Substrate Recognition of Nitrogenase-like Dark Operative Protochlorophyllide Oxidoreductase from Prochlorococcus marinus*

Chlorophyll and bacteriochlorophyll biosynthesis requires the two-electron reduction of protochlorophyllide a ringDbya protochlorophyllide oxidoreductase to form chlorophyllide a. A light-dependent (light-dependent Pchlide oxidoreductase (LPOR)) and an unrelated dark operative enzyme (dark operative Pchlide oxidoreductase (DPOR)) are known. DPOR plays an important role in chlorophyll biosynthesis of gymnosperms, mosses, ferns, algae, and photosynthetic bacteria in the absence of light. Although DPOR shares significant amino acid sequence homologies with nitrogenase, only the initial catalytic steps resemble nitrogenase catalysis. Substrate coordination and subsequent [Fe-S] cluster-dependent catalysis were proposed to be unrelated. Here we characterized the first cyanobacterial DPOR consisting of the homodimeric protein complex ChlL2 and a heterotetrameric protein complex (ChlNB)2. The ChlL2 dimer contains one EPR active [4Fe-4S] cluster, whereas the (ChlNB)2 complex exhibited EPR signals for two [4Fe-4S] clusters with differences in their g values and temperature-dependent relaxation behavior. These findings indicate variations in the geometry of the individual [4Fe-4S] clusters found in (ChlNB)2. For the analysis of DPOR substrate recognition, 11 synthetic derivatives with altered substituents on the four pyrrole rings and the isocyclic ring plus eight chlorophyll biosynthetic intermediates were tested as DPOR substrates. Although DPOR tolerated minor modifications of the ring substituents on rings A–C, the catalytic target ring D was apparently found to be coordinated with high specificity. Furthermore, protochlorophyllide a, the corresponding [8-vinyl]-derivative and protochlorophyllide b were equally utilized as substrates. Distinct differences from substrate binding by LPOR were observed. Alternative biosynthetic routes for cyanobacterial chlorophyll biosynthesis with regard to the reduction of the C8-vinyl group and the interconversion of a chlorophyll a/b type C7 methyl/formyl group were deduced.

Protochlorophyllide (Pchlide) 2 a is a metabolite during chlorophyll and bacteriochlorophyll biosynthesis. Two distinct strategies have evolved for the stereo-and regioselective two electron reduction of ring D of Pchlide a to form chlorophyllide (Chlide) a ( Fig. 1) (1)(2)(3). The light-dependent Pchlide oxidoreductase (LPOR; NADPH Pchlide oxidoreductase, EC 1.3.1.33) forms a complex with its substrates Pchlide a and NADPH. Absorption of light by the former induces the subsequent 17,18-trans reduction by the latter (4 -6). Because of this crucial light-dependent step, flowering plants can only synthesize chlorophylls in the presence of light (7,8). The second Pchlide a-reducing enzyme is the light-independent, dark operative Pchlide oxidoreductase (DPOR). It uses ATP to drive the production of Chlide a. Anoxygenic bacteria solely contain DPOR (9), whereas angiosperms produce only LPOR. Gymnosperms, mosses, ferns, algae, and cyanobacteria possess both LPOR and DPOR simultaneously (2). The presence of DPOR is responsible for the ability of these organisms to synthesize chlorophyll or bacteriochlorophyll in the dark. DPOR consists of three protein subunits that are termed ChlN, ChlB, and ChlL in chlorophyll-synthesizing organisms (10 -12) and BchN, BchB, and BchL in bacteriochlorophyll-synthesizing organisms (13). They share significant amino acid sequence similarities with the NifD, NifK, and NifH subunits of nitrogenase, respectively (12,14).
The first part of the proposed catalytic mechanism for DPOR strongly resembles the initial steps of the electron transfer processes of nitrogenase catalysis. Based on two different investigations using DPOR enzymes from Chlorobium tepidum and from Rhodobacter capsulatus (13,15) a "plant type" [2Fe-2S] ferredoxin was proposed to transfer one electron to the dimeric DPOR subcomplex BchL 2 containing an intersubunit [4Fe-4S] redox center that is coordinated by residues Cys 97 and Cys 131 (C. tepidum numbering) (13). Residues Lys 10 in the P-loop region and Leu 126 in the so called switch II region of BchL have been found to be crucial for the binding of ATP. Two molecules of ATP bound to the homodimeric BchL 2 subunit are required to overcome the kinetic barrier of 22.2 kJ mol Ϫ1 for Pchlide a * This work was supported by grants from the Deutsche Forschungsgemeinschaft. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1 To whom correspondence should be addressed. Tel.: 49-531-391-5808; Fax: 49-531-391-5854; E-mail: j.moser@tu-bs.de.
