The biochemical basis for stereochemical control in polyketide biosynthesis.

One of the most striking features of complex polyketides is the presence of numerous methyl- and hydroxyl-bearing stereogenic centers. To investigate the biochemical basis for the control of polyketide stereochemistry and to establish the timing and mechanism of the epimerization at methyl-bearing centers, a series of incubations was carried out using reconstituted components from a variety of modular polyketide synthases. In all cases the stereochemistry of the product was directly correlated with the intrinsic stereospecificity of the ketoreductase domain, independent of the particular chain elongation domains that were used, thereby establishing that methyl group epimerization, when it does occur, takes place after ketosynthase-catalyzed chain elongation. The finding that there were only minor differences in the rates of product formation observed for parallel incubations using an epimerizing ketoreductase domain and the nonepimerizing ketoreductase domain supports the proposal that the epimerization is catalyzed by the ketoreductase domain itself.

Bacterial flavodoxins support nitric oxide production by Bacillus subtilis nitric-oxide synthase.

Unlike animal nitric-oxide synthases (NOSs), the bacterial NOS enzymes have no attached flavoprotein domain to reduce their heme and so must rely on unknown bacterial proteins for electrons. We tested the ability of two Bacillus subtilis flavodoxins (YkuN and YkuP) to support catalysis by purified B. subtilis NOS (bsNOS). When an NADPH-utilizing bacterial flavodoxin reductase (FLDR) was added to reduce YkuP or YkuN, both supported NO synthesis from either L-arginine or N-hydroxyarginine and supported a linear nitrite accumulation over a 30-min reaction period. Rates of nitrite production were directly dependent on the ratio of YkuN or YkuP to bsNOS. However, the V/Km value for YkuN (5.2 x 10(5)) was about 20 times greater than that of YkuP (2.6 x 10(4)), indicating YkuN is more efficient in supporting bsNOS catalysis. YkuN that was either photo-reduced or prereduced by FLDR transferred an electron to the bsNOS ferric heme at rates similar to those measured for heme reduction in the animal NOSs. YkuN supported a similar NO synthesis activity by a different bacterial NOS (Deinococcus radiodurans) but not by any of the three mammalian NOS oxygenase domains nor by an insect NOS oxygenase domain. Our results establish YkuN as a kinetically competent redox partner for bsNOS and suggest that FLDR/flavodoxin proteins could function physiologically to support catalysis by bacterial NOSs.

Expression and characterization of the two flavodoxin proteins of Bacillus subtilis, YkuN and YkuP: biophysical properties and interactions with cytochrome P450 BioI.

The two flavodoxins (YkuN and YkuP) from Bacillus subtilis have been cloned, overexpressed in Escherichia coli and purified. DNA sequencing, mass spectrometry, and flavin-binding properties showed that both YkuN and YkuP were typical short-chain flavodoxins (158 and 151 amino acids, respectively) and that an error in the published B. subtilis genome sequence had resulted in an altered reading frame and misassignment of YkuP as a long-chain flavodoxin. YkuN and YkuP were expressed in their blue (neutral semiquinone) forms and reoxidized to the quinone form during purification. Potentiometry confirmed the strong stabilization of the semiquinone form by both YkuN and YkuP (midpoint reduction potential for oxidized/semiquinone couple = -105 mV/-105 mV) with respect to the hydroquinone (midpoint reduction potential for semiquinone/hydroquinone couple = -382 mV/-377 mV). Apoflavodoxin forms were generated by trichloroacetic acid treatment. Circular dichroism studies indicated that flavin mononucleotide (FMN) binding led to considerable structural rearrangement for YkuP but not for YkuN. Both apoflavodoxins bound FMN but not riboflavin avidly, as expected for short-chain flavodoxins. Structural stability studies with the chaotrope guanidinium chloride revealed that there is moderate destabilization of secondary and tertiary structure on FMN removal from YkuN, but that YkuP apoflavodoxin has similar (or slightly higher) stability compared to the holoprotein. Differential scanning calorimetry reveals further differences in structural stability. YkuP has a lower melting temperature than YkuN, and its endotherm is composed of a single transition, while that for YkuN is biphasic. Optical and fluorimetric titrations with oxidized flavodoxins revealed strong affinity (K(d) values consistently <5 microM) for their potential redox partner P450 BioI, YkuN showing tighter binding. Stopped-flow reduction studies indicated that the maximal electron-transfer rate (k(red)) to fatty acid-bound P450 BioI occurs from YkuN and YkuP at approximately 2.5 s(-1), considerably faster than from E. coli flavodoxin. Steady-state turnover with YkuN or YkuP, fatty acid-bound P450 BioI, and E. coli NADPH-flavodoxin reductase indicated that both flavodoxins supported lipid hydroxylation by P450 BioI with turnover rates of up to approximately 100 min(-1) with lauric acid as substrate. Interprotein electron transfer is a likely rate-limiting step. YkuN and YkuP supported monohydroxylation of lauric acid and myristic acid, but secondary oxygenation of the primary product was observed with both palmitic acid and palmitoleic acid as substrates.

