Oxidative transformations of amino acids and peptides catalysed by Cytochromes P450¹

Enzymatic oxidation is an essential reaction in both primary and secondary metabolism. Owing to the large variety of substrates and types of reactions to be performed, a myriad of oxidoreductases has evolved (see Table 1). In general, the electron transfer is mediated by cofactors that together with the protein environment create specific redox potentials. These characteristic potentials determine the kind of reaction that can be catalysed by a certain class of enzymes. A very abundant and prominent class of oxidoreductases comprises Cytochrome P450 monooxygenases that will be subject of this book chapter in the context of oxidation of amino acids and peptides.

In 1962 Cytochromes P450 (P450s) were described for the first time as coloured pigments in rat liver microsomes.⁹  As described in this publication P450s show a very characteristic absorption maximum at 450nm when the protein solution is saturated with CO and subsequently reduced with dithionite. This maximum was later shown to be unique for cytochromes bearing a prosthetic, thiolate coordinated low-spin b-type haem.¹⁰  The thiolate ligand was subsequently identified as an invariant cysteine residue which lies within a highly conserved region of the protein. High sequence conservation in proximity to the haem and in other sections of the protein now allows the facile identification of new P450s based on predicted open reading frames. In this way more than eleven thousand genes through all kingdoms of life have been classified as members of the P450 superfamily and grouped into families and subfamilies based upon their sequence homologies.⁷ 

The first X-ray structure of CYP101A1 (P450_CAM; P450 nomenclature will be given along with the general enzyme name when available) revealed a characteristic P450 fold that has been shown to be well conserved amongst the superfamily (Fig. 1).¹¹  The typical structure shows a high content of α-helices named in letters starting at the N-terminus (A to L) and a few β-strands. The most prominent helix is the I-helix spanning the whole molecule and coming in close proximity to the distal plane of the haem. The F and G helices, as the most mobile parts of the protein, are supposed to restrict the entry of potential substrates into the active site cavity.¹²  In the low spin, resting state the ferric haem (Fe(III)) is coordinated to a distal water molecule.

In the catalytic cycle of P450_CAM as depicted in Fig. 2, the entry of the substrate into the active site displaces the distal water molecule leading to high spin ferric haem (Fe(III)). This is observable as a shift in absorption in the visible spectrum (decrease of absorption at 420nm and increase at 390nm). Electron transfer from a redox partner leads to reduction of the haem iron followed by formation of a ferrous dioxy complex with molecular oxygen (Fe(II)⁺-O₂). Sequential transfer of a second electron and a proton lead to formation of Compound 0 (Fe(III)⁺-OOH). After a second protonation, the O-O bond is cleaved and water is released, which leads to the formation of the reactive ferryl-oxo porphyrin cation radical, Compound I (Fe(IV)⁺-O). Following hydrogen atom abstraction from the substrate, radical recombination leads to the formation of the hydroxylated product which is released from the active site while the monooxygenase relaxes to its ferric resting state.¹³ 

In vitro the proton and electron transfer steps can sometimes be circumvented by saturating the ferric haem with hydrogen peroxide, a process called the peroxide shunt pathway. The advantage of this technique is the simplification of the reaction as neither redox partners nor expensive cofactors (NADH or NADPH) are required.¹⁴ 

The interplay of P450s with electron transfer systems is almost as diverse as the reactions catalysed by this machinery. Hannemann et al. grouped all known P450-redox systems into ten classes characterised in part by the number of polypeptides involved (one, two or three) and localisation of the system (cytosolic or ER membrane bound).¹⁵  The most well-studied P450, P450_CAM, is the prototype of a bacterial member of class I. Electron transfer occurs between the 2Fe-2S iron-sulfur cluster of putidaredoxin (Pd) and the P450. An NADH-dependent FAD-containing ferredoxin reductase, putidaredoxin reductase (PdR), reduces the oxidised putidaredoxin after each electron transfer to the P450. Mitochondrial P450s share the same topology, but many systems are membrane-associated. Most liver-microsomal P450s, however, belong to class II, where the redox system is incorporated in a single polypeptide chain (diflavin reductases). Electrons are transferred from an FMN cofactor via an FAD cofactor to the P450. These two classes (I and II) are the most important ones and their members have been studied extensively.

Despite this, bacterial P450s are often found as “orphan” genes with no genes coding for potential redox partners in the surrounding regions. Biosynthetic gene clusters for glycopeptide antibiotics, for example, contain several genes coding for P450s but none for potential redox partners.¹⁶  In order to study these P450s the peroxide shunt pathway is applied and/or heterologous redox systems are tested. However, some P450s show strong specificity for their redox partners because of protein-protein interactions and the delicate regulation of redox potentials.¹⁷  In the P450_CAM catalytic cycle putidaredoxin reductase (PdR) can be replaced by palustrisredoxin reductase (PuxR) whereas the same substitution on ferredoxin level doesn't allow electron transfer.¹⁷  For other P450s, however, the palustrisredoxin/ palustrisredoxin reductase (Pux/PuxR) system has successfully been employed as heterologous electron donors.¹⁸  This indicates the potential to exploit P450s for biocatalytic processes even where an endogenous redox system has not been identified.

Cytochrome P450 enzymes play very important roles in all kingdoms of life. On the one hand they are involved in catabolic processes for the degradation of xenobiotics, such as pharmaceuticals, pesticides and other potentially detrimental compounds that are taken up by an organism.¹⁵  On the other hand P450s are key players in the biosynthesis of many secondary metabolites. In mammals these range from steroids and vitamins to fatty acids, whereas in prokaryotes many products are part of biological warfare between individual species²⁰  (e.g. glycopeptide antibiotics produced by Actinomycetes).^19,20  However, there are also examples of eukaryotes that take advantage of such metabolites in inter- and intraspecific competition.²¹  Many of these secondary metabolites are based on amino acids or peptides that undergo several modifications until they reach their full biological activity. The typical P450 catalysed reactions within this maturation process are hydroxylation, epoxidation, heteroatom oxidation or phenolic coupling. In some cases the substrates are free in solution whereas in other cases they are presented to the monooxygenase by a carrier protein.

A wide range of secondary metabolites in prokaryotes and eukaryotes are derived from amino acids and peptides. In particular, prokaryotic peptide derivatives have gained great importance as their antibiotic effects are exploited in clinical applications (penicillin,²²  cephalosporin,²³  vancomycin,²⁴  bottromycin A2²⁵ ). These peptides can either be of ribosomal (e.g. lantibiotics) or of non-ribosomal origin (e.g. glycopeptide antibiotics).

