Lyase Activities of CpcS- and CpcT-like Proteins from Nostoc PCC7120 and Sequential Reconstitution of Binding Sites of Phycoerythrocyanin and Phycocyanin β-Subunits*

Genes all5292 (cpcS2) and alr0617 (cpcS1) in the cyanobacterium Nostoc PCC7120 are homologous to the biliprotein lyase cpcS, and genes all5339 (cpcT1) and alr0647 (cpcT2) are homologous to the lyase cpcT. The functions of the encoded proteins were screened in vitro and in a heterologous Escherichia coli system with plasmids conferring biosynthesis of the phycocyanobilin chromophore and of the acceptor proteins β-phycoerythrocyanin (PecB) or β-phycocyanin (CpcB). CpcT1 is a regioselective biliprotein lyase attaching phycocyanobilin exclusively to cysteine β155 but does not discriminate between CpcB and PecB. The in vitro reconstitutions required no cofactors, and kinetic constants were determined for CpcT1 under in vitro conditions. No lyase activity was found for the lyase homologues CpcS2 and CpcT2, but complexes are formed in vitro between CpcT1 and CpcS1, CpcT2, or PecE (subunit of phycoviolobilin:α-phycoerythrocyanin isomerase lyase). The genes coding the inactive homologues, cpcS2 and cpcT2, are transcribed in N-starved Nostoc. In sequential binding experiments with CpcT1 and CpcS1, a chromophore at cysteine 84 inhibited the subsequent attachment to cysteine 155, whereas the inverse sequence generates subunits carrying both chromophores.

F. diplosiphon they had originally been identified as part of a C-phycoerythrin operon 8 (23). CpcS1 is a nearly universal bilin: Cys-84-phycobiliprotein lyase that catalyzes the regiospecific covalent attachment of PCB to cysteine 84 of allophycocyanin and the ␤-subunits of PEC (PecB) and CPC (CpcB) as well as of phycoerythrobilin to both subunits of C-phycoerythrin (26,27). Likewise, CpcT from Synechococcus PCC7002 catalyzes the regioselective attachment of PCB to cysteine ␤155 of CpcB (21). Nostoc PCC7120 has two rod phycobiliproteins, CPC and PEC, and two sets of EF-type lyases that are specific for the respective ␣-subunits (18). It also has a pair of genes homologous to cpcS (all5292 (cpcS2) and alr0617 (cpcS1)) and a pair of genes homologous to cpcT (all5339 (cpcT1) and alr0647 (cpcT2)) (14). We show that only one of each pair, cpcS1 and cpcT1, codes for active lyases that catalyze PCB attachment to cysteines ␤84 and ␤155, respectively, of both CPC and PEC. Using the two enzymes PCB can be attached correctly at both cysteines but only if they are used sequentially and in the order CpcT1 followed by CpcS1 because PCB at cysteine ␤84 interferes with attachment at cysteine ␤155.
The expressed proteins were first purified via Ni 2ϩ affinity chromatography on chelating Sepharose (Amersham Biosciences) (26). If necessary, the affinity-enriched proteins were further purified via fast protein liquid chromatography (Amersham Biosciences) with a Superdex 75 column developed with potassium phosphate buffer (KPB; 50 mM, pH 7.0) containing NaCl (150 mM) or with a DEAE fast flow column developed with a gradient of 0 to 1 M NaCl in KPB (20 mM, pH 7.1) (27).
Reverse Transcription PCR-Nostoc PCC7120 was grown in BG-11 medium at 28°C under continuous illumination from a fluorescent lamp. To induce heterocyst differentiation, filaments were washed and then resuspended in nitrogen-free medium (BG-11 0 ). Total RNA was isolated by the hot phenol method (31). The crude RNA was treated with 5 units of DNase I (Fermentas, Beijing) at 37°C for 30 min. One microgram of RNA was used in each reverse transcription-PCR experiment. Reverse transcription-PCR was performed using One Step RNA PCR Kit (TaKaRa, Dalian, China) according to the manufacturer's instructions.
