CpeS Is a Lyase Specific for Attachment of 3Z-PEB to Cys82 of β-phycoerythrin from Prochlorococcus marinus MED4*

In contrast to the majority of cyanobacteria, the unicellular marine cyanobacterium Prochlorococcus marinus MED4 uses an intrinsic divinyl-chlorophyll-dependent light-harvesting system for photosynthesis. Despite the absence of phycobilisomes, this high-light adapted strain possesses β-phycoerythrin (CpeB), an S-type lyase (CpeS), and enzymes for the biosynthesis of phycoerythrobilin (PEB) and phycocyanobilin. Of all linear tetrapyrroles synthesized by Prochlorococcus including their 3Z- and 3E-isomers, CpeS binds both isomers of PEB and its biosynthetic precursor 15,16-dihydrobiliverdin (DHBV). However, dimerization of CpeS is independent of bilins, which are tightly bound in a complex at a ratio of 1:1. Although bilin binding by CpeS is fast, transfer to CpeB is rather slow. CpeS is able to attach 3E-PEB and 3Z-PEB to dimeric CpeB but not DHBV. CpeS transfer of 3Z-PEB exclusively yields correctly bound βCys82-PEB, whereas βCys82-DHBV is a side product of 3E-PEB transfer. Spontaneous 3E- and 3Z-PEB addition to CpeB is faulty, and products are in both cases βCys82-DHBV and likely a PEB bound at βCys82 in a non-native configuration. Our data indicate that CpeS is specific for 3Z-PEB transfer to βCys82 of phycoerythrin and essential for the correct configuration of the attachment product.

Phycobilisomes are the major light-harvesting complexes of cyanobacteria, rhodo-phytes, and cryptophytes (1,2). The light-absorbing properties of corresponding phycobiliproteins result from up to four linear tetrapyrroles, known as bilins, covalently bound via thioether bonds to conserved cysteine residues of each ␣ and ␤ subunit. Bilins are derived from heme, which is converted to the first linear tetrapyrrole biliverdin IX␣ (BV IX␣) 2 by heme oxygenases (3). In cyanobacteria, BV IX␣ is further reduced by ferredoxin-dependent bilin reductases to 3Z-phycoerythrobilin (3Z-PEB) or 3Z-phycocyanobilin (3Z-PCB) 3 (4,5). Subsequent bilin attachment to phycobiliproteins is in most cases supported by members of the three known phycobiliprotein lyase families (E/F, S/U, and T) (6). These proteins are supposed to guide the reaction in a chaperone-like manner probably by conformational control of the bilin (7,8). In addition, some lyases act as isomerases and generate phycoviolobilin or phycourobilin upon attachment to the phycobiliprotein (9 -11). Whereas PCB, phycourobilin, or phycoviolobilin attaching lyases are well characterized, little is known about lyases specific for attachment of PEB to phycoerythrins (PE). We therefore investigated a putative PEB-transferring lyase encoded by P. marinus MED4 (12).
Prochlorococcus strains dominate most oceanic phytoplankton communities (13). They possess a remarkable pigment composition and depend on unique divinyl-chlorophyll antenna complexes instead of phycobilisomes for photosynthetic light harvesting (14,15), Despite the absence of phycobilisomes Prochlorococcus carries a special PE of unknown function (16 -18). Genes associated with phycobiliprotein maturation can be found in highly reduced prochlorococcal as well as in Prochlorococcus-infecting phage genomes indicating the importance of PE for cell fitness (12,19). Prochlorococcus occurs in two ecotypes adapted to specific niches (20). High-light strains inhabit the upper nutrient-depleted but light-irradiated layer of the ocean, whereas low-light strains colonize depths of up to 200 m, which are nutrient-rich but exposed to a low light intensity. Whereas low light-adapted strains encode ␣ and ␤ subunits of PE, HL strains like P. marinus MED4 carry only a degenerated form of ␤-PE (14,21). P. marinus MED4 possesses one of the smallest genomes of photosynthetic organisms but still encodes all components for phycobiliprotein maturation: ␤-PE (CpeB), enzymes for PEB and PCB biosynthesis (i.e. HO1, PebA, PebB, and PcyA) and the S-type lyase CpeS (12,22). We characterized CpeS from MED4 with respect to bilin specificity and affinity, binding kinetics, and transfer activity. Besides BV IX␣ and DHBV, both the 3Z-and 3E-isomers of PEB and PCB were included in this study to address the so far unstudied stereoselectivity of lyases. Our results demonstrate that CpeS specifically transfers 3Z-PEB to CpeB and ensures correct configuration of the attachment product. The functionality of the last component of the phycobiliprotein maturation machinery in Prochlorococcus is demonstrated underlining the relevance of PE in this unusual cyanobacterium.

