Structural Characterization of OxyD, a Cytochrome P450 Involved in β-Hydroxytyrosine Formation in Vancomycin Biosynthesis

Abstract

The cytochrome P450 OxyD from the balhimycin glycopeptide antibiotic biosynthetic operon of Amycolatopsis mediterranei is involved in the biosynthesis of the modified amino acid β-R-hydroxytyrosine, an essential precursor for biosynthesis of the vancomycin-type aglycone. OxyD binds the substrate tyrosine not free in solution, but rather covalently linked to the carrier protein (CP) domain of the non-ribosomal peptide synthase BpsD, exhibiting micromolar binding affinity to a tyrosine-loaded carrier protein construct. The crystal structure of OxyD was determined to 2.1-Å resolution, revealing a potential binding site for the carrier protein-bound substrate in a different orientation to that seen with the acyl carrier protein-bound P450_BioI (Cryle, M. J., and Schlichting, I. (2008) Proc. Natl. Acad. Sci. U.S.A. 105, 15696–15701). A series of residues were identified across known aminoacyl-CP-oxidizing P450s that are highly conserved and cluster in the active site or potential CP binding site of OxyD. These residues appear to be characteristic for aminoacyl-CP-oxidizing P450s, allowing sequence based identification of P450 function for this subgroup of P450s that play vital roles in the biosyntheses of many important natural products in addition to the vancomycin-type antibiotics. The ability to analyze such P450 function based upon sequence data alone should prove an important tool in the analysis and identification of new medicinally relevant biomolecules.

Introduction

Hydroxylated amino acid residues are widely utilized in nature as precursors for complex natural products (see Fig. 1). The oxidative modification of amino acids at the β-position is a process that occurs in many different classes of medicinally important biomolecules, including the vancomycin-type antibiotics (reviewed in Refs. 2, 3). These glycopeptide antibiotics have had great clinical success in treating Gram-positive bacterial infections that are resistant to other classes of antibiotics and function through blocking cell wall biosynthesis via complex formation with the cell wall precursor peptidoglycan peptidyl units terminating in -Lys-D-Ala-D-Ala (2, 3). The biosynthesis of the vancomycin peptide requires the incorporation of two amino acid residues derived from β-R-hydroxytyrosine. Another group of medicinally important molecules that contain hydroxylated amino acid residues is the aminocoumarin-containing angucycline family of antibiotic compounds, which include coumermycin A, clorobiocin, and simocyclinone and act upon bacterial DNA gyrase. β-R-Hydroxytyrosine is an essential precursor in the biosynthesis of these types of antibiotics, where it is needed for the formation of both the A (Fig. 1, blue) and B (Fig. 1, red) rings in the aminocoumarin core. Further examples of the incorporation of hydroxylated amino acids into the biosyntheses of medicinally important compounds can be found in the biosyntheses of the nikkomycin-type antibiotics, zorbamycin and echinomycin (Fig. 1) (4–7).

FIGURE 1.

Structures of medicinally important natural products derived from β-hydroxyamino acid residues. Groups derived from the β-hydroxyamino acids are shown in red; novobiocin and clorobiocin also possess an additional moiety derived from a β-hydroxytyrosine, shown in blue.

Despite the large structural difference in the compounds highlighted above and the difference in the biosyntheses of the compounds overall, the formation of the β-hydroxyamino acid precursors follows a conserved route. This involves the oxidation of the amino acid by a member of the cytochrome P450 (P450)² superfamily. P450s form a superfamily of heme containing monoxygenases that catalyze a vast array of biologically important transformations (8). The archetypical P450 reaction is the hydroxylation of unactivated C–H bonds, a difficult oxidation made more impressive by the ability of P450s to utilize their oxidative power both regio- and stereoselectively. Many P450s play important roles in natural product biosyntheses, where such selectivity is channeled toward the synthesis of complex secondary metabolites. P450s can catalyze a variety of different reactions, including epoxidation of double bonds, heteroatom oxidation, phenolic coupling, and multistep oxidative transformations (9).

The formation of β-hydroxy amino acid precursors by P450s differs from the majority of P450-catalyzed oxidation reactions in that the substrate is presented to the P450 in a carrier protein-bound form, rather than free in solution. P450s that utilize carrier protein-bound substrates form a small but significant subgroup of the superfamily. One example is P450_BioI, which is responsible for the carbon–carbon bond cleavage of acyl carrier protein-bound fatty acids during the first step of biotin biosynthesis in Bacillus subtilis (1). Additional examples are found in the biosynthesis of the vancomycin-type antibiotics: balhimycin (10), chloroeremomycin (11), and vancomycin, produced by different strains of Amycolatopsis (Fig. 1) (12, 13). In the biosyntheses of these compounds, the P450s OxyA, OxyB, and OxyC have been shown in vivo to be responsible for the oxidative phenolic coupling of the aromatic side chains found in the vancomycin aglycone (14–18).

The role of the fourth P450 in the vancomycin gene cluster, OxyD, has been investigated in vivo, where it has been shown to be involved in the biosynthesis of l-β-R-hydroxytyrosine (l-3-R-hydroxytyrosine), an essential precursor in the biosynthesis of the vancomycin aglycone (19, 20). OxyD has been shown to be transcribed along with two other genes, the first a non-ribosomal peptide synthase (BpsD), composed of single adenylation (A) and carrier protein (PCP) domains (21), whereas the second (Bhp) was originally annotated as a perhydrolase (20) but was recently shown to be a thioesterase (22). The likely roles for the three proteins can be inferred from the biosynthetic operons of other natural products that catalyze the formation of oxidized amino acids (23, 24): firstly tyrosine is selectively loaded onto the PCP domain of BpsD, which serves as a substrate for OxyD hydroxylation of tyrosine, followed in turn by thioester cleavage of β-R-hydroxytyrosine from BpsD by Bhp (Scheme 1).

SCHEME 1.

Formation of l-3-(R)-hydroxytyrosine in A. mediterranei that is required for balhimycin biosynthesis. A, adenylation domain; PCP, peptidyl carrier protein domain; and TE, thioesterase domain.

Biochemical characterization of the P450s that catalyze the oxidation of carrier protein-bound amino acids in novobiocin and nikkomycin biosynthesis has been performed (23, 25). Walsh and co-workers demonstrated that the oxidation of tyrosine to β-R-hydroxytyrosine is catalyzed by the P450 NovI, with tyrosine bound to another protein in the novobiocin biosynthetic operon, NovH. NovH resembles a non-ribosomal peptide synthase (NRPS), possessing an adenylation and a carrier protein domain and which displays a high degree of selectivity for l-tyrosine activation (24). Other aminocoumarin-containing antibiotics have had similar NRPS and P450 genes identified within the biosynthetic operons encoding their production, including those of clorobiocin (P450: cloI, NRPS: cloH) (26), simocyclinone (P450: simD1, NRPS: simD6) (27), and coumermycin A1 (P450: cumC, NRPS: cumD) (28). In the case of nikkomycin biosynthesis, the activation of a histidine residue by the NRPS NikP1 allows the oxidation of the bound histidine residue to β-R-hydroxyhistidine by the cytochrome P450 NikQ. Removal of β-R-hydroxyhistidine from NikP1 is then accomplished by NikP2, a thioesterase domain-like protein (23).

