|Home | About | Journals | Submit | Contact Us | Français|
Cofactor F420, a 5-deazaflavin involved in obligatory hydride transfer, is widely distributed among archaeal methanogens and actinomycetes. Owing to the low redox potential of the cofactor, F420-dependent enzymes play a pivotal role in central catabolic pathways and xenobiotic degradation processes in these organisms. A physiologically essential deazaflavoenzyme is the F420-dependent glucose-6-phosphate dehydrogenase (FGD), which catalyzes the reaction F420 + glucose-6-phosphate → F420H2 + 6-phospho-gluconolactone. Thereby, FGDs generate the reduced F420 cofactor required for numerous F420H2-dependent reductases, involved e.g., in the bioreductive activation of the antitubercular prodrugs pretomanid and delamanid. We report here the identification, production, and characterization of three FGDs from Rhodococcus jostii RHA1 (Rh-FGDs), being the first experimental evidence of F420-dependent enzymes in this bacterium. The crystal structure of Rh-FGD1 has also been determined at 1.5 Å resolution, showing a high similarity with FGD from Mycobacterium tuberculosis (Mtb) (Mtb-FGD1). The cofactor-binding pocket and active-site catalytic residues are largely conserved in Rh-FGD1 compared with Mtb-FGD1, except for an extremely flexible insertion region capping the active site at the C-terminal end of the TIM-barrel, which also markedly differs from other structurally related proteins. The role of the three positively charged residues (Lys197, Lys258, and Arg282) constituting the binding site of the substrate phosphate moiety was experimentally corroborated by means of mutagenesis study. The biochemical and structural data presented here provide the first step towards tailoring Rh-FGD1 into a more economical biocatalyst, e.g., an F420-dependent glucose dehydrogenase that requires a cheaper cosubstrate and can better match the demands for the growing applications of F420H2-dependent reductases in industry and bioremediation.
The unusual cofactor F420, a 7,8-didemethyl-8-hydroxy-5-deazariboflavin, was originally discovered in various archaea (Cheeseman et al. 1972) (Fig. (Fig.1).1). It was demonstrated that in both methanogenic and nonmethanogenic archaea, F420 represents a central catabolic redox cofactor involved in the oxidation of energy sources (e.g., H2 and formate) (Jacobson et al. 1982; Vitt et al. 2014; Tzeng et al. 1975a; Wood et al. 2003) and the reduction of cofactors such as NADP+ and tetrahydromethanopterin (Tzeng et al. 1975b; Warkentin et al. 2001; Hartzell et al. 1985; Aufhammer et al. 2005). In recent years, it has become clear by genome sequence analyses and biochemical studies that the deazaflavin cofactor is also utilized by numerous enzymes in actinobacteria, including Mycobacterium tuberculosis (Mtb)—the notorious causative agent of tuberculosis (Daniels et al. 1985). In actinomycetes, F420 was found to be involved in several important processes such as biosynthesis of antibiotics in Streptomyces spp. (e.g., tetracycline, lincosamide, and aminoglycoside) (Wang et al. 2013; Coats et al. 1989; Li et al. 2009a), degradation of coumarin derivatives (e.g., carcinogenic aflatoxins) (Taylor et al. 2010; Lapalikar et al. 2012b; Ahmed et al. 2015), and other aromatic compounds (e.g., picrate and related compounds) (Ebert et al. 1999; Heiss et al. 2002; Jirapanjawat et al. 2016). For mycobacteria, there is a compelling evidence that F420 is essential to render the bacilli persistent in hostile and challenging environments, such as anaerobic conditions, and oxidative and nitrosative stress (Hasan et al. 2010; Gurumurthy et al. 2013; Purwantini and Mukhopadhyay 2009). Interestingly, in vivo activation of the novel antitubercular nitroimidazole prodrugs—such as pretomanid (PA-824), delamanid (OPC-67683), and TBA-354—strictly requires a selective reduction of these prodrugs facilitated by an F420H2-dependent reductase (Stover et al. 2000; Matsumoto et al. 2006; Denny 2015). Owing to the newly discovered mode of action, these nitroimidazole compounds are highly promising as they exhibit no cross-resistance with the current front-line antitubercular drugs in vitro and even exert activity on non-replicating tubercle bacilli (Stover et al. 2000; Matsumoto et al. 2006; Singh et al. 2008). Delamanid (OPC-67683) was recently clinically approved for multidrug-resistant tuberculosis whereas pretomanid (PA-824) and TBA-354 are currently in phase III and phase I clinical trials, respectively (Tasneen et al. 2015).