reduction. The hydrolysis of the two ATP molecules might result in a decrease in the redox potential of the BchL 2bound [4Fe-4S] cluster allowing for the transfer of one electron onto a second protein complex composed of BchN and BchB. For this complex a heterotetrameric structure was determined and denoted as (BchNB) 2 (13,15,16). Site-directed mutagenesis experiments of BchN and BchB in combination with kinetic measurements revealed the presence of four cysteine residues crucial for DPOR catalysis. They indicated a significant decrease of the overall iron and sulfur content of (BchNB) 2 modified in residues Cys 21 , Cys 46 , and Cys 103 of BchN and residue Cys 94 of BchB. From these data, two intersubunit [4Fe-4S] redox centers were proposed for (BchNB) 2 (13). This first part of the DPOR catalytic mechanism clearly resembles the initial electron transfer in nitrogenase (13,17,18). However, the [4Fe-4S] clusters of (BchNB) 2 have no direct equivalent in nitrogenase. Instead, nitrogenase contains a complicated [8Fe-7S]-P cluster, which interestingly shares all of the four cysteine ligands that have been identified for C. tepidum DPOR (13,16). The two additional cysteine residues involved in P cluster coordination have no counterpart in the sequence of BchNB proteins.
The P cluster of nitrogenase is responsible for the electron transfer onto an additional metallo center (1Mo-7Fe-9S-1X-1homocitrate) located on subunit NifD, where the reduction of N 2 to NH 4 ϩ takes place (19,20). However, involvement of such an additional cofactor for DPOR catalysis was clearly ruled out, and a direct electron transfer from the [4Fe-4S] cluster of (BchNB) 2 onto the substrate was proposed (13). Obviously, the second part of DPOR catalysis differs distinctly from the nitrogenase system. To date no structural information about (BchNB) 2 is available, but it seems reasonable that the active site might be located in close proximity to the [4Fe-4S] cluster of (BchNB) 2 .
Marine cyanobacteria including Prochlorococcus marinus are presumably the most abundant photosynthetic organisms on earth (21). They contribute 30 -60% of the oceanic primary production (21)(22)(23)(24) and play a significant role in the global carbon cycle.
In comparison with other cyanobacteria, P. marinus has an unusual pigment composition (21 (26). The [8-vinyl]-chlorophylls are usually precursors for the synthesis of chlorophylls. The only other organisms where they have been found in significant amounts are chlorophyll biosynthesis mutants of C. tepidum (27) and maize (28,29).
Although LPOR uses both Pchlide a and [8-vinyl]-Pchlide a as a substrate, with a preference for the latter, there is no study on the substrate specificity of DPOR. To determine the exact location of the [8-vinyl]-reduction step in the biosynthetic pathway, we wanted to see whether DPOR from P. marinus is able to reduce Pchlide a and [8-vinyl]-Pchlide a. Furthermore, it has been shown that the 7-formyl group of chlorophyll b is mainly synthesized from chlorophyll a by the chlorophyllide a oxygenase (CAO) (30), whereas Pchlide a is only a poor substrate for CAO. Because of this observation, it was of interest to elucidate a potential role of Pchlide b as a substrate for DPOR catalysis.
In the present investigation we established a heterologous expression system for the DPOR enzyme from P. marinus, the first cyanobacterial DPOR to be analyzed. Three catalytical [4Fe-4S] clusters were characterized by EPR spectroscopy. To determine the structural elements of Pchlide that are critical for DPOR catalysis, we used chemically modified substrates. The results of these substrate recognition experiments allowed us to propose the location of individual reaction steps in the chlorophyll biosynthetic pathways.

EXPERIMENTAL PROCEDURES
Production and Purification of P. marinus DPOR in Escherichia coli-The gene chlN from P. marinus (SS120, Roscoff Culture Collection RCC156) was PCR-amplified using primers CAGCGAATTCATGAGCGGCTCAACG and GCCGTCGA-CTTAAACAGCTTCTAGAG and cloned into the BamHI and SalI sites of pGEX-6P-1 to yield pGEX-chlN. Subsequently, chlB from P. marinus was amplified using primers GACGGTCGA-CTCAATTTCACACAGGAAACAGTATTCATGGAACT-AACACTTTGA and GCTGTATGCGGCGCTTCAAGCTC-CGAAATGAG and cloned into the SalI and NotI sites of pGEX-chlN to generate plasmid pGEX-chlNB (The E. coli-specific ribosomal binding site implemented upstream of chlB to enhance the protein production is marked in bold letters.) Using primers AGCGGATCCATGACTACAACCTTAGC and TACCGTCGACACCCTAGTCAAAACC, the corresponding chlL gene was amplified into the BamHI and NotI sites of pGEX-6P-1 to yield plasmid pGEX-chlL.