Secondary metabolites derived from ribosomal peptides have a variety of appearances and functions. Their only common feature is that they are transcribed from genes and translated into a precursor peptide consisting of leader and core sequence.²⁶  The leader sequence serves as a guide through maturation of the core peptide by recruiting tailoring enzymes and it mediates the secretion of the product (see Fig. 3). The most abundant post-translational modifications are listed in Table 2 and have been reviewed in detail by McIntosh et al.²⁶ 

The mechanism and order of these maturation steps strongly depend on the substance class. In the discussion we will focus on the maturation of lantibiotics which involves the actions of P450 enzymes.³³ 

All genes contributing to the biosynthesis of a certain lantibiotic are organised within a gene cluster starting with lanA, the gene coding for the precursor peptide. All tailoring enzymes such as the lanthionine synthetase are encoded downstream of the lanA gene. Lanthionine synthetase is a homomeric or heteromeric bifunctional complex that catalyses the dehydration of serine or threonine and the subsequent addition of the thiol group of a cysteine.³⁴  This thioether forms the lanthionine bridges that provide extraordinary rigidity and stability against proteases to the peptides (see Fig. 7).³⁵  It has been shown for lacticin 481 that basal synthetase activity can be detected in the absence of the leader whilst the directionality of processing from N- to C-terminus is lost.³⁶  Furthermore, leader sequences were found to direct synthetase activity even in trans but with decreased efficiency compared to the intact precursor peptide.³⁶  Following the actions of lanthionine synthetase and other tailoring enzymes, the leader sequence mediates the recognition by cellular secretion systems and is cleaved from the mature peptide. Most lantibiotics have been shown to be inactive in presence of the leader. Thus, the leader peptide not only drives peptide maturation, it also takes part in secretion and self-immunity.³³ 

From a commercial point of view ribosomal peptides are of significant interest as they are easily modified to generate a range of new bioactive compounds. Simple exchange of codons in the core sequence can afford fully functional or even improved compounds as long as the leader remained unchanged.³⁷  Furthermore, a new application utilising the versatile tool box of naturally occurring post-translational modifications mediated by leader peptides called leader peptide assisted biosynthesis (LPB) has been established.³⁸  Target peptides can be expressed in fusion with a helper leader in order to improve peptide characteristics such as pharmacokinetics, bioavailability, stability, or conformational uniformity amongst others.^39–41  LBP can even be combined with the use of commercially available proteases, ensuring full control over the complete maturation and facilitating isolation of the final peptide.⁴² 

The other group of peptidic secondary metabolites is of non-ribosomal origin. These peptides are assembled by large enzyme complexes called Non-Ribosomal Peptide Synthetases (NRPS). The generic topology of an NRPS is characterised by a modular organisation where each module specifically incorporates a certain amino acid into the peptide (see Fig. 4). The modules are arranged in one or more polypeptide chains with the exact number of modules depending on the system. The minimal module consists of an Adenylation (A), a Peptidyl Carrier (P or PCP) and a Condensation (C) domain. These modules can be extended by further domains such as Epimerisation (E), Methyltransferase (Mt), Cyclisation (Cy), Oxidation (O), Reduction (R) or Thioesterase (Te) domains.⁴³  In the catalytic process the A-domain activates an amino acid by forming an acyl adenylate with ATP and transfers it onto the 4′-phosphopantetheinyl (Ppant) arm of the PCP domain, where it forms a reactive thioester.⁴⁴  The Ppant group shuttles amino acids from the active site of the A-domain to the active site of the C-domain. The C-domain bears two cavities where the amino acids (i) and (i+1) can be accommodated.⁴⁵  In these positions the free N-terminus of the tethered (i+1) amino acid attacks the Ppant-bound C-terminus of residue (i) leading to peptide bond formation. The nascent peptide chain remains attached to the (i+1) Ppant and is transferred to the active site of the next condensation domain (see Fig. 4). The final module of the NRPS contains an additional thioesterase domain that cleaves the thioester bond between the last PCP and the peptide. Depending on the system this step yields either a linear or a N-C macrocyclic peptide.⁴⁶ 

The number of additional domains in these modules (E, Mt, Cy etc.) already suggests that with this system peptides with a range of unusual amino acids (methylated, Mt-domain; cyclised, Cy-domain) and amino acid configurations (E-domain) can be produced. Further diversity is introduced by incorporation of non-proteogenic amino acids like hydroxyphenylglycine.⁴⁷  An array of tailoring enzymes such as methyltransferases, halogenases and oxygenases modify the peptide backbone to obtain highly stable and bioactive compounds. Specific transferases attach sugar or fatty acid decorations depending on subcellular localisation and other required features as e.g. membrane permeability.

It was shown that the actions of A-domains are highly specific for the side chains of their substrates⁴⁹  whereas C-domains are very specific towards the stereochemistry of amino acids.⁵⁰  These mechanisms ensure the correct sequence and stereochemistry of the generated peptide. These characteristics can be exploited for the analysis of newly discovered NRPSs; new NRPS gene clusters are identified by the high homology of A-domains (up to 56% sequence similarity) and from the active site residues of A- and C-domains one can predict the sequence and stereochemistry of the product peptide with bioinformatic tools.⁴⁹ 

The modularity of the NRPS offers a vast platform for biotechnological applications in order to generate new peptides. The obvious approach, namely genetic swapping of domains or whole modules, was soon proven to be difficult:⁵¹  only a few examples afforded functional peptides in acceptable yields.^52,53  Possible explanations are the high substrate specificities of the later synthetic domains, the inactivity of swapped domains due to incorrect boundary selection or inadequate communication between the domains and modules. Hahn and Stachelhaus identified short regions (15-25 amino acids) at the edges of NRPS subunits called communication-mediating domains (COM).⁵⁴  These domains are crucial for the assembly of the NRPS and the communication between the subunits and thereby impose additional complications on domain swapping.

Eppelmann et al. have managed to alter the substrate specificity of A-domains by site-directed mutagenesis.⁵⁵  However, this approach is only suitable for conservative changes, where few mutations are needed and the substrate is still tolerated by the downstream domains.

Ribosomal and non-ribosomal peptides are processed by a myriad of tailoring enzymes contributing to the rich structural and functional diversity of these secondary metabolites. One important modification is oxidation, e.g. oxidation of thiazolines, oxidative coupling of aromatic side chains, β-hydroxylations and epoxidations. These reactions can be catalysed by members of the P450 superfamily on amino acid and peptide level and will now be discussed.

P450-catalysed hydroxylation of amino acids and peptides is mainly found in secondary metabolic pathways. Gene cluster analyses have identified a large number of P450s implicated in such reactions, although in many cases only homology studies have been used for assigning function and still less is known about the exact substrates. In the biosynthesis of natural products derived from amino acids P450s often act as tailoring enzymes on PCP-bound substrates.⁵⁶  This is especially the case for the β-hydroxylation of amino acids, which will be discussed below.

