NapF Is a Cytoplasmic Iron-Sulfur Protein Required for Fe-S Cluster Assembly in the Periplasmic Nitrate Reductase*

The periplasmic nitrate reductase (Nap) is wide-spread in proteobacteria. NapA, the nitrate reductase catalytic subunit, contains a Mo-bisMGD cofactor and one [4Fe-4S] cluster. The nap gene clusters in many bacteria, including Rhodobacter sphaeroides DSM158, contain an napF gene, disruption of which drastically decreases both in vitro and in vivo nitrate reductase activities. In spite its importance in the Nap system, NapF has never been characterized biochemically, and its role remains unknown. The NapF protein has four polycysteine clusters that suggest that it is an iron-sulfur-containing protein. In the present study, a His6-tagged NapF protein was overproduced in Escherichia coli and purified anaerobically. The purified NapF protein was used to obtain polyclonal antibodies raised in rabbit, and cellular fractionation of R. sphaeroides followed by immunoprobing with anti-NapF antibodies revealed that the native NapF protein is located in the cytoplasm. This contrasts with the periplasmic location of the mature NapA. However, NapA could not be detected in an isogenic napF– strain of R. sphaeroides. The His6-tagged NapF protein displayed spectral properties indicative of Fe-S clusters, but these features were rapidly lost, suggesting cluster lability. However, reconstitution of the Fe-S centers into the apo-NapF protein was achieved in the presence of Azotobacter vinelandii cysteine desulfurase (NifS), and this allowed the recovery of nitrate reductase activity in NapA protein that had previously been treated with 2,2′-dipyridyl to remove the [4Fe-4S] cluster. This activity was not recovered in the absence of NapF. Taking into account the cytoplasmic localization of NapF, the presence of labile Fe-S clusters in the protein, the napF– strain phenotype, and the NapF-dependent reactivation of the 2,2′-dipyridyl-treated NapA, we propose a role for NapF in assembling the [4Fe-4S] center of the catalytic subunit NapA.

The periplasmic nitrate reduction (Nap) 1 system has been found in many different bacteria, and several physiological roles have been proposed depending on the organism, such as redox control to dissipate excess of reductant, anaerobic and aerobic denitrification, adaptation to anaerobic growth, and scavenging nitrate in nitrate-limiting environments (1)(2)(3)(4)(5). The function of the Rhodobacter sphaeroides DSM158 Nap system, which is encoded by the napKEFDABC transcription unit (6,7), is to regulate the cellular redox state by dissipating excess of reductant in cells growing under phototrophic conditions or with highly reduced carbon sources, such as butyrate or caproate (4,8). The Nap enzyme is a heterodimer composed of NapA, the Mo-bisMGD-containing catalytic subunit that also presents an N-terminal [4Fe-4S] center, and NapB, a diheme cytochrome c. NapC is a membrane-anchored cytochrome c that mediates the physiological electron transfer from the membrane quinol pool to the NapAB periplasmic heterodimer. The Nap system of R. sphaeroides also includes two putative small transmembrane proteins of unknown function (NapK and NapE), a soluble protein (NapD), which could be required for the maturation of the enzyme, and NapF, a putative 16-kDa cysteine-rich protein that has been suggested to bind four [4Fe-4S] centers (6,7). The R. sphaeroides wild-type strain lacks nitrite reductase, thus nitrite production can be used as an indicative of nitrate reductase activity in vivo. The mutational analysis on the R. sphaeroides nap genes reveals that the napABC genes are essential for periplasmic nitrate reduction, because mutants defective in these genes are unable to produce nitrite in vivo. Although the function of the NapF protein in the R. sphaeroides Nap system remains unknown, a mutation of the napF gene indicates that this protein is also required for in vivo nitrate reduction (7). In addition, loss of NapF results in a considerable decrease of the Nap activity assayed with methyl viologen (MV) as artificial electron donor (7). As reduced viologens seem to transfer electrons directly to the active site of the enzyme (9,10), it is unlikely that NapF participates in electron transfer to the periplasmic NapAB complex, even more if NapF is a cytoplasmic protein. Accordingly to this, the R. sphaeroides napC Ϫ strain is unable to produce nitrite in vivo but shows the same MV-dependent nitrate reductase activity as the wild-type strain (6,7). However, the subcellular localization of NapF is controversial due to the presence of a conserved twin arginine motif, which could be involved in the periplasmic targeting of the protein, although this motif is not followed by a hydrophobic region (7). The Escherichia coli Nap system includes two iron-sulfur proteins, NapG and NapH, in addition to NapF (11). Although napFGH genes are not essential for periplasmic nitrate reduction in E. coli (12), the loss of NapF results in a severe growth defect in a NapG ϩ H ϩ strain, but not in an napGH deletion mutant (13). It has been proposed that electrons can flow from menaquinol to the NapC subunit in E. coli. A membrane-anchored NapGH complex could act as a proton translocating dehydrogenase transferring electrons from ubiquinol and catalyzing an effective electron transfer to the periplasmic NapAB complex. Nevertheless, no function has been assessed to NapF in the Nap system of E. coli (11)(12)(13). Fe-S clusters act as cofactors in many different proteins such as electron carriers, environmental sensors, substrate transporters, or regulatory proteins participating in control of gene expression (14,15). Spontaneous in vitro Fe-S cluster assembly can occur, but in vivo Fe-S assembly requires accessory proteins (16). Studies of Fe-S cluster assembly in the Azotobacter vinelandii nitrogenase revealed the requirement of two proteins, NifS and NifU (14). Homologues of these two proteins are found in almost all organisms, from bacteria to humans. NifS is a pyridoxal phosphate-dependent cysteine desulfurase that mobilizes sulfur from L-cysteine. NifU is the Fe-S cluster scaffold upon which the nascent cluster is constructed to be transfer to an apo-protein (14). In E. coli, the isc and suf operons are required for Fe-S biogenesis of different iron-sulfur proteins (17)(18)(19)(20). The Isc system is the housekeeping Fe-S cluster assembly, whereas the Suf system is important for Fe-S biogenesis under stressful conditions. In this system, SufA plays a role as scaffold protein for assembly of iron-sulfur clusters and delivery to target proteins, SufS is a cysteine desulfurase that mobilizes the sulfur atom from cysteine and provides it to the cluster, and SufE binds to SufS and is responsible for a 50-fold stimulation of the cysteine desulfurase activity of SufS (19).
