Broadened Substrate Specificity of 3-Hydroxyethyl Bacteriochlorophyllide a Dehydrogenase (BchC) Indicates a New Route for the Biosynthesis of Bacteriochlorophyll a*

Background: Bacteriochlorophyll biosynthesis is fundamental for the photosynthetic capture of solar energy by photosynthetic bacteria. Results: 3-Hydroxyethyl chlorophyllide a was identified as a novel substrate of 3-hydroxyethyl bacteriochlorophyllide a dehydrogenase (BchC). Conclusion: The broadened substrate specificity of BchC would allow for a novel branched pathway for bacteriochlorophyll a biosynthesis. Significance: The observed plasticity of the pathway might be relevant for the engineering of photosynthetic organisms. Bacteriochlorophyll a biosynthesis requires formation of a 3-hydroxyethyl group on pyrrole ring A that gets subsequently converted into a 3-acetyl group by 3-vinyl bacteriochlorophyllide a hydratase (BchF) followed by 3-hydroxyethyl bacteriochlorophyllide a dehydrogenase (BchC). Heterologous overproduction of Chlorobaculum tepidum BchF revealed an integral transmembrane protein that was efficiently isolated by detergent solubilization. Recombinant C. tepidum BchC was purified as a soluble protein-NAD+ complex. Substrate recognition of BchC was investigated using six artificial substrate molecules. Modification of the isocyclic E ring, omission of the central magnesium ion, zinc as an alternative metal ion, and a non-reduced B ring system were tolerated by BchC. According to this broadened in vitro activity, the chlorin 3-hydroxyethyl chlorophyllide a was newly identified as a natural substrate of BchC in a reconstituted pathway consisting of dark-operative protochlorophyllide oxidoreductase, BchF, and BchC. The established reaction sequence would allow for an additional new branching point for the synthesis of bacteriochlorophyll a. Biochemical and site-directed mutagenesis analyses revealed, in contrast to theoretical predictions, a zinc-independent BchC catalysis that requires NAD+ as a cofactor. Based on these results, we are designating a new medium-chain dehydrogenase/reductase family (MDR057 BchC) as theoretically proposed from a recent bioinformatics analysis.


Bacteriochlorophyll a biosynthesis requires formation of a 3-hydroxyethyl group on pyrrole ring A that gets subsequently converted into a 3-acetyl group by 3-vinyl bacteriochlorophyllide a hydratase (BchF) followed by 3-hydroxyethyl bacteriochlorophyllide a dehydrogenase (BchC). Heterologous overproduction of Chlorobaculum tepidum
BchF revealed an integral transmembrane protein that was efficiently isolated by detergent solubilization. Recombinant C. tepidum BchC was purified as a soluble protein-NAD ؉ complex. Substrate recognition of BchC was investigated using six artificial substrate molecules. Modification of the isocyclic E ring, omission of the central magnesium ion, zinc as an alternative metal ion, and a non-reduced B ring system were tolerated by BchC. According to this broadened in vitro activity, the chlorin 3-hydroxyethyl chlorophyllide a was newly identified as a natural substrate of BchC in a reconstituted pathway consisting of dark-operative protochlorophyllide oxidoreductase, BchF, and BchC. The established reaction sequence would allow for an additional new branching point for the synthesis of bacteriochlorophyll a. Biochemical and site-directed mutagenesis analyses revealed, in contrast to theoretical predictions, a zinc-independent BchC catalysis that requires NAD ؉ as a cofactor. Based on these results, we are designating a new medium-chain dehydrogenase/reductase family (MDR057 BchC) as theoretically proposed from a recent bioinformatics analysis.
The capturing of the global energy demand vitally depends on the biosynthesis of chlorophylls (Chls) 2 and bacteriochloro-phylls (BChls). These tetrapyrrole pigments are essential for light capturing, the transfer of the obtained excitation energy, and the primary charge separation process of photosynthesis (1). The light absorption and redox properties of BChls and Chls are modulated by the peripheral substituents of the macrocycle. However, the most important diversification concerns the aromatic system (Fig. 1B): the chlorin ring structure characteristic for Chls and e.g. BChl c (reduced D ring at C17/C18; max, Qy-band Ϸ 650 -700 nm) and the bacteriochlorin ring structure specific for BChls (additionally reduced B ring at C7/C8; max, Qy-band Ϸ 770 -800 nm) (2).
The biosynthesis of the core ring structure of Chls and BChls starts with the ATP-dependent magnesium insertion into the fully conjugated protoporphyrin IX molecule. The following two reaction steps are then responsible for the formation of the fifth, isocyclic ring E. The propionate side chain at ring C is methylated, and the resulting magnesium protoporphyrin IX monomethyl ester is subsequently converted into 8-vinyl protochlorophyllide in a sophisticated oxidation/cyclization reaction. Then the reduction of the peripheral vinyl group at C8 results in the formation of protochlorophyllide a (Pchlide; compound 11; see Subsequently, reduction of the C17/C18 double bond of ring D leads to the formation of the chlorin ring structure of chlorophyllide a (Chlide; compound 12). Two unrelated enzymes have developed for the regio-and stereospecific reduction of the fully conjugated Pchlide ring system: light-dependent protochlorophyllide oxidoreductase performs a light-driven, NADPH-dependent reduction, whereas the ATP-dependent multisubunit enzyme dark-operative protochlorophyllide a oxidoreductase (DPOR) uses nitrogenase-like biochemistry (3)(4)(5)(6).
