ATP-driven Reduction by Dark-operative Protochlorophyllide Oxidoreductase from Chlorobium tepidum Mechanistically Resembles Nitrogenase Catalysis*

During chlorophyll and bacteriochlorophyll biosynthesis in gymnosperms, algae, and photosynthetic bacteria, dark-operative protochlorophyllide oxidoreductase (DPOR) reduces ring D of aromatic protochlorophyllide stereospecifically to produce chlorophyllide. We describe the heterologous overproduction of DPOR subunits BchN, BchB, and BchL from Chlorobium tepidum in Escherichia coli allowing their purification to apparent homogeneity. The catalytic activity was found to be 3.15 nmol min-1 mg-1 with Km values of 6.1 μm for protochlorophyllide, 13.5 μm for ATP, and 52.7 μm for the reductant dithionite. To identify residues important in DPOR function, 21 enzyme variants were generated by site-directed mutagenesis and investigated for their metal content, spectroscopic features, and catalytic activity. Two cysteine residues (Cys97 and Cys131) of homodimeric BchL2 are found to coordinate an intersubunit [4Fe-4S] cluster, essential for low potential electron transfer to (BchNB)2 as part of the reduction of the protochlorophyllide substrate. Similarly, Lys10 and Leu126 are crucial to ATP-driven electron transfer from BchL2. The activation energy of DPOR electron transfer is 22.2 kJ mol-1 indicating a requirement for 4 ATP per catalytic cycle. At the amino acid level, BchL is 33% identical to the nitrogenase subunit NifH allowing a first tentative structural model to be proposed. In (BchNB)2, we find that four cysteine residues, three from BchN (Cys21, Cys46, and Cys103) and one from BchB (Cys94), coordinate a second inter-subunit [4Fe-4S] cluster required for catalysis. No evidence for any type of molybdenum-containing cofactor was found, indicating that the DPOR subunit BchN clearly differs from the homologous nitrogenase subunit NifD. Based on the available data we propose an enzymatic mechanism of DPOR.

During chlorophyll and bacteriochlorophyll biosynthesis in gymnosperms, algae, and photosynthetic bacteria, dark-operative protochlorophyllide oxidoreductase (DPOR) reduces ring D of aromatic protochlorophyllide stereospecifically to produce chlorophyllide. We describe the heterologous overproduction of DPOR subunits BchN, BchB, and BchL from Chlorobium tepidum in Escherichia coli allowing their purification to apparent homogeneity. The catalytic activity was found to be 3.15 nmol min ؊1 mg ؊1 with K m values of 6.1 M for protochlorophyllide, 13.5 M for ATP, and 52.7 M for the reductant dithionite. To identify residues important in DPOR function, 21 enzyme variants were generated by site-directed mutagenesis and investigated for their metal content, spectroscopic features, and catalytic activity. Two cysteine residues (Cys 97 and Cys 131 ) of homodimeric BchL 2 are found to coordinate an intersubunit [4Fe-4S] cluster, essential for low potential electron transfer to (BchNB) 2 as part of the reduction of the protochlorophyllide substrate. Similarly, Lys 10 and Leu 126 are crucial to ATP-driven electron transfer from BchL 2 . The activation energy of DPOR electron transfer is 22.2 kJ mol ؊1 indicating a requirement for 4 ATP per catalytic cycle. At the amino acid level, BchL is 33% identical to the nitrogenase subunit NifH allowing a first tentative structural model to be proposed. In (BchNB) 2 , we find that four cysteine residues, three from BchN (Cys 21 , Cys 46 , and Cys 103 ) and one from BchB (Cys 94 ), coordinate a second intersubunit [4Fe-4S] cluster required for catalysis. No evidence for any type of molybdenum-containing cofactor was found, indicating that the DPOR subunit BchN clearly differs from the homologous nitrogenase subunit NifD. Based on the available data we propose an enzymatic mechanism of DPOR.
