Purification and Biophysical Characterization of a New [2Fe-2S] Ferredoxin from Azotobacter vinelandii, a Putative [Fe-S] Cluster Assembly/Repair Protein*

During the purification of site-directed mutant variants of Azotobacter vinelandii ferredoxin I (FdI), a pink protein, which was not observed in native FdI preparations, appeared to associate specifically with variants that had mutations in ligands to FdI [Fe-S] clusters. That protein, which we designate FdIV, has now been purified. NH2-terminal sequence analysis revealed that the protein is the product of a previously described gene, herein designated fdxD, that is in the A. vinelandii iscSUA operon that encodes proteins involved in iron-sulfur cluster assembly or repair. An apoprotein molecular mass of 12,434.03 ± 0.21 Da was determined by mass spectrometry consistent with the known gene sequence. The monomeric protein was shown to contain a single [2Fe-2S]2+/+ cluster by UV/visible, CD, and EPR spectroscopies with a reduction potential of −344 mV versus the standard hydrogen electrode. When overexpressed in Escherichia coli, recombinant FdIV holoprotein was successfully assembled. However, the polypeptide of the recombinant protein was modified in some way such that the apoprotein molecular mass increased by 52 Da. Antibodies raised against FdIV and EPR spectroscopy were used to examine the relative levels of FdIV and FdI in various A. vinelandii strains leading to the conclusion that FdIV levels appear to be specifically increased under conditions where another protein, NADPH:ferredoxin reductase is also up-regulated. In that case, the fpr gene is known to be activated in response to oxidative stress. This suggests that thefdxD gene and other genes in the iron-sulfur cluster assembly or repair operon might be similarly up-regulated in response to oxidative stress.

During the purification of site-directed mutant variants of Azotobacter vinelandii ferredoxin I (FdI), a pink protein, which was not observed in native FdI preparations, appeared to associate specifically with variants that had mutations in ligands to FdI [Fe-S] clusters. That protein, which we designate FdIV, has now been purified. NH 2 -terminal sequence analysis revealed that the protein is the product of a previously described gene, herein designated fdxD, that is in the A. vinelandii iscSUA operon that encodes proteins involved in ironsulfur cluster assembly or repair. An apoprotein molecular mass of 12,434.03 ؎ 0.21 Da was determined by mass spectrometry consistent with the known gene sequence. The monomeric protein was shown to contain a single [2Fe-2S] 2؉/؉ cluster by UV/visible, CD, and EPR spectroscopies with a reduction potential of ؊344 mV versus the standard hydrogen electrode. When overexpressed in Escherichia coli, recombinant FdIV holoprotein was successfully assembled. However, the polypeptide of the recombinant protein was modified in some way such that the apoprotein molecular mass increased by 52 Da. Antibodies raised against FdIV and EPR spectroscopy were used to examine the relative levels of FdIV and FdI in various A. vinelandii strains leading to the conclusion that FdIV levels appear to be specifically increased under conditions where another protein, NADPH:ferredoxin reductase is also up-regulated. In that case, the fpr gene is known to be activated in response to oxidative stress. This suggests that the fdxD gene and other genes in the iron-sulfur cluster assembly or repair operon might be similarly up-regulated in response to oxidative stress.
In 1962 the isolation of the first iron-sulfur ([Fe-S]) protein, Clostridium pasteurianum ferredoxin, was reported (1). Within a few years it was clear that a new class of proteins had been discovered containing clusters composed of Fe and S 2Ϫ atoms ligated to the protein by cysteine ligands (2). Since that time there has been an exponential growth in the discovery of new [Fe-S] proteins that has exploded recently due to gene sequenc-ing. Today, over 100 different types of [Fe-S] proteins have been isolated from all life forms, from the most primitive archaea and bacteria to the most advanced eucaryotes (3)(4)(5)(6)(7)(8)(9). These structurally diverse proteins are involved in critical electron transfer reactions, chemical catalysis of hydration/dehydration reactions (10 -11), regulation of gene expression (12)(13)(14)(15)(16)(17)(18), oxygen and iron sensing (12)(13)(14)(15)(16)19), or the generation and stabilization of radical intermediates (20 -23).
The multiplicity of [Fe-S] cluster structures is remarkable (3-9, 24 -25) with the simplest clusters containing 1, 2, 3, or 4 iron atoms. Much of what we know about these protein-bound clusters comes from studies of a class of [Fe-S] proteins collectively known as ferredoxins. Ferredoxins are loosely defined as small, soluble, generally acidic [Fe-S] proteins that function to transfer electrons from one protein to another (2)(3)(4)(5)(6)(7)(8)(9). In the past, it was believed that a single ferredoxin was a relatively nonspecific electron transfer agent that could participate in a number of different cellular processes (26), and it is still common to see the term "bacterial ferredoxin" or "plant ferredoxin" used in this generic sense. However, it is now clear that a single organism can contain numerous different types of ferredoxins, each of which is expected to have a specific function in the cell. Table I shows that the organism of interest in this study, Azotobacter vinelandii, appears to synthesize at least 12 different small, ferredoxin-like proteins. Only a few of these proteins have been purified and characterized to date (28 -31, 34 -35, 43-44) with others identified based on sequence motif and/or homology with proteins that have been purified from other organisms. One protein of particular relevance to this study was initially identified by Zheng et al. (32) who sequenced a gene cluster from A. vinelandii that encoded proteins that are likely to be involved in the assembly of [Fe-S] clusters. Within that gene cluster was a gene they designated fdx (Table I). That gene was expected to encode a [2Fe-2S] ferredoxin because it was homologous both in sequence and gene location to a [2Fe-2S] ferredoxin that had been purified previously from Escherichia coli (45).