In this study we report the heterologous expression in E. coli of the R. sphaeroides NapF protein fused to an N-terminal His 6 motif. NapF purification allowed the study of its spectroscopic properties and the isolation of anti-NapF antibodies raised in rabbit to assess the subcellular localization of the native protein in R. sphaeroides. Demonstration of the implication of the R. sphaeroides NapF protein in the assembly of the iron-sulfur center of the catalytic subunit (NapA) is also presented.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Plasmids-The bacterial strains and plasmids used in this work are listed in Table I. The E. coli strains were cultured in Luria-Bertani (LB) medium or on LB agar plates at 37°C (22). pBluescript and pSVB25 were used routinely in gene manipulation, and the pKPD60 vector was used for the construction of alkaline phosphatase gene fusions. E. coli cells harboring any of the plasmids used in this work were cultured in the presence of ampicillin at a final concentration of 100 g ml Ϫ1 .
Enzyme Assays, Analytical Methods, and Spectroscopic Analyses-Nitrate reductase was assayed with reduced methyl viologen as artificial electron donor, and the nitrite produced in the reaction was determined colorimetrically (25). Alkaline phosphatase activity was assayed in subcellular fractions of E. coli with p-nitrophenyl phosphate (24,26). Malate dehydrogenase activity was measured in subcellular fractions of R. sphaeroides by following NADH oxidation in a spectrophotometer at 340 nm (27). Succinate dehydrogenase was assayed in subcellular fractions of R. sphaeroides by the phenazine methosulfate-dependent reduction of dichlorophenolindophenol (28). Labile sulfur determination was performed by a reaction with N-N-dimethyl-p-phenylenediamine and FeCl 3 following described methods (29,30) and using a calibration curve with Na 2 S⅐9H 2 O in NaOH (under nitrogen atmosphere) as standard. Labile iron was assayed in samples heated at 80°C for 10 min, by reaction with sodium methasulfite and bathophenanthroline, as previously described (31), and using a calibration curve with FeSO 4 ⅐7H 2 O as standard. To determine iron and sulfur labile in the apo-NapF reconstitution assay, low molecular mass iron and sulfide compounds, not bond by the protein, were removed by gel filtration. Protein concentration was estimated by the Lowry procedure using bovine serum albumin (BSA) as standard (32).
The oxidized and reduced UV-visible spectra of NapF were obtained by oxidizing the protein with potassium ferricyanide or using dithionite as reductant, respectively. The spectra were recorded in a DU7500 (Beckman) spectrophotometer. The form B molybdopterin derivative was extracted from the native NapA protein and the dipyridyl-treated NapA samples by a modification of the procedure of Johnson and Rajagopalan (33). 1 ml of sample (0.5 mg ml Ϫ1 of enzyme) in 10 mM Tris-HCl (pH 7.0) was acidified to pH 2.5 with concentrated HCl. The samples were incubated in a boiling water bath for 20 min and centrifuged at 19,000 ϫ g for 10 min. Fluorescence excitation and emission spectra of the supernatants were recorded using a PerkinElmer Life Sciences LS-5 luminescence spectrometer. Surface-enhanced laser desorption ionization mass spectrometry was used to determine, through time of flight, the His 6 -tagged NapF molecular mass by using an immobilized metal affinity capture surface.