The apparent similarities of the mechanisms for β-hydroxyamino acid formation in these disparate classes of medicinally important compounds led to our interest in structurally characterizing the OxyD protein from Amycolatopsis mediterranei and to investigate the interaction between the P450 and its carrier protein-bound substrate. Here we describe the expression and purification of the OxyD protein from A. mediterranei, its interaction with tyrosine presented in a peptidyl carrier protein domain-bound form and the crystal structure of OxyD to 2.1-Å resolution, the first structure of a P450 involved in the oxidation of an amino acid charged carrier protein to be determined. We have identified a common motif present in all known P450s that catalyze carrier protein-bound β-hydroxyamino acid formation and visualized this binding interface upon the OxyD structure. This now allows the sequence-based identification of P450s involved in carrier protein-bound amino acid hydroxylation and also provides a platform for future rational redesign of a wide range of medicinally important biomolecules.

EXPERIMENTAL PROCEDURES

Cloning of oxyD and bpsD_pcp Genes

Synthetic genes encoding oxyD and the C-terminal region of the non-ribosomal peptide synthase BpsD (amino acids 496–581) from A. mediterranei, respectively, were obtained from Geneart (Regensburg, Germany) with concomitant incorporation of unique 5′-NdeI and 3′-HindIII restriction sites. The genes were codon optimized for Escherichia coli expression and supplied in pBluescript II plasmids. The genes were subjected to a restriction digest using NdeI and HindIII, purified by agarose gel electrophoresis, and cloned into the corresponding sites in the E. coli expression plasmid pET28a(+) (Novagen, Darmstadt, Germany) such that the expression of the genes was placed under the control of a T7 promoter. In these constructs the expressed protein includes an N-terminal hexahistidine tag and thrombin cleavage site. The resulting clones, pOxyD and pBpsD_PCP, were sequenced through their reading frames. Plasmids were then used to transform chemically competent BL21(DE3)pLysS E. coli cells.

A truncated construct (bpsD_spcp) encoding for residues 502–581 of the non-ribosomal peptide synthase BpsD from A. mediterranei was amplified from plasmid pBpsD_PCP using oligonucleotides sPCP_D1 GGTCGTCATATGA GCGGTCGTGCACCG and sPCP_D2 GCGG CCGCAAGCTTTTATTAGGTCAGTGC to introduce an 5′-NdeI restriction site and maintain the 3′-HindIII restriction site (underlined). The clone pBpsD_sPCP was then prepared as indicated above and used to transform chemically competent BL21(DE3) E. coli cells.

A synthetic gene encoding a hybrid construct of the C-terminal region of the non-ribosomal peptide synthase BpsD (amino acids 502–581) from A. mediterranei along with the first 14 residues replaced with MEKAPENETEKVLS (the first α-helix from the soluble published PCP_7S construct from Amycolatopsis orientalis (29), conserved residues underlined) was obtained from Geneart with concomitant incorporation of unique 5′-NdeI and 3′-HindIII restriction sites and an N-terminal start codon. The gene was codon optimized for E. coli expression and supplied in the plasmid pMA. The clone pBpsD_hPCP was then prepared as indicated above and used to transform chemically competent BL21(DE3) E. coli cells.

Cloning of the bpsD_pcp Domain as N-terminal Fusion Proteins

The hybrid construct bpsD_hpcp was amplified from plasmid pBpsD_hPCP using oligonucleotides sPCP_D3 AATTACCATGGGCGGTCGTGCACCGAGC and sPCP_D4 GGATCTCAGTGGTGGTGG to introduce a 5′-NcoI restriction site and retain the 3′-NotI restriction site present in the construct pBpsD_hPCP for expression screening with a fusion-protein vector series (30). The bpsD_hpcp gene thus amplified was subjected to a restriction digest using NcoI and NotI, purified by agarose gel electrophoresis, and cloned into the corresponding sites in the E. coli N-terminal fusion protein expression plasmid series reported by Bogomolovas et al. (30); each construct bears an internal hexahistidine tag and a tobacco etch virus protease cleavage site between the fusion protein and the cloned protein, with protein expression under the control of the T7 promoter. Single clones from each fusion protein construct were sequenced through the bpsD_hpcp reading frame to ensure that its sequence was without mutations before these clones were used to transform chemically competent BL21(DE3) E. coli cells. The plasmid identified as the best after test expression analysis (see below) was derived from plasmid pET-Trx_1b, encoding a thioredoxin fusion protein and is known as pTrx-BpsD_hPCP_1b. To alter the position of the hexahistidine tag from in between the fusion domains (plasmid 1b) to at the C terminus of the BpsD PCP domain (plasmid 1c), the hybrid construct bpsD_hpcp was amplified from plasmid pBpsD_hPCP using oligonucleotides sPCP_D5 TAGATCGCCCATGGAGAAAGCGCCGGAAAACG and sPCP_D6 TGAATCTCGAGGGTCAGTGCTGCATCCAGAAC to introduce a 5′-NcoI restriction site and a 3′-XhoI restriction site (underlined) without a stop codon. The bpsD_hpcp gene thus amplified was subjected to a restriction digest using NcoI and XhoI, purified by agarose gel electrophoresis, and cloned into the corresponding sites in the E. coli fusion protein expression plasmid pET-Trx_1c, with this construct bearing a C-terminal hexahistidine tag following the cloned protein sequence Trx-BpsD_hPCP and protein expression under the control of the T7 promoter (30). The resultant clone, pTrx-BpsD_hPCP_1c, was sequenced through the bpsD_hpcp reading frame to ensure that its sequence was without mutations. Plasmid pTrx-BpsD_hPCP_1c was then used to transform chemically competent BL21(DE3) E. coli cells.

Protein Expression and Purification

General Procedures

All purification steps were performed at 4 °C, with buffers filtered through a 0.2-μm filter and degassed prior to use. The determination of protein concentration was performed using the method of Bradford using bovine serum albumin as a reference protein (31).