At this point, BChl a and Chl a biosynthesis pathways diverge. Chlide is the immediate precursor of Chl a, which is formed via esterification of a polyisoprene alcohol to the C17 propionic acid moiety (7).
The synthesis of BChl a requires three additional enzymatic steps involving the following enzymes: (i) chlorophyllide a oxidoreductase (COR), the second enzyme of the pathway, that performs ATP-dependent nitrogenase-like catalysis (8,9); (ii) 3-vinyl bacteriochlorophyllide a hydratase (BchF), which is responsible for the hydroxylation of the 3-vinyl group; and (iii) 3-hydroxyethyl bacteriochlorophyllide a dehydrogenase (BchC), which catalyzes the subsequent oxidation of the 3-hydroxyethyl group into a 3-acetyl moiety.
Pioneering work for the elucidation of the bchF and bchC gene functions was the spectroscopic characterization and identification of accumulated BChl a precursor molecules in the respective mutants of different Rhodobacter strains (10 -15). Whole cell isotope labeling experiments revealed that the oxygen atom at C3 1 was derived from a water molecule using the purple bacteria Rhodobacter sphaeroides and Roseobacter denitrificans (16). Based on the spectroscopic analysis of the Rhodobacter capsulatus bchF mutant strains MB1003, KZR8G9, and CB1200 (10,11,17,18) and the bchC mutants R. capsulatus CW100 (13) and R. sphaeroides T127 and TB34 (12,19), the following two enzyme order variants of the BChl a biosynthetic pathway were concluded (compare also Fig. 5).
Reaction Sequence II Accordingly, the metabolite Chlide is located at the branching point of the pathway, implying extended substrate specificity for the enzymes COR and BchF.
Plasticity of the late steps of BChl/Chl biosynthesis was also proposed with respect to the 8-vinyl reduction step and for the formation of the 7-formyl group of Chl b on the basis of in vitro experiments using chemically modified substrate variants. An 8-vinyl derivative (instead of the 8-ethyl substrate) and the 7-formyl analog (instead of the 7-methyl substrate) were efficiently converted by DPOR (20).
The green sulfur bacterium Chlorobaculum tepidum is a strict anaerobe and an obligate photoautotroph. It is a well established model organism for the investigation of anoxygenic photosynthesis, in particular under conditions of very low light intensity (21)(22)(23). The main photosynthetic pigments of C. tepidum are BChl c, Chl a, and BChl a. The term "BChl c" names a group of pigments differing from other BChls and Chls: BChl c molecules contain a chiral hydroxylated center at C3 1 , lack a 13 2 -methylcarboxylate group, and possess a characteristic C20 methyl group together with a varying number of additional methyl groups at C8 2 and C12 1 (for a review, see Ref. 24). Despite the established designation as a BChl, all BChl c pigments possess a chlorin instead of a bacteriochlorin ring system (Fig. 1B). BChl c molecules are the major light-harvesting pigments that are present in large antenna structures called chlorosomes where they selfassemble into tubular nanostructures that are stabilized by interpigment magnesium ligation andand H-bonding interactions (25,26).
Chl a acts as the primary electron acceptor in the photochemical reaction centers of C. tepidum. BChl a also plays a fundamental role in the reaction center, and it is also found in antenna proteins of the chlorosome envelope (for a review, see Ref. 27).
With respect to the late steps of BChl a biosynthesis, only the sophisticated enzymology of the multisubunit COR enzyme has been well characterized (8, 9, 28 -31). Besides this, the bchF gene from C. tepidum has been shown to complement a bchFdeficient R. capsulatus mutant (32). No further biochemical characterization of any BchF or BchC enzyme is available from the literature to date.
In the present investigation, we made use of recombinant DPOR, BchF, and BchC proteins to analyze the BChl a biosynthesis pathway of C. tepidum. A series of artificial BchC substrates allowed for the characterization of the substrate specificity of BchC. These results, in combination with a DPOR-BchF-BchC pathway reconstitution assay, would allow for a novel branched biosynthetic route via 3-hydroxyethyl chlorophyllide a (3 1 -OH-Chlide; compound 12) as a BchC substrate. In addition, the cofactor specificity and metal requirement of BchC were characterized.
Production of a Cell-free Extract Containing C. tepidum DPOR-Production of C. tepidum DPOR subunits N, B, and L was mainly performed as described earlier (plasmid pGEX-bchNBL (5)) using 500 ml of LB medium containing 1 mM L-cysteine and 1 mM Fe(III) citrate to improve iron-sulfur cluster formation. All remaining steps were performed under anoxic conditions at an oxygen partial pressure below 1 ppm (oxygen detector, Coy Laboratories, Grass Lake, MI). Cells were harvested and resuspended in 10 ml of anoxic buffer 3 (100 mM HEPES-NaOH, 150 mM NaCl, 10 mM MgCl 2 , pH 7.5). The cell-free extract containing all three DPOR subunits was obtained by a single French press passage (16,000 p.s.i.) and subsequent ultracentrifugation in sealed anoxic tubes (4°C, 65 min, 130,000 ϫ g). The supernatant was stored as 500-l aliquots in anoxic bottles at Ϫ20°C for up to 3 months.
