INTRODUCTION

The antibiotic valanimycin (Fig. 1) is a naturally occurring azoxy compound isolated from the fermentation broth of Streptomyces viridifaciens MG456-hF10 (1). In addition to antibacterial activity, valanimycin exhibits potent antitumor activity against in vitro cell cultures of mouse leukemia L1210, P388/S (doxorubicin-sensitive), and P388/ADR (doxorubicin-resistant) (1). Valanimycin is a member of a growing class of natural products that contain the azoxy group. This group now includes the cycad toxins macrozamin and cycasin (2-5), the carcinogen elaiomycin (6-8), the antifungal agents maniwamycins A and B (9), and the nematocidal compounds jietacins A and B (10). A characteristic structural feature of these compounds is the presence of an N-N bond. A wide variety of natural products have been reported to contain N-N linkages (11), but little is known at present about the biochemistry of N-N bond formation.

Investigations of the biosynthesis of valanimycin have shown that it is derived from L-serine and L-valine and that valine is incorporated via the intermediacy of isobutylamine and isobutylhydroxylamine (Fig. 1) (12). Evidence has also been deduced for the intermediacy of a hydrazine derivative formed by the condensation of isobutylhydroxylamine with L-serine (13).

It has been shown that two enzymes are required for the conversion of isobutylamine to isobutylhydroxylamine (14). One of these enzymes is a flavin reductase that provides reduced flavin for the second enzyme, isobutylamine N-hydroxylase (IBAH).¹ IBAH utilizes the reduced flavin and molecular oxygen to catalyze the hydroxylation of isobutylamine to isobutylhydroxylamine. IBAH has been purified to homogeneity (14), and amino acid sequences obtained from the purified protein have been used to clone the gene (vlmH) encoding IBAH (15). Overexpression of this gene in Escherichia coli yielded a protein that exhibited IBAH activity in the presence of a partially purified flavin reductase from S. viridifaciens or an FMN reductase of Vibrio fischeri (15). Since the genes encoding secondary metabolic pathways in bacteria are generally clustered, cloning of the vlmH gene has allowed access to the gene (vlmR) encoding the flavin reductase gene of S. viridifaciens. In this report, we describe the cloning, analysis, and overexpression of this flavin reductase gene.

MATERIALS AND METHODS

Bacterial Strains, Phages, and Plasmids

The bacterial strains, phages, and plasmids are listed in Table I.

Table I. Bacterial strains, phages, and plasmids used Strain or plasmid Relevant properties Source or Ref. Strains   E. coli   DH5 F 80dlacZ M15 (lacZYA-argF)U169 deoR recA1endA1 hsdR17 (rk, mk+) phoA supE44 thi-1 gyrA96 relA1 Life Technologies, Inc.   DH10B FmcrA (mrr-hsdRMS-mcrBC) 80dlacZM15 lacX74 deoR recA1 araD139 (ara, leu)7697 galU galK rpsL nupG Life Technologies, Inc.   BL21 (DE3) F ompT hsdSB(rBmB)gal dcm (DE3) Novagen Plasmids   pET-24a(+) Multicloning site vector; Kanr, T7lac promoter Novagen   pET-28b(+) Multicloning site vector; Kanr, T7lac promoter Novagen   pPROEXHTa Multicloning site vector; Ampr, trc promoter Life Technologies, Inc.   pB634E3 7-kb EcoRI-EcoRI insert from phage 6.3.4 in pBluescriptII SK(); Ampr (15)   pB634E3Sac1 3-kb EcoRI-SacI insert from pB634E3 This study   pET24 (vlmRN1RC1) 0.6-kb NdeI-XhoI vlmR gene insert in pET-24a (+); Kanr, T7lac promoter This study   pET28 (vlmRN1RC1) 0.6-kb NdeI-XhoI vlmR gene insert in pET-28b (+); Kanr, T7lac promoter This study   pPROEX (vlmRN3RC3) 0.6-kb blunt end-HindIII vlmR gene insert in pPROEXHTa; Ampr, trc promoter This study

Media and Bacteriological Techniques

E. coli strains were grown in LB medium at 37 °C. Selection was made with 100 µg of ampicillin or 30 µg of kanamycin per ml of LB agar or other medium.

DNA Methods

Plasmid DNA was purified with a QIAprep Spin Plasmid Kit. DNA fragments were isolated from agarose gels using the QIAquick Gel Extraction Kit. PCR products were separated on agarose gels and purified from the gel. Digestion with restriction endonucleases and ligation experiments were carried out by standard procedures. Automated DNA sequencing was performed with an Applied Biosystems DNA sequencer at the Molecular Genetics Core Facility, University of Texas Houston Medical School, using universal and synthetic oligonucleotide primers.

Transformations

Transformations were carried out using commercially available competent E. coli cells. The procedures followed the protocols recommended by the manufacturers.

PCR Protocols

A 50-µl PCR reaction mixture contained 10 mM Tris·HCl (pH 8.3 at 25 °C), 50 mM KCl, 1.5 mM MgCl₂, 0.001% (w/v) gelatin, 200 µM each dNTP, 0.5 µM each PCR primer, DNA template, and Taq DNA polymerase (1.25 units). PCR tubes (0.65 ml) containing all the components except the polymerase were incubated at 94 °C for 0.5 min in a PCR personal cycler (Biometra). The PCR was initiated by addition of the polymerase to the PCR tubes. The temperature program was as follows: 94 °C, 0.5 min; 94 °C, 0.5 min; 65 °C, 1 min; 72 °C, 3 min; 30 cycles; 72 °C, 7 min.

