An extraordinary array of P450 (cytochrome P450) enzymes are encoded on the genome of the human pathogen Mycobacterium tuberculosis (Mtb) and in related mycobacteria and actinobacteria. These include the first characterized sterol 14α-demethylase P450 (CYP51), a known target for azole and triazole drugs in yeasts and fungi. To date, only two Mtb P450s have been characterized in detail: CYP51 and CYP121. The CYP121 P450 shows structural relationships with P450 enzymes involved in synthesis of polyketide antibiotics. Both P450s exhibit tight binding to a range of azole drugs (e.g. clotrimazole and fluconazole) and the same drugs also have potent effects on growth of mycobacteria (but not of e.g. Escherichia coli). Atomic structures are available for both Mtb CYP51 and CYP121, revealing modes of azole binding and intriguing mechanistic and structural aspects. This paper reviews our current knowledge of these and the other P450 systems in Mtb including recent data relating to the reversible conversion of the CYP51 enzyme between P450 (thiolate-co-ordinated) and P420 (thiol-co-ordinated) species on reduction of the haem iron in the absence of a P450 substrate. The accessory flavoprotein and iron–sulfur proteins required to drive P450 catalysis are also discussed, providing an overview of the current state of knowledge of Mtb P450 redox systems.
The cytochromes P450 (P450s; CYPs) are haem b-binding oxygenases with widespread roles in Nature . Their ability to catalyse reductive scission of molecular oxygen and to incorporate an atom of oxygen into substrate, often with exquisite regio- and stereo-selectivity, has enabled their exploitation in diverse biological pathways. The roles of P450s in mammals have long been known, but P450s are also widespread in prokaryotes, with mycobacteria and related actinobacteria having a particularly high ‘P450 density’ in their genomes (see e.g. [2,3]). In various species of Streptomyces, for instance, P450s function in synthesis of antibiotics such as pikromycin and nystatin [4,5]. However, in mycobacteria, the roles of the P450s are currently less well understood. In the human pathogen Mycobacterium tuberculosis (Mtb), there are 20 P450 enzymes encoded throughout the 4.4 Mb genome, and structural and mechanistic studies of some of these isoforms and redox partner proteins are beginning to shed light on catalytic properties and possible cellular function . Two Mtb P450s have been structurally characterized to date: CYP51, the first characterized bacterial sterol 14α-demethylase [7,8], and CYP121. The CYP121 isoform has been structurally resolved to high resolution (1.06 Å; 1 Å=0.1 nm) and has highlighted novel structural properties of P450s . The present paper contextualizes our current knowledge of the structure and function of these two enzymes, their interactions with substrates, ligands and redox partners and potential roles of these and other P450s and Mtb redox partner systems.
Mtb and the crisis of drug resistance
Once considered virtually eradicated as a leading cause of human mortality in the developed world, Mtb has made a terrifying return as the major bacterial cause of human death worldwide . There are two major reasons underlying this resurgence. The first revolves around the spread of HIV AIDS across the globe and the ability of Mtb to thrive in immune-compromised individuals. The second is the frightening development of drug resistance in Mtb that has occurred in the past few decades, with strains resistant to all frontline and second choice antibiotics now characterized . For instance, resistance to the leading drug isoniazid [INH (isonicotinic acid hydrazide)] is mediated by mutations in the catalase/peroxidase enzyme encoded by the katG gene . These inactivate the enzyme with respect to reductive activation of INH to a reactive radical and in turn the INH (isonicotinoyl) radical does not react non-enzymatically with free NAD(P)+ and the reaction products cannot inactivate target enzyme(s) including enoyl-ACP (acyl-carrier protein) reductase (InhA; essential for mycolic acid synthesis) and hydrofolate reductase (essential for nucleic acid synthesis) [13,14]. The leading drug rifampicin kills Mtb by binding to the RNA polymerase β-subunit (RpoB) and inhibiting mRNA synthesis. Mutations to the rifampicin-binding site (located in the main channel for double-stranded DNA entry into the polymerase) confer resistance (see e.g. [15,16]).