Protochlorophyllide (Pchlide) 2 is a central metabolite for the biosynthesis of chlorophylls (Chl) and bacteriochlorophylls (bChl). In photosynthetic organisms two distinct enzymes catalyze the stereospecific reduction of ring D of the aromatic Pchlide to form chlorophyllide (Chlide) (1-3) (Fig. 1). The first enzyme is the light-dependent Pchlide oxidoreductase (LPOR; NADPH Pchlide oxidoreductase, EC 1.3.1.33). To achieve the chemically difficult reduction of the aromatic ring system, the complex of monomeric LPOR and substrate Pchlide must absorb light energy prior to NADPH-dependent reduction (4 -6) making LPOR a dominant player in the light-dependent greening of flowering plants (angiosperms) (7,8). The second Pchlide-reducing enzyme is the light-independent (dark) Pchlide oxidoreductase (DPOR), which shares no amino acid sequence homology to the LPOR system. It consumes ATP to drive the production of Chlide. Some anoxygenic bacteria only encode DPOR (9), whereas angiosperms exclusively produce LPOR. Cyanobacteria, algae, and gymnosperms possess both LPOR and DPOR (1). DPOR enables these organisms to synthesize bChl or Chl in the dark.
In bChl-synthesizing organisms the genes bchN, bchB, and bchL encode the three subunits of DPOR. In Chl-synthesizing organisms the corresponding genes are known as chlN, chlB, and chlL (10 -12). Protein subunits of DPOR share significant amino acid sequence similarity with NifD, NifK, and NifH, the three subunits of nitrogenase. Among the subunits of DPOR, BchL is most similar to its counterpart NifH sharing 33% amino acid sequence identity (13). The remaining two DPOR subunits BchN and BchB are similarly related to NifD and NifK, although sequence identities at ϳ15% (for both comparisons) are not as pronounced (14).
BchN and BchB together contain only four highly conserved cysteines, compared with six cysteines coordinating the [8Fe-7S] P-cluster in NifD/NifK. Three are found in BchN (Cys 21 , Cys 46 , Cys 103 , C. tepidum numbering) and one in BchB (Cys 94 ). A bridging Fe-S center, although presumably smaller than the P-cluster, may thus be located at the interface of BchN and BchB. Residues involved in MoFe cofactor binding in NifD are not conserved in BchN implying that this part of dinitrogen reduction does not have a counterpart in reduction of Pchlide.
Several protein complexes of dinitrogenase reductase and the MoFe-protein have been trapped using AlF 4 Ϫ or BeF 3 Ϫ together with Mg-ADP, or by deleting Leu 127 in the ATPbinding motif of NifH (19). The [4Fe-4S] cluster of NifH 2 has, moreover, been shown to be a better reductant when in complex with the MoFe-protein, than in its uncomplexed form (20). These findings indicate that the BchL subunits of DPOR may similarly catalyze an ATP-driven transfer of elec-trons from a bound [Fe-S] cluster to catalytic protein complex BchNB containing a second Fe-S redox center. Interestingly ATP-dependent electron transfer is not restricted to nitrogenase. Instead, Fe-dinitrogenase reductase-like subunits also occur in enzymes such as 2-hydroxyacyl-CoA dehydratase, referred to as "archerases" due to their ability to "activate single electrons" by an ATP-induced change in the redox potential of a [4Fe-4S] center (21). DPOR from Rhodobacter capsulatus, studied using a homologous expression system, consists of BchL and a BchNB complex. Its enzymatic activity requires both ATP and a reducing agent such as dithionite (22,23), whereas ferredoxin is its physiological electron donor (22). Because R. capsulatus DPOR activity declines rapidly after affinity purification, crude cellular extracts were used in these experiments (22).
We have established the first heterologous milligram range production and purification system for Chlorobium tepidum DPOR in Escherichia coli, allowing the kinetic, spectroscopic, and structural characterization of recombinant wild-type and mutant BchNB and BchL including two [4Fe-4S] clusters and their coordinating cysteine residues.

