Identification of a Gene Essential for the First Committed Step in the Biosynthesis of Bacteriochlorophyll c

Abstract

Bacteriochlorophylls (BChls) c, d, and e are the major chlorophylls in chlorosomes, which are the largest and one of the most efficient antennae produced by chlorophototrophic organisms. In the biosynthesis of these three BChls, a C-13²-methylcarboxyl group found in all other chlorophylls (Chls) must be removed. This reaction is postulated to be the first committed step in the synthesis of these BChls. Analyses of gene neighborhoods of (B)Chl biosynthesis genes and distribution patterns in organisms producing chlorosomes helped to identify a gene (bciC) that appeared to be a good candidate to produce the enzyme involved in this biochemical reaction. To confirm that this was the case, a deletion mutant of an open reading frame orthologous to bciC, CT1077, was constructed in Chlorobaculum tepidum, a genetically tractible green sulfur bacterium. The CT1077 deletion mutant was unable to synthesize BChl c but still synthesized BChl a and Chl a. The deletion mutant accumulated large amounts of various (bacterio)pheophorbides, all of which still had C-13²-methylcarboxyl groups. A C. tepidum strain in which CT1077 was replaced by an orthologous gene, Cabther_B0031 from “Candidatus Chloracidobacterium thermophilum” was constructed. Although the product of Cabther_B0031 was only 28% identical to the product of CT1077, this strain synthesized BChl c, BChl a, and Chl a in amounts similar to wild-type C. tepidum cells. To indicate their roles in the first committed step of BChl c, d, and e biosynthesis, open reading frames CT1077 and Cabther_B0031 have been redesignated bciC. The potential mechanism by which BciC removes the C-13²-methylcarboxyl moiety of chlorophyllide a is discussed.

Introduction

Chlorophototrophs are organisms that rely on chlorophyll-based phototrophy as their principal energy source. Six bacterial phyla are currently known to include chlorophototrophs: Cyanobacteria, Proteobacteria, Chlorobi, Chloroflexi, Firmicutes, and Acidobacteria (1, 2). These organisms synthesize a wide variety of chlorophylls (Chls)² and bacteriochlorophylls (BChls) (3), which are always used as the primary electron donors in photosynthetic reaction centers and which also function in light-harvesting antennae (1). Three families of photochemical reaction centers are known. Homodimeric and heterodimeric type-1 reaction centers produce weak oxidants and strong reductants, and heterodimeric type-2 reaction centers produce strong oxidants and weak reductants (4, 5). These three families of reaction centers can be combined with various, independently evolved light-harvesting structures, which allow microorganisms to perform chlorophototrophy optimally as a function of the chemical and light environment (1).

Chlorosomes are highly efficient light-harvesting structures that are found in all chlorophototrophic members of Chlorobi, in some members of the Chloroflexi, and in a recently discovered organism, “Candidatus Chloracidobacterium thermophilum” (Candidatus C. thermophilum), which is currently the only known chlorophototrophic member of the Acidobacteria (1, 2). Chlorosomes can contain up to 250,000 BChl c, d, or e molecules (1, 6, 7), and these types of BChls uniquely occur in chlorosomes. Chlorosomes are the largest light-harvesting complexes found in nature; their size and efficiency permit chlorosome-containing organisms to grow under extremely low light intensities, which are unable to support the growth of organisms with smaller antenna complexes (8, 9).

In chlorosomes, the BChl c, d, or e molecules self-aggregate and form large, protein-independent suprastructures. By combining results from systems biology, cryo-electron microscopy, solid-state NMR, and molecular modeling, it was recently shown that these BChls form concentric nanotubes (10, 11). The self-assembly of BChl c, d, and e molecules into such structures is directly related to two distinctive features of these molecules (12, 13): the presence of C-3¹-hydroxyl group, which provides a ligand to an adjacent BChl molecule and the absence of C-13²-methylcarboxyl group, which reduces steric hindrance that would otherwise interfere with self-aggregation (14). BChl c, d, and e are also methylated at the C-8² and C-12¹ positions in some but not all organisms; these additional methyl groups influence the stereochemistry of the C-3¹ chiral center (15). BChl c and e, but not BChl d, also are methylated at the C-20 position, and BChl e has a formyl group instead of a methyl group at C-7 (3). These modifications cause important adjustments to the absorption maxima of the BChl aggregates, to the absorption bandwidth of the BChl aggregates, and to the amount of BChl produced per cell (16–18). Thus, the addition of the methyl groups to the BChl molecules helps organisms to achieve optimal light absorption under diverse, light-limiting growth conditions.

Using C. tepidum as a genetically tractable model organism, the biosynthetic pathway for the synthesis of BChl c_F (BChl c esterified with farnesol), BChl a_P (BChl a esterified with phytol), and Chl a_PD (Chl a esterified with Δ2,6-phytadienol) has been well characterized in green sulfur bacteria. Following the recent discovery of two types of 8-vinyl reductases in various green sulfur bacterial strains (19, 20),³ only two steps in the biosynthesis of BChl c/d/e are currently uncharacterized. The first is the removal of C-13²-methylcarboxyl group from chlorophyllide (Chlide) a to produce 3-vinyl-bacteriochlorophyllide d, and the second is the conversion of Bchlide c to Bchlide e. Mutational analyses of genes encoding all other steps in the BChl c biosynthetic pathway had strongly suggested that the reaction illustrated in Fig. 1 is the first committed step in the biosynthesis of BChl c, d, and e. The biosynthetic pathway leading from the “hub” intermediate, Chlide a, to BChl c, d, and e has previously been summarized (3, 21). As mentioned above, the removal of the C-13²-methylcarboxyl group is essential for the self-aggregation of BChl c, d, and e molecules into the concentric nanotubes found in chlorosomes. Thus, the identification of the gene(s) involved in this reaction are very important for a complete understanding of the biosynthesis of BChl c, d, and e and are additionally important for understanding how chlorosome-producing organisms evolved the capacity to synthesize these distinctive BChls.

FIGURE 1.

Removal of C-13²-methylcarboxyl moiety from chlorophyllide a. The first committed step in the biosynthesis of BChl c, d, and e. The C-13²-methylcarboxyl group is indicated by a circle on the chlorophyllide a molecule.