The scourge of drug resistance highlights the absence of new drug development for Mtb infections in recent years. The problems of Mtb drug resistance also came to prominence to coincide with determination of the genome sequence of Mtb strain H37Rv, which served to emphasize the large proportion of the genome given over to production of enzymes involved in lipid metabolism . This, in turn, reinforced importance of lipid metabolism to the viability of Mtb, which produces an extraordinary array of lipid molecules . A further revelation from the Mtb genome was the fact that 20 P450 enzymes were encoded, along with several ferredoxins and putative ferredoxin reductase flavoproteins that could act as redox partners for the P450s [2,18]. At the time, this was the largest P450 complement seen for a bacterium, although subsequent genome sequences for various mycobacteria and related actinobacteria demonstrated that these genera typically have substantial P450 complements (see e.g. ). Among the Mtb P450s was the first prokaryotic representative of the sterol 14α-demethylase P450 family, a target enzyme for azole and triazole drugs used as antifungals to inhibit synthesis of ergosterol in pathogens such as Candida albicans . This finding highlighted the possibility that this drug class might have activity against mycobacteria and this has indeed been demonstrated in preliminary studies on Mtb, M. smegmatis and related actinobacteria (see e.g. [20,21]). The characterization of the potential P450 target enzymes for these agents is clearly a priority, and to date two of the Mtb P450s have been studied in detail, revealing intriguing structural and catalytic properties.
CYP121: P450 structural detail at unprecedented levels
A problem encountered in the initial selection of Mtb P450 systems for expression and characterization was the lack of similarity, in most cases, to other P450s of defined function. As a result, substrate selectivity is difficult to predict from first principles for all but two or three of the Mtb P450s. The P450 encoded by the Mtb Rv2276 gene (CYP121) has approx. 30% amino acid sequence identity to characterized bacterial P450s involved in sulfonylurea herbicide degradation (Streptomyces griseolus CYP105A1) and in synthesis of the antibiotic erythromycin (Saccharopolyspora erythraea P450 eryF; CYP107A1) [22,23]. CYP107A1 has high affinity for, and has been structurally characterized as bound to, the azole antifungal ketoconazole . CYP121 was one of the tranche of Mtb P450s selected for expression and biochemical/structural analysis in our laboratory and has proven an important enzyme that has revealed novel aspects of P450 structure. In preliminary work, CYP121 was expressed and purified in an E. coli system. As with many Mtb enzymes expressed in E. coli, the bulk of the protein was found to be insoluble on lysis of transformant cells grown at 37°C. However, it was noted that insoluble enzyme was deep red in colour and had obviously incorporated haem. Thus it appeared likely that a fully folded holoprotein was produced in E. coli (as opposed to a misfolded ‘inclusion body’) and that intact CYP121 aggregated at this growth temperature and/or adhered to cellular membranes. Alteration of transformant growth and induction conditions effected a spectacular improvement in production of soluble CYP121, with growth at low temperature (∼20°C) and mild gene induction being major factors promoting soluble CYP121 production .
Purified CYP121 demonstrated spectroscopic properties typical of P450s using techniques such as UV–visible absorption, CD, EPR and resonance Raman spectroscopy. CYP121 was shown to contain cysteinate-co-ordinated haem b, which was predominantly low-spin in the resting form of the enzyme . As predicted, CYP121 showed high affinity for several azole antifungal drugs, as demonstrated by optical changes on azole binding, with a red shift of the major (Soret) haem band from ∼417 nm to 424 nm. In particular, clotrimazole, econazole and miconazole bound with Kd values <0.2 μM (Figure 1). Parallel studies of effects of these and other azoles on laboratory strains of mycobacteria and related actinomycetes (e.g. S. coelicolor) showed that these azoles had potent inhibitory effects on bacterial growth, but that growth of E. coli (itself devoid of P450s) was almost completely unaffected . Recent studies have confirmed the inhibitory effects of azoles on growth of Mtb itself (K.J. McLean, unpublished work).