EXPERIMENTAL PROCEDURES
Construction of Expression Plasmid-The P. marinus MED4 cpeS coding region was PCR-amplified from genomic DNA with primers encompassing recognition sites for EcoRI (5Ј-gaccatgattagaattcgttgacg-3Ј) and HindIII (5Ј-ccaagcttgcctgcagagaataat-3Ј). The PCR product was cloned into expression vector pASKIBA45ϩ (IBA) for N-terminal fusion of a Strep-tag to CpeS. The resulting plasmid was verified by sequencing. Construction of plasmid pET43a-PE for production of N-terminal NusA His-tagged CpeB was described previously (18).
Production and Purification of Recombinant Proteins-A culture of BL21(DE3) carrying the respective overexpression plasmid was incubated at 37°C to an A 580 of 0.6. After induction with isopropyl-␤-D-thiogalactopyranoside (100 M; pET43 derivative) or anhydrotetracycline (200 g/ml; pASKIBA45ϩderivative) cells were incubated for 18 h at 17°C, harvested by centrifugation, resuspended in sodium phosphate buffer (60 mM sodium phosphate, 100 mM NaCl, pH 7.5) and disrupted by two passages through a French press cell at 20,000 psi. After separation of cell debris the supernatant was loaded on a Talon metal affinity resin (Clontech; CpeB) or a Strep-Tactin-Sepharose (IBA; CpeS) column. Purification was carried out according to the manufacturer's instructions based on sodium phosphate buffer. If needed, purified CpeB was incubated with thrombin (10 cleavage units/mg CpeB) and dialyzed against cleavage buffer (60 mM sodium phosphate, 150 mM NaCl, 5 mM MgCl 2 , 2.5 mM CaCl 2 , pH 7.8). Tag-free CpeB was then purified by a second metal affinity chromatography. Purified proteins were dialyzed against buffer, which was in case of CpeB supplemented with 1 mM dithiothreitol. CpeB was additionally dialyzed twice against buffer to remove dithiothreitol. Concentrations of proteins were determined using the calculated molar extinction coefficient ⑀ 280 (23).
Chemical Modification of Cysteine Residues-CpeB was incubated with 10 mM iodoacetamide for 30 min and subsequently purified via Nap5 columns (GE Healthcare).
Spectroscopy-UV-visible absorption measurements were done using an Agilent Technologies 8453 spectrophotometer. Fluorescence spectroscopy was performed with an Aminco-Bowman AB2 spectrofluorimeter.
Determination of CpeS Bilin Specificity-Assays were performed in sodium phosphate buffer, pH 7.5, at room temperature. Lyase (10 M) and bilin (5 M) were incubated for 10 min before affinity purification of CpeS. Columns were washed with increasing NaCl concentrations (100 mM-1 M) and wash fractions screened for released bilins by absorption spectroscopy.
Stopped Flow Kinetics of Bilin Binding to CpeS-Stopped flow experiments were performed in an SFM-400 apparatus with MOS-200 optics (Bio-Logic) at 20°C in sodium phosphate buffer, pH 7.5. Bilin (2 M) was rapidly mixed with various concentrations of CpeS and the absorption increase at 594 nm (3E-PEB), 590 nm (3Z-PEB), and 607 nm (DHBV) was recorded. Ten time traces were accumulated and averaged each. An exponential equation was fitted to the experimental data yielding the k obs value for each concentration. All experiments were repeated two times.
Isothermal titration calorimetry-Thermodynamic parameters of CpeS interaction with 3E-PEB at 20°C in sodium-phosphate buffer pH 7.5 were determined using an isothermal titration microcalorimeter (MicroCal Auto-iTC 200 , GE Healthcare). The first 0.5 l of 3E-PEB (300 M) were injected into the temperature-controlled sample cell containing 200 l of CpeS (30 M). The change in heating power was monitored for 180 s until equilibrium was reached before the next 1.8-l injection was started. Background heat generation from dilution of 3E-PEB in buffer was measured to be irrelevant. The data were averaged over two ITC experiments.