For several years now we have been using a different ferredoxin, the extremely well characterized 7Fe FdI 1 from A. vinelandii (Table I), as a model to address certain basic questions in [Fe-S] protein biochemistry (46 -50). The general approach has been to construct, purify, and characterize numerous site-directed mutant variants of FdI. In the course of these studies we have observed that some, but not all variants of FdI, appear to copurify with a contaminating pink protein that is not seen during the purification of native FdI. Here we report the purification and characterization of that pink protein and its identification as the product of the fdx gene that is cotranscribed with genes encoding proteins that appear to be involved in the assembly of [Fe-S] clusters.
We have chosen to designate the pink protein as FdIV because FdI (28,(33)(34)(35)(36), a [2Fe-2S]-containing protein that is designated [Fe-S] II but is sometimes referred to as FdII (28, 30 -31), and FdIII (43) have all been characterized extensively from A. vinelandii. We suggest that the FdIV gene designation be changed from fdx to fdxD, in keeping with the FdI, fdxA nomenclature, reserving fdxC for the FdIII gene when it becomes available.

EXPERIMENTAL PROCEDURES
Strains-The A. vinelandii strains used in this study include AvOP that synthesizes native FdI at wild-type levels, LM100 which has a kanamycin resistance cartridge inserted in the fdxA gene and does not synthesize any FdI (33), and a number of strains that were designed to overexpress specific site-directed mutant variants of FdI using the previously described pKT230 system (51). The variants used here are C42D (52), C20S (49), C49S (52), ⌬T14/⌬D15 (50), and C39S (52). It should be noted that as indicated in the text, all of these proteins accumulate at much lower levels than native FdI such that the levels produced by the overproduction strains are approximately equivalent to, or less than, the levels of native FdI produced by wild-type A. vinelandii.
Protein Purification-All A. vinelandii cells were grown under N 2fixing conditions in a 200-liter New Brunswick fermentor (48). The protein purification was carried out at room temperature, and fractions were monitored by their absorbance at 405 nm. FdIV was purified from the strain overexpressing the C42D variant of FdI, because, as indicated below, that strain has the greatest accumulation of FdIV. The preparation was carried out anaerobically in the presence of 2 mM Na 2 S 2 O 4 except the final Mono-Q/FPLC column step. Cell-free extracts were prepared from 1 kg of cells, and the first DEAE-cellulose column was run as described previously for the purification of other FdI variants (51). From the first DEAE-cellulose column (51) the FdI fraction eluted at 70 -80% of the linear 0.1-0.5 M NaCl gradient as a very shallow, yet well resolved brown peak. The protein fraction was immediately diluted with 2 or 3 volumes of 0.1 M potassium phosphate buffer, pH 7.4, and loaded onto a 2.5 ϫ 5-cm DEAE-cellulose column that was preequilibrated with the same buffer. Following loading, the column was washed slowly at the elution rate of 1.5 ml/min for 5 h with 0.12 M KCl in the same buffer until a greenish-brown protein (FdIII) (43) eluted. The pink-colored fraction containing FdIV, which was visible on the column immediately following the greenish-brown protein, was then eluted with 0.3 M KCl. Ammonium sulfate was added slowly to 75% saturation with stirring. After a 30-min incubation the sample was centrifuged at 20,000 ϫ g for 20 min. The pellet was then collected and resuspended in 0.025 M Tris-HCl, pH 7.4, before loading onto a 1.5 ϫ 110-cm Sephadex G-75 (Amersham Pharmacia Biotech) column preequilibrated with 0.025 M Tris-HCl, pH 7.4, and 0.1 M NaCl. FdIV and FdI eluted as a single broad peak as observed by monitoring absorbance at 405 nm; however, FdIV actually moved slightly faster than FdI. For that reason the first third of the peak was collected and concentrated using a Centriplus-10 concentrator (Amicon) prior to further purification on a 1-ml Mono-Q/FPLC column (Amersham Pharmacia Biotech) with a flow rate of 1 ml/min and a linear gradient of 0.3-0.8 M NaCl in 0.05 M Tris-HCl, pH 8.0, over 20 ml. Two fractions were well resolved as follows: the red-colored FdIV fraction that eluted between 0.46 and 0.48 M NaCl, and a brown-colored FdI fraction that eluted between 0.54 and 0.58 M NaCl. The pink-colored FdIV fraction became red-colored when it was air-oxidized. Each fraction was collected, buffer-exchanged with 0.025 M Tris-HCl, pH 7.4, concentrated to 0.5 ml, and kept in the Ϫ20°C freezer until use. The yield for FdIV is ϳ0.5 mg per kg C42D cell.
