Structural Basis of Free Reduced Flavin Generation by Flavin Reductase from Thermus thermophilus HB8

Background: TTHA0420 is a flavin reductase, which makes free reduced flavin involved in a variety of fields.

Results: We determined the dual binding mode of the substrate and co-factor flavins of TTHA0420.

Conclusion: A specific motif YGG in the C terminus functions to regulate the alternative binding of NADH and substrate flavin.

Significance: Our results have mechanistic implications for the reductase with two flavins.

Abstract

Free reduced flavins are involved in a variety of biological functions. They are generated from NAD(P)H by flavin reductase via co-factor flavin bound to the enzyme. Although recent findings on the structure and function of flavin reductase provide new information about co-factor FAD and substrate NAD, there have been no reports on the substrate flavin binding site. Here we report the structure of TTHA0420 from Thermus thermophilus HB8, which belongs to flavin reductase, and describe the dual binding mode of the substrate and co-factor flavins. We also report that TTHA0420 has not only the flavin reductase motif GDH but also a specific motif YGG in C terminus as well as Phe-41 and Arg-11, which are conserved in its subclass. From the structure, these motifs are important for the substrate flavin binding. On the contrary, the C terminus is stacked on the NADH binding site, apparently to block NADH binding to the active site. To identify the function of the C-terminal region, we designed and expressed a mutant TTHA0420 enzyme in which the C-terminal five residues were deleted (TTHA0420-ΔC5). Notably, the activity of TTHA0420-ΔC5 was about 10 times higher than that of the wild-type enzyme at 20–40 °C. Our findings suggest that the C-terminal region of TTHA0420 may regulate the alternative binding of NADH and substrate flavin to the enzyme.

Introduction

Free reduced flavins are involved in a variety of biological functions. In the flavin-diffusible monooxygenase (FDM)² system, for example, free reduced flavin is first generated by the flavin reductase component of the enzyme and then transferred to the oxygenase component for activation. The flavin reductase-luciferase system, which mediates bioluminescence, is well studied (1, 2). The FRP-luciferase coupled reaction can utilize free reduced flavin through free diffusion or direct transfer. Flavin reductases have also been isolated from nonluminous organisms, where they are involved in reduction of ferritin-Fe(III) to Fe(II) (3, 4), oxidation of aromatic compounds (5–7), degradation of chelating agents (8), desulfurization of fossil fuels (9), biosynthesis of antibiotics (10–13), and activation of ribonucleotide reductase (14, 15).

In recent years, the flavin reductase component of FDM has been the subject of both functional and structural analyses, which have provided some understanding of the mechanism by which the flavin is reduced. For example, crystallographic analysis has shown that the flavin reductase PheA2 from Bacillus thermoglucosidasius is a homodimer; that each subunit is organized around a six-stranded antiparallel β-barrel and a capping α-helix; and that the enzyme catalyzes the NADH-dependent reduction of free FAD via the ping-pong bisubstrate-biproduct mechanism (6). The reduced FAD is then used by the larger monooxygenase component (PheA1) to hydroxylate phenols to the corresponding catechols. PheA2 contains a dual binding cleft for substrate NADH and FAD, which alternate binding during catalysis. The structures of two enzymes from a similar family, HpaC_St from Sulfolbus tokodaii (5) and HpaC_Tt from Thermus thermophilus HB8 (16), have also been solved. In those enzymes, the substrate flavin binding site is thought to be in close proximity to the NADH binding site, but the precise location of the sites remained unclear.

TTHA0420 in T. thermophilus HB8 was predicted to be a flavin reductase after it was found to contain a flavin reductase motif using Pfam. Specifically, the main motif is GDH, which is also conserved in PheA2 or HpaC. However, we found that this enzyme belongs to a subclass of flavin reductases having a specific feature in the C-terminal region, based on multiple sequence alignment with a wide array of other known flavin reductases.

In the present study, we determined the crystal structure of the apo- and holoenzymes of TTHA0420 from T. thermophilus HB8 at a resolution of 2.0 and 1.9 Å, respectively. Our findings not only reveal the co-factor flavin binding site but also provide insight into the substrate flavin binding site. These structural and functional studies suggest that the C-terminal region of TTHA0420 may regulate the alternate binding of NADH and substrate flavin to the enzyme. Furthermore, the fact that TTHA0420 has no coupled oxygenase component suggests the free reduced flavin could be involved in functions such as reduction of Fe(III) from ferrisiderophore (15, 17).