The removal of the C-13²-methylcarboxyl group from Chlide a is a reaction that is very similar to the conversion of pheophorbide a (Pheide a) to PyroPheide a during the Chl a degradation in plants, algae, and some cyanobacteria (22–24). Two enzymes have been identified that catalyze this reaction by apparently different mechanisms. Pheophorbidase is a methylesterase that belongs to the α/β-hydrolase (esterase/lipase) protein superfamily, and it hydrolyzes the C-13²-methylcarboxyl group to produce C-13²-carboxyl-PyroPheide a (22, 25). Following the removal of the methyl group, spontaneous decarboxylation occurs to produce PyroPheide a. Alternatively, pheophorbide demethoxycarbonylase removes the C-13²-methylcarboxyl group of Pheide a in a concerted manner with no carboxylate intermediate (22). Pyropheophytin production was also reported in Euglena gracilis (26), although neither the enzyme involved nor the reaction mechanism is known. No homologs of pheophorbidase occur in the sequenced genomes of BChl c, d, or e-synthesizing organisms, and the gene product responsible for pheophorbide demethoxycarbonylase has not yet been identified. Thus, a presently unknown gene product(s) must catalyze the removal of C-13²-methylcarboxyl group in bacteria that produce chlorosomes.

In this study, we describe the identification of a gene denoted bciC (ORF CT1077 in C. tepidum), which encodes a product that is essential for the synthesis of BChl c. The bciC gene was initially identified by examining gene neighborhoods and by performing phylogenetic profiling analyses. To verify that BciC played a role in BChl c biosynthesis, a C. tepidum null mutant was constructed by deleting the bciC gene. This mutant was unable to synthesize BChl c and accumulated a variety of Bpheides. By characterizing these compounds in detail, we show that BciC is essential for the removal of the C-13²-methylcarboxyl group of Chlide a. The biochemical and physiological properties of the ΔbciC mutant are described, and the potential enzymatic function of BciC is discussed.

EXPERIMENTAL PROCEDURES

Bioinformatics Analysis

A phylogenetic profiling analysis similar to the one used by Ito et al. (20) was used to identify genes likely to be involved in BChl c, d, or e biosynthesis. The genomes of Chlorobi and Chloroflexi strains (listed in Fig. 2) were downloaded from Integrated Microbial Genomes website and Genbank^TM and that of the Candidatus C. thermophilum strain B was recently determined in our laboratory.⁴ This selection included the available genomes of all organisms that synthesize BChl c, d, or e as well as closely related chlorophototrophs that do not. An all-against-all BLASTP search of translated genes of all these genomes was performed using an e-value cutoff of 10⁻⁵. Phylogenetic profiles of all translated genes were created by collecting the e-values of their reciprocal best hits in each genome. A value of 1 was assigned to those without reciprocal best hits. A PERL script was used to search the output scores for genes with phylogenetic profiles matching the theoretical best model. The theoretical model consisted of a score of 0 for each BChl c, d, or e-synthesizing strain and a score of 1 from each strain unable to synthesize BChl c, d, or e.

FIGURE 2.

Gene neighborhood and phylogenetic profile indicate bciC and homologs are possibly involved in the biosynthesis of BChl c, d, and e. A, gene neighborhood containing bciC homolog in Candidatus C. thermophilum genome. The white arrow denotes the bciC homolog (Cabther_B0031). bchH encodes the large subunit of magnesium chelatase; hemA encodes glutamyl-tRNA reductase; bchL encodes a subunit of protochlorophyllide reductase; bciB denotes the homolog of a cyanobacterial-type 8-vinyl protochlorophyllide reductase gene identified by Ito et al. (20); bchF encodes C-3 vinyl hydratase; bchQ and bchR encode C-8² and C-12¹ methyltransferases (3, 17), respectively. B, BLASTP results of BciC of Candidatus C. thermophilum against proteins of chlorophototrophic members of the phyla Chlorobi, Chloroflexi, and Acidobacteria. Strains used in this search were as follows: for Acidobacteria, Candidatus C. thermophilum strain B; for Chloroflexi (from left to right): Oscillochloris trichoides DG6, Chloroflexus aurantiacus Y-400-fl, Chloroflexus aurantiacus J-10-fl, Chloroflexus aggregans DSM 9485, Roseiflexus sp. RS-1, Roseiflexus castenholzii DSM 13941; for Chlorobi (from left to right): Chloroherpeton thalassium ATCC 35110, Chlorobium phaeovibrioides DSM 265, Chlorobium luteolum DSM 273, Chlorobium clathratiforme DSMZ 5477, Chlorobium phaeobacteroides BS1, Prosthecochloris aestuarii DSM 271, Chlorobium chlorochromatii CaD3, Chlorobium ferrooxidans DSM 13031, Chlorobium phaeobacteroides DSM 266, Chlorobium limicola DSM 245, Chlorobaculum parvum NCIB 8327, and Chlorobaculum tepidum TLS, Chlorobaculum limnaeum DSMZ 1677^T.

Materials and Growth Conditions

The wild-type strain of C. tepidum, growth media, and standard growth conditions used in this study have been described previously in detail (27). The light intensity was ∼90 μmol m⁻² s⁻¹ for growth of C. tepidum cells prior to transformation and throughout the transformation and segregation processes. The light intensity was increased to ∼300 μmol m⁻² s⁻¹ for the ΔbciC deletion mutant after complete segregation to promote more rapid growth. Cells of the wild-type and a strain in which the C. tepidum bciC gene was replaced with that of Candidatus C. thermophilum were grown under identical light intensity (∼300 μmol m⁻² s⁻¹) and temperature (42 °C) to facilitate comparisons of their pigment contents.