Production and Purification of C. tepidum DPOR in E. coli-
The complete bchNBL-operon from C. tepidum (strain ATCC 49652) was PCR-amplified using primers CAGGATCCATGA-TGCCGGTTTCAAG and CAATGCGGCCGCTTACTGCC-AGCCACC and cloned into BamHI and NotI sites of pGEX-6P-1 to yield pGEX-bchNBL. The bchL gene was analogously amplified using primers CAGGATCCATGAGTTTAGTATT-GGCC and CAATGCGGCCGCTTACTGCCAGCCACC and cloned into BamHI and NotI sites of pGEX-6P-1 to generate pGEX-bchL. The plasmids were transformed into E. coli BL21(DE3) Codon Plus RIL to produce the BchNBL complex or only BchL. Cells were cultivated at 17°C in 500 ml of LB medium with 1 mM Fe(III)-citrate. Protein production was induced by 50 M isopropyl-␤-thiogalactoside at 0.6 OD 578 . After 16 h of cultivation 1.7 mM dithionite was added and cultivation was continued without agitation for 2 h at 18°C in an anaerobic chamber (Coy Laboratories, Grass Lake, MI) to allow [Fe-S] cluster formation. All remaining steps were performed under anaerobic conditions at oxygen partial pressure below 1 ppm (Oxygen detector, Coy Laboratories, Grass Lake, MI). Solutions were N 2 -saturated prior use. Cells were centrifuged, resuspended in 15 ml of lysis buffer (100 mM HEPES-NaOH (pH 7.5), 10 mM MgCl 2 , 1 M glycerol, 10 mM dithiothreitol), and disrupted by a single passage through a French press at 1500 p.s.i. into an anaerobic bottle. Following 60 min of centrifugation at 175,000 ϫ g at 4°C, the supernatant was added to 500 l of glutathione-Sepharose (GE Healthcare) equilibrated with lysis buffer. After washing with 20 ml of phosphate-buffered saline containing 10 mM dithiothreitol (wash buffer) the recombinant fusion protein GST-BchN in complex with BchB or GST-BchL alone were eluted using 2 ml of 50 mM Tris-HCl (pH 8.0) containing 10 mM reduced glutathione. Fractions containing the BchNB complex or GST-BchL were identified by SDS-PAGE.  Purification of the Ternary BchNBL Complex-To obtain a stable ternary DPOR complex, all steps of the affinity purifications were carried out in the presence of 2 or 10 mM BeF 3 Ϫ , or 100 or 200 M AlF 4 Ϫ and 5 mM Mg-ADP. Cell-free extracts were incubated with metallofluorides and Mg-ADP 30 min prior to affinity chromatography.
Point substitutions in BchN were introduced using the oligonucleotides: Protein Concentrations-Concentrations of GST fusion proteins in crude cellular extracts were determined using the GSTdetection module (GE Healthcare). The concentration of purified proteins was determined employing the BCA (bicinchoninic acid) protein assay kit (Pierce) and bovine serum albumin as a standard.
N-terminal Amino Acid Sequence Determination-Edman degradation was used to confirm the identity of purified proteins and quantify protein subunits from Western blots.
UV-Visible Light Absorption Spectroscopy-UV-visible light spectra of purified recombinant BchNB complexes and BchL were recorded using a V-550 spectrophotometer (Jasco, Groß Umstadt, Germany) under strict anaerobic conditions.
Iron Determination Method-Protein-bound iron was determined colorimetrically with o-phenantroline after acid denaturation of purified BchNB or BchL (24). The iron content of wild-type BchNB and BchL was confirmed by commercial inductively coupled plasma mass spectrometry (ICP-MS) (CURRENTA Bayer-Analytics, Leverkusen, Germany).
Detection of Potential Flavins-Standard detection methods for FMN and FAD cofactors were performed as described elsewhere (25,26).
Analysis of a Potential Heme Cofactor-The presence of a potential heme cofactor was screened as described elsewhere (27).
