Phytobilins are linear tetrapyrrole molecules synthesized by plants, algae, and cyanobacteria that function as the direct precursors of the chromophores of the light-harvesting phycobiliproteins and of the photoreceptor phytochrome (Beale, 1993; Hughes and Lamparter, 1999). The pathways of phytobilin biosynthesis have been elucidated by biochemical fractionation of plant and algal extracts, by overcoming a blocked step with exogenous putative intermediates, and by analysis of linear tetrapyrrole-deficient mutants (Beale and Cornejo, 1991a, 1991b, 1991c; Terry et al., 1993). These studies indicate that the biosynthesis of phytobilins shares common intermediates with heme and chlorophyll biosynthetic pathways to the level of protoporphyrin IX, at which point the latter two pathways diverge by metalation with iron or magnesium (Beale, 1993). Phytobilins are derived from heme, which is converted to biliverdin IXα (BV), the first committed intermediate in their biosynthesis. In red algae, cyanobacteria, and plants, this interconversion is accomplished by ferredoxin-dependent heme oxygenases that are related in sequence to the mammalian heme oxygenase (Cornejo et al., 1998; Davis et al., 1999; Muramoto et al., 1999). Although they catalyze the same reaction, mammalian heme oxygenases use an NADPH-dependent cytochrome P450 reductase to generate reducing power for heme catabolism (Maines, 1988).

The metabolic fate of BV differs in mammals, cyanobacteria, and plants, with BV being metabolized by different reductases with unique double-bond specificities (Figure 1). Mammalian biliverdin IXα reductase (BVR), an NAD(P)H-dependent enzyme that catalyzes the two-electron reduction of BV at the C10 methine bridge to produce bilirubin IXα (BR), was the first of these enzymes to be discovered (Maines and Trakshel, 1993). A similar enzyme, encoded by the gene bvdR, was identified recently in cyanobacteria (Schluchter and Glazer, 1997). Cyanobacteria and red algae also possess novel ferredoxin-dependent bilin reductases for the synthesis of the linear tetrapyrrole precursors of their phycobiliprotein light-harvesting antennae complexes (Beale and Cornejo, 1991a, 1991b, 1991c; Cornejo et al., 1998). Primarily on the basis of studies with the red alga Cyanidium caldarium, these investigators proposed that the biosynthesis of the two major phycobiliprotein chromophore precursors, phycoerythrobilin (PEB) and phycocyanobilin (PCB), requires two ferredoxin-dependent bilin reductases and several double-bond isomerases. The first bilin reductase catalyzes the two-electron reduction of BV at the C15 methine bridge to produce the BR isomer 15,16-dihydrobiliverdin (DHBV), whereas the second bilin reductase catalyzes the conversion of 15,16-DHBV to 3Z-PEB, a formal two-electron reduction of the C2 and C3¹ diene system. In C. caldarium, an additional enzyme mediates the isomerization of 3Z-PEB to 3Z-PCB, both of which appear to be isomerized to their corresponding 3E isomers before assembly with the nascent phycobiliprotein apoproteins (Beale and Cornejo, 1991a, 1991b, 1991c). More recent studies lend support for a similar pathway of PCB and PEB synthesis in cyanobacteria (Cornejo and Beale, 1997). In contrast with mammals and phycobiliprotein-containing organisms, plants and green algae reduce BV to 3Z-PΦB by the ferredox-in-dependent enzyme PΦB synthase, which targets the 2,3,3¹,3²-diene system for reduction (Terry et al., 1995; Wu et al., 1997). In plants, 3Z-PΦB is isomerized to its 3E isomer, which appears to be the immediate precursor of the phytochrome chromophore (Terry et al., 1995). The green alga Mesotaenium caldariorum possesses a second bilin reductase activity that catalyzes the reduction of the 18-vinyl group of PΦB to produce 3Z-PCB (Wu et al., 1997). These investigations also revealed that 3E-PCB is the natural phytochrome chromophore precursor in this organism.

Figure 1. Pathway of the Biosynthesis of Bilin Pigments.