Subcellular Fractionation-Subcellular fractionations of E. coli and R. sphaeroides were carried out as previously described (34). Cells were harvested by centrifugation and washed in 100 ml of 0.05 M Tris-HCl buffer (pH 8.0). A cell pellet was obtained by centrifugation and resuspended in 50 ml of sucrose buffer (75 mM Tris-HCl, pH 8.0; 20 mM EDTA; 100 mM NaCl; 0.5 M sucrose) and 100 g ml Ϫ1 lysozyme and then incubated for 30 min at 30°C. After centrifugation for 30 min at 19,000 ϫ g, the supernatant (periplasmic fraction) was separated from the pellet (spheroplasts), which was resuspended in 75 mM Tris-HCl, pH 8.0, buffer. The spheroplasts were then broken by cavitation (3 pulses of 5 s at 90 watts), and the cell extract was centrifuged at 200,000 ϫ g for 45 min. Then, the supernatant (cytoplasmic fraction) was separated from the pellet (membrane fraction). Several enzymatic activities were measured as markers of purity of each subcellular fraction of R. sphaeroides. Thus, periplasmic nitrate reductase (25) was found in the periplasmic fraction, malate dehydrogenase activity (27) was detected in the cytoplasmic fraction, and succinate dehydrogenase (28) was found in the membrane fraction. Western Blots and Heme or Protein Staining-For electrophoretic separation, samples were loaded onto polyacrylamide gels, with 14% (w/v) resolving gels and 5% stacking gels. These gels were used in Western blots, heme analysis, or protein staining with Coomassie Brilliant Blue or silver. Immunoprobing analyses to detect the His 6 -tagged NapF protein in E. coli were performed by using monoclonal antipolyhistidine clone his-1 from mouse ascites fluid as primary antibody and anti-mouse IgG alkaline phosphatase conjugate from goat as second antibody. Immunoprobing analyses to detect the native NapF protein in R. sphaeroides were carried out with polyclonal anti-NapF antibodies raised in rabbit and anti-rabbit IgG alkaline phosphatase conjugate from goat. Immunoprobing analyses to detect the NapA protein in R. sphaeroides were carried out with polyclonal anti-NapA antibodies and anti-rabbit IgG alkaline phosphatase conjugate from goat. Heme staining gels were carried out with dimetoxybenzidine dihydrochloride (35). The reaction was developed with 0.5 M sodium citrate (pH 4.4) and 300 l of 30% H 2 O 2 . Silver staining gels were performed in the presence of sodium acetate, sodium thiosulfate, 25% glutaraldehyde, 2.5% silver nitrate, and 37% formaldehyde (36).
DNA Manipulations-DNA manipulations were performed by using standard procedures (22). A fusion between the 5Ј-end of the napF gene and the alkaline phosphatase gene (phoA) from the plasmid pKPD60 (24) has been performed. For this purpose, the 2.17-kb PstI/BamHI fragment that contains the R. sphaeroides napKEFD genes and the 5Ј-end of napA was ligated into the vector pALTER to generate pAL-TER-PB. To create in the 26th triplet of the napF gene a recognition site for the restriction enzyme XhoI, an amplification by PCR with the primers 5Ј-TGTCGGCCTCGAGCGTCCAGGGC-3Ј (XhoI site underlined) and 5Ј-CCACGCACTTTGCCTCGAGATC-3Ј was performed. The restriction enzyme XhoI was used to digest the isolated PCR product, generating a 1.2-kb fragment that contains the napKE genes and the 5Ј-end of the napF gene. This fragment was then ligated into the pKPD60 vector previously digested with XhoI and SalI to generate the pKPD60X construct. In the alkaline phosphatase measurements, the pKPD60 vector was used as a positive control, because this plasmid contains the signal peptide of the periplasmic cytochrome c 550 of Paracoccus denitrificans fused to the phoA gene. A negative control was performed by digestion of the pKPD60 vector with XhoI and SalI and further re-ligation to generate the pKPD60⌬X plasmid, thus removing a 0.5-kb fragment, which contains the cytochrome c 550 signal peptide. A napA-phoA gene fusion was also constructed by inserting the 2.17-kb PstI/BamHI fragment, which contains the napKEFD genes and the 5Ј-end region of the napA gene, into the pSVB25 vector to generate pSVB25X60. This construct was digested with BamHI, and the linear fragment was partially filled-in with the Klenow polymerase in the presence of dGTP and dATP. Thus, compatible ends were generated to be cloned into the vector pKPD60, previously digested with XhoI and partially filled-in with the Klenow polymerase in the presence of dCTP and dTTP, to produce the last construct pKPD60B*X* with the desired in-frame napA-phoA gene fusion. The plasmids pKPD60, pKPD60X, pKPD60⌬X, and pKPD60B*X* were sequenced to check that all the constructions were made in the correct reading frame, and were introduced into E. coli strains DH5␣ and XL1-blue. To generate the His 6tagged NapF recombinant protein, the PCR mutagenesis technique was performed by using as template the plasmid pFR24 carrying the nap-KEFD genes and the 5Ј-end of the napA gene, which presents a BamHI restriction site, and the primers: 5Ј-CTGACAGCCGTGGATCCTATC-CCGC-3Ј (BamHI site underlined) and 5Ј-GGAAACAGCTATGAC-CATG-3Ј. The PCR product was digested with BamHI and cloned into pQE32, to generate pQE32/napF. This plasmid was sequenced to confirm that the His 6 tag was correctly fused to NapF. The napF gene sequence has been deposited in the EMBL, GenBank TM , and DDBJ Nucleotide Sequence Databases under the accession number Z46806.
Overproduction and Purification of NapF-The E. coli JM109 strain harboring the pQE32/napF plasmid was cultured anaerobically at 37°C. Anaerobic conditions were achieved by filling completely 500-ml screw-capped bottles with LB culture medium supplemented with ampicillin (100 g ml Ϫ1 ). When the cultures reached an A 600 of 0.4, a heat-shock of 10 min at 42°C was carried out and 50 M of IPTG were added, keeping the cultures at 25°C overnight for the NapF induction.