Quantification of Phycocyanobilin and Protein-PCB was prepared as described before (15). PCB concentrations were determined spectroscopically using an extinction coefficient ⑀ 690 ϭ 37,900 M Ϫ1 cm Ϫ1 in methanol, 2% HCl (32). Protein concentrations were determined with the Bradford method using bovine serum albumin as the standard (33).
SDS-Polyacrylamide Gel Electrophoresis-SDS-PAGE was performed with the buffer system of Laemmli (34). The gels were stained with Coomassie Brilliant Blue R for the protein and with ZnCl 2 for bilin chromophores (35).
Complex Formation among Lyases and Receptor Proteins-One of the untagged proteins, CpcS1, CpcT1, NblB, CpcS2, CpcT2, CpcE, CpcF, PecE, or PecF, was incubated overnight with His-tagged CpcS1, CpcT1, NblB, CpcS2, CpcT2, CpcE, CpcF, PecE, or PecF at 4°C. The mixtures were then loaded on a Ni 2ϩ affinity column, washed 3 times with 5 column volumes of start buffer containing varying concentrations of NaCl (see below), once with the same buffer containing in addition 50 mM imidazole, and finally with the same buffer containing 0.5 M imidazole. Three parallel experiments were carried out with NaCl concentrations of 0.2, 0.5, or 1 M in all washing steps. The eluate from the last wash was analyzed by SDS-PAGE.
Spectroscopy-UV-visible absorption spectra were recorded with a Lamda 25 spectrophotometer (PerkinElmer Life Sciences, Shanghai). Formation of the photochromic PVB-PecA (i.e. ␣-PEC) in the lyase reaction was monitored by the absorption at 570 nm and by double-difference spectroscopy of the reversible photoreaction of the PVB chromophore, as previously described (36). Fluorescence spectra were recorded with an LS 45 spectrofluorimeter (PerkinElmer Life Sciences) and are not corrected. Circular dichroism (CD) was measured with a J-810 CD spectrometer (Jasco, Munich, Germany).
Kinetic tests were carried out with purified proteins as published (7,17,18,26). K m , V max , and k cat were calculated from Lineweaver-Burk plots using Origin V7 (Origin Lab Corp., Munich, Germany).
For mass spectrometry, chromoproteins (10 M) were digested with trypsin (40 M) in KPB (100 mM, pH 7.0) for 4 h at 37°C. After purification (27), the isolated chromopeptides were analyzed by mass spectrometry in positive ion mode using a Q-Tof Premier mass spectrometer (Waters Micromass Technologies, Manchester, NH) with a nanoelectrospray ionization source.
The site selectivity was verified by the absorption and fluorescence maxima ( max ϭ 625 nm) of the purified PCB chromoproteins (Fig. 2). These values compare well to the respective band positions assigned to the ␤155-chromophores of PEC and CPC from M. laminosus (41)(42)(43). The absorption and fluorescence spectra were very similar when PecB or CpcB from Nostoc PCC7120 were used as acceptor proteins, and the extinction coefficients and fluorescence yields of all products ( Table 2) are typical of native phycobiliproteins (43). Two of the products, PCB-PecB(C84A) and PCB-CpcB(C84S), were purified and analyzed in more detail.

TABLE 1 Estimation of chromoproteins produced in E. coli with enzymes from Nostoc PCC7120 and apoproteins from M. laminosus in the absence of any lyase and under catalysis of one or two of CpcS1, CpcT1, CpcS2, and CpcT2
Chromoprotein production was evaluated from the fluorescence of the supernatants of the broken E. coli cells (26). Data were obtained from two independent measurements. In each column they are given relative to the yield of chromoprotein formed under catalysis of CpcT1 only.

Lyase
Chromoprotein production a

PCB-PecB(C84A) PCB-PecB PCB-CpcB(C84S) PCB-CpcB
Product of PCB attached at Cys-84 due to catalysis of CpcS1 (26). The productivity of PCB attached to CpcB and PecB under catalysis of only CpcS1 is given for comparison in the last row.