Pchlide Reduction Assay-DPOR activity was measured in 250-l assays in 100 mM Hepes-NaOH (pH 7.5), 2 mM ATP, 5 mM MgCl 2 , 20 M Pchlide, and 10 mM dithiothreitol as electron donor and an ATP regenerating system (20 mM creatine phosphate and 20 units of creatine phosphokinase/assay). This standard DPOR assay was supplemented with 30 -100 l of cell-free E. coli extract containing overproduced GST-BchN, BchB, and BchL. In DPOR reconstitution assays, 20 g of BchNB complex and 20 -60 g of GST-BchL fusion protein were added. After 30 -70 min at 35°C under strict anaerobic conditions in the dark, reactions were stopped by 1 ml of acetone. Pchlide and Chlide in supernatants obtained by centrifuging for 10 min at 12,000 ϫ g were spectrophotometrically quantified using extinction coefficient ⑀ 626 ϭ 30.4 mM Ϫ1 cm Ϫ1 for Pchlide (23) and ⑀ 665 ϭ 74.9 mM Ϫ1 cm Ϫ1 for Chlide (28). To investigate the ability of ferredoxin to function as elctron donor, the reconstitution assay was supplemented with a ferredoxin regenerating system (13 mM glucose 6-phosphate, 1.1 unit of glucose-6-phosphate dehydrogenase from Torula yeast (Sigma), 1.64 mM NADP ϩ , and 0.025 units of ferredoxin-NADP ϩ oxidoreductase from spinach (Sigma)). The assay was also supplemented with commercially available ferredoxins from Clostridium, Porphyra, or spinach (Sigma) or with recombinantly produced ferredoxin from Synechococcus (29) at a concentration of 25 g/ml, respectively.
Activation Energy of DPOR Catalysis-To determine the activation energy of ATP-driven DPOR catalysis, the standard assay was repeated between 291 and 319 K. The initial rate of Chlide formation was used to calculate the rate constant at each temperature and the activation energy was determined by the Arrhenius equation, where k is the rate constant, E act the activation energy, R the gas constant (8.314 kJ mol Ϫ1 K Ϫ1 ), T the temperature (K), and c a reaction constant.
CN Ϫ Inhibition of DPOR-To analyze the inhibitory effect of CN Ϫ on DPOR catalysis, standard DPOR assays were carried out in the presence of 0 -60 mM sodium cyanide.
Molecular Mass Determination-Analytical gel permeation chromatography was performed using a Superdex 200 HR 10/30 column (GE Healthcare), equilibrated with wash buffer containing 100 mM NaCl. The column was calibrated with protein standards (Molecular Weight Marker Kit MW-GF 1000, Sigma) at a flow rate of 0.5 ml min Ϫ1 . A 15-l sample of purified BchNB or BchL (ϳ45 g) was run under identical conditions (30).
Molybdenum Cofactor Analysis-To identify a possible molybdenum cofactor coordinated by BchNB, the nit-1 reconstitution assay supplemented with 4 g of BchNB was used (31). 1.2 mg of BchNB were used to detect the stable oxidation product FormA of the cofactor (32). Subsequent product dephosphorylation, sample preparation, and HPLC analysis were performed as published (33).
DPOR Production by Strains of E. coli Defective in Molybdenum Uptake-DPOR was produced by transforming pGEX-bchNBL* into E. coli RK-5207, a strain unable to import molybdenum from the medium (34,35). To overcome the phenotype of the RK-5207 mutant, DPOR was concurrently produced in the presence of 1 and 10 mM Na 2 MoO 4 in LB medium. DPOR activity was tested in cell-free extracts.
Molybdenum and Vanadium Determination by ICP-MS-To rule out the involvement of molybdenum or vanadium in DPOR catalysis ICP-MS was performed (36,37). Samples containing 5 g of purified protein were applied to the ICP-MS.

RESULTS AND DISCUSSION
C. tepidum DPOR Produced in E. coli Is Enzymatically Active-The complete bchNBL operon of the thermophilic green sulfur bacterium C. tepidum was cloned into a standard E. coli expression vector, producing the DPOR subunit BchN as an N-terminal glutathione S-transferase fusion protein (pGEX-bchNBL). However, amounts of recombinant proteins and DPOR activity were found to be low. Replacing the C. tepidum-type ribosomal binding site upstream of bchB (CCAACGAACCAA) and bchL (CCTCTATCATAC) with an E. coli type ribosomal binding site (GGAAACAGTATT) resulted in an expression vector (pGEX-bch-NBL*) that allows significant amounts of BchB and BchL to be produced (data not shown). Concomitantly, the DPOR activity of E. coli cell extracts increases from 0.29 to 3.15 nmol min Ϫ1 mg Ϫ1 (standard DPOR assay, 35°C) (Fig.  3B, traces a-d), a significant improvement in specific activity compared with ϳ40 pmol min Ϫ1 mg Ϫ1 for DPOR from R. capsulatus cell-free extracts (23). Despite distinct homology to nitrogenase, DPOR activity does not require further enzymes, such as molybdenum cofactor biosynthesizing or incorporating proteins, known to be essential to nitrogenaseproducing organisms. Orthologues of genes encoding these proteins are not found in the genome of the E. coli production host.