Despite the extensive biochemical analysis of the phytobilin biosynthetic pathways in plants, algae, and cyanobacteria, the low levels of bilin reductase expression have hindered efforts to clone these enzymes. Using a genetic approach, Kohchi and colleagues (2001) recently cloned the HY2 locus of Arabidopsis, which encodes the enzyme PΦB synthase. The studies reported here were undertaken to identify HY2-related genes in the protein and nucleic acid databases. Using cloning, expression, and biochemical characterization, our investigations revealed three new classes of ferredoxin-dependent bilin reductases with either unique substrate or product specificities.

Figure 2. Multiple Sequence Alignment of the Identified Bilin Reductases.

On the basis of the biochemical data presented here, we name these HY2-related cyanobacterial loci after their roles in the biosynthesis of PCB (i.e., pcyA) and PEB (i.e., pebA and pebB). One of these proteins, the product of locus slr0116 (i.e., pcyA) in the genome of the cyanobacterium Synechocystis sp PCC6803, appears to be part of an operon with a putative response regulator located 62 bp upstream (Ashby and Mullineaux, 1999). Interestingly, this response regulator belongs to the OmpR subfamily for which a mutation (ycf27) was shown to cause a reduced energy transfer from the phycobilisomes to photosystem I (Ashby and Mullineaux, 1999). pcyA-related open reading frames (orfs) also were found in the oxyphotobacterium Prochlorococcus sp MED4 (CCMP1378), which is also known as Prochlorococcus marinus MED4, in the marine cyanobacterium Synechococcus sp WH8102, and in the nitrogen-fixing, heterocyst-forming filamentous cyanobacteria Anabaena sp PCC7120 and Nostoc punctiforme. Among the other identified HY2-related genes are two orfs, orf236 and orf257, from the marine cyanobacterium Synechococcus sp WH8020 that lie adjacent to each other within the major phycobiliprotein gene cluster (Wilbanks and Glazer, 1993a, 1993b). These orfs, which encode the proteins Ycp2_SYNPY and Ycp3_SYNPY, appear to be part of a three-gene operon containing an upstream orf of unknown function, orf200. A similar operon was identified in Synechococcus sp WH8102. The genomes of N. punctiforme and Prochlorococcus, both the MED4 and SS120 (CCMP1375) subspecies, also contain similar operons. In contrast to the N. punctiforme and Anabaena operons, an upstream orf in the Prochlorococcus operons exhibits a striking similarity to the ferredoxin-dependent heme oxygenase gene HY1 (Davis et al., 1999; Muramoto et al., 1999) and its homologs in the cyanobacterium Synechocystis sp PCC6803 (Cornejo et al., 1998). On the basis of their roles in PEB biosynthesis shown in this study, we name these orfs pebA and pebB.

PSI-BLAST iterations also identified a weak relatedness of HY2 to the red chlorophyll catabolite reductase (RCCR) from barley and Arabidopsis. RCCR is involved in chlorophyll catabolism and catalyzes the ferredoxin-dependent reduction of the linear tetrapyrrole, red chlorophyll catabolite (RCC), to yield the primary fluorescent chlorophyll catabolite (Wüthrich et al., 2000). These investigators showed that RCCR was incapable of reducing BV to either bilirubin or PΦB (Wüthrich et al., 2000). Interestingly, the sequence similarity between RCCR and the other HY2-related proteins is so weak that TBLASTN searches using the two RCCR sequences failed to identify HY2 or other HY2-related proteins present in the publicly available databases (Wüthrich et al., 2000). This divergence undoubtedly reflects the unusual substrate specificity of the RCCR for bilins derived from chlorophyll catabolism.

Phylogenetic analysis of the HY2-related family of proteins was performed using a heuristic parsimony search with a modified PAM250 weighting matrix and the program PAUP* version 4.0 (see Methods). A single tree obtained with this analysis (Figure 3) revealed four clades of HY2-related proteins with strong bootstrap support: PcyA, PebA, PebB, and RCCR. We noted that HY2 lies within in the PebB clade.

Figure 3. Phylogenetic Tree of the HY2 Family of Ferredoxin-Dependent Bilin Reductases.

Figure 4. SDS-PAGE of Affinity-Purified Bilin Reductases.