Production of a C. tepidum BchF-and BchC-containing Extract-The resuspended pellet from 500 ml of an E. coli culture containing pACYC-FC was passed through a French press (16,000 p.s.i.), and the crude cellular extract was stored on ice.
Solubilization of BchF-The resuspended pellet from 500 ml of E. coli culture overproducing His 6 -BchF (pACYC-F) was disrupted by a single French press passage (16,000 p.s.i.). For the BchF activity assays, this crude cellular extract was stored on ice for up to 6 h. His 6 -BchF was solubilized from the host membrane fraction. For this purpose, the crude cellular extract was subjected to low speed centrifugation (4°C, 60 min, 4,000 ϫ g). The resulting supernatant was then subjected to ultracentrifugation (4°C, 60 min, 146,000 ϫ g). The obtained supernatant was discarded, and the membrane pellet was washed twice with 3 ml of ice-cold buffer 4 (50 mM Tris-HCl, 20 mM MgCl 2 , pH 8.0). Subsequently, membrane proteins were solubilized from the membrane fraction by stirring at 4°C for 2 h in 7 ml of solubilization buffer (buffer 4 containing 0.6% (v/v) Triton TM X-100). After another ultracentrifugation step (4°C, 60 min, 146,000 ϫ g), the supernatant containing the solubilized membrane proteins was stored on ice for subsequent activity measurements. Integrity of the His 6 -BchF protein fraction was confirmed by N-terminal sequencing.
Purification of C. tepidum BchC-The resuspended pellet from 1 liter of E. coli cell culture overproducing GST-BchC (from plasmid pGEX-C) was thawed on ice for 12 h, 2 l Benzonase (Merck Millipore) was added, and the cells were disrupted by a single passage through a French press at 16,000 p.s.i. Cell debris was removed by ultracentrifugation (4°C, 65 min, 130,000 ϫ g), and the supernatant was applied to 1 ml of Protino glutathione-agarose 4B (Macherey-Nagel) equilibrated with buffer 2 at room temperature. The column flow-through was applied to the resin twice to increase the yield of the affinity chromatography. After washing with 10 ml of buffer 2, the loaded column was incubated with 60 units of PreScission protease (GE Healthcare) for 90 min at 22°C under slight agitation. The liberated BchC protein was eluted with 3 ml of buffer 2 and stored at 4°C.
Determination of Protein Concentration-The concentration of purified protein fractions was determined using the Bradford reagent (Sigma-Aldrich) according to the manufacturer's instructions with bovine serum albumin dissolved in buffer 2 as a standard.
N-terminal Amino Acid Sequence Determination-Automated Edman degradation was used to confirm the identity of the solubilized BchF and the purified BchC protein.
Chelator Treatment of BchC-Purified BchC samples were incubated in the presence of EDTA or EGTA at concentrations of 10 or 100 mM (75 min at 48°C). After a centrifugation step (10 min, 12,100 ϫ g, room temperature), protein solutions were buffer-exchanged to buffer 2 using an "illustra" NAP-5 column (GE Healthcare) according to the manufacturer's instructions. Protein samples were concentrated using an Amicon Ultra-0.5 centrifugal filter device (10,000 nominal molecular weight limit; Merck Millipore) and subjected to BchC activity assays.
Artificial Substrates of BchC-Compounds 1-6 ( Fig. 1, compare B and C) were dissolved in DMSO. Their respective syntheses and/or isolations are cited in Fig. 1C.
BchC Activity Assays in the Presence of Artificial Substrates-Qualitative in vitro activity assays containing 3.2-8.4 M BchC were performed in the presence of 1 mM NAD ϩ (or alternatively 1 mM NADH) and 750 nM-23 M substrates 1-6 in a total volume of 250 l of buffer 1. Reactions were incubated for 30 min at 34°C in the dark and subsequently stopped by addition of 500 l of acetone. Insoluble assay components were sedimented (two times, 17,000 ϫ g, 10 min, 4°C), and the supernatant was analyzed by UV-visible light absorption spectroscopy as described elsewhere (9).
The specific activity of BchC in the presence of compounds 1-3, 5, and 6 was determined in the linear range of the activity assay (100 nM-1 M purified enzyme and 8 M substrate) in the presence of 1 mM NAD ϩ or NADH in a volume of 550 l. Compounds 1 and 5 were epimeric 3 1 R/S mixtures; hence an effective substrate concentration of ϳ4 M was implied (60). At selected time points, 100-l samples were taken and mixed with 400 l of acetone. Samples were analyzed as described above. The following extinction coefficients were used to calculate specific activities: Zn-3-acetyl-Pheide a, ⑀ ϭ 65.2 mM Ϫ1 cm Ϫ1 (33); Zn-3-acetyl-13 2 -OH-Pheide a, ⑀ ϭ 65.2 mM Ϫ1 cm Ϫ1 (33); BPheide a, ⑀ ϭ 67.5 mM Ϫ1 cm Ϫ1 (34); Zn-BPheide a, ⑀ ϭ 67.7 mM Ϫ1 cm Ϫ1 (34). It is worth noting that esterified pigments (at C17 3 ) show an extinction coefficient of the Q Y band identical to that observed for the non-esterified derivatives. Chelator-treated BchC samples (3.2 M) were assayed in the presence of compound 1 (10 M) or 5 (2.6 M) for 30 min.
