Isolation and biochemical characterization of two soluble iron(III) reductases from Paracoccus denitrificans Jir ˇ ı ´ Mazoch 1 , Radek Tesar ˇ ı ´ k 2 , Vojte ˇ ch Sedla ´ c ˇ ek 1 , Igor Kuc ˇ era 1 and Jaroslav Tura ´ nek 2 1 Department of Biochemistry, Faculty of Science, Masaryk University, Brno, Czech Republic; 2 Department of Immunology, Veterinary Research Institute, Brno, Czech Republic Two soluble enzymes (FerA and FerB) catalyzing the reduction of a number of iron(III) complexes by NADH, were purified to near homogeneity from the aerobically grown iron-limited culture of Paracoccus denitrificans using a combination of anion-exchange chromatography (Seph- arose Q), chromatofocusing (Mono P), and gel permeation chromatography (Superose 12). FerA is a monomer with a molecular mass of 19 kDa, whereas FerB exhibited a molecular mass of about 55 kDa and consists of probably two identical subunits. FerA can be classified as an NADH:flavin oxidoreductase with a sequential reaction mechanism. It requires the addition of FMN or riboflavin for activity on Fe(III) substrates. In these reactions, the apparent substrate specificity of FerA seems to stem exclu- sively from different chemical reactivities of Fe(III) com- pounds with the free reduced flavin produced by the enzyme. Observations on reducibility of Fe(III) chelated by vicinal dihydroxy ligands support the view that FerA takes part in releasing iron from the catechol type siderophores synthes- ized by P. denitrificans. Contrary to FerA, the purified FerB contains a noncovalently bound redox-active FAD coen- zyme, can utilize NADPH in place of NADH, does not reduce free FMN at an appreciable rate, and gives a ping- pong type kinetic pattern with NADH and Fe(III)-nitrilo- triacetate as substrates. FerB is able to reduce chromate, in agreement with the fact that its N-terminus bears a homo- logy to the previously described chromate reductase from Pseudomonas putida. Besides this, it also readily reduces quinones like ubiquinone-0 (Q 0 ) or unsubstituted p-benzo- quinone. Keywords: iron(III) reductase; NADH:flavin oxidoreduc- tase; Paracoccus denitrificans. Ferric iron [Fe(III)] reductases, which catalyze the reduction of various complexed forms of Fe(III), are of ubiquitous occurrence in nature [1]. Conversion of Fe(III) to more soluble Fe(II) generally increases the bioavailability of iron and occurs either outside the cell [in conjunction with an upward Fe(II) transport] or intracellularly (e.g. iron release from the accumulated ferrisiderophores) [2]. In bacteria, environmental Fe(III) can serve as a hydrogen sink, allowing the regeneration of NAD + [3], or even as a terminal electron acceptor in an anaerobic respiration associated with translocation of H + and generation of a transmembrane potential difference [4,5]. Fe(III) reductases constitute a heterogeneous group of enzymes in respect to their biochemical properties. The majority of the known bacterial Fe(III) reductases resides in the cytoplasm, consists of one small polypeptide chain lacking any distinguishable prosthetic group, reduces Fe(III) at the expense of NADH or NADPH and requires flavinasanelectrontransfermediator[6–13].Onthe contrary, Fe(III) reductases of Fe(III) respirers are c-type cytochromes excreted into the medium or exposed at the cell surface where they can make direct contact with the solid metal oxides [14,15]. A representative of eukaryotic type enzymes is the FRE1 Fe(III) reductase of Saccharomyces cerevisiae, a transmembrane flavocytochrome b with prop- erties similar to those of the NADPH oxidase of human neutrophils [16]. In some instances the observed reduction of Fe(III) possibly represents a side activity of an enzyme that may catalyze a different reaction in vivo, as exemplified by the enzymes flavin reductases [12], sulfite reductase [17] or dihydropteridine reductase [18]. The facultative anaerobe Paracoccus denitrificans posses- ses a flexible respiratory chain able to incorporate consid- erable amounts of iron-containing proteins in response to the changed growth conditions [19,20]. This ability depends in part upon a capacity of the microbe to mobilize iron from external and internal sources. Evidence for a high-affinity iron-scavenging system involving low molecular weight, Fe(III)-specific siderophores was first presented by Tait [21] who isolated three iron-binding catechol compounds from the culture supernatant of P. denitrificans and provided initial data on their structure, metal-binding properties and enzymes needed for their biosynthesis. Since then, several Correspondence to I. Kuc ˇ era, Department of Biochemistry, Faculty of Science, Masaryk University, Kotla ´ r ˇ ska ´ 2, CZ-61137 Brno, Czech Republic. Fax: + 420 541129623, Tel.: + 420 541129351, E-mail: ikucera@chemi.muni.cz Abbreviations: Bis/Tris, bis(2-hydroxyethyl)amino-tris(hydroxy- methyl)methane; DHB, 2,3-dihydroxybenzoate; FNR, fumarate and nitrate reductase regulator (Escherichia coli); FnrP, FNR homologue from P. denitrificans; NfsB, oxygen-insensitive nitroreductase of E. coli; PVDF, poly(vinylidene difluoride); pyrocat, pyrocatechol; QR, quinone reductase. (Received 30 September 2003, revised 27 November 2003, accepted 5 December 2003) Eur. J. Biochem. 271, 553–562 (2004) Ó FEBS 2004 doi:10.1046/j.1432-1033.2003.03957.x thorough investigations into the mechanism, stereospecific- ity and kinetics of siderophore-mediated iron transport in P. denitrificans have been carried out [22–24], but informa- tion on the enzymology of Fe(III) reduction still remains scarce and incomplete. Tait [21] described formation of Fe(II) from the Fe(III)–siderophore complex upon incuba- tion with NADH, succinate and an ultracentrifuged and dialyzed cell-free extract while Dailey and Lascelles [25] noticed reduction of Fe(III) citrate by NADH or succinate as electron donors in the presence of crude cell extracts containing membranes. More recently, Kucera and Match- ova [26] succeeded in discriminating between the reduction of Fe(III) sulfate by NADH catalyzed by cytosolic and membrane fractions. The former reaction differed markedly from the latter by about 10-times lower apparent K m value for NADH, by a discernible stimulatory effect of