Azole binding to Mtb CYP121
The atomic structure of CYP121 was solved to 1.06 Å, allowing the first truly atomic level description of a P450. The structure revealed closest similarity to CYP107A1, although there are marked differences in active site construction. The CYP121 active site cavity is large (1337 Å3), but constricted around the haem site and dominated by Ser237, which hydrogen-bonds to a water molecule that is the sixth ligand to the haem iron and to the Arg387 side chain. Ser237 also hydrogen-bonds to the carbonyl backbone of Ala233. The presence of Arg387 in the immediate vicinity of the haem is relatively unusual for P450s and suggests an anionic substrate or functional group recognized . The constrained environment around the haem raised interesting questions relating to how bulky azoles could co-ordinate with the haem iron. Recently, we have solved the atomic structure of a CYP121–fluconazole complex, which indicates a novel mode of haem iron binding achieved via a bridging water molecule (H.E. Seward, A. Roujeinikova, K.J. McLean, A.W. Munro and D. Leys, unpublished work). The CYP121 structure demonstrated unequivocally that haem could be bound in P450 enzymes in two orientations, related by a 180° ‘flip’ about the CHα-Fe-CHδ axis of the haem macrocycle . The haem is also severely kinked at one of the pyrrole rings (by ∼30° towards the distal side) by the side chain of Pro346, the residue immediately following the cysteinate ligand to the iron (Figure 2). The significance of the haem distortion is uncertain, but it may impact on redox potential of the haem iron. Proline residues are found in other P450s in the same position (e.g. plant CYP71A and B families, and eukaryotic CYP7A and 7B enzymes involved in oxidation of cholesterol and of dehydroepiandrosterone and related steroids) (see e.g. [25,26]) and in Mtb P450s CYP141 and CYP128. CYP128 is of particular interest as an Mtb P450 that appears essential for strong growth of the organism in culture .
CYP121 active site and haem distortion
While CYP121 structural data did not provide clear pointers to the type of substrate molecules recognized, studies of binding to Mtb lipid fractions do indicate CYP121 has affinity for some of the complex lipids produced by Mtb (K.J. McLean, unpublished work). Accurate structural detail also identifies CYP121 residues that probably define a relay system for delivering protons to haem iron during oxygen scission . While substrate selectivity remains to be defined for CYP121, both substrate class recognized and atomic structure are known for another Mtb P450: the sterol demethylase CYP51.
CYP51: structural plasticity, azole binding and reversible inactivation
The CYP51 sterol demethylase family is one of the most evolutionarily ancient of the P450 superfamily, and in fungi is the major target for the azole group of antibiotics . The discovery of a CYP51 gene in Mtb provided the first evidence that CYP51 was present in prokaryotes . Subsequently, CYP51s have been identified in a number of other mycobacteria and actinobacteria (e.g. S. coelicolor) . The Mtb CYP51 was purified and an unusual instability of its carbon monoxy complex was observed . The Fe(II)CO complex provides the spectroscopic hallmark by which P450s are defined, with the haem Soret band shifting to approx. 450 nm. For Mtb CYP51, formation of the P450 species is followed by its collapse over minutes to one with Soret maximum close to 420 nm: the P420 species (Figure 3). P420 has long been known as an inactive state of P450, formation of which can be induced by e.g. denaturants and high-pressure treatment (see e.g. ). However, spectroscopic studies indicate that protonation of the thiolate ligand to the haem iron (as opposed to major structural disruption with loss of haem co-ordination) underlies P420 formation in many cases . In Mtb CYP51, P420 collapse is dramatically retarded by binding the substrate-like molecule oestriol, and in P450 EpoK the formation of P450 from P420 can actually be induced by binding the substrate epothilone D . In Mtb CYP51, complete reoxidation of the P420 Fe(II)CO complex to the Fe(III) species enables reformation of P450, demonstrating the reversibility of P420 formation and its reconversion into P450 on reoxidation. Anaerobic spectral studies of ferrous CYP51 suggest that thiol co-ordination develops on haem reduction, with (in this case at least) formation of the P420 CO complex being simply a marker for a protonation and inactivation process that occurs in the ferrous substrate-free enzyme .