Both plasmids were individually transformed into E. coli BL21(DE3) Codon Plus RIL cells to produce the ChlNB complex or the ChlL subunit, respectively. The cells were aerobically cultivated at 25°C in 500 ml of LB medium containing 1 mM Fe(III)-citrate and 1 mM L-cysteine. At an A 578 of 0.5, protein production was induced by the addition of 25 M isopropyl-␤-thiogalactoside. After 16 h of cultivation, 1.7 mM dithionite was added, and cultivation was continued without agitation for 3 h at 18°C in an anaerobic chamber (Coy Laboratories, Grass Lake, MI) to allow [Fe-S] cluster formation. All of the remaining steps were performed under anaerobic condi- tions (95% N 2 , 5% H 2 , Ͻ1 ppm O 2 ). The solutions were N 2 saturated prior use. The cells were harvested by centrifugation. The bacterial cell pellet was resuspended in 15 ml of lysis buffer (100 mM Hepes/NaOH, pH 7.5, 10 mM MgCl 2 , 150 mM NaCl, 10 mM dithiothreitol) and disrupted by a single passage through a French press at 1500 p.s.i. into an anaerobic bottle. Following centrifugation for 60 min at 175,000 ϫ g at 4°C, the supernatant was applied to 1 ml of glutathione-Sepharose (GE Healthcare) equilibrated with lysis buffer. After washing with 20 ml of phosphate-buffered saline containing 10 mM dithiothreitol (washing buffer), the recombinant fusion protein GST-ChlN in complex with ChlB or alternatively GST-ChlL alone were eluted using 2 ml of lysis buffer containing 10 mM glutathione in its reduced form. Alternatively, ChlNB or ChlL were liberated from bound GST via PreScission TM protease (GE Healthcare) treatment.
Fractions containing the ChlNB complex or ChlL were identified by SDS-PAGE. Determination of Native Molecular Mass-Analytical gel permeation chromatography was performed using a Superdex 200 HR 10/30 column (GE Healthcare), equilibrated with lysis buffer. 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 200-l sample of purified (ChlNB) 2 or ChlL 2 (ϳ0.6 mg) was run under identical conditions. For preparative gel permeation chromatography up to 5 ml of purified (ChlNB) 2 or ChlL 2 were run at 1.5 ml min Ϫ1 on a Superdex 200 26/60 column (GE Healthcare) under strict anaerobic conditions. The eluted proteins were detected at 280 nm and collected in 1.5-ml aliquots.
Determination of Protein Concentration-The BCA protein assay kit (Pierce) was used according to the manufacturer's instructions with bovine serum albumin as a standard.
N-terminal Amino Acid Sequence Determination-Automated Edman degradation was used to confirm the identity of purified proteins.
Iron Determination Method-The iron content of the purified (ChlNB) 2 complex and ChlL 2 were confirmed colorimetrically with bathophenanthroline after acid denaturation (31).
Preparation of EPR Samples-Sample preparation was carried out in an anaerobic chamber. Purified P. marinus (ChlNB) 2 and ChlL 2 were concentrated to 77 and 50 M, respectively, using an Amicon stirred ultrafiltration cell (Millipore, Bedford, MA) equipped with a 50,000-Da compound excluding ultrafiltration membrane. Ten l of a sodium dithionite solution (100 mM) was added to 100 l of protein solution and incubated for 10 -30 min. Where indicated, 100 l of (ChlNB) 2 were supplemented with 10 l of ATP (2 mM) and 50 l of ChlL. Control samples contained protein fractions as purified. The proteins were finally transferred to quartz EPR tubes with 4-mm outer diameters and frozen in liquid nitrogen. EPR Spectroscopy-9.5-GHz X-Band EPR spectra were recorded on a Bruker ESP300E spectrometer equipped with a rectangular microwave cavity in the TE 102 mode. For temperature control at 10,000, the sample was kept in an Oxford ESR 900 helium flow cryostat with an Oxford ITC4 temperature controller. The microwave frequency was detected with an EIP frequency counter (Microwave Inc., San Jose, CA). The magnetic field was calibrated using a Li/LiF standard with a known g value of 2.002293 Ϯ 0.000002 (34). Base-line corrections, if required, were performed by subtracting a background spectrum, obtained under the same experimental conditions from a sample containing only buffer solution. Simulations of the experimental EPR spectra have been carried out with the program EasySpin (35).
Protochlorophyllide Preparation-Pchlide (substrate 1; see Table 1 and Fig. 4) was isolated from the bchL-deficient R. capsulatus strain ZY-5 (36) and subsequently purified by affinity purification in combination with preparative high performance liquid chromatography according to Ref. 13.