Purification of the Heterotetrameric BchNB Complex-DPOR was maintained under strict anaerobic conditions throughout, including during cell harvesting, cell disruption, protein purification, and activity assays. BchN was purified from E. coli pGEX-bchNBL* extracts. GST-BchN fusion protein was coupled to glutathione-Sepharose, washed extensively, and either eluted intact with 10 mM glutathione (Fig. 3A, lane 2 coli extracts from cells containing pGEX-bchL after isopropyl ␤-Dthiogalactopyranoside induction, affinity chromatography on glutathione-Sepharose, extensive washing, and glutathione elution. B, absorption spectra of standard DPOR assays using E. coli cell extracts or reconstitution assays after 20 min at 35°C and acetone extraction. Trace a, standard DPOR assay containing 30 l of E. coli extract; trace b, assay mixture without dithionite; trace c, assay mixture without ATP; trace d, control reaction without cell extract; trace e, reconstitution assay using 20 g of purified (BchNB) 2 and 20 g of purified BchL 2 ; trace f, control reaction using 20 g of (BchNB) 2 but no BchL 2 ; trace g, control reaction using 20 g of BchL 2 but no (BchNB) 2 . C, determination of the activation energy of DPOR catalysis. In an Arrhenius plot the logarithm of activity (ln k, where k is the initial rate of Chlide formation in the standard DPOR assay) is plotted versus the reciprocal of the absolute temperature in K. D, cyanide inhibition of DPOR. The activity of standard DPOR assays (15 min at 35°C) in the presence of 0 -60 NaCN is plotted against the concentration of NaCN. 50% inhibition of DPOR is achieved at 36 mM NaCN.
or liberated from bound GST by PreScission protease treatment (Fig. 3A, lane 3). SDS-PAGE analyses confirm the masses of GST-BchN (M r ϭ 74,000) or BchN (M r ϭ 47,000) and reveal stoichiometric amounts of a second protein (M r ϭ 59,000) (Fig.  3A, lanes 2 and 3). By N-terminal amino acid sequencing BchB was identified in equimolar amounts. The molecular mass of the BchNB complex was estimated to be M r ϭ 210,000 by gel filtration (Fig. 4C) implying a stoichiometry of 2BchN (46,643 Da each) and 2BchB (58,968 Da each). Henceforth the complex is therefore denoted (BchNB) 2 . Overall, we purified 2 mg of pure (BchNB) 2 complex per liter of medium.
Purification of the Homodimeric BchL Complex-To allow reconstitution of DPOR, the subunit BchL was separately purified from E. coli extracts. The GST-BchL fusion protein was purified by affinity chromatography and eluted as GST fusion protein with 10 mM glutathione buffer or cleaved off GST by PreScission protease. SDS-PAGE analysis indicates a single protein of M r ϭ 57,000 for GST-BchL (Fig. 3A, lane 4) or M r ϭ 30,000 for BchL. Analytical gel filtration revealed a relative molecular mass of about 60,000 (Fig. 4C) for BchL indicating that purified BchL from C. tepidum is homodimeric, henceforth denoted BchL 2 . Overall, 1.5 mg of pure BchL 2 were recovered from 5 liters of E. coli culture.
Reconstitution of DPOR Activity-To reconstitute DPOR activity in vitro, 20 g (0.1 nmol) of (BchNB) 2 and 20 -60 g (0.7-2 nmol) of BchL were assayed using 20 M Pchlide in the presence of 10 mM dithionite as reductant, 2 mM ATP, and an ATP regenerating system. After 20 min incubation under anaerobic conditions at 35°C, the reactions were stopped by adding 80% (v/v) acetone. Photometric analysis indicated that the reconstituted C. tepidum enzyme efficiently converts Pchlide (absorbance maximum 626 nm) into Chlide (absorbance maximum 665 nm; Fig.  3B, trace e). Chlide production was not detected if BchNB (trace f), BchL (trace g), dithionite, or the ATP regenerating system were omitted, or when the assay was performed under aerobic conditions (data not shown).