To determine whether the recombinant HY2-related proteins possess bilin reductase activity, we used a coupled holophytochrome assembly assay to analyze crude protein extracts from E. coli expressing these proteins for their ability to convert BV to phytobilins under standard PΦB synthase assay conditions (M.T. McDowell and J.C. Lagarias, manuscript submitted for publication). Figure 5A shows that crude bacterial lysates containing GST fusions of mHY2 and PcyA_SYNY3 (PcyA_ANASP; not shown) all exhibited BV reductase activities, yielding phytobilin products that could combine with the cyanobacterial phytochrome Cph1 apoprotein (apoCph1) to yield phytochrome difference spectra. The bilin metabolites incubated with apoCph1 resulted in different maxima and minima, suggesting that the various proteins reduced BV to distinct products. Both PcyA-containing extracts produced a BV metabolite(s) that gave spectra identical to those of the PCB adduct of apoCph1, with difference maxima at 655 nm and minima at 705 nm (Yeh et al., 1997). Figure 5A shows that both difference peaks of the mHY2 metabolites were markedly red shifted, with maxima at 670 and 730 nm, which is characteristic of the PΦB adduct of apoCph1 (Yeh et al., 1997; Kohchi et al., 2001). Identical results were obtained using the purified recombinant HY2 and PcyA proteins (data not shown). Similarly, E. coli extracts lacking HY2 or PcyA proteins failed to metabolize BV to bilin products that could functionally assemble with apoCph1 (data not shown).

Figure 5. Phytochrome Difference Spectra and Phytofluor Fluorescence Spectra of Recombinant Cyanobacterial Phytochrome (Cph1) Incubated with Reaction Metabolites.

In contrast to the results for PcyA and HY2, no phytochrome difference spectrum was observed when the BV metabolites from reactions containing PebA_SYNPY, PebB_SYNPY, or a 1:1 mixture of the two Synechococcus-derived proteins were incubated with apoCph1. To determine whether fusion to GST is responsible for inhibiting the enzyme activity of these proteins, GST was removed by protease digestion and the full-length proteins were purified (Figure 4). Neither the purified proteins nor the 1:1 mixture of PebA and PebB were able to convert BV to a bilin product(s) that yielded a photoactive adduct with apoCph1 (Figure 5A). The observation that coincubation of a 1:1 ratio of PebA and PebB with BV elicited a color change of the assay mixture from bluish-green to pink suggested that these proteins converted BV to bilins unable to form a photoconvertible holophytochrome. It is noteworthy that this pronounced color change was not observed when either PebA or PebB was assayed separately. This strongly implied that the PebA/PebB mixture could convert BV to PEB, the precursor of the phycobiliprotein C-phycoerythrin. To test this hypothesis, BV-derived bilin metabolites from PebA, PebB, and PebA/PebB were incubated with apoCph1, and the mixtures were analyzed spectrofluorometrically for the production of the fluorescent PEB–apoCph1 “phytofluor” adducts (Murphy and Lagarias, 1997). Only the PebA/PebB product mixture yielded a highly fluorescent compound, whose excitation and emission spectra were consistent with the formation of a phytofluor (Figure 5B). This result suggested that PebA and PebB were both required for the conversion of BV to PEB.

Figure 6. HPLC Analysis of the BV Metabolites of PebA, PebB, PcyA, and HY2 Bilin Reductases.

A mixture of PebA and PebB effectively converted all of the BV to two colored pigments, one purple (retention time 9 min) and the other pink (retention time 10.5 min), whose retention times differed from that of 15,16-DHBV (Figure 6). Both bilin metabolites have absorption maxima in acetone:20 mM formic acid (50:50, v/v) near 580 nm (data not shown). Because PebB could not metabolize BV, these results suggest either that the 15,16-DHBV product of PebA was metabolized by PebB to the purple and pink bilins or that PebB forms a complex with PebA to alter its product profile. That 15,16-DHBV was a substrate for PebB was demonstrated by incubation of PebB with HPLC-purified 15,16-DHBV. In this case, the same two bilin products were observed (data not shown). HPLC coelution experiments showed the purple and pink pigments to be the 3E and 3Z isomers of PEB, respectively (data not shown). Both pigments are chemically stable in the HPLC mobile phase, eluting as single peaks after purification and reinjection. Moreover, both HPLC-purified pigments form phytofluors upon incubation with apoCph1, indicating that these are configurational isomers of PEB. HPLC-purified 15,16-DHBV from the PebA-mediated reduction of BV, however, was unable to form a fluorescent adduct with apoCph1 (data not shown).

