Chlorophyll a Synthesis in Rhodobacter sphaeroides by Chlorophyll Synthase of Nicotiana tabacum

Simple Summary Photosynthetic (PS) organisms utilize light energy via their PS apparatuses which contain PS pigments. Since a PS organism usually has limited wavelength regions of light absorption determined by the property of their pigments, introducing heterologous pigments can expand the light absorption regions, which could enhance the efficiency of light utilization. Rhodobacter sphaeroides is a bacteriochlorophyll a (BChl a)-containing PS bacterium which has been widely used in photosynthesis research and biotechnological applications. In this study, we attempted to introduce chlorophyll synthase (ChlG) into R. sphaeroides to synthesize chlorophyll a (Chl a) under BChl a formation. It was challenging because BChl a and its precursors can inhibit ChlG. However, ChlGs of angiosperm plants were found to possess resistance against these inhibitions. Thus, we selected ChlG of Nicotiana tabacum, which had the highest resistance among angiosperm ChlGs and achieved the formation of Chl a in the presence of BChl a in R. sphaeroides by expression of the enzyme. Abstract The production of phytylated chlorophyll a (Chl aP) in Rhodobacter sphaeroides, which uses phytylated bacteriochlorophyll a (BChl aP), is the first step in expanding the light absorption spectra. Unlike the chlorophyll synthase (ChlG) of the Synechocystis sp. PCC6803, ChlGs of angiosperms, including Arabidopsis thaliana, Nicotiana tabacum, Avena sativa, and Oryza sativa, showed bacteriochlorophyll synthase activity and resistance to inhibition by bacteriochlorophyllide a (BChlide a), geranylgeranylated BChl a (BChl aGG), and BChl aP, collectively called bacteriochlorins. Among the angiosperm ChlGs, N. tabacum ChlG had the highest bacteriochlorophyll synthase activity and resistance to inhibition by bacteriochlorins. Expression of N. tabacum chlG in R. sphaeroides resulted in the formation of free Chl aP in the presence of BChl aP during photoheterotrophic growth, even though reactive oxygen species were generated.


Introduction
Rhodobacter sphaeroides is a purple anoxygenic photosynthetic (PS) bacterium that synthesizes phytylated bacteriochlorophyll a (BChl a P ). Another group of oxygenic PS organisms, including cyanobacteria, algae, and plants, possess phytylated chlorophyll a (Chl a P ). Chlorophyll synthase (ChlG) and bacteriochlorophyll synthase (BchG) catalyze the prenylation (esterification) of chlorophyllide a (Chlide a) and bacteriochlorophyllide a (BChlide a), respectively, with the C 20 isoprenoid of the geranylgeranyl (GG) moiety to form geranylgeranylated Chl a (Chl a GG ) and BChl a (BChl a GG ) ( Figure 1). The GG groups of Chl a GG and BChl a GG are subsequently reduced by ChlP and BchP (geranylgeranyl reductases) to form Chl a P and BChl a P , respectively. Alternatively, phytyl pyrophosphate (PPP) is formed by the reduction in GG-pyrophosphate (GGPP) by ChlP and BchP, and PPP can be used as a substrate for ChlG and BchG [1,2] (Figure 1).

Figure 1.
Chl a and BChl a biosynthesis in R. sphaeroides. Chlide a is metabolized by BchF and BchXYZ, and subsequently by BchC to form BChlide a. BChl aGG is synthesized via the ping-pong mode by BchG using GGPP and BChlide a as the first and second substrates, respectively. The GG group of BChl aGG is then reduced by BchP to yield BChl aP. Alternatively, PPP is first formed from GGPP by BchP and subsequently used as a substrate by BchG to produce BChl aP. Similarly, Chl aGG/Chl aP are synthesized by ChlG/ChlP. ChlP activity can be substituted by BchP [8]. Generally, ChlG would be inhibited by BChlide a, BChl aGG, and BChl aP (bacteriochlorin) (red box). Similarly, BchG would be inhibited by Chlide a, Chl aGG, and Chl aP (collectively called chlorin). However, this study focused on the analysis of recombinant R. sphaeroides synthesizing more BChl a than Chl a; therefore, BchG inhibition by chlorin was negligible. Nicotiana tabacum ChlG (NtChlG), which is resistant to bacteriochlorin inhibition, was used to produce Chl aP together with BChl aP in R. sphaeroides. Carbon numbers and ring designations are illustrated in the BChlide a structure. Differences between Chlide a and BChlide a are denoted by shades in the Chlide a structure.
In this study, we aimed to produce Chl a in R. sphaeroides under BChl a-forming conditions via heterologous expression of chlG. Accordingly, ChlG must be resistant to BChl a, and its precursor BChlide a. ChlGs from angiosperms (Arabidopsis thaliana, Nicotiana tabacum, A. sativa, and Oryza sativa) and other PS organisms were analyzed to determine Figure 1. Chl a and BChl a biosynthesis in R. sphaeroides. Chlide a is metabolized by BchF and BchXYZ, and subsequently by BchC to form BChlide a. BChl a GG is synthesized via the ping-pong mode by BchG using GGPP and BChlide a as the first and second substrates, respectively. The GG group of BChl a GG is then reduced by BchP to yield BChl a P . Alternatively, PPP is first formed from GGPP by BchP and subsequently used as a substrate by BchG to produce BChl a P . Similarly, Chl a GG /Chl a P are synthesized by ChlG/ChlP. ChlP activity can be substituted by BchP [8]. Generally, ChlG would be inhibited by BChlide a, BChl a GG , and BChl a P (bacteriochlorin) (red box). Similarly, BchG would be inhibited by Chlide a, Chl a GG , and Chl a P (collectively called chlorin). However, this study focused on the analysis of recombinant R. sphaeroides synthesizing more BChl a than Chl a; therefore, BchG inhibition by chlorin was negligible. Nicotiana tabacum ChlG (NtChlG), which is resistant to bacteriochlorin inhibition, was used to produce Chl a P together with BChl a P in R. sphaeroides. Carbon numbers and ring designations are illustrated in the BChlide a structure. Differences between Chlide a and BChlide a are denoted by shades in the Chlide a structure.
In this study, we aimed to produce Chl a in R. sphaeroides under BChl a-forming conditions via heterologous expression of chlG. Accordingly, ChlG must be resistant to BChl a, and its precursor BChlide a. ChlGs from angiosperms (Arabidopsis thaliana, Nicotiana tabacum, A. sativa, and Oryza sativa) and other PS organisms were analyzed to determine whether Chl a formation in the presence of BChl a/BChlide a is possible. Among the enzymes assessed, ChlG, with the highest resistance to bacteriochlorin inhibition, was selected to produce Chl a in R. sphaeroides.

