Mechanistic Studies of the Phytochromobilin Synthase HY2 from Arabidopsis*

Phytochromobilin (PΦB) is an open chain tetrapyrrole molecule that functions as the chromophore of light-sensing phytochromes in plants. Derived from heme, PΦB is synthesized through an open chain tetrapyrrole intermediate, biliverdin IXα (BV), in the biosynthesis pathway. BV is subsequently reduced by the PΦB synthase HY2 in plants. HY2 is a ferredoxin-dependent bilin reductase that catalyzes the reduction of the A-ring 2,3,31,32-diene system to produce an ethylidene group for assembly with apophytochromes. In this study, we sought to determine the catalytic mechanism of HY2. Data from UV-visible and EPR spectroscopy showed that the HY2-catalyzed BV reaction proceeds via a transient radical intermediate. Site-directed mutagenesis showed several ionizable residues that are involved in the catalytic steps. Detailed analysis of these site-directed mutants highlighted a pair of aspartate residues central to proton donation and substrate positioning. A mechanistic prediction for the HY2 reaction is proposed. These results support the hypothesis that ferredoxin-dependent bilin reductases reduce BV through a radical mechanism, but their double bond specificity is decided by strategic placement of different proton-donating residues surrounding the bilin substrate in the active sites.

Phytochromobilin (P⌽B) 2 is an open chain tetrapyrrole chromophore critical for light-sensing phytochromes to regulate growth and development of plants. The phytochromebound P⌽B absorbs light energy and proceeds with reversible structural rearrangement to alter the biochemical activities of phytochrome. P⌽B is covalently linked to apophytochrome through a thioether bond between a conserved cysteine residue on the apoprotein and the ethylidene group on the A-ring of P⌽B (1). The biosynthesis of P⌽B has been shown to reside in the plastids, where heme is first linearized by a heme oxygenase into the reaction intermediate biliverdin IX␣ (BV) and then subsequently reduced by a P⌽B synthase (2)(3)(4)(5).
PcyA is the most extensively studied FDBR enzyme. Previous biochemical analysis has identified a two-electron reduced intermediate, 18 1 ,18 2 -dihydrobiliverdin IX␣, present in the PcyA reaction, indicating that D-ring reduction precedes A-ring reduction ( Fig. 1B) (10). Two organic radical intermediates have also been detected in the two double bond reduction steps, strongly supporting the proposed mechanism for sequential electron-coupled proton transfer in the PcyA-catalyzed BV reduction (11). Mutagenesis studies have shown a histidine/aspartate ion pair playing the essential role in the active site, which is further confirmed by the crystal structure of PcyA (12)(13)(14). Combining the structural information and biochemical data, a more detailed mechanism for PcyA reaction has been proposed (12). The histidine/aspartate ion pair functions as the central proton provider, which first works together with a glutamate residue to donate two protons to the D-ring exo-vinyl group and then with water molecules to mediate proton transfers from two possible proton relays to the A-ring 2,3,3 1 ,3 2 -diene system.
In contrast to the understanding of PcyA, the catalytic mechanism of other FDBRs is still not well elucidated. With the activity that is common in most FDBRs, HY2 is the ideal candidate to investigate A-ring 2,3,3 1 ,3 2 -diene reduction (ethylidene group formation), as well as the global mechanism for FDBR catalysis. Here, we report that an organic radical species can be detected in HY2-catalyzed A-ring 2,3,3 1 ,3 2 -diene reduction, suggesting that BV reduction catalyzed by HY2 proceeds as a series of electron transfers from ferredoxin accompanied by proton donations from the enzyme. The results also reveal that the radical mechanism is universal to FDBRs. Mutagenesis identified the conserved Asp 256 residue as being critical for protonation. Another aspartate residue, Asp 116 , is important for substrate positioning in the active site. Based on the biochemical information, a catalytic mechanism for HY2-catalyzed BV reduction is proposed.
