PEB (phycoerythrobilin) is one of the major open-chain tetrapyrrole molecules found in cyanobacterial light-harvesting phycobiliproteins. In these organisms, two enzymes of the ferredoxin-dependent bilin reductase family work in tandem to reduce BV (biliverdin IXα) to PEB. In contrast, a single cyanophage-encoded enzyme of the same family has been identified to catalyse the identical reaction. Using UV–visible and EPR spectroscopy we investigated the two individual cyanobacterial enzymes PebA [15,16-DHBV (dihydrobiliverdin):ferredoxin oxidoreductase] and PebB (PEB:ferredoxin oxidoreductase) showing that the two subsequent reactions catalysed by the phage enzyme PebS (PEB synthase) are clearly dissected in the cyanobacterial versions. Although a highly conserved aspartate residue is critical for both reductions, a second conserved aspartate residue is only involved in the A-ring reduction of the tetrapyrrole in PebB and PebS. The crystal structure of PebA from Synechococcus sp. WH8020 in complex with its substrate BV at a 1.55 Å (1 Å=0.1 nm) resolution revealed further insight into the understanding of enzyme evolution and function. Based on the structure it becomes obvious that in addition to the importance of certain catalytic residues, the shape of the active site and consequently the binding of the substrate highly determines the catalytic properties.
- bilin reductase
- EPR spectroscopy
- phycoerythrobilin synthase (PebS)
FDBRs (ferredoxin-dependent bilin reductases) are a class of enzymes involved in reducing the haem metabolite BV (biliverdin IXα) to form several individual open-chain tetrapyrroles (phycobilins) used for light-perception or light-harvesting in plants and cyanobacteria . FDBRs are distinct from BV reductases in mammals or cyanobacteria which are mainly involved in catabolic degradation of BV to bilirubin [2,3]. Currently several members of the FDBR family are known. The first cloned member was PΦB (phytochromobilin) synthase (HY2) from Arabidopsis thaliana, producing PΦB, the chromophore of the photoreceptor phytochrome . On the basis of the amino acid sequence of PΦB synthase, additional members of the FDBR family were identified [5,6]. Among these are enzymes, which likewise catalyse a two-electron reduction, but also two members that catalyse a formal four-electron reduction. The most common target for reduction within the FDBR family is the 2,3,31,32-diene system of the A-ring (Figure 1); however, only PΦB synthase acts on the substrate BV directly, thereby producing PΦB. All other A-ring reductions target intermediates of four-electron reductions leading to the cyanobacterial pigments PCB (phycocyanobilin) and PEB (phycoerythrobilin). Specifically, the second reduction performed by PcyA (PCB:ferredoxin oxidoreductase) targets the A-ring of 181,182-DHBV (dihydrobiliverdin), an isolatable intermediate in the reduction of BV to PCB . The previously identified cyanophage PebS (PEB synthase) on the other hand utilizes the intermediate 15,16-DHBV to produce PEB [6,8]. The identical A-ring reduction is also performed by the two-electron reducing PebB (PEB:ferredoxin oxidoreductase). This enzyme acts in tandem with PebA (15,16-DHBV:ferredoxin oxidoreductase), which reduces BV at the C-15–C-16 double bond to produce 15,16-DHBV [5,9]. Both enzymes are proposed to function in close contact, and metabolic channeling of 15,16-DHBV has been postulated . In contrast with this, PebS realizes a perfect metabolic channeling of the same intermediate by combining the two activities of PebA and PebB in one enzyme. Although the sequence identity between PebS and PebA is rather low (27%), they do, however, serve as a great paradigm of enzyme evolution and function. We previously presented the crystal structure and biochemical analysis of cyanophage PebS [8,10]. Like PcyA, the first crystallized member of the FDBR family [11,12], PebS shows an α/β/α-sandwich fold, with a central substrate-binding site parallel to the plane of the sheet . The binding of the substrate BV is rather flexible, which was reflected in the different binding modes observed. Concurrent with these different binding modes is also the flexibility of Asp206, one of two aspartate residues (Asp105 and Asp206) shown to be critical for the reaction. Although Asp206 seems to be important for the A-ring reduction activity of PebS, Asp105 is proposed to be involved in both enzymatic steps which both proceed via a tetrapyrrole radical intermediate . Interestingly, both residues are highly conserved within the whole FDBR family (Supplementary Figure S1 at http://www.BiochemJ.org/bj/439/bj4390257add.htm) and a common function has been discussed [5,10,13].