The cytoplasmic fraction of E. coli containing the heterologous NapF protein was loaded onto a nickel-nitrilotriacetic acid-agarose (Qiagen) column, and an imidazole gradient was performed (up to 250 mM imidazole) to elute the recombinant His 6 -tagged NapF protein.
Isolation and Purification of the Anti-NapF Polyclonal Antibodies-Polyclonal antibodies raised against the purified NapF were obtained by following the method previously described by Diez and López-Ruiz (37). A volume of 250 l containing 135 g of purified NapF in 50 mM Tris-HCl, pH 8.0, was diluted up to 2 ml in phosphate-buffered saline buffer (135 mM NaCl, 2.6 mM KCl, 1.5 mM KH 2 PO 4 , 8 mM Na 2 HPO 4 , pH 8.0), mixed with 2 ml of adjuvant complete Freund (Difco Laboratories), and homogenized to obtain an uniform sample. The immunization process was carried out during 59 days. Rabbit-blood samples were collected in the presence of heparin, centrifuged at 4°C and 19,000 ϫ g for 15 min, and after incubation at 56°C for 15 min, were frozen at Ϫ80°C until use. Immunoglobulins were partially purified by a 40% ammonium sulfate precipitation. The pellet was washed with 1.75 M ammonium sulfate and resuspended in 17.5 mM phosphate buffer, pH 7.0. Lipoproteins were eliminated by centrifugation, and the supernatant was loaded onto an ionic exchange DEAE-Sephacel column with the phosphate buffer described above. The fractions that contributed at the first peak at 280 nm were collected, representing the IgG fraction. The anti-NapF antibodies were stored at 4°C until use.
In Vitro Reconstitution of the Iron-Sulfur Centers of NapF-The assembly of the iron-sulfur clusters to the apo-NapF protein was performed under anaerobic conditions (nitrogen atmosphere) in a total volume of 1 ml, by adding: 12.5 mM Tris-HCl buffer, pH 7.4; 50 mM KCl; 5 nmol of NapF; 1 mM L-cysteine; 2.5 mM DTT; 0.3 nmol of cysteine desulfurase (NifS) from A. vinelandii; and 2 mM ferrous ammonium sulfate, as previously described (38). The reconstitution was monitored in a Hitachi U-3310 spectrophotometer by recording scans between 350 and 700 nm. During the reconstitution process, a yellow-brown color was developed that correlated with the formation of an absorption peak at 420 nm. When NapF, NifS, DTT, Fe(II), or cysteine was omitted from the reconstitution mixture, there was no increase in absorbance at 420 nm nor formation of the yellow-brown color. An additional control was performed by using 0.4 nmol of BSA instead of NapF, and no peak at 420 nm was observed.
Generation of an Inactive NapA Form and Its Reactivation by NapF-To generate an inactive NapA protein devoid of its [4Fe-4S] cluster, the purified NapA protein was incubated with the iron chelator 2,2Ј-dipyridyl (DP-treated NapA) under anaerobic conditions (nitrogen atmosphere). The sample, in a total volume of 0.6 ml, contained: 12.5 mM Tris-HCl buffer, pH 7.4; 0.39 mol of NapA; 7.8 mol of 2,2Јdipyridyl; 1 mM sodium dithionite; and 5 mM MgATP. The reaction mixture was incubated at room temperature for 20 min (39) and was loaded onto a Sephadex G-25 column to remove low molecular mass compounds. The fractions were eluted in 12.5 mM Tris-HCl buffer, pH 7.4. This sample was used as the DP-treated NapA form in the nitrate reductase reconstitution assay that was carried out under anaerobic conditions (nitrogen atmosphere) in a total volume of 1 ml, by adding: 12.5 mM Tris-HCl buffer, pH 7.4; 16 nmol of DP-treated NapA; 2.5 nmol of NapF; 1 mM cysteine; 2.5 mM DTT; 0.3 nmol of cysteine desulfurase (NifS) from A. vinelandii; and 2 mM ferrous ammonium sulfate. Samples were collected at different times, and nitrate reductase was assayed with methyl viologen as reductant. When the reconstitution assay was performed using 0.4 nmol of BSA instead of NapF, or when NapF, NifS, DTT, Fe 2ϩ , or cysteine was omitted, nitrate reductase activity was not recovered.

RESULTS
Sequence Analysis and Characteristics of the NapF Protein of R. sphaeroides DSM158 -NapF has four polycysteine clusters that suggest that it is an iron-sulfur-containing protein (Fig. 1). In addition, sequence comparison of the NapF proteins shows the presence of a conserved twin arginine motif at the Nterminal end, which could act as a possible signal for its translocation to the periplasm through the Tat pathway (Fig. 1A). However, this twin arginine motif is not followed by the conserved region XFLK and the hydrophobic sequence required for the exportation to the periplasm (40, 41). Thus, a cytosolic localization of NapF is more likely. The hydropathy profile reveals that most of NapF is very hydrophobic, with the cysteine clusters for binding of iron-sulfur centers located in these hydrophobic regions (Fig. 1B), although NapF does not contain transmembrane helices.