Bacterial Strains and Growth Conditions
R. sphaeroides 2.4.1 [10] was used as the wild-type (WT) strain (Table S1) and cultured in Sistrom's succinate-based (Sis) minimal medium [11] at 28 • C, as described previously [12]. Briefly, R. sphaeroides was grown aerobically with agitation at 250 rpm, and exponentially growing cells were used as the inoculum for anaerobic cultures, which were grown anaerobically in the dark in the presence of 75 mM dimethyl sulfoxide (DMSO) or photoheterotrophically at 10 W/m 2 light. In order to alleviate reactive oxygen species (ROS) formation, photoheterotrophic growth was performed in Sis medium supplemented with 2 mM DMSO as an antioxidant [13]. Escherichia coli was cultured at 37 • C in Luria-Bertani (LB) medium [14] with shaking at 250 rpm. For R. sphaeroides, tetracycline (Tc) and kanamycin (Km) were used at a final concentration of 1 µg/mL and 25 µg/mL, respectively. For Escherichia coli, ampicillin (Ap), Km, and Tc were used at final concentrations of 50 µg/mL, 25 µg/mL, and 20 µg/mL, respectively.

Construction of Plasmids
Details of the primers and plasmids used in this study are listed in Tables S2 and S3, respectively. Amino acid sequences of BchGs/ChlGs examined in this study and detailed procedures of plasmid construction are included in the Supplementary Materials.

Construction of R. sphaeroides Mutants
The target gene was internally deleted using pLO1, as described previously [15]. E. coli S17-1 [16] was transformed using recombinant pLO1 carrying a mutated target gene insert and subsequently mobilized into R. sphaeroides through conjugation. Single-crossover recombinants with the Km resistant (Km r ) phenotype were selected on Sis agar plates containing Km and were subsequently segregated on 15% (w/v) sucrose to obtain doublecrossover recombinants, which were sensitive to Km. Double-crossover recombination was confirmed by genomic DNA polymerase chain reaction (PCR) analysis for the target gene.

Isolation and Spectral Analysis of the Membrane Fraction from R. sphaeroides
Cells were harvested and resuspended in 10 mM phosphate-buffered saline (PBS) at pH 7.4, followed by lysis using a high-pressure homogenizer (Microfluidizer TM , Microfluidics Corporation, Newton, MA, USA) at 20,000 psi. The homogenate was centrifuged at 6000× g at 4 • C for 10 min, and the supernatant was subjected to ultracentrifugation at 150,000× g at 4 • C for 1 h. The supernatant, after ultracentrifugation, was decanted as a cytosolic fraction. Membranes in the pellet were solubilized in PBS supplemented with 1% (w/v) n-dodecyl-β-D-maltoside (DDM), followed by continuous mixing at 4 • C for 1 h. Insoluble material was removed by centrifugation at 12,000× g at 4 • C for 5 min. The total protein in the membrane fraction was determined using the Lowry method [18]. Before absorption spectra analysis using a UV-Vis spectrophotometer (UV 2550; Shimadzu, Kyoto, Japan), the samples were adjusted to the same protein level.

Pigment Extraction from R. sphaeroides
Pigments were extracted from cells using an acetone/methanol (7/2, v/v) mixture. The sample was vortexed vigorously for 1 min and centrifuged at 14,000× g at 4 • C for 5 min. The supernatant was filtered using a syringe filter (0.45 µm pore size, Whatman, Maidstone, UK), and the filtrate was subjected to HPLC analysis. The HPLC system was composed of an LC-6AD dual pump (Shimadzu, Japan) equipped with a Zorbax ODS column (Agilent Technologies, Santa Clara, CA, USA; particle size, 5 µm; diameter × length, 4.6 × 250 mm), a fluorescence detector, and a UV-Vis absorbance detector.