Protein Expression, Purification, and Site-directed Mutagenesis-For expressing HY2 protein, the Arabidopsis mHY2 coding region without the predicted N-terminal transit peptide was subcloned into the Escherichia coli expression vector pTYB12 (New England Biolabs) to construct pTYB12-mHY2 (2). E. coli strain BL21 containing pTYB12-mHY2 was generated to express chitin-binding domain-mHY2. The bacterial cells were grown at 37°C in 500-ml batches of Luria-Bertani medium containing ampicillin (100 g/ml) to A 600 ϭ 0.6 -1.0. Cultures were induced by the addition of 0.1 mM isopropyl ␤-D-thiogalactopyranoside, incubated overnight at 16°C, and harvested subsequently by centrifugation. Tag-free mHY2 was purified according to instructions supplied by the manufacturer (New England Biolabs). Tag-free mHY2 protein was further purified with a Superdex 200 size exclusion column pre-equilibrated with 25 mM TES-KOH (pH 8.5), 100 mM KCl, and 10% (v/v) glycerol. All site-directed mutants were generated in pTYB12-mHY2 using the QuikChange site-directed mutagenesis kit (Stratagene). Mutant mHY2 proteins were expressed and purified following a method similar to that used for the wild-type protein.
Expression and purification of recombinant Nostoc sp. PCC 7120 PcyA were performed as described previously (10). Recombinant Synechococcus sp. PCC 7002 ferredoxin (Fd) was purified as described previously (16). Purified Fd was dialyzed against 25 mM TES-KOH (pH 8.5) and 100 mM KCl. HY2, PcyA, and Fd preparations were flash-frozen in liquid nitrogen and stored at Ϫ80°C prior to use.
Enzyme Assay and Spectroscopic Analysis-Steady-state bilin reductase assays were performed similarly to those described previously (10)   subsequently evaporated to dryness using a SpeedVac concentrator. HPLC analysis and anaerobic single turnover mHY2 assays were performed as described previously (11) with a slight modification. For single turnover assay, the concentration of FNR was increased from 0.0025 to 0.00625 units/ml. UV-visible absorption and freeze-quench EPR spectroscopic measurements were performed as described previously (11). For measuring the UV-visible spectra of BV complexes of wildtype HY2, PcyA, and site-directed mutants, 10 M BV and enzymes were mixed and incubated at 25°C for 5 min prior to measurement. To compare the efficiency of BV binding between wild-type HY2 and site-directed mutants, the long-toshort wavelength absorption ratios (A 650 /A 380 ) of enzyme-BV complexes were measured. Ratios for free and wild-type HY2bound BV were set as 0 and 100%, respectively.
For EPR measurements, the single turnover assay was used with the exception that the concentrations of mHY2, BV, Fd, and FNR were all increased 4-fold (to 40 M, 40 M, 40 M, and 0.025 units/ml, respectively), and 10% spectroanalytical grade glycerol was included in the reaction mixtures. At various time points after the addition of NADPH, 200-l aliquots were withdrawn from reaction mixtures, transferred to 4-mm quartz EPR tubes, and immediately frozen in liquid nitrogen. Continuous wave EPR studies of mHY2 were performed using a Bruker E580 spectrometer. EPR spectra at 70 K were acquired using an X-band microwave frequency of 9.50 GHz with a modulation amplitude of 4 G. CD analysis was performed on a Jasco J-715 CD spectrophotometer. CD spectra were recorded at room temperature under constant nitrogen gas flow with a 0.1-cm sample light pass.
Structure Simulation and Modeling-The structure of PebS with bound BV (Protein Data Bank code 2VCK) (see accompanying article (26)) was used as a template to generate holoprotein homology models of residues 48 -283 of Arabidopsis HY2 using the multiple sequence alignment of FDBRs and the Discovery Studio 2.0 program package (Accelrys). The resulting binary model for HY2 was manually edited in coot to lessen steric clashes and then subjected to energy minimization in Discovery Studio 2.0 (17).

Absorption Spectrum of HY2-bound BV Implies That the Conformation Is Different Compared with PcyA-bound BV
Molecule-Absorption of bilins is easily affected by their conformation, protonation state, and chemical environment (18). In the case of BV and as shown in Fig. 2, free BV gives a low long-to-short wavelength absorption ratio of 0.27 in solution. PcyA-bound BV has been predicted to adopt a cyclic porphyrin-like configuration based on the observation of an absorption increase at a long wavelength (ratio of 0.48) and a blue shift in the absorption maximum to 650 nm (10). This is confirmed by the crystal structure of the PcyA-BV binary structure (14). The absorption change in the long wavelength band could be primarily from the change in both the conformation and protonation state of bound BV. So far, many ionizable residues in the active site of PcyA have been selected to investigate their functions in binding and catalysis. Critical residues have been identified (12,13). None of the mutations of these critical resi-dues give a significant loss of the 650 nm band. We therefore believe that the long wavelength absorption is due mainly to the conformational change upon binding. In contrast to PcyA, HY2-bound BV showed a similar blue shift at a long wavelength but with a larger absorption, giving a long-to-short absorption ratio of 0.73. The results indicate that HY2-bound BV is cyclic but with a slightly extended conformation compared with PcyA-bound BV.