Overproduction and Purification of the NapF Protein-A His 6 -tagged NapF protein was overproduced in E. coli and purified, as described under "Experimental Procedures." The R. sphaeroides NapF protein was synthesized as a fusion protein with a His 6 motif at the N terminus of the protein. This fusion was constructed in the pQE32 vector (Qiagen) under the lac promoter, for IPTG induction when expressed in E. coli JM109. Subcellular fractions of the strains JM109 (pQE32) and JM109 (pQE32/napF) of E. coli were isolated from cells cultured anaerobically with IPTG. Western blots of the subcellular fractions using polyclonal anti-His tag antibodies revealed that His 6 -tagged NapF protein was expressed and localized in the cytoplasm of the strain JM109 (pQE32/napF) but was absent in the cytoplasmic fraction of the control strain JM109 (pQE32) of E. coli (Fig. 2A). The His 6 -NapF protein was not detected in the periplasmic fraction of the strain JM109 (pQE32) or in the periplasmic fraction of the strain JM109 (pQE32/napF) ( Fig.  2A). This result indicates that in E. coli the recombinant NapF protein is only localized in the cytoplasmic fraction. Purification of this soluble His 6 -tagged NapF protein was undertaken from the cytoplasmic fraction, which was isolated and loaded onto a nickel-nitrilotriacetic acid-agarose column. The recombinant NapF protein was eluted using a gradient between 5 and 250 mM imidazole. The different chromatographic fractions were analyzed in SDS-polyacrylamide gels by either Coomassie Blue stain (not shown) or silver stain (Fig. 2B). These gels A, amino acid sequence comparison of the NapF proteins from R. sphaeroides (7), Sinorhizobium meliloti (accession number NP435922), Pseudomonas aeruginosa (accession number NP249867), and E. coli (11). Identical amino acid residues in at least three of the four sequences are marked in bold. The putative twin arginine motif for periplasmic targeting (40) is underlined, and the four-cysteine motifs for binding of the iron-sulfur clusters are indicated with double bars. B, the hydropathy plot of the R. sphaeroides NapF protein is shown according to Kyte and Doolittle (44). The amino acid sequence is also shown, and the four-cysteine motifs are marked with double bars. revealed successful purification of a 16-kDa protein with a yield of 80 g l Ϫ1 . Analysis of NapF by surface-enhanced laser desorption ionization revealed a molecular mass of 17,626 Da. This value is almost identical to the predicted mass of 17,614 Da for the His 6 -tagged NapF protein (not shown).
UV-visible Spectra of the Iron-Sulfur Protein NapF-The UV-visible spectra of the recombinant NapF purified under anaerobic conditions showed a peak at 303 nm when oxidized with potassium ferricyanide (not shown) and two peaks at 310 nm and 420 nm when reduced with dithionite (Fig. 2C), indicating that NapF is an iron-sulfur protein. However, the 420-nm peak, which showed an absorbance value of 0.18, was rapidly lost suggesting that iron-sulfur centers of NapF are very labile. An approach for in vitro attachment of iron-sulfur clusters to the apo-protein was undertaken (38), in which apo-NapF protein was incubated anaerobically in the presence of L-cysteine, ferrous ammonium sulfate, DTT, and NifS (cysteine desulfurase) of A. vinelandii. UV-visible spectra were recorded during the reconstitution process, and an increasing peak at 420 nm was observed (Fig. 3). In addition, the sample was turned yellow-brown colored mostly within 60 min. This color change and the 420-nm peak were not observed in a control assay where BSA was used instead of NapF. These features were also absent when NapF, NifS, DTT, Fe 2ϩ , or cysteine was omitted from the reconstitution mixture. Before reconstitution, the contents of acid labile iron and sulfide were 1 mol and 0.9 mol per mol of NapF, respectively. During reconstitution, these contents were 3 mol of Fe and 2.5 mol of S 2Ϫ mol Ϫ1 NapF after 15-min incubation and 7.7 mol of Fe and 7.5 mol of S 2Ϫ mol Ϫ1 NapF after 90 min. When reconstitution was performed for more than 90 min, an excess of iron and sulfur produced precipitation that correlated with a color change from yellowbrown to dark brown. The 420-nm peak could not be detected when anaerobic conditions were omitted. Subcellular Localization of NapF in R. sphaeroides DSM158 -A subcellular localization of the R. sphaeroides NapF protein is essential to assess its physiological function. To investigate the NapF localization, a fusion between the 5Ј-end of the R. sphaeroides napF gene and the phoA gene was generated, and the alkaline phosphatase activity was assayed in periplasmic and cytoplasmic fractions of the E. coli cells transformed with this construct (Table II). A fusion between the phoA gene and the periplasmic cytochrome c 550 of Paracoccus denitrificans (pKPD60 plasmid), which presents a signal peptide for translocation through the Sec pathway, was used as a positive control. As a negative control, the sequence of the cytochrome c 550 signal peptide was deleted (pKPD60⌬SX). In addition, the 5Ј sequence of the napA gene of R. sphaeroides coding for a typical twin arginine signal peptide was fused to the phoA gene (pKPD60B*X*). Recently, it has been described that the NapA protein is translocated to the periplasm by the Tat pathway (41,42). However, cells carrying the napF-phoA FIG. 2. Subcellular localization of the recombinant His 6tagged NapF in E. coli, with purification and UV-visible spectrum of the purified NapF. A, localization of His 6 -NapF in E. coli by immunoprobing with anti-His 6 antibodies: lane a, periplasmic fraction isolated from the strain JM109 (pQE32); lane b, periplasmic fraction isolated from the strain JM109 (pQE32/napF); lane c, cytoplasmic fraction isolated from the strain JM109 (pQE32); lane d, cytoplasmic fraction isolated from the strain JM109 (pQE32/napF). The immunodetection analysis was carried out four times with a total protein concentration in the different subcellular fractions in a range of 20 -100 g ml Ϫ1 . B, the recombinant NapF protein was purified by a nickel-nitrilotriacetic acid-agarose column, and several chromatographic fractions eluted by using an imidazole gradient were loaded onto a polyacrylamide gel, which was silver-stained. C, spectrum of the dithionitereduced NapF. Protein concentration was 0.4 mg ml Ϫ1 .

