An Unusual Ferryl Intermediate and Its Implications for the Mechanism of Oxacyclization by the Loline-Producing Iron(II)- and 2-Oxoglutarate-Dependent Oxygenase, LolO.

N-Acetylnorloline synthase (LolO) is one of several iron(II)- and 2-oxoglutarate-dependent (Fe/2OG) oxygenases that catalyze sequential reactions of different types in the biosynthesis of valuable natural products. LolO hydroxylates C2 of 1-exo-acetamidopyrrolizidine before coupling the C2-bonded oxygen to C7 to form the tricyclic loline core. Each reaction requires cleavage of a C-H bond by an oxoiron(IV) (ferryl) intermediate; however, different carbons are targeted, and the carbon radicals have different fates. Prior studies indicated that the substrate-cofactor disposition (SCD) controls the site of H· abstraction and can affect the reaction outcome. These indications led us to determine whether a change in SCD from the first to the second LolO reaction might contribute to the observed reactivity switch. Whereas the single ferryl complex in the C2 hydroxylation reaction was previously shown to have typical Mössbauer parameters, one of two ferryl complexes to accumulate during the oxacyclization reaction has the highest isomer shift seen to date for such a complex and abstracts H· from C7 ∼ 20 times faster than does the first ferryl complex in its previously reported off-pathway hydroxylation of C7. The detectable hydroxylation of C7 in competition with cyclization by the second ferryl complex is not enhanced in 2H2O solvent, suggesting that the C2 hydroxyl is deprotonated prior to C7-H cleavage. These observations are consistent with the coordination of the C2 oxygen to the ferryl complex, which may reorient its oxo ligand, the substrate, or both to positions more favorable for C7-H cleavage and oxacyclization.

Synthesis of 6,6- and 7,7-Difluoro-1-acetamidopyrrolizidines and Their Oxidation Catalyzed by the Nonheme Fe Oxygenase LolO.

LolO, a 2-oxoglutarate-dependent nonheme Fe oxygenase, catalyzes both the hydroxylation of 1-exo-acetamidopyrrolizidine (AcAP), a pathway intermediate in the biosynthesis of the loline alkaloids, and the cycloetherification of the resulting alcohol. We have prepared fluorinated AcAP analogues to aid in continued mechanistic investigation of the remarkable LolO-catalyzed cycloetherification step. LolO was able to hydroxylate 6,6-difluoro-AcAP (prepared from N,O-protected 4-oxoproline) and then cycloetherify the resulting alcohol, forming a difluorinated analogue of N-acetylnorloline and providing evidence for a cycloetherification mechanism involving a C(7) radical as opposed to a C(7) carbocation. By contrast, LolO was able to hydroxylate 7,7-difluoro-AcAP (prepared from 3-oxoproline) but failed to cycloetherify it, forming (1R,2R,8S)-7,7-difluoro-2-hydroxy-AcAP as the sole product. The divergent LolO-catalyzed reactions of the difluorinated AcAP analogues provide insight into the LolO cycloetherification mechanism and indicate that the 7,7-difluorinated compound, in particular, may be a useful tool to accumulate and characterize the iron intermediate that initiates the cycloetherification reaction.

Evidence for Modulation of Oxygen Rebound Rate in Control of Outcome by Iron(II)- and 2-Oxoglutarate-Dependent Oxygenases.