OxyD Expression and Purification

pOxyD-transformed BL21(DE3)pLysS cells were grown overnight at 37 °C in TB plus kanamycin (25 mg/liter) to provide a starter culture for expression. 5 × 2 liter of TB plus kanamycin (25 mg/liter) was then inoculated with 1% (v/v) of overnight culture and grown at 37 °C to an absorbance (600 nm) of 0.5, whereupon the temperature was reduced to 18 °C and 0.5 mm δ-aminolevulinic acid was added. Expression of OxyD was induced 20 min later using 0.1 mm IPTG. After 24 h at 18 °C, a further 0.5 mm of δ-aminolevulinic acid was added to the cultures, which were then grown for a further 24 h. The cell pellet was collected by centrifugation at 4 °C (5,000 × g) and resuspended in lysis buffer (50 mm Tris·HCl (pH 7.4), 50 mm NaCl, 1 mm phenylmethylsulfonyl fluoride, 0.5 mm dithioerythritol). The cells were lysed using two passes through a fluidizer (Microfluidics, Newton, MA). The lysis solution was clarified by centrifugation (120,000 × g) and loaded onto a 5-ml Ni-NTA Superflow column (Qiagen) that had been pre-equilibrated in nickel wash buffer (50 mm Tris·HCl (pH 8.0), 300 mm NaCl, and 20 mm imidazole). The column was washed with 5 column volumes (CV) of nickel wash buffer before the bound protein was eluted with 1.5 CV of nickel elution buffer (50 mm Tris·HCl (pH 8.0), 300 mm NaCl, and 250 mm imidazole). The eluted fractions were concentrated, and buffer was exchanged into QA-buffer (50 mm Tris·HCl (pH 8.0), 50 mm NaCl, and 0.5 mm EDTA) using NAP-25 columns (GE Healthcare). The protein was purified by anion-exchange chromatography using a Mono Q column (GE Healthcare) attached to an Äkta fast-protein liquid chromatography system (GE Healthcare). Following loading, the column was washed with 8 CV of QA buffer before the bound protein was eluted with a 15-CV gradient with a final concentration of 40% 50 mm Tris·HCl (pH 8.0), 1 m NaCl, and 0.5 mm EDTA. The eluted fractions were analyzed by SDS-PAGE, and appropriate fractions were pooled, concentrated, and subjected to gel filtration on a Superose-12 column (GE Healthcare) attached to an Äkta fast-protein liquid chromatography system using exchange buffer (50 mm Tris·HCl (pH 8.0), 100 mm NaCl). The eluted fractions were analyzed by SDS-PAGE, appropriate fractions were pooled and concentrated using an Amicon Ultra centrifugal filter with a 30,000 molecular weight cut-off (Millipore, Bedford, MA), and aliquots were flash frozen in liquid nitrogen and stored at −80 °C. Final protein yield of OxyD from the 10 liters of culture was 20 mg. The identity of the purified protein as OxyD was confirmed by MALDI-TOF MS peptide map fingerprinting of a tryptic digest of the excised SDS-PAGE protein band.

Attempted Expression of BpsD_PCP Constructs

Appropriately transformed BL21(DE3) cells were grown overnight at 37 °C in TB plus kanamycin (25 mg/liter) to provide a starter culture for expression. 50-ml test cultures of TB plus kanamycin (25 mg/liter) were inoculated with 1% (v/v) of overnight culture and grown at 37 °C to an absorbance (600 nm) of 0.5, whereupon the temperature was reduced to 18 °C. Expression of the PCP_D construct was induced 10 min later using 0.1 mm IPTG. After 24 h the cell pellet was collected by centrifugation at 4 °C (5,000 × g), the pellet resuspended in lysis buffer, and the cells were lysed using sonication. The lysis solution was clarified by centrifugation (120,000 × g), with the whole cell and soluble and insoluble protein fractions analyzed by SDS-PAGE. No construct tested afforded any overexpression of BpsD_PCP protein.

Test Expression of pET-Fusion BpsD_PCP Constructs

Appropriately transformed BL21(DE3) cells were grown overnight at 37 °C in TB plus kanamycin (25 mg/liter) to provide a starter culture for expression of each fusion protein. 50-ml test cultures of TB plus kanamycin (25 mg/liter) were inoculated with 1% (v/v) of overnight culture and grown at 37 °C to an absorbance (600 nm) of 0.5, whereupon the temperature was reduced to 18 °C. Expression of the PCP_D construct was induced 10 min later using 0.1 mm IPTG. After 12 h the cell pellet was collected by centrifugation at 4 °C (5,000 × g), the pellet was resuspended in lysis buffer, and the cells were lysed using sonication. The lysis solution was clarified by centrifugation (120,000 × g), with the whole cell and soluble and insoluble protein fractions analyzed by SDS-PAGE. All N-terminal fusion protein constructs (see Ref. 30) afforded overexpression of BpsD_PCP fusion proteins, although not all resulted in the production of soluble protein. The identity of all fusion proteins were confirmed by MALDI-TOF MS peptide map fingerprinting of tryptic digests of the excised SDS-PAGE protein bands corresponding to each individual fusion protein.

Expression of and Purification of pTrx-BpsD_hPCP_1b Construct

Appropriately transformed BL21(DE3) cells were grown overnight at 37 °C in TB plus kanamycin (25 mg/liter) to provide a starter culture for expression. 2 × 1 liter cultures of TB plus kanamycin (25 mg/liter) were inoculated with 1% (v/v) of overnight culture and grown at 37 °C to an absorbance (600 nm) of 0.4, whereupon the temperature was reduced to 18 °C. Expression of the Trx-PCP_D_1b construct was induced 15 min later using 0.1 mm IPTG. After 12 h the cell pellet was collected by centrifugation at 4 °C (5,000 × g), the pellet was resuspended in lysis buffer, and the cells were lysed using sonication. The lysis solution was clarified by centrifugation (120,000 × g) and loaded onto a 1-ml Ni-NTA Superflow column that had been pre-equilibrated in nickel wash buffer. The column was washed with 5 CV of nickel wash buffer before the bound protein was eluted with 1.5 CV of nickel elution buffer. SDS-PAGE analysis of the eluted protein revealed large amounts of thioredoxin protein without the BpsD_PCP fusion. The identity of both the full-length and truncated proteins was confirmed by MALDI-TOF MS peptide map fingerprinting of tryptic digests of the excised SDS-PAGE protein bands. Due to proteolytic cleavage, the decision was made to switch to a C-terminal hexahistidine tag fusion protein to avoid co-purification of truncated fusion proteins along with the desired full length construct.

Expression of and Purification of pTrx-BpsD_hPCP_1c Construct

Appropriately transformed BL21(DE3) cells were grown overnight at 37 °C in TB plus kanamycin (25 mg/liter) to provide a starter culture for expression. 10 × 100 ml cultures of TB plus kanamycin (25 mg/liter) were inoculated with 1% (v/v) of overnight culture and grown at 37 °C to an absorbance (600 nm) of 0.4, whereupon the temperature was reduced to 18 °C. Expression of the Trx-PCP_D_1c construct was induced 15 min later using 0.1 mm IPTG. After 12 h the cell pellet was collected by centrifugation at 4 °C (5,000 × g), the pellet was resuspended in lysis buffer, and the cells were lysed using sonication. The lysis solution was clarified by centrifugation (120,000 × g) and loaded onto a 1-ml Ni-NTA Superflow column that had been pre-equilibrated in nickel wash buffer. The column was washed with 5 CV of nickel wash buffer before the bound protein was eluted with 1.5 CV of nickel elution buffer. SDS-PAGE analysis of the eluted protein revealed >95% pure full-length protein, which was buffer exchanged against exchange buffer, concentrated using an Amicon Ultra centrifugal filter with a 5,000 molecular weight cut-off divided into aliquots, and flash frozen in liquid nitrogen before being stored at −80 °C. The identity of the purified protein as Trx-BpsD_PCP was confirmed by MALDI-TOF MS peptide map fingerprinting of a tryptic digest of the excised SDS-PAGE protein band.