Preparation of BChlide a, Chlide a, BChl a GG , and Chl a GG
BChlide a and Chlide a were prepared from BChl a P and Chl a P (Sigma-Aldrich, St. Louis, MO, USA), respectively, using chlorophyllase of A. thaliana (AtChlase). C-terminally strep-tagged AtChlase was purified from E. coli BL21 (DE3) (Stratagene, San Diego, CA, USA) (Table S1) transformed with pChlase (Table S3). Cells were cultured until the OD 600 reached 0.6, and subsequently, anhydrotetracycline (Sigma-Aldrich) was added at a final concentration of 0.2 µg/mL. Cells were incubated for another 12 h and harvested by centrifugation at 6000× g at 4 • C for 5 min. The cell pellet was resuspended in Buffer A (50 mM Tris-Cl, pH 8.0, 150 mM NaCl), and cells were lysed via sonication on ice for 5 min for a total of three times. The sample was centrifuged at 8000× g at 4 • C for 15 min, and the supernatant was loaded onto Strep-Tactin TM resin (IBA Life Sciences, Göttingen, Germany). The resin was washed with Buffer A, and the AtChlase was eluted with Buffer A containing 2.5 mM desthiobiotin.
BChl a GG was prepared from BP (Table S1), and Chl a GG was prepared from BZFP-WSCP-Ntc (Table S1), similar to the method used in a previous study [8]. The cells were grown anaerobically in Sis medium supplemented with 75 mM DMSO in the dark. BChl a GG and Chl a GG were extracted using dioxane and purified via Sepharose CL-6B column chromatography, as described previously [20].
The extinction coefficients of BChl a GG and Chl a GG were similar to those of BChl a P and Chl a P , respectively. Equivalent levels of BChlide a were produced when either BChl a GG or BChl a P at the same absorbance were digested with AtChlase, and similar results for Chlide a were obtained with Chl a GG and Chl a P . The molar extinction coefficients of BChl a GG and Chl a GG were estimated to be 83.9 mM −1 cm −1 at 771 nm and 85.6 mM −1 cm −1 at 662 nm [19], respectively. 2.7. HPLC Determination of BChl a GG , Chl a GG , BChl a P , and Chl a P BChl a GG , Chl a GG , BChl a P , and Chl a P were detected by HPLC as described previously [3], with minor modifications. Samples were injected into a 200 µL injection loop and eluted using linear gradients of acetone/water mixture for 29 min at a flow rate of 1 mL/min. Solvent compositions were as follows: 0 min, 70% acetone; 2 min, 70% acetone; 4 min, 82% acetone; 15 min, 88% acetone; 19 min, 100% acetone; 24 min, 100% acetone; and 29 min, 70% acetone. Quantification of BChl a GG and BChl a P was performed by monitoring the absorbance at 771 nm, and Chl a GG and Chl a P were evaluated at 662 nm. Commercial BChl a P and Chl a P (Sigma-Aldrich) were used to generate standard curves for the quantification of BChl a GG (BChl a P ) and Chl a GG (Chl a P ), respectively.

Heterologous Expression and Quantification of BchG and ChlG in E. coli
E. coli BL21 (DE3) transformed with pET-Rsb, pET-Syc, pET-Atc, pET-Ntc, pET-Asc, and pET-Osc (Table S3) were cultured at 37 • C in 30 mL LB medium containing Km until OD 600 reached 0.6. Then, 1 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) was added, and cells were further incubated at the same temperature for 6 h. Cells were harvested by centrifugation at 4000× g and 4 • C for 5 min and washed once with PBS. Cell pellets were resuspended in 5 mL of Buffer B (120 mM potassium acetate, 10 mM magnesium acetate, 50 mM HEPES-KOH, pH 7.6, 14 mM mercaptoethanol, and 10% glycerol) [3], followed by sonication on ice for 5 min for a total of three times. The resulting sample was centrifuged at 8000× g and 4 • C for 5 min, and the supernatant (cell lysate) was used as the enzyme source.

Kinetic Analysis of BchG and ChlG Activities
BchG and ChlG activities were determined as described previously [3]. Cell lysates (1 mg protein) were mixed with 200 µL of Buffer B supplemented with 0.5 mM ATP, 50 µM GGPP (or 50 µM PPP), and varying concentrations of BChlide a (10-80 µM) for BchG activity assay or Chlide a (5-40 µM) for ChlG activity assay. The reaction mixture was incubated at 30 • C for 3 h for BchG activity and for 30 min for ChlG activity. Inhibitions of BchG and ChlGs were analyzed in the presence of inhibitors BChlide a/Chlide a, BChl a GG/ Chl a GG , and BChl a P /Chl a P at 10, 20, and 40 µM, respectively.
The reaction was stopped by adding 800 µL acetone, and the mixture was centrifuged at 12,000× g and 4 • C for 5 min. The supernatant was filtered and injected into the HPLC system, and BChl a and Chl a were determined as described earlier. Kinetic parameters, including inhibition constant, were calculated from the data sets of initial reaction rate V 0 by nonlinear regression fitting to competitive inhibition model using SigmaPlot ver. 14 (Systat Software, San Jose, CA, USA). All measurements were independently repeated three times, and data are shown as the mean ± standard deviation (SD).

Determination of Intracellular ROS Levels
ROS levels were determined as described previously [24]. Cell culture aliquots were centrifuged at 12,000× g and 4 • C for 1 min, and the pellet was resuspended in Sis medium. The oxidation-sensitive fluorescent probe, 2,7-dihydrodichlorofluorescein diacetate (H 2 DCFDA; Sigma-Aldrich), was added to the cell suspension at a final concentration of 10 µM. Samples were then incubated at 30 • C for 30 min, and fluorescence was measured at 525 nm (after excitation at 492 nm) using a microplate reader (EnSpire; PerkinElmer, Waltham, MA, USA). Samples were evaluated in triplicate, and the controls were not treated with H 2 DCFDA. All measurements were performed independently in triplicates, and the data are presented as the mean ± SD.
Singlet oxygen levels were determined using Singlet Oxygen Sensor Green (SOSG, Invitrogen, Waltham, MA, USA). Cells were grown aerobically, and aliquots (5% volume ratio) of cells in the exponential phase were inoculated into a fresh Sis medium (supplemented with 50 µM SOSG) at a final OD 660 of 0.1. Cells were grown photoheterotrophically for 1 h, and the fluorescence of reacted SOSG at 525 nm was evaluated after excitation at 504 nm. Cells grown in the dark for the same period were used as controls. Quantification was performed independently in triplicates, and the data are presented as mean ± SD.

Statistical Analysis
All results are presented as mean ± SD. In order to assess differences between the two samples, the Student's t-test was applied to the statistical analysis, in which significant differences were annotated by the following symbols: ns, p > 0.05; *, p ≤ 0.05; **, p ≤ 0.01; ***, p ≤ 0.001.