A 100-µl PCR reaction mixture contained 20 mM Tris·HCl (pH 8.8 at 25 °C), 2 mM MgSO₄, 10 mM KCl, 10 mM (NH₄)₂SO₄, 0.1%(v/v) Triton X-100, 100 mg/ml bovine serum albumin, 200 µM each dNTP, 0.5 µM each PCR primer, DNA template, and 3.75 units of cloned Pfu DNA polymerase (Stratagene). PCR tubes (0.65 ml) containing all the components except the polymerase were incubated at 95 °C for 0.5 min in a PCR personal cycler (Biometra). The PCR was initiated by addition of the polymerase to the PCR tubes. The temperature program was as follows: 95 °C, 0.5 min; 95 °C, 0.5 min; 60 °C, 0.5 min; 72 °C, 3 min; 28 cycles; 72 °C, 8 min.

Sequence Analysis

DNA sequence assembly and restriction site analysis were performed with Sequencher, Macintosh version 3.0. Sequence analyses were performed with the Wisconsin GCG Package, version 8.1-UNIX, including MAP, BESTFIT, ISOELECTRIC, BLAST, MOTIFS, CODONPREFERENCE, and PILEUP. Pattern searches were also carried out with PSCAN at the ISREC-Server,² The computations for BLAST were performed at the NCBI using the BLAST network service. Alignment of amino acid sequences was achieved as follows. First, an initial multiple sequence alignment of the sequences was created with PILEUP. The alignment was then manually adjusted according to information obtained from the BLAST searches. The aligned sequences thus obtained were displayed using SeqVu, Macintosh version 1.01.

Construction of Expression Plasmids

Two PCR primers (RN1, 5-AAA CAT ATG ACC CCC TCG GCG GCT GCC AC-3; and RC1, 5-TGG CTC GAG CTA CTC AAG GGA GGT GAC AGC GC-3) were designed to allow the introduction of an NdeI site on the upstream side of the gene and an XhoI site on the downstream side. A DNA fragment (0.6 kb) containing the vlmR gene was amplified by PCR (protocol A) from pB634E3Sac1 using RN1 and RC1 as primers. The PCR product was purified from an agarose gel and then digested with NdeI and XhoI. The resulting fragment was isolated from an agarose gel and cloned into NdeI-XhoI-digested plasmid vector pET24a (+). The resulting recombinant plasmid pET24(vlmRN1RC1) was amplified in E. coli DH5 cells, and the insert was sequenced to ensure that the correct construction had been obtained. The pET24(vlmRN1RC1) construct was then introduced into E. coli BL21(DE3) for expression of the vlmR gene.

Two PCR primers (RN3, 5-ATG ACC CCG TCT GCT GCT GCT ACC GGC C-3; and RC3, 5-TTT AAG CTT CTA CTC AAG GGA GGT GAC AGC AGC GCT GGG-3) were designed to allow the introduction of an HindIII site on the downstream side. In RN3, 4 bases were changed (underlined) so that the codons for the first eight amino acids corresponded to those preferred by E. coli. A DNA fragment (0.6 kb) containing the vlmR gene was amplified by PCR (protocol B) from pB634E3Sac1 using RN3 and RC3 as primers. The purified PCR product was then digested with HindIII, and the resulting fragment was isolated from an agarose gel. The purified restriction fragment was cloned into EheI-HindIII-digested pPROEXHTa vector. The resulting recombinant plasmid pPROEX(vlmRN3RC3) was amplified in E. coli DH5 cells, and the insert was sequenced to ensure that the correct construction had been obtained. The pPROEx(vlmRN3RC3) construct was then introduced into E. coli DH10B for expression of the vlmR gene.

The expression plasmid pET24(vlmRN1RC1) was digested with NdeI and XhoI. The resulting 0.6-kb fragment was isolated from an agarose gel and then cloned into NdeI-XhoI digested pET-28b (+) vector. The resulting recombinant plasmid pET28(vlmRN1RC1) was amplified in E. coli DH5 cells, and the insert was sequenced to ensure that the correct construction had been obtained. The pET28(vlmRN1RC1) construct was then introduced into E. coli BL21(DE3) for expression of the vlmR gene.

Heterologous Expression of vlmR in E. coli

Expression of VlmR from pET24(vlmRN1RC1) and pPROEX(vlmRN3RC3) was carried out following the guidelines of the manufacturers of the expression vectors. Expression of VlmR from pET28(vlmRN1RC1) used the following procedure. LB broth (100 ml, 30 µg/ml kanamycin) was inoculated with 4 ml of frozen stock of E. coli BL21(DE3) cells harboring pET28(vlmRN1RC1) and incubated at 37 °C at 300 rpm for about 1.5 h. The culture was stored at 4 °C overnight. The next day, the cells were collected by centrifugation (4,000 × g, 10 min) and resuspended in LB broth (5 ml, 30 µg/ml kanamycin). Four ml of this cell suspension was used to inoculate LB broth (4 × 200 ml, 30 µg/ml kanamycin, in four 1-liter flasks). The broth was incubated at 37 °C and 300 rpm for about 2 h to reach an A₆₀₀ of about 0.7. IPTG (8 ml of a 100 mM solution) was then added and the culture was incubated at 37 °C and 300 rpm for an additional 3 h. The cells were collected by centrifugation and stored at 20 °C.