Size Exclusion Chromatography-An GE Healthcare Superdex 75 HR10/300 GL size exclusion column was equilibrated in sodium phosphate buffer at a flow rate of 0.5 ml/min. Standards with known M r (i.e. bovine serum albumin, 66,000; carbonic anhydrase, 29,000; cytochrome c, 12,400; and aprotinin, 6,500) were applied to the column, and their elution volumes were determined spectroscopically. CpeS and tag-free CpeB were chromatographed under identical conditions.

3Z-PEB⅐␤Cys 82 Lyase CpeS
Chromophore Transfer Assays-Bilin (15 M) was incubated with or without CpeS (30 M) for 5 min. Total binding of bilin to CpeS was confirmed via absorption spectroscopy. In control experiments, affinity purified CpeS⅐bilin was used. CpeB (30 M) was added, and reactions were followed spectroscopically. After 45 min, CpeB was isolated by affinity chromatography and further analyzed. In competition assays, bilin (3.5 M) was added to CpeS, CpeB, or premixed CpeS/CpeB (7 M each), and absorption and fluorescence emission spectra were taken immediately.

Purification of Recombinant CpeS and CpeB-P. marinus
MED4 possesses an unusual green light-absorbing ␤-PE (CpeB) with one PEB bound at Cys 82 (18). Adjacent to the cpeB gene, the putative S-type lyase CpeS is encoded. Besides CpeS, no other lyase homolog was identified in MED4. To study whether the putative lyase is sufficient for bilin attachment to ␤-PE, corresponding proteins were overproduced in E. coli and purified via affinity chromatography (supplemental Fig. S1). CpeS was obtained to almost 100% purity, CpeB to an estimated purity of Ͻ50%. Because CpeB was produced with an N-terminal NusA-tag, impurities most likely represent NusA breakdown products, which were removable by affinity purification of thrombin-cleaved CpeB. However, truncated CpeB was unstable and therefore only employed in small-scale control experiments to rule out an influence of either tag or contaminations on bilin binding assays. In addition, covalent binding of bilins to contaminating proteins was ruled out by zinc-blot analysis (supplemental Fig. S2).
CpeS Binds DHBV and Both PEB Isomers-Prochlorococcus possesses the functional biosynthetic machinery for production of PEB and PCB (37). Although PEB is the chromophore of ␤-PE, no PCB-binding protein has been identified in MED4 so far.
Because members of the S/U type lyase family are characterized by a low bilin specificity (6,8), we tested CpeS for binding of all bilins synthesized by MED4. We included 3Z-and 3E-isomers of PCB and PEB to address the largely unstudied bilin stereoselectivity of lyases. Absorption spectra of BV IX␣, 3E-PCB, and 3Z-PCB were only slightly influenced by CpeS (data not shown). Mixing of CpeS with DHBV, 3E-PEB, or 3Z-PEB immediately resulted in the formation of bluish, nonfluorescent complexes with increased long wavelength absorption at 607 nm (DHBV), 594 nm (3E-PEB), or 590 nm (3Z-PEB) (Fig. 1), indicating binding of bilins in a stretched conformation. During subsequent affinity purification of lyase-bilin mixtures, washing with increasing salt concentrations lead to release of total BV IX␣, 3E-PCB, and 3Z-PCB at low salt concentrations (200 mM NaCl), whereas DHBV, 3E-PEB, and 3Z-PEB were retained, even at high salt concentrations (1 M NaCl) (data not shown). Absorption spectra of eluted complexes were comparable with those obtained before purification indicating stable binding of DHBV, 3E-PEB, and 3Z-PEB to CpeS. To investigate whether this stability is due to covalent binding, SDS-PAGE and subsequent zinc-blot analyses were performed (supplemental Fig. S2A). No zinc-induced red fluorescence was detected indicating a noncovalent interaction.