Single Turnover BchC Experiments-The purified BchC-NAD ϩ complex was analyzed in 250 l of buffer 1 in the presence of 2.9 M substrate 5. A negative control (without protein) and experiments with increasing concentrations of BchC (0.8, 2.0, 4.0, 6.0, and 8.0 M) were incubated in the absence of external cofactors for 3 h at 34°C in the dark. A control assay was analogously performed in the presence of 4 M BchC and 1 mM NAD ϩ .
Isolation of Chlide-Chlide was isolated from R. capsulatus strain CB1200 (11) as outlined before using RCV 2/3 PY medium containing 12 mg/liter FeSO 4 under dimmed light conditions (37). The pigment was dissolved in DMSO and stored at Ϫ20°C.
Reconstitution of the DPOR-BchF-BchC Pathway-Coupled enzymatic assays were initiated in an anoxic chamber (Coy Laboratory Products Inc.). DPOR activity from 40 l of the cell-free E. coli extract (preparation described above) was analyzed in the presence of 5 M Pchlide, 0.72 mM dithionite as an artificial electron donor, 5 mM dithiothreitol, 2 mM ATP, and an ATPregenerating system (20 mM creatine phosphate and 20 units of creatine phosphokinase) in 250 l of buffer 3 (34°C, 600 rpm, 60 min, in the dark). Coupled DPOR-BchF assays were subsequently completed (under oxic conditions) by the addition of 100 l of a crude cellular BchF extract or alternatively 100 l of a purified, detergent-solubilized BchF sample (as described above). Coupled DPOR-BchF-BchC experiments were either supplemented with 1.6 -3.9 M purified BchC, or alternatively, 100 l of a crude cellular extract from a BchF/BchC-overproducing E. coli strain were used. As cofactor, 500 M NAD ϩ (or alternatively NADP ϩ ) were used. For all experiments, buffer 1 was added to a final volume of 500 l, and samples were incubated for another 30 min at 34°C. Assays were stopped by the addition of 1 ml of acetone, and the pigment composition was analyzed as described above.
HPLC Analysis of Intermediates of the Reconstituted Pathway-HPLC analyses in combination with multiwavelength fluorescence detection were used to further characterize the intermediates of the reconstituted pathway. Samples from all pathway reconstitution experiments were analyzed on a Jasco HPLC system equipped with an FP-1520 multiwavelength fluorescence detector using an UltraSep ES RP18 column (SEPSERV) according to a modified method described before (38). Samples of 20 l were separated at a flow rate of 0.5 ml/min at 30°C using a linear gradient of 25-100% (v/v) acetonitrile in H 2 O containing 0.005% (v/v) triethylamine over 54 min.
Metal Determination-The zinc content of the recombinant purified BchC protein was determined by inductively coupled plasma mass spectrometry (Currenta GmbH and Co. OHG, Leverkusen, Germany). In a parallel affinity purification, a metal-independent GST-protein (alanyl-phosphatidylglycerol synthase (40)) was purified and analogously analyzed by mass spectrometry to assess the background level of the experimental approach used.

C. tepidum BchF Is an Integral Membrane
Protein-In the current literature, a 3-vinyl bacteriochlorophyllide a hydratase activity is ascribed to the BchF protein (ORF CT1421) due to the observation of substrate accumulation in the respective R. capsulatus mutant strains (10,11). Bioinformatics analysis of the BchF amino acid sequence using the TMpred program (41) revealed four putative membrane-spanning helices comprising the transmembrane segments Ile 23 -Tyr 42 , Thr 56 -Glu 77 , Phe 94 -Gln 113 , and His 118 -Gln 140 according to C. tepidum numbering. The recombinant His 6 -BchF protein was solubilized from the total membrane fraction (Fig. 1A,  left). Bacterial cells were disrupted (Fig. 1A, lane 2), and residual unbroken cells (and potential inclusion bodies) were removed by low speed centrifugation. The supernatant was then subjected to high speed centrifugation (supernatant; lane 3), and the sedimented (membrane) fraction was solubilized. After another high speed centrifugation, the final purification step revealed a single dominant BchF protein band (lane 4, arrow) that was confirmed by N-terminal amino acid sequencing. The SDS-PAGE mobility of BchF (M r ϳ 25,000) significantly differed from the calculated molecular mass of 20,008 Da as observed for other transmembrane proteins (42).
Purification of the Soluble BchC Protein from C. tepidum-The subsequent enzyme of the pathway has been annotated as BchC (ORF CT1422) due to substrate accumulation in mutants of different Rhodobacter species (11)(12)(13)(14)(15). The recombinant GST-BchC fusion protein was localized in the soluble cytosolic fraction of the E. coli host used (Fig. 1A, right, lane 6). The GST-tagged protein was purified by affinity chromatography (lanes 7-9), and subsequently the native BchC protein was liberated by on-column PreScission protease treatment (lanes 10 -12). Overall, 3 mg of an almost homogeneous protein were obtained from 1 liter of cell culture. The integrity of this protein with a relative molecular weight of 37,000 (calculated molecular mass, 37,068 Da) was confirmed by N-terminal amino acid sequencing. The purified BchC was stable for several weeks when stored at 4°C.