Protein Purification

All of the steps in the purification protocol were performed at room temperature, except for the initial sonication and centrifugation which were carried out at 4 °C. Cells (2.8 g wet weight) from 800 ml of a culture of BL21(DE3) containing pET28(vlmRN1RC1) were suspended in about 60 ml of ice-cold buffer D (20 mM Bis-Tris·HCl, 100 mM NaCl, pH 8.0, 10% w/v glycerol) and sonicated for 4 min (70% power, Branson sonifier) while cooling in an ice bath. After removal of cell debris by centrifugation at 24,000 × g for 10 min, 59 ml of crude cell extract were obtained (Table II). The extract was loaded onto a TALON (CLONTECH) metal affinity column (10-ml bed volume, 10 mm inner diameter) equilibrated with buffer D. The column was washed with 20 ml of buffer D, 50 ml of buffer E (20 mM Bis-Tris·HCl, 100 mM NaCl, pH 8.0), 50 ml of buffer F (20 mM Bis-Tris·HCl, 20 mM imidazole, 100 mM NaCl, pH 8.0), and 8 ml of buffer G (20 mM Bis-Tris·HCl, 300 mM imidazole HCl, 100 mM NaCl, pH 8.0). The overexpressed protein was then eluted with 30 ml of buffer G, and the eluant was mixed with 3 ml of glycerol to give a solution of the purified His-tag protein (hFR) (33 ml, 49.5 mg of protein). Cleavage buffer (3.6 ml) (200 mM Tris·HCl, pH 8.4, 1.5 M NaCl, 25 mM CaCl₂) and biotinylated human thrombin (31.7 units) (Novagen) were then added to the solution of the purified hFR, and the mixture was incubated for 3 h at 25 °C to remove the His-tag. At the end of the incubation, 750 µl of 50% v/v streptavidin-agarose slurry were added to the mixture, and incubation at 25 °C was continued for 30 min. The agarose was then removed by filtration. The filtrate was desalted and the buffer exchanged with buffer D by using a Sephadex G-25 column (68-ml bed volume, 20 mm inner diameter). The desalted sample was loaded onto a TALON column (10-ml bed volume, 10 mm inner diameter) equilibrated with buffer D, and the column was then washed with 20 ml of buffer D, 50 ml of buffer E, and 8 ml of buffer F. The flavin reductase derived by thrombin cleavage (tFR) was eluted with 32 ml of buffer F, and 4 ml of glycerol was added to the eluant.

Table II. Purification of recombinant flavin reductase The purification utilized cells from 800 ml of an E. coli culture (2.8 g wet weight). Step Total protein Total units Specific activity mg µmol NADPH/min units/mg Crude extract 145 12 0.08 TALON fraction 49.5 5287 107 Thrombin treatment 25.7 3611 141

Inhibition of the FAD Reductase Activity by Crude Cell Extract

Crude cell extract from E. coli BL21(DE3) containing pET28(vlmRN1RC1) (1 ml, protein concentration, 3.2 mg/ml) was loaded onto a CentriCon-10 concentrator (10,000-Da cutoff) and spun at 3000 × g for 1 h. About 200 µl of filtrate was obtained, and the protein was retained in the upper chamber. Assays were then carried out using 0.35 µg of tFR in the presence of 5 µl of either the filtrate or the concentrated protein solution. The filtrate did not inhibit the reductase activity, whereas the concentrated protein solution completely inhibited the reductase activity (less than 2% residual activity). Addition of 5 µl of the unconcentrated crude cell extract to the assay mixture also completely inhibited reductase activity.

Test for FAD Loss during TALON Affinity Chromatography of hFR

To 5 ml of a crude cell extract from E. coli BL21(DE3) containing pET28(vlmRN1RC1) with a protein concentration of 3.2 mg/ml, 68 µl of a solution of FAD (14.7 mM) in buffer D was added to give a final FAD concentration of 200 µM. After being incubated at 25 °C for 3 h, the mixture was loaded onto a TALON column (1-ml bed volume, 10 mm inner diameter) equilibrated with buffer D. The column was washed with 2 ml of buffer D, 5 ml of buffer E, 5 ml of buffer F, and 0.5 ml of buffer G. The hFR fraction was then eluted with 3 ml of buffer G. The hFR concentration was determined to be 28.6 µM (0.67 mg/ml), and the A₄₅₀ was found to be 0.1718. Assuming that the enzyme-bound FAD has the same absorption coefficient as free FAD (11.3 mM¹ cm¹), the concentration of FAD in the hFR fraction is calculated to be 15.2 µM.

Spectral Measurements

Absorption spectra were measured with a Hewlett-Packard 8452A Diode Array Spectrophotometer equipped with a thermostated cell holder. Enzyme activities were determined with a 5-s cycle time and 0.5-s integration time. The concentrations of substrates and cofactors were determined spectrophotometrically using the following absorption coefficient data: NADH, 6.22 mM¹ cm¹ (340 nm); NADPH, 6.22 mM¹ cm¹ (340 nm); FAD, 11.3 mM¹ cm¹ (450 nm), 4.78 mM¹ cm¹ (340 nm); FMN, 12.2 mM¹ cm¹ (450 nm), 5.45 mM¹ cm¹ (340 nm); riboflavin, 12.2 mM¹ cm¹ (450 nm), 5.85 mM¹ cm¹ (340 nm).

Determination of Hydrogen Peroxide Concentration

The sample solution (0.5 ml) was mixed with 0.5 ml of a solution containing 2.5 mM 4-aminoantipyrine and 170 mM phenol in a 1.5-ml cuvette, and a background spectrum was taken. Horseradish peroxidase (0.5 µl) (5.25 units/µl) (Sigma, Type X) was added, and the mixture was incubated for 3 min at 25 °C. The A₅₁₀ was measured to determined the hydrogen peroxide concentration in the mixture using a value for ₅₁₀ of 6.58 mM¹ cm¹ (16).