The discovery of this novel antimycobacterial class of drugs is attracting an increasing interest in F420-dependent enzyme research (Taylor et al. 2013). Due to the unique redox potential (−340 mV) of F420, which is lower than that of FAD (−220 mV) and even of the classical hydrogen carrier NAD(P)+ (Jacobson and Walsh 1984; de Poorter et al. 2005), F420H2-dependent enzymes are capable of catalyzing hydrogenation of a wide range of organic compounds which are otherwise recalcitrant to reductive activation such as enones (Taylor et al. 2010; Lapalikar et al. 2012b; Lapalikar et al. 2012a) and imines (Coats et al. 1989; Li et al. 2009a; Li et al. 2009b) in various heterocycles (Schrittwieser et al. 2015). These enzymes thus hold the promise of being highly valuable in industrial biotechnology and bioremediation, and can be exploited as a complement to the available toolboxes for asymmetric chemical synthesis (Taylor et al. 2013; Greening et al. 2016; Ney et al. 2016).
As most bacterial F420-dependent enzymes are involved in catalyzing reductions, several F420-dependent dehydrogenases have evolved with the purpose to maintain a cytosolic reservoir of reduced F420 (F420H2). In mycobacteria and other actinomycetes, glucose-6-phosphate dehydrogenases (FGDs) seem to be the main producer of F420H2, by catalyzing the reaction F420 + glucose-6-phosphate → F420H2 + 6-phosphogluconolactone (Fig. (Fig.1).1). FGD was first identified in Mycobacterium smegmatis and subsequently in other actinomycetes, including Mtb (Purwantini and Daniels 1996; Purwantini et al. 1997; Purwantini and Daniels 1998). Since the identification of the first FGD two decades ago in Daniels’ lab (Purwantini and Daniels 1996), only two FGDs from actinomycetes, namely M. smegmatis and Mtb, have been characterized in detail (Bashiri et al. 2007; Bashiri et al. 2010). These two FGDs share 37% sequence similarity and belong to an F420-dependent enzyme subgroup within the luciferase-like hydride transferase family. The affinity of both enzymes for F420 and glucose-6-phosphate (G6P) falls in a comparable range, facilitating the release of the resulting reduced cofactor to be consequently exploited by downstream F420H2-dependent enzymes. Heterologous expression in Escherichia coli of both FGDs was found to be troublesome, often resulting in formation of inclusion bodies. Structural characterization of an FGD from Mtb has been recently described (Bashiri et al. 2008).
Rhodococci are high G+C content, Gram-positive aerobic, non-sporulating actinomycetes of high biotechnological and environmental importance due to their ability to catalyze an array of unique enzymatic reactions (van der Geize and Dijkhuizen 2004). A recent bioinformatic study suggested that Rhodococcus jostii RHA1 is among the actinomycetes that carry the largest number of F420-dependent enzymes. It was predicted to possess at least 104 deazaflavoenzymes (Selengut and Haft 2010). Nevertheless, up to date, there is no experimental evidence for the presence of deazaflavoenzymes in R. jostii RHA1. Therefore, in this work, we aimed to (1) verify the existence of FGDs in R. jostii RHA1 (Rh-FGD) by heterologous expression of putative FGD-encoding genes in E. coli, (2) characterize the catalytic properties of the identified enzyme(s), and (3) obtain and analyze the crystal structure of a Rh-FGD.
R. jostii RHA1 was grown in lysogeny broth (LB) at 30 °C; after which, genomic DNA was extracted using the GenElute Bacterial Genomic DNA kit from Sigma. Three putative fgd genes, RHA1_RS43115, RHA1_RS10755, and RHA1_RS43570, were amplified from R. jostii RHA1 genomic DNA using Phusion High-Fidelity DNA polymerase (Thermo Scientific) and the corresponding pairs of primers as listed in Table Table1.1. The purified PCR products (100–200 ng) were treated with 0.5 U Taq polymerase (Roche) and 0.75 mm dATP by incubation at 72 °C for 15 min to introduce the 3′-A overhangs. The resulting insert DNA fragments were ligated into the pET-SUMO vector according to the instruction manual of the Champion pET SUMO expression system (Invitrogen). The construction of the Rh-FGD1 mutants K197N, K258N, and R282Q was done by using the QuikChange® mutagenesis method with primers (Table (Table1)1) designed by the web-based QuikChange® Primer Design Tool (Agilent Technologies) and the pET-SUMO-RHA1_RS43115 plasmid as template. All constructs were confirmed by sequencing.