FMN, and by the inability of succinate to substitute for NADH. The present report describes the results of studies leading to the conclusion that at least two distinct soluble enzymes of P. denitrificans exhibit an activity of Fe(III) reductase. These enzymes were purified and their molecular and catalytic properties were analyzed. The data obtained indicate that one of them can be assigned to the NADH-dependent flavin reductases while the other carries a bound flavin prosthetic group and bears a resemblance to a previously described enzyme capable of reducing chromate [27]. Materials and methods Organism and cultivation Paracoccus denitrificans, strain CCM 982 was obtained in freeze-dried form from the Czech Collection of Micro- organisms (CCM), Masaryk University, Brno, Czech Republic. It was maintained on agar plates and stored at 4 °C. Growth was achieved via batch cultivation using minimal succinate medium, containing (in m M ) sodium succinate, 50; KH 2 PO 4 , 33; Na 2 HPO 4 ,17;NH 4 Cl, 50; MgSO 4 , 1.1 and Fe(III) citrate, 0.5 l M . The final pH was adjusted to 7.3 by dropwise addition of 0.1 M NaOH. Separate 750 mL aliquots were autoclaved in 3-L Erlenmeyer flasks, inoculated with 20 mL of a pregrown culture and incubated for 10–12 h at 30 °Conareciprocal shaker at 180 r.p.m. Cells were sedimented by centrifuga- tion (6200 g,20minat4°C), washed twice with 50 m M Tris/HCl buffer, pH 7.4, and then taken up in this buffer. Preparation of the enzymes For a typical purification  10 g wet cells harvested from 3-L culture were used. Chromatography was run on a Pharmacia FPLC apparatus. Step 1: Preparation of cell-free extract. Cell-free extracts were made by passing the cell suspension twice through a 33-mL X-Press cell (AB Biox, Sweden), treating with 4 lgof deoxyribonuclease (Sigma) per mL for 30 min at room temperature to reduce the viscosity caused by DNA and removing the cellular debris by centrifugation at 109 000 g for 40 min in a rotor-type 45 TI (Beckman). The Fe(III) reductase-containing extracts were stored at )15 °C until commencement of purification. Step 2: Ion exchange chromatography. About 40 mL of cell-free extract, prepared as described above, were loaded onto an HP Sepharose Q (Pharmacia) column (XK26: 10 cm · 2.6 cm) pre-equilibrated with 20 m M Tris/HCl, pH 7.4, and washed by the same buffer. The proteins were eluted in two phases with 300 mL of linear gradient of 0–0.6 M NaCl and 50 mL of linear gradient of 0.6–1 M NaCl at a flow rate of 0.8 mLÆmin )1 .Fractionsof5mL were collected and those displaying Fe(III) complex reduc- tase activity were pooled (each activity peak separately), concentrated in an ultrafiltration cell (Amicon) through YM-3 membrane (Millipore) and then dialyzed against 20 m M Tris/HCl, pH 7.4. Step 3: Chromatofocusing. Two buffer systems in this chromatography on MonoP HR 5/20 (Pharmacia) column (20 · 0.5 cm) for two distinct enzymes were employed. First, starting with 0.075 M Tris/acetate, pH 9.3 used Polybuffer 96, second one started with 0.025 M Bis-Tris/ iminodiacetic acid, pH 7.1, and Polybuffer 74 (Pharmacia) was applied here, according to instructions of the manu- facturer. The flow rate was 0.3 mLÆmin )1 during sample loading and 0.5 mLÆmin )1 during focusing. Fractions of 0.5 mL were collected. Pooled active fractions were desalted on PD-10 columns (Supelco) equilibrated by the start buffer for the next step. Step 4 (optional): Chromatofocusing. The same condi- tions as in the previous step with Polybuffer 74 were used. Peak fractions were collected. Step 5: Gel permeation chromatography. Maximum 0.5 mL of sample was loaded onto Superose 12 (Pharmacia) analytical column (30 cm · 1 cm) equilibrated by 50 m M Tris/HCl, pH 7.4 containing 150 m M NaCl. The flow rate after sample application was 0.5 mLÆmin )1 . For determin- ation of molecular masses, the same column was calibrated by marker proteins including alcohol dehydrogenase (141 kDa), albumin (67 kDa), ovalbumin (43 kDa), chy- motrypsinogen A (25 kDa) and ribonuclease A (13.7 kDa). Step 6 (optional): Preparative native electrophoresis. In some instances (e.g. for N-terminal sequenation), final purification of the enzymes was performed from the activity bands on native gels (see below). The proteins were recovered by electroelution using electroelution tubes and a Mini Protean Cell (Bio-Rad) with Tris/glycine buffer, pH 8.3 [no sodium dodecyl sulfate (SDS)] at constant current 8 mA per one electroelution tube for 5.5 h. Determination of protein concentration Protein concentration was determined by the method of Lowry et al. [28] using bovine serum albumin as the standard. Protein elution profiles from chromatographic columns were also monitored by measuring fractional absorbance at 280 nm. Enzyme assays Fe(III) reductase activity was determined according to a variant of the method by Dailey and Lascelles [25]. The 554 J. Mazoch et al. (Eur. J. Biochem. 271) Ó FEBS 2004 assay measures the formation of the Fe(II)(ferrozine) 3 complex at 562 nm (absorption coefficient ¼ 28 m M )1 Æ cm )1 )in25m M Tris/HCl buffer, pH 7.4, 30 °C, with 0.8 m M ferrozine, 0.05 m M FMN and 0.2 m M Fe(III) complex. The reaction was started by the addition of 0.15 m M NADH. Chromate reductase activity was assayed at 30 °Cin 0.5-mL reaction mixtures containing 25 m M Tris/HCl buffer, pH 7.4, 0.045 m M K 2 CrO 4 ,0.15m M NADH, and 5–30 lL of enzyme preparation. Chromate concentration was quantified by adding samples to the reagent consisting of 0.1 M H 2 SO 4 and 0.01% 1,5-diphenyl carbazide and measuring the absorbance at 540 nm [27]. Quinone reductase and flavin reductase activities were monitored by following the disappearance of the NADH absorbance at 340 nm in 25 m M Tris/HCl buffer, pH 7.4, 30 °C. Concentrations of electron acceptors were 100 l M for quinones when measuring quinone reductase and 50 l M for FMN in flavin reductase assay. Stock solutions of quinones were prepared as 5 or 10 m M in 20% acetone (v/v). Stopped-flow experiments A JASCO J-810 stopped-flow spectrophotometer with a dead time of 5.2 ms and a cell path length of 1.5 mm was used to measure redox reactions of Fe(III) complexes with chemically reduced FMN. A solution of 1.25 m M FMN in 