Overproduction of FdIV in E. coli-Overproduction of FdIV in E. coli was accomplished by constructing a fdxD gene cartridge in vitro and cloning this cartridge into the pT 7 -7 plasmid (53) such that fdxD gene expression was controlled by the T 7 phage transcriptional and translational regulatory elements. The fdxD gene cartridge was constructed by using an isolated 1.0-kilobase pair DNA fragment that was generated by EcoRI restriction enzyme digestion of an isolated fragment of A. vinelandii strain DJ116 (⌬nifS) chromosomal DNA (32). This DNA fragment was cloned into pUC119, and the plasmid was designated pDB946. pDB946 was subsequently used as a template for PCR amplification of the fdxD coding sequence. PCR amplification of fdxD was performed essentially as recommended by the supplier of Amplitaq (Perkin-Elmer). Cycling temperatures were 95°C for 1 min, 60°C for 2 min, and 72°C for 3 min. The PCR primers used were 5Ј-CATGCATAT-GCCGCAGATCGTTTTTC-3Ј and 5Ј-CTACGGATCCCTCAGTGCTGT-TCC-3Ј. The amino acid coding region for FdIV is underlined in the first primer sequence shown above. Following fdxD amplification, the gene cartridge was digested with the restriction enzymes NdeI and BamHI and ligated to NdeI-and BamHI-digested pT 7 -7 DNA. Proper orientation of fdxD within the resulting hybrid plasmid (pDB1024) such that the T 7 gene-10 promoter directs fdxD transcription was confirmed by restriction enzyme digestion of isolated plasmid DNA. For isolation of FdIV overproduced in E. coli, plasmid pDB1024 was transformed into the host strain BL21(DE3), and the transformed cells were maintained on LB media supplemented with 100 g/ml ampicillin. Cells were grown in 500-ml batches in LB media at 30°C and were induced for FdIV production at approximately 160 Klett units (red filter) by the addition of lactose to 1% (w/v) final concentration. Following addition of lactose, cells were cultured for an additional 2 h, harvested by centrifugation, and frozen at Ϫ20°C until use.
The purification of recombinant FdIV was carried out aerobically. E. coli cells containing the overproduced FdIV were ruptured by sonication in 0.025 M Tris-HCl, pH 7.4. Extracts were then ultracentrifuged in a Beckman Ty35 rotor at 35,000 rpm for 20 min at 4°C. Cell-free lysate was then loaded onto a 1.  NaCl was run. The red protein eluted at approximately 0.5 M NaCl, and the peak fractions were loaded onto a 5 ϫ 44-cm Sephacryl S-300HR (Amersham Pharmacia Biotech) equilibrated and run in 0.025 M Tris-HCl, pH 7.4, and 0.1 M NaCl. Column eluents were monitored at A 405 using an Amersham Pharmacia Biotech UV-1 optical detector and control unit.

Protein Sequencing and Determination of Molar Extinction
Coefficient-The NH 2 -terminal protein sequencing was carried out at the Biotechnology Resource Facility at the University of California, Irvine, and also at the W. M. Keck Facility at Yale University. The protein samples used for protein sequencing at the University of California, Irvine, were obtained by excising the predominant protein band of an FdIV sample from a Coomassie Blue-stained SDS-12% PAGE and eluting the protein into 10 mM ammonium bicarbonate. The protein sample was then modified using 2-mercaptoethanol and 4-vinylpyridine so that the modified cysteines could be detected. The sample used for sequencing at Yale University was prepared by excising the protein band from a Coomassie Blue-stained polyvinylidene fluoride membrane (Millipore) onto which an FdIV sample resolved in an SDS-12% PAGE was electroblotted. Twenty seven residues were solved at the University of California, Irvine (PQIVFLPHEVHCPQGRVVEAATGTSIL, note that the underlined residues were identified with ambiguity). Twenty residues were identified at Yale University, PQIVFLPHEVHCPQGRV-VEA. The molecular weight of FdIV was determined at the W. M. Keck Facility at Yale University using quadruple-time-of-flight mass spectrometry using an electrospray interface. The electrospray ionization technique was preferentially chosen with the hope of detecting signals corresponding to both holoprotein with the [Fe-S] cluster bound and the apoprotein. The mass spectrometry using MALDI ionization technique was performed at the Biotechnology Resource Facility at the University of California, Irvine. Quantitative triplicate amino acid analyses of Mono-Q/FPLC purified FdIV was carried out at the W. M. Keck Facility at Yale University. The protein content for determination of the molar extinction coefficient was estimated from the nanomoles of Ala, Leu, and Phe using the composition and M r 12,375 (for apoprotein) calculated from the gene sequence and was corrected for holoprotein. Here the cluster composition was determined.