Construction of Mutants

Open reading frame CT1077 (bciC) was inactivated by transformation by using a construction in which the coding sequence was replaced by the aadA gene, which confers resistance to streptomycin and spectinomycin. Two DNA fragments, one directly upstream (∼750 bp) and one directly downstream of CT1077 (∼750 bp), were amplified from genomic DNA by PCR using primers 1077Upfor and 1077Uprev and 1077Dnfor and 1077Dnrev (see supplemental Table 1 for sequences of all primers used in this study). These resulting amplicons were digested with the appropriate restriction enzymes and were sequentially cloned into plasmid pUC19 using PstI/KpnI and KpnI/EcoRI restriction sites, respectively. The aadA resistance cassette was excised from pSRA2 plasmid (28) and inserted into the introduced KpnI site between the upstream and downstream fragments. PCR screening was used to identify clones in which the aadA cassette was in the same transcriptional orientation as CT1077 (Fig. 3A). This construction was transformed into C. tepidum cells as described previously (28).

FIGURE 3.

Physical maps and gel electrophoresis of PCR products showing construction and segregation of deletion and replacement mutants of bciC in C. tepidum. Scheme showing the construction of the bciC deletion (A) and replacement (B) mutants in C. tepidum. C, agarose gel electrophoresis of PCR products of ORF CT1077 locus of C. tepidum wild-type (lane 1), ΔbciC mutant (lane 2), and bciC replacement strain ΔbciC::bciC^CAB-aadA (lane 3) using primers (small arrows labeled segfor and segrev in A and B) upstream and downstream of ORF CT1077.

C. tepidum ORF CT1077 was replaced with the orthologous gene from Candidatus C. thermophilum by using a similar approach. Upstream and downstream fragments were amplified using primers 1077Upfor and 1077UprevNcoI and 1077Dnfor and 1077Dnrev, which introduced an NcoI site at the position equivalent to the start codon of CT1077. These fragments were assembled in pUC19 using the same restriction sites as above. Cabther_B0031 was amplified from genomic DNA by PCR using primers B0031for and B0031rev. This amplicon was inserted into the resulting plasmid from the previous step between the NcoI and KpnI sites, and the aadA gene was then inserted as described above. In the resulting construction, ORF CT1077 was completely replaced from its start to stop codons, and the construct had the aadA marker downstream as a selectable marker for transformation (Fig. 3B). The resulting plasmids were linearized by digestion with AhdI and used to transform wild-type C. tepidum as described (27, 28). Segregation of the mutants was confirmed by PCR by using primers upstream and downstream of CT1077 (Fig. 3), and the resulting amplicons were sequenced to verify that no unwanted mutations had occurred during PCR amplification.

Pigment Analysis

Cell cultures were grown to late exponential phase (OD_{600 nm} = 1.6–1.8) and collected by centrifugation. Cells in 1.0 ml of culture were resuspended in aqueous buffer (10 mm KH₂PO₄, 50 mm NaCl, pH = 7.0), and whole-cell absorption spectra from 350 to 950 nm were recorded using a Genesys spectrophotometer (ThermoSpectronic, Rochester, NY). Pigments were extracted by brief sonication of cells after resuspension in methanol, and cell debris and precipitated protein was removed by centrifugation. The absorption spectra of the total pigment extracts in methanol were recorded in the same manner. For HPLC and HPLC-MS analyses, pigments were similarly extracted in acetone:methanol (7:2, v/v). The extracted pigments were analyzed by reversed-phase HPLC on an Agilent 1100 series HPLC system (Agilent Technologies, Santa Clara, CA) using an analytical Discovery C18 column (4.6 mm × 25 cm) (Supelco, Sigma-Aldrich). The solvents and gradient protocols used have been described previously (19). Solvent A was 62.5% water, 21% methanol, 16.5% acetonitrile containing 10 mm ammonium acetate, whereas solvent B was 50% methanol, 20% acetonitrile, and 30% ethyl acetate. The elution program starts at 20% solvent B at 0.75 ml min⁻¹ and increases to 70% solvent B at 1 ml min⁻¹ gradually in 10 min. It then increases to 100% solvent B, 1 ml min⁻¹ over 30 min followed by 100% solvent B, 1 ml min⁻¹ for another 10 min. It then gradually changes to starting solvent conditions and flow rate over 10 min. ChemStation software (version B.02.01; Agilent Technologies) was used to analyze the chromatograms and absorption spectra of eluates. (B)Chl concentrations were calculated from absorbance using molar extinction coefficients of 86 liter g⁻¹ cm⁻¹ at 667 nm for BChl c, 79.2 liter g⁻¹ cm⁻¹ at 665 nm for Chl a and 60 liter g⁻¹ cm⁻¹ at 770 nm for BChl a (28). Protein concentrations were determined using the Invitrogen Quant-iT Protein Assay kit.

Mass Spectrometry

Samples for HPLC-MS and MS/MS were only prepared from the bciC deletion mutant. Pigments were extracted similarly as described above in ∼15 ml acetone/methanol mixture from the cells harvested from 500-ml culture, and after the resulting extract was concentrated under a nitrogen stream down to ∼1 ml, the pigment solution was loaded on a larger (10 mm × 25 cm) Discovery C18 purification column (Supelco, Sigma-Aldrich). The same HPLC system and procedure described above, but with 3-fold higher flow rate, were used to separate the (B)Pheides from other molecules. The eluted (B)Pheides were collected and dried under flow of nitrogen. The dried pigments were dissolved in methanol (1 ml) and used for HPLC-MS and MS/MS analyses. HPLC-MS analyses were carried out using the analytical HPLC column and solvent program described above on an Agilent 1100 series HPLC system (Agilent Technologies) connected to an LCT Premier mass spectrometer (Waters, Milford, MA). To increase ionization efficiency, formic acid (98%) was introduced post-column into the eluent stream at a rate of 5 μl min⁻¹ with a syringe pump (29). MS/MS analyses for ions of selected masses were carried out on a 3200QTRAP mass spectrometer (Applied Biosystems, Foster City, CA).