25 m M Tris/HCl, pH 7.4, containing PtO 2 (0.1 mgÆmL )1 )as a catalyst [29] was made anaerobic with argon bubbling and then a stream of hydrogen gas was introduced to convert FMN into FMNH 2 . Following the sedimentation of the catalyst, the yellowish supernatant was placed in one syringe, and the second syringe was filled with a buffered solution of 1.6 m M ferrozine and 0.4 m M Fe(III) complexed with an appropriate ligand. After 1 : 1 mixing, changes in absorbance at 562 nm were recorded. Data from at least six repetitions were averaged and processed with the software package supplied with the instrument, giving the desired initial velocity values. Mass spectrometry The mass spectra of proteins were recorded on a Bruker Reflex IV mass spectrometer. Prior analysis, the samples were dialyzed against 10 m M Tris/HCl, pH 7.4, and then concentrated to a final value of about 1–10 lgofproteinper lL using a Speed-Vac concentrator (Savant, Holbrook, USA). One microliter each of sample was mixed with 2 lLof matrix solution (ferulic acid, saturated in 50% acetonitrile, 1% trifluoroacetic acid). Aliquots (0.6 lL) were placed on the target and left to dry. The spectra were taken in positive- ion linear mode. External mass calibration was performed using the molecular ions from the bovine insulin at 5734.5 Da and the horse myoglobin at 16952.6 Da 100 single shot spectra were accumulated in each mass spectrum. Gel electrophoresis One-dimensional SDS gel electrophoresis was carried out in 10–15% polyacrylamide gels [30], containing 0.1% SDS, using a Mini Protean II (Bio-Rad) vertical electrophoretic system. Staining was carried out using Coomassie brilliant blue R-250. Standard molecular mass markers were from Fluka. Non-denaturing electrophoresis was carried out in the same manner, but with omission of SDS from the gel running and loading buffers, and the sample was not pretreated under denaturing conditions. Staining was car- ried out using Coomassie blue R-250 or by zymogram analysis. The latter entailed overlay of the gel with 0.15 m M NADH and 0.8 m M ferrozine in 50 m M Tris/HCl buffer, pH 7.4, containing 0.05 m M FMN. Fe(III)nitrilotriacetic acid solution in the same buffer was added to the resulting concentration of 0.2 m M . Bands of Fe(III) reductase activity were detected as red zones against a slightly reddish background. Spectroscopic measurements Absorption spectra of the enzymes were taken with an Ultrospec 2000 UV/visible spectrophotometer (Pharmacia) using a 1-cm light path cell. Fluorescence spectra were obtained using an LS 50B spectrofluorimeter (Perkin Elmer). Thin-layer chromatography analysis of flavins About 0.4 mg of the purified flavoprotein, dissolved in 0.1 mL of buffer, was mixed with 0.9 mL of methanol and the mixture was heated at 95 °C for 15 min. After cooling on ice and centrifugation at 12 000 g for 5 min, the supernatant was dried with a Speed-Vac concentrator. Water (10 lL) and methanol (10 lL) were added to the residue, and the solution was analyzed by TLC employing a cellulose gel plate (DC-Alufolien Cellulose, 0.1 mm thick- ness; Merck AG, Darmstadt, Germany) and the upper layer of an n-butyl alcohol/glacial acetic acid/water (4 : 1 : 5) mixture as the solvent phase. Methanol/water (1 : 1) solution of riboflavin, FMN and FAD were spotted as reference and the migration of the compounds were visualized with UV light. N-Terminal sequence analysis The highly purified enzymes were subjected to SDS/PAGE (15% polyacrylamide) and then electroblotted onto a poly(vinylidene difluoride) membrane (Bio-Rad). Selected bands were cut out of the Coomassie Blue-stained blots and analyzed by the Edman method with a 491 protein sequencer (Perkin-Elmer Applied Biosystems) according to the standard program (PL PVDF Protein). Homology searching and sequence alignments Homology searching against the databases was performed with the Basic Logical Alignment Search Tool ( BLAST )on NCBI Server (http://www.ncbi.nih.nlm.gov). Alignments of the sequences were produced using CLUSTAL W Multiple Sequence Alignment software on NPS@Server (http:// npsa-pbil.ibcp.fr). Kinetic data analysis Input data represented the mean values from at least three replicates. Nonlinear regression analysis was performed Ó FEBS 2004 P. denitrificans Fe(III) reductases (Eur. J. Biochem. 271) 555 with the kinetic software package EZ - FIT developed by Perrella [31]. Results Choice of a Fe(III) substrate for the enzyme assay Most of Fe(III) reductases have no known native substrates, but they exhibit a fairly weak specificity and can reduce many artificial Fe(III) complexes. Therefore, we explored a series of five compounds (Fe(III)EDTA, Fe(III)EGTA, Fe(III)nitrilotriacetic acid, Fe(maltol) 3 and FeCl 3 ) as poten- tial artificial substrates for measuring the activity in cell-free extracts by the ferrozine assay. When no flavin was added exogenously, we found the following specific activities [nmol Fe(II)(ferrozine) 3 Æs )1 Ægprotein )1 , means of three replicates]: 2, 151, 146, 16, 2, whereas with 50 l M FMN the respective values were: 5, 210, 392, 323, 83. These results told us that the Fe(III)nitrilotriacetic acid substrate together with FMN would best serve the purpose of activity measuring during enzymes purification. Contrary to some [8,13], but in accordance with other [10] previous research, our assay for the enzymes from P. denitrificans was not significantly sensitive to oxygen, as air removal from the reaction mixture with a stream of argon increased the rate of Fe(II)(ferro- zine) 3 production by 10% maximum. Based on this finding, all further measurements could be performed under ambient oxygen pressure. Purification of the Fe(III) reductases Preliminary experiments established that cultivation of P. denitrificans in a growth medium with limiting initial levels of Fe(III) citrate (0.5 l M instead of the normally used concentration of 30 l M ) increased the specific activity of Fe(III) reductase in the cell-free extract about 1.5-fold. Accordingly, the enzyme was isolated from the iron- deficient cells that had reached the stationary phase after aerobic growth. A summary of the data of the purification is given in Table 1 for a typical preparation. Aspects