SDS-PAGE ( Fig. 3) showed proteins of the expected size that strongly fluoresced in the presence of Zn 2ϩ , as is characteristic for bilins covalently bound to the proteins (35). The biosynthesized PCB-CpcB(C84S) had the typical CD spectrum of PCB chromophores in native phycobiliproteins (42); that is, with a positive visible band and a negative near-UV band, but the CD of PCB-PecB(C84A) and, therefore, its conformation was different (Fig. 4). In this case, the "typical" spectrum resembling that of native ␤-PEC was only obtained when the second chromophore, PCB-Cys-84, was also attached to PCB-Cys-155-PecB by the action of CpcS1 (supplemental Fig. S1, see below). The far-UV CD spectra are typical for ␣-helical proteins. Regioselective attachment in the chromoproteins is finally supported also by HPLC of peptic digests. The chromopeptides obtained from PecB(C84A) and CpcB(C84S) were subsets of those obtained from the respective wild-type subunits isolated from M. laminosus (Fig. 1), and these subsets are complementary to those obtained previously from the respective chromoproteins reconstituted in the presence of the ␤84-lyase, CpcS1 (26). The point of attachment was confirmed by mass spectrometry of the major chromopeptides (isolated by HPLC) of the tryptic digest (Table 3). Peptic chromopeptides were less suitable for mass spectrometry due to the lower specificity of peptic digestion and the lack of charged amino acids in some peptides. From the data we conclude that CpcT1 catalyzes the site-selective attachment of PCB to cysteine ␤155 in both PecB and CpcB. 9 When the above PEC and CPC ␤-subunits and mutants were synthesized heterologously in E. coli, the cells showed strong fluorescence; the corresponding excitation spectra were similar to the respective absorption spectra (supplemental Fig. S2). This excludes the possibility that reconstitution took place after cell disruption (26). A spontaneous in vivo addition of PCB to the apobiliproteins (6, 44 -47) was further excluded by a series of control experiments, where pCDF-cpcT1 was omitted in the transformation of BL21(DE3), and hence, the lyase was absent in the cells after induction. Under these conditions only trace amounts were formed of the chromoproteins (PCB-PecB(C84A), PCB-PecB, PCB-CpcB(C84S), or PCB-CpcB) when compared with the production in the presence of CpcT1 (Table 1).
Incubation of PCB and affinity-purified PecB(C84A) or CpcB(C84S) with CpcT1 produced the same chromoproteins as the respective E. coli system (Fig. 2). This system, however, has a lower fidelity because some chromophore binding was also observed in Cys-84 in mutants in which cysteine ␤155 had been replaced by isoleucine (i.e. PecB(C155I) and CpcB(C155I)). These products, however, had red-shifted absorption and fluorescence maxima, and the fluorescence intensity was greatly reduced (supplemental Fig. S3); both are typical for spontaneous PCB addition (and for partial oxidation to mesobiliverdin) (45,47). In conclusion, CpcT1 catalyzes the site-selective attachment of PCB to cysteine ␤155 in both PecB and CpcB. This site selectivity corroborates the identification of CpcT as a PCB:Cpc-Cys-␤155 lyase in Synechococcus PCC7002 (21) and shows at the same time that of the two homologues present in Nostoc PCC7120, only CpcT1 is active. The roles of CpcS2 10 and CpcT2 still remain unclear. Their transcription was studied in Nostoc PCC7120 by reverse transcription-PCR under different growth conditions. Neither gene was transcribed when cells were grown in nitrogen-replete medium, but transcripts for both genes were detected when cells were grown in N-free medium (supplemental Fig. S4). They do not appear, therefore, to be pseudogenes.