This present study has been undertaken to understand the structural and mechanistic details discriminating between the two cyanobacterial enzymes and the phage enzyme. Although PebS is able to reduce the intermediate 15,16-DHBV further to the final product, the PebA reaction is terminated at this point. PebB on the other hand has a distinct substrate specificity for the intermediate 15,16-DHBV. In an interdisciplinary approach we have combined X-ray crystallography with biochemical and biophysical characterization of WT (wild-type) and mutant proteins to gain a deeper understanding of the evolution of these three enzymes.
Unless otherwise specified, all chemical reagents were ACS grade or better. Glucose-6-phosphate dehydrogenase, NADP+, catalase, glucose, glucose 6-phosphate, spectrophotometric grade glycerol, trifluoroacetic acid, 4-methylmorpholine and amino acids were purchased from Sigma–Aldrich. HPLC grade formic acid, acetone and acetonitrile were purchased from J.T. Baker. BV was obtained from Frontier Scientific.
Production and purification of recombinant proteins
PebA and PebB from Synechococcus sp. WH8020 were expressed and purified, and the concentration was determined as described previously . The purified proteins contained eight additional N-terminal residues from the expression vector. SePebA [SeMet (selenomethionine)-labelled PebA] was produced using M9 minimal medium. At 15 min prior to induction with 100 μM IPTG (isopropyl β-D-thiogalactopyranoside), 100 mg/l lysine, phenylalanine and threonine, 50 mg/l isoleucine, valine and leucine, and 60 mg/l SeMet were added to the culture. Expression and purification for labelled protein followed the same procedure as for unlabelled PebA with the exception that all buffers contained 5 mM dithiothreitol. Ferredoxin from cyanophage P-SSM2 and FNR (ferredoxin:NADP+ oxidoreductase) from Synechococcus sp. WH7002 were prepared as described previously .
All protein variants were generated by site-directed mutagenesis from pGEX-6P-1_PebBSyn and pGEX-6P-1_PebBSyn  using the QuikChange® site-directed mutagenesis kit (Stratagene). The primers used are listed in Supplementary Table S1 (at http://www.BiochemJ.org/bj/439/bj4390257add.htm).
Anaerobic bilin reductase assay, EPR and HPLC measurements
Bilin reductase activity assays, EPR measurements, HPLC analyses and preparative production of 15,16-DHBV were performed under anaerobic conditions as described previously . To start the reaction, an NADPH-regenerating system was used . Only the reaction of the PebA_D84N variant was started with an excess of NADPH (10 electron equivalents, 50 μM final concentration) instead of the NADPH-regenerating system to reduce unspecific reduction of accumulating radical species. For PebB, the following modifications were used. The assay was performed at 20°C with 0.1 μM FNR and 25 units of catalase. After 2 min, 10 μM PebB_WT or PebB variant was added to the reaction mixture. The reaction was stopped after a total of 10 min.
EPR measurements were performed as described previously . The following modifications were applied only for the PebB reaction. The reaction was performed at 20°C with 37.5 units of catalase, 0.032 μM FNR and 4 μM ferredoxin using PebA_WT and BV as the substrates. After 2 min, PebB was added to the PebA–DHBV complex at a 1:1 ratio.
Crystallization conditions were screened by the sitting-drop vapour diffusion method utilizing the Cryos, PEG and PACT Suites (Qiagen) applying 200/100 nl and 100/100 nl mixtures of the protein solution (10–20 mg/ml)/reservoir solution incubated at 18°C in the dark. Conditions were further optimized with the hanging-drop vapour diffusion method. Substrate BV was added in a 2-fold excess. PebA crystallized at protein concentrations between 12 and 20 mg/ml in 0.1 M Hepes (pH 7) and 28% PEG [poly(ethylene glycol)] 4000.