Protein Characterization-The antibody for FdIV was produced in rabbits in Bethyl Laboratories, Inc. (Montgomery, TX), using routine immunization protocols. The final antiserum was obtained 1 week after the sixth antigen was injected into the rabbit. To determine iron content, samples were digested, and the analysis was carried out as described previously using FeCl 3 ⅐6H 2 O to generate a standard curve with FdI and FdIII as controls (54). Redox titration of FdIV was performed spectrophotometrically at room temperature using 5Ј-deazariboflavin/ potassium oxalate as a photoreductant system and benzyl viologen as a chemical mediator (E 0 Ј ϭ Ϫ320 mV versus standard hydrogen electrode) as described (55). The sample was anaerobically prepared in a glove box (oxygen level Ͻ 1 ppm) and was sealed with a rubber stopper and an aluminum cap. A 1-cm light path cuvette, in a 1-ml volume, contained 5 M 5Ј-deazariboflavin, 15 mM potassium oxalate, 45 M benzyl viologen, and 40 M FdIV in 0.05 M Tris-HCl, pH 8.0. The mixture was exposed to the light of a slide projector (5-300 s), and absorbance changes were monitored after 30 s equilibration in darkness. A water-containing vessel was installed between the sample and light source to avoid overheating the sample during the illumination. The percentage of [2Fe-2S] cluster reduction was determined from the absorbance change at 440 nm, which is an isosbestic point between the oxidized and reduced forms of benzyl viologen. The total reduction of [2Fe-2S] was estimated from the absorbance change after 2 mM Na 2 S 2 O 4 was added to sample. The extent of benzyl viologen reduction was calculated from the absorption at 600 nm after correction for the absorbance of Fd at this wavelength.
EPR spectra were obtained using a Bruker ESP300E spectrometer, interfaced with an Oxford liquid helium cryostat. The EPR measurement for spin quantitation was performed either at 30k for 100 M purified FdIV or 25k for 100 M recombinant FdIV at microwave power where the EPR signal of FdIV is linearly proportional to (power) 1/2 . Cu 2ϩ -EDTA (100 M), which was used as a standard, was recorded at the same EPR setting and temperature, where the signal is also linearly increasing as a function of (power) 1/2 . The concentration of spin was calculated by double integration of signal over the entire sweep width (56). For anaerobic EPR measurement, the samples were loaded into the EPR tubes in an O 2 -free (O 2 Ͻ 1 ppm) glove box (Vacuum Atmosphere, Hawthorne, CA). The tubes were sealed with a cap, wrapped with parafilm, and then transferred to a liquid nitrogen Dewar.
Absorption spectra were recorded in 0.5-ml quartz cuvettes on a Hewlett-Packard 8452A diode array spectrophotometer. CD spectra were obtained using a Jasco J-500C spectropolarimeter. CD measurements were carried out using small volume cylindrical cells with fused quartz windows.
Attempted Insertional Inactivation of fdxD Gene in A. vinelandii-Kanamycin resistance cartridges were inserted into the fdxD gene in order to disrupt its transcription. Plasmid pDB946 was digested with EcoRV, and kanamycin cartridges derived from pUC-KAPA were inserted. Kanamycin cartridges were inserted so that they are transcribed in the same (pDB1015) or opposite orientation (pDB1016) as the fdxD gene.
The Level of Both FdIV and FdI in FdI Variants-EPR and Western blot methods were performed to quantify the amount of FdIV and FdI present in the cell. Because of the low abundance of protein in whole cells, we prepared concentrated protein samples in the following way. The same amount (350 g) of cell paste from site-directed variants (C42D, C20S, and ⌬T14/⌬D15), LM100 (FdI deletion), and AvOP (containing wild-type FdI) was thawed, broken, and centrifuged at 8,000 rpm for 2 h. The supernatant was loaded onto a 2.5 ϫ 15-cm anion exchange (DE52, Whatman) column equilibrated with 0.05 M Tris-HCl, pH 8.0. The column was washed with 2 volumes of 0.025 M Tris-HCl, pH 7.4, followed by 300 ml linear gradient from 0.1 to 0.5 M NaCl. FdIV and FdI eluted together as a single peak as monitored by A 405 . The peak was collected and concentrated to ϳ5 ml in an Amicon cell fitted with a YM-3 membrane.

RESULTS AND DISCUSSION
How FdIV Came to Our Attention-Over the past several years we have purified, or attempted to purify, a large number of site-directed mutant variants of the 7Fe FdI from A. vinelandii (47)(48)(49)(50)52). Many of these proteins presented unique purification challenges, but in all cases our initial attempts began with a procedure that was developed for native FdI involving a first column separation on DEAE-cellulose with a linear NaCl gradient. In all cases the columns were monitored by measuring the absorbance at 405 nm, and in all cases an FdI peak was observed in the expected position, relative to internal marker proteins, the MoFe and Fe proteins of nitrogenase (51,54). The size of the absorbance peak varied depending upon how much of a particular FdI variant accumulated in vivo, but the position of the peak was invariant. It was upon going to the second step, using another DEAE-cellulose column, or the third step, using a size exclusion column (Sephadex G-50), that significant differences in the color of the FdI fractions were observed. For some, but certainly not all variants, the protein fractions appeared pink, instead of the normal brown color for native FdI. Often in these cases it was never possible to purify successfully the FdI variant, or the FdI variant was ultimately purified only with great difficulty and in very low yield. Table  II summarizes the qualitative observations obtained by us over the past several years.