deserving comment are as follows: (i) Assay of the fractions in the HP-Sepharose Q elution profile identified two peaks of Fe(III) reductase activity, eluting at approximately 50–150 m M and 250–400 m M NaCl. They were termed ÔFerAÕ and ÔFerBÕ, respectively. Peak fractions were pooled separately and purification was continued for each pooled peak of activity. (ii) The presence of two Fe(III) reductases in cell-free exacts could also be directly demonstrated by native gel electrophoresis of the cell-free extracts followed by activity staining, which produced two bands with the intensity ratio of about 10 : 1. The more intense band corresponded to the enzyme with a higher electrophoretic mobility towards the anode. (iii) Preparative chromatofocusing proved to be an efficient purification step, which could be used twice for FerB, but not for FerA due to less resistance to pH changes. Chromatofocusing enabled us to determine isoelectric points of FerA and FerB as 6.9 and 5.5, respectively. These values are in accord with the observed chromatographic behavior of the enzymes at the Sepharose Q step. (iv) During optimization of the purification protocol the possibility of employing a hydrophobic interaction chro- matography was also noticed. After loading the test sample on a Phenyl Superose column (Pharmacia) or on a Hema- Bio 1000 Phenyl column (Tessek, Prague) and applying a 1.5–0 M ammonium sulfate gradient, FerA was typically eluted at 80–90% of the gradient compared to FerB eluted at 45–50%. FerA thus seems to be considerably more hydrophobic than FerB. Determination of molecular mass SDS/PAGE analysis gave single Coomassie blue staining bands at a position of an approximate M r of 18 000 for both the purified FerA and FerB (Fig. 1). These values were further refined by applying a MALDI-MS analysis, which revealed peaks at m/z 18 814 and 18 917 for FerA and at m/z 20 196 for FerB. We are currently unable to identify unequivocally the reason for the presence of two peaks in the spectrum of FerA. As the enzymes could be separated in a column of Superose 12, the same column was used to estimate their molecular sizes. By calibration of the column with standard proteins of known molecular masses, the molecular masses of native FerA and FerB were estimated as 25 kDa and Table 1. Purification of P. denitrificans Fe(III) reductases. The activity was determined following absorbance of Fe(II)(ferrozine) 3 complex at 562 nm. Specific activity is expressed as enzyme activity per mg of total protein. Treatment Volume (mL) Total protein (mg) Total activity (nkat) Specific activity (nkatÆmg )1 ) Purification factor (-fold) Yield (%) FerA Cytosolic extract 25.5 517.4 120.3 0.23 1 100 Sepharose Q 2.8 7.81 59.1 7.57 32.9 49.13 Polybuffer exchanger 96 2.8 0.155 15.9 102.58 446 13.22 Superose 12 HR 10/30 4.0 0.028 9.17 327.50 1424 7.62 FerB Cytosolic extract 25.5 517.4 85.50 0.17 1 100 Sepharose Q 2.85 23.73 38.37 1.62 9.8 44.9 Polybuffer exchanger 74 1.5 0.434 16.47 37.93 230 19.3 Polybuffer exchanger 74 1.0 0.069 5.22 75.23 457 6.1 Superose 12 HR 10/30 0.6 0.024 2.25 93.89 568 2.6 556 J. Mazoch et al. (Eur. J. Biochem. 271) Ó FEBS 2004 55 kDa, respectively. From these results, we concluded that FerA exists in a monomeric form in the native state and that FerB has an oligomeric structure (a homodimer or, less probable, a homotrimer). N-Terminal sequence comparison The N-terminal amino acid sequences of FerA and FerB were determined and are presented in Fig. 2(A). A BLAST search revealed sequence similarity of 73% and 65% (sequence identity of 50% in both cases) of this region in FerB with chromate reductase from Pseudomonas putida andflavinreductaseformPseudomonas syringae (Fig. 2B), respectively. On the contrary, no significant similarity was found between the N-terminus of FerA and any known oxidoreductase. Spectral properties Measurement of the UV-visible absorption spectrum of FerA (about 95% pure, 0.04 mgÆmL )1 ) showed no chro- mophore uniquely attributable to this enzyme. On the other hand, the solution of FerB was yellowish in color and exhibited absorption bands at 380 and 448 nm (Fig. 3), diagnostic for the presence of a flavin group. When 0.4 m M NADH was added, the absorption peak at 448 nm disappeared, indicating a reducibility of the bound flavin by the physiological electron donor. The possession by FerB of a flavin cofactor was also apparent from the fluorescence Fig. 1. SDS/PAGE followed by silver staining. Lane 1, 10 lgproteinof the starting cell-free extract; lane 2, 0.3 lgofpurifiedFerA;lane3, 1 lg of purified FerB; lane 4, molecular mass markers indicated (in kDa) on the right. Fig. 2. N-Terminal amino acid sequences of P. denitrificans FerA and FerB (A) and comparison of the N-terminal amino acid sequence of FerB with the sequences of other oxidoreductases (B). N-Terminal sequences were determined by Edman degradation. Multiple alignment of chosen sequences was performed using CLUSTAL W programme. Fre, flavin reductases of Pseudomonas syringae pv. syringae B728a (gi:23469150) and of P. syringae pv. tomato DC3000 (gi:28870863); CR, chromate reductase of Pseudomonas putida (gi:14209680); ?, a hypothetical protein (sequence derived from the known DNA sequence). Identical (w), strongly similar (:) and weakly similar (.) residues are indicated. Fig. 3. Absorption spectra of as isolated (—) and NADH reduced (- ) FerB. The spectrum of FerB, 0.2 mg proteinÆmL )1 , were measured under aerobic conditions in 25 m M Tris/HCl buffer (pH 7.4), employing the buffer solution as a reference, with a light path of 1 cm. The reduced spectrum was obtained 5 min after the addition of 0.4 m M NADH. Ó FEBS 2004 P. denitrificans Fe(III) reductases (Eur. J. Biochem. 