Structure determination and refinement
Oscillation data of SeMet-labelled protein crystals were collected at 100K at the SLS (Swiss Light Source) on beamline X10SA. Data were processed using XDS . As a test set, 5% of the data were randomly assigned. Data statistics are given in Table 1. The Matthews coefficient was estimated to 2.1 Å3/Da for one molecule per asymmetric unit (1 Å=0.1 nm).
Phases were determined from a 2 Å dataset collected at the Se-K-edge. AutoSharp readily located the two Se-atoms in the asymmetric unit . ARP/wARP was used to autotrace PebA in the resulting map . This model was then used with ARP/wARP to rebuild the model with the second 1.55 Å PebA dataset. The resulting model was improved using alternating cycles of manual correction in COOT  and automatic refinement in PHENIX . The first six residues from the expression vector, residues 128 and 129, and the last three residues were not modelled due to missing density. The model has been deposited at the PDB under accession number 3X9O.
RESULTS AND DISCUSSION
All members of the FDBR family are radical enzymes
Cyanobacteria use the dual enzyme system PebA and PebB to produce the phycobilin PEB, whereas a single enzyme PebS encoded by a cyanophage does the same job. We used comparative enzymology to understand the similarities and differences between PebA, PebB and PebS which are homologous with each other and which all belong to the FDBR family. This work was furthermore intended to gain insight into enzyme evolution. Although PebA and PebB have in parts already been studied biochemically , we reinvestigated their enzymatic properties under anaerobic conditions. This is essential to detect and stabilize possible tetrapyrrole radical intermediates, as shown for other members of the FDBR family, including the previously investigated cyanophage PebS [10,13,19]. In contrast with the aerobic reduction of BV to 15,16-DHBV by PebA, which showed a decrease of absorption at 690 nm and a concomitant increase at 590 nm , two additional absorption maxima are observed under anaerobic conditions. In the course of the reaction an increase and further decrease at ~440 nm and ~750 nm was monitored (Figure 2A, top panel). Absorbance at these wavelengths has been attributed to bilin radical intermediates [10,13,19]. EPR studies confirmed this assumption. For PebA, the strongest EPR signal was observed 1 min after the start of the reaction (Figure 2B), when no product was yet detectable via HPLC (results not shown). The appearance of this signal followed similar kinetics as the absorption at ~440 nm and ~750 nm. In addition, the decrease of radical signal was accompanied by an increase of product formation, indicating that the radical observed is a BV radical.
For the first time we monitored the PebB reaction which uses 15,16-DHBV as a substrate. In order to facilitate proper 15,16-DHBV delivery, a completed PebA reaction was employed. This was preferred over external supply of 15,16-DHBV because metabolic channeling of 15,16-DHBV from PebA to PebB has been postulated and might be involved in proper delivery and subsequent binding of 15,16-DHBV . This is underlined by the observation that 15,16-DHBV externally supplied to the PebS_D206N variant binds differently than the intermediate produced by itself . Owing to the lack of a molar absorption coefficient and the instability of 15,16-DHBV, this approach provides an easy and accurate way to supply PebB with equimolar amounts of substrate. First, a regular PebA reaction employing a PebA–BV complex was used. After 2 min almost all BV was converted into 15,16-DHBV, which remained bound to PebA (PebA–DHBV). Subsequently, PebB_WT was added leading to an immediate transfer and binding of 15,16-DHBV to PebB (PebB–DHBV) as observed by a shift of 15,16-DHBV absorbance from 590 nm (PebA–DHBV) to ~605 nm (PebB–DHBV) (Figure 3A and Supplementary Figure S2 at http://www.BiochemJ.org/bj/439/bj4390257add.htm). Almost simultaneously a strong increase at 682 nm was observed, most probably representing the formation of radical intermediates which were again confirmed by EPR measurements (Figure 3B). A small EPR signal was still detected at 2 min due to a not fully completed PebA reaction. Addition of PebB_WT resulted in an increased radical signal, which disappeared with the completion of the reaction at 10 min. At this time point the formation of PebB-bound 3Z-PEB (PebB–PEB) was detected at 548 nm (Supplementary Figure S2). This is in agreement with the PebS reaction where the final product is also 3Z-PEB . Herewith we prove that all different members of FDBRs act via radical intermediates.