Synthesis of (Bacterio)pheophytins

Pigments were extracted from the cells of a 2-liter culture of the bciC deletion mutant, concentrated, and purified by preparative HPLC as mentioned above. Two fractions, containing ∼1 mg of purified BChl a_P and ∼0.1 mg of purified Chl a_PD, respectively, were collected separately and dried under nitrogen. Pheophytin (Pheo) a_PD was synthesized by demetallation of Chl a_PD by adding 10 ml of HCl (2%, v/v) to Chl a_PD dissolved in 10 ml of methanol. Pheo a_PD was extracted from the mixture using ethyl ether. The ether phase was washed sequentially with 4% (w/v) NaHCO₃ and water, and the resulting pigment extract was dried. Bacteriopheophytin (Bpheo) a_P, 3-acetyl-Pheo a_P and 3-(1-hydroxyethyl)-Pheo a_P were synthesized from BChl a_P by procedures similar to those described by Kunieda et al. (30). Briefly, Bpheo a_P was synthesized by demetallation of BChl a_P as described above for Pheo a_PD. Oxidation of the B-ring of Bpheo a_P was performed by incubation with 2,3-dichloro-5,6-dicyano-p-benzoquinone in acetone. After extraction and washing, the product was 3-acetyl-Pheo a_P. 3-(1-hydroxyethyl)-Pheo a_P was synthesized by incubating Bpheo a_P with NaBH₄ in isopropanol to reduce 3-acetyl group followed by subsequent oxidation of the B-ring using 2,3-dichloro-5,6-dicyano-p-benzoquinone.

RESULTS

Identification of Potential BChl c, d, and e Biosynthesis Gene

Fig. 2A shows a region of the recently completed genome of Candidatus C. thermophilum, which produces chlorosomes and synthesizes BChl c (2).⁴ All of the genes in this cluster, which are probably co-transcribed as an operon, are predicted to play a role in the synthesis of protoporphyrin IX (hemA) or BChls (bchH, bchL, bciB, bchF, bchQ, and bchR) except for one open reading frame (Cabther_B0031) with an unknown function. Thus, it seemed reasonable to hypothesize that Cabther_B0031 was also involved in (B)Chl biosynthesis. Because the genes encoding all but one enzyme required for the synthesis of BChl a, Bchl c, and Chl a were already known and identified in the Candidatus C. thermophilum genome, it seemed highly likely that Cabther_B0031 encoded the remaining enzyme. The unidentified gene should encode an enzyme for the removal of C-13²-methylcarboxyl group, which was thought to be the first committed step in the synthesis of BChl c, BChl d, and BChl e from Chlide a.

A phylogenetic profiling analysis was performed to obtain further support for this hypothesis. Assuming that all organisms that can synthesize BChl c, d, or e use the same gene product(s) to catalyze the removal of C-13²-methylcarboxyl group, the gene(s) of interest should be found in all of these organisms and should be absent from organisms that are unable to synthesize BChl c, d, or e. Fig. 2B shows the results of a phylogenetic profiling analysis based on this reasoning, which was used to identify genes whose phylogenetic profiles best fit the anticipated distribution pattern (see “Experimental Procedures” for details of the analysis). Homologs of Cabther_B0031 matched the predicted pattern perfectly (Fig. 2B). It was fully understood that this analysis was based on an assumption that might be incorrect. Furthermore, even if it the assumption was correct, genes identified by this type of analysis could be involved in functions other than the specific enzymatic reaction (e.g. the gene product could regulate some aspect of BChl c, d, and e biosynthesis). Despite these reservations, the gene neighborhood and phylogenetic profiling analyses provided strong and complementary evidence that Cabther_B0031 and its orthologs were candidates to encode a protein required for the first committed step in BChl c, d, and e biosynthesis.

(B)Chl Contents of ΔCT1077 Deletion Mutant

To investigate the function of the gene family that includes Cabther_B0031, a deletion mutant of the orthologous gene in C. tepidum (CT1077) was constructed (Fig. 3, A and C). PCR analysis showed that this deletion mutant segregated completely (Fig. 3, A and C, compare lanes 1 and 2). Unlike the dark green wild-type cells of C. tepidum, the ΔCT1077 mutant cells were orange-colored. This color phenotype is characteristic of cells that are deficient in, if not completely devoid of, BChl c. For example, the mutant cells were similar in appearance to those of a previously characterized bchK null mutant, which was also unable to synthesize BChl c (28). Fig. 4A compares the absorption spectrum of whole cells of the wild-type (solid line) and the mutant (dashed line). The absorption peaks at 460 and 747 nm due to BChl c were completely missing. Only very weak and broad absorbance bands at ∼800 nm and in the region from 400 to 500 nm were observed. Similar results were observed by Frigaard et al. (28) for the bchK mutant. The absorption band near 800 nm was derived from a combination of BChl a-binding proteins, including CsmA, Fenna-Matthews-Olson protein, and the reaction centers (28, 31, 32).

FIGURE 4.

Absorption spectra of whole cells and extracted pigments from C. tepidum wild-type (solid line), ΔbciC mutant (dashed line) and the bciC replacement strain ΔbciC::bciC^Cab-aadA (dotted line). A, absorption spectra of whole cells in aqueous buffer (see “Experimental Procedures” for detail). B, absorption spectra of extracted pigments in methanol. Spectra were normalized at 600 nm.

Fig. 4B shows the absorption spectra of pigments extracted from wild-type cells (solid line) and the ΔCT1077 mutant (dashed line). The spectrum for the methanol extract of the mutant cells lacked the strong absorption peaks at 435 and 667 nm that are characteristic of BChl c. The absorption peaks at 750 and 665 nm suggested that molecules related to BChl a and Chl a, respectively, were still produced in the mutant. Chlorobactene, which absorbs at ∼490 nm (33), was present in both extracts as expected. These absorption spectra showed that the ΔCT1077 deletion mutant was completely unable to synthesize BChl c and chlorosomes. However, the mutant could still normally synthesize other pigments (viz. BChl a, Chl a, and carotenoids).

HPLC analyses of pigments extracted from the ΔCT1077 deletion mutant and wild-type cells confirmed the conclusions derived from the absorption spectra. As shown in Fig. 5A, BChl c_F was not detected in the mutant strain, BChl c esterified by other, more hydrophobic alcohols, and Bchlide c were likewise not detected in the HPLC analyses of pigment extract from the mutant cells. However, BChl a_P and Chl a_PD were detected at similar levels (Fig. 5 and Table 1). These results clearly showed that deletion of ORF CT1077 led to an inability to synthesize BChl c, and only this (B)Chl, in C. tepidum. Thus, it was tentatively concluded that the product of ORF CT1077 was probably involved exclusively in the biosynthetic pathway leading from Chlide a to BChl c. However, it was not yet clear whether this product was essential for the removal of C-13²-methylcarboxyl group.