Synthesis of Amino Acid Derivatives and PCP Modification

Synthesis of SNAc and CoA Tyrosine and Phenylalanine Derivatives

l-Tyrosine-CoA and l-phenylalanine-CoA were synthesized from the appropriate t-butoxycarbonyl-protected amino acids (Bachem, Bubendorf, Switzerland) and CoA (USB, Staufen, Germany) using the peptide coupling conditions reported by Strieker et al. (32) and employing a modified purification protocol utilizing a Source 15RPC column (1.7 ml, GE Healthcare) attached to an Äkta fast-protein liquid chromatography system. Purification of the aminoacyl-CoA compounds utilized an initial washing step of 7 CV with 0.1% (v/v) aqueous trifluoroacetic acid, followed by a 15-CV gradient elution rising to 40% (v/v) 0.1% (v/v) trifluoroacetic acid in acetonitrile (elution volume: l-tyrosine-CoA, 18.6 ml; l-phenylalanine-CoA, 21.4 ml). Correct fractions were identified by MALDI-TOF MS, pooled, and lyophilized before being stored at −80 °C prior to use. l-Tyrosine-SNAc, d-tyrosine-SNAc, and l-phenylalanine-SNAc were prepared from the appropriate t-butoxycarbonyl-protected amino acids and SNAc using the coupling conditions reported by Ehmann et al. (33) with a final purification utilizing a Source 15RPC column with an initial washing step of 7 CV with 0.1% (v/v) aqueous trifluoroacetic acid, followed by a 15-CV gradient elution rising to 20% (v/v) 0.1% (v/v) trifluoroacetic acid in acetonitrile (elution volume: l-tyrosine-SNAc, 17.9 ml; d-tyrosine-SNAc, 17.9 ml; and l-phenylalanine-SNAc, 21.8 ml). Correct fractions were identified by MALDI-TOF MS, pooled, and lyophilized before being stored at −20 °C prior to use.

Trx-BpsD_PCP Modification with Sfp

Aminoacyl and holo-Trx-BpsD_PCP fusion proteins and fluoresceinyl Trx-BpsD_PCP fusion proteins were prepared using the promiscuous phosphopantetheinyl B. subtilis transferase Sfp (34) in the following procedures: Apo Trx-BpsD_PCP (30 μm) was converted into holo-Trx-BpsD_PCP by incubation with Sfp (0.2 μm) for 30 min at 37 °C in reaction buffer containing 50 mm Tris·HCl (pH 7.4, room temperature), 10 mm MgCl₂, 90 μm CoA, and 2 mm dithioerythritol. Protein modification was confirmed by MALDI-TOF MS. In a further test of the ability of the Trx-BpsD_PCP construct to be post-translationally modified by the action of CoA derivatives and Sfp, apo Trx-BpsD_PCP (50 μm) was converted to fluoresceinyl Trx-BpsD_PCP by incubation with Sfp (0.2 μm) for 30 min at 37 °C in reaction buffer containing 50 mm Tris·HCl (pH 7.4), 10 mm MgCl₂, 250 μm fluoresceinyl-CoA (prepared by incubation of 1 mm CoA and 1 mm iodofluorescein in 50 mm NaP_i buffer, pH 8.5, for 2 h at 24 °C) and 2 mm dithioerythritol. Protein modification was confirmed by MALDI-TOF MS. Apo Trx-BpsD_PCP (30 μm) was converted to tyrosyl and phenylalanyl Trx-BpsD_PCP by incubation with Sfp (0.2 μm) for 30 min at 37 °C in reaction buffer containing 50 mm Tris·HCl (pH 7.4, room temperature), 10 mm MgCl₂, 90 μm aminoacyl-CoA, and 2 mm dithioerythritol. Following the completion of the reactions the aminoacylated Trx-BpsD_PCP proteins were buffer-exchanged against 20 mm Tris·HCl (pH 7.4) and 50 mm NaCl using NAP-5 desalting columns before being concentrated using an Amicon Ultra centrifugal filter with a 5000 molecular weight cut-off to a final concentration of 150–180 μm. Product formation was confirmed by MALDI-TOF MS.

Binding of holo-Trx-BpsD_PCP, Aminoacyl Trx-BpsD_PCPs, and Amino Acid SNAc Thioesters to OxyD

The binding of substrates to OxyD was performed using UV-visible spectroscopy (V-630 spectrometer, JASCO, Gross-Umstadt, Germany) by monitoring the change in spin state of the heme iron due to water ligand displacement caused by substrate binding and thus a change from hexacoordinated iron (III) with an absorption maximum at 419 nm to pentacoordinated iron (III) with an absorption maximum at 392 nm. OxyD (1.6 μm) in a solution with a final buffer composition of 20 mm Tris·HCl (pH 7.4) and 50 mm NaCl was titrated with aliquots of substrate in the same buffer (150 μm holo-Trx-BpsD_PCP; 150 μm phenylalanyl Trx-BpsD_PCP; 180 μm tyrosyl Trx-BpsD_PCP; 800 μm amino acid SNAc thioesters) and the change in spin state of the heme iron monitored using the difference in absorbance between 419 and 392 nm. Reactions were performed in triplicate for SNAc thioesters and in duplicate for Trx-BpsD_PCP substrates. The binding of substrates was then fitted to the appropriate equation using the program GraFit 5 (Trx-BpsD_PCP substrates fitted using a one-site binding equation, [Bound ligand] = (capacity·[free ligand])/(K_d + [free ligand]); SNAc substrates fitted using a cooperative binding equation, θ = ([Ligand]ⁿ·capacity)/((K_a)ⁿ + [ligand]ⁿ), where θ is the fraction of ligand binding sites filled, n is the Hill coefficient, and K_a is the ligand concentration producing half occupancy). The percentage spin state change was calculated using the extinction coefficient values reported for P450_cam (35, 36) and P450_terp (37).

