Flavocytochrome P450 (cytochrome P450) BM3 is an intensively studied model system within the P450 enzyme superfamily, and is a natural fusion of a P450 to its P450 reductase redox partner. The fusion arrangement enables efficient electron transfer within the enzyme and a catalytic efficiency that cannot be matched in P450 systems from higher organisms. P450 BM3's potential for industrially relevant chemical transformations is now recognized, and variants with biotechnological applications have been constructed. Simultaneously, structural and mechanistic studies continue to reveal the intricate mechanistic details of this enzyme, including its dimeric organization and the relevance of this quaternary structure to catalysis. Homologues of BM3 have been found in several bacteria and fungi, indicating important physiological functions in these microbes and enabling first insights into evolution of the enzyme family. This short paper deals with recent developments in our understanding of structure, function, evolution and biotechnological applications of this important P450 system.
P450s (cytochromes P450) are oxygenases known for their catalytic versatility and the breadth of substrate selectivity exhibited across their enzyme superfamily . The vast majority of P450s catalyse reductive scission of molecular oxygen bound to their haem iron and subsequent formation of a molecule of oxygenated (frequently hydroxylated) substrate and a molecule of water . P450-catalysed regio- and stereo-selective oxygenation of molecules is exploited for numerous physiological roles, including steroid manufacture , modification of drugs and xenobiotics leading to excretion , bacterial synthesis of polyketide antibiotics  and production of lipid mediators of cellular function . Their central roles in human physiology and pharmacology are well known and there are 57 human P450s . However, determination of various prokaryotic genome sequences has led to the appreciation that e.g. actinobacteria, mycobacteria and bacilli have sizeable ‘CYPomes’ and exploit P450 chemistry for diverse purposes. For example, the human pathogen Mycobacterium tuberculosis encodes 20 P450s on its genome, and the model organism Bacillus subtilis has seven [8,9].
P450s require electron delivery for reductive activation of bound dioxygen, and in eukaryotes the redox partner is usually CPR (NADPH-P450 reductase), a diflavin enzyme that first catalyses hydride transfer from NADPH to FAD, then internal electron transfer to an FMN cofactor, and finally sequential delivery of two electrons from FMN to P450 haem iron . Eukaryotic P450s and CPRs (forming class II P450 redox systems) are almost invariably integral membrane proteins, whereas prokaryotic P450s (at least those characterized structurally) are cytosolic. Most prokaryotic P450s communicate with two redox partners: an NAD(P)H-binding and FAD-containing reductase (similar to the FAD-binding domain of CPR) and a ferredoxin or flavodoxin that shuttles electrons between the reductase and the relevant P450 (class I P450 systems) . However, analysis of fatty acid oxygenase activity in the soil bacterium Bacillus megaterium by Fulco's group in the 1980s revealed a major ‘outlier’ in the P450 superfamily [12–14]. P450 BM3 (formally CYP102A1) catalysed NADPH-dependent hydroxylation of several long-chain fatty acids at the ω–1 through ω–3 positions, but the activity was contained within a single 119 kDa polypeptide [12,13]. Cloning and analysis of the amino acid sequence of P450 BM3 indicated that the enzyme was formed by gene fusion of a P450 (N-terminal) to a CPR (C-terminal) and that neither of these domains possessed membrane anchor regions found in eukaryotic homologues . Thus BM3 was the first prokaryotic class II P450 system, and was a natural fusion of soluble P450 and CPR enzymes . The last two decades have seen intensive studies to determine structural and catalytic properties of P450 BM3, and its concomitant rise to prominence as a model system within the P450 superfamily. Atomic structures of its haem (P450) domain have been solved in the presence and absence of fatty acid substrate, revealing major conformational differences, and the FMN-containing (flavodoxin-like) domain was also structurally resolved [16–18] (Figure 1). Detailed mutagenesis studies revealed key roles for several amino acids in fatty acid and haem binding, electron transfer and control of substrate selectivity (see, e.g., [19–21]). However, BM3 still holds surprises and recent data have shed light on novel aspects of its structure and function, its catalytic mechanism and its biotechnological potential (see, e.g., ), as discussed below.