Pchlide Reduction Assay-DPOR activity was measured in 125-l assays containing 100 mM Hepes/NaOH, pH 7.5, 2 mM ATP, 5 mM MgCl 2 , 13 M Pchlide, and 2 mM dithiothreitol as an artificial electron donor. An ATP-regenerating system consisting of 20 mM creatine phosphate and 20 units/assay creatinephosphokinase was employed. DPOR assays contained 100 pmol of purified (GST-ChlNB) 2 and 200 pmol of purified (GST-ChlL) 2 . Standard assays were incubated under strict anaerobic conditions for 5 min up to 70 min at 25°C in the dark. For the determination of the temperature optimum for P. marinus DPOR, tests were performed at 15-40°C for 20 min. The reactions were stopped by adding 500 l of acetone. After centrifugation for 30 min at 12,000 ϫ g, Pchlide a and Chlide a in the supernatant were spectrophotometrically quantified by using an extinction coefficient of ⑀ 626 ϭ 30.4 mM Ϫ1 cm Ϫ1 for Pchlide a (16) and ⑀ 665 ϭ 74.9 mM Ϫ1 cm Ϫ1 for Chlide a (37). For the analysis of DPOR substrate recognition, the Pchlide reduction assay was performed in the presence of substrate analogs (see Table 1).
DPOR Substrate Competition Assay-All of the substrate analogs showing reduced activity in the Pchlide reduction assay were tested for their ability to inhibit DPOR catalysis. Therefore, Pchlide reduction assays were performed in the presence of 13 M Pchlide and additionally supplemented with the Pchlide derivatives found not to sustain DPOR activity in concentrations up to 20 M. Chlide formation was monitored spectrophotometrically.
Substrate Binding Assay-2 mg of purified (GST-ChlNB) 2 were bound to 200 l of glutathione-Sepharose (GE Healthcare) and incubated with 200 l of 25 M Pchlide (or the corresponding substrate analogs) for 10 min. After washing with 5 ϫ 200 l of lysis buffer (GST-ChlNB) 2 was eluted with 300 l of 15 mM glutathione in lysis buffer and analyzed for bound pigments by UV-visible spectroscopy.

RESULTS AND DISCUSSION
Purification and Biochemical Characterization of P. marinus DPOR-GST-ChlN and ChlN each complexed with stoichiometric amounts of untagged ChlB were purified using glutathione-Sepharose (GE Healthcare) ( Fig. 2A, lanes 2 and 3). Their identity was confirmed by Edman degradation, further indicating a molar ratio of 1.1/1 for ChlN/ChlB. The relative native molecular mass of the purified ChlNB complex determined by size exclusion chromatography was 210,000, indicating a tetrameric (ChlNB) 2 complex (ChlN ϭ 46,199 Da, ChlB ϭ 58,729 Da). Twenty mg of the (ChlNB) 2 complex were purified from 1.5 liters of E. coli culture. Preparative size exclusion chromatography revealed a second minor elution peak (elution volume 208.5 ml) in addition to the dominant elution peak representing the highly purified (ChlNB) 2 complex (elution volume 190 ml) (Fig. 3B). This additional protein was identified as ChlN alone. Obviously, the slightly increased expression level for GST-ChlN over ChlB also resulted in the purification of free ChlN in the initial affinity purification step.
Analogously, DPOR subunit ChlL was chromatographically purified ( Fig. 2A, lanes 4 and 5). Gel permeation chromatography showed a relative native molecular mass of 60,000 for ChlL, indicating a ChlL 2 homodimer (ChlL ϭ 32,395 Da). Overall 6 mg of ChlL 2 were purified per liter of E. coli culture. From these results we conclude that the cyanobacterial DPOR from P. marinus features the same overall subunit architecture as the C. tepidum and R. capsulatus enzymes (13,16).
Clusters-Anaerobically purified (ChlNB) 2 is brownish in color, and the UV-visible spectra revealed an absorption maximum at 428 nm characteristic for [4Fe-4S] clusters. This absorption peak was bleached upon oxygen exposure (24 h) (Fig.  3A) or EDTA treatment (15 mM, 4 h) as described before (38,39). The determination of iron and sulfur contents of highly purified protein fractions yielded 7.7 mol of iron and 6.9 mol of sulfur/mol of (ChlNB) 2 . This is in good agreement with 2 [4Fe-4S] clusters/(ChlNB) 2 complex (13), and a recent study that revealed a weak [4Fe-4S] ϩ1 EPR signal for the BchNB complex of R. capsulatus after coincubation with subunit BchL and ATP (40).