Protein Crystallization, Data Collection, and Structure Determination

Crystals were grown by hanging drop, vapor diffusion at 20 °C. The OxyD protein (8 mg/ml) was mixed (1:1) with the reservoir solution (0.1 m Bis-Tris (pH 6.8), 18% (v/v) polyethylene glycol 2000 monomethyl ether) and equilibrated against the reservoir solution. After 60 days red diamond-shaped crystals (∼100 μm length) had formed. The crystals were then passed through four cryoprotectant solutions with increasing concentrations of glycerol until the final cryoprotectant solution (0.1 m Bis-Tris (pH 6.8), 18% (v/v) polyethylene glycol 2000 monomethyl ether, and 25% (v/v) glycerol) was obtained. The crystals were then flash cooled in liquid nitrogen for data collection. Two native OxyD data sets were collected at the X10SA beamline at the Swiss Light Source at the Paul Scherrer Institute (Villigen, Switzerland) with the crystals kept at 100 K during data collection. The data were processed using the XDS program suite and scaled using XSCALE (38) (see Table 2). The crystal symmetry belongs to space group P2(1) with two OxyD molecules per asymmetric unit. The OxyD structure was solved by molecular replacement using the program PHASER (39) and a search model consisting of P450_BioI CYP107H1 (Protein Data Bank code 3EJB, Chain B), including residues 21–51, 80–166, and 228–395. Iterative manual model building and refinement were performed using COOT (40) and REFMAC (41) with TLS refinement (42) following simulated annealing procedure performed in CNS (43). During several cyclic rounds of refinement with REFMAC and manual rebuilding, heme, glycerol, and solvent molecules were included in the model. TLS input files were generated using the TLS-Motion Determination Server (44, 45). Structure validation was performed using MOLPROBITY (46) and PROCHECK (47). Structure-based sequence alignments were carried out with SSM (48) as implemented in COOT, with secondary structure analysis of the final structure performed using the DSSPcont server (49) and comparisons to known structures performed using DaliLite (50). Electrostatic properties were calculated using the program APBS (51) as implemented in PyMOL (52). All structural figures were prepared using PyMOL (52).

TABLE 2.

Crystallographic data for OxyD

	OxyD native
Data collection
Space group	P21
Cell dimensions a, b, c (Å), β (°)	65.9, 61.1, 100.6, and 102.5
Molecules/asymmetric unit	2
X-ray source	SLS X10SA
Wavelength (Å)	0.99986
Resolution (Å)a	40.0-2.1
Unique reflections	44,661
Rsyma	0.06 (0.30)
I/σ(I)a	19.2 (4.5)
Completeness (%)a	97.5 (92.4)
Redundancy	7.9
Refinement
Wilson B-factor (Å2)	32.8
Resolution in refinement	20.0-2.1
Rwork/Rfreeb	0.208/0.233
TLS groups	A6–A135, A136–A190, A191–A224, A225–A358, A359–A396, B6–B44, B45–B138, B139–B227, B228–B357, B358–B396
No. of atoms
Protein	5985 (A6–A176, A181– A396; B6–B396)
Heme	86
Glycerol	60
Water	204
B-factors (Å2)
Protein	22.2
Heme	23.8
Glycerol	61.0
Water	26.8
r.m.s.d.
Bond lengths (Å)	0.006
Bond angles (°)	0.884
Ramachandran statisticsc	90.1/9.1/0.6/0.1%e
Ramachandran statisticsd	97.5/2/2f
PDB code	3MGX

^a Numbers in parentheses correspond to the highest resolution shell (2.2-2.1 Å).

^b R_work = Σ‖F_o| − |F_c‖/Σ|F_o|, calculated from the working reflection set; R_free was calculated in the same manner using the 5% test set reflections.

^c Calculated by PROCHECK (47); percentage is shown for the protein residues in most favored/additional allowed/generously allowed/disallowed regions.

^d Calculated by MOLPROBITY (46); percentage is shown for the protein residues in most favored regions, disallowed residues, and bad rotamers.

^e Residue in disallowed region Asn-296A (loop region residue prior to a glycine).

^f Residues in disallowed regions Gly-39A and Gly-39B (loop region residues in a serine/glycine-rich loop).

Sequence Alignments and Homology Modeling

Sequence alignments were performed using the ClustalW2 server at EMBL-EBI (53). Similar sequences were determined using BLAST (54) as implemented by the ExPASy Proteomics Server (55). Secondary structure prediction was performed using the Jpred3 server (56).

Generation of BpsD PCP Domain Model

A homology model of the BpsD PCP domain was generated from the known structure of a PCP from the tyrocidine A NRPS (PDB code 2JGP) (57) using SWISS-MODEL (58); this template structure was chosen, because it contains the closest related PCP domain to the BpsD PCP domain with a crystal structure available in the PDB. The PCP model thus generated was docked onto the structure of OxyD using the program HEX 4.2 (59). Two of the docking solutions afforded possible binding orientations in terms of geometry and distance of the post-translationally modified serine reside to the heme iron. Similar conformations were observed with the docking of the apo-PCP molecule onto OxyD using the programs Patchdock (60) and Gramm (61).

RESULTS

OxyD Protein Expression

P450 OxyD was successfully overexpressed in BL21(DE3)pLysS cells, affording soluble protein (2 mg/liter expression culture) following cell lysis and clarification, albeit with high levels of protein expression resulting in inclusion bodies. An initial Ni-NTA affinity purification of the clarified cell lysate was followed by anion exchange to remove minor contaminants, whereas a final gel-filtration step was employed both as a final polishing step and to buffer exchange the protein prior to storage at −80 °C. The final protein purity, expressed as the absorbance of the heme Soret divided by the absorbance of aromatic residues at 280 nm was found to be 1.1, which compares well to the reported values of other bacterial P450s (35–37). OxyD appears to be monomeric as determined by gel filtration.

BpsD_PCP Protein Expression

To investigate the binding of carrier protein loaded with tyrosine by OxyD, attempts were made to express the isolated PCP domain of BpsD. Unlike the previously reported successful expression of the PCP6S and PCP7S subunits from the vancomycin NRPS proteins in E. coli (29), the expression of the PCP domain of BpsD and an N-terminally truncated construct were both unsuccessful in E. coli. Attempts to optimize the N-terminal sequence of the BpsD_PCP were also unsuccessful in yielding any protein, with attempts made to modify the N-terminal portion of the first α-helix to resemble that observed in the PCP-7S domain (29). The use of a vector series containing different N-terminal fusion proteins to enhance solubility was attempted (30). All fusion proteins showed overexpression of the desired proteins, with the majority exhibiting a reasonable soluble fraction. The construct initially used for expression of the BpsD PCP domain was a thioredoxin (Trx) fusion protein containing a hexahistidine tag between the Trx and BpsD_PCP domains. Truncation of the fusion protein by proteolytic degradation occurred at high levels, so a construct containing a C-terminal hexahistidine tag was utilized instead for expression of the Trx-BpsD_PCP fusion protein. This construct allowed both the successful expression of the Trx-BpsD_PCP fusion protein and allowed purification via single step Ni-NTA affinity purification to >95% purify as established by SDS-PAGE.