Kinetic Properties of DPOR Catalysis-To establish the K m values for C. tepidum DPOR we determined the initial velocity of Chlide formation over a broad range of Pchlide, ATP, and dithionite concentrations using the outlined DPOR assay and saturating concentrations for the residual substrates. For Pchlide, ATP, and the artificial co-substrate dithionite, Chlide formation followed Michaelis-Menten kinetics. The apparent K m for Pchlide was found to be 6.1 M, slightly lower than the 10.6 M measured for DPOR from R. capsulatus (22). The K m of C. tepidum DPOR for ATP was 13.5 M, and 52.7 M for dithionite (supplemental data Fig. 1).
Thermodynamic Properties of Pchlide Reduction by C. tepidum DPOR-Initial rates of Chlide formation were used to calculate rate constants at various temperatures. An Arrhenius plot of ln k versus 1/T indicates an activation energy for DPOR electron transfer of 22.2 kJ mol Ϫ1 (Fig. 3C), only slightly higher than the 18.8 kJ mol Ϫ1 reported for light-driven Pchlide reduction by LPOR (6) but dramatically lower than the 90 -97 kJ mol Ϫ1 required for the electron transfer from the Fe-protein to the MoFe-protein of nitrogenase (38,39). The nitrogenase reaction is exergonic by ϳ64 kJ mol Ϫ1 (40). Overall, nitrogenase requires 16 ATP molecules to overcome the kinetic barrier for the reduction of nitrogen (40). Pchlide reduction is similarly  ϭ 443,000). The inferred mass of BchNB is M r ϭ 210,000 Ϯ 10,000 indicating a (BchNB) 2 tetramer, whereas that of BchL is M r ϭ 60,000 Ϯ 5,000 indicating a BchL dimer (BchL 2 ). APRIL 18, 2008 • VOLUME 283 • NUMBER 16 exergonic with a ⌬H ϳ40 kJ mol Ϫ1 (5). As the kinetic barrier to DPOR catalysis is only a quarter that of nitrogenase, at least 4 molecules of ATP would be required to transfer two electrons during the reduction of Pchlide. The even number of ATP molecules would match the dimeric nature of BchL 2 , whereas electron transfer presumably occurs via the inter subunit [Fe-S] cluster as in nitrogenase.

DPOR from C. tepidum, Chlorophyll Biosynthesis, and Nitrogenase
CN Ϫ Resistance of DPOR-To establish whether CN Ϫ also uncouples ATP hydrolysis from electron transport in DPOR as observed in nitrogenase (41), the effect of CN Ϫ on DPOR catalysis was analyzed with the standard DPOR assay. Plotting the DPOR activity (Chlide formation) against CN Ϫ concentration (Fig. 3D) indicates that 36 mM NaCN is required to reduce DPOR activity by half, significantly higher than the 5 mM NaCN for A. vinelandii nitrogenase (Mo-nitrogenase) (42). The metal centers of (BchNB) 2 are thus clearly not as sensitive with respect to CN Ϫ as those of nitrogenase, indicating distinct differences in the metal centers of DPOR and nitrogenase.