FIGURE 5.

HPLC elution profiles of extracted pigments from C. tepidum ΔbciC mutant (A) and wild-type (B) at 667 nm. See detailed elution times, molecular weights, absorption spectra, and probable structures of eluates in peaks 1 to 5 (A) in Fig. 6 and Table 2.

TABLE 1.

BChl c, BChl a, and Chl a contents (μg per mg protein) in cells of C. tepidum wild-type, bciC deletion mutant, and bciC replacement strain ΔbciC::bciC^Cab-aadA

Cells	BChl c	BChl a	Chl a
Wild-type	124 ± 8	3.7 ± 0.5	0.42 ± 0.03
ΔbciC mutant	0	4.3 ± 0.3	0.37 ± 0.05
ΔbciC::bciCCAB-aadA strain	110 ± 12	3.8 ± 0.2	0.41 ± 0.03

As shown in Fig. 5A (peaks 1–5), the CT1077 deletion mutant accumulated large amounts of pigments that eluted very early from the reversed-phase column, which indicated that these molecules were much more hydrophilic than BChl c_F. As will be shown below, these molecules were (B)Pheides. Similar amounts of (B)Pheides also accumulated in a bchK mutant (28). These (B)Pheides seemed likely to be derived from accumulated (B)Chlide intermediates resulting from the loss of a particular enzymatic function in BChl c biosynthesis. Thus, the identification of these molecules could provide essential evidence to identify which enzyme function was missing in the ΔCT1077 mutant.

Identification of (B)Pheides in ΔCT1077 Deletion Mutant

The structures of the (B)Pheides in peaks 1–5, Fig. 5A were deduced on the basis of their absorption spectra (Fig. 6), their elution times, comparisons with chemically prepared standard compounds, and their molecular masses obtained from HPLC-MS analysis (Table 2). Peaks 4 and 5 had identical absorption spectra (e.g. Fig. 6C, solid line for peak 4), which was nearly identical to that of Pheo a (Fig. 6C, broken line; 34). The molecular mass of the compound in peak 4 (592 Da) was same as that of Pheide a, but the mass of the compound in peak 5 had a molecular mass of 606 Da. The increased elution time indicated that the compound in peak 5 was more hydrophobic, and it was surmised that this compound was a methylated form of Pheide a. There are three methyltransferases in C. tepidum that could modify Chlide a: BchU (methylation at C-20), BchQ (methylation at C-8²), and BchR (methylation at C-12¹). Because the absorption spectra of the compounds in peaks 4 and 5 were identical to Pheo a, the methylation must have occurred at either C-12¹ or C-8² but not directly on the tetrapyrrole ring at C-20 (16, 17, 35, 36). Based on previous studies that showed that methylation by BchR at C-12¹ is nearly complete, whereas methylation at C-8² is less probable in C. tepidum, it was concluded that peak 5 probably contained [12-Et]-Pheide a (for further details, see “Discussion”).

FIGURE 6.

Absorption spectra of pheophorbides accumulated in C. tepidum ΔbciC mutant (peaks 1–5 in Fig. 5A) and corresponding pheophytins. A, deduced structure of eluates in peak 1, 3-(1-hydroxyethyl)-Pheide a, and comparison of its absorption spectra (solid line) and that of Pheo a (dashed line). B, deduced structure of eluates in peaks 2 and 3, 3-acetyl-Pheide a, Bpheide a, and their C-12 methylated derivatives, and comparison of the absorption spectra of eluates in peak 2 and 3 (thick solid line), that of 3-acetyl-Pheo a (dashed line) and Bpheo a (dotted line), and calculated composite absorption spectrum of the two (B)Pheos (thin solid line). C, deduced structure of eluates in peaks 4 and 5, Pheide a, and its C-12 methylated derivative, and comparison of the absorption spectra of eluates in peaks 4 and 5 (solid line) and that of Pheo a (dashed line). Absorption spectra were normalized to the same values for Q_y peaks in A and C. Absorption spectra of the two (B)Pheos were normalized so that the sums of absorption at 682 and 750 nm were the same as those of the eluates. The absorbance of the two (B)Pheos were added to generate a calculated, composite spectrum of the mixture of the two (B)Pheos.

TABLE 2.

Molecular masses of pheophorbides in each elution peak of a HPLC analysis of pigments in C. tepidum ΔbciC mutant (Fig. 5A)

Peak no.	Elution time	Molecular mass(es)
	min	m/z
1	10.0	610
2	11.6	608, 610
3	12.4	622, 624
4	14.0	592
5	14.9	606

The 682 and 750 nm peaks in the absorption spectra of the compounds eluted in peaks 2 and 3 (Fig. 6B) seemed to represent the Q_y peaks of the mixture of two different compounds, a chlorin and a bacteriochlorin, rather than a Q_y peak and secondary peak of a single compound. HPLC-MS analyses confirmed this deduction. The molecular masses of the two major molecular ions in peak 2 differed by two mass units (608 and 610 Da), the expected difference between a double bond and a single bond. These two compounds eluted together, which indicated that these compounds had very similar hydrophobicity. Therefore, the reduced double bond was expected to be a CC bond rather than a CO bond; if the B-ring of a chlorin were reduced, a bacteriochlorin would be the expected product. Moreover, these compounds eluted earlier than Pheide a, which suggested that they should include an additional polar group compared with Pheide a (e.g. a keto or hydroxyl group). The 750 nm Q_y peak and a shoulder around 525 nm in the absorption spectrum (Fig. 6B) matched Bpheo a (32). Additionally, the molecular mass of Bpheide a (610 Da) was identical to that of one of the compounds detected in peak 2. Thus, it was concluded that one of the compounds in peak 2 was Bpheide a. The other compound, a chlorin with a mass of 608 Da and with the same side chains as Bpheide a, would thus be 3-acetyl-Pheide a. Its absorption spectrum, with a Q_y peak at 682 nm, was very similar to that reported for 3-acetyl-Pheo a (37). As described previously for peaks 4 and 5, the two compounds in peak 3 were 14 mass units heavier than those in peak 2. For the same reasons as described above, it was concluded that the two compounds in peak 3 were [12-Et]-Bpheide a and [3-acetyl, 12-Et]-Pheide a.