Proteins were initially expressed in E. coli BL21(DE3), grown in Terrific broth containing 50 μg/mL kanamycin, 1% (w/v) glucose, and induced with 1 mm isopropyl β-D-1-thiogalactopyranoside (IPTG) at 24 °C when the cells reached OD600 ~0.7–0.8. To overcome the insolubility of the overexpressed proteins in E. coli BL21(DE3), the expression hosts were changed to E. coli C41(DE3) (Lucigen) for both the wild-type and mutant Rh-FGDs. The culture conditions were kept the same as for E. coli BL21(DE3), except for the addition of 0.2% (w/v) glucose. The cells were grown until late stationary phase and harvested by centrifugation at 4600×g for 10 min (Beckman–Coulter JA-10 rotor, 4 °C). Cells were resuspended in lysis buffer (50 mm KPi pH 7.8, 400 mm NaCl, 100 mm KCl, 10% (v/v) glycerol, 1 mm β-mercaptoethanol, 20 mm imidazole) and disrupted by sonication using a VCX130 Vibra-Cell (Sonics & Materials, Inc., Newtown, USA) at 4 °C (5 s on, 10 s off, 70% amplitude, total of 5 min). Following centrifugation at 20000×g for 45 min (Beckman–Coulter JA-25.5 rotor, 4 °C) to remove unbroken bacteria and cellular debris, the supernatant was applied onto a 5-mL HisTrap HP column (GE Healthcare) pre-equilibrated in the same buffer. The recombinant enzyme with the His-SUMO tag was eluted with a gradient from 20 to 500 mm imidazole in the same buffer. Fractions containing the pure enzyme as indicated by SDS-PAGE and FGD activity assay were pooled, desalted to remove imidazole, and concentrated in a 30-kDa MWCO Amicon (Milipore) centrifugal filter unit. Protein concentration was estimated using the Waddell’s method (Waddell 1956).
To obtain the native enzyme, the His-SUMO tag was cleaved by incubating with 1% (mol/mol) SUMO protease (Invitrogen) for 2 h at 4 °C. The His-SUMO tag, uncleaved protein, and SUMO protease were removed by applying the cleavage mixture onto a second HisTrap column. The native enzyme was concentrated and finally purified through a Superdex 200 10/300 GL (GE Healthcare) in 10 mm HEPES pH 7.5, 100 mm NaCl, 10% (v/v) glycerol, 1 mm β-mercaptoethanol prior to crystallization experiments.
Analysis of Rh-FGD1 thermostability was based on the unfolding temperature, T m, determined by the use of the Thermofluor® technique (Pantoliano et al. 2001) with a Bio-Rad C1000 Touch Thermal Cycler (Bio-rad Laboratories, Inc.) in 96-well plates. Each well had a final volume of 25 μL containing 1.6 μm Rh-FGD, 5 × SYPRO Orange (Invitrogen), buffers, and/or additives. The protein start buffer was exchanged to 50 mm KPi pH 7.8 and 150 mm NaCl for the buffer screen and to 50 mm KPi pH 7.8, 500 mm NaCl, and 100 mm KCl for the additive screen. The compositions of the buffers and additives are described in Boivin et al. (Boivin et al. 2013).
FGD activity was routinely monitored by following the reduction of F420 at 420 nm, 25 °C, and pH 7.5 using an absorption coefficient ε420 nm of 41.4 mm −1 cm−1 (Eirich et al. 1978; Purwantini et al. 1992) in a V-650 spectrophotometer from Jasco (IJsselstein, The Netherlands). F420 was isolated from M. smegmatis as previously described (Bashiri et al. 2010; Isabelle et al. 2002) (M. smegmatis mc2 4517 and the plasmid pYUBDuet-FbiABC were kind gifts from Dr. G. Bashiri, the University of Auckland, New Zealand). The assay mixture typically contained 50 mm Tris/HCl pH 7.5, 300 mm NaCl, 1 mm β-mercaptoethanol, 1 mm EDTA, 100 nm enzyme, 20 μm F420, and 1 mm glucose-6-phosphate (G6P) in a final volume of 500 μL. For steady-state kinetics, 10 nm enzyme was used in the same buffer except for the experiments with glucose that were performed with 500 nm enzyme. Kinetic data were analyzed using nonlinear regression to the Michealis–Menten equation using GraphPad Prism v. 6.0 (GraphPad Software Inc., La Jolla, CA, USA). For the pH optima determination, the reactions contained 40 mm Britton-Robinson buffer (Britton and Robinson 1931), 100 nm enzyme, and 20 μm F420 and were initiated by adding 1 mm G6P. In the experiments, enzyme activity was monitored at 401 nm (an isosbestic point of F420; ε401 nm = 25 mm −1 cm−1) (Jacobson et al. 1982; DiMarco et al. 1990) for 5 min.