Dimeric CpeS Binds Two Molecules of DHBV, 3E-PEB, or 3Z-PEB-S-type lyases can occur in different oligomeric states (38 -40). Therefore, we performed size exclusion chromatography of CpeS in the presence and absence of bilins. In the absence of bilins, CpeS eluted as a complex of ϳ47.1 kDa ( Fig.  2A), which correlates well with the calculated mass of a homodimer (46.6 kDa). Elution profiles in the presence of bilins were almost identical with protein and bilin eluting simultaneously as CpeS⅐bilin complexes of ϳ50.8 kDa (DHBV), 50.1 kDa (3E-PEB), and 50.3 kDa (3Z-PEB) (Fig. 2B). These data suggest CpeS dimer formation prior to bilin binding.
CpeS Binds DHBV and Both PEB Isomers with Similar Binding Kinetics and High Affinity-As CpeS⅐bilin complexes were formed immediately after mixing, binding kinetics of both components were analyzed by stopped flow experiments. CpeS and bilins were rapidly mixed at different ratios and absorption increases at 594 nm (3E-PEB), 590 nm (3Z-PEB), or 607 nm (DHBV) were recorded. A single exponential function was fitted to the experimental data, from which the observed rate constant k obs was extracted (supplemental Fig. S4). The k obs values were plotted against the protein concentration (Fig. 3). Linear regression yielded the association rate constants k on as the slopes of the straight lines with standard errors in the range of 10%. 3E-PEB shows the fasted association rate constant (2 M Ϫ1 s Ϫ1 ) followed by 3Z-PEB (1.4 M Ϫ1 s Ϫ1 ) and DHBV (0.9 M Ϫ1 s Ϫ1 ) (Table 1). However, k on values were in the same range especially when considering possible inexactness in DHBV or 3Z-PEB concentration due to the unknown extinction coefficients. The ordinate intercept values, which correspond to the dissociation rate constant k off , are close to 1 s Ϫ1 . As this value is small, the standard error is of the same magnitude so that a precise value for k off cannot be obtained. Nevertheless,

TABLE 1 Rate constants and thermodynamic parameters for the interaction of CpeS with bilins
The association rate constants (k on ) were determined by stopped flow experiments, binding constant (K a ), enthalpy (⌬H 0 ), and entropy change (⌬S 0 ) by ITC. From those values, K d ϭ 1/K a and k off ϭ k on ϫ K d were calculated. ND, not determined.

3Z-PEB⅐␤Cys 82 Lyase CpeS
a value close to 1 s Ϫ1 can be confirmed by calculation from the k on value and the K d value obtained from ITC (see below), which yields exactly K d ϫ k on ϭ k off ϭ 1.4 s Ϫ1 .
CpeS⅐3E-PEB complex affinity was measured by ITC. 3Z-PEB and DHBV were not included due to insufficient amounts available. Experiments were repeated two times with an example of a result presented in Fig. 4, and mean values are shown in Table 1. The stoichiometry of CpeS⅐3E-PEB was n ϭ 1.2 Ϯ 0.04 in reasonable agreement with the spectroscopically determined 1:0.89 ratio in the complex. The resulting K d value of the CpeS⅐3E-PEB complex was 0.7 M, which represents tight binding and which is driven by both favorable enthalpy (⌬H 0 ϭ Ϫ6.1 Ϯ 0.5 kcal/mol) as well as entropy (⌬S 0 ϭ 7.3 Ϯ 1.3 cal/mol/deg) changes upon complex formation. The K d determined for the CpeS⅐3E-PEB complex is in range with values obtained for other lyases (8,39,41,42).