Except for small shifts in the positions of the Soret and Q_y peaks, the absorption spectrum of the compound in peak 1 was very similar to that of Pheide a. Its molecular mass (610 Da) was 18 Da more than that of Pheide a. Considering that this compound eluted first and was thus the most hydrophilic, it seemed likely that this compound would be a hydration product derived from Pheide a. C. tepidum has two hydratases, BchF and BchV, which could add water to the C-3 vinyl group of Pheide a to form a hydroxyethyl side chain on Chlide a in C. tepidum (3). It was thus tentatively concluded that peak 1 contained 3-(1-hydroxyethyl)-Pheide a.

To confirm these tentative identifications, (B)Pheos with the same head groups as each of the primary (B)Pheides (i.e. not the methylated derivatives) were synthesized and absorption spectra of these (B)Pheos were compared with those of the (B)Pheides. Pheo a was synthesized from Chl a, and its absorption spectrum was essentially identical to those of the compounds in peaks 4 and 5 (Fig. 6C), which were thus identified as Pheide a and [12-Et]-Pheide a. A similar result was observed for 3-(1-hydroxyethyl)-Pheo a, which was synthesized from BChl a, and the compound in peak 1 was thus confirmed to be 3-(1-hydroxyethyl)-Pheide a (Fig. 6A). Bpheo a and 3-acetyl Pheo a were synthesized from BChl a, and their absorption spectra were compared with those of the compounds in peaks 2 and 3 (Fig. 6B). The Q_y absorption peaks of Bpheo a and 3-acetyl Pheo a matched those of the two peaks of the compounds in these peaks with absorption maxima at 750 nm and 682 nm, respectively. Soret peaks at 358 and 410 nm and secondary peaks at around 525 nm of the two (B)Pheos were also observed in the spectra corresponding to peaks 2 and 3. The composite absorption spectrum of the two (B)Pheos was calculated, and this spectrum was very similar to the in-line absorption spectrum of a mixture of eluates. Small differences in the absorption spectra of (B)Pheides and the synthesized (B)Pheos probably were the result of solvent differences, which are known to affect the absorption properties of (B)Chls (34). In summary, the close agreement between the absorption spectra of the tentatively identified (B)Pheides and the chemically synthesized standards, together with the mass spectral data, confirm the structural assignments shown in Fig. 6.

The important and common feature of all of the various (B)Pheides accumulated in the mutant was the presence of the C-13²-methylcarboxyl group. To confirm the presence of this key side group, MS/MS analyses were performed on three selected ions of different compound classes (corresponding to the four types of (B)Pheides with different structures (see Fig. 6)). The primary fragment ions M⨥ of all three protonated positive ions [M + H]⁺ lost a mass of 60, which equals the mass of a methylcarboxyl group and a proton (Table 3). These results were consistent with previous fragmentation studies of tetrapyrroles with methylcarboxyl groups (38). These data strongly supported the conclusion that the product of ORF CT1077 was essential for the removal of the C-13²-methylcarboxyl group of Chlide a in C. tepidum.

TABLE 3.

MS/MS fragmentation results of protonated positive ions [M + H]⁺ of selected pheophorbides accumulated in C. tepidum ΔbciC mutant

Molecular mass of [M + H]+ (m/z)	Molecular mass of primary fragment (m/z)
593	533
609	549
611	551

Other Characteristics of ΔbciC Deletion Mutant

Previous studies have shown that mutants unable to synthesize BChl c grow much more slowly than wild-type cells at low light intensity (16, 17, 28). However, wild-type cells exhibit light saturation and grow more slowly at supersaturating light intensities, possibly because of photoinhibition due of exciton quenching in chlorosomes (28). The growth rate of the ΔbciC mutant was compared with that of the wild-type at light intensities between 93 and 307 μmol photons m⁻² s⁻¹. As shown in Table 4, the growth rate of the wild-type was maximal at ∼200 μmol photons m⁻² s⁻¹ and decreased at higher intensities. However, the growth rate of the ΔbciC mutant increased 4-fold over the same light intensity range (Table 4). The ΔbciC mutant grew ∼10-fold slower than the wild-type at the lowest light intensity but only grew 2-fold slower at the highest light intensity. It is possible that the growth rate of the ΔbciC mutant would have continued to increase at higher light intensities, but this was not tested. Very similar growth behavior as a function of light intensity was observed for a bchK mutant (28).

TABLE 4.

Growth rates of C. tepidum wild-type and ΔbciC mutant at different light intensities

Light intensities (μmol s−1 m−2)	Growth rate (h−1)
	Wild-type	ΔbciC mutant
93	0.27	0.025
155	0.28	0.055
241	0.26	0.071
307	0.20	0.10

Interestingly, the content of (B)Chls and (B)Pheides in the ΔbciC mutant changed with time as the culture was subcultured. Immediately after complete segregation of the mutant had occurred, the ΔbciC mutant synthesized much more Chl a than the wild-type. Moreover, most of the Chl a produced had phytol as the esterifying alcohol rather than Δ2,6-phytadienol as found in the wild-type (Fig. 7A). The mutant initially synthesized mainly Chl a_P instead of Chl a_PD, which was the usual Chl a in wild-type. The Chl a:BChl a ratio in the mutant in this initial growth stage was 1:2, compared with 1:10 in the wild-type. These changes in Chl a disappeared after three to five rounds of serial subculturing, and the (B)Chl a contents returned to the same levels as found in wild-type cells. The (B)Chl a contents remained stable after the initial three to five serial subcultures, but (B)Pheides accumulation did not. Although the types of (B)Pheides accumulated seemed to be the same, the mutant accumulated much less (<3%) (B)Pheides after >20 serial subcultures (Fig. 7B).