Structures of SF and SB P450 BM3 haem domains
P450 BM3: built for catalytic efficiency
BM3's fusion arrangement provides inherent advantages for efficiency of the system with respect to other P450s. The enzyme is catalytically self-sufficient, requiring only dioxygen, NADPH and fatty acids for activity. Recruitment of partner enzymes is not required . In addition, hydroxylase activity of BM3 with fatty acids is the highest reported for any P450 mono-oxygenase (>15000 min−1 with arachidonate), a testament to the efficiency of electron transfer reactions within the flavocytochrome . In particular, the rate of hydride transfer from NADPH to FAD (biphasic with a fast rate of ∼760 s−1 at 25°C) and of electron transfer from FMN to haem (>220 s−1 in the myristic acid-bound form at 25°C) are severalfold faster than for mammalian P450 redox systems . Mutagenesis studies have successfully identified residues critical for efficient reactivity with NADPH. These include Cys999, a component of a ‘catalytic triad’ of amino acids (with Ser830 and Asp1044 in BM3) important for NADPH binding and hydride transfer in the diflavin reductase enzyme family, which also includes mammalian CPR and NOS (nitric oxide synthase) (Figure 2) [23–25]. In the human cancer-related diflavin reductase NR1 (novel reductase 1), two of the three triad residues are absent, and NR1 has the lowest rate of FAD reduction of all diflavin reductases at approx. 1 s−1 . A spectacular switch of BM3 coenzyme specificity towards NADH was achieved by mutation of Trp1046 (to alanine or histidine), where the aromatic side chain of Trp1046 shields the FAD isoalloxazine ring and must be displaced by coenzyme as a prelude to hydride transfer [27,28]. In W1046A/H mutants, this barrier is completely or partially removed, enabling enhanced reactivity with NADH . Given the much lower cost of NADH compared with NADPH, this has important ramifications for exploitation of BM3 for biotechnologically relevant transformations, as discussed below.
The FAD-binding region of P450 BM3
In attempting to understand reasons for BM3's efficient electron transport system, the pivotal role of the FMN-containing domain must be considered. The FMN domain must communicate both with the FAD/NADPH-binding domain and with the haem domain, and the atomic structure of rat CPR (and latterly of neuronal NOS reductase) indicates that FAD-to-FMN electron transfer occurs directly between the cofactors in a conformation that is unlikely to be compatible with subsequent direct FMN-to-haem electron transfer [10,29]. Instead, movement of the FMN domain (in BM3 and related enzymes) must occur, allowing consecutive interactions of the FMN domain with its electron-donating and -accepting partner domains . Recent studies have demonstrated that a dimeric form of P450 BM3 is catalytically active as a fatty acid hydroxylase, with electron transfer occurring between the FMN domain of monomer 1 and the haem domain of monomer 2 . Thus the ‘swinging arm’ model for the FMN domain must be considered in light of these findings, and conformational dynamics may not be as dramatic as might be required to facilitate electron transfer from FAD-to-FMN-to-haem within a BM3 monomer.
Biotechnological applications come to fruition
For such an efficient P450 as BM3, there is the attractive possibility of active site engineering to produce variants capable of industrially or biotechnologically relevant transformations. There have been spectacular successes in this arena, using both rational mutagenesis (based on atomic structures of the haem domain) and forced evolution approaches to alter BM3 substrate selectivity, or to improve its catalytic efficiency with existing substrates. Among the most notable studies are the following. The double mutant L75T/L181K was generated with a view to promoting binding and catalysis with short-chain fatty acids, through creation of a binding motif for the substrate carboxylate group deep in the active site cavity. This variant considerably improved turnover with hexanoic acid and octanoic acid . Removal of the Arg47/Tyr51 motif at the mouth of the active site (which interacts with the carboxylate group of long-chain fatty acids) diminished catalytic efficiency with all substrates of chain length >C6, reinforcing the importance of this motif in initial interactions with substrates as they enter the BM3 active site. Other rational mutagenesis studies have improved BM3's activity with polycyclic aromatic hydrocarbons (A264G/R47L/Y51F)  and have produced a regio- and stereo-selective arachidonic acid 14S,15R epoxygenase . Conversion of specificity towards testosterone and drug-like molecules (e.g. dextromethorphan and amodiaquine) was achieved in an R47L/F87V/L188Q triple mutant . Random mutagenesis coupled with screening for desired activities (forced evolution) has also generated useful variants. Most notably, mutants conferring alkane hydroxylase activity on the P450 were generated [35,36]. In these variants, several mutations are distant from the substrate-binding regions and may impact on conformational properties of the P450.