Artificial Substrates of BchC-Synthetic modified substrate molecules were tested to identify important determinants for substrate recognition (20,33,43,44). Six chemically synthesized BchC substrates (containing various modifications at rings A, B, and E as well as different central metal ions) were analyzed in in vitro experiments using purified BchC. All pigments were characterized on the basis of the respective Q Y absorption band (red-most absorption maximum; compare cited literature of individual compounds listed in Fig. 1C). Specific activities for all substrates were determined in the linear range of the enzyme assay by measuring specific substrate depletion or product formation.
In the present literature, the bacteriochlorin 3 1 -OH-BChlide (16) is described as the natural substrate of BchC (this substrate has a chiral center at C3 1 , but stereochemical aspects have not been considered in the present investigation). We first tested the demetallated equivalent of this substrate for enzymatic activity. As shown in Fig. 2A (bottom left), the NAD ϩ -dependent oxidation of 3 1 -OH-BPheide 3 (1) resulted in a bathochromic absorption shift of 36 nm that gave rise to a new Q Y band at 753 nm, which is indicative of the formation of BPheide a. A specific activity of 1.1 nmol min Ϫ1 mg Ϫ1 was observed. This result indicated that the presence of the central magnesium ion was not a prerequisite for BchC substrate recognition.
Consequently, compounds with an altered central metal ion (e.g. zinc or palladium) were also considered as artificial substrates of BchC. Compounds Zn-BPheide a (2) and Pd-BPheide a (3), both carrying a 3-acetyl group, were subjected to in vitro activity assays in the presence of 1 mM NADH to assess the reverse BchC reaction. For Zn-BPheide a, a hypsochromic absorption shift of 56 nm and a specific activity of 64.9 nmol min Ϫ1 mg Ϫ1 were observed due to the formation of Zn-3 1 -OH-BPheide a (Fig. 2A, bottom right). By contrast, experiments in the presence of Pd-BPheide a (3) did not reveal BchC activity above the detection limit of the assay used (0.4 nmol min Ϫ1 mg Ϫ1 ). Obviously, the presence of the Pd 2ϩ central metal ion

. Localization and purification of BchF and BchC analyzed by SDS-PAGE and spectral characteristics of artificial BchC substrates and related products.
A, left, the heterologously produced BchF membrane protein was isolated from the total E. coli membrane fraction by Triton X-100 solubilization. Lane 1, molecular mass markers (relative molecular mass (ϫ1,000) are indicated); lane 2, crude cellular extract; lane 3, supernatant after first high speed ultracentrifugation step; lane 4, supernatant after Triton X-100 solubilization and subsequent high speed ultracentrifugation containing the solubilized BchF protein with an apparent molecular mass of 25 kDa (arrow). A, right, affinity purification of BchC. GST-tagged BchC was immobilized using Protino glutathioneagarose 4B, and the target protein BchC was liberated by on-column PreScission protease treatment. Lane 5, molecular mass markers; lane 6, cell-free E. coli extract; lanes 7 and 8, flow-through of the GST affinity chromatography; lane 9, wash fraction; lanes 10 -12, elution fractions containing untagged BchC. B, core structures of porphyrin (P), chlorin (C), and bacteriochlorin (B) pigments. Ring systems A-E are indicated. Substituents R 1 -R 4 and the chelated central atom are indicated in C. C, structural and spectroscopic details of artificial substrates (left) and the resulting BchC reaction products (right). Substrates 2, 3, and 6 were used to analyze the BchC-catalyzed back-reaction in the presence of NADH. Efficient substrate utilization for compounds 1, 2, 5, and 6 was determined. a, mixture containing R and S stereoisomers; b, this study, measured in acetone, 50 mM Tris-HCl, pH 8.0 (volume ratio, 2:1); c, in diethyl ether; d, in acetone; n.a., not available. Note that the 3-(1-hydroxy)-ethyl substituent is often referred to as 3-hydroxyethyl or 3 1 -OH in the present literature. -, no product formation.
hampers the BchC catalyzed (back) reaction. When 3 1 -OH-BChl a (4) was analyzed in the presence of NAD ϩ , no change in the spectral properties was observed. It was concluded that the presence of a bulky hydrophobic ester side chain on ring D abolished enzymatic substrate oxidation.
The conjugated system of bacteriochlorins (Fig. 1, B and C, compounds 1-4) is mainly responsible for the absorption properties of BChl a and its derivatives. The subsequent BchC activity assays make use of chlorins showing Q Y bands at ϳ100nm-shorter wavelengths (compounds 5 and 6; Fig. 1, B and C). Until now, no biosynthetic chlorin intermediates have been described as a substrate of BchC (Fig. 5).