FIG. 3. Reconstitution of the iron-sulfur clusters of NapF.
The reconstitution was performed in the presence of L-cysteine, NifS (cysteine desulfurase) of A. vinelandii, ferrous ammonium sulfate, and DTT. The spectra were recorded at several times from 350 to 700 nm during 90 min. Line a, 0 min; line b, 15 min; line d, 90 min. An assay control replacing the protein NapF for BSA is also shown. Lines c 1 /c 2 correspond to spectra of 0 and 90 min of this assay control, respectively. fusion (pKPD60X) or the napA-phoA fusion (pKPD60B*X*) were devoid of alkaline phosphatase activity in the periplasm (Table II). This result indicates that the twin arginine motif does not allow the translocation of alkaline phosphatase, thus making inadequate this approach to investigate the subcellular location of NapF. Therefore, to determine the subcellular localization of the R. sphaeroides NapF protein, the purified NapF was used to obtain polyclonal antibodies, which were titrated by dot blotting (Fig. 4A). After a partial purification, the polyclonal anti-NapF antibodies raised in rabbit were used in Western blots with the subcellular fractions isolated from R. spha-eroides. This immunological analysis showed that NapF is only present in the cytoplasm of the wild-type cells and is absent in the cytoplasm of the isogenic napF Ϫ mutant strain (Fig. 4B). The NapF protein was not detected in the periplasmic fractions of the wild-type or the napF Ϫ strains (Fig. 4B). When the membrane fractions of the wild-type or the napF Ϫ strains were used, NapF was also undetectable (not shown). As a positive control, 5 g of purified NapF was used (Fig 4B). To check the purity of each subcellular fraction of R. sphaeroides, several enzymatic activities were measured. The periplasmic nitrate reductase was only found in the periplasm, malate dehydrogenase was only detected in the cytoplasm, and succinate dehydrogenase was only found in the membrane fraction (not shown). This immunological analysis was carried out four times with different amounts of total protein in a range of 0.1-0.4 mg ml Ϫ1 , indicating that the native NapF protein is localized in the cytoplasm of R. sphaeroides DSM158.
Phenotype of the napF Ϫ Strain from R. sphaeroides-Nitrate reduction is severely impaired in an isogenic napF Ϫ mutant strain of R. sphaeroides both in vivo and in vitro (7). It is worth noting that MV donates electrons directly to the active site of the nitrate reductase (9,10), and, for this reason, a mutant in the napC gene shows the same MV-dependent Nap activity as the wild-type strain (7). On the contrary, the napF Ϫ mutant shows only a very low activity with MV as artificial electron donor (7). This low Nap activity of the napF Ϫ mutant strain is not due to a decrease on nap gene expression, because this mutant showed even a slightly higher level of the c-type cytochrome NapB than the wild-type in a heme-stained gel (Fig.  5A). In addition, an immunological analysis revealed that levels of the catalytic subunit NapA in the soluble (periplasmic and cytoplasmic) fractions of the napF Ϫ strain are undetectable (Fig. 5B). NapA was also undetectable in membrane fractions FIG. 4. Subcellular localization of NapF in R. sphaeroides DSM158. A, titration of serum extraction in the presence of primary antibodies. Anti-NapF polyclonal antibodies were raised in rabbit as described under "Experimental Procedures." Dilutions of purified NapF (0.5 mg ml Ϫ1 ) were used: a, 1:50 dilution; b, 1:10 dilution; c, 1:5 dilution. The top dots represent the serum extraction at a 1:100 dilution containing the primary antibodies, and the bottom dots represent the preimmune serum at a 1:100 dilution. B, localization of NapF in R. sphaeroides DSM158 by immunoprobing with the anti-NapF polyclonal antibodies: a, cytoplasmic fraction isolated from the wild-type strain; b, cytoplasmic fraction isolated from the napF Ϫ strain; c, 5 g of purified NapF; d, periplasmic fraction isolated from the wild-type strain; e, periplasmic fraction isolated from the napF Ϫ strain. The immunodetection analysis was carried out four times with a total protein concentration in the different subcellular fractions in a range of 0.1-0.4 mg ml Ϫ1 .

FIG. 5. Heme staining gel of periplasmic fractions and immunoprobing of soluble fractions isolated from R. sphaeroides strains.