There is some irony in the fact that BM3 has been so extensively engineered to achieve novel catalytic properties, while its true physiological role in Bacillus megaterium remains unclear. Functions such as detoxification of polyunsaturated fatty acids or metabolism of branched chain fatty acid components of the cell membrane have been postulated [37,38]. However, what is becoming increasingly clear from the burgeoning genomic databases is that there are several other BM3-like enzymes in other bacteria and lower eukaryotes.
The expanding CYP102 family
P450 BM3 stood alone as the only P450–CPR fusion enzyme for several years, until discovery and characterization of P450 foxy in the fungus Fusarium oxysporum in the mid 1990s. P450 foxy is a membrane-associated enzyme and is classified in a different gene family to BM3 (CYP505A1), although they are clearly evolutionarily related and are approx. 41 and 35% identical in amino acid sequence of P450 and reductase domains respectively . BM3 and CYP505A1 both catalyse subterminal hydroxylation of saturated fatty acids . The era of genome sequencing impacted on P450 science soon after the discovery of CYP505A1, and revealed the presence of BM3-like enzymes in numerous other microbes. In Bacillus subtilis, two homologues were identified, and their P450/CPR fusion composition was confirmed by expression of the CYP102A2 and CYP102A3 genes and characterization of the enzymes . CYP102A2/A3 catalyse fatty acid hydroxylation, but show different selectivity profiles to BM3 and strong activity with branched chain fatty acids prevalent in the Bacillus subtilis membrane. Moreover, co-operative equilibrium binding of selected substrates and atypical kinetics suggestive of co-binding of two substrate molecules were observed for CYP102A2/A3, indicating important differences in catalytic properties to BM3 . Homologues are recognized in other bacilli, and in numerous other bacteria, including: Bradyrhizobium japonicum, Streptomyces avermitilis, Burkholderia cepacia, Rhodopseudomonas palustris and the heavy metal-tolerant Ralstonia metallidurans. BM3-like enzymes are also found in fungi such as Aspergillus nidulans and Neurospora crassa.
P450 BM3 and its homologues that have been characterized enzymatically were shown to possess fatty acid hydroxylase activity. However, perhaps not all the BM3-like flavocytochromes are fatty acid oxygenases. For instance, the BM3-like enzyme from the actinomycete Actinosynnema pretiosum is in a gene cluster directing synthesis of the antitumour agent ansamitocin. This molecule is a macrocyclic lactam, and the 1005-amino-acid BM3-like enzyme (Asm30) is postulated to perform an epoxidation reaction in the final stages of ansamitocin synthesis .
An alignment of amino acid sequences of 15 bacterial BM3-like P450–CPR fusion enzymes demonstrates strong conservation of several residues shown to play important catalytic roles in BM3. These include Phe87 (a regulator of regioselectivity of substrate oxygenation), Phe393 (pivotal in control of haem redox potential and reactivity with oxygen) and the catalytic triad residues (Ser830, Cys999 and Asp1048) in the reductase domain that are important for efficient coenzyme binding and hydride transfer from NADPH [23,33,43] (Figure 2). A large section of the BM3 haem domain I helix is also completely conserved. Interestingly, the Arg47/Tyr51 motif proven important (both structurally and enzymatically) for binding the carboxylate group of fatty acid substrates is not retained in any of the bacterial homologues analysed. Phylogenetic analysis of this grouping also demonstrates that the Bacillus CYP102 enzymes are evolutionarily divergent from BM3-like enzymes in the other bacteria, perhaps reflecting adaptation for different physiological roles. Moreover, BM3 (CYP102A1) appears to be somewhat distinct from the other Bacillus enzymes and to have diverged at an early stage in the evolution of this group (Figure 3).