Angiosperm ChlGs Prenylated BChlide a
As AsChlG formed BChl aGG and also synthesized Chl aGG in vitro in the presenc BChlide a [9], we examined whether ChlGs from other PS organisms possess such br substrate specificities. A bchG-deleted mutant (BG1) [2] and a bchZ/bchF-deleted mu (BZF1) [17] of R. sphaeroides accumulated BChlide a, and Chlide a in cells (Figure 1), these mutants could be used to examine BchG and ChlG activity, respectively. BG1 BZF1 were transformed to express bchGs/chlGs from diverse PS organisms (Table S1). recombinant strains were cultured anaerobically in the dark condition with DMSO mM), and extracted pigments were analyzed by HPLC to assess the formation of intra lular BChl aP or Chl aP. BG1 transformants carrying bchGs of PS bacteria (purple su Allochromatium vinosum; green nonsulfur, Chloroflexus aurantiacus; green sulfur, Chloro ulum tepidum; and the aerobic anoxygenic PS bacterium, Roseobacter denitrificans) syn sized BChl aP as BG1 carrying R. sphaeroides bchG (BG1-Rsb) did ( Figure 2). Surprisin BG1 carrying chlGs of angiosperm (A. thaliana, N. tabacum, A. sativa, and O. sativa) synthesized BChl aP, whereas BG1 carrying other chlGs (vascular plants, Picea sitche (gymnosperm) and Selaginella moellendorffii (lycophyte); moss, Physcomitrella patens; al Auxenochlorella protothecoides and Bathycoccus prasinos (green algae), Chondrus crispus Cyanidioschyzon merolae (red algae); cyanobacterium, Synechocystis; green sulfur ba rium, C. tepidum) failed ( Figure 2). In contrast, all BZF1 transformants carrying chlGs thesized Chl aP, while all BZF1 carrying bchGs failed ( Figure 3). Another homolo NtchlG (named NtchlG2 hereafter) was found in the genome of N. tabacum, which sho 99% similarity (99% identity) to NtChlG. The formation of BChl aP by NtChlG2 was sim to that of NtChlG. (Table S4). The results are summarized in Table 1. Thus, angiosp ChlGs were confirmed to possess both ChlG and BchG activity, whereas other BchGs ChlGs only possess either of the activities (Table 1).  pigments were extracted from the cells and subjected to HPLC analysis. BChlide a and BChl aP are indicated by arrows, which were detected by absorbance at 771 nm. A.U., Arbitrary Unit.  BG1 is incompetent for PS growth because it does not synthesize BChl aP while accumulating BChlide a. Only an enzyme able to prenylate the BChlide a to BChl a can restore  BG1 is incompetent for PS growth because it does not synthesize BChl a P while accumulating BChlide a. Only an enzyme able to prenylate the BChlide a to BChl a can restore the PS growth [25]. The BchGs of PS bacteria (A. vinosum, C. aurantiacus, C. tepidum, and R. denitrificans) restored the photoheterotrophic growth of BG1-like BG1-Rsb (Figure 4c,d). ChlGs of angiosperm (A. thaliana, N. tabacum, A. sativa, and O. sativa) also restored the growth despite the extended lag periods and decreased growth rates compared to BG1-Rsb ( Figure 4a). As expected, other ChlGs (P. sitchensis, S. moellendorffii, P. patens, A. protothecoides, B. prasinos, C. crispus, C. merolae, Synechocystis, and C. tepidum) did not support the photoheterotrophic growth of BG1 (Figure 4a,b). These results further corroborated the intracellular BChl a P formation of BG1-carrying angiosperm chlGs (Figure 2). the PS growth [25]. The BchGs of PS bacteria (A. vinosum, C. aurantiacus, C. tepidum, and R. denitrificans) restored the photoheterotrophic growth of BG1-like BG1-Rsb (Figure 4c,d).
ChlGs of angiosperm (A. thaliana, N. tabacum, A. sativa, and O. sativa) also restored the growth despite the extended lag periods and decreased growth rates compared to BG1-Rsb ( Figure 4a). As expected, other ChlGs (P. sitchensis, S. moellendorffii, P. patens, A. protothecoides, B. prasinos, C. crispus, C. merolae, Synechocystis, and C. tepidum) did not support the photoheterotrophic growth of BG1 (Figure 4a,b). These results further corroborated the intracellular BChl aP formation of BG1-carrying angiosperm chlGs (Figure 2). Geranylgeranyl reductases (BchP and ChlP) reduce Chl aGG to produce Chl aP. Alternatively, PPP is formed first from GGPP by BchP and ChlP and subsequently used as a substrate for ChlG to form Chl aP [1,2] (Figure 1). The heterologous expression of SychlG with endogenous bchP in R. sphaeroides mutants grown under microoxic conditions, in which metabolic steps specific to BChl aGG formation were interrupted, resulted in the formation of Chl aP together with minor Chl a species possessing dihydroGG and tetrahy-droGG groups [8]. Similarly, Chl aP was formed predominantly by the various ChlGs examined in BZF1 grown in the dark in the presence of DMSO (75 mM) (Figure 3), which signifies the reduction in Chl aGG to Chl aP by BchP of R. sphaeroides. Minor Chl a was barely detectable in recombinant BZF1 cells (Figure 3), which may have been due to differences in the culture conditions.