The 740 nm near-infrared absorption observed in the spectrum of PcyA-bound BV has been shown to be an indication of the primary protonation of the BV substrate, which is the determinant for subsequent proton transfer steps (11,13). However, the 740 nm absorption was not detected in the absorption spectrum of HY2-bound BV. This suggests that the initial protonation step in the catalytic reaction of HY2 may be different from that in PcyA.
Spectral Change upon the Addition of Electrons to the HY2-BV Complex Indicates the Formation of a Transient Radical Intermediate-We then monitored the HY2 catalytic reaction by UV-visible spectroscopy with anaerobic bilin reductase assay by sequentially adding 1 eq of electron (i.e. 5 M NADPH) to 10 M HY2-BV complex (Fig. 3A). After adding the first equivalent of electrons (1 eq of e Ϫ ), a significant spectral change was observed mainly at the long wavelength absorption. Absorption at 650 nm started to decrease accompanied by the shifting of the long wavelength absorption maximum to 615 nm, indicating the formation of P⌽B following time. This result was confirmed by the HPLC analysis of reaction products 10 min after adding NADPH (Fig. 3B, 1 eq. e Ϫ trace). Notable absorption increased at 740 nm, followed by a slow decay, indicating that a new species of pigment was produced. A similar phenomenon has been detected in the PcyA reaction, representing the formation of transient radical species during double bond reductions (11). The results suggest that HY2-catalyzed A-ring diene reduction also possibly proceeds with a radical intermediate. Sequential additions of 2 and 3 electron eq showed similar results but with less absorption change at 740 nm and more decrease at 650 nm, giving more P⌽B production (Fig. 3B, 2 and 3 eq. e Ϫ traces). To detect the radical species, the single turnover anaerobic assay for binary BV complexes with HY2 proteins was performed and monitored by both UV-visible and EPR spectroscopy. As shown in Fig. 4A, wild-type HY2 rapidly converted BV into P⌽B in 5 min with an excess equivalent of electrons added. Compared with 1 eq of electron reduction (Fig. 3A), the absorption change at 740 nm was more significant, and a slight change at 470 nm was observed. Aliquots were taken from the reaction mixture at different time points after adding NADPH, frozen in liquid nitrogen, and measured by EPR spectroscopy. An isotropic EPR signal at G Ϸ 2 with a peak-to-trough width of 15 G was detected (Fig. 4B). The radical signal appeared rapidly and slowly decayed with similar kinetics as the absorbance change at 740 nm (Fig. 4C). These spectral properties are similar to those of the PcyA reaction but with the difference that the HY2 radical can be detected at higher temperature (70 K), suggesting that the HY2-BV radical is more stable and well protected by the enzyme during reaction (11). These results confirm that a transient radical intermediate is produced in the HY2-catalyzed BV reduction.

Mutagenesis Study Identifies Asp 116 , Asp 256 , and Lys 255 in the Reduction of the A-ring Double
Bond-The double bond reduction requires proton donations from enzymes. We then envisaged identifying the proton-donating residues on HY2 that are critical for the catalytic steps. Several possible catalytic residues, including Asp 116 , Cys 131 , and Lys 255 , have been proposed based on the homology model of the HY2 active site constructed from the holo-PcyA structure (12). These residues are all present in two conserved regions of the multiple sequence alignment of FDBRs (supplemental Fig. S1A). The two conserved regions represent the main part of the central ␤-sheet and C-terminal ␣-helices in the PcyA structure (supplemental Fig. S1B) (12,14,19). Combining the information from the homology model and sequence alignment, we selected several conserved ionizable residues within the two conserved regions for mutagenesis. All of the mutant proteins were expressed, purified to homogeneity, and measured for steady-state bilin reductase activity ( Table 1). The long-to-short wavelength ratios in the absorption spectra of the enzyme-BV complexes were measured to compare the relative ability of BV to bind to wild-type and mutant HY2 proteins.