Iron(II)- and 2-oxoglutarate-dependent (Fe/2OG) oxygenases generate iron(IV)-oxo (ferryl) intermediates that can abstract hydrogen from aliphatic carbons (R-H). Hydroxylation proceeds by coupling of the resultant substrate radical (R•) and oxygen of the Fe(III)-OH complex ("oxygen rebound"). Nonhydroxylation outcomes result from different fates of the Fe(III)-OH/R• state; for example, halogenation results from R• coupling to a halogen ligand cis to the hydroxide. We previously suggested that halogenases control substrate-cofactor disposition to disfavor oxygen rebound and permit halogen coupling to prevail. Here, we explored the general implication that, when a ferryl intermediate can ambiguously target two substrate carbons for different outcomes, rebound to the site capable of the alternative outcome should be slower than to the adjacent, solely hydroxylated site. We evaluated this prediction for (i) the halogenase SyrB2, which exclusively hydroxylates C5 of norvaline appended to its carrier protein but can either chlorinate or hydroxylate C4 and (ii) two bifunctional enzymes that normally hydroxylate one carbon before coupling that oxygen to a second carbon (producing an oxacycle) but can, upon encountering deuterium at the first site, hydroxylate the second site instead. In all three cases, substrate hydroxylation incorporates a greater fraction of solvent-derived oxygen at the site that can also undergo the alternative outcome than at the other site, most likely reflecting an increased exchange of the initially O2-derived oxygen ligand in the longer-lived Fe(III)-OH/R• states. Suppression of rebound may thus be generally important for nonhydroxylation outcomes by these enzymes.

Methods for the preparation and analysis of the diterpene cyclase fusicoccadiene synthase.

Prenyltransferases are terpene synthases that combine 5-carbon precursor molecules into linear isoprenoids of varying length that serve as substrates for terpene cyclases, enzymes that catalyze fascinating cyclization reactions to form diverse terpene natural products. Terpenes and their derivatives comprise the largest class of natural products and have myriad functions in nature and diverse commercial uses. An emerging class of bifunctional terpene synthases contains both prenyltransferase and cyclase domains connected by a disordered linker in a single polypeptide chain. Fusicoccadiene synthase from Phomopsis amygdali (PaFS) is one of the most well-characterized members of this subclass and serves as a model system for the exploration of structure-function relationships. PaFS has been structurally characterized using a variety of biophysical techniques. The enzyme oligomerizes to form a stable core of six or eight prenyltransferase domains that produce a 20-carbon linear isoprenoid, geranylgeranyl diphosphate (GGPP), which then transits to the cyclase domains for the generation of fusicoccadiene. Cyclase domains are in dynamic equilibrium between randomly splayed-out and prenyltransferase-associated positions; cluster channeling is implicated for GGPP transit from the prenyltransferase core to the cyclase domains. In this chapter, we outline the methods we are developing to interrogate the nature of cluster channeling in PaFS, including enzyme activity and product analysis assays, approaches for engineering the linker segment connecting the prenyltransferase and cyclase domains, and structural analysis by cryo-EM.

Engineering Substrate Channeling in Assembly-Line Terpene Biosynthesis.

Fusicoccadiene synthase from P. amygdala (PaFS) is a bifunctional assembly-line terpene synthase containing a prenyltransferase domain that generates geranylgeranyl diphosphate (GGPP) from dimethylallyl diphosphate (DMAPP) and three equivalents of isopentenyl diphosphate (IPP), and a cyclase domain that converts GGPP into fusicoccadiene, a precursor of the diterpene glycoside Fusicoccin A. The two catalytic domains are linked by a flexible 69-residue polypeptide segment. The prenyltransferase domain mediates oligomerization to form predominantly octamers, and cyclase domains are randomly splayed out around the prenyltransferase core. Previous studies suggest that substrate channeling is operative in catalysis, since most of the GGPP formed by the prenyltransferase remains on the protein for the cyclization reaction. Here, we demonstrate that the flexible linker is not required for substrate channeling, nor must the prenyltransferase and cyclase domains be covalently linked to sustain substrate channeling. Moreover, substrate competition experiments with other diterpene cyclases indicate that the PaFS prenyltransferase and cyclase domains are preferential partners regardless of whether they are covalently linked or not. The cryo-EM structure of engineered "linkerless" construct PaFSLL, in which the 69-residue linker is spliced out and replaced with the tripeptide PTQ, reveals that cyclase pairs associate with all four sides of the prenyltransferase octamer. Taken together, these results suggest that optimal substrate channeling is achieved when a cyclase domain associates with the side of the prenyltransferase octamer, regardless of whether the two domains are covalently linked and regardless of whether this interaction is transient or locked in place.