Hydroxylated amino acids are widely used intermediates in the biosynthesis of a variety of complex secondary metabolites. Many substances with antibiotic, antiviral or antifungal properties are among these natural products and are highly interesting for biomedical research. Somewhat surprisingly, most P450-catalysed oxidations on the β-position of amino acids are performed on carrier protein-bound substrates. The desired amino acid is recognised by the A-domain of an NRPS and loaded onto the PCP domain. The aminoacyl-PCP then serves as a substrate for a P450, which catalyses the oxidation of the β-position of the tethered amino acid: this reaction often proceeds with high stereoselectivity. Following this oxidation, the hydroxylated amino acid is released by a thioesterase domain (see Fig. 5) and further metabolised. β-Hydroxylation of tyrosine, tryptophan, histidine and valine residues have been found to proceed via this pathway and it is hypothesised that restricting oxidation to amino acids in their PCP-bound form is used to control the fraction of amino acids diverted into secondary metabolism.⁵⁷ 

Aminocoumarin antibiotics. Aminocoumarin antibiotics, including novobiocin, clorobiocin, coumermycin A₁ or the simocyclinones (Fig. 6) are produced by certain Streptomyces species and show antiobiotic activity against Gram-positive bacteria. They effect their antibiotic activity by inhibiting DNA replication through binding to bacterial DNA gyrase.⁵⁷  The coumarin core of these antibiotics is derived from L-tyrosine, which in the first biosynthetic step is oxidised to (2S,3R)-β-OH-tyrosine. From the novobiocin biosynthetic gene cluster of Streptomyces spheroides, novI and novH have been shown to encode a P450 and a di-domain NRPS consisting of a tyrosine-selective A domain and a C-terminal PCP domain. In vitro experiments with these proteins revealed the high substrate-selectivity of NovH for L-tyrosine and demonstrated the direct oxidation of tyrosine bound to NovH with NovI (CYP163A1), which exclusively generates (2S,3R)-β-OH-tyrosine.⁵⁷  In a divergence from the typical scheme for formation of such hydroxylated amino acids, their cleavage from the PCP occurs through a subsequent enzymatic transformation rather than thioesterase-mediated cleavage.⁵⁸  In the biosynthetic gene clusters of clorobiocin, coumermycin and simocyclinone the genes encoding the NovI homologues CloI (CYP163A2),⁵⁹  CumD,⁶⁰  and SimI (CYP163A3)⁶¹  have been identified but not further evaluated. In the biosynthetic pathways of novobiocin and clorobiocin L-β-hydroxytyrosine serves additionally as precursor for the prenylated benzoic acid moiety of these antiobiotics.

Glycopeptide antibiotics. L-β-R-hydroxytyrosine is also found in glycopeptide antibiotics, such as vancomycin and teicoplanin (see 2.4.1, Fig. 13). In the biosynthesis of vancomycin, β-hydroxytyrosine is provided by the interplay of the di-domain NRPS synthase BpsD, the P450 OxyD (CYP146A1) and the thioesterase Bhp. In vitro studies performed with purified OxyD gave greater insight into substrate binding by the P450.⁶²  OxyD equally accepts PCP-bound tyrosine or phenylalanine, suggesting that substrate specificity is determined mainly by the respective A-domain of the NRPS. No binding was observed for the isolated amino acids.⁶²  Amino acid SNAc thioesters – well-established substrate analogues used for the investigation of NRPS systems⁶³  – showed only unusual, cooperative binding behavior. An X-ray crystal structure of OxyD revealed a rather open active site, supporting the fact that the amino acid has to be inserted into the active site attached to the phosphopantetheinyl linker. Thus, only PCP-loaded substrates were able to bind to OxyD, with dissociation constants in low micromolar range.⁶² 

Nikkomycin antibiotics. Nikkomycins are a group of dipeptidyl nucleoside antibiotics produced by Streptomyces tendae Tü901 and Streptomyces ansochromogenes. They act as strong competitive inhibitors of chitin synthase by mimicking its natural substrate UDP-N-acetylglucosamine thereby inhibiting growth of insects and fungi.⁶⁴  They consist of two unnatural amino acids, hydroxypyridylhomothreonine (HMT) and an N-glycosidically modified aminohexuronic acid.⁶⁵  The nucleoside part of nikkomycins can contain as nucleobase either uracil (nikkomycin J, Z) or 4-formyl-4-imidazolin-2-one (nikkomycin I, X), whereas L-histidine has been shown to be the biosynthetic precursor for the latter. In S. tendae Tü901 the NRPS NikP1 selects and loads L-histidine onto its PCP-domain. β-Hydroxylation of the PCP-bound histidine residue is carried out by P450 NikQ (CYP162A1), affording (2S,3R)-3-OH-histidine.⁶⁶  In the nikkomycin gene cluster of S. ansochromogenes, sanQ is the corresponding P450-encoding gene (CYP162A2).⁶⁷  Following P450-catalysed oxidation, the (2S,3R)-3-OH-histidine is released from NikP1 by the thioesterase NikP2 and further metabolised in the nikkomycin synthesis pathway. In vitro oxidation studies using purified proteins verified and highlighted the high substrate selectivity and stereospecificity achieved by the interplay of NikP1 and NikQ.⁶⁵ 

A P450 is also involved in the biosynthetic pathway of the second unusual amino acid in nikkomycins, HMT. Mutation studies revealed that the P450s NikF (CYP105K1) in S. tendae Tü901 and SanH (CYP105K2) in S. ansochromogenes catalyse the aromatic hydroxylation of the pyridyl residue of this amino acid.^66,67  Strains with disrupted nikF/sanH genes only produced non-hydroxylated pyridylhomothreonine containing nikkomycins. As mutations in sanI (apparently encoding a ferredoxin) cause a down-regulation of SanH activity, it has also been suggested that SanI is the natural redox partner of SanH. The natural redox partners for NikQ and SanQ have not been identified.

Quinomycin antibiotics. Quinomycins are produced by various species of Streptomyces and can exhibit highly potent antibacterial, antiviral and/or antitumor activities. Characteristic for these C₂-symmetric, cyclic depsipeptides like echinomycin,⁶⁸  thiochoraline⁶⁹  or SW-163D,⁷⁰  are two intercalative chromophores, either quinoxaline-2-carboxylic acid (QXC) or 3-hydroxyquinaldic acid (HQA), attached to the peptide core. Both chromophores are derived from L-tryptophan, but two different biosynthetic pathways for the chromophores were originally suggested. In the pathway initially proposed for HQA the P450 TioI hydroxylates quinaldic acid in the final step to HQA.⁶⁹  In contrast, feeding experiments with deuterated (2S,3S)-β-hydroxytryptophan revealed that stereospecific β-hydroxylation of PCP-bound L-tryptophan by the TioI-homologue Ecm12 happens at the beginning of QXC biosynthesis.⁷¹  Further investigations have identified 3-hydroxy-L-kynurenine, derived from hydroxytyrosine, as a common intermediate in biosynthesis of both chromophores, strongly suggesting that β-hydroxylation of a PCP-bound amino acid by a P450 initiates secondary metabolite biosynthesis.^72,73  In an interesting development for the biological production of such compounds, the biosynthetic pathways of echinomycin⁷⁴  and SW-163D⁷⁵  have been re-engineered by Wantanabe and colleagues in a plasmid-based system allowing the production of bioactive natural products in E. coli.