Cysteine Residues Cys 97 and Cys 131 of BchL Coordinate an Intersubunit Iron-Sulfur Cluster-Purified, GST-tagged BchL 2 is brownish in color. A UV-visible light absorption spectrum recorded under strict anaerobic conditions reveals an absorption peak at 428 nm typical for a [4Fe-4S] cluster (Fig. 4B, spectrum a). The absorption in the range of 390 -428 nm bleaches after addition of the reductant dithionite (2 mM, 1 h) (Fig. 4B,  spectrum b), whereas oxidation by oxygen exposure for 15 h (Fig. 4B, spectrum c) as well as EDTA treatment (10 mM, 2 h) completely abolishes the absorption in this area of the spectrum. This observation is in accordance to UV-visible spectra obtained for proteins containing iron-sulfur redox clusters (43)(44)(45). By applying an extinction coefficient for [4Fe-4S] clusters of ⑀ 410 ϭ 15 mM Ϫ1 cm Ϫ1 (44, 46) we calculated 0.9 mol of [4Fe-4S] cluster per mol of BchL 2 . This ratio was confirmed by ICP-MS quantification revealing 3.62 mol of iron/mol of BchL 2 . Amino acid sequence alignments of BchL proteins reveal five conserved cysteines. Each cysteine was individually substituted by serine to identify potential iron-sulfur cluster ligands. Only variants C97S and C131S cause a significant loss in DPOR activity to Ͻ0.5%, whereas C38S and C162S retain 100% and C239S 80% of wild-type activity. Correspondingly, the iron and sulfur content of BchL variants C97S and C131S only were reduced (Table 1). These cysteines thus presumably coordinate the inter-subunit [4Fe-4S] cluster, analogous to Cys 97 and Cys 132 in NifH 2 . These results are in good agreement with a recent study for R. capsulatus DPOR, where an EPR signal for a [4Fe4S] cluster was obtained (47).
Residues Lys 10 and Leu 126 of BchL Are Essential to Catalysis-In nitrogenase, binding of NifH 2 to (NifDK) 2 is dependent upon ATP-induced conformational changes. Subsequent ATP hydrolysis induces further conformational rearrangements that decrease the redox potential of the NifH 2 [4Fe-4S] cluster, transferring a single electron to (NifDK) 2 , followed by the dis-

iron, and sulfur content of mutant DPOR enzymes
The specific enzymatic activity of the wild-type DPOR (3.15 nmol min Ϫ1 mg Ϫ1 ) was set as 100%. The iron and sulfur contents per mol of purified wild-type (BchNB) 2 and BchL 2 were set as 100%. DPOR protein variants bchN C46S, bchN C103S, bchL C97S, bchL C131S, bchL K10A, and bchL ⌬L126 were additionally tested with 260 g of (BchNB) 2 and 200 g of BchL 2 per assay with an increased incubation time for up to 4 hs. Obtained activities were below to the detection limit of the assay at 15.75 nmol min Ϫ1 mg Ϫ1 . sociation of NifH 2 from (NifDK) 2 . Two sequence motifs are crucial to ATP-binding and subsequent signal transmission. The "switch II" motif DVLGDVVCGG conformationally relays ATP binding to the [4Fe-4S] cluster inter alia by repositioning cluster ligand Cys 132 . Deleting Leu 127 of the switch II motif traps NifH 2 in a conformation preventing hydrolysis of ATP thus inactivating nitrogenase (48). This "trapped" NifH 2 variant forms a stable ternary complex with (NifDK) 2 allowing simple co-purification. The N-terminal P-loop region GXXXXGK(S/T) coordinates the ␥-phosphate of ATP primarily by Lys 15 (A. vinelandii numbering). Replacing Lys 15 by glutamine completely inactivates nitrogenase despite ATP binding being unaffected.
Lys 15 and Leu 127 of NifH correspond to Lys 10 and Leu 126 of BchL, conserved in all BchL protein sequences. Mutating bchL to produce the variants K10A and L126⌬ (Leu 126 deleted) in the plasmid pGEXbchNBL* results in a specific activity Ͻ0.5% of wild-type DPOR. Lys 10 and Leu 126 are thus essential for DPOR catalysis indicating that ATP hydrolysis and electron transfer in DPOR may be analogous to nitrogenase. Attempts to purify a ternary BchNBL complex using either BchL variant L126⌬ or Mg-ADP in combination with metallofluorides (AlF 4 Ϫ and BeF 3 Ϫ ), in analogy to nitrogenase, were not successful, indicating differences in the transient interaction of BchL 2 with (BchNB) 2 to nitrogenase (18).