Identification of critical residues for catalytic activity
Central to the proposed reaction mechanism of PebS are two aspartate residues Asp105 and Asp206, both involved in interactions with the pyrrole nitrogens upon substrate binding . Both are shown to be essential for the complete reduction of BV to PEB by PebS and are highly conserved throughout the family of FDBRs (Supplementary Figure S1). To study the role of the corresponding residues in PebA and PebB (Table 1), several variants were characterized in anaerobic bilin reductase assays. The resulting products were analysed by HPLC and UV–visible spectroscopy.
When the aspartate residue at position 84 in PebA is changed into an asparagine residue, no conversion of BV is observed. Interestingly, the D84N variant appears to stabilize a radical intermediate, as indicated by a fast increase and very low decrease of an absorption maximum at 740 nm (Figure 2A). When the carboxylic side chain is retained, but exchanged by a longer side chain which is expected to sterically interfere with substrate binding, the resultant PebA variant D84E is still able to convert BV into 15,16-DHBV at a similar efficiency (Figure 2A). However, the 15,16-DHBV produced is significantly more unstable (results not shown), suggesting that Asp84 is also involved in stabilization of the enzyme–product complex. This catalytic behaviour is in agreement with data of the PebS_D105N/E variants which showed the same properties for the first reduction .
Similar results were obtained for the asparagine variant of the homologous amino acid residue in PebB (PebB_D107N). A stabilization of a radical intermediate was observed both in UV–visible (increase of absorption at 670 nm) and EPR spectroscopy (Figure 3). Consequently, no product formation was detected (results not shown). Interestingly, retaining the carboxylic side chain of this residue in a glutamate variant (PebB_D107E) did not rescue the activity, indicating that space constraints might be crucial for reduction. However, this variant is still able to bind 15,16-DHBV in a manner similar to the WT protein (Supplementary Figure S2) and trace amounts of the product PEB were monitored by HPLC (Table 1).
When Asp205 is changed into an asparagine residue, PebA retains its activity. The D205N variant converted its substrate BV into 15,16-DHBV (Figure 2A). In contrast, the homologous variant of PebB (PebB_D231N) showed a complete loss of PEB formation. An increase of absorption at 666 nm with no further decay again suggested the stabilization of a radical intermediate (Figure 3A). The EPR signal observed increased over time with the highest signal intensity at 10 min (Supplementary Figure S3 at http://www.BiochemJ.org/bj/439/bj4390257add.htm). PebB_D231E on the other hand showed only slightly decreased activity with significant product formation.
The data presented clearly demonstrate that the individual activities of PebA and PebB with their dependency on important catalytic residues are combined in the phage enzyme PebS. Specifically, Asp105 is important for both reductions, whereas Asp206, although conserved, is only critical for A-ring reduction .
Overall structure of PebA
In order to determine the structural differences between PebS and PebA, which only catalyses the first reduction of BV to 15,16-DHBV, the structure of SeMet-labelled PebA with BV IXα was solved by the single wavelength anomalous dispersion method and was refined at a 1.55 Å resolution (Table 2). Six residues of the linker peptide used in the expression construct and the last three residues of the mature protein were not modelled due to missing electron density. In addition, the two residues Asn128 and Gly129 could not be placed for the same reason. Clear density for the substrate allowed unambiguous placing of BV into the binding site. Identification of the correct orientation was possible due to the asymmetric vinyl substituents of the A- and D-ring.
The structure of PebA represents the first structural view of an FDBR catalysing a two-electron reduction. Together with the different structures available for the four-electron reduction enzymes PcyA and PebS, we are now able to make a clearer discrimination between structural features preserved throughout the family and features with distinct differences, which might be key elements in controlling the stereoselective reactions. The overall structure consists of a central seven-stranded antiparallel β-sheet, which is, from both sides, flanked by a total of six α-helices (Figure 4). All structurally solved FDBRs share the presence of this central seven-stranded β-sheet [8,11] which forms a mostly hydrophobic basis of the substrate-binding site. In addition, the binding pocket is always formed by two long helices (PebA:H5–H6, PebS:H3–H4 and PcyA:H7–H8) connected by the ‘D-loop’ (residues connecting H5 and H6). The position of these helices is similar in all structures, but variations of the linker geometry occur between enzymes and upon substrate binding. Indeed, PebA can be superimposed with low RMSDs (root mean square deviations) of 2.02 Å and 2.39 Å to PebS (PDB code 2VCK, chain C) and PcyA (PDB code 2D1E) respectively. The largest structural variations are located in the ‘lid’ (loop between helix H3 and strand S7 and by helices H5 and H6) in response to substrate binding in both PebS and PcyA (Figure 4) [8,20]. In addition, large differences in length and folding of the lid region exist between the different FDBRs. In PebA and PcyA helical elements are found, which are completely missing in PebS (Supplementary Figure S4 at http://www.BiochemJ.org/bj/439/bj4390257add.htm). Both the changes upon substrate binding and differences in structure make this area the most promising target for interaction of PebA with both ferredoxin (as proposed in ) and PebB.