271) 557 spectrum where the enzyme excited at 446 nm shows a fluorescence emission with a maximum at 506 nm. Identification of the flavin moiety of FerB Exposure of FerB to heated methanol released a yellow soluble compound from the holoenzyme. Thin-layer chro- matography of the methanol extract gave a single fluores- cent spot with a R f of 0.075. Under identical conditions, riboflavin, FMN and FAD exhibited R f values of 0.40, 0.19 and 0.075, respectively. Thus, we concluded that FerB contains a noncovalently bound FAD. Substrate specificity In order to investigate the catalytic abilities of the isolated enzymes, a specificity study was undertaken with Fe(III)ni- trilotriacetic acid and NADH as reference substrates. The rate of reduction of different Fe(III) compounds, under the conditions specified, is shown in Table 2. FerA has an absolute requirement for the flavin and cannot effectively use NADPH as an electron donor in place of NADH. A substi- tution of riboflavin for FMN led to a decrease by 73% in the reaction rate (not shown in Table 2). Unlike FerA, FerB does not require exogenous flavin and utilizes both NADH and NADPH. Probably the most striking difference between FerA and FerB lies in the inability of the latter enzyme to reduce Fe(III) complexed by vicinal dihydroxy groups, as present in maltol, pyrocatechol and 2,3-dihydroxybenzoic acid. In line with this, we also observed that FerA, but not FerB, exhibited a reducing activity towards the complex of Fe(III) with a crude isolate of parabactin, the natural catecholic siderophore of P. denitrificans (data not shown). By monitoring the absorbance at 340 nm, we observed that the addition of FMN to FerA resulted in a significant rate of NADH oxidation also in the absence of Fe(III), which suggested the presence of an NADH : FMN oxido- reductase activity. Its value typically amounted to 330 nmol NADHÆs )1 Æmg protein )1 . For this reason we tested whether the reduction of Fe(III) could be due to the FMNH 2 formed by FerA. This possibility gained support from the stopped-flow experiments in which we compared the rates of a direct nonenzymatic reduction of Fe(III) complexes by the chemically prepared FMNH 2 .Therelative rate values presented in the last column of Table 2 match more closely those of FerA (r ¼ 0.92) rather than those for FerB (r ¼ 0.77), demonstrating that FerA indeed may act primarily as a flavin reductase, and, per se, not mediate the subsequent redox reactions of Fe(III). The identification of a sequence homology between FerB and chromate reductases (Fig. 2B) prompted an investigation of whether or not the Paracoccus enzyme can reduce chromate. The positive answer is conveyed by the results in Fig. 4, showing a significant speed-up in Table 2. Substrate specificity of Fe(III) reductases from P. denitrificans, compared with the chemical reactivity of FMNH 2 . Experimental details are described in the Material and methods section. The Fe(III) complexes were used at a final concentration of 0.2 m M . The relative reaction velocity for NADPH was related to that of NADH as 100%. The relative rates for the different Fe(III) complexes were related to that of Fe(III)nitrilotriacetic acid as 100%. The relative rate zero means a relative reaction velocity of > 0.1%. Substrate Relative rate of Fe(II)(ferrozine) 3 complex formation FerA FerB No enzyme +50 l M FMN –FMN +50 l M FMN FMNH 2 NADH 100 a 100 b 100 – NADPH 0 82.3 69.0 – Fe(III)nitrilotriacetic acid 100 a 100 b 100 100 c Fe(III)EGTA 125.5 54.2 54.9 106.4 Fe(III)EDTA 3.8 2.4 1.1 5.5 Fe(III)(maltol) 3 91.4 3.9 3.1 83.1 FeCl 3 29.0 0.7 0.7 – Fe(III)citrate 65.8 3.7 8.4 36.4 Fe(III)(pyrocat) 3 37.3 0 0 13.0 Fe(III)(DHB) 3 76.8 0 0 28.0 a 100% ¼ 2.04 · 10 )7 molÆs )1 Æmg protein )1 ; b 100% ¼ 8.43 · 10 )9 molÆs )1 Æmg protein )1 ; c 100% ¼ 3.73 · 10 )5 M Æs )1 . Fig. 4. Chromate reductase activity of FerB. Reaction mixture (0.5 mL) containing 0.045 m M chromate, 0.15 m M NADH and 25 m M Tris/HCl buffer, pH 7.4, with (d) or without (s)5lg of FerB were incubated at 30 °C. Aliquots were withdrawn at the indicated time intervals and analyzed for the concentration of chromate. 558 J. Mazoch et al. (Eur. J. Biochem. 271) Ó FEBS 2004 the reduction of chromate by NADH, caused by the presence of FerB. FerB also functioned as an efficient quinone reductase. For a series of four quinones, the relative activities were found as follows: 2,3-dimethoxy-5-methyl- 1,4-benzoquinone (ubiquinone-0), 100%; p-benzoquinone, 77%; menadione, 6%; duroquinone, 0.01% (100% ¼ 2.42 lmol NADHÆs )1 Æmg protein )1 ). The enzyme did not catalyze the oxidation of NADH by FMN at an appreci- able rate (specific activity less than 1 nmol NADHÆs )1 Æ mg protein )1 ). Kinetic studies of reactions catalyzed by FerA and FerB When studying the Fe(III) reductase reaction of FerA, there are effectively three compounds to be considered in a kinetic analysis: NADH, FMN and complexed Fe(III). Initial velocity studies were carried out by varying the concentra- tions of NADH in the presence of a series of concentrations of Fe(III)nitrilotriacetic acid while maintaining FMN at a constant concentration. As illustrated in Fig. 5(A), recipro- cal plots are patterns of lines intersecting to the left of the vertical axis. The FMN reductase reaction of FerA was analyzed with NADH and FMN in the absence of Fe(III). Under these conditions, kinetic patterns similar to those found above were obtained (Fig. 5B). This behavior indi- cates a sequential mechanism with NADH and FMN as substrates. The results obtained on studying the Fe(III) reductase reaction of FerB are in Fig. 6. It can be seen that the reciprocal plot for NADH as variable substrate at a series of concentrations of Fe(III)nitrilotriacetic acid consists of a series of parallel lines. Kinetic patterns of this type occur in cases of ping-pong mechanisms. At the highest concentra- tion of NADH the initial velocity of the enzyme turned out to be slower than that predicted by the hyperbolic law. Similar substrate inhibition was also observed in the 1/v vs. 1/[Fe(III)nitrilotriacetic acid] plot; this issue was, however, not pursued further at this stage. Table 3 lists values of the kinetic parameters for individ- ual substrates, as estimated by computer fitting of velocity data of Figs 5 and 6 to the appropriate kinetic equations. Besides the data for Fe(III)nitrilotriacetic acid in Table 3, we also determined the kinetic parameters for some other Fe(III) complexes reducible by FerB. In the presence of a saturation concentration (0.15 m M )ofNADH,thevalues of K m and k cat are: (0.4 ± 0.2) m M and (0.91 ± 0.02) s )1 for Fe(III)EGTA (1.6 ± 0.8) m M and (0.16 ± 0.03) s )1 for Fe(III)EDTA (0.4 ± 0.2) m M and (0.08 ± 0.01) s )1 for