A, the periplasmic fractions were isolated as described under "Experimental Procedures" from the wild-type, napF Ϫ , and napB Ϫ strains, loaded onto a polyacrylamide gel, and stained to detect the heme groups. B, immunoprobing with anti-NapA antibodies of soluble (periplasmic and cytoplasmic) fractions from the wild-type and napF Ϫ strains. A control with the purified NapA protein was also included. These analyses were performed four times with a total protein concentration in the different subcellular fractions in a range of 0.1-0.4 mg ml Ϫ1 . of the napF Ϫ strain (not shown). The absence of the catalytic subunit NapA in the napF Ϫ strain is in agreement with an essential role of NapF in the NapA maturation, probably in the assembly of its [4Fe-4S] cluster.
Reconstitution of the Nitrate Reductase Activity of the DPtreated NapA Protein by NapF-The generation of an inactive NapA protein devoid of its [4Fe-4S] cluster was undertaken as previously described for the nitrogenase (39). The purified NapA protein was incubated anaerobically in the presence of 2,2Ј-dipyridyl (DP), as indicated under "Experimental Procedures." About 88% of the MV-dependent nitrate reductase activity was lost after 20 min of incubation with this iron chelator. The molybdenum cofactor was still present in the DPtreated NapA form as deduced from the fluorescence excitation and emission spectra of the extracted form B molybdopterin derivative (not shown). The content of acid-labile iron in the NapA protein before the treatment was 4.6 mol mol Ϫ1 NapA and after the 2,2Ј-dipyridyl treatment was 0.4 mol mol Ϫ1 NapA, suggesting that the [4Fe-4S] cluster was absent in the DP-treated sample. The reconstitution of the nitrate reductase activity was carried out anaerobically with the DP-treated NapA protein, in the presence of NifS (cysteine desulfurase) of A. vinelandii, L-cysteine, ferrous ammonium sulfate, DTT, and the NapF protein. Under these conditions, a rapid increase in the nitrate reductase activity assayed with reduced methyl viologen as electron donor was observed (Fig. 6). This increase in the nitrate reductase activity was not found when NapF was omitted or replaced by BSA (Fig. 6). In addition, in the absence of NifS, DTT, Fe 2ϩ , or cysteine, reconstitution of the reductase activity was not observed. The reactivation of NapA was also sensitive to pretreatment of NapF with air, and this is in accordance with the absence of the 420-nm peak in the NapF reconstitution assay under aerobic conditions. These results indicated that iron-sulfur centers in NapF are required for the recovery of the nitrate reductase in the DP-treated NapA protein. DISCUSSION The periplasmic nitrate reduction system has been found in a phylogenetically wide range of bacteria. Although most of these Nap systems include the napF gene, no function has been assessed for this component (7,(11)(12)(13). The NapF sequence analysis reveals that polycysteine clusters are indicative of the presence of iron-sulfur clusters (Fig. 1). The conserved twin arginine motif in the N terminus of NapF suggests that this protein could be translocated to the periplasm through the export Tat pathway. This system is specific for folded metalloproteins, which presents a hydrophobic signal peptide (26 -58 amino acids) with the conserved RRXFLK motif (40 -42). However, the NapF sequence lacks the conserved residues and the hydrophobic region following this twin arginine motif (Fig. 1A), and thus, a cytoplasmic localization can be considered. In addition, the NapF hydropathy plot reveals that NapF does not contain transmembrane helices but is a very hydrophobic protein (Fig. 1B).
To investigate this further, an anaerobic overproduction of a His 6 -tagged NapF protein was carried out in E. coli cells. The recombinant protein was localized in the cytoplasm with anti-His 6 antibodies ( Fig. 2A) and purified with an imidazole gradient (Fig. 2B). The dithionite-reduced spectrum of the Histagged NapF showed a peak at 420 nm (Fig. 2C) that suggests the presence of Fe-S centers in the molecule (43). Because the spectrum was performed with a total protein concentration of 22.7 M, and the absorbance at 420 nm was about 0.18, it can be deduced an absorption coefficient (⑀) at 420 nm of about 8 mM Ϫ1 cm Ϫ1 . It has been described previously that the ⑀ 420 nm is ϳ4 mM Ϫ1 cm Ϫ1 for proteins with one [4Fe-4S] center (43), and therefore, only ϳ50% of the expected [4Fe-4S] centers are present in the purified NapF. Further attempts to determine labile iron and sulfur by analytical procedures were unsuccessful, because the sample lost the 420-nm peak very rapidly, thus making it difficult to assign a specific type of Fe-S centers to the NapF protein. In light of this, the in vitro attachment of iron-sulfur clusters to NapF has been successfully undertaken (38), increasing the peak at 420 nm during reconstitution (Fig.  3). This correlates with the fact that the sample was turned brown-colored mostly within 90 min. In addition, iron and sulfur labile determinations in the reconstituted NapF protein reveals that ϳ50% of the expected [4Fe-4S] are present in the protein. The presence of the 420-nm peak in the optical spectrum and the development of the yellow-brown color are typical for [4Fe-4S] clusters formation (38,43). However, the presence of other types of Fe-S clusters, such as [2Fe-2S], or the formation of dimers/multimers of NapF that may act as a functional unit can not be excluded.