Other examples. There are further examples where P450s have been identified or suspected to catalyse β-hydroxylations on PCP-bound amino acid substrates. In the biosynthesis of the bleomycin antibiotic zorbamycin and the highly potent proteasome inhibitor salinosporamide A,⁷⁶  slightly different amino acids are subjected to β-hydroxylation by a P450. In the former case, gene cluster analysis implicated β-hydroxyvaline as an intermediate synthesised by the interplay of the NRPS ZbmVIIb and the P450 ZbmVIIc,⁷⁷  while in the latter the P450 SalD (CYP163B1) has been shown to hydroxylate the PCP-bound unusual amino acid L-3-cyclohex-2′-enylalanine.^76,78  In both cases the biosynthetic route is continued by fusion of the hydroxylated PCP-bound amino acid with the PKS machinery.

Beside β-hydroxylation of PCP-bound amino acids, P450s are also found to catalyse a variety of hydroxylation reactions on peptide-derived substrates in secondary metabolism (see Fig. 7). The ribosomally-synthesised, 23 amino acid-long lantibiotic microbisporicin is produced by the actinomycete Microbispora corallina. A P450, MibO, has been identified in the microbisporicin gene cluster and is believed to be responsible for the conversion of a proline residue into the unusual 4-hydroxyproline and 3,4-dihydroxyproline residues found in the product peptide.²⁷ 

The ribosomally-synthesised bottromycin antibiotics⁷⁹  exhibit activity against methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci by selectively blocking the aminoacyl-tRNA binding site of bacterial ribosomes. Following identification and analysis of the bottromycin gene cluster from Streptomyces bottropensis, it has been hypothesised that the thiazole moiety, which is important for the biological activity of several secondary metabolites, is generated by β-hydroxylation of a cysteine-derived thiazoline ring (Fig. 7 path A). P450 BmbI (highly similar to CYP283A1) is believed to oxidise the thiazoline moiety and this is followed by subsequent elimination of water and decarboxylation.²⁸  However, in another study Müller and colleagues discuss as an alternative thiazole biosynthetic route with the oxidative decarboxylation step of cysteine as the first step, followed by a cyclodehydration step (Fig. 7 path B).⁸⁰  Therefore, further data are necessary to identify the correct mechanism.

The antibiotic skyllamycin, a cyclic depsipeptide produced by several strains of Streptomyces, has been identified as a highly potent inhibitor of the platelet-derived growth factor signaling pathway. Skyllamycin A and B feature, besides an unusual α-hydroxylated glycine residue, three β-hydroxylated amino acids ((2S,3S)-β-hydroxyphenylalanine, (2S,3S)-β-hydroxy-O-methoxytyrosine and (2S,3S)-β-hydroxyleucine). Interestingly, only one P450 (Sky32, CYP163B3) is responsible for all three β-hydroxylation reactions. This was proven by inactivation of the sky32 gene leading to the isolation of non-β-hydroxylated skyllamycin from the knockout strain, which clearly indicates that hydroxylation happens on the peptide substrate and not before peptide assembly. This was additionally supported by feeding experiments: hydroxylated amino acids were not able to restore wild-type peptide production.⁸¹  The selectivity of this P450 has also now been clarified in vitro.¹³⁶ 

The natural product himastatin (see Fig. 9) possesses several structural features derived from different P450-catalysed oxidation reactions. The P450 HmtN acts as a regio- and stereoselective γ-hydroxylase of an unusual D-piperazic acid residue leading to a monomeric himastatin precursor.⁸² 

Bacterial P450s involved in secondary metabolism mostly catalyse only one reaction. However, in the biosynthesis of the phytotoxic thaxtomin A, two consecutive hydroxylation reactions (see Fig. 10) are performed by the P450 TxtC (CYP264A1).⁸³  Remarkably, TxtC modifies two very different sites on the substrate, in terms of both their structure and reactivity: one reaction is hydroxylation of a tertiary aliphatic carbon on the diketopiperazine moiety and the second is the hydroxylation of an aromatic ring. The mechanistic details of these transformations remain to be elucidated.

P450s are well known to oxidise alkene residues to epoxides: they can also perform the oxidation of aromatic rings, where the intermediate is generally believed to be, or at least is most simply rationalised as, a highly reactive epoxide. Whilst not widely prevalent in the oxidation of amino acids and peptides by P450s, there are a limited number of examples that centre upon the apparent epoxidation of tryptophan residues, with resultant interception of the epoxide by a nitrogen atom within the parent substrate to generate a new heterocycle in the product. The examples identified here include compounds sharing an indolactam core (Fig. 8), and the cyclic depsipeptides himastatin and the kutznerides.

Lyngbyatoxin. The best-characterised example of aryl oxidation leading to bond formation by a P450 is from lyngbyatoxin biosynthesis (Lyngbya majuscula), where the P450 LtxB (CYP107CC1) is responsible for the formation of the nine membered ring of the (-)-indolactam V moiety.⁸⁴  The substrate for this P450 consists of a dipeptide alcohol produced by an NRPS from N-methyl-L-valine and L-tryptophan. This is released from the NRPS via reduction to the C-terminal alcohol, catalysed by an NADPH-dependent reductase. The cyclisation of this dipeptide fragment is then performed by LtxB, resulting in the attachment of the nitrogen of the N-methyl-L-valine residue to the indole moiety. The mechanism of this reaction is most simply rationalised by the epoxidation of the indole ring by LtxB, followed by nucleophilic attack of the N-methyl-L-valine amine to open the epoxide and subsequent dehydration to restore the aromaticity of the indole ring.⁸⁵  The reaction of LtxB with the substrate N-methyl-L-valine-L-tryptophanol has been performed in vitro with purified enzyme, where the (-)indolactam V product was identified. Additionally, the specificity of LtxB for the N-methyl-L-valine residue was explored through the use of reduced dipeptides with various residues in this position. The results of turnover experiments showed that LtxB can tolerate a number of aliphatic residues in this position: norleucine-, norvaline- and isoleucine-containing substrates are oxidised at around two-thirds of the efficiency of the natural substrate and a leucine-containing substrate oxidised at one-third of the efficiency of the natural substrate. A phenylalanine substitution at this position abolished activity of LtxB. The differences in oxidation are somewhat unusual due to the comparable binding strengths reported for the different reduced dipeptides to LtxB, although slight alterations in the binding mode for non-native substrates may well be sufficient to prevent productive enzymatic function. It is also possible that the low efficiency of the electron transfer system employed exacerbated the differences in product formation due to low product yields in all cases. One curious feature of LtxB is the presence of an N-terminal MbtH domain: these domains of 50–70 amino acids are commonly found encoded in biosynthetic gene clusters encoding non-ribosomal peptides, although LtxB is the first case of such a domain fused to a P450.⁸⁶  The function of MbtH domains has only recently been identified, where it has been shown that they are important for NRPS-mediated peptide synthesis through adenylation domain activation and NRPS di-domain dimerisation.^87,88  The possible role of the N-terminal MbtH domain in LtxB could therefore be to either simply maintain the P450 in close proximity to the NRPS machinery, or it could also be to control the rate of dipeptide formation dependent upon the presence of the P450.