An aspartate residue (Asp 102 in Nostoc PcyA) has been shown to function as the proton donor in PcyA-catalyzed BV reduction (13). There are three conserved aspartate residues in the two conserved regions of HY2, Asp 116 , Asp 146 , and Asp 256 . To test their functional roles, aspartate-to-asparagine substitutions were first made to determine whether the ionization ability of these aspartate residues is essential for HY2 activity. The D116N mutant still retained the ability to bind substrate but with only 1.5% relative activity of the wild-type protein, suggesting that Asp 116 is important for BV reduction. The D146N mutant completely lost catalytic activity but also the ability to bind BV, indicating that this mutation affects mainly substrate binding. A similar result was obtained from a mutagenesis study of the same conserved aspartate residue of PcyA, which was also found from the crystal structure to hydrogen bond to another serine residue to stabilize the overall fold (12)(13)(14)19). This shows that the HY2 protein may have a similar fold to that of PcyA.
To our surprise, the mutation of Asp 256 to asparagine abolished HY2 activity without affecting substrate binding. Although Asp 256 is universally conserved in the FDBR family, FIGURE 3. Absorption spectra and HPLC analysis of HY2-catalyzed BV reduction under anaerobic conditions. A, the time course of BV reduction by HY2 was monitored spectrophotometrically following sequential addition of 1-3 molar eq of reductant (i.e. 5 M NADPH) to samples containing 10 M HY2-BV complex, 0.00625 units/ml FNR, 10 M Fd, and an oxygen-scavenging system as described under "Experimental Procedures." Absorption spectra were monitored for 10 min; 1-min intervals are shown. Twenty-minute periods were provided between subsequent additions of NADPH. Absorbance increases followed by decays at 740 nm are indicated by double arrows, and absorbance decreases at 650 nm and increases at 615 nm are indicated by single arrows. B, reaction mixture aliquots (200 l) corresponding to the spectra shown in A were withdrawn, and extracted bilin pigments were analyzed by reversed-phase HPLC. The elution positions of BV, 3Z-P⌽B, and 3E-P⌽B are indicated by arrows, and the absorbance at 650 nm was monitored.
mutation of the corresponding aspartic acid residue of PcyA (Asp 217 ) has been shown to have no effect on activity (13). Structural data for PcyA also indicate that the carboxylic side chain of this residue initially points into the binding pocket in the apo structure but becomes solvent-exposed upon substrate binding (12,14,19). In HY2, this residue apparently plays an important role in BV reduction. This result is further supported by data from a new member of the FDBR family, PebS, which catalyzes the reduction of BV into PEB with the C-15-C-16 double bond and A-ring diene reduction activities (9,26).
To address the importance of Asp 116 and Asp 256 , additional amino acid substitutions were generated. Replacement of both residues with histidine also completely abolished HY2 activity, indicating the requirement of their acidic side chains for HY2 reaction. However, the D256E mutant retained only partial activity, showing the importance of the configuration or length of the Asp 256 carboxylic side chain.
In addition to the aspartate mutations, we also selected several residues, including Cys 131 , Asn 133 , Arg 252 , and Lys 255 , for mutagenesis. Based on previous homology modeling results, these residues were predicted to reside in the active site of HY2, being close to bound BV. HY2 with a mutation of Cys 131 to alanine retained almost full activity, indicating that this residue is not the proton donor. Although asparagine is not ionizable, the N133A mutant produced only partial activity, indicating that this residue may play other roles in the active site, such as stabilizing BV conformation through hydrogen bonding. The R252Q mutant lost catalytic activity and the ability to bind substrate, indicating that this mutation affects substrate binding. Mutation of Lys 255 showed a severe effect on catalytic activity, suggesting that this residue may be important for P⌽B production. Because of the high pK a for lysine, the catalytic role of Lys 255 needs to be further investigated. Taken together with the steady-state activity of HY2 mutants, we found several residues in HY2, including Asp 116 , Asp 256 , and Lys 255 , that may be involved in the process of BV A-ring 2,3,3 1 ,3 2 -diene reduction.