UV Characterization and Substrate Binding

UV-visible spectroscopy of the purified OxyD protein showed a typical substrate free P450 absorption spectrum, with λ_max at 419 nm and α/β bands at 536 nm and 572 nm. Titration of l-tyrosine into a solution of OxyD gave no change in the absorption of the protein even at high tyrosine concentrations (1 mm), indicating that free l-tyrosine does not bind to the active site of OxyD, as there was no perturbation of the heme iron (Table 1). Titration of OxyD with imidazole also did not result in any change in absorption, indicating no binding of imidazole even at concentrations as high as 25 mm. Titration of OxyD with l-phenylalanine and l-tyrosine loaded Trx-BpsD_PCP protein showed a shift in λ_max from 419 to 392 nm, indicating a spin state change of the heme iron caused by a binding interaction with an average dissociation constant of 14 ± 3 μm for the modified Trx-BpsD_PCP proteins (Fig. 2A and supplemental Fig. S1). There was no difference in affinity for the l-tyrosine-loaded Trx-BpsD_PCP (13 ± 4 μm) compared with the l-phenylalanine-loaded protein (14 ± 2 μm), although there was a decrease in the percentage spin state shift from 40% for l-tyrosine-loaded Trx-BpsD_PCP to 31% for l-phenylalanine-loaded Trx-BpsD_PCP. Additionally, and as expected, the binding of holo-Trx-BpsD_PCP (including the CoA-derived linker but not the bound amino acid) to OxyD revealed no spin state shift upon increasing concentrations of the substrate. In all cases, the use of the fusion protein was necessary due to the instability of the isolated PCP domain. These results are consistent with those obtained for the binding of PCP-loaded peptides by OxyB, in terms of both spin state change and dissociation constant (17 ± 5 μm) (29). They also indicate that only a fraction of the OxyD molecules are able to bind at any one point in time to the PCP-loaded substrates.

TABLE 1.

Spectroscopic determination of substrate binding to OxyD

Substrate	Binding mode	Kd/Ka	Spin shift
			%
l-Tyr	No binding		0
l-Tyr SNAc	Cooperative (n = 3)	11 ± 1 μm	92
l-Phe SNAc	Cooperative (n = 4)	12 ± 1 μm	75
l-Tyr Trx-BpsD_PCP	Single site	13 ± 4 μm	40
l-Phe Trx-BpsD_PCP	Single site	14 ± 2 μm	31

FIGURE 2.

Titration curves obtained probing the binding of l-tyrosine-loaded Trx-BpsD_PCP (A) and l-tyrosine SNAc to OxyD (B). X-axis, substrate concentration in μm; y-axis, change in the difference in absorbance between 419 nm and 392 nm.

The use of amino acid SNAc thioesters as substrate analogues for investigating non-ribosomal peptide synthases and biosynthetically related proteins is well established (22, 32, 33), with the advantage of much simpler substrate preparation and product analysis. The use of amino acid SNAc thioesters for binding studies with OxyD was therefore attempted. However, in this case the behavior of these substrates does not mimic tyrosyl-PCP binding to OxyD. A comparable situation regarding the appropriateness of SNAc thioesters for aminoacyl PCP P450 oxidases has been observed with NovI, whereby l-tyrosine SNAc was not hydroxylated by NovI (24). l-Tyrosine SNAc binds to OxyD in a cooperative manner (with a Hill constant (n) of 3 ± 0.2), with a K_a of 11 ± 1 μm, and a high percentage spin state shift (92%) (Fig. 2B). The binding of l-phenylalanine SNAc to OxyD displays similar behavior (K_a of 12 ± 1 μm), with a high Hill constant (n = 4 ± 0.8) and a high percentage spin state shift (75%). The binding of these small molecules thus does not represent a good model for PCP-loaded substrate binding to OxyD. Additionally, qualitative analysis of d-tyrosine SNAc, l-tyrosine CoA, and l-phenylalanine CoA binding to OxyD also reveals cooperative behavior (data not shown). These results indicate that the complete PCP-binding partner is necessary to obtain relevant binding data for OxyD substrates.

Overall Structure of OxyD

The structure of OxyD adopts the canonical P450 fold, with predominantly α-helical secondary structure surrounding a non-covalently linked heme (protoporphyrin IX) moiety (Fig. 3). In total there are 14 α-helices bearing the designations A′ to L and 3 β-sheets bearing the designations β-1 to β-3. The two OxyD monomers (A and B) in the asymmetric unit of the unit cell are structurally very similar (root mean square deviation (r.m.s.d.) of 0.4 Å for Cα atoms) (Table 2).

FIGURE 3.

Diagram showing the overall structure of OxyD (glycerol molecules shown in yellow, majority of OxyD shown in gray with certain structural motifs colored for clarity, and heme shown in red) with structural motifs labeled.

Active Site Architecture

The active site of OxyD contains the conserved catalytic residues and typical structure found in P450s, which includes a kink in the I-helix where the highly conserved acid/alcohol residue pair (Glu-238/Thr-239 in OxyD) controls the protonation of intermediate oxygen species during oxygen activation (Fig. 4A) (62). There is a water molecule bound to the heme iron at a distance of 2.0 Å, and at least one glycerol molecule is also present in the active site of OxyD (Fig. 4A). The glycerol molecules are in similar conformations in the two monomers of the asymmetric unit, with an oxygen atom the closest atom to the heme iron (4.3 Å and 4.4 Å for chain A and B, respectively) and hydrogen bonding to the heme-bound water (2.8 Å and 3.0 Å for chain A and B, respectively).

FIGURE 4.

A, active site of OxyD showing critical P450 catalytic residues and those interacting with the heme propionate groups, with hydrogen bonding distances indicated (critical catalytic residues shown in magenta, heme interacting residues shown in blue, heme-bound water shown in green, glycerol molecules shown in yellow, heme shown in orange, OxyD shown in gray). B, hydrogen bonding interactions between residues from different structural elements controlling the geometry of the active site pocket, with hydrogen bonding distances indicated (F or G helix residues shown in yellow, I-helix residues shown in cyan, B–B₂ loop residues shown in magenta, β1-sheet residues shown in green, C-terminal loop residues shown in orange, heme shown in red, and OxyD shown in gray).

The active site is composed of several structural elements, one of them being the N-terminal and central regions of the I-helix (residues 228–239), which extends across the face of the heme. The opposite side of the heme is enclosed by a β-strand and the immediately preceding residues (residues 281–287), which form part of the β-1 sheet. Above this beginning of the β-strand, residues Val-382 and Val-383 from the long C-terminal loop impinge upon the active site, whereas the opposite side of the active site, containing the heme propionate groups, is formed by the N- and C-terminal portions of the B–B₂ loop (in total, residues 69–88). The active site is rather open above the heme, with the C-terminal residues of the F-helix and the N-terminal residues of the G-helix forming a cap above the I-helix (residues 171–188). The long loop between the F and G helices is pushed away from the top of the active site, where it forms crystal contacts with the first β-strand in the β-1 sheet of another OxyD molecule. It is possible that this region closes upon substrate binding to the P450, whereas the conformation adopted by the central portion of the B–B₂ loop may also rearrange upon substrate binding.