Thermodynamic and biophysical characterization of cytochrome P450 BioI from Bacillus subtilis.

Cytochrome P450 BioI (CYP107H1) from Bacillus subtilis is involved in the early stages of biotin synthesis. Previous studies have indicated that BioI can hydroxylate fatty acids and may also perform an acyl bond cleavage reaction [Green, A. J., Rivers, S. L., Cheesman, M., Reid, G. A., Quaroni, L. G., Macdonald, I. D. G., Chapman, S. K., and Munro, A. W. (2001) J. Biol. Inorg. Chem. 6, 523-533. Stok, J. E., and De Voss, J. J. (2000) Arch. Biochem. Biophys. 384, 351-360]. Here we show novel binding features of P450 BioI--specifically that it binds steroids (including testosterone and progesterone) and polycyclic azole drugs with similar affinity to that for fatty acids (K(d) values in the range 0.1-160 microM). Sigmoidal binding curves for titration of BioI with azole drugs suggests a cooperative process in this case. BioI as isolated from Escherichia coli is in a mixed heme iron spin state. Alteration of the pH of the buffer system affects the heme iron spin-state equilibrium (higher pH increasing the low-spin content). Steroids containing a carbonyl group at the C(3) position induce a shift in heme iron spin-state equilibrium toward the low-spin form, whereas fatty acids produce a shift toward the high-spin form. Electron paramagnetic resonance (EPR) studies confirm the heme iron spin-state perturbation inferred from optical titrations with steroids and fatty acids. Potentiometric studies demonstrate that the heme iron reduction potential becomes progressively more positive as the proportion of high-spin heme iron increases (potential for low-spin BioI = -330 +/- 1 mV; for BioI as purified from E. coli (mixed-spin) = 228 +/- 2 mV; for palmitoleic acid-bound BioI = -199 +/- 2 mV). Extraction of bound substrate-like molecule from purified BioI indicates palmitic acid to be bound. Differential scanning calorimetry studies indicate that the BioI protein structure is stabilized by binding of steroids and bulky azole drugs, a result confirmed by resonance Raman studies and by analysis of disruption of BioI secondary and tertiary structure by the chaotrope guanidinium chloride. Molecular modeling of the BioI structure indicates that a disulfide bond is present between Cys250 and Cys275. Calorimetry shows that structural stability of the protein was altered by addition of the reductant dithiothreitol, suggesting that the disulfide is important to integrity of BioI structure.

Flavocytochrome P450 BM3 mutant A264E undergoes substrate-dependent formation of a novel heme iron ligand set.

A conserved glutamate covalently attaches the heme to the protein backbone of eukaryotic CYP4 P450 enzymes. In the related Bacillus megaterium P450 BM3, the corresponding residue is Ala264. The A264E mutant was generated and characterized by kinetic and spectroscopic methods. A264E has an altered absorption spectrum compared with the wild-type enzyme (Soret maximum at approximately 420.5 nm). Fatty acid substrates produced an inhibitor-like spectral change, with the Soret band shifting to 426 nm. Optical titrations with long-chain fatty acids indicated higher affinity for A264E over the wild-type enzyme. The heme iron midpoint reduction potential in substrate-free A264E is more positive than that in wild-type P450 BM3 and was not changed upon substrate binding. EPR, resonance Raman, and magnetic CD spectroscopies indicated that A264E remains in the low-spin state upon substrate binding, unlike wild-type P450 BM3. EPR spectroscopy showed two major species in substrate-free A264E. The first has normal Cys-aqua iron ligation. The second resembles formate-ligated P450cam. Saturation with fatty acid increased the population of the latter species, suggesting that substrate forces on the glutamate to promote a Cys-Glu ligand set, present in lower amounts in the substrate-free enzyme. A novel charge-transfer transition in the near-infrared magnetic CD spectrum provides a spectroscopic signature characteristic of the new A264E heme iron ligation state. A264E retains oxygenase activity, despite glutamate coordination of the iron, indicating that structural rearrangements occur following heme iron reduction to allow dioxygen binding. Glutamate coordination of the heme iron is confirmed by structural studies of the A264E mutant (Joyce, M. G., Girvan, H. M., Munro, A. W., and Leys, D. (2004) J. Biol. Chem. 279, 23287-23293).