Methylpendolmycin. The deep-sea bacterium Marinactinospora thermotolerans forms the indolactam-type antibacterial natural products methylpendolmycin and pendolmycin, which share the same core structure as the lyngbyatoxins discussed above.⁸⁹  Gene cluster analysis and inactivation has shown that the P450 encoded by mpnC is responsible for the formation of the indolactam core. The MpnC protein is highly homologous to LxtB (44% amino acid identity, 73% similarity) and also possesses an N-terminal MbtH-domain, albeit one that is slightly shorter than the one found in LtxB. The major methylpendolmycin product formed in M. thermotolerans differs in the structure of the reduced dipeptide P450 precursor, which is N-methyl-L-leucine-L-tryptophanol: this was shown to accumulate when mpnC was inactivated. From the difference in precursor structure, MpnC would appear to favour oxidation of dipeptide fragments containing N-Me-leucine instead of N-methyl-valine, although the selectivity may also be enforced due to the selectivity of the NRPS module responsible for incorporation of this amino acid. In vitro selectivity studies with LtxB confirm that both substrates are processed by LtxB, although no direct competition studies are available as of yet.

Himastatin. Another example of a P450-catalysed transformation equivalent to aromatic epoxidation followed by amine addition to afford a ring closed product is found in the biosynthesis of himastatin in Streptomyces himastatinicus.⁸²  Himastatin is a novel antibiotic compound that is composed of two cyclic depsipeptides joined by a biaryl linkage through a pyrroloindole moiety. The pyrroloindole, derived from L-tryptophan, has been shown by gene inactivation studies to be formed by the action of the P450 encoded by hmtT on the monomer of the cyclic depsipeptide. The mechanism of formation of the (2R,3aR,8aR)-3a-hydroxyhexahydropyrrolo-[2,3b]indole 2-carboxylic acid moiety is believed to follow a similar mechanism to that of LtxB and MpnC, with initial epoxidation of the tryptophan indole ring followed by attack of a nitrogen group to open the epoxide and form the pyrroloindole moiety: the major difference here is that the attacking nitrogen is an amide that links the neighbouring D-threonine residue and originates from the tryptophan itself. This results in the formation of a five-, rather than a nine-, membered ring as in the case of LtxB/MpnC. Additionally, there is no elimination of water to restore the aromaticity of the indole ring: rather the hydroxyl group produced from ring opening of the epoxide remains in the final pyrroloindole structure.

Kutznerides. The kutzneride antifungal and antibacterial agents are cyclic depsipeptides with a number of unusual residues isolated from Kutzneria sp. 744.⁹⁰  These include a pyrroloindole moiety, albeit with 2-chlorine substituents upon the 6-membered ring. The similarity of this moiety to that found in himastatin is striking, and the biosynthesis of the moiety is highly similar: a P450 encoded by the gene ktzM with high homology to that encoded by hmtT (63% sequence identity and 87% sequence similarity) has been identified in the kutzneride biosynthetic gene cluster. It is thus hypothesised that the synthesis of the dichloropyrroloindole moiety occurs via the epoxidation of a tryptophan moiety in the cyclic depsipeptide in a similar manner to that seen in himastatin.

Natural products with N-oxidised functionalities are less common. Yet more uncommon is the implication of P450s in these kinds of oxidation reactions which are – especially in prokaryotic organisms – more generally performed by flavin monooxygenases or non-haem iron-dependent dioxygenases.⁹¹  Nevertheless, some highly interesting examples have been delineated where P450s are involved in N-oxidations of amino acid-derived natural products.

Thaxtomins. The cellulose biosynthesis inhibiting thaxtomin phytotoxins are produced by plant-pathogenic Streptomyces species. They belong to the class of cyclic dipeptides (diketopiperazines) which are produced by cyclodipeptide synthases utilising tRNA-loaded amino acids in an ATP-dependent manner.⁹²  Thaxtomin is biosynthesised by the NRPS-catalysed condensation of L-4-nitrotryptophan and L-phenylalanine. The nonproteogenic amino acid L-4-nitrotryptophan is generated by direct nitration of the indole ring through the actions of P450 TxtE, utilising nitric oxide produced by the NO synthase TxtD (see Fig. 10A). This is highly unusual, as nitro groups are typically derived from oxidation of an amine. However, by feeding NO-releasing compounds to a txtD inactivated mutant Challis and coworkers could show utilisation of NO by TxtE for the oxidative nitration of L-tryptophan.⁹³  Using recombinant TxtE, the regio- and substrate-selectivity of the nitration reaction could be clarified (dissociation constant for L-tryptophan and TxtE: 60±6µM) and a mechanism with NO and O₂ as co-substrates has been proposed (see Fig. 10B).⁹³  Thus, in the biosynthesis of thaxtomin a unique and for P450s very uncommon oxidation reaction type has been realised. TxtE may therefore be an interesting candidate for expanding the scope of P450-catalysed reactions in biotechnological applications.

Pulcherriminin. In Bacillus subtilis the cyclic dipeptide cyclo-L-leucyl-L-leucyl (cLL) is oxidatively transformed by the P450 CypX (CYP134A1) into pulcherriminic acid, which is a precursor of the extracellular iron chelator pulcherriminin.⁹⁴  In vitro experiments with purified CypX revealed that CypX binds to cLL with a low micromolar affinity (dissociation constant: 24.5±0.5µM), with similar affinities determined for compounds with larger hydrophobic side chains.⁹⁵  It is probable that the high substrate specificity of the preceding cyclodipeptide synthase in B. subtilis reduces the need for P450-mediated substrate selectivity. The crystal structure of CypX revealed an interesting alteration of the I-helix, with the alcohol in the typically highly conserved acid/alcohol pair (responsible for correct protonation of reactive intermediates) replaced by a proline residue. The role of the missing alcohol residue may be fulfilled by a highly ordered active site water network or may not be necessary because of an alteration in the mechanism of oxygen activation due to the oxidative transformations performed by this P450.

Pulcherriminic acid formation involves a three-step oxidation with transformations of the diketopiperazine nitrogen atoms to the respective N-oxides and additionally aromatisation of the diketopiperazine ring. Aromatisation is proposed to occur either via hydroxylation and elimination of water or via an electron transfer reaction (see Fig. 11). In turnover experiments of cLL with CypX and a variety of different redox partners two potential singly oxidised intermediates were observed, showing that different oxidative pathways for CypX diketopiperazine oxidation potentially exist.⁹⁵ 

Nocardicin A. Nocardicin A, a monocyclic β-lactam antibiotic produced by the actinomycete Nocardia uniformis features a rare oxime functionality. Nocardicin A is built up from two hydroxyphenylglycine amino acids that are attached to the serine-derived β-lactam ring. Further, a homoserinyl side chain is attached to one of the aromatic side chains.⁹⁶  Townsend and colleagues identified NocL as the first prokaryotic P450 able to oxidise an amine to a syn-oxime and proposed a two-step mechanism via successive N-hydroxylations.^91,97  Following the hydroxylation steps, the neighboring amide bond could then facilitate elimination of water to generate a nitroso species and promote tautomerisation to the syn-oxime (see Fig. 11). Intramolecular hydrogen-bonding was also invoked to account for the greater abundance of syn-oxime in comparison to anti-oxime.⁹¹  Furthermore, in vitro studies performed by Townsend et al. demonstrated that NocL only converts nocardicin C to nocardicin A, suggesting the homoserinyl chain is crucial for NocL activity.⁹⁷ 