CpeS Transfers PEB but Not DHBV to Dimeric CpeB-To verify whether CpeS is sufficient for correct bilin addition to ␤-PE, we compared spontaneous and lyase-mediated attachment of bilins to CpeB by recording the fluorescence emission of assembled holo-CpeB. DHBV was neither spontaneous nor under catalysis of CpeS attached to CpeB (Fig. 5, A and D). This was further verified by zinc-blot analysis (supplemental Fig.  S2B) and HPLC of CpeB peptides (data not shown). Spontaneous and CpeS-mediated reactions of both PEB isomers with CpeB resulted in formation of fluorescent complexes with covalently bound bilins ( Fig. 5 and supplemental Fig. S2B). Reaction products were stronger fluorescent when obtained (i) with 3Z-PEB and (ii) in presence of CpeS. The course of spontaneous and lyase-supported reactions were significantly different. While in the presence of CpeS, fluorescence intensities increased constantly (Fig. 5, E and F), maximal fluorescence was reached in the absence of CpeS after 10 min followed by an intensity decrease (Fig. 5, B and C). CpeS-mediated reactions as well as the spontaneous reaction of 3Z-PEB with CpeB were largely completed within 40 min, whereas addition of free 3E-PEB to CpeB was not completed within 90 min. This was confirmed by following spontaneous addition of PEB to CpeB via absorption spectroscopy (supplemental Fig. S5). Mixing of CpeB with either 3E-PEB or 3Z-PEB resulted instantly in a blue shift of the long wavelength absorption maximum of bilins. Whether this blue shift reflects the formation of a thioether linkage between CpeB and PEB, conformational changes of the bilin or a combination of both is unclear. The blue-shifted maximum decreased while simultaneously a red-shifted absorption maximum increased. In the case of 3Z-PEB, the reaction was completed after 40 min (supplemental Fig. S5, C and D). 3E-PEB addition to CpeB was not completed until at least 2 h (supplemental Fig. S5, A and B). Lyase-mediated transfer reactions were not    Fig. S3). As described above, CpeS is a dimer with two bilins bound. To verify the oligomerization state of the lyase target protein CpeB, size exclusion chromatography with tag-free CpeB was performed (supplemental Fig. S6). After incubation with 3E-PEB for 2 h, CpeB eluted mainly as complex of ϳ45 kDa as judged by the absorbance of bound bilin (supplemental Fig.  S6A). This correlates well with the calculated mass of a CpeB homodimer (43.4 kDa). An additional broad shoulder corresponding to either higher ordered oligomers or aggregates was detected. These CpeB fractions had no bilin bound. In the absence of bilin, the elution profile of CpeB was essentially the same as in the presence of 3E-PEB (supplemental Fig. S6B). However, the amount of higher ordered oligomers or aggregates increased in the absence of bilin. These data suggest interaction of dimeric CpeS with dimeric CpeB. One might speculate that both Cys 82 sites in the CpeB dimer are supplied with PEB before the CpeS⅐CpeB complex falls apart.
CpeS Is a 3Z-PEB-transferring Lyase-CpeB was affinity purified after 45 min of incubation with either free bilin or CpeS⅐bilin. As shown in Fig. 6 and Table 2, CpeB chromophorylation products differed significantly in their spectroscopic properties. Holo-CpeB generated by incubation with free bilins or with CpeS⅐3E-PEB exhibited at least two absorption maxima, with one at ϳ610 nm typical for phycobiliprotein-bound DHBV (43,44). Only CpeS⅐3Z-PEB-mediated reactions resulted in formation of holo-CpeB with identical spectroscopic properties as native ␤-PE from P. marinus MED4 (18).
Bilin types and binding sites in CpeB were determined by analysis of CpeB bilin peptides via HPLC (supplemental Fig. S7) and MALDI-TOF MS and MS/MS (supplemental Fig. S8 and Table 3). CpeS-mediated addition of 3Z-PEB to CpeB resulted in binding of PEB to Cys 82 , whereas transfer of 3E-PEB yielded in addition Cys 82 -DHBV. Spontaneous addition of 3E-PEB and 3Z-PEB resulted in formation of Cys 82 -DHBV and of a second product whose identification via MALDI-TOF failed as for unknown reasons no bilin peptides were detectable. Absorption spectra of this product were nearly identical to those of lyase-produced Cys 82 -PEB, but elution times differed significantly. The PEB attachment site of spontaneous bilin addition to CpeB was indirectly narrowed down to Cys 82 by several lines of evidence: (i) binding of PEB to the CpeB tag or to impurities was ruled out by repeating experiments with truncated CpeB (data not shown); (ii) modification of CpeB cysteine residues using iodoacetamide abolished bilin binding (data not shown) (18); (iii) CpeB Cys 82 mutants do not form fluorescent adducts with 3E-PEB in vitro (18). Due to the nearly identical absorbance of bilin peptides 1 and 2, we concluded that both carry PEB at ␤Cys 82 . Differences in HPLC retention times of spontaneously and CpeS-generated ␤Cys 82 -PEB peptides must be due to altered conformation or configuration of the bilin peptide.