Sequential Addition of PCB to Cys-84 and Cys-155-It has been previously shown that CpcS1 acts as a PCB:Cys-␤84-phycobiliprotein lyase capable of attaching PCB to cysteine ␤84 of PecB and CpcB, respectively, and resulting in PCB-Cys-84-PecB and PCBCys-84-CpcB (26). We now tested whether by combined catalysis of CpcT1 and CpcS1, holo-␤-PEC and ␤-CPC could be generated in the E. coli system. pho1-pcyA, pETDuet-cpcT1, and pCDF-cpcS1 were co-transformed into E. coli BL21(DE3) together with pCOLA-pecB; the purified phycobiliprotein formed, however, had an absorption maximum at 602 nm, which is redshifted from that of ␤-PEC isolated from M. laminosus ( max ϭ 598 nm) (42). The same reconstitution based on pCOLA-cpcB resulted in a chromoprotein with an absorption maximum at 618 nm, which again is red-shifted compared with that of ␤-CPC isolated from M. laminosus ( max ϭ 609 nm) (41,43). By contrast, however, the fluorescence maxima were in both cases the same for the reconstituted chromoprotein and the respective isolated subunit (629 and 644 nm for ␤-PEC and ␤-CPC, respectively). In both ␤-PEC and ␤-CPC, PCB-Cys-84 absorbs at a longer wavelength than PCB-Cys-155 and is the predominantly fluorescing chromophore (41)(42)(43). The data, thus, indicate the correct attachment of PCB to Cys-84 but an incomplete or incorrect reconstitution of PCB-Cys-155, which absorbs at shorter , and PCB-CpcB (F); spectra were recorded of the supernatants of broken cells (solid lines) and after the Ni 2ϩ affinity chromatography and subsequent dialysis against start buffer (dashed lines). All acceptor proteins were from M. laminosus; PecB and CpcB from Nostoc PCC7120 gave very similar results (see Table 2).

TABLE 2 Quantitative absorption and fluorescence data of the reconstituted and biosynthesized PCB-PecB(C84A), PCB-PecB, PCB-CpcB(C84S), and PCB-CpcB
Proteins were purified by Ni 2ϩ affinity chromatography and dialysis. The data were averaged from two independent measurements.

Phycobiliprotein
Absorption a Fluorescence max (Q Vis/uv ) wavelengths than PCB-Cys-84. This is also supported by a comparison with the respective subunits carrying only PCB-Cys-84 (Fig. 5A).
CpcT1 and CpcT2 form stable complexes (see below). Therefore, "inactive" CpcT2 was tested to determine whether it affects the double-reconstitution by CpcS1 and CpcT1 by using a tetra-plasmidic E. coli system containing pho1-pcyA, pcpcS1-cpcT1, pETDuet-cpcT2, and pCOLA-pecB (or pCOLA-cpcB). The reconstituted products, however, again had red-shifted absorption maxima when compared with the respective isolated subunits, but the fluorescence maxima were the same (Fig.  5B). A similar test was also negative when both cpcS2 and cpcT2 were co-expressed in E. coli; that is, in a system containing both CpcS1, CpcT1, and CpcS2, CpcT2. Finally, we also attempted the co-expression of alr0616 (nblB), coding for NblB, which is homologous to CpcE and involved in the degradation of phycobiliproteins (48). However, there was again no effect (not shown). It is concluded, therefore, that the established Cys-84 and Cys-155 lyases, alone or in combination with their homologues, are not yet sufficient to fully reconstitute holo-␤-PEC or ␤-CPC in E. coli. In all experiments, the expression of the respective genes was verified by SDS-PAGE of the cell lysates; the failure of PCB addition to Cys-␤155 despite the presence of CpcT1 must, therefore, have another reason.
The presence of more than one binding site raises the possibility that a particular sequence may be required for chro-   mophore addition. For instance, a chromophore at Cys-␤84 could inhibit the subsequent addition of the second chromophore to Cys-␤155. This is indeed suggested by the follow-ing experiments using sequential application of the lyases; the apoprotein, CpcB, was either first treated with CpcS1 and then with CpcT1 or vice versa. In a first set of experiments, both reactions were performed in vitro. When PCB was first attached by CpcS1 at Cys-84 of CpcB, this chromophore inhibited the subsequent correct attachment of the second chromophore at Cys-155; the red-shifted absorption of the final product is similar to that of the reconstitution product shown in Fig. 5, A and B, which was obtained in E. coli expressing both cpcT1 and cpcS1. When, conversely, PCB was first attached to Cys-155 of CpcB, then the second PCB was attached to Cys-84 of the purified PCB-Cys-155-CpcB. This second addition, however, is incomplete, and the chromophore is also partly oxidized as judged from the red-shifted shoulder in the absorption spectrum (Fig. 5, C and D). Such oxidations are frequent side products during the spontaneous addition reaction (45,47).