FIGURE 7.

Comparisons of pigment contents in C. tepidum ΔbciC mutant after different numbers of serial subcultures. A, comparison of partial HPLC profiles showing BChl a and Chl a contents in ΔbciC mutant cells immediately after complete segregation (dashed line) and those after three to five subcultures (solid line), which was very similar to wild-type (Table 1). Pheides accumulated in both type of cells were nearly identical (not shown). B, comparison of partial HPLC profiles showing Pheides accumulated in the mutant cells after three to five subcultures (solid line) and those after >20 subcultures (dotted line). BChl a and Chl a contents in both type of cells were identical (not shown). HPLC profiles in both panels were normalized to equivalent BChl a contents. ΔbciC mutant cells used in all other analyses were those subcultured three to five times after complete segregation.

CT1077 Replacement Mutant

A strain in which ORF CT1077 was replaced by its ortholog, Cabther_B0031, from Candidatus C. thermophilum and an aadA cassette for selection, was constructed (see Fig. 3B). After transformation and restreaking, the strain was verified by analytical PCR and DNA sequence analysis of the resulting amplicon. The (B)Chl content (Table 1) and the absorption spectra of whole cells (Fig. 4A) and extracted pigments (Fig. 4B) of the complemented strain in which CT1077 was replaced by Cabther_B0031—aadA were nearly identical to those of the wild-type. These data showed that the phenotype of the ΔCT1077 deletion mutant did not result from polarity effects of the aadA insertion on the adjacent upstream or downstream genes (see Fig. 3, A and B). These results also indicated that the Cabther_B0031 product was functional in C. tepidum. A second strain, in which CT1077 was replaced with the orthologous gene from Chloroherpeton thalassium, the earliest diverging member of the Chlorobiaceae (21), had the same characteristics (data not shown).

DISCUSSION

The ΔCT1077 deletion mutant of C. tepidum completely lacked BChl c and exclusively accumulated (B)Pheides with C-13²-methylcarboxyl groups. These results clearly demonstrated that the CT1077 gene product is essential for the removal of the C-13²-methylcarboxyl moiety from Chlide a, which is expected to be the first committed step in BChl c biosynthesis (3). Moreover, the complementation studies with CT1077 orthologs from Candidatus C. thermophilum and C. thalassium, whose sequences are highly divergent from that of CT1077 (supplemental Fig. 1), suggested that these gene products are probably sufficient for the removal of this moiety from Chlide a. Thus, to acknowledge the role of the CT1077 product in BChl c biosynthesis, ORF CT1077 was renamed bciC.

Despite repeated attempts using several different genes, all attempts to express bciC genes heterologously in Escherichia coli and Synechococcus sp. PCC 7002 were unsuccessful. Therefore, the specific enzymatic activity of BciC, i.e. whether it is a pheophorbidase (methylesterase) or a demethoxycarbonylase, could not be demonstrated directly. It is unclear why attempts to overexpress bciC failed, but the fact that BciC is predicted to be an integral membrane protein with five predicted trans-membrane α-helices (predicted by TMHMM version 2.0) might be a major contributing factor. However, the BciC ortholog from Candidatus C. thermophilum was able to replace the function of BciC in C. tepidum very effectively. The sequences of the BciC proteins from these two very distantly related organisms were only 28% identical (Candidatus C. thermophilum BciC). Furthermore, BciC from C. thalassium was only 48% identical to BciC of C. tepidum. Therefore, it seemed highly unlikely that BciC from Candidatus C. thermophilum could efficiently replace C. tepidum BciC, if BciC were to be a component of a multisubunit enzyme complex. Therefore, it seems likely that BciC alone is sufficient to catalyze the removal of the C-13²-methylcarboxyl group of Chlide a. It is also unlikely that BciC plays a regulatory role in BChl c biosynthesis. The ΔbciC mutant strain was not leaky; moreover, the ΔbciC mutant never exhibit phenotypic reversion despite very strong back-selection pressure from maintaining cells at low light intensity for more than two years. These and other observations are consistent with BciC playing an essential role in catalysis.

As discussed above, reactions similar to that shown in Fig. 1 occur during Chl degradation in plants and algae, and it is known that two mechanisms exist. In the first, a two-step process is initiated by a pheophorbidase, which demethylates the C-13²-methylcarboxyl group of Chlide a and thereby produces an intermediate that can spontaneously lose the C-13²-carboxyl group. In the second mechanism, a concerted reaction occurs, in which the C-13²-methylcarboxyl group of Chlide a is lost without the formation of a carboxylate intermediate. The reaction mechanism putatively catalyzed by BciC in the biosynthesis of BChl c is not known, and arguments for either mechanism can be made. Methylesterases are often members of recognizable protein superfamilies, and the observation that BciC has no homologs in the databases and that BciC protein family has no conserved serine or cysteine residues (supplemental Fig. 1) might indicate that it is a pheophorbide demethoxycarbonylase. On the other hand, the fact that BciC has no homologs in the genome of Chlamydomonas reinhardtii, in which pheophorbide demethoxycarbonylase activity has been reported (22), might indicate otherwise. Methylesterases often have rather simple catalytic requirements, but BciC could represent a completely novel family of such enzymes. Other than in bacteria that can synthesize BChl c, d, or e, a BciC homolog can currently be found in only one organism, the diatom Phaeodactylum tricornutum CCAP 1055/1, and interestingly, no plant-type pheophorbidase homolog exists in its genome. This suggests that BciC might replace the activity of a pheophorbidase in that organism. Because we have shown that BciC can be heterologously expressed in C. tepidum, further studies of the enzyme expressed in this organism will be required to determine the mechanism employed by BciC.