To determine the subcellular localization of NapF in R. sphaeroides DSM158, the purified NapF protein was used to obtain A similar model has been described for the Suf system of E. coli (19). FIG. 6. Reactivation of the DP-treated NapA form by the NapF protein. The whole assay was carried out as indicated under "Experimental Procedures" and included the DP-treated NapA, NapF and NifS (circles). Two negative assays without NapF, either with only DPtreated NapA and NifS (triangles) or with DP-treated NapA, NifS, and BSA instead of NapF (squares) are also shown. The NapF concentration used was 40 g. specific antibodies raised in rabbit against the recombinant NapF protein. Immunoprobing with the polyclonal antibodies anti-NapF of the subcellular fractions of R. sphaeroides revealed that the native NapF protein is localized in the cytoplasm of the R. sphaeroides cells (Fig. 4B) and is absent in the periplasmic fraction (Fig. 4B) or in the membrane fraction of this strain (not shown). As a negative control, subcellular fractions of an isogenic napF Ϫ mutant strain were used, and NapF was never detected. As a positive control, 5 g of purified NapF was used (Fig. 4B). This result confirms that NapF is localized in the cytoplasm of R. sphaeroides DSM158. A different approach to demonstrate the NapF localization by alkaline phosphatase fusion was unsuccessful due to the inability of the Tat pathway to export a functional alkaline phosphatase, as revealed by the napA-phoA fusion (Table II). The PhoA protein is translocated by the Sec route in an unfolded state and acquires its native conformation in the periplasm, whereas the Tat pathway seems to translocate folded proteins that acquire their cofactors in the cytoplasm (40), as described for the NapA protein (41,42). Therefore, it can be deduced that the twin arginine signal peptides do not allow the periplasmic translocation of a functional alkaline phosphatase. However, the immunoprobing with polyclonal antibodies anti-NapF demonstrated that NapF has a cytoplasmic localization in R. sphaeroides DSM158. This cytoplasmic location of NapF contrasts with the periplasmic localization of the mature NapA but is compatible with a role of this protein in processing or assembling the iron-sulfur center of NapA, rather than as an electron donor to the periplasmic NapAB complex.
On the other hand, nitrate reduction is severely impaired in a napF Ϫ mutant strain of R. sphaeroides, and only a very low nitrate reductase activity is observed both in vivo and in vitro (7). Interestingly, the NapA protein was undetectable in the napF Ϫ strain (Fig. 5B), probably because, in this mutant, the NapA protein lacks its [4Fe-4S] cluster and this inactive protein is degraded in the cytoplasm. This result agrees with a possible role of NapF in the assembling of the [4Fe-4S] center of NapA prior to its translocation to the periplasm. To confirm this function, an inactive DP-treated NapA protein was obtained by incubation of the native NapA protein with 2,2Јdipyridyl. Fluorescence spectra of the extracted form B molybdopterin derivative and iron content determination revealed that the DP-treated NapA protein contains the molybdenum cofactor but lacks the [4Fe-4S] center. The reconstitution of the MV-nitrate reductase of NapA was efficiently catalyzed by NapF (Fig. 6). When cysteine, DTT, or NifS were omitted from the reconstitution assay, the active form of NapA could not be detected (Fig. 6). In addition, when the reconstituted NapF protein is exposed to air, the 420-nm peak decreases rapidly suggesting that its iron-sulfur centers are labile in the presence of oxygen. No further nitrate reductase activity recovery can be achieved with this sample. These results confirm that NapF requires in the first instance the presence of cysteine, DTT and NifS to acquire its iron-sulfur centers and then, NapF participates in the [4Fe-4S] cluster assembly of NapA (Fig. 7). A similar model has been proposed for the Suf system of E. coli with SufA acting as a scaffold protein for assembly of ironsulfur clusters and delivery to target proteins, and SufS showing a cysteine desulfurase activity (19). It is worth noting that NapA is translocated to the periplasm by the Tat pathway and Tat-deficient strains accumulate in the cytoplasm an active cofactor-containing NapA protein (41), suggesting that NapF is transferring the [4Fe-4S] cluster to NapA in the cytoplasm independently of the fact of its further exportation to the periplasm.
In conclusion, this work shows that NapF is synthesized as a cytoplasmic protein in R. sphaeroides. This makes it unlikely that the putative iron-sulfur clusters of the cytoplasmic NapF are involved in electron transfer to the terminal periplasmic NapA nitrate reductase. In addition, a napF Ϫ mutant strain of R. sphaeroides is impaired in the synthesis of correctly localized and active NapA. This protein binds a Mo-bisMGD cofactor and one [4Fe-4S] cluster, and both cofactors are attached to the protein in the cytoplasm prior to export by the Tat translocase. This raises the possibility of a role for NapF in the biogenesis of the [4Fe-4S] center of NapA. This is also in agreement with the presence of highly labile iron-sulfur clusters in NapF and was clearly demonstrated by the NapF-dependent reconstitution of the DP-treated NapA form. We propose that NapF interacts with NapA in the cytoplasm for the assembly of the [4Fe-4S] center and avoids its translocation via the Tat machinery, probably by the interaction of the NapF N-terminal twin arginine motif with the Tat components, until the biogenesis of the [4Fe-4S] cluster of NapA is completed.