Fe(III)citrate, and (0.4 ± 0.2) m M and (0.015 ± 0.003) s )1 for Fe(III) chloride. The kinetics of the reduction of Fe(III) ligated with maltol could not be measured reliably due to the difficulty caused by an intensive coloration of the complex. The fact that all the obtained values K m are higher than 0.2 m M confirms that the measurements reported in Table 2 were performed under conditions of first-order Fig. 5. Initial-velocity kinetics of the reactions catalyzed by FerA. Measurements were performed with 0.2 lgFerAproteinÆmL )1 in 25 m M Tris/HCl buffer, pH 7.4 at 30 °C. (A) Fe(III) reductase reac- tion velocity as a function of NADH concentration in the presence of eight (s), 12 (d), 20 (h), 40 (j)and67(n) lM Fe(III)nitrilotriacetic acid at constant concentrations of FMN (50 l M ) and ferrozine (0.8 m M ). (B) Flavin reductase reaction velocity as a function of FMN concentration in the presence of 5 (s), 10 (d), 20 (h)and30(j) l M NADH. In this case both ferrozine and Fe(III)nitrilotriacetic acid were absent. Data are presented as double-reciprocal plots. Fig. 6. Initial-velocity kinetics of the Fe(III) reductase reaction cata- lyzedbyFerB.Measurements were performed at 30 °Cwith 3 lgFerBproteinÆmL )1 in 25 m M Tris/HCl buffer, pH 7.4, contain- ing 0.8 m M ferrozine. Reciprocal initial velocity is plotted against the reciprocals of NADH concentration at a series of fixed concentrations of Fe(III)nitrilotriacetic acid equal to 40 (s), 66.7 (d), 100 (n)and200 (m) lM. Ó FEBS 2004 P. denitrificans Fe(III) reductases (Eur. J. Biochem. 271) 559 kinetics and hence the relative rates listed here, being proportional to the k cat /K m ratios, correctly express the substrate specificity of FerB. Discussion In this study, we have demonstrated a coexistence of two types of soluble Fe(III) reductases, FerA and FerB, in the soil bacterium Paracoccus denitrificans. Both enzymes are flavin-dependent, but in different ways: a flavin serves as a dissociable cosubstrate for FerA (a flavin reductase-type enzyme) and as a tightly bound redox-active cofactor of FerB (a flavoprotein-type enzyme). Flavin reductases are present in a variety of organisms [12]. Extensive studies have been conducted, for instance, with enzymes from E. coli [7,32] or luminous bacteria Vibrio harveyi [33–40] and Vibrio fischeri [41–44]. Although these enzymes catalyze similar reactions, they differ in respect to substrate specificities and reaction mechanisms. Accordingly, they can be classified into D, P and G types based on whether NADH, NADPH or both NADH and NADPH are utilized. Obviously, FerA falls into the D -type. Another classification criterion distinguishes between flavin reductases lacking a detectable prosthetic group and those having a noncovalently bound flavin cofactor. Considering the negative result of the spectroscopic search for a bound flavin group and the sequential kinetic pattern obtained (Fig. 5), we conclude that the first possibility applies, namely that FerA accelerates a direct redox reaction between NADH and a flavin. In this respect, it seems to be similar to the D -type flavin reductase of V. harveyi [35]. The observed substrate specificity of FerA (Table 2) adds to the previous appreciation of the biological role of a free reduced flavin as a reducing agent for Fe(III) [2,12,13,32] and broadens it to account for the reduction of the catechol- type Fe(III) complexes. The high effectiveness of the redox process mediated by free flavin is reflected by a low value of the apparent Michaelis constant for the Fe(III) complex with nitrilotriacetic acid (Table 3). In the case of FerB, the distinctive spectral features (Fig. 3) and the ping-pong kinetic patterns (Fig. 6) make it likely that the enzyme shuttles between the forms containing oxidized and reduced bound flavin during the catalytic cycle. The fact that FerB appears rather inefficient in reducing some substrates which are otherwise prone to uncatalyzed reactions with free FMNH 2 (Table 2) suggests that the protein moiety contributes significantly to the observed substrate specificity. Although more direct infor- mation on the three-dimensional structure of FerB is obviously required before any firm conclusion can be drawn as to how this specificity is achieved, we can tentatively imagine two situations. In the first, the reduced flavin binds more tightly than the oxidized one, so that the cofactor becomes poorer electron donor and reactions of electron acceptors with low standard redox potentials are hampered. Our voltammetric measurements (unpublished results) have revealed that the Fe(III)/Fe(II) pairs complexed with maltol, pyrocatechol and dihydroxybenzoate, which do not accept electrons from FerB, have indeed the lowest standard redox potentials among the iron compounds listed in Table 2. Alternatively, FerB may carry a buried substrate binding cavity whose size allows for interaction with smaller electron acceptors like 1 : 1 Fe(III) complexes, chromate or qui- nones, but not with the rather bulky 1 : 3 Fe(III) complexes or exogenous free flavins. A survey of the literature indicates that FerB shares some biochemical properties with other enzymes. Its relatedness to the earlier described chromate reductase of Pseudomonas putida [27] is apparent from a match in the N-terminal sequence (Fig. 2B) and the ability to reduce chromate at a comparably high rate (Fig. 4). The subunit and native masses of FerB found here agree favorably with the values of 20 and 50 kDa reported before for Pseudomonas enzyme [27]. However, there is a marked difference in isoelectric points. While FerB is an acidic protein (pI 5.5 by chromatofocusing), pI of chromate reductase was claimed to be higher than 7.0 from chromatographic behavior; our own calculation based on the known amino acid sequence (gi:14209680) and exploiting the pI/Mw tool [45] on http:// www.expasy.org gave the value of 8.55. Unfortunately, the Pseudomonas enzyme was not tested for the presence of flavin nor the activity towards electron acceptors other than chromate, and hence a more detailed comparison with FerB cannot be presented at the moment. It may also be of interest to mention here the existence of the FAD-contain- ing DT-diaphorase [46] and