Anaerobic Assay Reveals That Asp 256 Is the Important Proton Donor-Because our steady-state assay cannot produce quantitative kinetic data, a more detailed analysis of the pre-steadystate catalysis reaction for these mutants was performed. To examine whether mutant enzymes can support proton transfer for BV reduction, single turnover anaerobic assays for the binary BV complexes with mutant HY2 proteins were monitored by UV-visible spectroscopy with an excess equivalent of electrons added. Compared with the wild-type reaction (Fig. 5,  upper left panel), BV was also fully converted into P⌽B but with FIGURE 4. EPR measurement of the bilin radical produced during the HY2 reaction. EPR spectroscopic measurement of BV complexes of HY2 was performed as described under "Experimental Procedures." A, UVvisible absorption spectra of the HY2 reaction. Time courses of BV reduction, monitored spectrophotometrically every 1 min after the addition of 10 molar electron eq of reductant (i.e. 200 M NADPH) to samples containing 40 M preformed HY2-BV complex, were obtained as described previously (11). Absorbance increases followed by decays at 470 and 740 nm are indicated by double arrows, and absorbance decreases at 650 nm and increases at 615 nm are indicated by single arrows. B, EPR spectra of the HY2-BV reaction. For EPR measurement, 200-l aliquots were withdrawn at 0, 0.5, 1.5, and 5 min after NADPH addition and immediately frozen in liquid nitrogen. C, comparison of EPR signals and absorption change in the HY2 reaction. The intensity of EPR signals was calculated from the sum of peak high and trough depth. Relative EPR signal intensity and relative absorbance at 740 nm normalized to the percent of the signal at 0.5 min are shown. AU, absorbance units.

TABLE 1 Relative activities of wild-type HY2 and site-directed mutants
Steady-state bilin reductase assays were performed as described under "Experimental Procedures." Integrated peak areas of 3Z/3E-P⌽ B reaction products from HPLC profiles were determined as a percentage of that of wild-type HY2. To compare the efficiency of BV binding between wild-type HY2 and site-directed mutants, longto-short wavelength absorption ratios of enzyme-BV complexes were measured. Ratios for free and wild-type HY2-bound BV were set as 0 and 100%, respectively. The reaction with the K255Q mutant also yielded P⌽B in the single turnover assay but with much slower kinetics compared with the wild-type reaction (Fig. 5, upper right panel). This result suggests that the lysine mutant can support proton transfers to the BV substrate. Because K255Q is affected only in the steady-state reaction, it is possible that Lys 255 is involved in the substrate binding or product release step. It also has been shown that positively charged residues in ferredoxin-interacting proteins participate in the electrostatic interaction with ferredoxin for electron transfer (20 -24). Therefore, Lys 255 may function in the interaction between HY2 and ferredoxin. Further investigation needs to be performed to support this hypothesis.

Protein
Interestingly, BV reduction was completely abolished in the D256N reaction (Fig. 5, lower right panel). The addition of 10 eq of electrons led to the absorption increase at 470 and 740 nm, which reached a maximum without the following decay. Similar results have been observed in a Nostoc PcyA study of mutations of the critical ion-pairing residues His 85 and Asp 102 (13). This result strongly indicates that one of the proton donation steps is defective in the D256N mutant. The appearance of an absorption increase at 470 and 740 nm in the D256N-BV reaction is possibly an indication of radical accumulation.
Radical Accumulation in the D256N-BV Reaction Confirms the Catalytic Role of This Residue-To ascertain that the absorption change during the D116N and D256N reactions corresponds to the radical intermediate, we performed freeze-quench EPR measurement. Aliquots were taken from the reaction mixtures at different time points after the addition of NADPH, frozen in liquid nitrogen, and measured by EPR spectroscopy at 70 K. The result from the D116N-BV reaction confirmed that the appearance of a strong radical signal had similar kinetics to the absorbance increase and decay at 670 and 730 nm (Fig. 6, A and C). In contrast, the D256N mutant produced a stable radical species that accumulated following time (Fig. 6, B and  D). This result supports our prediction that Asp 256 is one of the proton donors. Taken together, these data reveal that the A-ring diene reduction of BV catalyzed by HY2 also proceeds through a transient radical intermediate. Asp 256 plays an important role in one of the proton donation steps.