The observed active site geometry is enforced by a number of inter-residue hydrogen bonds that hold the secondary structure elements in their given conformation (Fig. 4B). These include the interactions of the F and G helices with the I-helix, mediated through the interaction of the backbone carbonyl of the F-helix residue Asn-169 with the guanidine nitrogen atom of the I-helix residue Arg-241 (2.7 Å) and through the interaction of a histidine nitrogen atom of the G-helix residue His-188 with the carboxylate group of the I-helix residue Asp-230 (2.7 Å). The C-terminal portion of the B–B₂ loop interacts with the I-helix through hydrogen bonds between the backbone amide nitrogen atoms of Met-87 and Val-88 and the carbonyl side-chain oxygens of the I-helix residues Asn-228 and Asn-231 (2.9 and 3.0 Å, respectively) (Fig. 4B). The β-1 strand residue His-284 forms interactions with two other secondary structure elements in close proximity to the heme. These interactions are mediated by hydrogen bonding between the histidine side-chain nitrogen atoms, and the backbone nitrogen and carbonyl oxygen atoms of residues Ile-381 and Ile-72 (2.7 and 3.0 Å, respectively). The interaction with Ile-72 also makes it more likely that this section of loop containing the non-polar residues Met-Met-Ile runs across the top of the β-1 strand even after the binding of the PCP-bound substrate.

Structural Comparisons to Other P450s

Analysis of the closest related P450 structures to OxyD that are present in the Protein Data Bank was performed on a sequence level using BLAST and on a structural level using the DALI server. The two approaches produced essentially the same list of ten P450s with similarity to OxyD. OxyD shares the highest structural similarity with CYP124 and CYP125 from Mycobacterium tuberculosis (for CYP124, PDB code 2WM4 and r.m.s.d. (Cα) 2.4 Å (63); for CYP125, PDB code 3IW0 and r.m.s.d. (Cα) 2.4 Å (64)), although several other P450 structures also share similar r.m.s.d. (Cα) values (supplemental Table S1). Overall, the most significant differences between these structures and that of OxyD include the substrate recognition sites (65): the B–B₂ loop region, the relative orientation of the F and G helices to the B–B₂ region, the “open” nature of the loop connecting the F and G helices, and alterations in the length and orientations of loop extensions of the first loop in the β-1 sheet (supplemental Fig. S2).

Alignments with Other Amino Acid PCP-oxidizing P450s

Individual sequence alignments of OxyD with seven other tyrosyl-PCP-, tryptophanyl-PCP-, histidinyl-PCP-, and valyl-PCP-oxidizing P450s were performed, revealing an average sequence identity of 35% and an average sequence similarity of 50% (supplemental Table S2). In addition to the conserved P450 heme binding motif, regions of high conservation were identified that occur in the P450 substrate recognition sites (65).

With the ability to localize these areas upon the OxyD structure, concentrations of conserved residues can be seen occurring toward the surface opening of the active site and in the active site itself (Fig. 5). In particular, these regions include the N and C termini of the B–B₂ loop (Gly-69, Ile-72, Ser-82, Gly-83, Gly-84, Met-86, Val-89, and Ser-90), the C-terminal region of the F-helix (His-171, Ala-172, Phe-173, and Gly-174), the N-terminal region of the G-helix (Ala-187, His-188, Thr-189, Glu-190, and Val-193), the N-terminal region of the active site I-helix (Asn-228, Cys-229, and Gly-235), and the β-1 sheet adjacent to the heme (Ala-282, Met-283, and His-284) (Table 3).

FIGURE 5.

Conserved residues found in the alignment of PCP-bound amino acid P450s localized on the structure of OxyD. B–B₂ loop residues are shown in magenta, F-helix residues are orange, G-helix residues are yellow, I-helix residues are cyan, β-1 sheet residues are green, heme is dark gray, and OxyD is gray with secondary structure elements labeled.

TABLE 3.

Conserved sequence regions in the alignment comparisons of known P450s with similar substrates to OxyD

Four of these residues correspond to residues involved in maintaining the relative geometry of the secondary structural elements (Fig. 4B), with the rest in the correct position for either interaction with the PCP or forming the active site around the bound substrate. Alignment of these 23 residues among the sequences of the PCP-oxidizing P450s revealed a very high degree of conservation of these residues, with very few altered to non-comparable chemistries (17 such mutations across 230 total residues). Most differences from the consensus sequence are seen in the sequence of the valine-oxidizing P450 ZbmVIIc, which is somewhat unsurprising given that this substrate is also most distinct from the others analyzed here. Alignment of the ten most similar structures to OxyD, none of which catalyze the oxidation of aminoacyl-PCPs but three of which have carrier protein-bound substrates, indicated that the particular sequence regions indicated above are able to distinguish P450s that interact with PCP-bound amino acid substrates compared with other substrates (supplemental Table S3). Of the 23 residues identified above, a maximum of 7 similar residues were present in one sequence (that of CYP130), whereas the majority of sequences shared only 2–4 similarities.

Alignment and Modeling of PCP Domains That Interact with Amino Acid-oxidizing P450s

Alignment of the PCP domains from NovH, NikQ, ZmbVIIb, and Ecm12 were made with the PCP domain from BpsD. All sequences exhibit ∼40% sequence identity and 50–60% similarity upon alignment, with the predicted secondary structure elements also matching well (supplemental Table S4). The sequence similarity around the post-translationally modified serine residue at the beginning of the predicted helix α-2 is the highest, because this region exhibits the motif required for phosphopantetheinyl transferases to convert the apo-PCP domains into the phosphopantetheine-modified holo forms. Allowing for single amino acid miss-alignments per residue in a five-way sequence alignment, the region of highest conservation not including the phosphopantetheinyl transferase motif is the C-terminal portion of helix α-2, followed by the C-terminal region of the α-1 helix and the short helix α-3. The majority of the conserved regions in α-1 and α-4 are predicted to project into the hydrophobic pocket between the helices, indicating that the potential interaction interface of the PCPs is centered on the α-2 helix and possibly extending to include the α-3 helix. Docking studies of a homology-modeled BpsD PCP domain afford two possible docking solutions where the post-translationally modified serine residue is within 20 Å of the heme iron (Fig. 6). Both these solutions indicate a preferential binding for the apo-PCP in the junction of the C terminus of the I-helix and the G-helix of OxyD, although it must be stressed that this docking does not explore the potential structural rearrangement of the OxyD structure upon interaction with the tyrosyl-PCP and as such acts merely as a guide to future interaction studies between the PCP and OxyD. It does, however, indicate an interface involving helices α-2 to α-3, which appears feasible given the sequence alignment analysis. The surface charge distribution of OxyD indicates that the active site of OxyD is predominantly negatively charged, with the exception of one residue in the G-helix (Arg-185), while the docked PCP structures are both oriented in such a way as to present a large patch of positive charge at the potential docking interface (Fig. 7).

FIGURE 6.

Docked homology models of the PCP domain of BpsD onto OxyD. OxyD is shown in gray, critical active site residues are colored as in Fig. 6, PCP-model one is cyan, PCP-model two is cyan, and selected OxyD secondary structure elements are labeled; A and B differ in a 45° rotation.

FIGURE 7.

Surface charge of OxyD (A, top view) and docked PCP-models (90° rotation of A into the page; PCP-model one (B), and PCP-model two (C)). OxyD possesses a predominantly negatively charged active site, whereas the PCP domains are both oriented to present a positively charged surface toward OxyD (model of OxyD shown without surface charge for clarity, charged displayed as a surface of blue (positive) and red (negative) charges, OxyD is gray, heme is displayed in gray, PCP-model one is cyan, and PCP-model two is magenta).