Alternative phosphate-sugar substrates for FGD were screened in a SynergyMX microplate spectrophotometer (BioTek) using 96-well plates with clear bottom. The reaction mix (200 μL) contained 100 nm enzyme, 10 mm substrate, and 29.6 μm F420 in the same buffer as described in the general spectrophotometric assay. The tested compounds for substrate profiling were d-glucose, d-mannose-6-phosphate, d-fructose-6-phosphate, α-d-glucose-1-phosphate, α-d-galactose-1-phosphate, and d-glucosamine-6-phosphate. The absorbance of F420 at 420 nm was monitored in intervals of 45 s for 1 h.
Native Rh-FGD1 crystals were obtained using the sitting-drop vapor diffusion technique at 20 °C by mixing equal volumes of 9.0 mg/mL protein in 10 mm HEPES pH 7.5, 10% (v/v) glycerol, 100 mm NaCl, 1 mm β-mercaptoethanol and of the mother liquor containing 0.16 m ammonium sulfate, 0.08 m sodium acetate pH 4.6, 20% (w/v) PEG 4000, and 20% (v/v) glycerol. X-ray diffraction data were collected at the PXI and PXIII beamlines of the Swiss Light Synchrotron in Villigen, Switzerland (SLS), and at the ID23-1 beamline of the European Synchrotron Radiation Facility in Grenoble, France (ESRF). Image integration and data scaling were processed with MOSFLM (Battye et al. 2011) and programs of the CCP4 suite (Winn et al. 2011). Detailed data processing statistics are shown in Table Table2.2. The Rh-FGD1 structure was initially solved by molecular replacement using MOLREP (Vagin and Teplyakov 2010) using the coordinates of FGD1 from M. tuberculosis (PDB ID code 3B4Y) (Bashiri et al. 2008) as the search model devoid of all ligands and water molecules. Model building and structure analysis was carried out with COOT (Emsley and Cowtan 2004) whereas alternating cycles of refinement was performed with REFMAC5 (Murshudov et al. 1997). Figures were created by CCP4mg (McNicholas et al. 2011); atomic coordinates and structure factors were deposited in the Protein Data Bank under the PDB ID code 5LXE.
Three genes encoding putative homologs of Mtb-FGD (accession number KBJ40183) (Bashiri et al. 2007; Bashiri et al. 2008) were identified by BLAST: RHA1_RS43115 (WP_011600337.1), RHA1_RS10755 (WP_011595003.1), and RHA1_RS43570 (WP_011600440.1) (with 84, 84, and 83% sequence identity to Mtb-FGD1, respectively). These genes were amplified from R. jostii RHA1 genomic DNA, cloned into the pET-SUMO vector, and expressed in E. coli C41(DE3) as N-terminal SUMO-hexahistidine-fused proteins using IPTG as an inducer. The cultivation conditions were optimized for the production of the soluble and active proteins, resulting in a 48-h growth at 24 °C with 1 mm IPTG in Terrific broth as the most effective condition. By testing the cell extracts containing all three different proteins (RHA1_RS43115, RHA1_RS10755, and RHA1_RS43570 referred to as Rh-FGD1, Rh-FGD2, and Rh-FGD3, respectively), it was found that they all exhibit FGD activity. Rh-FGD1 and Rh-FGD2 exhibited comparable specific activity whereas Rh-FGD3 was >20-fold less active. We focused our exploration on the best expressed FGD, namely Rh-FGD1. Typically, approximately 80 mg of pure Rh-FGD1 was obtained from 1 L of TB culture. Furthermore, it is worth noting that Rh-FGD1 is flanked by genes putatively encoding a 6-phosphogluconate dehydrogenase and a G6P isomerase. This strongly suggests that Rh-FGD1 is indeed a glucose-6-phosphate dehydrogenase.