DISCUSSION
HY2 is a plant-type FDBR that synthesizes the P⌽B chromophore precursor for phytochrome (2). With the activity of A-ring double bond reduction, HY2 is able to generate the ethylidene group for thioether linkage between P⌽B and apophytochrome (1). It also functions with the activity that exists in the biosynthesis pathways of phycocyanobilin and PEB (8). This study was undertaken to investigate the catalytic mechanism of HY2 from Arabidopsis. Results from spectroscopic assay indicate that HY2 reduces the A-ring double bond through a transient radical intermediate similar to PcyA. We also observed that a number of ionizable residues are involved in the reduction steps. In particular, Asp 256 is a candidate for serving as one of two proton donors. The second proton donor is possibly not a protein residue. In the following discussion, we compare the differences between HY2 and other FDBRs and outline the mechanistic implications of the possible reaction steps for HY2-catalyzed BV reduction. We also suggest a revised homology model for the HY2-BV binary complex.
Comparison of HY2, PcyA, and PebS Activities-Our current knowledge of the catalytic mechanism of FDBRs comes mostly from the data on PcyA (10 -13). Based on mutagenesis and structural data, a histidine/aspartate ion pair in the active site of PcyA was shown to be critical for protonation steps in both Aand D-ring double bond reductions of BV (12,13). However, sequence comparison between FDBR members indicates that this ion pair can be found only in PcyA, implying that the catalytic mechanism proposed for PcyA may not be applicable to all FDBRs (13). In this study, we first found that HY2 also catalyzes BV reduction through a transient radical intermediate. The EPR signal is very similar to that of the A-ring endo-vinyl radical in the PcyA reaction (11). These results suggest that all FDBRs may reduce BV with a similar step that receives the electrons directly from reduced ferredoxin to facilitate proton donation from residues adjacent to reacting atoms of BV.
However, replacing the ionizable residues in the active site of HY2 has produced interesting results showing the different catalytic roles of aspartate residues in different FDBRs. In PcyA, the carboxylic side chain of Asp 220 between helices H7 and H8 swings out of the active site to become solvent-exposed upon BV binding (12,14,19). Biochemical data also support that the ionization ability of this residue is not essential for PcyA activity (13). In our study, we found that the corresponding aspartate residue in HY2, Asp 256 , is actually critical for A-ring double bond reduction. The proton donation ability of the carboxylic side chain of Asp 256 is essential for BV reduction into P⌽B, suggesting the direct involvement of this residue in HY2 catalysis steps.
Another critical aspartic acid residue in the PcyA active site is Asp 105 , the primary proton donor in the Asp/His ion pair in the PcyA reaction (12,13). However, its corresponding residue in HY2 is a nonionizable residue, Asn 133 . Our data from this study also show that mutation of Asn 133 has no significant effect on P⌽B synthesis, indicating that this residue is not important for the HY2 catalytic reaction.
The second residue in the ion pair of PcyA, His 88 , has been shown to play central roles in both exo-vinyl and A-ring diene double bond reductions (12,13). This residue functions mainly in proton transfers to carbonyl oxygens on A-and D-rings for subsequent tautomerizations. However, the histidine residue is present only in the PcyA family (13). The comparable residue in HY2 is Asp 116 , which was predicted to be located in the HY2 active site (12). Although our steady-state assay initially showed that the D116N mutant loses catalytic activity, which could be defective in substrate/cofactor binding, chemical reaction, product release, and other steps, results from single turnover assay indicate that it is still able to produce P⌽B with slower kinetics compared with wild-type HY2. This result clearly shows that this aspartate residue in HY2 has different functions compared with His 88 in PcyA. The strong red absorp- tion during the D116N-BV reaction showed that a significant change in conformation and/or protonation state happened in the BV molecule during reduction. This is a strong indication that Asp 116 functions in positioning the BV molecule in the binding pocket.