Biochemical studies of ferredoxin-dependent bilin reductases from algae and plants indicated that the 3Z isomers of PEB, PCB, and PΦB were the primary metabolites of these enzymes, with the formation of the 3E isomer requiring distinct bilin isomerase(s) (Beale and Cornejo, 1991a, 1991b, 1991c; Cornejo and Beale, 1997; M.T. McDowell and J.C. Lagarias, submittedmanuscript ). Our results show that both bilin isomers are produced with recombinant HY2, PcyA, and PebA/PebB proteins. We believe that the production of the 3E isomers occurred because of the presence of glutathione in the assay mixture and because of heating in the Speed-Vac concentrator. In this regard, glutathione-mediated 3Z to 3E isomerization of phycobilins has been reported for bilin reductases from C. caldarium (Beale and Cornejo, 1991c). Preliminary experiments performed with GST fusion proteins that did not come in contact with reduced glutathione or with proteins that were eluted from the affinity column by protease digestion greatly increased the relative amount of 3Z isomers produced (data not shown). Heating that occurred during concentration also contributed to the formation of the 3E isomers. If the drying time was reduced, only 3Z isomers were detected (data not shown). Therefore, we conclude that in all cases the 3Z isomers are the primary reaction product of these reductases and that the production of the 3E isomers occurs by non–enzyme-mediated side reactions caused by heat and reduced glutathione.

Using a combination of protein-based pattern searches of genomic databases, phylogenetic analysis, and biochemical characterization, these investigations establish that the HY2 family of ferredoxin-dependent bilin reductases can be subdivided into five classes: PcyA, PebA, PebB, HY2, and RCCR families (Figure 3). This classification system is supported by the distinct substrate preference and double-bond regiospecificity of representative members of each bilin reductase subfamily. PcyA, PebA, and HY2 all recognize BV as a substrate, yet each yields different bilin products. PebB and RCCR possess unique bilin substrates (i.e., 15,16-DHBV and RCC, respectively), and neither metabolizes BV (Wüthrich et al., 2000; this study). Biochemical analyses of representatives of the three new classes of bilin reductases identified here, PcyA, PebA, and PebB, document their involvement in the biosynthesis of the phycobiliprotein chromophore precursors PCB and PEB.

The MED4 and SS120 subspecies of the oxyphotobacteria Prochlorococcus also possess peb operons very similar to those of Synechococcus sp WH8020 and WH8102, except that the former possess upstream genes related to heme oxygenase. This strongly suggests that both oxyphotobacterial subspecies can synthesize PEB. In this regard, genes encoding the α and β subunits of a novel phycoerythrin have been identified in the SS120 subspecies of Prochlorococcus (Hess et al., 1996, 1999). It also has been shown that this unusual phycoerythrin plays a role in light harvesting in this ecotype (Lokstein et al., 1999), which is adapted for photoautotrophic growth at great ocean depths where light is limited. This observation is consistent with the lack of phycoerythrin genes in the high light–adapted MED4 ecotype. Although the enzymatic activities of Prochlorococcus PebA and PebB have not been determined experimentally, our phylogenetic reconstructions suggest that these proteins may be functional orthologs of the Synechococcus and Nostoc enzymes. Further analysis of the bilin biosynthetic pathways in Prochlorococcus and marine cyanobacteria such as Synechococcus sp WH8020 will be interesting, because the shorter wavelength–absorbing phycourobilin chromophores are major constituents of their phycoerythrins (Ong and Glazer, 1991; Hess et al., 1996). Although we have identified PCB and PEB biosynthetic enzymes in both organisms, it remains to be determined whether either these or other enzymes play a role in phycourobilin biosynthesis.