Cyanogenic glucosides. Cyanogenic glucosides are β-glucosides of amino acid-derived α-hydroxynitriles. With the exception of some arthropod clades, these compounds are produced by a large variety of plants and act as defense compounds by releasing hydrogen cyanide enzymatically when the organism sustains insect attack.⁹⁸  In general, the entire biosynthetic pathway of cyanogenic glucosides is encoded by just three genes, two encoding membrane-bound P450s and one a UDP-glycosyltransferase. It has been shown by in vitro experiments that the first P450 (a CYP79) acts as a multifunctional enzyme, catalysing two sequential N-hydroxylations of the amine of an amino acid followed by a dehydration/decarboxylation reaction step, thus producing an aldoxime as a free intermediate. The second P450 (a CYP71) performs two reactions, first catalysing the dehydration of the aldoxime to a nitrile and next hydroxylating the Cα-carbon to generate a cyanohydrin, which is then transformed into the glucoside (Fig. 12).^99,100  These cyanogenic glucosides are derived from several aromatic (phenylalanine/tyrosine) as well as aliphatic amino acids (isoleucine/leucine/valine). All members of the CYP79 family feature some unique amino acid substitutions resulting in an increase in polar and charged residues in the active site. According to Møller and colleagues these unusual substitutions may have evolved either due to the unusual catalytic activity of CYP79 enzymes or due to the high hydrophilicity of their amino acid substrates and intermediates compared to most other P450 substrates.^21,98 

Interestingly, the biosynthetic pathways to cyanogenic glucosides in plants and insects proceed using essentially the same intermediates with highly related enzymes. This is an interesting example of convergent evolution in two different kingdoms.⁹⁸ 

Of all the classes of chemical transformations affected by P450s upon amino acids or peptides, the coupling of aromatic side chains has been the most widely studied both in vivo and in vitro. This is due to the role that these transformations, catalysed by P450s, play in the biosynthesis of the glycopeptide antibiotics. These compounds, highly modified heptapeptides produced via the action of NRPSs, are still in clinical use and provide last-resort therapeutics with action against Gram-positive bacteria such as methicilin resistant S. aureus. Additional examples of phenolic and aryl cross coupling catalysed by P450s have been reported in the biosynthesis of other NRPS-synthesised natural products (arylomycin A2 and himastatin), in staurosporine biosynthesis and in the oxidation of cyclic dipeptides.

Structural classes of glycopeptide antibiotics. A very large number of different glycopeptide antibiotics have been identified. These have largely been assigned to one of five classes, which are determined by the crosslinking state of the amino acid side chains that form the peptide and the decorating groups found attached to the aglycone.¹⁰¹  From a perspective of P450-mediated aromatic coupling, there are three main classes: the vancomycin-type (three side chain cross links: AB biaryl coupling, C-O-D phenolic coupling and D-O-E phenolic coupling); the teicoplanin-type (four side chain cross links: vancomycin-type plus additional F-O-G phenolic coupling); and the complestatin-type (two side chain cross links: B-O-D phenolic coupling and DF biaryl coupling; additional A-O-C phenolic coupling in some cases).¹⁰¹  In all gene clusters for such compounds identified to date, there exists a number of P450s implicated in the formation of such cross links.¹⁰²  These proteins widely referred to as Oxy proteins (CYP165 family), exhibit high degrees of similarity across multiple gene clusters, which has made the assignment of predicted function much more straightforward. Comparison of the gene clusters of complestatin and the vancomycin type glycopeptides show a high similarity between P450s encoded by oxyB and comJ for example, whilst the other P450 (encoded by comI) has lower predicted homology to all other Oxy proteins. The product of comI expression can be assigned to the unusual biaryl tryptophan-hydroxyphenylglycine cross coupling found in the complestatin-type peptides.^103,104  Comparative analysis of the vancomycin- and teicoplanin-types shows that there is an additional P450 present in the teicoplanin-type gene clusters that is related to that encoded by oxyA, but more distantly than the true oxyA gene in the cluster.^{19,48,105,106}  Given the proximity of the additional ring in teicoplanin to that formed by OxyA (see below), the assignment of this new P450 to that responsible for the F-O-G ring catalysis becomes somewhat instinctive; this was later confirmed by inactivation experiments (see below).

Reaction order of glycopeptide antibiotic cross coupling P450s. In the biosynthesis of the vancomycin type glycopeptides, the P450s encoded by oxyA, oxyB and oxyC have been shown via inactivation studies in vivo to be responsible for the D-O-E, C-O-D and AB ring junctions respectively. Initially, the inactivation of the oxyA gene led to the isolation of heptapeptides bearing the C-O-D ring only:¹⁰⁷  this could be deduced from an analysis of the metabolites. Only linear peptides were found when oxyB was inactivated, indicating that the enzyme encoded by this gene was the first to act on the peptide and installing the C-O-D ring.⁴⁷  The oxyC inactivation experiments showed the presence of peptides bearing both C-O-D and D-O-E rings, thus allowing the assignment of the AB ring to OxyC and the D-O-E ring to OxyA. By combining the data from the inactivation of the oxy genes, the order of the P450s involved in peptide oxidation was revealed as OxyB, OxyA and finally OxyC.⁴⁷  Additional gene disruption experiments hinted that the peptide substrates for the Oxy proteins were still associated with the NRPS machinery, as disruption of elements of the NRPS caused the isolation of peptides with only the C-O-D ring installed.¹⁰⁸  A comparable inactivation study of the four P450s found in the gene cluster encoding a teicoplanin-like glycopeptide A47934 was also performed, with the isolation of the peptides produced by the inactivation mutants leading to an order of P450 action.¹⁰⁹  This again showed that the C-O-D ring is installed first by OxyB, with the remaining order of the last three enzymes not as easy to deduce as with the vancomycin type systems. The assignment of function to the P450s was clear however, with the AB ring again installed by OxyC, the D-O-E ring by OxyA and the F-O-G ring by the new P450 OxyE, which has its highest homology to OxyA.¹⁰⁹ 

In an elegant in vivo study, Süssmuth and coworkers engineered the NRPS machinery to incorporate an additional hydroxyphenylglycine (Hpg) residue into the peptide between the two normal Hpg amino acids and identified the new biosynthetic products.¹¹⁰  Metabolite analysis indicated that the NRPS now produced an octapeptide as had been intended, although the C-O-D ring was the only one installed in the peptide product (see Fig. 14 A). This indicated a somewhat broad peptide substrate tolerance for OxyB, a finding that has also been confirmed in vitro with shorter peptide substrates (see below). Despite the lack of in vitro data for OxyC, further gene inactivation studies have given some insights into the selectivity of this enzyme for alternative peptide substrates. Using strains where the formation of the final amino acid residue, dihydroxyphenylglycine (Dpg), was interrupted, feeding of modified Dpg residues (3-hydroxy, 3-methoxy, 3-hydroxy-5-methoxy, dimethoxy) indicated that these residues could all be incorporated into peptides that were then fully modified by the Oxy enzymes (see Fig. 14 B): this naturally implies that OxyC does not require the Dpg residue to be absolutely conserved to allow it to perform the AB biaryl ring closure, but rather that a single meta-oxygen substituent is sufficient for the enzyme to function.¹¹¹  An additional in vivo study with relevance for OxyC showed that the Dpg-knockout strain discussed above would produce very small amounts of a novel vancomycin-like aglycone in the absence of Dpg: this involved the incorporation of 4-hydroxyphenylglycine as the 7^th amino acid and the formation of a modified AB ring of 13 (rather than the standard 12) atoms (see Fig. 14 C).¹⁰⁸  Whilst this product was inactive as an antibiotic, it clearly demonstrated that the OxyC enzyme possesses some flexibility in its requirements for AB ring formation and gives hope for the future application of such P450s as catalysts in aglycone diversification.