Enzyme Assay

All enzyme assays were carried out at 25 °C. NADPH oxidation activity was measured by monitoring the initial rate of the decrease in absorption at 340 nm. The standard assay mixture (1 ml) contained 20 mM Bis-Tris·HCl, pH 7.5, 40 µM FAD, 100 µM NADPH, and the enzyme sample. One unit of enzyme activity was defined as amount of the enzyme required to oxidize 1 µmol of NADPH per min. The ability of the FAD reductase to support the N-hydroxylation of IBA was assayed in the presence of a large molar excess of the overexpressed N-hydroxylase. The assay was performed at 30 °C for 30 min in 100 µl of 100 mM Bis-Tris·HCl buffer, pH 7.5, in a 1.7-ml microcentrifuge tube. The assay mixture contained 100 mM NaCl, 10 µM FMN, 4 mM NADPH, 10 mM isobutylamine, 0.4 units/µl catalase, 1.40 µM overexpressed isobutylamine N-hydroxylase, and 80 nM tFR. After the incubation, the concentration of isobutylhydroxylamine was determined to be 168 µM by HPLC analysis (15).

FAD Reductase Activity and Effect of Molecular Oxygen

An incubation mixture (3 ml) containing 20 mM Bis-Tris·HCl, pH 7.5, 41 µM FAD, and 99 µM NADPH was placed in a cuvette (3.5 ml, 1-cm light path), equipped with a magnetic stirbar and rubber septum. Prepurified argon was introduced into the solution by means of a needle through the septum, and a second needle inserted into the septum was connected to a tube serving as the outlet. The end of the outlet tube was immersed in mineral oil. The incubation mixture was flushed with argon for 4 min to remove oxygen. The UV-visible spectrum was then determined (Fig. 2, curve A). A solution of tFR (30 µl, 2.1 µg, 3.0 units) was injected into the cuvette to initiate the enzymatic reaction, and the mixture was stirred at room temperature under argon. The yellow color disappeared immediately. After 4 min, another spectrum was determined (Fig. 2, curve B). This spectrum indicated that the concentration of the remaining FAD was 4 µM, and that of the remaining NADPH was 66 µM. Oxygen gas was then bubbled into the incubation mixture for 2 min, whereupon the yellow color returned. The UV-visible spectrum of the reoxidized solution (Fig. 2, curve C) indicated that the concentration of FAD was 41 µM, whereas that of the remaining NADPH was 4 µM. By using the peroxidase assay, the hydrogen peroxide concentration in the reoxidized incubation mixture was found to be 89 µM.

Identification of the Flavin Component of the Reductase

About 0.5 mg of the His-tag fusion protein expressed in E. coli DH10B from pPROEX(vlmRN3RC) was suspended in a mixture of 100 µl of 20 mM sodium phosphate buffer, pH 7.5, and 0.9 ml of methanol. The suspension was heated at 95 °C for 15 min. After centrifugation, the supernatant was dried with a Speed-Vac concentrator. Water (10 µl) and methanol (10 µl) were added to the residue, and the solution was analyzed by TLC on a K6F silica gel plate using the upper layer of an n-butyl alcohol:glacial acetic acid:water (4:1:5) mixture as the developing solvent. Aqueous solutions of FAD, FMN, and riboflavin were spotted as references, and the compounds were visualized with UV light. Under these conditions, FAD, FMN, and riboflavin exhibited R_F values of 0.06, 0.16, and 0.43, respectively. The enzyme-derived sample gave a single fluorescent spot with an R_F of 0.06, indicating that the flavin present in the recombinant flavin reductase is FAD.

Determination of the Stoichiometry and Dissociation Constant for Binding of FAD to tFR

A 1.5-ml mixture of tFR (0.44 mg/ml, 20.5 µM) and FAD (116.6 µM) in buffer F was dialyzed against 500 ml of buffer D for 2 days at 25 °C using a Pierce Slide-A-Lyzer 10,000 M_r cutoff dialysis cassette. The protein concentration of the resulting solution was determined to be 0.51 mg/ml, i.e. 23.7 µM. Using the solution outside the dialysis cassette as background, the UV-visible spectrum of the protein solution was determined (Fig. 3, curve B). The spectrum was similar to that of free FAD (Fig. 3, curve A) and gave an A₄₅₀ of 0.161. Assuming that the enzyme-bound FAD has the same absorption coefficient as free FAD (11.3 mM¹ cm¹), the concentration of enzyme-bound FAD is calculated to be 14.2 µM. Thus 60% of the tFR bound FAD. Therefore. the concentration of free FAD was calculated to be 0.3 µM, and K_d was about 0.5 µM¹.

1

1

A 2-ml mixture of tFR (0.44 mg/ml, 20.5 µM) and FAD (100 µM) was incubated at 25 °C for 30 min. The mixture was loaded into a CentriCon-10 concentrator (10,000 M_r cutoff) and spun at 3000 × g for 15 min. On the basis of the assumption that the enzyme-bound FAD has the same absorption coefficient as free FAD, the concentrations of FAD in the filtrate and retentate were determined from the A₄₅₀ to be 90.0 and 125.9 µM, respectively. Thus the concentration of enzyme-bound FAD was 35.9 µM. The protein concentration of the retentate was determined to 0.83 mg/ml, i.e. 38.6 µM. Therefore 93% of the tFR bound FAD at a saturating concentration of FAD, suggesting that 1 mol of FAD was bound per mol of tFR monomer.

One-half a milliliter of a stock solution of tFR in buffer D (12.64 µM) was mixed with 0.5 ml of an FAD stock solution in buffer D, and the mixture was incubated at 25 °C for 20 min. Five such mixtures were made in triplicate from different FAD stock solutions (5, 6, 7, 9, and 18 µM). The final concentrations of tFR and FAD ([E]_t and [FAD]_t) were calculated from those of the stock solutions and confirmed by measurement of the A₄₅₀. Each mixture (15 total) was loaded onto a CentriCon-10 concentrator (10,000 M_r cutoff) and spun at 4000 × g for 3 min so that about 0.15 ml of filtrate was obtained. The concentration of FAD in the filtrate ([FAD_f) was determined from the A₄₅₀, and the concentration of enzyme-bound FAD ([FAD]_b) as [FAD]_t  [FAD]_f. Assuming that the total concentration of FAD binding sites for tFR is [BS]_t, then [FAD]_b = [BS]_t  ([FAD]_b/[FAD]_f) × K_d. Scatchard analysis ([FAD]_b plotted against [FAD]_b/[FAD]_f) (Fig. 4) gave a K_d of 0.70 ± 0.05 µM and a [BS]_t of 6.35 ± 0.1 µM. Since the concentration of tFR was 6.32 µM, the ratio of FAD binding sites to tFR was 1.0:1. This suggests that 1 mol of tFR contains 1 FAD binding site.