While purifying the FdI [4Fe-4S] cluster ligand variant, C42D, we observed that even the first column FdI fraction appeared to be pink and that a large amount of pink material eluted from subsequent columns in the same fractions as the C42D FdI which we were identifying by cross-reaction to FdI antibodies. To determine whether the pink color arose from an unusual form of FdI or from a contaminating protein, a variety of purification methods was employed in attempts to separate the pink color from C42D FdI. As described under "Experimental Procedures," FdIV was ultimately separated from partially purified C42D FdI on a Mono-Q/FPLC column to give a fraction that ran as a single band on SDS-PAGE that did not react with FdI antibody. A routine purification procedure was then developed to yield about 0.5 mg of FdIV per kg of C42D cell paste.
As shown in Fig. 1, purified FdIV runs on an SDS-12% polyacrylamide gel at a position corresponding to ϳ25,000, slightly slower than native FdI. It has previously been observed that FdI migrates abnormally on SDS-12% PAGE as if its molecular weight were ϳ24,000 rather than 12,000 (57). Such anomalous behavior on SDS-PAGE has previously been re-ported for other small and highly charged proteins (the isoelectric point of FdI calculated from amino acid sequence is 4.03) (58). The similar purification properties and electrophoretic behaviors of FdIV and FdI suggested that they were likely to have similar charges and molecular weights. As expected, FdIV is a monomer that it runs on Superdex 75/FPLC with almost the same retention time as FdI, which is monomer. It is also a highly acidic protein. The isoelectric point calculated from the amino acid sequence as shown in Fig. 2 is 4.29. Once FdIV was purified, antibodies were raised against it, and the antiserum does not cross-react with FdI (data not shown).
Identification of FdIV-For identification, the gel-purified protein was subjected to NH 2 -terminal protein sequencing with cysteine modification. Twenty residues were unambiguously identified, PQIVFLPHEVHCPEGRVVEA. Sequence homology searches in the data base Protein Data Base ϩ SwissProt ϩ PIR ϩ GenBank TM translations were performed. As shown in Fig. 2, the results revealed the presence of a sequence with a perfect match in A. vinelandii (32), and sequences with high identity in Pseudomonas aeruginosa (59), E. coli (45,60), Hemophilus influenzae (61), and Buchnera aphidicola (62). Searches in the Unfinished Microbial Genome data base have found similar sequences in Vibrio cholerae, Shewanella putrefaciens, Salmonella typhi, Yersinia pestis, Pasteurella multocida, Bordetella pertussis, Neisseria meningitidis, Neisseria gonorrhoeae, and Actinobacillus actinomycetemcomitans. Sequence comparisons indicate that FdIV is the gene product of the fdx gene in the A. vinelandii iscSUA operon and consists of 111 residues with 12,435 Da (deduction from the amino acid sequence) with the first 2 Mets missing in the purified protein (Fig. 2). The molecular mass determined by electrospray mass spectrometry was 12,434.03 Ϯ 0.21 Da, which can be attributed to apoprotein (average mass 12,435 Da) (data not shown). We further confirmed this by observing the same molecular mass for the fdx gene product with MALDI ionization techniques (data not shown). The MALDI technique requires use of an organic chemical matrix (saturated sinapinic acid was used in this study) dissolved in 30% acetonitrile and 0.1% trifluoroacetic acid (63). That procedure is expected to destroy the ironsulfur cluster. Therefore the molecular mass obtained from the MALDI techniques should correspond to the apoform, which it does. Unlike what has been reported for other [Fe-S] proteins (64 -66), in this electrospray mass spectrometry study we could not detect the second peak corresponding to holoprotein at 178 Da above that of the apoprotein, 12,435 Da. Fig. 2 only the protein from E. coli has been purified and characterized and shown to contain a single [2Fe-2S] 2ϩ/ϩ cluster (45). In that case the four cysteine ligands to the cluster were suggested to be those shown in Fig. 2 based on comparison with human (67) and Pseudomonas putida (68) ferredoxins whose [2Fe-2S] 2ϩ/ϩ clusters are known to be ligated by a Cys-X 5 -Cys-X 2 -Cys-X 35 -Cys motif. This pattern is highly conserved among all 14 sequences shown in Fig. 2. In addition, the sequences around the four cysteine residues in the motif are also highly similar. If these four cysteines are indeed the ligands for [2Fe-2S] 2ϩ/ϩ clusters, this pattern represents a subclass of ferredoxin-type [2Fe-2S] 2ϩ/ϩ cluster binding motifs, which is shared by "vertebrate or bacterial"-type ferredoxins (45, 69 -70). This is in contrast to the Cys-X 4 -Cys-X 2 -Cys-X 29 or 30 -Cys motif found in "plant"-type Fds, which participate in photosynthetic electron transfer processes in plants and in blue-green algae (71)(72). The pink color of Fd IV, as well as its sequence, is consistent with the protein containing a [2Fe-2S] 2ϩ/ϩ cluster as a prosthetic group. The nature of this center was therefore investigated and identified by spectroscopic analyses and biochemical methods.

