Many polyketides are synthesized by a class of multifunctional enzymes called type I modular polyketide synthases (PKSs). Several reports have described the power of predictively altering polyketide structure by replacing individual PKS domains with homologues from other PKSs. For example, numerous erythromycin analogues have been generated by replacing individual methylmalonyl-specific acyl transferase (AT) domains of the 6-deoxyerythronolide B synthase (DEBS) with malonyl-, ethylmalonyl-, or methoxymalonyl-specific domains. However, the construction of hybrid PKS modules often attenuates product formation both kinetically and distributively. The molecular basis for this mechanistic imperfection is not understood. We have systematically analyzed the impact of replacing an AT domain of DEBS on acyl-AT formation, acyl-CoA:HS-NAc acyl transferase activity, acyl-CoA:ACP acyl transferase activity (nucleophile charging), acyl-SNAc:ketosynthase acyl transferase activity (electrophile charging), and beta-ketoacyl ACP synthase activity (condensation). As usual, domain junctions were located in interdomain regions flanking the AT domain. Kinetic analysis of hybrid modules containing either malonyl transferase or methylmalonyl transferase domains revealed a 15-20-fold decrease in overall turnover numbers of the hybrid modules as compared to the wild-type module. In contrast, both the activity and the specificity of the heterologous AT domains remained unaffected. Moreover, no defects could be detected in the ability of the heterologous AT domains to catalyze acyl-CoA:ACP acyl transfer. Single turnover studies aimed at directly probing the ketosynthase-catalyzed reaction led to two crucial findings. First, wild-type modules catalyzed chain elongation with comparable efficiency regardless of whether methylmalonyl-ACP or malonyl-ACP were the nucleophilic substrates. Second, chain elongation in all hybrid modules tested was seriously attenuated relative to the wild-type module. Our data suggest that, as currently practiced, the most deleterious impact of AT domain swapping is not on the substrate specificity. Rather, it is due to the impaired ability of the KS and ACP domains in the hybrid module to catalyze chain elongation. Consistent with this proposal, limited proteolysis of wild-type and hybrid modules showed major differences in cleavage patterns, especially in the region between the KR and ACP domains.

The potential of actinomycetes to produce natural products has been exploited for decades. Recent genomic sequence analyses have revealed a previously unrecognized biosynthetic potential and diversity. In order to rationally exploit this potential, we have developed a sequence-guided genetic screening strategy. In this "genome mining" approach, genes that encode tailoring enzymes from natural product biosyntheses pathways serve as indicator genes for the identification of strains that have the genetic potential to produce natural products of interest. We chose halogenases, which are known to be involved in the synthesis of halometabolites as representative examples. From PCR screening of 550 randomly selected actinomycetes strains, we identified 103 novel putative halogenase genes. A phylogenetic analysis of the corresponding putative halogenases, and the determination of their sequential context with mass spectrometric analysis of cultures filtrates revealed a distinct correlation between the sequence and secondary metabolite class of the halometabolite. The described screening strategy allows rapid access to novel natural products with predetermined structural properties.

Considerable progress has been achieved in derivatization of glycopeptide antibiotics by using genetic engineering and in vitro enzymatic approaches. In this issue of Chemistry & Biology, the identification and application of a glycopeptide-specific sulfotransferase by Lamb et al. expands the tool box of biocombinatorial synthesis.