Recently, a new member of the FDBR family, PebS, has been identified from cyanophage P-SSM2 (9). PebS synthesizes PEB through two double bond reductions in the 15,16-carbons and A-ring 2,3,3 1 ,3 2 -diene system. More recently, the crystal structure of the PebS-BV binary complex was also solved (see accompanying article (26)). The overall structure of PebS is similar to that of PcyA, with an ␣/␤/␣-sandwich fold. The BVbinding site is also located in a pocket between the central ␤-sheet and C-terminal ␣-helices. These results support the hypothesis that the FDBR family has similar structural scaffolds but with different proton-donating residues in the active sites. Furthermore, the side chain of the conserved aspartate residue (Asp 206 in PebS) is pointing toward the A-ring and in hydrogenbonding distance to the carbonyl oxygen (26). It is likely that this conserved aspartic acid residue in HY2 and PebS has a similar function in A-ring diene reduction. By analyzing the biochemical data of HY2 and PebS structure, we believe that HY2 is more similar to PebS at least in A-ring reduction. Taken together, these data reflect that although FDBRs share a similar structural scaffold, they have evolved different proton-donating residues in active sites for BV reduction.
Mechanistic Prediction for Substrate Binding-Based on spectroscopic measurements of the PcyA-BV complex and structural information, bound BV is in a porphyrin-like conformation to fit into the binding pocket for subsequent proton transfers from the histidine residue to both carbonyl O-1 and O-19 (10,12,14). The proposed mechanism indicates that the carbonyl O-19 of bound BV is first protonated, resulting in the absorption increase in the near-infrared region of the PcyA-BV spectrum. The O-protonation is contributed by both Asp 105 and His 88 together. Defects in either residue produce inactive enzymes and lead to loss of the near-infrared absorption (13). In the case of HY2, the aspartate/histidine ion pair responsible for primary O-protonation is missing, as is the spectroscopic signature in the near-infrared region. These phenomena indicate that the HY2 reaction is possibly not initiated by a primary protonation. Analysis of the D256N mutant indicated that the ability of this residue to donate a proton is required for A-ring double bond reduction. Furthermore, the absorbance change at 740 nm during reduction steps is similar to the change in the PcyA-18 1 ,18 2dihydrobiliverdin reaction, which involves a proton transfer step through the carbonyl oxygen to C-2 on the A-ring (12). Therefore, we propose that the carbonyl oxygen on the A-ring is important for the proton transfer from the carboxylic side chain of Asp 256 in HY2. Our data also show that the ionization property of Asp 116 is important for stabilizing the conformation of bound BV during reduction but is not essential for P⌽B production. We then conclude that Asp 116 may form a second hydrogen bond with the proton on the A-ring nitrogen atom. Both hydrogen bondings could generate a less basic condition on the A-ring to localize the subsequently transferred electron in the tetrapyrrole ring system. Alternatively, Asp 256 , Asp 116 , or both can function as the primary proton donors for the protonation on both carbonyl oxygens. This can generate the cationic BV substrate for subsequent electron transfers.
Mechanistic Prediction for A-ring Reduction-Our study has established that HY2 reduces the BV A-ring 2,3,3 1 ,3 2 -diene system through a sequential reaction of electron transfers. The sequential electron transfers are accompanied by proton donations in the active site. Results from mutagenesis and spectroscopic analysis of possible catalytic residues in the predicted HY2 active site have identified only one proton-donating residue. The observation that the D256N-BV reaction accumulates stable radical intermediates not only indicates that Asp 256 is a proton donor but also suggests that electron transfer could occur without proton transfer. We therefore hypothesize that upon the first electron transfer, the proton on Asp 256 that hydrogen bonds to the carbonyl oxygen is abstracted by the BV anion radical to produce the neutral radical with O-1 protonation. Alternatively, the bound BV can also be protonated first by Asp 256 , Asp 116 , or both residues on carbonyl oxygens to form cation species and then attract the electron(s) from reduced ferredoxin(s). The accepted electron(s) is possibly relocated on the C-3 2 atom due to the change in substrate conformation and environment. Following the conformational change in the BV radical, the transferred proton on O-1 could be rearranged to form the C-2 protonated radical species with an enol-to-keto tautomerization (Fig. 7, Radical intermediate). A second electron transfer accompanied by a second proton transfer to the C-3 2 atom then yields the final product, 3Z/3E-P⌽B. Finally, the product is released due to the loss of bilin-protein interaction and the increase in negative charge in the active site. The different radical species should be able to be distinguished by advanced EPR techniques such as electron nuclear double resonance and electron spin echo envelope modulation.