FdIV Contains a [2Fe-2S] 2ϩ/ϩ Cluster-Of the proteins shown in
Iron content analysis for FdIV gives 2.0 Ϯ 0.1 iron atoms per molecule when the ⑀ 458 ϭ 7,071 M Ϫ1 cm Ϫ1 , determined from triplicate amino acid analysis for protein quantitation. In the same assay, the 7Fe containing FdI and 8Fe containing FdIII gave 7.0 Ϯ 0.1 and 7.7 Ϯ 0.1 iron atoms per molecules, respectively. These results indicate that FdIV contains either a [2Fe-2S] cluster or two rubredoxin-like 1Fe centers. The possibility of having two rubredoxin-like iron centers was ruled out because FdIV contains only 6 cysteine residues per molecule, which cannot accommodate formation of two rubredoxin-like 1Fe centers. This is further supported by the observation that we could not detect any EPR signal with g values of 4.3 and 9.5, which arise from the 3/2 and 1/2 kramers' doublets, respectively, of an S ϭ 5/2 multiplet of rubredoxin in the dithionitereduced state (data not shown) (73)(74).
The UV/visible spectrum of oxidized FdIV, shown in Fig. 3, is distinct from those obtained for 7Fe FdI and 8Fe ferredoxins (such as FdIII) (43) and is indistinguishable from that of bacterial type [2Fe-2S] ferredoxins, exhibiting peaks at 414 and 458 nm and a broad maximum with a shoulder near 322 and 337 nm (45). The extinction coefficient of 7,071 M Ϫ1 cm Ϫ1 at 458  nm is a little lower than those obtained for other [2Fe-2S] Fds that are reported to be between 8,000 and 10,000 M Ϫ1 cm Ϫ1 (75)(76). Upon reduction of FdIV with 2 mM dithionite, the absorbance in the visible region is greatly reduced, and characteristically, a new peak at 545 nm which makes FdIV turn "pink" in color appears (55,77). Such a spectral feature has previously been observed for the closely related E. coli [2Fe-2S] ϩ Fd and the bacterial type [2Fe-2S] ϩ FdV from Rhodobacter capsulatus, but not for plant-type [2Fe-2S] ϩ ferredoxins (77). The spectrum of oxidized FdIV is recovered when reduced FdIV is exposed to air (data not shown), indicating that the protein is stable in the presence of O 2 .
The CD spectra of oxidized and dithionite-reduced FdIV, shown in Fig. 4, are very similar to those of other [2Fe-2S] 2ϩ/ϩ ferredoxins (45,77). The [2Fe-2S] cluster exhibits intense visible CD compared with other biological [Fe-S] clusters, and CD spectra are more sensitive than the corresponding absorption spectra to the protein folding in the vicinity of the cluster (78 -79). For the oxidized protein the negative peaks at 550 and 500 nm, and large positive peaks at 435 nm, with a shoulder, are characteristic features of [2Fe-2S] 2ϩ clusters. This [2Fe-2S] 2ϩ CD spectrum is quite different from those exhibited by 7Fe or 8Fe ferredoxins (43,57). The presence of a shoulder in the main peak and the separate bands in the range from 250 to 350 nm are also used to differentiate between plant-type [2Fe-2S] 2ϩ ferredoxins and vertebrate or bacterial-type [2Fe-2S] 2ϩ ferredoxins. Again the environment around the [2Fe-2S] 2ϩ cluster in FdIV, as observed by CD, is quite similar to vertebrate or bacterial-type ferredoxins.
EPR analysis of purified FdIV in the oxidized form shows no signal (Fig. 5A). Upon reduction by dithionite, the spectrum exhibits a nearly axial signal with apparent g values of 2.03 and 1.95. This signal may be attributed to a [2Fe-2S] ϩ cluster with an S ϭ 1/2 spin state (Fig. 5B), which comes from the anti-ferromagnetic coupling of ferric S ϭ 5/2 and ferrous S ϭ 2 spin states. Spin integration of the EPR signal of spectrum B in signal intensity was still detectable at temperatures as high as 70k (data not shown). The power saturation dependence is further consistent with a [2Fe-2S] ϩ cluster (data not shown). Taken together, the power and temperature dependence rule out the possibility that the EPR signal arises from a [4Fe-4S] ϩ cluster.
The oxidation-reduction potential of the FdIV [2Fe-2S] 2ϩ/ϩ cluster was measured by anaerobic titration using benzyl viologen as a potential indicator and 5Ј-deazariboflavin/potassium oxalate as a photoreductant. The absorbance change was measured at 440 nm to eliminate the contribution of benzyl viologen to the signal as described under "Experimental Procedures." By using the Nernst equation (Equation 1) and plotting the E h versus the log([FdIV ox ]/[FdIV red ]), the midpoint redox potential was calculated to be approximately Ϫ344 Ϯ 10 mV versus standard hydrogen electrode at pH 7.8 (best fitted with a slope of 52 mV in case of n ϭ 1) (Fig. 6). The potential is similar to the related E. coli [2Fe-2S] 2ϩ/ϩ Fd, which was known to be approximately Ϫ380 mV (77) and is within the range of other [2Fe-2S] 2ϩ/ϩ ferredoxins.