NtChlG Exhibited the Highest Catalytic Efficiency for BChlide a among the Angiosperm ChlGs Examined
For functional comparison, kinetic analyses were performed using the angiosperm ChlGs (Table 2). RsBchG and SyChlG, which have the activities of BchG and ChlG alone, respectively, were included as controls. Both enzymes work in the ping-pong model using GGPP and PPP as the first substrate and BChlide a and/or Chlide a as the second substrate [2,26]. Km and kcat values were determined for BChlide a and Chlide a in the presence of GGPP and PPP, respectively. C-terminally His6-tagged BchG and ChlG were overexpressed in E. coli, and the cell lysates were used as enzyme sources. In order to calculate kcat, the amount of enzyme in the cell lysates was quantified via western blot analysis using Geranylgeranyl reductases (BchP and ChlP) reduce Chl a GG to produce Chl a P . Alternatively, PPP is formed first from GGPP by BchP and ChlP and subsequently used as a substrate for ChlG to form Chl a P [1,2] (Figure 1). The heterologous expression of SychlG with endogenous bchP in R. sphaeroides mutants grown under microoxic conditions, in which metabolic steps specific to BChl a GG formation were interrupted, resulted in the formation of Chl a P together with minor Chl a species possessing dihydroGG and tetrahy-droGG groups [8]. Similarly, Chl a P was formed predominantly by the various ChlGs examined in BZF1 grown in the dark in the presence of DMSO (75 mM) (Figure 3), which signifies the reduction in Chl a GG to Chl a P by BchP of R. sphaeroides. Minor Chl a was barely detectable in recombinant BZF1 cells (Figure 3), which may have been due to differences in the culture conditions.

NtChlG Exhibited the Highest Catalytic Efficiency for BChlide a among the Angiosperm ChlGs Examined
For functional comparison, kinetic analyses were performed using the angiosperm ChlGs (Table 2). RsBchG and SyChlG, which have the activities of BchG and ChlG alone, respectively, were included as controls. Both enzymes work in the ping-pong model using GGPP and PPP as the first substrate and BChlide a and/or Chlide a as the second substrate [2,26]. K m and k cat values were determined for BChlide a and Chlide a in the presence of GGPP and PPP, respectively. C-terminally His 6 -tagged BchG and ChlG were overexpressed in E. coli, and the cell lysates were used as enzyme sources. In order Biology 2023, 12, 573 9 of 16 to calculate k cat , the amount of enzyme in the cell lysates was quantified via western blot analysis using an anti-His 6 -tag antibody. PpfC-His 6 of similar size was used as the quantification standard ( Figure S1). NtChlG showed the highest catalytic efficiency for BchG activity among the angiosperm enzymes, which was two to three-fold higher than those of the other enzymes ( Table 2). The kinetic properties of NtChlG2 for BChlide a and Chlide a (Table S4) were similar to those of NtChlG (Table 2). NtChlG had six to eight-fold higher K m and two-fold lower k cat for BChlide a compared to RsBchG, which resulted in approximately 12-to 18-fold lower catalytic efficiency (k cat /K m ) ( Table 2). Nevertheless, NtChlG supported the photoheterotrophic growth of BG1 (Figure 4a). Conversely, angiosperm ChlGs demonstrated catalytic efficiency for Chlide a comparable to that of SyChlG ( Table 2).
The k cat /K m values for BchG and ChlG activities were higher when GGPP was used instead of PPP. Consistently, lower K m and higher k cat were observed with GGPP when compared with the corresponding values with PPP. Thus, the geranylgeranylated pigment may be preferentially formed and subsequently converted to a phytylated pigment. The highest catalytic efficiency of NtChlG for BChlide a was also consistent with the shortest lag period and highest growth rate of BG1 expressing NtchlG (BG1-Ntc) among BG1 expressing other angiosperm chlG genes (Figure 4a).
Because angiosperm chlGs were expressed in E. coli, their BchG activities had to be confirmed directly in the plant. As a representative, N. tabacum plants were germinated and cultured in light for three weeks. The leaves were harvested, and their lysates were prepared and used as enzyme sources for analysis of native ChlG ( Figure S2). The relative levels of NtChlG and NtChlG2 in the leaf lysates were not analyzed. However, the native leaf lysates exhibited intrinsic BchG activity. The apparent K m values of the leaf lysates for BChlide a and Chlide a (Table S5) were not significantly different from those of NtChlG in the E. coli lysate ( Table 2).