The binding of FAD to tFR was also examined by a stopped-flow study with an MC 200 Monochromator equipped with a MilliFlow Stopped Flow Reactor (SLM Instruments, Inc.). One syringe contained a solution of tFR (4 mg/ml) in buffer D, and a solution of FAD in buffer D was placed in the other syringe. The stopping syringe was set at 0.15 ml. The two solutions were rapidly mixed in a 1:1 volume ratio, and the fluorescence intensity at 530 nm (excitation wavelength 450 nm) was monitored as a function of time. The integration time was 0.3 ms. For each FAD concentration, 10 reactions were carried out, and the data were averaged. To determine the value of the observed binding constant k_obs, the equation F = F  F × exp(k_obs × t) + F_S (where F is fluorescence intensity at time t; F is fluorescence intensity at infinite time; F_S is fluorescence intensity difference between the fluorescence at infinite time and at time 0) was used for nonlinear fitting. Four different FAD concentrations were used. According to the equation k_obs = k₁ + k₁ × [FAD], the association rate constant k₁ and the dissociation rate constant k₁ determined from Fig. 5 were 10.3 ± 0.2 µM¹ s¹ and 9.7 ± 4.4 s¹, respectively. Thus, K_d was calculated as 0.9 µM.

Spectral Properties of the FAD-Reductase Complex

tFR in buffer F containing glycerol was desalted and the buffer exchanged with buffer D by using a PD-10 column. The resulting tFR solution had a concentration of 22.9 µM (0.49 mg/ml). Two ml of the tFR solution was mixed with 6 µl of 7.22 mM FAD solution to give a FAD final concentration of 21.66 µM. Based upon a K_d of 0.70, the concentration of the enzyme-FAD complex in the FAD/tFR solution was calculated to be about 18.6 µM. A 21.66 µM FAD solution in buffer D was prepared, and spectra of the FAD/tFR, tFR, and FAD solutions were determined using buffer D as background. The fluorescence measurements were carried out with an Aminco-Bowman Series 2 Luminescence Spectrometer using identical conditions for all three solutions. CD spectra for all three solutions were measured with a Aviv Circular Dichroism Spectrometer (model 62ADS). All the spectra were taken at 25 °C.

Native Molecular Weight Determination

The native molecular weight for tFR was determined by using a Superose 6 FPLC gel filtration column with 50 mM sodium phosphate buffer, pH 7.0, at a flow rate of 0.5 ml/min. The column was calibrated by using Bio-Rad gel filtration standards.

Other Methods

SDS-PAGE was performed according to the method of Laemmli (17). A sample of total E. coli proteins was obtained by suspending the cells in SDS-gel loading buffer and incubating the mixture at 90 °C for 5 min. Isoelectric focusing (pH range 9-11) was performed according to the method of Robertson et al. (18). After electrophoresis, the gels were immersed in staining solution (0.25% (w/v) Coomassie Brilliant Blue R250 in methanol:distilled water:acetic acid (4.5:4.5:1.0)) and gently agitated for 10 min. The gels were then destained with destaining solution (staining solution minus Coomassie Brilliant Blue R250) until the protein bands showed clearly. Protein concentrations were determined by the method of Bradford (19) using bovine serum albumin as the standard.

RESULTS

Plasmid pB634E3 containing the vlmH gene (15) was digested with SacI to give three fragments that were cloned into pBluescript II SK() to give pB634ESac1, pB634ESac2, and pB634ESac3. The complete DNA sequence of the insert in pB634E3Sac1 was then obtained using universal and designed primers. The sequence analysis indicated the presence of three complete open reading frames (ORFs). All three ORFs were in the same orientation, and the first and second ORFs were in the same reading frame. The second and third ORFs (492 and 585 base pairs, respectively) appeared to be translationally coupled (20) (Fig. 6). The first ORF (1135 base pairs, 310-1446) corresponded to the previously cloned vlmH gene. A BLAST analysis of the deduced amino acid sequences for the second and third ORFs indicated that the protein derived from ORF2 exhibits a low similarity to cytochrome c oxidases, and the protein derived from ORF3 exhibits similarities to several flavin reductases. This suggested that ORF3 might encode a flavin reductase. Accordingly, ORF3 was named vlmR. The vlmR gene encoded a polypeptide of 194 amino acids (Fig. 7), with a calculated molecular mass of 21,265 daltons and a calculated isoelectric point of 10.2. The vlmR gene exhibited an overall G + C content of 67.7%, with an average G + C content of 78.5% at the third codon position. These values are somewhat lower than the average values for Streptomyces genes (21). No recognizable ribosome binding site was found near the start codon of either ORF2 or the vlmR gene.

