Enzymatic studies on aromatic prenyltransferases

Aromatic prenyltransferases (PTases), including ABBA-type and dimethylallyl tryptophan synthase (DMATS)-type enzymes from bacteria and fungi, play important role for diversification of the natural products and improvement of the biological activities. For a decade, the characterization of enzymes and enzymatic synthesis of prenylated compounds by using ABBA-type and DMATS-type PTases have been demonstrated. Here, I introduce several examples of the studies on chemoenzymatic synthesis of unnatural prenylated compounds and the enzyme engineering of ABBA-type and DMATS-type PTases.

UbiA-type PTases possess a conserved (N/D)DXXD motif for binding of Mg 2+ ion and diphosphate that is also conserved in the isoprenyl diphosphate synthases [9,10]. The enzymes in this group are observed in the ubiquinone and menaquinone biosynthesis [10], membrane lipids biosynthesis in archaea [21], in the biosynthesis of prenylated aromatic secondary metabolites in plants [1], and fungal meroterpenoid biosynthesis [22]. On the other hand, ABBAtype and DMATS-type PTases from microorganisms are soluble proteins and do not contain diphosphate and metal ion binding motif [11][12][13][14][15][16][17][18]20]. The soluble aromatic PTases are involved in the biosynthesis of secondary metabolites in bacteria and fungi.
In the present review, several examples of the recent studies on chemoenzymatic synthesis and the enzyme engineering of soluble ABBA-type and DMATS-type PTases to generate unnatural prenylated aromatic compounds are provided.