Angiosperm ChlGs Showed Resistance to Inhibition by Bacteriochlorin
To produce Chl a P in R. sphaeroides, ChlG must be resistant to bacteriochlorin inhibition. The following clues suggested that angiosperm ChlGs would possess the resistance. First, AsChlG prenylated Chlide a in the presence of BChlide a, the amount of which was comparable to that of the reaction without BChlide a in a previous study [9]. This is unlike SyChlG, which is competitively inhibited by BChlide a [2]. Second, angiosperm ChlGs synthesized Chl a P in BG1, which accumulates intracellular BChlide a (Figure 5b). Accordingly, ChlGs of angiosperm plants such as A. thaliana, N. tabacum, A. sativa, and O. sativa were examined for resistance to BChlide a inhibition to determine whether it is a common characteristic of the angiosperm enzymes. In order to select angiosperm ChlGs with the highest resistance to BChlide a, the inhibition constants (K i ) for BChlide a were determined. Because ChlG is situated in the cell membrane, BChl a GG and BChl a P were also examined to determine whether they can act as inhibitors. Remarkably, SyChlG was competitively inhibited by BChlide a [2] (Figure S3g,j) and as well as by BChl a GG and BChl a P (Figure S3h,i,k,l). Similarly, RsBchG was competitively inhibited by Chlide a [2] ( Figure S3a,d) and also by Chl a GG and Chl a P (collectively called chlorin) ( Figure S3b,c,e,f). The K i values of SyChlG for bacteriochlorin were similar to the corresponding constants of RsBchG for chlorin (Table 3). However, the K i values of angiosperm ChlGs for bacteriochlorin were much larger than those of SyChlG (Table 3), indicating that the enzyme activities are highly resistant to bacteriochlorin. Among them, NtChlG had the highest K i values for all bacteriochlorin inhibitors and was subsequently selected to produce Chl a P under the biosynthetic pathway for BChl a P of photoheterotrophically growing R. sphaeroides in the following experiments. To produce Chl aP in R. sphaeroides, ChlG must be resistant to bacteriochlorin inhibition. The following clues suggested that angiosperm ChlGs would possess the resistance. First, AsChlG prenylated Chlide a in the presence of BChlide a, the amount of which was comparable to that of the reaction without BChlide a in a previous study [9]. This is unlike SyChlG, which is competitively inhibited by BChlide a [2]. Second, angiosperm ChlGs synthesized Chl aP in BG1, which accumulates intracellular BChlide a (Figure 5b). Accordingly, ChlGs of angiosperm plants such as A. thaliana, N. tabacum, A. sativa, and O. sativa were examined for resistance to BChlide a inhibition to determine whether it is a common characteristic of the angiosperm enzymes. In order to select angiosperm ChlGs with the highest resistance to BChlide a, the inhibition constants (Ki) for BChlide a were determined. Because ChlG is situated in the cell membrane, BChl aGG and BChl aP were also examined to determine whether they can act as inhibitors. Remarkably, SyChlG was competitively inhibited by BChlide a [2] (Figure S3g,j) and as well as by BChl aGG and BChl aP ( Figure S3h,i,k,l). Similarly, RsBchG was competitively inhibited by Chlide a [2] ( Figure  S3a,d) and also by Chl aGG and Chl aP (collectively called chlorin) ( Figure S3b,c,e,f). The Ki values of SyChlG for bacteriochlorin were similar to the corresponding constants of RsBchG for chlorin (Table 3). However, the Ki values of angiosperm ChlGs for bacteriochlorin were much larger than those of SyChlG (Table 3), indicating that the enzyme activities are highly resistant to bacteriochlorin. Among them, NtChlG had the highest Ki values for all bacteriochlorin inhibitors and was subsequently selected to produce Chl aP under the biosynthetic pathway for BChl aP of photoheterotrophically growing R. sphaeroides in the following experiments. BG1-Rsb, BG1-Syc, and BG1-415 were used as controls. Pigments were extracted from the membranes and quantified by HPLC; the obtained results are listed on the right side of the corresponding spectra (b). Membrane (red) and cytosolic fractions (blue) of photoheterotrophically growing BG1-Rsb and BG1-Ntc were separated and adjusted to the same volume as that (2 mL) used for resuspension of cells harvested for the analysis of absorption spectra (c). Pigments were extracted from the membrane, and cytosolic fractions and the relative levels of BChl aP and Chl aP were quantified by HPLC (c). BChl aP and Chl aP were detected at low levels in cytosolic fractions, which were ascribed to the contaminated membrane debris. ND, not detected. . BG1-Rsb, BG1-Syc, and BG1-415 were used as controls. Pigments were extracted from the membranes and quantified by HPLC; the obtained results are listed on the right side of the corresponding spectra (b). Membrane (red) and cytosolic fractions (blue) of photoheterotrophically growing BG1-Rsb and BG1-Ntc were separated and adjusted to the same volume as that (2 mL) used for resuspension of cells harvested for the analysis of absorption spectra (c). Pigments were extracted from the membrane, and cytosolic fractions and the relative levels of BChl a P and Chl a P were quantified by HPLC (c). BChl a P and Chl a P were detected at low levels in cytosolic fractions, which were ascribed to the contaminated membrane debris. ND, not detected.