EXPERIMENTAL PROCEDURES

Subcloning of Wild-type Enzyme

The wild-type TTHA0420 gene was amplified by polymerase chain reaction (PCR) from T. thermophilus HB8 genomic DNA. The PCR product was then cloned into the pET11a expression vector, and the resulting construct was confirmed by DNA sequencing.

C-terminal Five-residue Deletion Mutant

The cDNA encoding TTHA0420 was amplified using the N-terminal primer 5′-ggg cat atg atg aat ctg gag gcc aag aag aag-3′ and the C-terminal primer 5′-ggg gga tcc tta gcc cgt gtc cca cat gac cag-3′, after which the PCR product was cloned into the NdeI/BamHI site of the pET15b expression vector. The resulting construct encoded a TTHA0420 deletion mutant lacking the C-terminal five residues (TTHA0420-ΔC5), which was confirmed by DNA sequencing.

Purification of Wild-type Enzyme

Selenomethionine-substituted protein was expressed in the Escherichia coil methionine auxotroph strain B834 (DE3). Cells were grown in M9 medium supplemented with selenomethionine for 12 h at 37 °C, after which they were harvested by centrifugation, suspended in buffer containing 20 mm Tris-HCl buffer (pH 8.0) and 500 mm NaCl, sonicated, and centrifuged to remove the cell debris. The supernatant was then incubated for 10 min at 70 °C and centrifuged, and the supernatant was applied to a Resource-PHE column equilibrated with 50 mm sodium phosphate buffer (pH 7.0) containing 1.5 m (NH₄)₂SO₄. After washing with the equilibration buffer, the protein was eluted with a linear gradient of 1.5 to 0.7 m (NH₄)₂SO₄ in 50 mm sodium phosphate buffer (pH 7.0). The protein-containing fractions were dialyzed against 20 mm Tris-HCl buffer (pH 8.0). Finally the protein was loaded onto a Resource-Q column equilibrated with 20 mm Tris-HCl buffer (pH 8.0). After washing with the equilibration buffer, the protein was eluted with a linear gradient of 0–4.5 m NaCl in 20 mm Tris-HCl buffer (pH 8.0), and the protein-containing fractions were pooled and dialyzed against 20 mm Tris-HCl buffer (pH 7.0). The protein fractions were separated into two peaks, one yellow and one clear. The amount of clear fraction was much larger than the amount of yellow fraction, so we used the clear fraction for the crystallization for apoenzymes. Holoenzyme was prepared by adding FAD for the crystallization.

Purification of TTHA0420-ΔC5 Mutant

The TTHA0420-ΔC5 mutant was expressed in E. coli BL21 (DE3) cells grown in SB medium for 3 h at 37 °C with constant shaking. Once the absorbance at 600 nm reached 0.7, 1 mm IPTG was added to induce expression of the mutants. After incubation for an additional 3 h, the induced cells were harvested by centrifugation, suspended in buffer containing 50 mm Tris-HCl buffer (pH 7.4), 1 mm EDTA, and 2 mm PMSF, sonicated, and centrifuged to remove the cell debris. The supernatant was then incubated for 10 min at 70 °C and centrifuged, after which it was applied to a chelating Sepharose Fast Flow column equilibrated with 20 mm Tris-HCl buffer (pH 7.0) containing 500 mm NaCl and 20 mm imidazole. After washing with the equilibration buffer, the protein was eluted with 100, 200, 300, and 500 mm NaCl. Finally, the resultant eluate was loaded onto a MonoQ column equilibrated with 20 mm Tris-HCl buffer (pH 7.0), and the protein was eluted with a linear gradient of 0–1 m NaCl in 20 mm Tris-HCl buffer (pH 7.0). The protein-containing fractions were pooled and dialyzed against 20 mm Tris-HCl buffer (pH 7.0). The His tag on TTHA0420-ΔC5 was removed using a thrombin cleavage capture kit (Novagen).

Flavin Reductase Assay

TTHA0420-dependent oxidation of NADH was measured to assess the enzyme activity. The initial rate of NADH oxidation was determined in a BECKMAN COULTER DU 640 spectrophotometer by monitoring A₃₄₀ using Δϵ = 6,200 m⁻¹ cm⁻¹ for NADH. The reaction was run in a 1.0-ml quartz cuvette with a 1-cm light path. The assay mixture included 20 mm potassium phosphate buffer (pH 7.0), 200 μm NADH, and 50 μm FAD (FMN), and measurements were made at 70 °C. Values were corrected for oxidation of NADH in the absence of substrate.