DISCUSSION
S/U-type lyases are classified into five groups (45). Until now, only members of groups CpcS-I, CpcS-III, and CpcU were well

3Z-PEB⅐␤Cys 82 Lyase CpeS
studied. These lyases attach PCB to ␤Cys 84 of phycocyanin or phycoerythrocyanin and to ␣Cys 82 and ␤Cys 82 of allophycocyanin (38,39,46). 4 While the function of CpcV group lyases is enigmatic, S-type lyases of groups CpeS and CpeU were speculated to attach PEB to PE subunits, due to their encoding in gene clusters associated to PE maturation (45). We have now characterized the first CpeS-like lyase from P. marinus MED4 and demonstrated that CpeS is a bilin lyase specific for attachment of 3Z-PEB to ␤Cys 82 of PE. S/U-type lyases exhibit different oligomerization states: heterodimeric, homodimeric, or monomeric (38 -40). As P. marinus MED4 CpeS is homodimeric, S-type lyases appear to be mainly dimeric with the exception of Nostoc sp. PCC7120 CpcS (39). However, the assumption of a CpcS monomer is based on missing self-interaction in pulldown assays, which might be due to stable dimer formation prior to experiments.
Dimeric CpeS binds two molecules of bilin with similar binding kinetics for 3E-PEB, 3Z-PEB, and DHBV. DHBV is a biosynthetic intermediate transferred from the ferredoxin-dependent bilin reductase PebA to its homolog PebB probably by metabolic channeling (4). PebB, in turn, catalyzes the reduction of DHBV to 3Z-PEB. Considering direct transfer of DHBV from PebA to PebB binding of DHBV by CpeS in vivo seems unlikely. In line with that assumption, CpeS does not attach DHBV to CpeB in vitro. However, DHBV-transferring lyases are at least feasible for cryptophytes with DHBV-containing phycobiliproteins (47). As lyases are also able to detach bilins (8,41,48), CpeS might remove incorrectly bound DHBV from CpeB, but this could not yet be verified in preliminary studies (data not shown).
Although CpeS binds 3E-PEB, 3Z-PEB, and DHBV tightly, there is no complex formation with BV IX␣, 3E-PCB, or 3Z-PCB. The common difference between these two groups of bilins is the reduction state of the methine bridge between C15 and C16. CpeS bound bilins have a reduced 15,16 bond and thereby a new chiral center at C16 in the R configuration (49). In contrast, PCB-specific lyase-isomerase PecE/PecF from Mastigocladus laminosus binds only bilins with a 15,16 double bond (50). Thus, binding of the correct bilin may be ensured in vivo by a specific interaction of lyases with the 15,16 bond of either PCB or PEB. However, CpcS1 from Nostoc sp. PCC7120 and CpcE/CpcF from Synechococcus sp. PCC7002 bind both PEB and PCB (8,35,41). One might argue that these species synthesize only PCB but not PEB, so that discrimination between these two bilins is unnecessary. Further comparing studies of bilin binding by PCB-and PEB-transferring lyases must be carried out to address this interesting hypothesis.