Heterologous Expression of NtchlG Resulted in the Formation of Free Chl a P in R. sphaeroides while Generating ROS
The phenotype of BG1-Ntc (BG1 carrying NtchlG) was analyzed intensively with BG1-Rsb, BG1-Syc (BG1 carrying SychlG), and BG1-415 (BG1 carrying empty vector) as controls. The BG1-Ntc showed phototrophic growth with a 5-to 10-h lag period and a two-fold extended doubling time compared with that of BG1-Rsb (Figures 4a,c, and 5a). As expected, SyChlG did not support the photoheterotrophic growth of BG1 (BG1-Syc) (Figures 4b and 5a). A small peak was detected at 680 nm (PC 680 ) in the membrane absorption spectrum of BG1-Ntc (Figure 5b), but no such peak was observed in the cytosolic fraction (Figure 5c). Chl a P of BG1-Ntc amounted to 5% of BChl a P (Figure 5b). As no PC 680 was observed in the cytosolic fractions of BG1-Ntc (Figure 5c), the membrane fraction of BG1-Ntc was subjected to clear native polyacrylamide gel electrophoresis (CN-PAGE) [27] to analyze free pigments ( Figure S4). Free Chl a P migrates faster than protein-bound Chl a P in CN-PAGE and can be easily detected in the lower part of the lanes via fluorescence under UV-A light. No free pigment was observed in BG1-Rsb, whereas free Chl a P was detected in the BG1-Ntc. Similar bands were observed with BG1 carrying other angiosperm chlGs ( Figure S4). Thus, the PC 680 formation is ascribed to free Chl a P accumulated in the membrane of cells grown under photoheterotrophic conditions. Formation of free Chl a P at a much lower level (5%) than total BChl a P further suggests that Chlide a is utilized to a small extent from the main metabolic stream toward BChlide a (Figure 5b).
In addition, WT-Ntc (WT carrying NtchlG) was analyzed with WT-Syc (WT carrying SychlG) and WT-415 (WT carrying empty vector) as controls. The growth of WT-Ntc was retarded compared to that of WT-415, whereas the growth of WT-Syc was not affected by the expression of the chlG gene ( Figure S5a). PC 680 was barely observed in the membrane spectrum of WT-Ntc ( Figure S5b). Nonetheless, Chl a P was still detected at only 2% of total BChl a P , which was lower than that of BG1-Ntc (5%). The absolute level of Chl a P in WT-Ntc (0.65 ± 0.02 nmole/mg protein) was also lower than that in BG1-Ntc (0.77 ± 0.07), which was due to the altered metabolic flow to BChl a P by the presence of ChlG in WT and BG1. Thus, we used BG1-Ntc to achieve higher Chl a P formation in R. sphaeroides.
RsBchG was inhibited by chlorin to a similar degree as SyChlG was inhibited by bacteriochlorin (Table 3). However, BChl a P levels in WT-Ntc were not different from that in the control cell WT-415 ( Figure S5b). Therefore, BChl a P formation by intrinsic RsBchG was unaffected by the heterologous expression of NtchlG in R. sphaeroides, which was due to the low level (2%) of Chl a P formation compared with BChl a P ( Figure S5b).
Next, we aimed to elucidate why the retarded growth rate and extended lag period were observed in BG1-Ntc (Figures 1 and 5a). Since aerobically grown seed cells were routinely inoculated into a fresh medium for the main culture at a 5% volume ratio, oxygen could be introduced into the photoheterotrophic culture of BG1-Ntc. Because free Chl a P was present in the membranes of photoheterotrophically growing BG1-Ntc (Figures 5b and S4), the general ROS levels were examined during photoheterotrophic growth. The general ROS levels of BG1-Ntc were higher than those of BG1-Rsb during the lag period and exponential growth (Figure 6b). Since DMSO is known to scavenge hydroxyl radicals [13], 2 mM DMSO was added at the onset of the photoheterotrophic culture. While the growth of the control cells (BG1-Rsb) was unaffected by the addition of 2 mM DMSO (Figure 6a), the general ROS level was decreased to a larger extent by 2 mM DMSO in BG1-Ntc (Figure 6b), resulting in the disappearance of the growth lag ( Figure 6a). However, the cell growth rate remained unaffected by treatment with DMSO. Singlet oxygen was measured using SOSG, and no difference was observed between the cells examined ( Figure 6c). Thus, the prolonged lag of BG1-Ntc resulted due to the elevated levels of general ROS; however, the reason for the slow growth rate during the exponential phase remains unclear. Next, we aimed to elucidate why the retarded growth rate and extended lag period were observed in BG1-Ntc (Figures 1 and 5a). Since aerobically grown seed cells were routinely inoculated into a fresh medium for the main culture at a 5% volume ratio, oxygen could be introduced into the photoheterotrophic culture of BG1-Ntc. Because free Chl aP was present in the membranes of photoheterotrophically growing BG1-Ntc (Figures 5b  and S4), the general ROS levels were examined during photoheterotrophic growth. The general ROS levels of BG1-Ntc were higher than those of BG1-Rsb during the lag period and exponential growth (Figure 6b). Since DMSO is known to scavenge hydroxyl radicals [13], 2 mM DMSO was added at the onset of the photoheterotrophic culture. While the growth of the control cells (BG1-Rsb) was unaffected by the addition of 2 mM DMSO (Figure 6a), the general ROS level was decreased to a larger extent by 2 mM DMSO in BG1-Ntc (Figure 6b), resulting in the disappearance of the growth lag (Figure 6a). However, the cell growth rate remained unaffected by treatment with DMSO. Singlet oxygen was measured using SOSG, and no difference was observed between the cells examined (Figure 6c). Thus, the prolonged lag of BG1-Ntc resulted due to the elevated levels of general ROS; however, the reason for the slow growth rate during the exponential phase remains unclear. Figure 6. The photoheterotrophic growth of BG1-Ntc (a) was examined in the presence (closed symbol, +) or absence (open symbol, −) of 2 mM DMSO. BG1-Rsb was used as the control. General ROS levels were determined during the growth period using H2DCFDA (b). Singlet oxygen levels were determined using SOSG 1 h after the onset of photoheterotrophic growth (c). Significances were assessed using the Student's t-test: ns, p > 0.05.

Discussion
Chl a production in R. sphaeroides would be the first step in the construction of Chl aPcontaining PS complexes. Recently, water-soluble Chl-binding protein (WSCP) of Brassica oleracea was produced in R. sphaeroides mutants that accumulated Chlide a [8], and the inhibition of SyChlG by BChlide a and BChl aP was prevented by blocking the metabolic steps specific to BChlide a formation. Consequently, the resulting mutant lacked a BChl aP-containing PS complex [8]. Interestingly, heterologous expression of Prochlorococcus marinus (a cyanobacterium) chlG (PmchlG) in R. sphaeroides WT resulted in Chl a accumulation [28]. However, the initial attempt to produce Chl aP using PmChlG failed in both the WT and BG1 strains ( Figure S6). Chl aP expression was observed in BZF1 alone, which