Determination of Temperature Optimum

The temperature optima for wild-type TTHA0420 and TTHA0420-ΔC5 were determined using 20 mm potassium phosphate buffer (pH 7.0) containing 50 μm FAD and 200 μm NADH, and the reaction time was 20 min. The assay temperature was varied from 20 to 90 °C.

Kinetic Analysis

For kinetic analysis, the NADH:flavin oxidoreductase activity of TTHA0420 was determined at 40 °C in 20 mm potassium phosphate buffer (pH 7.0). To estimate the kinetic parameters, the NADH concentration was varied at fixed flavin concentrations and vice versa. Kinetic parameters were determined from saturation curves and fitted with the Michaelis-Menten equation. To gain insight into the mechanism underlying the catalytic activity, NADH:flavin oxidoreductase activity was determined as a function of the FAD concentration at several constant NADH levels and as a function of the NADH concentration at several constant FAD levels. Reciprocal initial velocities were plotted against reciprocal substrate concentrations and fitted with a straight line determined by linear regression.

Crystallization of Apo- and Holoenzymes

Crystals of the selenomethionine-substituted apoenzyme were grown using the sitting drop vapor diffusion method, in which 1 μl of protein solution (9.25 mg/ml) was mixed with an equal volume of reservoir solution containing 200 mm diammonium tartrate (pH 6.6) and 25% PEG 3350 at 20 °C. Crystals of the wild type and holoenzyme were grown using the hanging drop vapor diffusion method, in which 2 μl of the enzyme solution (9.25 mg/ml) containing 5 mm FAD were equilibrated against a reservoir containing 100 mm HEPES (pH 8.0), 2% PEG 400, and 2.2 m ammonium sulfate at 20 °C.

Data Collection of Apo- and Holoenzymes

Data for the apoenzyme were measured using a CCD detector Jupiter 210 at the synchrotron radiation source at the BL26B2 station at SPring-8, Harima, Japan and were processed using CrystalClear (Rigaku/MSC). Multiwavelength anomalous dispersion (18) data were collected from a single crystal of selenomethionine TTHA0420 under cryoconditions (−180 °C) using mixed paraffin oil/Paratone-N. Data for the holoenzyme were measured using an ADSC Quantum315 CCD detector system (Area Detector Systems) on beamline KEK-BL5 at the Photon Factory in Tsukuba, Japan under cryoconditions (−180 °C) in mixed paraffin oil/Paratone-N and were processed using HKL2000. The data collection statistics are summarized in Table 1.

TABLE 1.

Data collection and refinement statistics

Values in parentheses are for the highest resolution shell.

	Apo-crystal			Holo-crystal
	Peak	Edge	Remote
Data collection
Beamline		BL26B2 (Spring-8)		KEK-BL5 (Photon Factory)
Detector		Jupiter 210		ADSC Quantum 315
Wavelength (Å)	0.9791	0.9795	0.9000	1.0000
Space group		P41212 (a = b = 63, 75 Å, c = 80, 45 Å)		I432 (a = b = c = 132, 97 Å)
Resolution range (Å)		40.0–1.9		50.0–1.9
No of reflections (Unique)	122, 500 (13,600)	121,070 (13,571)	124,405 (13,607)	673,191 (16,153)
Completeness (%)	99.7 (100.0)	99.5 (99.9)	99.7 (100.0)	100.0 (100.0)
Rsym (%)a	5.7 (7.9)	3.2 (4.3)	3.8 (4.9)	11.0 (29.2)
Redundancy	9.0 (9.3)	8.9 (9.1)	8.8 (9.1)	41.7 (43.2)
Mean I/ σI	28.5 (18.6)	50.9 (40.0)	37.9 (29.6)	8.8 (5.3)
Refinement
Resolution		34-2.0		20-1.9
Rwork (%)b		21.2		20.0
Rfree (%)c		24.1		22.8
Root mean square deviations
Bond length (Å)		0.005		0.013
Bond angles (degrees)		1.22		1.51
Cofactors, water molecules		129 wat		1 FAD, 1 FAD, 69 wat
Ramachandran plot
Most favored regions (%)		97.5		97.5
Additionally allowed regions (%)		2.5		2.5

^a R_sym = Σ_hΣ_i|I_i(h) − 〈I(h)〉|/Σ_hΣ_i|I_i(h)|, where I_i(h) is the intensity measurement for a reflection h, and 〈I(h)〉 is the mean intensity for this reflection.