Rh-FGD1 displayed an optimum for activity on glucose-6-phosphate at pH 7.5–8.0 (Fig. (Fig.2);2); this is somewhat similar to the FGDs from Mtb [6.5–7.0 (Bashiri et al. 2008)] and from M. smegmatis [two separate pH optima: 6.0 and 8.0 (Purwantini and Daniels 1996)]. For further studies on Rh-FGD1, pH 7.5 was chosen to monitor FGD activity.
Rh-FGD1 is strictly dependent on F420 as coenzyme. The enzyme did not show any significant activity when NAD+, NADP+, FAD, or FMN was used as alternative electron acceptors. Rh-FGD1 was also found to be highly specific for G6P as electron donor. All tested alternative phosphate-sugars displayed significantly lower activity when compared to G6P. 10 mm d-mannose-6-phosphate, d-fructose-6-phosphate, and d-glucosamine-6-phosphate reached only 1.1, 4.8, and 2.8% of the rate obtained with 1 mm G6P, respectively. The free anomeric carbon C1 of the sugar is crucial for the dehydrogenation as no detectable FGD activity was observed with α-d-glucose-1-phosphate and α-d-galactose-1-phosphate. Rh-FGD1 accepted d-glucose as substrate, although with very low catalytic activity.
The thermostability of Rh-FGD1 was evaluated by determining apparent melting temperatures (T m) using the Thermofluor® technique (Pantoliano et al. 2001). This revealed that Rh-FGD1 represents a stable enzyme, exhibiting T m values above 35 °C in most common buffer systems (Fig. (Fig.3).3). The best stabilizing buffers were HEPES, citrate, and phosphate. Several additives were found to have significant effects on the thermostability of Rh-FGD1. NaCl, glycerol, and divalent cations (e.g., Mg2+, Ca2+, and Mn2+) exerted marked effects, resulting in T m values of above 55 °C. The stabilizing effect of NaCl depends greatly on its concentration: an increase in NaCl concentration from 50 mm to 1 m (in either HEPES or Tris/HCl) drastically elevates the T m by around 20 °C. Based on these findings, we typically stored Rh-FGD1 in a phosphate-based buffer with both NaCl and glycerol as additives. Remarkably, the enzyme can retain >90% of its activity after 1 year when being stored at −80 °C.
For determining the steady-state kinetic parameters with F420 and glucose-6-phopshate as substrate, Rh-FGD1 activity was monitored following the decrease in absorbance at 420 nm associated with the reduction of F420. The kinetic data did fit well when using the Michaelis–Menten kinetic model. The kinetic parameters for the natural substrates G6P and F420 were determined (Table (Table3)3) by keeping one of the substrates constant (F420 at 20 μm or G6P at 2.0 mm, respectively), while varying the other substrate concentration. The apparent K m values for G6P and F420 are 0.31 mm and 3.8 μm, respectively. The K m value for F420 is very similar to that observed with FGDs from Mtb (K d = 4.5 μm) and M. smegmatis (4 μm) (Purwantini and Daniels 1996; Bashiri et al. 2008). The K m value for G6P is closer to that for Mtb-FGD (0.1 mm) whereas it is much lower than the equivalent value from FGD in M. smegmatis (1.6 mm). The observed differences in K m values for G6P can partly be explained by different levels of G6P in these organisms; e.g., it is known that mycobacterial cells can contain high levels of G6P (Hasan et al. 2010; Purwantini and Daniels 1996).
As the FGD-catalyzed reaction involves two substrates, G6P and F420, we set out to decipher which mechanism is operative for Rh-FGD1, namely a ping-pong, sequential, or random mechanism. Both substrate concentrations were varied, and the F420 reduction rates were measured accordingly. Increasing concentrations of both substrates G6P and F420 resulted in an increase in reaction rates, suggesting that the reaction occurs via a ternary complex Rh-FGD1:G6P:F420. This is best illustrated by the observed intersection of the lines when double reciprocal values of the reaction rates and substrate concentrations are plotted (Fig. (Fig.4).4). Whether these two substrates bind in an ordered or a random manner, however, remains to be further investigated, e.g., by product inhibition or tracer studies with radioactive labeled substrates.