The analysis of the complestatin-type peptide modification has only been performed by comparison to the other systems. This suggests initial formation of the B-O-D ring (OxyB like ComJ) followed by the (unique) formation of the DF biaryl system, presumably by the other P450 in the gene cluster, ComI.¹⁰³ 

Structural and functional characterisation of glycopeptide antibiotic cross coupling P450s. Initial attempts to investigate the Oxy proteins in vitro did not afford typical P450 binding spectra and the efficiency of enzymatic turnover was minimal, if present at all. With clues collected from both in vivo and in vitro sources, it was suggested that the Oxy proteins could well bind to the peptides not free in solution but rather still bound to an NRPS carrier protein domain.¹⁰⁸  One particular clue was the structural characterisation of the OxyB and OxyC enzymes from the vancomycin gene cluster: both structures revealed extremely open active sites, which would be expected for a substrate as large as a peptide-PCP.^16,112  The same open active site architecture was later seen also for the teicoplanin F-O-G ring forming enzyme, OxyE.^113,114  The investigation of OxyB from vancomycin has been undertaken extensively by the Robinson group, who initially showed excellent turnover of hexapeptide substrates upon both PCP-domains from the sixth and seventh amino acid incorporation modules from the corresponding NRPS (see Fig. 4).^115,116  PCP-domain seven was the best behaved in both binding and turnover experiments and the effects of peptide modification were explored in detail; this included establishing that OxyB had a wide range of tolerance for alternative peptide lengths, with peptides loaded on PCP-domain seven and composed of three, five, six and seven amino acids all being oxidised. Curiously, typical type-I P450 binding spectra could not be gathered for the PCP-loaded smaller peptide substrates (Fig. 15 A). The methylation or acetylation of the N-terminus of the peptide did not appear to effect binding and oxidation of the PCP-peptides by OxyB. OxyB was also able to bind and oxidise a PCP-loaded peptide substrate with the incorrect stereochemistry of the tyrosine residue at position six of the peptide (Fig. 15 B).^116–118  The ring formation process was also shown to require oxygen and recently the effect of substituents (β-hydroxylation, aromatic chlorination) on the tyrosine residue was also investigated (Fig. 15 C).^117,119 

To date, no further in vitro data has been obtained for any other Oxy protein. Mechanistically, a number of alternatives can be envisaged for P450-catalysed oxidative aromatic coupling – these either invoke hydrogen atom abstraction from the phenol group of the aromatic residues by the highly reactive Compound I species or epoxidation of the aromatic ring (similar to that invoked for the epoxidation of tryptophan previously discussed).¹²⁰  The intermediates following from these initial steps then would be one of the following: a diradical species, a geminal diol or an α-ketoether. In the final two cases, the phenolic cross-link is already established at this point of the reaction. Re-establishment of aromaticity or phenolic coupling of the diradical is then the final step to unify the pathways at the desired product. The mechanism of oxidative aromatic coupling in the biosynthesis of glycopeptide antibiotics has been investigated both in vitro by Robinson and in vivo by the group of Spencer. Use of an atmosphere of ¹⁸O₂ during the in vitro oxidation of PCP-peptides by OxyB showed that there was no incorporation of labeled oxygen into the cross linked peptide products.¹¹⁷  This argues against both mechanisms that invoke a geminal diol, unless the elimination of water from the diol occurs with complete stereoselectivity. The in vivo experiments of Spencer used the supplementation of cultures producing chloroerymomycin with 4-hydroxyphenylglycine (Hpg) that was labeled with deuterium atoms on the aromatic ring and possessed an ¹⁸O atom in the phenol group.¹²⁰  They could show that the product aglycone contained labeled Hpg residues with total retention of the phenolic oxygen, which indicated that there was no loss of the phenolic oxygen during the P450-catalysed cross linking reactions. This result was in agreement with the in vitro experiments for OxyB and indicates that the mechanism for OxyA and OxyC catalysed coupling follows a similar route with total retention of the oxygen atoms of the Hpg residues. This also argues against mechanisms invoking a geminal diol intermediate, suggesting a mechanism either involving diradical formation or formation of a keto-intermediate. The mechanisms can be differentiated by the requirements of both phenol rings to closely approach the haem: the diradical mechanism could well occur over a longer range through proton-coupled electron transfer (see later) and would not require both aromatic rings to closely approach the haem, whilst the keto-mechanism would require that both phenol groups are within direct hydrogen abstraction range of the activated haem. Future structural studies would no doubt be of use to assist in further unraveling the mechanism of aromatic coupling in glycopeptide biosynthesis.

Additional examples of P450-catalysed phenolic and aryl coupling in the biosynthesis of NRPS-produced peptides are found in the cases of arylcomycin A2 (Streptomyces roseosporus) and himastatin (Streptomyces hygroscopicus).

Arylomycin A2. Arylomycin A2 (Fig. 16) is a member of the arylomycin family of antibacterial agents, which exert their activity by targeting type I signal peptidases – highly conserved and essential proteins located on the extracellular side of the cytoplasmic membrane.¹²¹  The arylomycins are composed of a hexapeptide backbone modified by a long chain acyl group on the amino-terminus. The peptide is synthesised by an NRPS, with the acyl chain also incorporated into the peptide by a starter condensation domain.¹²²  In contrast to the glycopeptide backbones, the arylomycin peptide is mainly composed of residues with smaller side chains (Ala, Ser, Gly) and only two larger aromatic moieties at position four (L-Hpg) and six (L-Tyr). They form a fused biaryl structure in the final natural product, with gene disruption studies confirming the candidate for catalysing the aryl coupling as the P450 encoded by aryC.¹²³  This P450 has a relatively high amino acid homology to that of the glycopeptide P450 OxyC, which also catalyses biaryl coupling.¹²²  Additionally, with such a similarity in structure there is also a high likelihood that the P450 acts upon the NRPS-bound, rather than the free, peptide. This example is thus the first to extend glycopeptide-like P450-catalysed aromatic coupling activity to new NRPS-biosynthetic systems, indicating that yet more examples may well be waiting for discovery by future isolation or gene-sequencing based methods.