Because of the inability to identify a second proton donor for the A-ring diene reduction, we hypothesize that the second proton is obtained from a water molecule in the active site or directly from bulk solvent. A similar mechanism has been pro- FIGURE 7. Proposed mechanism for HY2-catalyzed A-ring diene reduction. HY2-bound BV is in a cyclic porphyrin-like conformation. Bound BV receives one electron and one proton to produce the BV radical intermediate. The primary proton donor could be Asp 256 . The second electron transfer, accompanied by subsequent proton transfer (possibly from a water molecule), yields the final product, 3E/3Z-P⌽B. P, the propionate group.
posed for the diene reduction in PcyA (12). Several conserved water molecules have also been found in the active site of the PcyA and PebS crystal structures (12,14,19,26). It is possible that the second electron transfer generates a BV anion radical, which is a strong base to extract the proton from a water molecule adjacent to the C-3 2 atom. This reaction can occur either in the active site or upon product release. Further isotope experiments need to be performed to support this hypothesis.
Structural Prediction of the HY2 Active Site upon BV Binding-A homology model of HY2 based on the PcyA-BV binary structure was generated previously (12). However, spectral data from our study show that the HY2-bound BV substrate may be positioned with a more extended conformation compared with the PcyA-bound BV molecule. In addition, several residues were predicted to play catalytic roles in the HY2 reaction, such as Cys 131 and Lys 255 . However, results from our mutagenesis and spectroscopic analysis indicate that these residues are not directly involved in the reaction steps. These data reveal that the previous HY2 model needs to be revised. Interestingly, the PebS structure recently solved has a similar fold and substratebinding pocket as the PcyA structure (26). The universally conserved aspartate residue in C-terminal ␣-helices (Asp 256 in HY2) is well positioned in the active site, and hydrogen bonding to the BV molecule is ideal for proton donation. Comparison of the CD spectra of HY2-, PcyA-, and PebS-bound BV also shows that the geometry of HY2-bound BV is similar to that of PebSbound BV, giving a more left-handed helical conformation (supplemental Fig. S3) (25,26). We therefore selected the PebS binary structure as the template for homology modeling to propose the possible location of catalytic residues in the HY2 active site.
As shown in the active site of the resulting HY2 model, the carboxylic side chain of Asp 256 points toward the A-ring and is in hydrogen-bonding distance to the carbonyl oxygen (supplemental Fig. S4). The position fits with our prediction that Asp 256 is responsible for the first proton transfer through O-1 to C-2. Asp 116 in the central ␤-sheet is in a good position to hydrogen bond to the A-ring nitrogen. It may function in positioning the A-ring for the subsequent proton transfer. Lys 255 was predicted as a possible proton donor in the previous model. However, our single turnover assay has shown that the Lys 255 mutant is still able to metabolize BV to P⌽B but with a slower reaction rate. This result can be explained by our revised model showing the side chain of Lys 255 as being exposed to solvent. It is possible that Lys 255 contacts ferredoxin for electron transfer. Involvement of charged residues in the electrostatic interaction between ferredoxin and its interacting partners has already been demonstrated in several studies (20 -24).
Based on our biochemical approach, all residues predicted to be adjacent to the C-3 2 atom have been shown to have no direct catalytic function. This result further supports our prediction that the water molecule is likely to play a role in the second protonation step. We cannot rule out the possibility that water is also involved in the initial protonation step. Conserved water molecules have been found in the active site of both PcyA and PebS binary structures (12,14,19,26). These molecules are shown to be in hydrogen-bonding distance to either the catalytic residues or reactive atoms of BV. They could function as the donors themselves or shuttling protons in the relay systems. Structural information will help us to elucidate these predictions in detail.
Future Studies-Although our biochemical data have provided the information necessary to predict a possible catalytic mechanism, experimental work is still needed to elucidate the precise reaction steps. We believe that advanced EPR techniques such as electron nuclear double resonance and site-specific spin labeling will provide more information on the chemical environment in the active site and mechanistic prediction. Structural information on the HY2 apoprotein and HY2-BV binary complex not only is required for the mechanistic prediction but also will be helpful in protein engineering to exchange activities between different FDBRs. The engineered HY2 will be an ideal candidate to generate chromophore analogs for studying detailed functions of phytochromes in plants.