Collapse of the Mtb CYP51 P450 complex
The atomic structure of Mtb CYP51 was solved in complex with the azole inhibitor fluconazole, demonstrating direct co-ordination of haem iron by a triazole nitrogen atom . Subsequently, structures of the CYP51 were solved in the presence of the substrate-like molecule oestriol and in the absence of ligand . These structures proved that a major structural reorganization accompanies CYP51 substrate binding. Importantly, comparative modelling of the positions of azole resistance-conferring mutations in fungal CYP51s indicates that they map mainly to conformationally dynamic sites in Mtb CYP51 (including the C helix and loops connecting B/C and G/H helices), rather than to active site regions defining binding of substrates/ligands . This provides important clues as to how fungal resistance to azoles may develop without compromising substrate interactions. Thus mutations that alter CYP51 protein dynamics may diminish affinity for azole drugs through preventing closure of a binding site, while still enabling such closure on binding substrate(s) .
Turnover studies with sterols (including dihydrolanosterol and the plant sterol obtusifoliol) proved that Mtb CYP51 was a bona fide sterol 14α-demethylase and that activity was supported by both heterologous and homologous redox partners (see below). However, gene knockout analysis indicated that CYP51 was not a gene required for optimal Mtb growth, and a full sterol biosynthesis pathway is not recognized in Mtb . Thus a full understanding of CYP51's physiological function and relevance to pathogenesis has yet to be developed. Potentially, this involves interactions with host sterol molecules.
The Mtb redox partner apparatus
Most P450s do not function in isolation and require redox partners for electron delivery to facilitate substrate oxygenation. In classical bacterial ‘class I’ systems, ferredoxins are electron donors to P450s and receive their electrons from NAD(P)H via FAD-binding reductases . In Mtb, several ferredoxins are recognized by genome analysis and four are of particular interest in the context of P450 chemistry and Mtb biology. The ferredoxins encoded by genes Rv1786 and Rv0763c are immediately adjacent to P450s (CYP143 and CYP51 respectively) and are probable redox partners. Fdx (product of Rv0763c) has been shown to encode a 3Fe-4S ferredoxin and to transfer electrons to CYP51 . While ferredoxins FdxA (Rv2007c) and FdxC (Rv1177) have not been characterized, the genes were shown either to be required for optimal Mtb growth (FdxC), or to be induced by cellular stresses including those mimicking ones Mtb may encounter in the infective state (hypoxia and low pH) [36,37]. With respect to reductase systems, the adrenodoxin reductase-like FprA has been structurally and catalytically characterized and shown to support CYP51 catalysis [38,39]. Another flavoprotein reductase, FdrA (Rv0688), was also shown to drive Mtb CYP51 sterol demethylation . Thus functional partner proteins for Mtb P450s have been established and priorities now lie in the area of establishing roles of P450s in Mtb physiology and pathogenicity.
Structural analysis of Mtb CYP51 and CYP121 has provided important data furthering our understanding of P450 structural features and conformational dynamics. Priorities for these systems are determination of biochemical functions in vivo, particularly given the demonstration that both enzymes have high affinity for azole antibiotics, and that antimycobacterial and antitubercular, properties of azoles have been established [7,18,20,21]. A drawback considered with use of azoles as antituberculars is cross-reactivity with human P450s (see e.g. ). However, systemically tolerated azoles display strong affinity for Mtb P450s such as CYP51, CYP121 and CYP132. Defining a single Mtb P450 target for such drugs will be difficult, but this in itself may be an advantage to their use. If two or more azole-binding P450s are pivotal to Mtb viability and/or pathogenicity, then development of azole drug resistance will be dramatically diminished. Given the sad recent history of Mtb resistance to other antibiotics, surely this is a compelling reason to characterize in greater detail the Mtb P450s and their interactions with azoles.
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).