^b R_work = Σ_h‖F_o| − |F_c‖/Σ_h|F_o|.

^c R_free was calculated with randomly selected reflections (5%).

Structural Determination and Refinement of Apo- and Holoenzymes

The positions of the selenium atoms were determined with the program SOLVE (19) followed by density modification using RESOLVE (20). Finally, the statistics showed the space group to be P4₁22. The initial model was built automatically by RESOLVE, which assigned 90% of the amino acids. The rest of model was built manually using XtalView (21), and the refinement was carried out using CNS (22). The coordinates were checked, and the secondary structure was defined using PROCHECK. The structure of the holoenzyme was determined by molecular replacement using the program Molrep (23) with the structure of the apoenzyme (Protein Data Bank entry 1WGB) as a model. The refinement was carried out using Refmac5 (23). The electron density of FAD could be seen at the expected position, based on the structures of other TC-FDM family enzymes (cofactor flavin binding site). In addition, an unexpected electron density for another FAD was found (substrate flavin binding site). The electron density of the flavin mononucleotide portion of the substrate FAD was clearer than the other atoms, so we only assigned the flavin mononucleotide portion to the electron density, not FAD. The coordinates were then checked, and the secondary structure was defined using PROCHECK. The refinement statistics are summarized in Table 1.

RESULTS

Crystal Structure of Apo- and Holoenzymes

The structure of the apoenzyme was determined using the multiwavelength anomalous dispersion method (18) and refined at 2.0 Å resolution to an R-factor of 21.2% (R_free = 24.1%). In addition, the structure of the holoenzyme was determined using Molrep (23) and refined at 1.9 Å resolution to an R-factor of 20.0% (R_free = 22.8%). The overall structures of the apo- and holoenzymes are similar, with a root mean square deviation of 1.10 Å. TTHA0420 consists of an antiparallel β-barrel with a capping α-helix (α2) and forms a homodimer related by a molecular 2-fold rotation axis, which is a non-crystallographic symmetry operation (Fig. 1, A and B). A search for structurally similar proteins using the DALI program (24) revealed the structure of TTHA0420 to have homology with four previously described proteins: ferric reductase (FeR) from Archaeoglobus fulgidus (Protein Data Bank entry 1IOS, 24% sequence identity) (25), PheA2 from Geobacillus thermoglucosidasius (Protein Data Bank entry 1RZ1, 25% sequence identity) (26), HpaC_Tt from T. thermophilus HB8 (Protein Data Bank entry 2ECR, 30% sequence identity) (16), and HpaC_St from S. tokodaii (Protein Data Bank entry 2D37, 24% sequence identity) (5). In all five of these proteins, including TTHA0420, the core of the subunit consists of a six-stranded antiparallel β-barrel with a capping α-helix that interacts with the ribityl phosphate group of flavin. The holo-crystal structure includes both co-factor flavin and substrate flavin. The 2F_o − F_c electron density maps around two flavins are shown in Fig. 1, B and C. As far as we know, this is the first report showing the structure of substrate flavin bound to a member of this enzyme family. Despite the overall structural similarity of TTHA0420 to the other enzymes, its C terminus has a unique conformation not seen in the other enzymes (Fig. 1D).

FIGURE 1.

Overall structure of TTHA0420 from T. thermophilus HB8. A, structure of the apo-TTHA0420 dimer. The monomers are depicted in yellow and cyan, respectively. Four α-helices and 11 β-sheets are indicated in the yellow monomer. B, structure of holo-TTHA0420. Cofactor FAD (yellow) and substrate FAD (magenta) are depicted in stick representations. The 2F_o − F_c electron density map was drawn at 1 σ around each flavin. For clarity, the two other flavins on the cyan monomer are not shown. C, close-up of the two flavins in the binding region. D, ribbon diagrams of TTHA0420 (gray), FeR (yellow), PheA2 (cyan), HpaC_Tt (green), and HpaC_St (orange). The C-terminal five residues of TTHA0420 are shown in magenta. Molecular graphics figures were created using PyMOL (available on the World Wide Web).