Comparison of FdIV Purified from A. vinelandii and Recombinant
FdIV from E. coli-In order to obtain larger quantities of FdIV for crystallization and affinity chromatography experiments (80), the fdxD gene was used to construct a hybrid plasmid where the expression of the fdxD was placed under the control of the T 7 transcriptional and translational control elements. As described under "Experimental Procedures," FdIV was successfully expressed in E. coli in soluble form using this system, and the protein was easily purified with yields of ϳ0.8 mg per g of cells. The air-oxidized, isolated gene product appeared to be red-colored, indicating that the [2Fe-2S] 2ϩ/ϩ cluster was successfully assembled into the polypeptide in E. coli. The successful overproduction of several other ferredoxins in their native forms in E. coli had previously been reported including the [2Fe-2S] 2ϩ/ϩ adrenodoxin (81), a [2Fe-2S] 2ϩ/ϩ ferredoxin from heterocysts of Anabaena strain 7120 (82), the 2 [4Fe-4S] 2ϩ/ϩ ferredoxin from C. pasteurianum (83)(84), and a [3Fe-4S] [4Fe-4S] ferredoxin from R. capsulatus (85). To see whether the FdIV [2Fe-2S] 2ϩ/ϩ cluster was correctly assembled into the polypeptide, we characterized the recombinant FdIV using the same spectroscopic and biochemical techniques we had used for the native protein. The UV/visible spectra, CD spectra, EPR signal, and the temperature and magnetic dependence of that signal of recombinant FdIV were indistin-guishable with those of purified FdIV (data not shown).
Given the identity of the spectra obtained for the recombinant and native proteins, we were surprised to observe that they did not migrate in the same place on SDS-PAGE. Fig. 7 shows that the recombinant protein migrates faster than the native protein on SDS-12% PAGE indicating either that it is smaller or that it has less negative charge or both. NH 2 -terminal sequencing of recombinant FdIV showed that the sequence starts with Pro as the first amino acid, as does the purified native FdIV (Fig. 2) so the difference in migration behavior cannot be attributed to a NH 2 -terminal deletion. The discrepancy between the two proteins was confirmed by mass spectrometry (data not shown). As indicated above, the mass of 12,434.03 Ϯ 0.21 Da obtained for the native apoprotein was as expected based on the gene sequence. In contrast, a mass of 12,486.01 Ϯ 0.89 Da, 52 Da mass units larger, was obtained for the recombinant apoprotein. Thus the recombinant protein appears to be modified in some way that makes it larger by 52 Da mass units, while making it run faster on SDS-PAGE suggesting that the modification may involve the removal of negative charge or the addition of positive charge. Another possibility is the formation of an intra-molecular cross-link, which alters its mobility on the SDS-PAGE and its molecular mass by 52. At present we are unable to identify the specific nature of the 52-Da modification; however, we do know that the change does not affect either the assembly of the [2Fe-2S] 2ϩ/ϩ cluster or the conformation of the protein near the cluster because the spectral properties are not changed.
Attempted Deletion of the fdxD Gene-A. vinelandii has a high frequency reciprocal recombination system that has been used to disrupt a variety of nonessential genes (e.g. fdxA and nifF). Disruption of the fdxD gene was desired not only as a probe of the function of the protein but also as a means of eliminating a tenacious contaminant encountered during the purification of some FdI variants (Table II). Wild-type A. vinelandii was transformed with plasmids that would result, upon recombination, in strains that would have the fdxD gene disrupted by the insertion of a kanamycin resistance cartridge (see "Experimental Procedures"). However, after initial selection for the kanamycin resistance phenotype, the selected colonies were observed to revert back to kanamycin sensitivity. This behavior indicates that the fdxD gene is required for proper cellular function of A. vinelandii.
Relative Amounts of FdI and FdIV in Strains Expressing FdI Variants-The purification and characterization of FdIV described above provides two new tools to examine the relative amounts of FdIV and FdI in various A. vinelandii strains. First, antibodies raised separately against the two proteins can be used because each antibody only cross-reacts with the protein it was raised against. Based on the qualitative information shown in Table II, we selected three strains expressing FdI variants to compare with the wild-type strain AvOP. The strains expressing C20S FdI, C42D FdI, and ⌬T14/⌬D15 were chosen because the levels of FdI accumulated by these strains appeared to cover wide ranges of FdI accumulation, roughly similar expression (⌬T14/⌬D15), a little lower FdI expression (C20S FdI), and significantly lower FdI expression (C42D FdI) when compared with the levels of FdI accumulated by the wild-type strain, an observation that is confirmed by the comparison shown in Fig. 8A. Therefore, it might be possible to probe the relationship between FdI and FdIV by looking at FdIV expression level in these strains. In addition, all three of these proteins were stable enough to be purified (50,52). Initially as a control, we also compared the FdIV levels in LM100, an A. vinelandii strain that does not synthesize FdI. Fig. 8B compares the levels of FdIV in the FdI fractions collected from the first column for the A. vinelandii strains we tested. The first surprise in this comparison is that the levels of FdIV seem to be the greatest in LM100, the strain that does not synthesize any FdI. Another surprise is that significant FdIV was observed using antibodies for wild-type and ⌬T14/⌬D15 even though none of the protein was observed during the purification of FdI from those stains. In contrast, for C20S, which has FdI (Fig. 8A) and FdIV levels comparable to wild-type (Fig.  8B), FdIV was observed during the purification of FdI. At present we are unable to explain why FdIV presents itself as a serious problem during the purification of some FdI variants and not for others, but clearly there is more to the problem than simply the relative levels of the two proteins. The one observation shown in Fig. 8B that was predicted in advance is that C42D does have more FdIV than the other strains.