Discussion
Chl a production in R. sphaeroides would be the first step in the construction of Chl a Pcontaining PS complexes. Recently, water-soluble Chl-binding protein (WSCP) of Brassica oleracea was produced in R. sphaeroides mutants that accumulated Chlide a [8], and the inhibition of SyChlG by BChlide a and BChl a P was prevented by blocking the metabolic steps specific to BChlide a formation. Consequently, the resulting mutant lacked a BChl a Pcontaining PS complex [8]. Interestingly, heterologous expression of Prochlorococcus marinus (a cyanobacterium) chlG (PmchlG) in R. sphaeroides WT resulted in Chl a accumulation [28]. However, the initial attempt to produce Chl a P using PmChlG failed in both the WT and BG1 strains ( Figure S6). Chl a P expression was observed in BZF1 alone, which did not produce BChlide a ( Figure S6b). The differences between previous reports [28] and our study remain to be determined.
The results demonstrate that angiosperm ChlGs possess BchG activity. The physiological relevance of this activity in plants is unknown. Evolutionary analysis of ChlGs and BchGs using the maximum likelihood method with the Le-Gascuel model [29] revealed that angiosperm ChlGs constitute a subclade distinct from the other ChlGs examined ( Figure S7). Moreover, resistance to inhibition by bacteriochlorin (BChlide a, BChl a GG , and BChl a P ) is specific to angiosperm ChlGs, and this property is distinct from other ChlGs, including SyChlG. The two properties (BchG activity and resistance of ChlG activity to bacteriochlorin inhibition) of angiosperm ChlGs are likely to be correlated through the interaction of the substrate and inhibitor at the active site of the enzyme. Nevertheless, the exact biochemical correlation requires structural elucidation of ChlG and BchG.
Chlide a (C3-vinyl, C7-methyl, C7-8 double bond) and BChlide a (C3-acetyl, C7methyl, C7-8 single bond) have structural differences in the functional group at C3 (ring A) and the reduction state between C7 and C8 (ring B) ( Figure 1). The structural differences between Chlide a and BChlide a may be distinguished by the active sites of ChlG and BchG. AsChlG activity was examined using Chlide a, BChlide a, Chlide b (C3-vinyl, C7-aldehyde, C7-8 double bonds), and acetyl-Chlide a (C3-acetyl, C7-methyl, C7-8 double bond) [9]. The activity of AsChlG for Chlide b and acetyl-Chlide a was comparable to that of Chlide a, whereas that of BChlide a was reduced [9]. This result is corroborated by the lower specificity constant of AsChlG for BChlide a compared to that for Chlide a, and similar results were obtained for other angiosperms ChlGs (Table 2). Furthermore, AsChlG was able to utilize the Chlide a-derivative as a substrate, in which C7 was exchanged to hold a bulky phenylamino group. The resulting activity was modestly lower than that observed for Chlide a [30]. Thus, AsChlG activity was largely unaffected when the Chlide substrate was altered at C3 and C7, but the state of reduction between C7-8 was critical for the activity of AsChlG.
AsChlG has been shown to have higher activity in the presence of GGPP than in the presence of PPP [1]. Similarly, RsBchG and SyChlG showed lower K m values for BChlide a and Chlide a, respectively, in the presence of GGPP [2], and similar results were obtained for angiosperm ChlGs (Table 2). Thus, GGPP is preferred relative to PPP by BchG and ChlGs examined in this study.
Remarkably, RsBchG and SyChlG were inhibited by Chl a and BChl a, respectively (Table 3). When the C 20 moiety of the inhibitor (Chl a/BChl a) was the same as that of the first substrate (GGPP and PPP), the inhibition constant K i was higher than that observed when they were different. Although a detailed interpretation of the results requires structural elucidation of both ChlG and BchG, the results may reflect the interaction between the prenylated active site and the prenyl group of the inhibitor (Chl a/BChl a).
The photoheterotrophic growth of R. sphaeroides was affected when NtchlG was expressed. As growth retardation was not observed with SychlG ( Figure S5), Chl a P accumulation was thought to affect cell growth. Free Chl a P generated ROS and DMSO (2 mM) was used to prevent oxidative damage. However, DMSO failed to restore the growth rate to that observed in control. One possible cause is the inhibition of BChl a P incorporation into the PS complex by Chl a P . The reaction center (RC) complex of R. sphaeroides has four BChl a P and two bacteriopheophytin a P (Bpheo a P , Mg 2+ -free form of BChl a P ) molecules. When pheophytin a P (Pheo a P , Mg 2+ -free form of Chl a P ) is provided to an isolated RC complex, the Bpheo a P of the RC complex can be easily replaced by Pheo a P in the presence of a mild detergent, such as lauryldimethylamine oxide, with heating at 42 • C [31][32][33][34][35]. The modified RC complex containing Pheo a P illustrated the electron transfer kinetics with a decreased quantum yield for both triplet special pair formation and quinone reduction compared with the native RC [34,35]. BChl a P of light-harvesting complex 2 (LH2) complex of R. sphaeroides can also be exchanged for Chl a P , which was proposed to exhibit carotenoid-to-BChl energy transfer that was different from the native complex [36,37]. Thus, Chl a P produced by NtChlG can affect the binding of BPheo a P and BChl a P to the RC and LH2 complexes, respectively, resulting in unexpected detrimental effects on PS growth.

Conclusions
In this study, angiosperm ChlGs were kinetically characterized and found to exhibit ChlG activities that are resistant to bacteriochlorin inhibitions. Among the angiosperms ChlGs studied, N. tabacum chlG demonstrated the best activity, and its expression produced Chl a P in R. sphaeroides under photoheterotrophic conditions in which BChl a P coexisted. The results of this study will be useful for building functional Chl a P -containing LHC in R. sphaeroides, which is currently underway.  Table S4: Kinetic parameters of NtChlG2; Figure S1: Quantification of BchG and ChlG in E. coli lysate; Figure S2: ChlG and BchG activities in leaf lysates of N. tabacum; Table S5: K m of native ChlG from leaf lysates of N. tabacum for BChlide a and Chlide a; Figure S3: Inhibition plots of RsBchG and SyChlG for chlorin and bacteriochlorin, respectively; Figure S4: Detection of free Chl a P in the membranes of BG1 expressing angiosperm chlGs; Figure S5: Effect of NtchlG expression on the photoheterotrophic growth of WT; Figure S6: Chl a P formation by Prochlorococcus marinus ChlG in R. sphaeroides; Figure S7: Evolutionary analysis of BchGs and ChlGs; Amino acid sequence of BchG and ChlG; Construction of plasmids; Clear-native PAGE; Kinetic analysis of BchG and ChlG activities in leaf extract of N. tabacum. References [2,8,11,16,17,27,29,[38][39][40][41][42] are cited in the supplementary materials.