To establish that the vlmR gene encodes a flavin reductase, the gene was cloned into the expression vector pET-24a (+). The recombinant vector containing the vlmR gene, pET24(RN1RC1), was introduced into E. coli BL21(DE3) and expression was induced with IPTG. Although this vector system uses the powerful T7 promoter, only a very low level expression was obtained as judged by SDS-PAGE. Changing bases in the 5-region of the gene to convert the first eight codons of the vlmR gene to those found in highly expressed E. coli genes had no beneficial effect on protein expression. The vlmR gene was next cloned into the expression vector pPROEXHTa to give pPROEX(RN3RC3) which was introduced into E. coli DH10B. This construct also utilized E. coli codons for the first eight amino acids. On induction with IPTG, the VlmR protein was expressed from this construct at a high level and in soluble form as an N-terminal His-tag fusion protein which was then purified with a TALON affinity column. The purified fusion protein displayed a very weak yellow color and exhibited flavin reductase activity. However, this form of the VlmR protein exhibited a tendency to precipitate from solution.

Since it appeared that the presence of an N-terminal His-tag leader sequence allowed expression of the vlmR gene, the vlmR gene was cloned into pET-28b (+) to give pET28(vlmRN1RC1). In this construct, the codon usage of the vlmR gene was not altered. The pET-28 vector encodes an N-terminal His-tag and a 20 amino acid spacer that contains a thrombin cleavage site (MGSSHHHHHHSSGLVPRGSH). pET28(vlmRN1RC1) was introduced into E. coli BL21(DE3) and expression induced with IPTG. After induction, a soluble protein (hFR) with a mass of about 26 kDa (calculation = 23.5 kDa) was produced at a very high level (Fig. 8). The pET28(vlmRN1RC1) construct was used to prepare larger amounts of the VlmR protein since it gave a higher yield of soluble protein than the pPROEx(RN3RC3) construct.

The VlmR protein expressed from pET28(vlmRN1RC1) was purified from the crude cell extract in one step, using a TALON affinity column (Table II). The purified fusion protein constituted about 35% of the total soluble protein. The purified fusion protein was shown to possess very high flavin reductase activity and was able to provide reduced flavin to IBAH for IBA hydroxylation. The N-terminal His-tag was then removed by thrombin cleavage. The cleavage conditions were controlled so that only about 70% of the fusion protein was digested to avoid possible second site cleavage. The thrombin-treated flavin reductase (tFR) was separated from the fusion protein and His-tag peptide fragment using a TALON column (Table II and Fig. 8). The purified tFR also exhibited very high flavin reductase activity, and it was able to provide reduced flavin to IBAH for IBA hydroxylation. Control experiments indicated that the high molecular weight fraction of the crude extract from the same strain strongly inhibits flavin reductase activity.

The subunit molecular mass of the tFR was determined to be ~23 kDa by SDS-PAGE. The native molecular mass of the tFR was measured to be ~36 kDa by gel filtration, which suggests that the protein exists as a homodimer. The pI of tFR was determined by isoelectric focusing to be 10.0. The nature of the thrombin cleavage site in the pET-28b-(+) vector is such that three amino acids (Gly-Ser-His) will remain at the N terminus of VlmR giving a protein with a predicted mass of 21,547 daltons and a calculated isoelectric point of 10.2.

The His-tag form of the flavin reductase expressed from the pPROEXHTa vector in E. coli DH10B exhibited a very weak yellow color, suggesting the presence of a bound flavin. The bound flavin was identified as FAD by denaturation of the protein and analysis of the released flavin by thin layer chromatography. In contrast, the purified flavin reductase obtained from the pET28(vlmRN1RC1) construct in E. coli BL21(DE3) was a colorless protein with no absorption between 300 and 800 nm. The stoichiometry of FAD binding to the apo form of tFR was determined by three methods. All three methods indicated that approximately 1 mol of FAD was bound per mol of tFR monomer. A Scatchard analysis (Fig. 4) indicated a FAD dissociation constant of 0.7 µM, whereas measurement of the dissociation constant by stopped flow methods gave a value of 0.9 µM. A plot of k_obs versus [FAD] obtained from the stopped flow experiments is shown in Fig. 5. The fact that FAD was a substrate rather than a cofactor for the enzyme was demonstrated by the complete reduction of an excess of FAD by the enzyme by excess NADPH under anaerobic conditions (Fig. 2). When oxygen was readmitted to the incubation mixture, the absorption at 450 nm due to FAD reappeared, and H₂O₂ was produced. Apparent kinetic data for substrates were determined at fixed concentrations of the second substrate (Table III). The kinetic data indicated that NADPH was a much better substrate than NADH and that the enzyme preferred FAD over FMN and riboflavin. At 40 µM FAD, the enzyme activity exhibited a linear dependence on NADH over a concentration range from 50 to 300 µM, and no evidence of saturation was observed. Since K_m and V_max could not be determined under these conditions, k_cat/K_m was calculated from the slope of the line generated by a double-reciprocal plot and from the enzyme concentration.

Table III. Apparent kinetic data for substrates Substrate Second substrate kcat Km kcat/Km s1 µM s1/µM1 NADPHa 40 mM FAD 62 32  ± 2 1.9 NADHb >300 1.8 × 103 FADa 100 mM NADPH 51 3.6  ± 0.4 14.2 FMNa 45 8.4  ± 1 5.4 Riboflavina 6.7 11.1  ± 2 0.6 a Used 0.173 µg (133 units/mg) of tFR in a 1-ml reaction volume. b Used 1.73 µg (133 units/mg) of tFR in a 1-ml reaction volume and an NADH concentration of 50-300 µM.

The crude cell extract from E. coli BL21(DE3) pET28 (vlmRN1RC1) displayed a faint yellow color, but purified hFR was colorless and exhibited no UV absorption between 300 and 800 nm. To examine the possible loss of bound FAD during purification, a large molar excess of FAD was added to the crude cell extract before TALON affinity chromatography. After chromatography, the concentration of hFR was found to be 28.6 µM, and the FAD concentration was 15.2 µM. Consequently, it appears that only a portion of enzyme-bound FAD is removed during the purification process, which suggests that most of the reductase is expressed in the apo form. Since the reductase was expressed at a high level, it is conceivable that the host E. coli might not be able to produce sufficient FAD to convert all of the overexpressed enzyme into the holo form. An experiment using the same expression system and a longer expression time (5.5 h) also produced hFR that contained no FAD after purification.