Cofactor FAD Binding Site

The cofactor flavin binding site is essentially the same as that in other small component TC-FDM enzymes (5, 25–27). Cofactor FAD binds to TTHA0420 within a wide groove, with its isoalloxazine ring positioned at the dimer interface. The groove is surrounded by the α1, α2, and α3 helices and by the β1 and β2 strands. The isoalloxazine ring is situated just below Tyr-157 in the C terminus. The 2,4-pyrimidinedione moiety of the isoalloxazine ring forms a hydrogen bond network with the side chain atoms of Asn-34, Trp-35, Gly-49, and Lys-51 (Fig. 2A). The main chain and side chain nitrogen atoms of Asn-34 are hydrogen-bonded to the FAD N5 atom. The dimethylbenzene portion of the isoalloxazine ring is situated between the α2 and α3 helices and is surrounded by three hydrophobic residues: Leu-17, Phe-83, and Met-150. The ribityl phosphate moiety of FAD forms hydrogen bonds with the side chain atoms of Thr-32 and Phe-82. The FAD ribose and adenine moieties form hydrogen bonds with His-55.

FIGURE 2.

Cofactor and substrate FAD binding sites in TTHA0420. A, residues involved in the binding of cofactor FAD (yellow) to TTHA0420 are shown in a yellow and orange stick representation. The dimeric partner's Phe-41 is shown in a green stick representation. B, substrate FAD and cofactor FAD bound to TTHA0420 are shown in magenta and yellow stick representations, respectively. The dimeric partner's Phe-41 is shown in a green stick representation. Arg-11, His-129, and Tyr-157 are shown in yellow orange stick representations.

Substrate Flavin Binding

In the holoenzyme, we also observed bound substrate flavin, which had not been resolved in PheA2. The isoalloxazine ring of the substrate flavin is positioned on the dimeric partner's Phe-41 residue, and the 2,4-pyrimidinedione moiety of the N5 atom of the isoalloxazine ring forms a hydrogen bond network with the side chain atoms of Arg-11 (Fig. 2B). The interaction between the substrate flavin and the enzyme appears weak because it uses mainly the parallel π-π interaction between the substrate flavin's isoalloxazine ring and Phe-41 (dimeric partner) (Fig. 2B).

Modeling of NADH Binding

To identify the NADH binding site in TTHA0420, we initially attempted to co-crystallize TTHA0420 and NADH but were unable to do so. Therefore, we next compared the structure of the NADH binding site of TTHA0420 with those of PheA2 (26) and HpaC_Tt (16). NADH-dependent flavin reductases contain a conserved GDH (glycine, aspartic acid, histidine) motif in their active sites (28). Within the GDH motif, the histidine residue is thought to be particularly important for retention of the nicotine ring of NADH (28). Judging from the sequence alignment of the NADH-dependent flavin reductases identified using ClustalW, His-129 is conserved among all of these enzymes (Fig. 3A), which is indicative of its likely importance for NADH binding. On the other hand, Tyr-157 in the C terminus of TTHA0420 is not conserved in PheA2 or HpaC (this residue is only conserved in a subclass of flavin reductases; see below) and is situated over the cofactor flavin, probably preventing NADH binding. Therefore, using the structure of PheA2 as a template, we constructed a NAD-bound model with the C-terminal five residues deleted (Fig. 3B). This model suggests that the reduction from NADH via co-factor FAD proceeds if the C terminus is released from the enzyme.

FIGURE 3.

A, sequence alignment of TTHA0420 and other FDM family enzymes. The sequences of HpaC_Tt from T. thermophilus HB8, HpaC_St from S. tokodaii, PheA2 from G. thermoglucosidasius, and FeR from A. fulgidus were aligned using ClustalW. Fully conserved residues are shown as asterisks. The C-terminal five residues of TTHA0420 are shown in a box. The conserved residues in the flavin reductase subfamily of TTHA0420 are marked in red (see “Discussion” and supplemental Fig. S1). B, model of NAD docking in TTHA0420-ΔC5. The NAD is modeled into the site based on the structure of PheA2-NADH. The C-terminal five residues were deleted for the modeling (TTHA0420-ΔC5).