The crystal structure of Rh-FGD1 was determined at 1.47 Å resolution by molecular replacement using Mtb-FGD1 devoid of all ligands (PDB ID code 3B4Y) (Bashiri et al. 2008) as the search model. The asymmetric unit contains two enzyme monomers forming a compact dimer (Fig. (Fig.5a),5a), which is also observed in solution as estimated by gel permeation analysis (data not shown), similarly to the mycobacterial homolog (1.0 Å rmsd difference for 610 pairs of Cα atoms) (Fig. (Fig.5b).5b). The very good quality of the electron density enabled us to model several residues in a double conformation and to identify a residue with a cis peptide bond in proximity of the active site (Fig. (Fig.5c).5c). Only residues 254–263 in subunit A, and 250–279 in subunit B lack clear electron density and were therefore excluded from the final model. Each Rh-FGD1 monomer is comprised of residues 1–334, forming an (α/β)8 TIM-barrel, with the active site typically located at the C-terminus of the barrel, as observed in Mtb-FGD1 (Bashiri et al. 2008). As indicated by the Dali server (Holm and Rosenstrom 2010), this protein topology is shared also with other homologous members of the luciferase-like hydride transferase family, including a secondary alcohol dehydrogenase (Adf) and a methylene-tetrahydromethanopterin reductase (Mer) (34 and 25 sequence identity with Rh-FGD1, respectively) (Aufhammer et al. 2004; Aufhammer et al. 2005). The two Rh-FGD1 molecules present in the asymmetric unit are essentially identical, as indicated by an overall rmsd difference of 0.55 Å in Cα atomic positions of 302 residues, except for a segment comprising residues 41–49, which was excluded in the noncrystallographic symmetry restrained refinement. The dimer interface area is rather large, burying approximately 2000 Å2 [as analyzed by the program PISA (Krissinel and Henrick 2007)] and accounting for ~14% of the monomer’s surface. Unless explicitly stated, hereafter, we will refer to monomer A for describing the structure.
All attempts to elucidate the structure of Rh-FGD1 in its holoenzyme form, i.e., with the F420 cofactor bound, were unsuccessful. Nonetheless, the obtained overall structure is substantially identical to that of the F420-bound Mtb-FGD1 and the architecture of the active site is conserved. Therefore, the F420 molecule was tentatively modeled in the Rh-FGD1 as a result of the superposition of the mycobacterial enzyme structure (Fig. (Fig.5b,5b, c). In particular, the high - quality electron density clearly indicates the presence of a well-ordered nonprolyl cis-peptide bond between Ser72–Val73 constituent of a bulge at the end of a β strand close to the presumed binding site of the F420 isoalloxazine ring (Fig. (Fig.5c).5c). This unusual cis-peptide is highly conserved in this enzyme family, being consistently observed in Mtb-FGD1, Adf, and Mer (joining Ser74–Val75, Cys72–Ile73 and Gly61–Val62, respectively). This bulge is essential as it serves as a backstop to hold the isoalloxazine ring from its re-face, bending the deazaisoalloxazine ring into a butterfly conformation (Aufhammer et al. 2005; Bashiri et al. 2008; Aufhammer et al. 2004). The F420 binding pocket is largely identical among FGDs, Adf, and Mer wherein the deazaisoalloxazine ring locates at the innermost part of the pocket and the hydrophilic polyglutamate tail extends into the solvent (Figs. (Figs.5d5d and and6).6). The most noticeable difference between the various structures of F420-binding proteins is a helical coil region located at the C-terminus of the TIM-barrel, creating a sort of lid element that stabilizes cofactor binding (Fig. (Fig.6).6). In Rh-FGD1, the sequence for this structural element is shorter than that of the homologous enzymes and corresponds to the disordered part (residues 254–263 in monomer A). The high flexibility of this region may correlate with a more dynamic interaction with the cofactor and may well explain the difficulty in obtaining the Rh-FGD1 structure in its holoenzyme form.
In Mtb-FGD1, a citrate molecule, most likely derived from the crystallization solution, was found to bind adjacent to the F420 molecule and later proved to be a competitive inhibitor for Mtb-FGD1. Citrate occupies a cavity with a size that can fit G6P in an orientation that is suitable for catalysis. This allowed the modeling of G6P into the active site of Mtb-FGD1, revealing highly conserved residues involved in substrate binding and catalysis (Bashiri et al. 2008) (Fig. (Fig.5d).5d). It has been postulated that in Mtb-FGD1, the phosphate moiety of G6P occupies a positively charged pocket constituted by side chains of Lys198, Lys259, and Arg283 (corresponding to Lys197, Lys258, and Arg282 in Rh-FGD1, respectively) (Bashiri et al. 2008) (Fig. (Fig.5b).5b). In Rh-FGD1, out of the three residues, the position of Lys258 is unknown as it is part of the disordered region. Instead, Lys197 and Arg282 are visible and adopt a similar conformation with respect to the corresponding residue in Mtb-FGD1. Nevertheless, a part of their side chains lacks clear electron density (Cδ, Cϵ, and Nζ of the former and the guanidinium group of the latter), which indicates a much higher flexibility. Sequence alignment indicated that the three residues are strictly conserved in proteins exhibiting FGD activity (Bashiri et al. 2008). Rv0132c—sharing 36% sequence identity with Mtb-FGD1 and previously annotated as Mtb-FGD2—does not contain these phosphate group binding residues and consistently showed no such assigned activity (Bashiri et al. 2012). In fact, Rv0132c was later proven to be an F420-dependent hydroxymycolic acid dehydrogenase and was proposed to be an unprecedented antitubercular target that may also be inhibited by the novel drug pretomanid (PA-824) (Purwantini and Mukhopadhyay 2013).