The C. tepidum ΔbciC mutant accumulated a variety of (B)Pheides. The identities of these (B)Pheides revealed information about the effects of C-13²-methylcarboxyl group on the substrate specificities of other BChl c biosynthesis enzymes. No (B)Chls or (B)Pheos esterified with farnesol were detected; this result shows that BchK, the BChl c synthase, can only esterify substrates that lack a C-13²-methylcarboxyl group. Because none of the detected (B)Pheides were methylated at the C-20 position, BchU is likewise unable to methylate substrates that have a C-13²-methylcarboxyl group. Interestingly, BchK and BchU are thought to catalyze the last two steps in the synthesis of BChl c (21). (B)Pheides methylated at the C-8² or C-12¹ positions were detected, but these methylated species were much less abundant than the unmethylated homologs, and the (B)Pheides generally carried only a single methyl group. This is very different from the situation for BChl c synthesis in wild-type cells, in which nearly all BChl c homologs are singly methylated at C-12¹ and most molecules also are singly or doubly methylated at C-8² (17). Thus, the C-13²-methylcarboxyl group in the substrates apparently lowered the activities of BchQ and BchR. From the analyses performed, it was not clear whether the (B)Pheides were predominantly methylated by BchQ, BchR, or both. However, because nearly all BChl c in wild-type cells is methylated at the C-12¹ position, but BChl c is not as extensively methylated at the C-8² position, it seems reasonable to assume the methylated (B)Pheides were mostly methylated at the C-12¹ position by BchR. Because they do not function exclusively in the BChl c biosynthesis pathway, it was not surprising that the C-13²-methylcarboxyl group had little or no effect on BchF and BchV (both are 3-vinyl-Bchlide hydratases).

The (B)Chl and (B)Pheide contents of the ΔbciC mutant changed as the strain was maintained in the laboratory. Initially, the mutant synthesized much more Chl a than the wild-type, and most of the Chl a was esterified with phytol rather than Δ2,6-phytadienol. The increased level of Chl a was anticipated, because a direct effect of eliminating BciC activity should be the accumulation of its substrate, Chlide a, which is also the substrate for the synthesis of Chl a by ChlG. The reduction of geranylgeranyl-PPi to form the esterifying alcohols of BChl a (i.e. phytol) and Chl a (i.e. Δ2,6-phytadienol) are only catalyzed by one enzyme, BchP, in C. tepidum (39, 40). Previous studies had suggested that such selective esterification might be achieved by strict substrate specificity of BchP or the (B)Chl synthases, BchG for BChl a and ChlG for Chl a. However, the observation that the ΔbciC mutant initially synthesized Chl a esterified with phytol strongly suggests that this hypothesis is incorrect. It remains unclear how wild-type C. tepidum and other related organisms manage to synthesize exclusively Chl a_PD and BChl a_P. For example, Harada et al. (40) suggested that another tail-modifying enzyme exists that would reintroduce a double bond into phytol. The concurrence of increased Chl a content, which was ∼5-fold higher than the level in wild-type cells, with the change in esterifying alcohol might indicate that the kinetics of the involved enzyme(s) and the availability of the substrate(s) could be the major contributing factor. Chl a_PD levels could exceed those required for the assembly of reaction centers, which would cause Chl a_PD to be available as a substrate for BchP for a much longer period of time. This could result in the reduction of Δ2,6-phytadienol to phytol. Together with previous results showing that BChG lacks substrate specificity for esterifying alcohols (39, 40), the presence of elevated levels of Chl a_P in the ΔbciC mutant strongly suggests that the multistep reduction of geranylgeranyl-PPi by BchP happens after, or both before and after, its esterification to (B)Chlide a by BchG or ChlG.

However, the mutant quickly adjusted to the ΔbciC mutation, probably because of rapidly selected secondary mutations, and could again synthesize (B)Chl a like the wild-type after only a few serial subcultures. This suggested that C. tepidum significantly favors Chl a_PD over Chl a_P in its reaction center. The ΔbciC mutant cultures no longer accumulated (B)Pheides after more than 20 serial subcultures. (B)Pheides are useless byproducts of tetrapyrroles that would naturally have been used to produce BChl c, and thus, they represent a large waste of resources for the mutant cells. It was interesting that the ΔbciC mutant was able to eliminate this wasteful metabolism in a relatively short period of time.

All organisms that can synthesize BChl c, d, or e, and whose genomes are available, have a homolog of bciC. Because the product of the bciC gene from the earliest diverging green sulfur bacterium, C. thalassium, was active in complementing the C. tepidum ΔbciC mutant, it is reasonable to assume that the products of bciC genes from all chlorophototrophic Chlorobi strains have the same activity. Furthermore, because BciC from a member of the phylum Acidobacteria had the same activity as C. tepidum BciC, it is likely that homologs in members of the Chloroflexi are also functionally orthologous. If this reasoning is correct, then all organisms that employ chlorosomes as light-harvesting antennae apparently use the same gene for the essential, first committed step in the synthesis of BChl c, d, and e from Chlide a. BciC sequences from Chloroflexi and Acidobacteria were the sequences that were most distantly related to BciC of members of the Chlorobi (<30% identity between sequences from different phyla). Thus, the phylogenetic relationships among BciC sequences are consistent with the relationships of the organisms and their phyla (supplemental Fig. 2), and it therefore seems unlikely that any of these genes were obtained through recent horizontal gene transfer. Supporting this assessment, gene neighborhoods surrounding bciC, which were conserved within a phylum, were completely different across members of the three relevant phyla. Notably, only in the genome of Candidatus C. thermophilum was bciC co-localized with other genes for (B)Chl biosynthesis. In 15 other organisms, including C. tepidum, the bciC gene was co-localized with genes completely unrelated to (B)Chl biosynthesis. This was one reason that CT1077 was not considered to be a probable candidate to encode the missing activity in previous attempts to identify this gene by phylogenetic profiling.

Similar to the recent identification of a variant 8-vinyl protochlorophyllide reductase (20), the recent and rapid increase in the availability of genomes of chlorophototrophs played a crucial role in the identification of bciC, the identification of which apparently completes the biosynthetic pathway for BChls c and d. By applying a combination of bioinformatic, genetic and biochemical approaches, the bciC gene product was shown to be essential for the first committed step in conversion of Chlide a into BChls c, d, and e. As more and more genomes of chlorophototrophic bacteria are sequenced, approaches similar to those described here should soon make it possible to answer the other outstanding questions concerning the biosynthesis of (B)Chls (e.g. how the C-7 formyl group is introduced into BChl e and how BChl b is synthesized).