Flavin Reductase Activity of Wild-type TTHA0420

We initially measured the NADH-dependent flavin reductase activity of TTHA0420. To determine the enzyme specificity, FAD, FMN, NADH, and NADPH were tested as substrates. As found with PheA2 (7), the electron donor and acceptor with the highest activities were NADH and FAD, respectively. Almost no activity was observed with NADPH. The temperature optimum for the wild-type enzyme was 70 °C (Fig. 4A), and the pH optimum was 7.0 (data not shown).

FIGURE 4.

Measurement of TTHA0420 activity. A, the temperature optimum for wild-type TTHA0420 was determined using 20 mm potassium phosphate buffer (pH 7.0) containing 50 μm FAD and 200 μm NADH. The assay temperature was varied from 20 to 90 °C. B, temperature optimum for TTHA0420-ΔC5. The assay conditions were the same as in A. C, double reciprocal plots of the enzyme activity plotted as a function of FAD concentration at different NADH concentrations: 10 (●), 20 (■), 40 (▴), 80 (○), and 160 (□) μm. NADH:flavin oxidoreductase activity of TTHA0420-ΔC5 was determined at 40 °C in 20 mm potassium phosphate buffer (pH 7.0). D, double reciprocal plots of the enzyme activity plotted as a function of the NADH concentration at different FAD concentrations: 1 (●), 2 (■), 4 (▴), and 8 (○) μm. The assay conditions are the same as in C.

The Activity Measurements of TTHA0420-ΔC5 Mutants

The three-dimensional structure of TTHA0420 suggests that the C-terminal five residues stack on the NADH binding site. We speculated that this configuration keeps the enzyme activity low. To test that idea and to determine the function of C-terminal region, we designed and expressed a TTHA0420-ΔC5 mutant in which the five C-terminal residues (positions 155–159) were deleted. As with the wild-type enzyme, the temperature optimum for TTHA0420-ΔC5 was 70 °C (Fig. 4B). Moreover, at lower temperatures (20–40 °C), the activity of TTHA0420-ΔC5 was about 10 times higher than that of wild-type TTHA0420 (Fig. 4, A and B). This suggests that at high temperatures, the C-terminal region of TTHA0420 is released from the NADH binding site, leading to an increase in enzyme activity.

Catalytic Mechanism of TTHA0420

Initial velocity measurements aimed at assessing the dependence of the catalytic activity of TTHA0420 on FAD and NADH were performed to provide insight into the kinetics of the enzyme's catalytic mechanism. When the NADH concentration was varied at several fixed FAD concentrations, the double reciprocal plots produced were well fitted by a set of parallel lines (Fig. 4, C and D). This indicates that TTHA0420 makes use of the ping-pong bisubstrate-biproduct reaction mechanism. The calculated K_m values for FAD and NADH were 4.3 and 25 μm, respectively.

DISCUSSION

TTHA0420 Forms Subclass of Flavin Reductase

TTHA0420 was predicted to be a flavin reductase after it was found to contain a flavin reductase motif using Pfam. The sequence alignment, including PheA2 and HpaC, showed that the key motif is GDH(125–127). However, based on multiple sequence alignment with a wide array of other known flavin reductases, we also found that TTHA0420 belongs to a particular subclass of flavin reductases (supplemental Fig. S1). In a multiple sequence alignment using the top 60 sequences obtained in a protein Blast search, members of this subclass contain several characteristic features, including a YGG sequence (residues 157–159) in their C terminus, as well as Phe-41 (this is also conserved except the last one shown in supplemental Fig. 1) and Arg-11. Interestingly, 15 residues of 23 conserved residues are in close proximity to the two flavin binding sites (Fig. 5). This strongly suggests that the co-factor and substrate flavins are in the same configuration in this subclass of flavin reductases. It may also provide insight into the binding mode of substrate flavin in flavin reductase family enzymes, including PheA2 and HpaC.

FIGURE 5.

23 conserved residues in flavin reductase subfamily of TTHA0420 are shown (see “Discussion” and supplemental Fig. S1). 15 residues are in especially close proximity to the binding sites for two flavins (purple). Phe-41, Ser-40, Pro-43, Pro-44, and Leu-45 are from the dimer's partner (cyan except for Phe-41 (purple)).