Purified tFR exhibits no UV absorption between 300 and 800 nm, no fluorescence emission at an excitation wavelength of 450 nm, and a weak and featureless CD spectrum between 300 and 800 nm. Binding of FAD to tFR causes a slight change in the UV spectrum of FAD (Fig. 3, curve B). A difference spectrum between free FAD and tFR-bound FAD exhibits negative peaks at 490, 460, and 356 nm and a positive peak at 396 nm (Fig. 3, curve C). In the experiments for the determination of K_d (method 3), the FAD concentrations of the FAD/tFR mixture derived from the concentration of FAD stock solution and from the A₄₅₀ of the mixture using an ₄₅₀ of 11.3 mM¹ cm¹ were consistent with each other. This suggests that tFR-bound FAD has the same absorption coefficient at 450 nm as free FAD.

The binding of FAD to tFR leads to a dramatic increase in the intensity in the fluorescence emission spectrum of FAD (Fig. 9). However, the fluorescence excitation spectrum of free and bound FAD does not display any shifts in absorption maxima (data not shown). The CD spectrum of the FAD·tFR complex (Fig. 10) displays a large positive peak at 370 nm and a large negative peak of 455 nm.

A FASTA data base search did not identify any proteins with significant similarities to VlmR. Searches of the PROSITE data base with MOTIFS and PSCAN also did not reveal any significant matches between the VlmR protein and motifs in the data base. A BLAST search revealed significant similarities between the VlmR protein and a small number of other proteins. A multiple sequence alignment of the VlmR protein with the related proteins identified by the BLAST search is presented in Fig. 11. The alignments between VlmR and these proteins are found throughout the amino acid sequences. Some of the aligning proteins are known to be flavin reductases, and others are hypothetical proteins of unknown function. The alignments exhibited by these hypothetical proteins suggest that these proteins are likely to be flavin reductases. Sequence comparisons between VlmR and a number of potentially related proteins using the BESTFIT program revealed that the closest match between VlmR and proteins of known function was with the SnaC protein which is an NADH:FMN oxidoreductase from Streptomyces pristinaespiralis (Table IV) (22). Other notable matches included the ActVB protein from Streptomyces coelicolor (23) and the NmoB/NtaB protein from Cheletobacter heintzii (24-26), both of which are NADH:FMN oxidoreductases.

Table IV. Comparison between VlmR and related proteins with BESTFIT program Gap weight, 3.0; length weight, 0.1.  Protein 1 Protein 2 Result Abbreviationa Length (aa) Compared range (aa) Abbreviationa Length (aa) Compared range (aa) Percent similarity Percent identity Z score Svf.VlmR 194 1-194 Scs.F1m3 160 1-160 50.0 27.8 11.2 Sps.SnaC 176 1-176 49.7 28.3 11.0 Sys.Fmnp 578 405-578 43.4 25.4 10.3 Eco.HpaC 170 1-170 42.9 24.1 10.2 Chi.NmoB/NtaB 322/321 1-200 47.3 26.9 9.8 Eco.F152 152 1-152 46.1 24.3 9.3 Sys.Hyp1 160 1-160 49.4 30.4 9.1 Sys.Hyp2 597 425-597 48.5 27.3 8.0 Scr.ActVB 177 1-177 51.4 29.1 7.1 Sys.Fp 573 400-573 48.8 26.5 5.4 Vfi.Fr 218 1-218 41.6 16.3 2.5 Vhi.Fr 237 1-237 46.3 18.3 0.1 Vhi.Frp 240 1-240 39.4 13.7 0 Vfi.Fr 218 1-218 Vhi.Frp 240 1-240 38.8 19.6 0.9 Vfi.Fr 218 1-218 Vhi.Fr 237 1-237 43.1 17.3 0 Vhi.Frp 240 1-240 Vhi.Fr 237 1-237 42.4 17.5 0 a Abbreviations and accession numbers for proteins are as follows: Svf.VlmR, the vlmR gene product; Chi.NmoB/Ntab, nitrilotriacetate monooxygenase component B from C. heintzii ATCC 29600 (NmoB L49438, g1119210; NtaB U39411, g148206); Scr.ActVB, NADH:FMN oxidoreductase from S. coelicolor (X63449, g46811); Sps.SnaC, NADH:FMN oxidoreductase from S. pristinaespiralis (U21216, g940352); Eco.HpaC, coupling component of 4-hydroxyphenylacetate hydroxylase from E. coli (Z29081, g452842); Sys.Hyp1, hypothetical protein from Synechocystis sp. PCC6803. (D64000, g1001553); Scs.Flm3, flm3 region hypothetical protein 1 from Synechococcus sp. PCC 7942 (L19521, g1084165, g310858); Eco.F152, f152 hypothetical protein from E. coli (AE000202, g1787242); Sys.Fmnp, potential FMN protein from Synechocystis sp. PCC6803 (D90900, g1651803); Sys.Hyp2, hypothetical protein from Synechocystis sp. PCC6803 (D90914, g1653555); Sys.Fp, putative flavoprotein from Synechocystis sp. PCC6803 (D64003, g1001242); Vfi.Fr, NADH:FMN oxidoreductase from V. fischeri (D17743, g517169); Vhi.Frp, NADPH:FMN oxidoreductase from V. harveyi (U08996, g478986); Vhi.Fr, NADH:FMN oxidoreductase from V. harveyi (D14674, g498172). b aa, amino acids.