DISCUSSION

OxyD Binds PCP-bound Amino Acid Substrates

The binding of substrates to P450 OxyD has been investigated by monitoring the changes in heme environment upon substrate binding to the P450. This indicates that binding of amino acids to OxyD does not occur unless the BpsD PCP domain and prosthetic phosphopantetheine linker are attached to the amino acid. This is understandable in biosynthetic terms, because this then allows the diversion of amino acids into secondary metabolism to be more tightly controlled without the indiscriminate oxidation of amino acids otherwise required for protein biosynthesis. The presence of the PCP domain and linker is a major determinant in binding selectivity for OxyD, with either phenylalanine of tyrosine loaded PCPs exhibiting similar binding efficiency to OxyD.

Identification of a Common Motif in P450s Oxidizing Amino Acid-bound PCPs

Sequence alignments of seven P450s with similar substrates to OxyD have revealed 23 highly conserved residues across 6 different regions of the structure, which correspond to the substrate recognition sites in P450s (65). These regions of similarity within the amino acid-bound PCP-oxidizing P450s may be used in distinguishing these P450s from P450s catalyzing the oxidation of other substrates. The lowest internal agreement for any amino acid-bound PCP-oxidizing P450 with the common sequence identified for the 23 highly conserved residues was 70% for ZmbVIIc, which despite the different nature of the amino acid substrate in this case (valine) still exhibits a high level of sequence identity.

The degree of conservation of the 23 “fingerprint” residues that aminoacyl-PCP binding P450s display compared with P450s catalyzing different oxidative transformations is low, allowing these regions to be used to differentiate amino acid-bound PCP-oxidizing P450s from P450s catalyzing the oxidation of different substrates. The highest similarity found comparing the conserved amino acid-bound PCP-oxidizing P450 sequence to that of a P450 with a different substrate is 30% (CYP130), a figure close to the sequence identity for the protein as a whole (27%). All other structurally similar proteins gave lower similarity in matches to the regions of similarity identified within the amino acid-bound PCP-oxidizing P450s than their overall sequence identity with OxyD.

The regions identified as highly conserved in the amino acid-bound PCP-oxidizing P450s are appropriately placed around the active site to suggest that they represent a common binding motif for these proteins; thus they help to present a binding surface almost exclusively dominated by negative charges. The oxidative selectivity normally built into the active sites of P450s is reduced in the case of PCP-bound substrates, firstly due to the PCP carrier molecule and prosthetic linker presenting a large binding surface for recognition and secondly due to the nature of the upstream processes that lead to the loading of the PCP domain of the non-ribosomal peptide synthase that acts as the P450 substrate. The selectivity of the adenylation domains of NRPSs for selecting one amino acid is very high, thus reducing the need for the P450 to select the correct amino acid through a highly optimized active site. This lowered selectivity is seen in the equivalent binding of PCPs to OxyD loaded with either phenylalanine of tyrosine, and explains the apparent lack of coordinating residues for the tyrosine phenol group within the active site. This does, however, lead to the situation that the PCP becomes the major determining factor in substrate binding to amino acid-bound PCP-oxidizing P450s.

Identification of P450 Function from Primary Sequence Analysis

The identification of common motifs in PCP-bound amino acid P450 oxidases suggests that analysis of primary sequence data can reveal whether the substrate for a P450 is an aminoacyl-PCP. Using this type of analysis, it would appear that CYP185A1 (CypLB) from Streptomyces tubercidicus I-1529 (66) encodes an aminoacyl-PCP oxidase (Table 3). This hypothesis is supported by the presence of an upstream non-ribosomal peptide synthase and a downstream isolated thioesterase domain, in a similar arrangement to that seen with OxyD. Unfortunately, the sequence of the adenylation domain of the upstream NRPS was not included in the submitted sequence data, making the exact nature of the amino acid substrate unclear. It would appear, however, that, based upon the sequence analysis and the surrounding genes, that CYP185A1 is a PCP amino acid hydroxylase. The determination of P450 substrate based upon analysis of the protein sequence has been attempted for many P450s with only limited success, unless there is high sequence similarity to a P450 of known function. The ability to assign PCP-bound amino acid oxidation to a P450 based upon the identified sequence motifs constitutes a powerful tool in the future analysis and discovery of new biomolecules.

Interaction of the PCP Domain with OxyD

Given the importance of the role of the PCP domain in substrate binding to amino acid-bound PCP-oxidizing P450s, it would be expected that the similar binding surfaces formed by these P450s would require PCP domains to possess high degrees of similarity; in fact, the similarity of the PCP domains in the systems discussed here is over 50% in all cases and is often much higher (supplemental Table S4). The alignment of predicted secondary structure matches >90% in all cases, with the greatest sequence conservation centered on the long α-2 helix (the post-translationally modified serine is found in a loop prior to the α-2 helix). The surface charge distributions of the PCP models indicate an overall positive charge across the surface of the PCP formed by the α-2 and α-3 helices, complementary to the overall negative charge seen on the OxyD binding surface. Docking a model of the apo-PCP with Hex onto OxyD affords two favorable binding conformations, where these regions of complementary charge interact (Fig. 7). Crucially, these possible arrangements present the post-translationally modified serine residue closer to the active site heme than that seen in the case of P450_BioI, an arrangement made necessary by the fact that in P450_BioI the sites of oxidation in the fatty acid substrate (C7 and C8 of the fatty acid chain) are ∼9.1 Å from the thioester linkage with the acyl carrier protein phosphopantetheine linker. In the case of OxyD, however, the substrate is a PCP-bound tyrosine residue, where the distance from the tyrosine β-carbon to the thioester linkage of the PCP phosphopantetheine linker is only 4.1 Å. This implies a much more intimate binding of the two proteins, with the PCP having to reside closer to the active site of OxyD. It must be stressed, however, that the docked models do not reflect the potential for rearrangement of the OxyD protein upon PCP binding. The docked models are, however, consistent with a wider opening of the active site of OxyD to accommodate the PCP binding partner.

Conclusions

Cytochrome P450 OxyD, found in the balhimycin biosynthetic operon of A. mediterranei, is the first PCP-bound amino acid P450 hydroxylase to have been structurally characterized. Comparative alignment of P450s catalyzing the oxidation of PCP-bound amino acid substrates and visualization on the OxyD structure have revealed a highly conserved “fingerprint” for such P450s, corresponding to the probable PCP binding site. The importance of the PCP domain in substrate binding is highlighted by the cooperative binding of non-physiological shortened SNAc thioesters, whereas the PCP-bound substrate exhibits low micromolar affinities. Crucially, the identification of a conserved P450 PCP-bound amino acid binding site makes it possible to assign such P450 substrate specificity based solely upon identified sequence motifs. With the prevalence of hydroxylated amino acid residues in many important natural products, this establishes an important tool with implications for the future discovery of new biomolecules.