Following reduction of the P. marinus (ChlNB) 2 complex with 10 mM dithionite, a complex EPR signal was observed that was tentatively attributed to reduced [4Fe-4S] ϩ1 clusters. Simulation of the obtained EPR spectrum was only possible by a superposition of two clusters (FeS-I and FeS-II) with different g values (Fig. 3C). FeS-II exhibited the typical spectral shape of a [4Fe-4S] ϩ1 cluster, having a rhombic g tensor. The g values deduced for this cluster from the simulation were g 1 ϭ 2.13(1), g 2 ϭ 2.008(5), and g 3 ϭ 1.89(1); the numbers in parentheses are the estimated errors in the last digit. The other cluster, FeS-I, exhibited an almost axial g tensor showing values of g 1 ϭ 2.115(5), g 2 ϭ 1.935(2), and g 3 ϭ 1.917 (2). These values are in a range where reduced [4Fe-4S] or [2Fe-2S] clusters could be expected (41,42). Based on the iron and sulfur analyses, a [4Fe-4S] cluster was deduced. The different g values of the FeS-I and FeS-II clusters indicate a different geometry, possibly because of differences in the coordination between the ChlN and ChlB subunits. No evidence for high spin states greater than S ϭ 1/2 (e.g. S ϭ 3/2, S ϭ 5/2) was obtained. The superimposed EPR spectrum of both clusters changed shape at higher temperatures (T ϭ 20 K), indicating different relaxation behavior for the [4Fe-4S] ϩ1 clusters or magnetic coupling between them. At temperatures above 20 K, the intensity of the EPR signals rapidly declined.
Reducing the (ChlNB) 2 complex in the presence of additional ATP and ChlL did not further increase the EPR signal intensity. The observed EPR signal intensity was in agreement with the rate of reduced (ChlNB) 2 estimated from the bleaching in the UV-visible spectra.
Fractions of concentrated ChlN alone (37 M) were brownish as well, and UV-visible spectroscopic analysis exhibits an absorption maximum at 415 nm indicative for [3Fe-4S] clusters (data not shown). However, following reduction with 10 mM dithionite, ChlN samples were EPR silent. From these findings we conclude that isolated ChlN still has the ability to coordinate a [Fe-S] center, albeit a residual [3Fe-4S] cluster that did not give rise to an EPR signal.
The properties of these [Fe-S] clusters agree with the results of a previous mutagenesis study for the identification of [4Fe-4S] cluster coordinating amino acid residues. Three cysteinylligands were identified on protein subunit BchN from C. tepidum, and only one ligand was located on corresponding subunit BchB (13). All three Cys-residues of ChlN are 100% conserved in the known protein sequences from cyanobacterial sources.
The results of our new cyanobacterial ChlL protein are in agreement with previous EPR studies for R. capsulatus BchL (g 1 ϭ 2.03, g 2 ϭ 1.94, and g 3 ϭ 1.86) (43). From these experiments and the results of an earlier mutagenesis study (13), we conclude that the DPOR ChlL 2 complex from P. marinus coordinates one intersubunit [4Fe-4S] cluster.
Reconstitution of DPOR Activity from Recombinant Purified (ChlNB) 2 and ChlL 2 -The functional DPOR enzyme was reconstituted by supplementing 100 pmol of purified (ChlNB) 2 complex and 200 pmol of ChlL 2 with 13 M Pchlide a 1, 2 mM dithionite as reductant and 2 mM ATP in combination with an ATP-regenerating system. UV-visible spectroscopic analyses of the acetone-extracted pigments demonstrated the effective reduction of Pchlide a 1 (absorbance maximum, 626 nm) to Chlide a (absorbance maximum, 665 nm) (Fig. 2B, spectrum a). This catalysis was abolished upon exposure to oxygen (data not shown). Likewise, no Chlide a was formed in the absence of either (ChlNB) 2 , ChlL 2 , dithionite, or ATP (Fig. 2B, spectra b-e). With this standard assay, a temperature optimum of 25°C was determined for our cyanobacterial DPOR system, which is in agreement with the mesophilic habitat of P. marinus. A specific activity of 910 pmol min Ϫ1 mg Ϫ1 was determined at 25°C. In a previous study for the moderate thermophilic C. tepidum DPOR system, a comparable specific activity of 3.12 nmol min Ϫ1 mg Ϫ1 was obtained (13). The initial velocity of product formation was measured over a broad range of substrate concentrations, whereas concentrations of the cosubstrates ATP and dithionite were kept at saturating concentrations. P. marinus DPOR catalysis followed Michaelis-Menten-type kinetics with a K m value of 6.9 M for Pchlide a 1, which again is comparable with K m ϭ 6.1 M obtained for C. tepidum DPOR (13).  2 and 100 M ChlL 2 exhibit significant absorption maxima at 428 nm. Following oxygen exposure for 24 h, these absorption maxima were bleached. B, preparative gel permeation chromatography of (ChlNB) 2 . 60 mg of purified (ChlNB) 2 after protease cleavage were analyzed on a Superdex 200 HR 26/60 column under anaerobic conditions (95% N 2 , 5% H 2 , Ͻ1ppm O 2 ) using a flow rate of 1 ml/min and monitoring the absorbance at 280 nm. EPR spectra of (ChlNB) 2 (C) and ChlL 2 (D) after reduction with dithionite were recorded at 10,000 and 9.5 GHz using a microwave power of 10 milliwatts. The obtained EPR spectra are accumulated from 32 measurements. Simulated spectra and the three g values are given. The spectra for (ChlNB) 2 exhibit a superposition of two nonidentical [4Fe-4S] clusters. For FeS-I, an alternative g 1 value of 2.045(5) (instead of g 1 ϭ 2.115(5); see text) is possible, however, with lower quality of the fit.