Evolutionary analysis of BM3 and related flavocytochromes P450
Structural features of P450 BM3
Atomic structural studies indicate that a major reorganization occurs between SF (substrate-free) and SB [substrate (fatty acid)-bound] forms of BM3 haem domain (Figure 1). An interesting question relates to whether this transition is induced by substrate binding itself or simply reflects different conformational states accessible by the P450. Our studies of an A264E mutant (in which Glu264 replaces water as distal ligand to the haem iron) revealed that both SF and fatty acid-bound forms of A264E haem domain crystallize in the conformation observed for the SB form of wild-type haem domain . In addition, the A264E variant (from optical titrations) binds fatty acids tighter than does wild-type BM3 . Thus an intriguing possibility is that the SB and SF (and perhaps other) conformations of the haem domain are in dynamic equilibrium, and that the SB conformation has higher affinity for substrate than the SF form .
The atomic structure of the haem–FMN domain of BM3 (residues 1–649 of the 1048-amino-acid protein) confirmed the FMN domain's flavodoxin-like fold, and also provided a firm structural basis for inability of the protein to stabilize the blue semiquinone form of FMN . Blue semiquinones are a hallmark of virtually all flavodoxins, but the FMN semiquinone is thermodynamically destabilized in favour of the hydroquinone in BM3, with one electron reduction of the FMN domain resulting in transient formation of the red (anionic) semiquinone [46,47]. Reasons underlying the unusual redox behaviour of the BM3 FMN domain probably include the proximity to the FMN of two lysine residues (Lys572/Lys580) and possibly a protonated histidine residue (His539). Also, a rigid FMN binding ‘loop’ region in BM3 FMN domain may be unable to relocate in the one electron reduced form in order to form hydrogen-bonding interactions that stabilize a neutral semiquinone. Such conformational changes and stabilizing interactions are observed in bacterial flavodoxins .
The final ‘building block’ in the P450 BM3 structure is the FAD/NADPH-binding domain, and we have recently determined the structure of this domain. The structure confirms close homology to the relevant domain of rat CPR. In the NADP(H)-binding site, the side chain of Trp1046 shields the FAD isoalloxazine ring, and the catalytic triad of Ser830, Cys999 and Asp1044 are positioned to assist in NADPH binding and catalysis of flavin reduction . In the absence of a structure for intact flavocytochrome P450 BM3, the fact that structures for all its component domains are now available will facilitate construction of realistic and experimentally tractable models to describe the organization of the catalytically relevant dimeric state of the enzyme .
Major questions remaining
P450 BM3 is a paradigm system in the P450 superfamily. It continues to attract interest as a model for studies of general P450 mechanism and for exploitation as a biocatalyst. A full-length structure of the BM3 flavocytochrome is still to be determined, and crystallography is complicated by its large multidomain construction, with flexible domains. The fact that the protein has a quaternary (dimeric) structure and that there may be monomer, dimer and possibly higher oligomers in solution provides a further headache in attempts to obtain crystals . Routes to solving this problem may include restriction of BM3's conformational mobility and prevention of non-specific inter- and intra-molecular interactions that give rise to heterogeneity in solution. Mutations that ‘lock’ the haem domain conformation or which prevent formation of non-specific disulfide bridges are avenues we are investigating as a means of isolating flavocytochrome P450 BM3 crystals [44,45]. However, it should be noted that conformational changes drive catalysis and electron transfer in this system, and the understanding of these movements and their regulation is another key area for future study. A final major target for BM3 relates to its exploitation as a biocatalyst and addressing the perception that P450s are ‘fragile’ enzymes. With recent demonstrations that BM3-related CYP4 enzymes have haem covalently linked via an acidic I helix residue, an attractive mutagenic route to stabilizing BM3 haem binding is opened up . Similarly, the weak binding of FMN in BM3 (by comparison with e.g. flavodoxins) could be addressed by protein engineering strategies. A quarter of a century after Armand Fulco's discovery, research on the BM3 system continues at pace and the catalytic potential of the enzyme is now truly being realized.
8th International Symposium on Cytochrome P450 Biodiversity and Biotechnology: Independent Meeting held at Swansea Medical School, Swansea, Wales, U.K., 23–27 July 2006. Organized and Edited by D. Kelly, D. Lamb and S. Kelly (Swansea, U.K.).
We acknowledge financial support from the U.K. Biotechnology and Biological Sciences Research Council (grants C15314, C19757 and BBS/B/06288) and the EU (grant NM4TB; code: 01893).