X-ray crystallographic analyses of phycobiliproteins have revealed that the stereochemistries of the thioether linkages to distinct cysteine residues are not all the same (Schirmer et al., 1987; Schmidt et al., 1987). Therefore, we propose that the different stereochemistries arise from the use of the 3Z and 3E isomers of the phycobilin precursor as substrates for assembly to distinct cysteinyl moieties. Beale and Cornejo (1991b) have identified a bilin isomerase that catalyzes the conversion of 3Z-PEB to 3Z-PCB in C. caldarium, which appears to be the sole pathway for PCB biosynthesis in this organism. More recently, a lyase/isomerase from the cyanobacterium Mastigocladus laminosus was described that is involved in both the isomerization of PCB to phycoviolobilin and its covalent attachment to apo-α-phycoerythrocyanin (Zhao et al., 2000). On the basis of these results and the diversity of bilin isomers found in phycobiliproteins from marine cyanobacteria, cryptomonads, and oxyphotobacteria (Ong and Glazer, 1991; Hess et al., 1996; Wedemayer et al., 1996), it is likely that numerous bilin isomerases are present in these oxygen-evolving photosynthetic organisms.

Unlike HY2 and PebB, members of the PebA family target the 15,16 double bond of BV for reduction. To evolve the PebA and PebB/HY2 subfamilies, we envisage a duplication of a ferredoxin-dependent BV reductase gene and subsequent divergence in a marine cyanobacterium growing in a light-limited environment. Such an environment would provide the selection pressure favoring evolution of the biosynthetic pathway for PEB, whose incorporation into phycoerythrin extends the light-harvesting wavelength range of their phycobilisomes. Depending on the rooting of the HY2 family tree, the comparative branch lengths of the PebA and PebB/HY2 families on the phylogenetic tree suggest that the A ring reductases are more ancient, with the 15,16 reductases evolving more recently. On the basis of these inferences, we speculate that a cyanobacterial progenitor of plant chloroplasts possessed a bilin reductase with an A ring reductase regiospecificity. The progenitor of present day cyanobacteria likely would have possessed the ability to synthesize PCB, an essential component of their allophycocyanin-containing phycobilisome core. Thus, the common pebA/pebB ancestor might have resembled present-day pcyA genes, which encode atypical BV reductases that catalyze the four-electron reduction of BV to PCB. To date, pcyA genes appear to be present in all cyanobacteria, whereas a peb operon is lacking in the phycoerythrin-deficient cyanobacterium Synechocystis sp PCC6803.

The role of the pebA, pebB, and pcyA genes in Prochlorococcus sp MED4 remains a mystery. Members of this genus are distinguished by the presence of integral membrane antennae complexes that contain divinyl chlorophyll a₂ and b₂ and by the lack of phycobilisomes (Partensky et al., 1999). Functional phycoerythrins have been detected only for the SS120 subspecies. As such, these organisms have been thought by some to be descendants of the class of prokaryotic photosynthetic organisms whose endosymbiosis led to higher plant chloroplasts. Phylogenetic analyses using 16S rRNA indicate that this probably is not the case, because Prochlorococcus species appear more similar to marine Synechococcus species than to chloroplasts (Urbach et al., 1998). These analyses also suggest that Prochlorococcus evolved more recently from a phycobilisome-containing ancestor that resembled a marine Synechococcus species. The need for pebA, pebB, and pcyA genes for phycobilin biosynthesis in this ancestor is self-evident, and such genes may not yet have been lost from Prochlorococcus species. It is conceivable that these BV reductases are required to make bilin chromophore precursors of light receptors, such as the phytochromes (Hughes and Lamparter, 1999). Although phytochrome-like genes are abundant in some cyanobacterial genomes, none are present in the genome of Prochlorococcus sp MED4 (data not shown). Alternatively, BV reductases may be needed to drive heme oxygenase, whose role in iron metabolism is well documented (Poss and Tonegawa, 1997a, 1997b; Richaud and Zabulon, 1997; Schmitt, 1997).