The (ChlNB) 2 Complex Binds the Substrate Pchlide a-Fol-
lowing incubation of purified (ChlNB) 2 bound to glutathione-Sepharose with 25 M of the substrate Pchlide a, (ChlNB) 2 was eluted with Pchlide a bound to the protein complex. We determined a ratio of 1.4 mol Pchlide/mol (ChlNB) 2 . These results are consistent with observations made for R. capsulatus DPOR (40) and the proposal of two active sites/(ChlNB) 2 complex (13). Pchlide a binding was observed without ATP, dithionite, or subunit ChlL being present, suggesting substrate binding by the catalytic (ChlNB) 2 complex to be the initial step in DPOR catalysis.
Zinc Protopheophorbide a Is an Efficient DPOR Substrate-For the analysis of the substrate specificity of chlorophyll biosynthetic enzymes, the use of zinc derivatives instead of magnesium-coordinating tetrapyrroles is well established (44 -46). When the standard DPOR assay was supplemented with 13 M zinc Ppheide a 2, a bathochromic absorption shift of 40 nm was observed, analogously to the reaction with the natural magnesium-containing substrate Pchlide a. By measuring the specific absorption of the reaction product zinc Pheide a, the initial velocity of product formation was measured over a broad range of substrate concentrations, whereas concentrations of the cosubstrates were kept at saturating concentrations. Zinc Pheide a formation followed Michaelis-Menten kinetics as described for the natural magnesium-containing substrate Pchlide a 1 with a K m value of 8.5 M (6.9 M for Pchlide a) and a specific activity of 730 pmol min Ϫ1 mg Ϫ1 (910 pmol min Ϫ1 mg Ϫ1 for Pchlide a). These are only minor differences, indicating that replacement of magnesium by zinc does not affect the reaction significantly. Zinc Ppheide a 2 and the corresponding 10 modified zinc derivatives used in this study are therefore considered suitable as putative substrates for the analysis of DPOR substrate recognition. Furthermore, we made use of eight magnesium-containing chlorophylls and chlorophyll biosynthetic intermediates, which were also tested as substrate analog. The absorption maxima and extinction coefficients of the individual substrates employed in this study are summarized in Table 1. All of the specific activities for the zinc-or magnesium-containing substrates analyzed in this study were related to values obtained for zinc Ppheide a 2 or Pchlide a 1, respectively.
Zinc Protopheophorbides with Altered Substituents on Ring A-In the first series of potential substrates, the C3 vinyl group on ring A was altered into a more polar formyl group (zinc Ppheide d 3) and into a more bulky acetyl group (zinc [3-acetyl]-Ppheide a 4). The C3 vinyl group was finally modified with a large phenylamino group (zinc 3 1 -phenylamino-Ppheide a 5) (Fig.  4A). The DPOR enzyme tolerated the polar formyl substituent of substrate 3, as indicated by a specific activity of 85% compared with substrate 2. However, compounds 4 and 5 did not result in any detectable DPOR activity. These results indicate the critical involvement of ring A in substrate recognition. Because the 3-formyl group in the accepted substrate 3 is of similar polarity as the 3-acetyl group in compound 4, which is not a substrate, we concluded that polarity at this site is of less importance for substrate recognition than size. In accordance with this hypothesis, compounds 4 and 5 neither functioned as competitive inhibitors in inhibition assays nor did they bind to the purified (ChlNB) 2 complex (Table 1).
Protochlorophyllide and Zinc Protopheophorbides with Altered Substituents on Ring B-The second set of putative substrates included compounds with alterations of the C7 methyl group into a formyl group (Pchlide b 6; zinc Ppheide b 7) and a phenylamino derivative (zinc 7 1 -phenylamino-Ppheide a 8).