However, when a BchC in vitro activity assay was performed in the presence of Zn-3 1 -13 2 -di-OH-Pheide a 3 (5) and 1 mM NAD ϩ , a bathochromic absorption shift of 24 nm and a specific activity of 21.4 nmol min Ϫ1 mg Ϫ1 were observed ( Fig. 2A, top  left). Obviously, the chlorin ring structure of compound 5 did not impede BchC catalysis. Furthermore, neither the presence of an additional 13 2 -hydroxyl on ring E nor the central zinc ion hampered BchC catalysis. This result was substantiated by using the chlorin Zn-3-acetyl-Pheide a (6), which might represent a natural chlorin reaction product of BchC (with magnesium replaced by zinc). In this reverse BchC assay, Zn-3-acetyl-Pheide a (6) was reduced in the presence of NADH to Zn-3 1 -OH-Pheide a (10), showing an almost identical Q Y band at 653 nm as the above mentioned Zn-3 1 -13 2 -di-OH-Pheide a (Q Y band at 652 nm; see Fig. 2A, top right). A specific activity of 172.5 nmol min Ϫ1 mg Ϫ1 was determined.
In summary, the chlorin substrates Zn-3-acetyl-Pheide a (6) and Zn-3 1 -13 2 -di-OH-Pheide a (5) showed a significantly increased specific activity when compared with the related bacteriochlorin compounds Zn-BPheide a (2) and 3 1 -OH-BPheide a (1) of the present investigation. Based on these results, it was proposed that 3 1 -OH-Chlide (13) might also be a natural substrate of BchC. Such additional substrate utilization would reflect an alternative biosynthetic route for the synthesis of BChlide as depicted in the lower part of Fig. 5A. These results are further supported by a C. tepidum mutagenesis study. A ⌬bciC mutant impaired in the removal of the 13 2 -COOCH 3 substituent accumulated 3-acetylpheophorbide a together with bacteriopheophorbide a (plus their monomethyl derivatives; Ref. 45).
Reconstitution of the DPOR-BchF-BchC Pathway-To substantiate this alternative biosynthetic pathway further, coupled activity assays in the presence of the highly oxygen-labile DPOR enzyme were performed in an anoxic chamber. All pathway reconstitution experiments made use of an E. coli strain for the well established polycistronic overproduction of subunits BchN, BchB, and BchL of DPOR (5). The corresponding DPOR  Fig. 1C. B, determination of specific activities in the linear range of the enzyme assay. Abs, absorption. activity assay (containing an ATP-regenerating system and the DPOR substrate Pchlide; Q Y band at 628 nm) clearly revealed formation of Chlide as indicated by a new Q Y peak at 667 nm (Fig. 3A, dotted line).
This specific Q Y Chlide peak was no longer present in coupled DPOR-BchF assays using either a crude cellular extract from a BchF-overproducing E. coli strain (Fig. 3A, dashed line) or alternatively the solubilized BchF membrane protein (Fig.  3B, dashed line). A new absorption maximum at 662 nm was indicative for the formation of the proposed 3 1 -OH-Chlide (13) (Fig. 3A, dashed line; for references of Q Y absorption maxima see Table 1).
In extending this reaction pathway, DPOR-BchF-BchC reconstitution experiments were carried out using either crude cellular BchF and BchC extracts (Fig. 3A, solid line) or alternatively a BchF extract and the purified BchC protein (Fig. 3B,  solid line). Obtained results revealed a new Q Y band at 680 nm indicative of the formation of 3-acetyl-Chlide (14). In low temperature fluorescence experiments (at 77 K), this compound revealed a characteristic fluorescence at 694 nm upon excitation at 470 nm. These determined fluorescence parameters allowed for the specific detection of 3-acetyl-Chlide (14) in mixtures containing Pchlide, Chlide, and/or 3 1 -OH-Chlide (13).
The individual steps of the reconstituted pathway were further analyzed by reversed phase HPLC and multiwavelength fluorescence detection. DPOR catalysis resulted in Chlide formation ( ex ϭ 439 nm and em ϭ 670 nm; eluting at 21.8 min) as depicted in Fig. 4A. Subsequently, the DPOR-BchF-dependent formation of 3 1 -OH-Chlide (13) ( ex ϭ 439 nm and em ϭ 662 nm; double peak eluting at 10.0 and 10.8 min) is indicated in Fig.  4B. This metabolite was efficiently oxidized due to the presence of BchC and NAD ϩ , resulting in the formation of 3-acetyl-Chlide (14) ( ex ϭ 470 nm and em ϭ 694 nm; eluting at 14.4 min) as indicated in Fig. 4C. Identical results were observed in experiments using crude cellular BchF extract, solubilized His 6 -BchF, or purified BchC (Fig. 3B).
Based on the in vitro assays in the presence of artificial substrates and the series of analyses from the pathway reconstitution experiments, we conclude that BChl a biosynthesis in C. tepidum might include an additional branching point that is based on the BchC-catalyzed conversion of 3 1 -OH-Chlide. This route is implemented in the reaction scheme depicted in Fig. 5  (7-14, 19, 35, 61-63).
Bioinformatics Analysis of BchC-Theoretical analyses related C. tepidum BchC to the medium-chain dehydrogenase/ reductase (MDR) protein superfamily (46 -49) comprising an N-terminal GroES-like domain fused to a C-terminal NAD(P) ϩ binding domain. MDRs can be mainly classified as zinc-containing or non-zinc-containing proteins using either NAD(H) or NADP(H) for catalyses (47).
Highest amino acid sequence identity values were observed for the comparison of BchC (50) with the well characterized enzymes sorbitol dehydrogenase (21%; Protein Data Bank code 1E3J) and threonine dehydrogenase (30%; Protein Data Bank code 3GFB). These orthologous MDRs perform zinc-and NAD(H)/NADP(H)-dependent catalysis.
