DEBS) genes from a stable plasmid system.