DPOR from P. marinus efficiently converted Pchlide b (substrate 6) into Chlide b, with a specific activity of 80% compared with substrate 1. It also reduced its zinc derivative zinc Ppheide b 7 to zinc Pheide b with a specific activity of 90% relative to substrate 2. As expected, both compounds were bound by The specific activity of P. marinus DPOR for Pchlide a (910 pmol min Ϫ1 mg Ϫ1 ) was set as 100% for all tested magnesium tetrapyrroles. Zinc Ppheide a gave a specific activity of 730 pmol min Ϫ1 mg Ϫ1 and was set as 100% for all tested zinc derivatives, respectively. Absorption maxima of the Qy band for each tetrapyrrole are shown. In the case of substrate utilization, the absorption maxima of the reaction product and extinction coefficients of the individual compounds are shown. The values in parentheses were estimated according to Ref. 49 (ChlNB) 2 . By contrast, compound 8 with a more bulky phenylamino group in C7 1 position was not a substrate. This indicates that also at this position steric hindrance is more important than polarity for binding and catalytic conversion. This conclusion is confirmed by inhibition experiments in which compound 8 failed to function as a competitive inhibitor for DPOR catalysis. Furthermore, compound 8 did not show detectable affinity for the (ChlNB) 2 complex in the substrate binding assay.

P. marinus DPOR Accepts 8-Ethyl and 8-Vinyl
Substrates-To test selectivity with regard to the C8 substituent, we made use of zinc [8-vinyl]-Ppheide a (substrate 9), which is the corresponding zinc derivative of [8-vinyl]-Pchlide a. Heterologously produced P. marinus DPOR was able to efficiently convert substrate 9; the specific activity was 80% of that for substrate 2, and the affinity (K m ϭ 6.7 M) was even higher compared with substrate 2. From these results a "flexible" binding of the ethyl/vinyl group in the C8 position was concluded.
Precursor Molecules of Pchlide a Biosynthesis Are Not Substrates for DPOR-Biosynthesis of Pchlide a includes the formation of magnesium protoporphyrin IX 10 and magnesium protoporphyrin IX monomethyl ester 11. Both lack the isocyclic ring E (Fig. 4B) but contain otherwise the same conjugation system and substituents as Pchlide a, thereby allowing for the analysis of the potential involvement of ring E in substrate recognition. Neither of the two tested substrates lacking this feature allowed for DPOR activity (Table 1), indicating clearly that the reduction of the C17-C18 bond can only take place after the formation of the isocyclic ring in the biosynthetic pathway.
Zinc Protopheophorbides with Altered Substituents on Ring E-To further specify substrate recognition of ring E by P. marinus DPOR, a series of four substrates modified at ring E were analyzed. First the methoxycarbonyl substituent at C13 2 was either removed (zinc pyro-Ppheide a 12) or substituted by a hydroxy group (zinc 13 2 OHpyro-Ppheide a 13). Next zinc 13 2 OMe-Ppheide a was tested. It contains the polar, larger methoxy group instead of the hydrogen at position C13 2 . This substitution stabilizes the stereochemistry at C13 2 ; therefore, both enantiomers were tested, with C13 2 S (substrate 14) and C13 2 R (substrate 15) configurations (Fig. 4B).
Even though the C13 2 and hydroxy residue of compound 13 are distinctly smaller than the respective C13 2 methoxycarbonyl group of zinc Ppheide a 2, DPOR reduced compound 13 with the same specific activity as zinc Ppheide a 2. Even zinc pyro-Ppheide a 12 carrying no side chain on C13 2 still resulted  OCTOBER 31, 2008 • VOLUME 283 • NUMBER 44 in 30% of DPOR activity, indicating that the intact methyl ester group is not a prerequisite for substrate recognition. However, activity is lost with zinc Ppheide derivatives carrying two C13 2 side chains, a methoxycarbonyl as well as a methoxy group. In this case, neither the C13 2 S (substrate 14) nor the C13 2 R (substrate 15) derivatives were accepted as substrates of DPOR. From these data we concluded that only the integrity of the carbon skeleton of ring E is essential, whereas the substituents on position C13 2 are not determinants for substrate recognition.

Substrate Recognition of Protochlorophyllide Oxidoreductase
Taken together, these findings clearly demonstrate that substrate recognition of DPOR differs considerably from the LPOR system; the latter does not accept compounds 12 and 13 as substrates (47).
Protochlorophylls and Chlorophylls with Altered Substituents on Ring D-Finally, a set of derivatives with modifications on pyrrole ring D were tested as DPOR substrates. Protochlorophylls a 16 and b 17 already contain the C 20 H 39 phytol ester at the propionate side chain at position C17, whereas the C17-C18 double bond is not reduced. Chlorophylls c1 18 and c2 19 contain an acrylic side chain with a C17 1 -C17 2 double bond instead (Fig. 4C). None of these D ring derivatives allowed for detectable DPOR activity. For the two phytyl esters, this may relate to the steric hindrance of the phytyl group, to its reduced polarity, or to both. Electronic effects are most likely responsible for the inactivity of DPOR with compounds 18 and 19. The additional C17 1 -C17 2 double bond is conjugated to the C17-C18 double bond that is reduced during Chlide formation. In agreement with other substrates, inactivity of DPOR seems to be related again to substrate recognition and binding, because the ring D derivatives 16 -19 were neither competitive inhibitors of DPOR catalysis, nor did they bind to the purified (ChlNB) 2 complex. By contrast chlorophyll c1 was shown to be a competitive inhibitor of LPOR catalysis (48). From these results we conclude a specific interaction with the propionate side chain at ring D during binding to the active site.