However, manual inspection of the BchC/MDR sequence alignment depicted in Fig. 6 (51, 52) did not reveal a coherent ligand pattern as expected for a classical zinc-dependent oxidoreductase. Accordingly, BchC was subjected to site-directed mutagenesis to identify key catalytic amino acid residues.
Activity Analysis of BchC Mutant Proteins-Highly conserved amino acid residues of BchC were deduced from an extended sequence alignment of BchC sequences (compare conservation pattern implemented in the BchC/MDR alignment shown in Fig. 6). A potential role of residues Ser 39 , Ser 42 , Glu 46 , Tyr 67 , Glu 68 , and His 141 as alternative zinc ligands was hypothesized. Accordingly, these residues were individually replaced by alanine residues, and the respective mutant proteins were subjected to a coupled BchC activity assay. For mutant proteins S39A and S42A, activities of about 7 and 40% were observed compared with the wild-type protein, whereas mutants BchC E46A, Y67A, E68A, and H141A did not reveal any detectable BchC activity. These results indicated a key catalytic role for residues Glu 46 , Tyr 67 , Glu 68 , and His 141 .
Chelator Treatment of BchC-Samples of purified BchC were preincubated in the presence of up to 100 mM EDTA or EGTA (75 min at 48°C). Subsequently, coupled BchC activity experiments and assays in the presence of substrate 3 1 -OH-BPheide a 3 (1) or Zn-3 1 -13 2 -di-OH-Pheide a 3 (5) did not reveal reduced BchC activities when compared with the respective control experiment (not shown). These results did not reflect the outcome of related experiments for well characterized zinc-dependent MDR enzymes: chelation of the zinc ions of yeast alcohol dehydrogenase (53) and L-threonine dehydrogenases from Pyrococcus furiosus and Thermococcus kodakaraensis (54,55) resulted in enzymatic inactivation. Accordingly, a zinc-independent BchC catalysis was considered.
Zinc Content of Recombinant BchC-The zinc content of the purified BchC enzyme and of an analogously purified zinc-independent reference protein (40) was analyzed by inductively coupled plasma mass spectrometry (MS). No metal content above the background level of the experimental approach used was determined.
Inductively coupled plasma MS analyses in combination with the chelator treatment experiments used were indicative of a

Spectroscopic characterization of the reconstituted pathway intermediates
Pigments from the coupled DPOR-BchF-BchC reconstitution assay shown in Fig. 3A were further characterized by absorption spectroscopy and low temperature fluorescence spectroscopy (in acetone, 50 mM Tris-HCl, pH 8.0; volume ratio, 2:1). Newly determined fluorescence parameters allow for the specific detection of 3-acetyl-Chlide (12) in mixtures containing Pchlide, Chlide, and/or 3 1 -OH-Chlide (13  zinc-independent BchC catalysis. These results were not in agreement with the initial bioinformatics analysis. However, the observed zinc-independent catalysis of BchC clearly reflects the absence of highly conserved ligands of a catalytically relevant metal ion (cysteine, histidine, and glutamate; Fig. 6). With respect to these results, it was not possible to clearly assign BchC to one of the established MDR families described in the literature (for a review, see Refs. 46 and 56).
Cofactor Specificity of BchC-NADP ϩ is a cofactor that commonly acts as a stereospecific electron acceptor within the framework of anabolic pathways. Therefore, a coupled BchC activity assay was also performed in the presence of 500 M NADP ϩ . This experiment revealed a low degree of product formation when compared with the related NAD ϩ -containing experiment (absorption spectrum depicted in Fig. 3C, short  dashes). However, a related control experiment without externally added nucleotide cofactor indicated an almost identical absorption spectrum (Fig. 3C, long dashes, overlapping the spectrum of the NADP ϩ -containing experiment).
Single turnover in vitro experiments containing purified BchC in the presence of Zn-3 1 -13 2 -di-OH-Pheide a (without external addition of nucleotide cofactor) also revealed product formation (as judged from the absorption at 676 nm). This enzymatic activity was efficiently stimulated by addition of 1 mM NAD ϩ . No such effect was observed in the presence of 1 mM NADP ϩ .
In agreement with these findings, a tightly bound NAD ϩ cofactor of BchC was hypothesized. The presence of this co-purified nucleotide was subsequently confirmed by single turnover experiments with increasing BchC concentrations. In  of BchC revealed gradually increasing product concentrations (0.1, 0.3, 0.5, 0.6, and 0.8 M) as judged by absorption measurements at 676 nm (dotted lines). A control experiment with NAD ϩ excess is shown in Fig. 3D (continuous line). From these results, we conclude that the overproduced C. tepidum BchC is purified (partially) in complex with its specific cofactor NAD ϩ .