DISCUSSION

The vlmR gene, which encodes an NADPH:FAD oxidoreductase, has been cloned from S. viridifaciens MG456-hF10 by chromosome walking and overexpressed in soluble form in E. coli. The resulting protein exhibited a high degree of flavin reductase activity in the presence of NADPH. Overexpression of the VlmR protein in E. coli proved to be problematic until it was discovered that the presence of a leader sequence coding for a His-tag at the 5-end of the gene allowed expression to occur. The difficulty in expression of the vlmR gene may be related to the fact that it is translationally coupled to the open reading frame (ORF2) that lies immediately upstream (Fig. 6). Since the vlmR gene is located in close proximity to the gene encoding isobutylamine N-hydroxylase (vlmH) on the S. viridifaciens chromosome (Fig. 6), it appears likely that the function of the VlmR protein is to provide reduced FAD to isobutylamine N-hydroxylase. This expectation is supported by preliminary evidence indicating that a mixture of VlmR and isobutylamine N-hydroxylase will efficiently catalyze the hydroxylation of isobutylamine to isobutylhydroxylamine. Since precursor incorporation experiments strongly support the intermediacy of isobutylamine and isobutylhydroxylamine in valanimycin biosynthesis (12, 13), it is reasonable to assume that both the vlmR and vlmH gene products are involved in valanimycin biosynthesis. However, definitive proof for the role of these genes in valanimycin biosynthesis will require that gene disruption experiments be performed.

A BLAST search indicated that VlmR is related to three proteins that are known to be NADH:FMN oxidoreductases: SnaC, ActVB, and NmoB/NtaB. These proteins and VlmR all exist as homodimers, bind flavins weakly, and utilize flavins as a substrate rather than as a cofactor. However, VlmR is unique in that it prefers FAD and NADPH and exhibits a pI of 10.0. Another unusual feature of VlmR is its strong preference for NADPH over NADH. This behavior is illustrated by the 1000-fold difference in the k_cat/K_m values for the two substrates at an FAD concentration of 40 µM (Table II). It is conceivable that the larger k_cat/K_m for NADPH is related to the fact that NADPH is more negatively charged than NADH and that VlmR will carry an excess of positive charges at biological pH due to its high pI. Apart from these considerations, the purpose that might be served by the high pI of the VlmR protein is unclear. The calculated pI values for SnaC, ActVB, and NmoB/NtaB fall between 5 and 7.

The fluorescence emission spectrum and CD spectrum of FAD are altered significantly upon binding to tFR. The increase in the intensity of the emission spectrum of the FAD may be due to movement of the adenine ring away from the isoalloxazine ring since stacking between these two rings quenches the fluorescence of free FAD (27). The optical activity of tFR-bound FAD apparent from the CD spectrum presumably results from asymmetric interactions between the bound FAD and the protein.

Sequence analysis suggests that VlmR, SnaC, NmoB/NtaB, and ActVB belong to the same family of flavin reductases. The hypothetical proteins Hyp1 (28), Flm3 (29) and F152³ from Synechocystis sp., Synechococcus sp., and E. coli, respectively, also appear to be members of this same family. Another group of hypothetical proteins from Synechocystis sp. Fmnp (31), Fp (28), and Hyp2 (32), are characterized by a C-terminal region that aligns with the VlmR, SnaC family. The larger size of these proteins suggest they may be multifunctional proteins that contain a flavin reductase domain.

The HpaC protein from E. coli (33) also aligns with the VlmR, SnaC family. This protein is part of a two-component enzyme system that catalyzes the conversion of 4-hydroxyphenylacetate to 3,4-dihydroxyphenylacetate. It has been reported that no NADH oxidation activity was detected in a cell-free extract of an E. coli strain harboring a plasmid that carried the hpaC gene (33). Consequently, HpaC was assigned the role of a coupling protein. However, when VlmR is overexpressed in E. coli, the NADPH oxidation activity in the crude cell-free extract is strongly inhibited (Table II). It therefore seems possible that HpaC is a flavin reductase and that the hydroxylation of 4-hydroxyphenylacetate in E. coli is catalyzed by a two-component system consisting of a flavin reductase and a hydroxylase.

Most luminous bacteria appear to contain multiple flavin reductases. For example, several flavin reductases have been cloned and sequenced from Vibrio sp. Examples include an NADPH-specific FMN reductase from Vibrio harveyi (Vhi.Frp) (34), an NADH-preferring FMN reductase from the same species (Vhi.Fr) (35), and an NADH-preferring FMN reductase from V. fischeri (Vfi.Fr) (30). None of these proteins was selected by the BLAST program as a possible match for VlmR. The lack of relationship between VlmR and the Vibrio flavin reductases was further confirmed by a BESTFIT comparison between VlmR and the three flavin reductases mentioned above (Table IV). A BESTFIT comparison between these three Vibrio proteins also revealed that they bear little resemblance to each other (Table IV). No sequence similarities were found between any other known flavoproteins and VlmR.

In summary, a gene (vlmR) encoding an NADPH:FAD oxidoreductase has been cloned from the valanimycin producer S. viridifaciens and overexpressed in soluble form. The overexpressed protein exhibited the expected enzymatic activity, thereby confirming the identity of the protein. The proximity of the vlmR gene to the gene (vlmH) coding for isobutylamine N-hydroxylase on the S. viridifaciens chromosome as well as evidence from enzymology and precursor incorporation experiments makes it likely that these two genes are part of a gene cluster associated with valanimycin biosynthesis. The derived amino acid sequence of the VlmR protein exhibited similarity to several other flavin reductases that may constitute a new family of flavin reductases. The successful overexpression of both the N-hydroxylase and the flavin reductase components of the isobutylamine hydroxylase system in soluble and active form now sets the stage for a more detailed enzymatic investigation of this novel amine hydroxylase.