Overall Substrate Recognition: A Proposal-From the presented data obtained from the analysis of 19 substrate analogs, a model can be deduced for the overall substrate recognition by DPOR from P. marinus. First, two molecules of Pchlide a are bound by the catalytic (ChlNB) 2 complex. During this process, several parts of the Pchlide a molecule are recognized with high specificity. Binding of ring A allowed polarity modifications of the C3 substituent (vinyl or formyl), but bulky ligands were not accepted. The same is true for ring B substituents. Polarity changes (C7-methyl or -formyl, C8-ethyl or -vinyl) were tolerated, whereas larger groups at C7 prevented substrate binding. Furthermore, the overall integrity of the isocyclic ring E is necessary for substrate recognition, whereas individual substituents of ring E are of minor importance as long as only one bulky group is bound at C13 2 . A methoxycarbonyl group on C13 2 may be removed or altered without detracting recognition of the substrate. However, substrate analogs devoid of ring E or carrying two substituents on C13 2 were not used as substrates. The least changes were tolerated at the C17 side chain of ring D next to the C17-C18 double bond, which is reduced during DPOR catalysis. Consequently, any modifications on the C17 propionate chain prevented substrate recognition and utilization.
Each of the tested substrate analogs preventing DPOR activity was neither a competitive inhibitor of catalysis nor found to bind to the (ChlNB) 2 complex. These findings suggest steric effects preventing those derivatives from entering the active site. Because all pyrrole rings are involved in substrate recognition, one might propose that the Pchlide molecule is buried in a active site cavity.
Variable Routes of Chlorophyll Biosynthesis Mediated by DPOR Activity-The 8-vinyl variant of Pchlide a has been shown to be a biosynthetic intermediate when chlorophylls and bacteriochlorophylls are synthesized in organisms containing LPOR. For LPOR it was determined that Pchlide a 1 and [8-vinyl]-Pchlide a equally function as substrates (49,50), indicating that the single divinyl-reductase enzyme is able to also convert [8-vinyl]-Pchlide a as well as [8-vinyl]-Chlide a (26, 51) (Fig.  5A). Investigation of a C. tepidum mutant lacking a 8-vinyl reductase and producing [8-vinyl]-bacteriochlorophylls demonstrated that [8-vinyl]-Pchlide is also a substrate for DPOR (27).
Even though no gene is found in the genome of P. marinus that encodes an enzyme for the reduction of the 8-vinyl group (52), we found that P. marinus DPOR utilizes Pchlide a 1 and zinc Ppheide a 2 as well as zinc [8-vinyl]-Ppheide a 9 as substrates. These results suggest that in organisms containing 8-ethyl chlorophylls, an 8-vinyl reduction step during chlorophyll synthesis in the dark may likely occur with [8-vinyl]-Pchlide a or [8-vinyl]-Chlide a as a substrate (26) (Fig. 5A).
Chlorophyll b is a naturally occurring pigment containing a C7 formyl instead of a C7 methyl group that is found in chlorophyll a. The oxygen at C7 was shown to be introduced into the C7 methyl group of Chlide a by a CAO (53)(54)(55). Although the CAO from Chlamydomonas reinhardtii was described to convert chlorophyll a to chlorophyll b (56), no oxygenation of Pchlide a to Pchlide b by CAO was shown to date. The direct function of Pchlide b as a light harvesting pigment in plants is controversially discussed in the literature (57)(58)(59). However, the adjustment of the chlorophyll a/b ratio in the "chlorophyll cycle" has been described as an important mechanism for the adaptation to various light conditions (60). With respect to this it is a relevant observation that P. marinus DPOR utilizes Pchlide a (zinc Ppheide a) as well as Pchlide b 6 and its corresponding Zn analog zinc Ppheide b 7. From these results one might conclude that Pchlide b is a natural precursor of b-type pigmentsinsomecyanobacteria.AbranchedpathwayviaPchlideb or via Pchlide a might then play an important role for the biosynthesis of chlorophyll b in P. marinus (Fig. 5B). When Pchlide b is a natural precursor, it remains open where the biosynthetic route divides into a-type and b-type chlorophyll precursors.