the FMN-containing flavin- Table 3. Michaelis constants (± SE) for the substrates of FerA and FerB. Values were calculated from the data in Figs 5 and 6. The best fitting kinetic models, selected according to the Akaike’s information criterion, correspond to a sequential bi–bi mechanism for FerA and to a ping-pong mechanism for FerB. Non-linear regression values of k cat are (14 ± 1) s )1 and (28 ± 6) s )1 for the Fe(III) reductase and FMN reductase activity of FerA, respectively, and (3.1 ± 0.5) s )1 for FerB. Substrate (variable concentration) Second substrate (fixed concentrations) Third substrate (constant concentration) K m (l M ) FerA NADH Fe(III)nitrilotriacetic acid FMN (50 l M ) 5.5 ± 0.7 Fe(III)nitrilotriacetic acid NADH FMN (50 l M )17±2 NADH a FMN – 11 ± 5 FMN a NADH – 20 ± 8 FerB NADH Fe(III)nitrilotriacetic acid – 2.6 ± 0.7 Fe(III)nitrilotriacetic acid NADH – 800 ± 200 a Fe(II)(ferrozine) 3 production was used to quantify reaction progress except in these cases where NADH oxidation by flavin was measured. 560 J. Mazoch et al. (Eur. J. Biochem. 271) Ó FEBS 2004 reductase [43,44] and nitroreductase [47] that resemble FerB in being homodimers of comparable molecular size, utilizing both NADH and NADPH as electron donors, exhibiting ping-pong kinetics and having a number of activities including that of quinone reductase. Indications exist that a quite small number of amino acid residues determine electron acceptor specificity of these enzymes. For example, a single point mutation was enough to convert NfsB, one of the oxygen-insensitive nitroreductases of Escherichia coli,to a highly active flavin reductase [48]. Constrains of similar type as in NfsB may exist in FerB and preclude in this way its reaction with free flavin. We hope that future compar- ative studies will clarify relationships (if any) between FerB and the other mentioned proteins on a molecular level. A part of this effort is currently aimed at the cloning of the gene for FerB for further characterization. Acknowledgements We thank Dr Z. Vobu˚ rka (Institute of Organic Chemistry and Biochemistry, AS CR, Prague) for performing N-terminal amino acid sequencing, Dr Z. Zdra ´ hal (Laboratory of Mass Spectrometry of Biomolecules, Faculty of Science, MU, Brno) for MALDI-MS analyses and Dr Z. Prokop (National Centre for Biomolecular Research, Faculty of Science, MU, Brno) for assisting us with the stopped-flow measurements. This work was supported by a grant from the Grant Agency of the Czech Republic to I. K. (203/01/1589) and partly by a contract from Ministry of Agriculture of the Czech Republic to J.T. (M-3-99-01). References 1. Schro ¨ der, I., Johnson, E. & de Vries, S. (2003) Microbial ferric reductases. FEMS Microbiol. Rev. 27, 427–447. 2. Cowart, R.E. (2002) Reduction of iron by extracellular iron reductases: implications for microbial iron acquisition. Arch. Biochem. Biophys. 400, 273–281. 3. Jones, J.G., Gardner, S. & Simon, B.M. (1983) Bacterial reduction of ferric iron in a stratified eutrophic lake. J. Gen. Microbiol. 129, 131–139. 4. Short, K.A. & Blakemore, R.P. (1986) Iron respiration-driven proton translocation in aerobic bacteria. J. Bacteriol. 167,729– 731. 5. Dobbin, P.S., Warren, L.H., Cook, N.J., McEwan, A.G., Powell, A.K. & Richardson, D.J. (1996) Dissimilatory iron(III) reduction by Rhodobacter capsulatus. Microbiology 142, 765–774. 6. Moody, M.D. & Dailey, H.A. (1985) Ferric iron reductase of Rhodopseudomonas sphaeroides. J. Bacteriol. 163, 1120–1125. 7. Fontecave, M., Eliasson, R. & Reichard, P. (1987) NAD(P)H: flavin oxidoreductase of Escherichia coli. A ferric iron reductase participating in the generation of the free radical of ribonucleotide reductase. J. Biol. Chem. 262, 12325–12331. 8. Halle, F. & Meyer, J M. (1992) Ferrisiderophore reductases of Pseudomonas: purification, properties and cellular location of the Pseudomonas aeruginosa ferripyoverdine reductase. Eur. J. Bio- chem. 209, 613–620. 9. Mazoy, R. & Lemos, M.L. (1996) Ferric-reductase activities in whole cells and cell fractions of Vibrio (Listonella) anguillarum. Microbiol. 142, 3187–3193. 10. Mazoy,R.,Lopez,E.M.,Fouz,B.,Amaro,C.&Lemos,M.L. (1999) Ferric-reductase activities in Vibrio vulnificus biotypes 1 and 2. FEMS Microbiol. Lett. 172, 205–211. 11. Vadas, A., Monbouquette, H.G., Johnson, E. & Schro ¨ der, I. (1999) Identification and characterization of a novel ferric reductase from the hyperthermophilic archaeon Archaeoglobus fulgidus. J. Biol. Chem. 274, 36715–36721. 12. Fontecave, M., Coves, J. & Pierre, J L. (1994) Ferric reductases or flavin reductases? Biometals 7, 3–8. 13. Coves, J. & Fontecave, M. (1993) Reduction and mobilization of iron by a NAD(P)H: flavin oxidoreductase from Escherichia coli. Eur. J. Biochem. 211, 635–641. 14. Myers, C.R. & Myers, J.M. (1997) Outer membrane cytochromes of Shewanella putrefaciens MR-1: spectral analysis, and purifica- tion of the 83-kDa c-type cytochrome. Biochim. Biophys. Acta 1326, 307–318. 15. Seeliger,S.,Cord-Ruwisch,R.&Schink,B.(1998)Aperiplasmic and extracellular c-type cytochrome of Geobacter sulfurreductans acts as a ferric ion reductase and as an electron carrier to other acceptors or to partner bacteria. J. Bacteriol. 180, 3686–3691. 16. Shatwell, K.P., Dancis, A., Cross, A.R., Klausner, R.D. & Segal, A.W. (1996) The FRE1 ferric reductase of Saccharomyces cere- visiae is a cytochrome b similar to that of NADPH oxidase. J. Biol. Chem. 271, 14240–14244. 17. Coves, J., Eschenbrenner, M. & Fontecave, M. (1993) Sulfite reductase of Escherichia coli is a ferrisiderophore reductase. Biochem. Biophys. Res. Commun. 192, 1403–1408. 18. Lee, P.L., Halloran, C., Cross, A.R. & Beutler, E. (2000) NADH- ferric reductase activity associated with dihydropteridine reductase. Biochem. Biophys. Res. Commun. 271, 788–795. 19. Sapshead, L.M. & Wimpenny, J.W.T. (1972) The influence of oxygen and nitrate on the formation of the cytochrome pigments of the aerobic and the anaerobic respiratory chain of Micrococcus denitrificans. Biochim. Biophys. Acta 267, 388–397. 20. Bosma, G., Braster, M., Stouthamer, A.H. & van Verseveld, H.W. (1987) Subfractionation and characterization of soluble c-type cytochromes from Paracoccus denitrificans cultured under various limiting conditions in the chemostat. Eur. J. Biochem. 