To further probe the roles of these residues, we generated single mutations, namely K197N, K258N, and R282Q Rh-FGD1. In comparison to the wild-type Rh-FGD1, the mutants showed a drastic decrease in catalytic efficiency for G6P, as indicated by K m values of two orders of magnitude higher than that of the wild-type enzyme (Table (Table3).3). The rate of catalysis was also considerably affected: The R282Q mutant virtually lost activity whereas the K197N and K258N mutants had a 4.5- and 30-fold lower k cat value, respectively, compared with the wild type. Disruption of the phosphate binding pocket may, to a certain extent, also affect the substrate specificity. In fact, when glucose was used as a substrate instead of G6P, the mutant K197N showed a slightly improved k cat/K m when compared with the wild-type enzyme (Table (Table3).3). These data unequivocally verified that the three targeted residues are crucial for the binding of the phosphate moiety of the G6P. Moreover, it might become possible to improve FGD activity towards glucose, e.g., by random mutagenesis of residues forming the G6P binding pocket.
F420 is an unusual redox cofactor originally found exclusively in a restricted number of microbes, such as archaea and actinomycetes. Astonishingly, a bioinformatics study in 2010 indicated that F420 can be much more widespread than previously thought and present in 11% of all sequenced bacteria and archaea (Selengut and Haft 2010). In particular, R. jostii RHA1 was predicted to contain at least 104 deazaflavoenzymes, an impressively large number. In line with this prediction, we present here the first experimental evidence for the presence of F420-dependent enzymes in R. jostii RHA1. More specifically, the R. jostii RHA1 genome encodes at least three FGDs: RHA1_RS43115 (WP_011600337.1), RHA1_RS10755 (WP_011595003.1), and RHA1_RS43570 (WP_011600440.1) (referred to as Rh-FGD1, Rh-FGD2, and Rh-FGD3, respectively). We have focused our exploration on Rh-FGD1, the best expressed one, characterized the kinetic properties and elucidated the structure of the apo protein at high resolution. On a cautionary note, it should be noticed that the Rh-FGD1 and Rh-FGD3 are plasmid encoded whereas Rh-FGD2 is instead encoded by a chromosomal gene. Preliminary tests (data not shown), however, indicated that Rh-FGD1 and Rh-FGD2 have comparable specific activity. This gene redundancy is generally believed to facilitate the high catabolic versatility in rhodococci (van der Geize and Dijkhuizen 2004; McLeod et al. 2006).
F420-dependent glucose-6-phosphate dehydrogenase has been suggested to be the primary enzyme responsible for the F420 reduction in several actinomycetal genera, including mycobacteria, thereby linking their central metabolism to the F420 reduction reaction (Purwantini and Daniels 1996). The main role of mycobacterial FGDs appears to be the generation of F420H2 as these bacilli also encode the conventional NADP+-dependent FGDs (Purwantini et al. 1997), which interestingly showed no significant phylogenetical relation to FGDs (Purwantini and Daniels 1998). A deletion of either fgd or fbiC—a gene involved in the biosynthesis of F420—renders these mycobacterial strains incapable of reducing xenobiotics via F420H2-dependent reductases (Taylor et al. 2010; Hasan et al. 2010; Stover et al. 2000; Manjunatha et al. 2006). Nevertheless, the physiological role of FGDs in Rhodococcus spp. remains largely unclear. It is well known that in Rhodococcus opacus and Nocardioides simplex, the reduced F420 is supplied mainly by F420:NAPDH oxidoreductases (FNOs) rather than FGDs (Ebert et al. 1999; Heiss et al. 2002; Ebert et al. 2001; Heiss et al. 2003). FNOs were found to be expressed from the same operon as the F420H2-dependent hydride transferases, which are responsible for the degradation of environmental nitroaromatic compounds such as picrate and 2,4-dinitrophenols (Ebert et al. 1999; Heiss et al. 2002; Ebert et al. 2001). However, it cannot be excluded that FGDs also play a (crucial) role in generating F420H2. As very little is known about the natural substrates of the F420H2-dependent enzymes in Rhodococcus spp., it can be speculated that FGDs are primarily responsible for providing the reductant for the endogenous metabolism, maintaining the redox homeostasis during normal growth or in response to oxidative stress as observed in mycobacteria. Several lines of evidence have revealed the pivotal role of G6P as an electron reservoir mobilized via FGDs in protecting mycobacteria against oxidative and nitrosative stress (Hasan et al. 2010; Gurumurthy et al. 2013). In fact, the presence of a NADP+-dependent FGD alone failed to render a M. smegmatis mutant deficient in FGD capable of surviving oxidative stress. Further investigations, e.g., gene deletion studies in combination with isotopic labeling metabolomics, are therefore necessary to decipher the precise role of FGDs in rhodococci (van der Geize et al. 2008).