Although CpeS binds bilins with a reduced 15,16 bond, only those with a further reduction of the A-ring diene system are transferred to CpeB, namely 3E-and 3Z-PEB. The A-ring diene system reduction of BV IX␣ results in two possible stereoisomeric ethylidene groups at the C3 carbon atom, 3Z and 3E. 3Z-isomers are the biosynthetic products of different ferredoxin-dependent bilin reductases 3 (4,5,28,29,37,51). However, 3E-isomers are often byproduct of bilin extraction methods and bilins isolated from phycobiliproteins by methanolysis are mainly 3E-configured. CpeS transfers both 3E-PEB and 3Z-PEB to CpeB, but only 3Z-PEB yields exclusively correctly bound PEB at ␤Cys 82 and holo-CpeB with identical spectroscopic features as native ␤-PE (18). 3E-PEB addition by CpeS is less effective and yields oxidized side products. Therefore, 3Z-PEB is the natural substrate for CpeS-mediated transfer to CpeB. Binding and transfer of 3E-and 3Z-isomers was shown for other lyases as well. In some cases, the functionality of lyases was proven in in vivo reconstitution systems and in in vitro assays. While the in vivo assays provide 3Z-PCB as substrate, the in vitro assays were performed with 3E-PCB. In agreement with the results of this study, CpcS1 and CpcT1 from Nostoc sp. PCC7120 are able to bind and transfer both PCB isomers (8,35,52). Although assays with 3Z-PCB result in a correctly bound chromophore, attachment of 3E-PCB by CpcT1 is of low fidelity and leads to partial oxidation to mesobiliverdin (52). Products of CpcS1 mediated 3E-and 3Z-PCB transfer to ␤-phycoerythrocyanin have nearly identical spectroscopic features as native ␤-phycoerythrocyanin, but only 3Z-PCB addition products match extinction coefficients of native ␤-phycoerythrocyanin (39). Synechococcus sp. PCC7002 CpcT transfers 3E-PCB and 3Z-PCB to phycocyanin but only spectroscopic data of the latter product are perfectly in agreement with those of native phycocyanin (53). In conclusion, lyases bind bilins with different stereochemistry of the C3 ethylidene group. Both isomers can be attached to phycobiliproteins but the 3E-isomer is only partially protected from oxidation. Thus 3Z-isomer specificity may be a common feature of all lyases. This is in agreement with 3Z-isomers as the primary product of bilin biosynthesis and points toward direct bilin transfer from the last biosynthetic enzyme to the lyase.
Binding of 3Z-PEB to CpeS is fast with an association rate constant of 1.4 M Ϫ1 s Ϫ1 and in competition assays with CpeB, added bilin was completely bound by the lyase (data not shown). As already speculated for other lyases (8), rapid bilin binding to CpeS may prevent nonenzymatic transfer to CpeB, which results in false addition products. Spontaneous addition of PEB to CpeB is of low yield and fidelity and addition products 4 The consensus numbering of bilin attachment sites is given.

TABLE 3 Identification of tryptic holo-CpeB bilin peptides via MALDI-TOF
Holo-CpeB was reconstituted with 3E-PEB, 3Z-PEB, CpeS⅐3E-PEB, or CpeS⅐3Z-PEB as indicated. Three different tryptic bilin peptides (peak nos. 1, 2, 3) of holo-CpeB were separated by HPLC and identified via MALDI-TOF. Bilin peptide sequences are given with the attached bilin in bold face in brackets. Calculated (Cal) and experimentally determined (Exp) m/z of bilin peptides are given. ND, not determined. are ␤Cys 82 -DHBV and most likely a ␤Cys 82 -PEB with an altered, incorrect configuration or conformation. Indeed, nonenzymatic addition of bilins to other phycobiliproteins was shown to be inefficient and to produce oxidized side products (11,35,38,39,46,53). In the case of CpcA, CpcB, PecA, and PecB from Mastigocladus laminosus, PCB adds spontaneously to the correct Cys 84 binding site but in a non-native configuration (54). Whereas bilins in solution are in all Z configuration and all syn-conformation (55)(56)(57), phycobiliprotein-bound bilins are mainly Z,Z,Z/anti,syn,anti-configured (58 -60). However, upon denaturation or digestion of the phycobiliprotein matrix bilins are released and adopt their Z,Z,Z/syn,syn-,syn-porphyrin-like form again. This is in line with the almost identical absorption of non-enzymatic and CpeS-catalyzed ␤Cys 82 -PEB peptides during HPLC. Differences in retention times might be due to differences in PEB binding to ␤Cys 82 . Bilin addition to phycobiliproteins generates two new chiral carbon atoms, C3 and C3Ј, with two possible configurations each. Whereas the configuration of C3 is always R, the configuration of C3Ј at binding sites ␣Cys 84 , ␣Cys 143 , and ␤Cys 84 of different phycobiliproteins is R, whereas it is S for binding sites ␤Cys 155 and ␤Cys 50/61 (58,59,(61)(62)(63). We therefore speculate that CpeS-mediated attachment of PEB yields ␤Cys 82 -C3'(R)-PEB, while non-assisted attachment of PEB results in formation of ␤Cys 82 -C3Ј(S)-PEB. However, verification of the C3Ј configuration by NMR failed so far, due to insufficient amounts. Current research in our laboratory will address this further to tackle this interesting question.