165, 665–670. 21. Tait, G.H. (1975) The identification and biosynthesis of sidero- chromes formed by Micrococcus denitrificans. Biochem. J. 146, 191–204. 22. Bergeron, R.J., Dionis, J.B., Elliott, G.T. & Kline, S.J. (1985) Mechanism and stereospecificity of the parabactin-mediated iron- transport system in Paracoccus denitrificans. J. Biol. Chem. 260, 7936–7944. 23. Bergeron, R.J., Weimar, W.R. & Dionis, J.B. (1988) Demon- stration of ferric 1-parabactin-binding activity in the outer mem- brane of Paracoccus denitrificans. J. Bacteriol. 170, 3711–3717. 24. Bergeron, R.J. & Weimar, W.R. (1990) Kinetics of iron acquisi- tion from ferric siderophores by Paracoccus denitrificans. J. Bac- teriol. 172, 2650–2657. 25. Dailey, H.A. Jr & Lascelles, J. (1977) Reduction of iron and synthesis of protoheme by Spirillum itersonii and other organisms. J. Bacteriol. 129, 815–820. 26. Kucera, I. & Matchova, I. (1997) Iron as a possible mediator of the oxic-to-anoxic transition in Paracoccus denitrificans. Biochem. Mol. Biol. Int. 43, 305–310. 27. Park, C.H., Keyhan, M., Wielinga, B., Fendorf, S. & Matin, A. (2000) Purification to homogeneity and characterization of a novel Pseudomonas putida chromate reductase. Appl. Environ. Microbiol. 66, 1788–1795. 28. Lowry, O.H., Rosebrough, N.J., Farr, A.L. & Randall, R.J. (1951) Protein measurement with the folin-phenol reagents. J. Biol. Chem. 193, 265–275. 29. Halle, F. & Meyer, J M. (1992) Iron release from ferrisider- ophores. A multi-step mechanism involving a NADH/FMN oxidoreductase and a chemical reduction by FMNH 2 . Eur. J. Biochem. 209, 621–627. 30. Laemmli, U.K. (1970) Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 227, 680–685. Ó FEBS 2004 P. denitrificans Fe(III) reductases (Eur. J. Biochem. 271) 561 31. Perrella, F.W. (1988) EZ-FIT: a practical curve-fitting micro- computer program for the analysis of enzyme kinetic data on IBM-PC compatible computers. Anal. Biochem. 174, 437–447. 32. Fieschi, F., Niviere, V., Frier, C., Decout, J L. & Fontecave, M. (1995) The mechanism and substrate specificity of the NADPH: flavin oxidoreductase from Escherichia coli. J. Biol. Chem. 270, 30392–30400. 33. Michaliszyn, G.A., Wing, S.S. & Meighen, E.A. (1977) Purifica- tion and properties of a NAD(P)H: flavin oxidoreductase from the luminous bacterium, Beneckea harveyi. J. Biol. Chem. 252, 7495–7498. 34. Jablonski, E. & DeLuca, M. (1977) Purification and properties of the NADH and NADPH specific FMN oxidoreductases from Beneckea harveyi. Biochemistry 16, 2932–2936. 35. Jablonski, E. & DeLuca, M. (1978) Studies of the control of luminiscence in Beneckea harveyi: properties of the NADH and NADPH: flavin oxidoreductases. Biochemistry 17, 672–678. 36. Lei, B., Liu, M., Huang, S. & Tu, S C. (1994) Vibrio harveyi NADPH-flavin oxidoreductase: Cloning, sequencing and over- expression of the gene and purification and characterization of the cloned enzyme. J. Bacteriol. 176, 3552–3558. 37. Tanner, J.J., Lei, B., Tu, S C. & Krause, K.L. (1996) Flavin reductase P: Structure of a dimeric enzyme that reduces flavin. Biochemistry 35, 13531–13539. 38. Liu, M., Lei, B., Ding, Q., Lee, J.C. & Tu, S C. (1997) Vibrio harveyi NADPH: FMN oxidoreductase: preparation and char- acterization of the apoenzyme and monomer-dimer equilibrium. Arch. Biochem. Biophys. 337, 89–95. 39. Lei, B. & Tu, S C. (1998) Mechanism of reduced flavin transfer from Vibrio harveyi NADPH-FMN oxidoreductase to luciferase. Biochemistry 37, 14623–14629. 40. Wang, H., Lei, B. & Tu, S C. (2000) Vibrio harveyi NADPH- FMN oxidoreductase Arg203 as a critical residue for NADPH recognition and binding. Biochemistry 39, 7813–7819. 41. Tu, S C., Becvar, J.E. & Hastings, J.W. (1979) Kinetic studies on the mechanism of bacterial NAD(P)H: flavin oxidoreductase. Arch. Biochem. Biophys. 198, 110–116. 42. Zenno, S., Saigo, K., Kanoh, H. & Inouye, S. (1994) Identification of the gene encoding the major NAD(P)H-flavin oxidoreductase of the bioluminescent bacterium Vibrio fischeri ATCC 7744. J. Bacteriol. 176, 3536–3543. 43. Inouye, S. (1994) NAD(P)H-flavin oxidoreductase from the bio- luminiscent bacterium, Vibrio fischeri ATCC 7744, is a flavopro- tein. FEBS Lett. 347, 163–168. 44. Koike, H., Sasaki, H., Kobori, T., Zenno, S., Saigo, K., Murphy, M.E.P., Adman, E.T. & Tanokura, M. (1998) 1.8 A ˚ crystal structure of the major NAD (P) H: FMN oxidoreductase of a bioluminiscent bacterium, Vibrio fischeri: overall structure, cofactor and substrate-analog binding, and comparison with related flavoproteins. J. Mol. Biol. 280, 259–273. 45. Bjellqvist, B., Hughes, G.J., Pasquali, Ch, Paquet, N., Ravier, F., Sanchez, J Ch, Frutiger, S. & Hochstrasser, D.F. (1993) The focusing positions of polypeptides in immobilized pH gradients can be predicted from their amino acid sequences. Electrophoresis 14, 1023–1031. 46. Ernster, L. (1987) DT-diaphorase: a historical review. Chem. Scripta 27A, 1–13. 47. Zenno, S., Koike, H., Tonokura, M. & Saigo, K. (1996) Gene cloning, purification, and characterization of NfsB, a minor oxy- gen-insensitive nitroreductase from Escherichia coli, similar in properties to FRase I, the major flavin reductase of Vibrio fischeri. J. Biochem. (Tokyo) 120, 736–744. 48. Zenno, S., Koike, H., Tonokura, M. & Saigo, K. (1996) Con- version of NfsB, a minor Escherichia coli nitroreductase, to a flavin reductase similar in biochemical properties to FRase I, the major flavin reductase in Vibrio fischeri, by a single amino acid sub- stitution. J. Bacteriol. 178, 4731–4733. 562 J. Mazoch et al. (Eur. J. Biochem. 271) Ó FEBS 2004 . Isolation and biochemical characterization of two soluble iron(III) reductases from Paracoccus denitrificans Jir ˇ ı ´ Mazoch 1 , Radek Tesar ˇ ı ´ k 2 ,. Republic Two soluble enzymes (FerA and FerB) catalyzing the reduction of a number of iron(III) complexes by NADH, were purified to near homogeneity from the aerobically grown iron-limited culture of Paracoccus. m M Tris/HCl, pH 7.4, and washed by the same buffer. The proteins were eluted in two phases with 300 mL of linear gradient of 0–0.6 M NaCl and 50 mL of linear gradient of 0.6–1 M NaCl at a flow rate of 0.8 mLÆmin )1 .Fractionsof5mL were