Soluble aromatic PTases
ABBA-type and DMATS-type aromatic PTases catalyze prenylation of dimethylallyl diphosphate (DMAPP) and/ or geranyl diphosphate (GPP) to aromatic compounds in bacteria and fungi. The ABBA-type PTases are identified from both of bacteria and fungi, and the CloQ from Streptomyces roseochromogenes var. oscitans is a first characterized ABBA-type of PTases in 2003, which is involved in the biosynthesis of clorobiocin [23]. Different from membranebound UbiA-type PTases, these enzyme reactions, except for NphB, are metal-independent enzymes [14]. The first crystal structure of ABBA-type PTase was solved with NphB in biosynthesis of the naphterpin derivatives [24]. The X-ray crystal structure of NphB showed the characteristic β/α barrel fold with antiparallel strands, which is completely distinct from UbiA-type PTases [25]. This enzyme group was later called as ABBA PTases due to their α-β-β-α PT folds (Fig. 1A) [13].
The EpzP and PpzP from Streptomyces cinnamonensis DSM 104 and Streptomyces annulatus 9663, respectively, catalyze prenylation toward phenazine [101]. The crystal structures of EzpP, the docking model with a substrate 5,10-dihydrophenazine-1-carboxylate (dhPCA), and A B Fig. 3 The substrate specificity of a TleC, MpnD, and b AtaPT toward prenyl donors the mutagenesis analysis provided the intimate structural details of the prenylation reaction mechanism. Based on these information together with the sequential comparison between EpzP and PpzP, the catalytic velocity of EpzP was improved by site-directed mutagenesis. V270F mutation was introduced to form π-stacking between dhPCA and Phe residue. As a result, the enzymatic activity of V270F mutant was increased five-times compared to wild type. Furthermore, the substitution of Ala285 with Gln residue to interact with a water molecule in the active site showed ~ 14-fold higher enzymatic activity than wild type.
The substrate specificity of FgaPT2 was altered by structure-based mutagenesis experiment. Lys174 residue in FgaPT2, proposed to abstract a proton from prenyl-attached arenium intermediate, was substituted with phenylalanine to stabilize the arenium intermediate and increase the interaction with benzene ring of non-genuine substrate L-tyrosine [26,103]. The K174F exhibited 4.9-times higher catalytic efficiency toward L-tyrosine than that of wild type, while the activity toward L-tryptophan was almost abolished. Interestingly, the K174F mutant catalyzes C3-prenylation reaction toward L-tyrosine and its analog 4-amino-L-phenylalanine, and N-prenylation reaction toward 4-amino-L-phenylalanine as a minor reaction [56]. The specificity for the prenylation of L-tyrosine and L-tryptophan was changed from 1:31 (wild type) to 208:1 (K174F mutant). Furthermore, saturation mutagenesis was performed at Arg244, interacting with carboxylate group of substrates [104]. The prenylation activities of 13 Arg244 mutants toward tryptophan-containing cyclic dipeptides were increased up to 76-times compared to wild type. Interestingly, the preferences for tryptophancontaining cyclic dipeptides, including cyclo-L-Trp-L-Leu, cyclo-L-Trp-D-Pro, cyclo-L-Trp-L-Pro, cyclo-L-Trp-Gly, cyclo-L-Trp-L-Trp, and cyclo-L-Trp-L-Phe of these mutants were also changed. For example, the wild-type, R244A, R244T, and R244Q prefer cyclo-L-Trp-D-Pro and cyclo-L-Trp-L-Leu, while R244G utilizes cyclo-L-Trp-L-Leu and cyclo-L-Trp-L-Trp as preferable substrates. The combination of the K174F and R244X mutations succeeded to alter the regiospecificity of prenylation from C4-regular prenylation to C3-reverse prenylation toward tryptophan-containing cyclic dipeptides (Fig. 5a) [105].
The structure-based engineering of DMATS to alter the product specificity was also achieved using FtmPT1 from Aspergillus fumigatus in fumitremorgins biosynthesis. FtmPT1 originally catalyzes C2 prenylation reaction to bevianamide F (cyclo-L-Trp-L-Pro). The structure analysis of FtmPT1 suggested that Tyr205 residue in FtmPT1 interacts with ketone-group of bevianamide F. The saturation-mutagenesis at Tyr205 revealed that the 15 mutants generate regularly C3-prenylated brevianamide F, but not at C2 position [106]. The substrate specificity A B C D Fig. 5 The enzyme reactions of engineered ABBA-type and DMATStype PTases. The enzyme reaction of a the FgaPT2 and its K174F/ K244X mutants, b the FtmPT1 and its Y205X mutants, c the TleC W97Y/A173M and W97Y/F170W/A173M mutants, and d FgaPT2 and its M328X mutants analysis of the two selected mutants Y205N and Y205L revealed that these mutants generated C3-reverse prenylated compounds as predominant products when cyclo-D-Trp-D-Pro, cyclo-D-Trp-L-Pro, and cyclo-L-Trp-D-Pro were used as substrates (Fig. 5b).
On the other hand, the engineering of substrate specificity toward prenyl donors was demonstrated using MpnD, and TleC. MpnD and TleC prefer to utilize C5 DMAPP and C10 GPP, respectively, and catalyze attachment of prenyl donor at the C-7 position of indolactam V in a reverse fashion [85]. The structural analysis of MpnD and TleC complexed with substrates suggested that the three amino acid residues Trp97, Phe170, and Ala173 in TleC and Tyr80, Trp157, and Met159 in MpnD regulate the selectivity of the length of prenyl donor and regiospecificity of prenylation position (Fig. 5c). Based on these observations, TleC A173M, TleC W97Y/A173M, and TleC W97Y/F170W/A173M, MpnD M159A were constructed and analyzed. The preference for prenyl donors of TleC A173M switched from GPP to the smaller DMAPP. On the contrary, M159A substitution in MpnD improved the GPP prenylation activity to generate lyngbyatoxin A, while DMAPP prenylation activity was decreased. Moreover, the TleC W97Y/A173M and TleC W97Y/F170W/ A173M mutants newly produced teleocidin A-2, C-19-epimer of lyngbyatoxin A, and 5-geranylindolactam V in addition to lyngbyatoxin A.
Similar manipulation of prenyl donor substrates was also performed using FgaPT2. The structure-based modeling of FgaPT2 with substrate suggested that the side chain of Met328 protrudes toward the active site and would decrease the size of active site. Thus, Met328 was thought to regulate the substrate specificity of the length of prenyl donor. The substitution with smaller side chain, including M328A, M328C, M328S, and M328G significantly increased the activity for GPP and FPP prenylation (Fig. 5d) [107]. Furthermore, the model also suggested that the active site residues Lue263 and Tyr398 could also interfere with terminal isoprene unit of FPP. The large-to-small substitution of Leu263 and Tyr398 with Ala and Phe, respectively, improved the FPP prenylation activity.

Conclusion
The development of the sequencing technology and the improvement of methodology to characterize the enzymes have accelerated the understanding of the biosynthesis of secondary metabolites. By using these techniques, dozens of ABBA-type and DMATS-type aromatic PTases were functionally and structurally characterized. The accumulation of the knowledge in enzymes provided the chance for the application and engineering of these aromatic PTases. The results presented in this review would be the model cases toward utilization of the secondary metabolite enzymes to generate structurally diversified and biologically active unnatural novel molecular scaffolds for drug discovery.   (2009)