Functional Role of C-terminal Five Residues and Free Flavin Generation Mechanism

The structural relation between the isoalloxazine rings of the cofactor and substrate flavins bound to TTHA0420 is unique in that they are different from other diflavin reductases, such as nitric-oxide synthase or P450 reductase (29) and l-proline dehydrogenase from Pyrococcus horikoshii or sarcosine oxidase from Corynebacterium (30, 31). Nonetheless, the dimethylbenzene portions are in close proximity: C7(substrate)–C8(cofactor) distance is 4.1 Å. In addition, Arg-11 and Phe-41 are particularly important for substrate flavin binding. Not only does Phe-41 support the binding of substrate flavin, it enables the dimethylbenzene portions of the two flavins to be situated in close proximity to one another (Fig. 2B). The role of the C terminus of TTHA0420 was inferred from the results of the functional and x-ray crystallographic analyses (Fig. 6). The structures of both apo- and holo-TTHA0420, which were revealed by x-ray crystallographic analysis at low temperature, showed the C terminus to be situated within the active site. We call this the on-state. This C terminus on-state appears to make NADH binding unlikely. Within the high temperature environment of T. thermophilus, however, the C terminus is much more flexible, and there should be an equilibrium between the on-state and the off-state. When the C terminus is in the off-state, NADH binds to the site, instead of Tyr-157, and acts as an electron donor for the reduction of cofactor flavin (Fig. 3B). At this stage, protonated His-129 (conserved His), which makes an ion pair with Asp-128, may play an important role for the counterpart of the anionic N1 atom of the isoalloxazine ring of FAD upon reduction of co-factor FAD. The C terminus can then return to the on-state position, enabling the oxidized substrate flavin to bind (Fig. 2B). Once bound, the substrate flavin is reduced as soon as the C terminus is off again. We suggest that the C terminus might alternate between the off-state for NADH binding and the on-state for the substrate flavin binding.

FIGURE 6.

Schematic representation of the generation of free flavin of TTHA0420 flavin reductase. The oxidation state and reduced state of co-factor flavin are shown in white and yellow, respectively. The oxidation state and reduced state of substrate flavin are shown in white and purple, respectively. The C-terminal region is shown in red. (See the “Discussion” for details.)

Finally, the mechanistic implication of the flavin reduction was summarized in Fig. 7. The reaction between the cofactor flavin and the pyridine nucleotide is a hydride transfer with one proton and two electrons. On the other hand, in the latter half-reaction from cofactor flavin to substrate flavin, the hydride transfer is not likely because the parallel and juxtaposed relationship among donor and acceptor is considered to be important (32). Therefore, we think that a two-step one-electron transfer mechanism is possible, as shown in nitric-oxide synthase or in the case of P450 reductase (33). In TTHA0420, two dimethylbenzene portions are closely situated within about 4 Å, similar to the FAD-FMN relationship in nitric-oxide synthase and P450 reductase, and the two dimethylbenzene portions seem to play an important role in the electron transfer.

FIGURE 7.

Proposed reaction mechanism in flavin reductase. I, His-129 is protonated by Asp-128 and Tyr-157 at first. II, reduction of cofactor flavin by the hydride transfer from NADH. The N1 atom of flavin is anionic, and the counterpart is cationic His-129. The distance between the N1 atom of flavin and the proton of His-129 is about 3.5 Å. III, substrate flavin binding instead of NAD. Tyr-157 locates again at the first position, and His-129 is deprotonated. IV, oxidation of the cofactor flavin and reduction of substrate flavin by two-step one-electron transfer via dimethylbenzene portions with a 4-Å distance. Substrate flavin (reduced) is produced via cofactor anionic reduced flavin.

Possible Role in Iron Reduction

There is no obvious gene around the TTHA0420 gene for an oxygenase component that would utilize free reduced flavin. We therefore speculate that TTHA0420 might play a role in the reduction of iron from Fe(III) to Fe(II) in ferric complexes or iron proteins. For example, evidence from microbial systems indicates that a flavin reductase is involved in the reductive activation of the iron center of ribonucleotide reductase (14, 34, 35) and in reducing ferrisiderophores for iron release (4, 36). Another interesting enzyme is FeR, which has the best similarity score with TTHA0420. FeR catalyzes iron reduction and is thought to have a binding site for Fe(III), although this site has not yet been observed on x-ray crystallography (25). TTHA0420 could act as a ferric reductase with a different mechanism of ferric reduction, one that makes use of free reduced flavin. In summary, based on the three-dimensional structure and functional analysis of TTHA0420-ΔC5, we conclude that TTHA0420 from T. thermophilus HB8 is flavin reductase, having a unique flexible C-terminal region, which is important for the substrate flavin binding.