Discussion
Fundamentals of the BChl a biosynthetic pathway were elucidated more than 40 years ago based on the spectroscopic analysis of accumulated precursors from different Rhodobacter mutant strains (57,58). This methodology already revealed that the DPOR reaction product Chlide is located at a central hub of the following branched biosynthetic pathways. Recently, evidence for Reaction Sequence I was also obtained from a thorough COR characterization (9). The present study FIGURE 6. Amino acid sequence alignment of C. tepidum BchC with zinc-containing and zinc-independent MDRs. Cons_BchC shows the conservation pattern for an alignment of 19 BchC sequences (not shown) and has been manually added to the amino acid sequence alignment of orthologous MDRs. Zinc-containing MDRs include L-threonine dehydrogenase from Pyrococcus horikoshii (TDH_Ph), L-threonine dehydrogenase from Thermococcus kodakaraensis (TDH_Tk), L-arabinitol 4-dehydrogenase from Neurospora crassa (LArDH_Nc), sorbitol dehydrogenase from Homo sapiens (SDH_Hs), and sorbitol dehydrogenase from Bemisia argentifolii (SDH_Ba). Zinc-independent MDRs include leukotriene B 4 12-hydroxydehydrogenase/15oxoprostaglandin 13-reductase from Cavia porcellus (LTD_Cp), quinone oxidoreductase from E. coli (QOR_Ec), quinone oxidoreductase from Coxiella burnetii (QOR_Cb), and enone oxidoreductase from Fragaria vesca (EO_Fa). The alcohol dehydrogenase from Pyrobaculum aerophilum (ADH_Pa) possesses a structural but no catalytic zinc ion. Catalytic zinc-binding residues are highlighted in yellow. Residues coordinating a water molecule ligating the catalytic zinc ion are marked in pink. Ligands for a structural zinc ion are highlighted in orange. Black boxes indicate conserved residues among MDRs with a catalytic zinc ion (information derived from an alignment of 11 zinc-containing MDR sequences; not shown). Highlighted in gray is the nucleotide binding motif. Arrows indicate C. tepidum BchC residues that were mutagenized in the present study. Cons_all shows the sequence conservation of the overall alignment. Asterisks (*) indicate fully conserved residues, colons (:) mark residues with strongly similar properties, and residues with weakly similar properties are indicated by a period (.). clearly reveals the BchF-dependent synthesis of 3 1 -OH-Chlide via Reaction Sequence II in C. tepidum. These results might indicate that biosynthesis of BChl a is parallel in C. tepidum and e.g. R. capsulatus with respect to Reaction Sequences I and II.
In the present investigation, the analysis of chemically modified BchC substrates allowed for the determination of the following additional route for the synthesis of BChl a.

Reaction Sequence III
BchC from C. tepidum is able to efficiently oxidize the bacteriochlorin substrate 3 1 -OH-BPheide a and the chlorin Zn-3 1 -13 2 -di-OH-Pheide a. In our subsequent pathway reconstitution experiments, 3 1 -OH-Chlide was identified as a BchC substrate that is located at this newly identified branching point. These findings are in agreement with a C. tepidum mutant study (45) and are further supported by a recent investigation for the COR enzyme from the purple bacterium R. denitrificans. An artificial substrate carrying a 3-acetyl group (instead of a 3-hydroxyethyl or a 3-vinyl group) was efficiently converted by COR (9). These findings might indicate that Reaction Sequence III is also relevant for BChl a biosynthesis in purple bacteria. The observed plasticity of the BChl biosynthetic pathways might be the evolutionary clue for the broad diversification of pigments under different environmental light conditions. The present investigation clearly indicates that C. tepidum BchC is an unusual zinc-independent dehydrogenase specifically using NAD ϩ as a redox cofactor. These findings are strongly supported by a recent bioinformatics investigation that assigns BchC to a newly defined family of MDRs using an automated algorithm for the refinement of the hidden Markov models used (47). According to these findings, BchC from C. tepidum might be loosely related to the zinc-independent enone oxidoreductase from Fragaria ϫ ananassa (strawberry) that catalyzes the NADH-dependent reduction of a polar CϭC double bond (Protein Data Bank code 4IDA (59) and Fig. 6). Based on the present literature, it was not possible to conclude a common reaction mechanism for the different zinc-independent MDRs characterized so far. In this study, critical involvement of residues Glu 46 , Tyr 67 , Glu 68 , and His 141 in BchC catalysis was demonstrated. However, the comparison of BchC with members of different MDR families (depicted in Fig. 6) does not reveal the conservation of these key catalytic BchC residues. Therefore, a differing substrate ligation/activation pattern in the active site of BchC must be concluded.
Nevertheless, the present study reveals an initial picture of C. tepidum BchC substrate recognition. The enzyme tolerates artificial substrates containing zinc instead of the central magnesium and the absence of a central metal ion. Furthermore, the presence of an additional 13 2 -OH group on ring E is accepted. However, a bulky side chain attached to ring D clearly abolishes BchC activity. These findings might indicate that mainly pyrrole rings A and D are involved in substrate recognition. Hydrogenation of a chlorin ring B to a bacteriochlorin changes the spatial orientations of the C7 and C8 substituents. The observed BchC activity with both chlorins and bacteriochlorins (compounds 1 and 5) might be indicative for an active site not covering the B ring system of the natural substrates. Last but not least, the coupled assays developed in this work are a basis for elucidating the details of BchC catalysis including the stereoselectivity for the configuration of C3 1 .
Author Contributions-J. M., D. J., and H. S. designed the study. J. M., D. J., C. L., and H. S. wrote the paper. C. L., S. K., S. P., and S. V. characterized BchF and BchC. H. S. supplied the artificial substrates. All authors analyzed the results and approved the final version of the manuscript.