In addition to the bilin reductases involved in phytobilin biosynthesis, a separate class exists of bilin reductases that are involved in chlorophyll degradation (Hörtensteiner et al., 2000; Wüthrich et al., 2000). The pathway of chlorophyll degradation that occurs during plant senescence is similar to the heme degradation pathway (Matile and Hörtensteiner, 1999). After dephytylation and magnesium removal, the chlorophyll macrocycle ring is opened by a monooxygenase that has yet to be cloned (Hörtensteiner et al., 1998). This is followed by a ferredoxin-dependent reduction of the bilin product catalyzed by the RCCR (Hörtensteiner et al., 2000; Wüthrich et al., 2000). RCCRs are the most diverged members of the ferredoxin-dependent bilin reductase family. Indeed, these enzymes have markedly different substrate specificities. It is notable that RCCRs catalyze a reduction very similar to that mediated by the PebA family (i.e., a 15,16 double-bond reduction). The structural determinants that are responsible for RCCR's unique substrate specificity and double-bond regiospecificity will be interesting to discover. Presumably, chlorophyll catabolism would be important for chlorophyll-containing prokaryotes; however, to date, RCCR genes are not readily identifiable in the genomes of any photosynthetic prokaryotes. It is possible that RCCR genes were lost, or alternately, that they evolved more recently from an HY2-like gene in the chloroplast endosymbiont progenitor, because they are found in cryptogams and plants (Hörtensteiner et al., 2000).

Examination of highly conserved residues in the entire HY2 family and those within each of the five classes of bilin reductases provides information regarding residues important to the protein structure, ferredoxin interaction site, and substrate/product specificity. In this regard, only a small number of residues are conserved in the entire HY2 family of enzymes. These include hydrophobic residues at positions 137, 157, 158, 256, and 314, Pro-151, Phe-221, Ser-222, and Asp-171 (Figure 2). The notable lack of conserved basic residues suggests that the propionyl moieties of the bilin substrates do not form salt linkages with the enzymes. The conserved hydrophobic residues proline and phenylalanine are likely to be involved in overall protein structure (i.e., folding). Alternately, they may form hydrophobic interactions with conserved regions of the various bilin substrates. The loss-of-function hy2-1 and hy2-104 alleles of phytochromobilin synthase from Arabidopsis support the critical role of Pro-151 in HY2's structure. The conserved serine and aspartate residues likely play catalytic roles, such as hydrogen bonding with the substrate and/or substrate protonation to make the bound bilin a better electron acceptor.

Despite the wide divergence of the HY2 family, we believe that these conserved residues indicate that the active sites of all members of this class are similar. We speculate that the distinct double-bond reduction specificities of the BV reductases (i.e., PcyA, PebA, HY2), the 15,16-DHBV reductases (i.e., PebB), and the RCCR families reflect the positioning of the respective substrates within the catalytic pocket. Because the A/B and C/D rings of BV are very similar but not identical, it is conceivable that the substrate binding sites of the PebA and HY2 enzymes are tailored to position BV in opposite orientations, favoring electron transfer to the bilin C/D ring or A ring, respectively. If this is true, then the PebB class might tether its 15,16-DHBV substrate in an orientation similar to that of the HY2 family, whereas RCC might be bound to RCCR in a manner similar to that in which BV is bound to PebA. Future studies will address the unique substrate/product specificity using domain swapping, site-directed mutagenesis, synthetic biliverdin substrates, and x-ray crystallography.

We thank Dr. Alex Glazer for generously providing us with Synechococcus sp WH8020 cells, Dr. Wendy Schluchter for Anabaena sp PCC7120 genomic DNA, Elsie Campbell for N. punctiforme cells, and Drs. Jane Lamerdin and E. Teresa Janecki for Prochlorococcus sp MED4 DNA. We thank Dr. Gabrielle Rocap for helpful information regarding Prochlorococcus nomenclature and Dr. Bob Kuzoff for help with phylogenetic analysis. Preliminary sequence data were obtained from the DOE Joint Genome Institute (JGI) at http://spider.jgi-psf.org/JGI_microbial/html/. Thanks also are due to Beronda Montgomery and Jennifer Santos for critically reading the manuscript. This work was supported in part by the U.S. Department of Agriculture National Research Initiative Competitive Research Grants program (Grant No. AMD-9801768 to J.C.L.), the Research for the Future Program (Grant No. JSPS-RFTF 00L01605 to T.K.), a Grant-in-Aid for Scientific Research on Priority Areas from the Ministry of Education, Sciences, Sports, and Culture of Japan to T.K., and a research fellowship from the Deutsche Forschungsgemeinschaft to N.F. (FR 1487/1-1).