Three-dimensional Structural Model of BchL-The amino acid sequence of BchL of DPOR shares a sequence identity of 33% with that of NifH of nitrogenase, allowing for three insertions/deletions involving nine residues distinguish BchL and NifH. Based on the structure of ATP-bound NifH 2 in complex with (NifDK) 2 (Protein data bank code 1M34) (49), we pro-duced a structural model of BchL (Fig. 5) using SwissModel (50) and Bragi (51). Conserved and conservatively replaced residues of BchL and NifH are located in the immediate vicinity of the [4Fe-4S] cluster, the BchL-BchL interface, the ATPbinding pocket, and the protein core of each subunit. Conversely, essentially no conservation of amino acids is observed on the surface of the BchL-monomer outside of the BchL-BchL interface. This is also true for the surface involved in binding (BchNB) 2 , indicating that the subunits of both DPOR and nitrogenase have evolved to optimize recognition within each complex but diverged to prevent cross-interaction between related subunits of alternate complexes. The conservation of residues at the dimer interface, the iron-sulfur cluster, and the ATP-binding site, supports the idea of a bridging [4Fe-4S] cluster in BchL 2 (see above) coordinated by cysteine residues Cys 97 and Cys 131 and that BchL 2 utilizes a related ATP-controlled gating mechanism to transfer electrons to (BchNB) 2 . The structural BchL 2 model, furthermore, places residues Cys 38 , Cys 162 , and Cys 239 within the hydrophobic core of each subunit, corroborating the interpretation that these cysteines are not involved in iron-sulfur cluster coordination. Substituting Cys 239 by serine presumably destabilizes the protein as a whole, explaining the observed partial loss of DPOR activity of the mutant C239S.
DPOR Catalysis Is Independent of a Molybdenum Cofactor-The presumed evolutionary relationship of DPOR to nitrogenase is supported by the similarity of the amino acid sequences of BchN, BchB, and BchL to those of NifD, NifK, and NifH as well as by the identical oligomeric state of (BchNB) 2 and (NifDK) 2 , and of BchL 2 and NifH 2 . This raises the question regarding conserved metal centers.
To identify a potential molybdopterin cofactor, (BchNB) 2 was analyzed by the nit-1 complementation assay. The mutant nit-1 of Neurospora crassa is deficient in molybdenum cofactor biosynthesis, producing a stable apo-form of nitrite reductase. In cell-free extracts of this mutant, nitrite reductase activity can be restored by adding molybdenum cofactor (31). Purified (BchNB) 2 is not able to complement this mutant. HPLC analyses failed to detect the stable oxidation product FormA-dephospho of molybdopterin in purified (BchNB) 2 . Next we produced DPOR in E. coli strain RK-5207 deficient in molybdenum uptake (52). Mo-protein synthesis in RK-5207 may be rescued by adding MoO 4 2Ϫ to the growth medium (34,35  the growth medium decreases DPOR activity by ϳ70%. Clearly, production of active DPOR in E. coli does not require molybdenum. Correspondingly, ICP-MS analyses of purified (BchNB) 2 indicated molybdenum and vanadium concentrations no higher than those of the negative controls. (BchNB) 2 , in contrast to the MoFe-protein of nitrogenase, thus does not bind nor require a molybdenum cofactor.
Proposed Enzymatic Mechanism for DPOR-Summarizing the above inferences, we propose that one of the two "plant type" [2Fe-2S]-ferredoxins in the genome of C. tepidum might be the natural electron donor of DPOR. The P-loop region (incorporating Lys 10 ) and the switch II region (Leu 126 ) are crucial for ATP-binding and dynamic positioning. We, furthermore, expect two molecules of ATP to bind per BchL dimer. Because the activation energy (22.2 kJ mol Ϫ1 ) for Pchlide reduction is a fourth that of nitrogenase, we propose that at least four molecules of ATP are required to overcome the kinetic barrier to the two-electron reduction of Pchlide. This scenario is in good agreement with the dimeric state of BchL 2 , presumably bridged by an intersubunit [4Fe-4S] cluster. The cluster would transfer a single electron to a second intersubunit [4Fe-4S] cluster coordinated by both BchN and BchB within the (BchNB) 2 complex. This second cluster then transfers the electron to the substrate, resulting in a reduced Pchlide radical. Consumption of another two ATP molecules by BchL 2 would then allow a second electron to follow the first, resulting in fully reduced Chlide.