Himastatin. Himastatin consists of a cyclic hexadepsipeptide core that is dimerised via an biaryl linkage at C5 from the tricyclic hexahydropyrroloindole moiety (see Fig. 9).⁸²  The formation of the biaryl crosslink has been shown by gene inactivation to be performed by the P450 encoded by hmtS, with the detection of only cyclic hexadepsipeptide monomers in the knockout strain. HmtS is the first P450 shown to catalyse such a biaryl coupling in peptide-based systems, although the precedent for such biaryl cross coupling reactions with identical monomers has already been shown for the CYP158A1 and CYP158A2 enzymes from Streptomyces coelicolor.^124–126  These P450s catalyse the formation of isomers of biflaviolin and triflaviolin from flaviolin monomers and have been well characterised in vitro. The exact mechanism of the biaryl crosslinking reaction still remains somewhat unclear however, due to the apparent requirement for monomer molecules to rearrange in the active site of the P450, as postulated in current mechanistic proposals.¹²⁴ 

cYY. cYY is a cyclic dipeptide formed from two tyrosine residues via a cyclodipeptide synthase (CDS) utilising aminoacyl-tRNAs as substrates (Fig. 17), and is the main cyclic dipeptide produced by Mycobacterium tuberculosis. In M. tuberculosis, the gene encoding the P450 CYP121A1 is organised in an operon-like structure with a CDS that was shown to catalyse cYY formation, which in turn led to the investigation of substrate properties of cYY with CYP121A1.¹²⁷  As this P450 has been shown to be essential for viability, there was naturally an interest in characterising this P450 that is a potential target for azole inhibitors. Belin and coworkers were able to demonstrate both binding of cYY to CYP121A1 (dissociation constant: 21.3±3.5μM) and additionally the P450-catalysed oxidative transformation of cYY into a biaryl linked product P1, where the crosslink was established ortho to the tyrosine phenol groups.¹²⁷  Additionally, they were able to solve the crystal structure of CYP121A1 with cYY bound in the active site, which showed the diketopiperazine ring and the distal tyrosine side chain essentially perpendicular to the haem plane, with the proximal tyrosine side chain approaching to within 6 Å of the haem iron (the structure of the unbound form of the enzyme was previously solved by Munro and co-workers).¹²⁸  Curiously, aside from Van der Waals contacts, the interactions of cYY with the protein were limited to one direct hydrogen bonding interaction, and there was surprisingly little alteration to the water network coordinated to the haem iron. The rearrangement of the molecule during catalysis would seem to be required from this structure to bring both phenol groups within an appropriate distance to the haem to allow hydrogen abstraction; an alternative explanation would be that the substrate bound structure does not represent the final active conformation of CYP121A1. This interpretation is supported by the fact that there is still a water molecule bound to the haem iron in the substrate-bound form: the iron-bound water molecule seems to be present in all low temperature experiments but can be displaced in room temperature titrations. One final alternative would be that the mechanism of such biaryl coupling relies upon proton-coupled electron transfer from the substrate to the haem via the water network. This would remove the need for dramatic substrate motion during the P450 active cycle and may indicate why CYP121A1 is sensitive to temperature induced spin state changes due to the necessity to maintain such a water network for activity. Such a mechanism has been invoked for chromopyrrolic acid coupling via P450 StaP (see below).

In a recent study, the selectivity of CYP121A1 for alternative cyclic dipeptide substrates has been biochemically and structurally determined.¹²⁹  The binding properties of a range of cyclic dipeptides and mimics to CYP121A1 was tested, which revealed that only structures with two aryl side chains (cYF, cYW, cY-Dopa) displayed binding within the same regime as the natural cYY substrate. Additionally, CYP121A1-mediated oxidation of these alternative substrates was either very inefficient (cYF), very unspecific (cYW) or not able to catalyse the formation of a comparable phenolic crosslinked product (cY-Dopa). As the cyclodipeptide synthase associated with CYP121A1 also produces varying amounts of cyclic dipeptides (including cYF and cYW) in addition to cYY,⁹²  the selectivity of CYP121A1 for cYY indicates that the phenolic crosslinking of this dipeptide is the probable role for this enzyme also in vivo.¹²⁹ 

Staurosporine. Whilst technically not a dipeptide, the biosynthesis of the indolocarbazole antitumour agent staurosporine (and the structurally related rebeccamycin) involves the P450-catalysed biaryl coupling of chromopyrrolic acid, formed ultimately from two molecules of tryptophan via a two-step enzyme catalysed process.¹³⁰  The P450 StaP (CYP245A1) catalyses the biaryl coupling of the indole rings at C5, with the three aglycone products identified from in vitro turnover experiments using StaP shown to occur via subsequent and rate limiting non-enzymatic processes (decarboxylation and oxidation) from a common intermediate (Fig. 18).^131,132  The crystal structures of StaP in both the substrate bound and free forms have been solved, with the substrate bound form showing three molecules of chromopyrrolic acid bound to the P450: one in the active site, one somewhat removed from the active site and a third molecule bound at a distant beta-sheet.¹³³  The molecule bound in the active site appears to be fixed in a so-called twisted butterfly conformation, with significantly more hydrogen bonding interactions present than was the case with CYP121A1. Additionally, the mechanistic postulate (supported by computational studies) is that there is no rearrangement of the indole rings during catalysis, but rather that the enzyme initially forms an indole cation radical via proton coupled electron transfer (PCET).^134,135  This cation can equilibrate to the distal indole ring, allowing the second PCET to occur from the proximal indole ring, establishing the diradical species that then undergoes biaryl coupling. This mechanism is reminiscent of that seen in the case of cytochrome c peroxidase, with the separate “storage” of two oxidising equivalents – one in the iron-oxo species and the other in a tryptophan indole cation radical.^134,135 

Oxidation of amino acids and peptides is a very common and highly useful modification in secondary metabolism. On the one hand oxidation reactions are crucial for conformational maturation and conferring stability and specific biological characteristics to certain compounds, whilst on the other hand they diversify the limited range of the ribosomal amino acid code through post-translational modification. Many of these secondary metabolites find biological or medicinal applications as cell culture additives, antibiotics, antivirals or anti-tumor agents, which makes understanding their biosynthesis important for exploitation and future modification of these compounds.

Cytochomes P450 are one of the most versatile superfamily of oxidoreductases that catalyse natural oxidations. They are found in all kingdoms of life and perform an array of reactions on a wide range of substrates. Due to their highly reactive nature, P450s can catalyse very challenging reactions such as C-H bond activations, coupled with high degrees of stereo- and regiospecificity. Thus, investigating the role of P450s in the oxidation of amino acids and peptides is not of pure biological or medicinal interest but it is also target of biotechnological applications in industry. It is clear that the enormous diversity of secondary metabolites has not been fully explored to this point and that the discovery and mechanistic investigation of P450s involved in these biosynthetic pathways has just begun. Thus, further important and exciting findings in P450-catalysed oxidations of amino acids and peptides are to be expected.