The most prominent feature of the binding pocket is the central polar centering ‘pin’, formed by Asp84 (PebA numbering used throughout for simplicity reasons) and the adjacent Asn67 on strands S5 and S4 respectively (Figures 5 and 6, and Supplementary Figure S5 at http://www.BiochemJ.org/bj/439/bj4390257add.htm). Both residues are involved in co-ordination of the polar pyrrol nitrogens and the carbonyl oxygens (Figures 5 and 6). Only for HY2, Asp84 is replaced by an asparagine residue, while at the same time the neighbouring Asn67 is changed to aspartate, thereby restoring the placement of a functional carboxylic group at this position . The importance of both residues for the HY2 catalytic mechanism is not yet fully understood and awaits detailed biophysical and structural analyses.
Inside the binding niche of PebA six water molecules are resolved. The BV propionate side chains are co-ordinated by salt bridges with three positively charged protein side chains (Lys93, Arg134 and Arg150). In contrast with PcyA and PebS, BV in PebA is bound in a roofed conformation, with both the A- and D-ring tilted 40° out of the plane and the A-ring buried deeper inside the pocket. The catalytic residue Asp84 is in an identical orientation in PebA as compared with Asp105 in PebS and PcyA and is therefore not the structural reason for the roofed BV binding. The other conserved aspartate residue Asp205 which is of catalytic importance for PebS, but has been shown to adopt different conformations in PebS, is rotated away from the active site in PebA. It is the first of three residues of the D-loop. This residue is not involved in direct ligand interaction, which is consistent with its non-catalytic role in PebA.
Structural flexibility of the active site is required for two consecutive reductions within one FDBR
Based on the structures and on the sequence alignment (Supplementary Figure S1), the binding pockets of all FDBRs are lined with hydrophobic and aromatic residues, making van der Waals and π-stacking interactions the dominating factor for substrate binding. Very few residues in the active site are chemically able to participate in substrate protonation and are found to be highly conserved amongst FDBRs. Also the general binding mode with the A-ring buried deeper seems to be identical throughout the FDBR family, regardless of the site of reduction (Supplementary Figure S5).
Considering the similarities between PebS and PebA, the question arises, which structural features of PebA are responsible for the termination of the BV reduction at the intermediate 15,16-DHBV, whereas it is directly processed to PEB in PebS? The binding pocket of PebA forces BV into a strained roofed conformation (Figure 5). This conformation results from steric restraints imposed by size variation of apolar protein side chains. For example, substitution of Met212 by a phenylalanine residue in combination with the substitution of Iso86 by valine in PebA forces the rotation of the D-ring of BV (Figure 6 and Supplementary Figure S5). In contrast, in PebS the pocket is much larger, leaving room for the tilting of the D-ring after reduction of the C-15–C-16 double bond to form the non-planar product 15,16-DHBV (Figure 6). It has been proposed that rearrangement of 15,16-DHBV in PebS is required for stereospecific reduction of the A-ring. Even though a structure of PebS with the bound intermediate 15,16-DHBV is still missing, the observed binding flexibility for BV supports such a rearrangement . In contrast, very little conformational variation for a variety of substrates has been observed in PcyA [11,20,22]. The intermediate 181,182-DHBV of the PcyA reduction still retains the conjugated π-system of BV, which extends over all four tetrapyrrole rings. Therefore no large structural changes have to be accomodated during the two successive reductions (Figure 6 and Supplementary Figure S5).