One disadvantage of the Western blot method of comparison is that it only provides semiquantitative information about the total protein present and cannot distinguish apoprotein and/or cluster-converted protein from holoprotein. Quantitative information about the relative amounts of FdI and FdIV holoproteins in the various strains can be obtained by EPR. This is possible because the [3Fe-4S] ϩ cluster of FdI exhibits a characteristic g ϭ 2.01 EPR signal in the oxidized state (Fig. 9A), conditions where FdIV is diamagnetic. Upon reduction with dithionite the [3Fe-4S] 0 cluster becomes EPR silent in the perpendicular mode, whereas the [2Fe-2S] ϩ cluster from FdIV exhibits a characteristic g ϭ 1.95 EPR signal (Fig. 9B). Fig. 9A compares the EPR spectra exhibited by the first column FdI fractions for the strains of interest. Clearly the amount of [3Fe-4S] ϩ containing FdI is highest in wild-type followed by C20S with much lower levels observed for both C42D and ⌬T14/⌬D15. In the latter case, however, it should be noted that the low level of [3Fe-4S] ϩ cluster is not expected to correlate with the levels of cross-reactive material present (Fig.   8A) because that mutation resulted in converting the FdI [3Fe-4S] cluster to a [4Fe-4S] cluster which does not exhibit an EPR signal in the oxidized state (50). 2 As expected, no FdI was detected in LM100. The good correlation between FdI levels observed with antibodies (Fig. 8A) and FdI levels observed by EPR (Fig. 9B) for C42D and C20S FdI indicate that the fractions contain mainly holoprotein. This is also consistent with the final yields of holoprotein for these two mutants once these fractions were purified to homogeneity.
The levels of FdIV observed by cross-reactivity with antibodies and by EPR are also similar. Fig. 9B shows that the highest levels of FdIV are observed in LM100 but significant levels are also observed in the wild type. Also consistent with the information in Fig. 8B C42D has higher levels of FdIV than does the wild type, and the levels of FdIV in C20S and ⌬T14/⌬D15 are lower.
Physiological EPR conditions for FdI are as follows: microwave frequency, 9.49 GHz; modulation amplitude, 0.64 mT; microwave power, 1 milliwatt; and temperature, 10 K; receiver gain, 5 ϫ 10 3 . For measurement of FdIV [2Fe-2S] ϩ , the sample was first reduced in 2 mM sodium dithionite, and EPR was carried out under the same condition as in FdI except change in temperature and receiver gain into 35 K and 2 ϫ 10 4 , respectively. A minor signal observed near g z of [2Fe-2S] ϩ in all FdI mutants is unidentified but, considering that the same signal is also observed in FdI-less LM100 strain, it does not represent [3Fe-4S] ϩ EPR signal of FdI. protein as the product of a previously sequenced gene which we now designate fdxD. This gene is located immediately downstream of the iscSUA-hscBA gene cluster and appears to be cotranscribed with this cluster (32). The iscSUA gene products are likely to catalyze critical reactions in the assembly and/or repair of [Fe-S] clusters. The hscBA gene products may serve as the molecular chaperone proteins that assist in the maturation of these [Fe-S] proteins (32). In this context it therefore seems likely that FdIV is somehow involved in the processes of [Fe-S] cluster formation and/or repair possibly by carrying out an essential electron transfer step. Our inability to construct a knock-out mutant further suggests that FdIV is essential for the survival of the cell.
FdIV originally came to our attention because it appeared to associate specifically with certain FdI variants that had altered [Fe-S] ligand composition, an observation that would be consistent with its role in [Fe-S] cluster assembly or repair. On the other hand we have now established that the protein has a very similar charge and molecular weight to FdI such that its apparent "association" with the FdI ligand variants could be attributed to the coincidental copurification of the two proteins. Indeed unlike the situation with FdI and its redox partner NADPH:ferredoxin reductase (57), all attempts to obtain crosslinked complexes of FdI and FdIV have been unsuccessful. Although these observations do not rule out direct contact between the two proteins, when combined with the observation that FdIV levels are dramatically increased in the FdI minus strain LM100, they do provide an alternative explanation, namely that FdIV levels are specifically increased in response to deletion of FdI (or some function of FdI disrupted in specific mutant strains). In this context it is important to note that in A. vinelandii another protein, ferredoxin reductase, has been shown to be overexpressed to the same levels via the same specific DNA sequence, in response to either deletion of FdI (33) or to addition of the superoxide propagator paraquat to cells (86). This oxidative stress response system is described elsewhere (86). 3 Future experiments will be directed toward examining the possibility that like the fpr gene, the iscSUA-hscBA-fdxD gene cluster, encoding proteins involved in [Fe-S] cluster assembly or repair, might be similarly up-regulated in response to oxidative stress or deletion of FdI.