The biocatalytic reduction of F420 has been carried out so far with the use of Mtb-FGD1 (Manjunatha et al. 2006). Such reduced F420 is essential in studying deazaflavin-dependent reductases. However, mycobacterial FGDs are poorly to moderately expressed as soluble protein when E. coli is used as a heterologous expression host (Purwantini and Daniels 1998; Bashiri et al. 2007; Manjunatha et al. 2006). To overcome this limitation, a dedicated M. smegmatis expression strategy was developed to enhance the solubility of mycobacterial proteins. The typical yield obtained was 7 mg of pure recombinant Mtb-FGD1 from 1 L of M. smegmatis culture (Bashiri et al. 2007). In contrast, we produced soluble Rh-FGD1 in rather high yield: 80 mg of pure protein L−1 of culture. The developed E. coli-based expression system facilitates the routine production of soluble FGD which can be used for the synthesis of reduced F420. Rh-FGD1 is a relatively fast enzyme, with a k cat of 17 s−1 for G6P (Table (Table3).3). In addition, Rh-FGD1 appears to be thermostable in most common buffers and additives (Fig. (Fig.3);3); upon storage at −80 °C, Rh-FGD1 retained >90% activity after 1 year. The observation that Rh-FGD1 displayed some activity, yet very low (Table (Table3),3), with glucose, a much cheaper substrate instead of G6P, hints to the possibility to engineer Rh-FGD1 into a more efficient F420-dependent glucose dehydrogenase. The first logical target for such tailoring efforts would be the phosphate binding pocket. Interestingly, when glucose was used as substrate, the mutant K197 N showed an improved k cat/K m value of 30% higher than that of the wild type. Therefore, by fine-tuning these residues by site-directed mutagenesis, one could obtain mutants with improved activity with the cheap cosubstrate glucose. Given its robustness and accessibility, Rh-FGD1 represents a potential candidate for the biocatalytic reduction of F420 in larger scale or in fusion with other valuable F420H2-dependent reductases in a redox self-sufficient whole-cell biotransformation.
This work was supported by a Ubbo Emmius fund from the University of Groningen, the Netherlands, via a research PhD scholarship awarded to QTN. We thank the European Synchrotron Radiation Facility (ESRF) and the Swiss Light Source (SLS) for providing beamtime and assistance, and the European Community’s Seventh Framework Programme (FP7/2007−2013) under BioStruct-X (Grants 7551 and 10205) for funding synchrotron trips. We are grateful to Dr. G. Bashiri at the Structural Biology Laboratory, School of Biological Sciences and Maurice Wilkins Centre for Molecular Biodiscovery, the University of Auckland, New Zealand, for generously providing a culture of M. smegmatis mc2 4517 and the plasmid pYUBDuet-FbiABC. S. Rovida, V. Piano, V. Speranzini, and F. Fiorentini were acknowledged for their experimental support.
This study was funded by a Ubbo Emmius scholarship from the University of Groningen, the Netherlands (awarded to QTN), and the European Community’s Seventh Framework Programme (FP7/2007−2013) under BioStruct-X (Grants 7551 and 10205).
The authors declare that they have no conflict of interest.
This article does not contain any studies with human participants or animals by any of the authors.
Andrea Mattevi, Phone: +39 0382 985525, Email: email@example.com.
Marco W. Fraaije, Phone: +31 50 363 4345, Email: firstname.lastname@example.org.