Asp205 is critical for A-ring reduction
On the basis of the results of the present study, and other biochemical and structural data [10–12,23] it is obvious that the conserved Asp84 residue is highly important for bilin reduction in PcyA, PebS, PebA and PebB. The only exception so far seems to be the above-mentioned HY2 protein where double exchange of an aspartate/asparagine pair occurs. Asp84 is well positioned to facilitate stereospecific protonation to both C-2 and C-16 to generate the R configuration at these carbon atoms. In contrast with the general importance of Asp84, Asp205, although conserved, is only important for A-ring reduction of BV (in HY2) and 15,16-DHBV (in PebS and PebB). For PebS we previously postulated a role for Asp205 as a proton shuttle to supply the protons for the second reduction. In PebA this proton shuttle is not required since only a two-electron reduction is catalysed, representing the first reduction in PebS. In both cases the active site seems to be able to supply the two protons necessary for the reaction either from Asp84 and/or water molecule(s). For the second reduction PebS requires additional delivery of protons from the surrounding medium as the intermediate stays tightly bound in the active site. Asp84 is proposed to fulfil this function. In contrast, it is hypothesized that PebB binds 15,16-DHBV into its protonated active site, analogous to BV binding to PebA and PebS, alleviating the need for an additional proton shuttle. We would expect at least residual function of the PebB_D231N variant if this residue is only relevant for proton delivery in PebB. Therefore an additional function of Asp205 in substrate co-ordination is required to fully explain the results of the present study. We believe that Asp205 will be involved in direct co-ordination of 15,16-DHBV, thereby affecting local reactivity of the substrate to support stereospecific protonation by Asp84. In turn, these data suggest a similar function of Asp205 also for PebS. As implicated by our biochemical data these two aspartate residues are similarly positioned in a PebB model (Supplementary Figure S5). Full understanding of these processes will require structures of PebA, PebB and PebS with bound 15,16-DHBV, as well as high-field EPR characterization of the substrate radicals. These experiments are currently underway in our laboratories.
FDBRs are a great paradigm to study enzyme evolution
On the basis of numerous biochemical and structural data on members of the FDBR family it becomes obvious that nature has retained one general fold to evolve several distinct enzymatic functions. Even key catalytic residues have been retained (i.e. Asp84 and Asp205), but still different reduction sites of the same or a similar substrate are targeted. This seems to be basically facilitated through the very specific positioning of the substrate molecule and substrate rearrangement after initial reduction in the four-electron reducing FDBRs. On the basis of our current knowledge we are quite confident that other FDBRs with different regiospecificities do exist in nature or can be engineered using directed evolution.
Andrea Busch and Nicole Frankenberg-Dinkel designed the study; Andrea Busch performed all of the biochemical experiments; Andrea Busch and Edward Reijerse performed the EPR experiments; Andrea Busch, Edward Reijerse and Wolfgang Lubitz analysed the EPR data; Eckhard Hofmann solved the crystal structure; Andrea Busch, Eckhard Hofmann and Nicole Frankenberg-Dinkel analysed the biochemical data; Andrea Busch, Eckhard Hofmann and Nicole Frankenberg-Dinkel wrote the paper.
This work was supported by the SFB 480 [Teilprojekt C6 and C8 (to N.F.D. and E.H.); and the Deutsche Forschungsgemeinschaft [grant number FR 148716-1 (to N.F.D.)]. A.W.U.B. received a Ph.D. fellowship from the Ruhr-University Bochum Research School.
Abbreviations: BV, biliverdin IXα; DHBV, dihydrobiliverdin; FDBR, ferredoxin-dependent bilin reductase; FNR, ferredoxin:NADP+ oxidoreductase; PΦB, phytochromobilin; PCB, phycocyanobilin; PcyA, PCB:ferredoxin oxidoreductase; PEB, phycoerythrobilin; PebA, 15,16-DHBV:ferredoxin oxidoreductase; PebB, PEB:ferredoxin oxidoreductase; PebS, PEB synthase; RMSD, root mean square deviation; SeMet, selenomethionine; WT, wild-type
- © The Authors Journal compilation © 2011 Biochemical Society