In the second set of experiments, the first reconstitution step was done in E. coli, and the second reconstitution step was done in vitro. Because we had previously observed that side reactions are reduced when crude E. coli extracts replaced purified protein for reconstitutions reactions (26), the second step was attempted in two versions using either the crude extract of the first addition reaction in E. coli or the affinity-purified sample. When PCB was attached first in E. coli to Cys-84, neither version proved successful. Treatment of purified PCB-Cys-84-CpcB with PCB in the presence of CpcT1 results in a product with broadened absorption that is untypical  for a native biliprotein. The situation is somewhat improved if the purification step for PCB-Cys-84-CpcB is omitted; the absorption is sharpened, indicative of a more specific binding, but still broadened compared with native ␤-subunits (Fig. 5E). This indicated that the ␤84-PCB interfered with the addition at Cys-␤155. If the sequence of additions was inverted, a broadened absorption was also obtained when purified PCB-Cys-155-CpcB (produced in E. coli) was treated in vitro with CpcS1 (Fig. 5F, dashed line). In this case, however, a (nearly) native ␤-subunit was obtained when the supernatant of broken E. coli cells containing unpurified PCB-Cys-155-CpcB was treated with PCB in the presence of CpcS1 (Fig. 5F, see also supplemental Fig. S1). The purified product from this sequential reconstitution had a narrow absorption ( max ϭ 612 nm), a high Q Vis/UV ratio, and an intense, narrow emission ( max ϭ 644 nm). Dual chromophorylation was confirmed by HPLC analyses (Fig. 6 and supplemental Fig. S5). Even under these optimum conditions, however, Cys-84 seemed to be incompletely occupied. First, the absorption maximum (612 nm, Fig. 5F) was still a little red-shifted as compared with that of holo-␤-CPC (609 nm (41)). Second, the abundance of the PCB-Cys-84-derived chromopeptides was somewhat reduced as compared with those obtained from isolated ␤-CPC (Fig. 6). Quantitatively, the PCB-Cys-84 and PCB-Cys-155 chromopeptide ratio of 0.7:1 indicates that 30% of Cys-84 still lacks PCB. Taken together, these experiments indicate that PCB bound to Cys-84 interferes with the subsequent, CpcT1-catalyzed binding of the second PCB to Cys-155, whereas PCB bound to Cys-␤155 allows the attachment of the second PCB to Cys-␤84 by CpcS1. Even in this case, however, optimum conditions must be maintained to avoid side reactions. Either some factor in the crude E. coli extract may assist correct binding or a cyanobacterial factor is missing.
As pointed out by one reviewer, one such cyanobacterial factor could be the phycocyanin ␣-subunit, because CpcA and CpcB form heterodimers also in the absence of chromophores. The addition of CpcA to the in vitro reconstitutions did not change the situation; however, there was in no case an improvement of chromophore binding. Similarly, the ␣-lyases CpcE/F or PecE/F were ineffective (not shown).
Reconstitutions of PecB gave very similar results. In particular, doubly chromophorylated, nearly native ␤-PEC was again obtained only when PCB was attached first in E. coli to Cys-155 of PecB with CpcT1, and then the supernatant was treated with PCB and CpcS1 for chromophorylation at Cys-␤84 (supplemental Fig. S1). Also, a similar occupancy ratio (0.7:1) was estimated from HPLC analysis of the chromopeptides (Fig. 6 and supplemental Fig. S5).
In control experiments we found no interference of the N-terminal tags on either the in vitro or the E. coli reconstitutions (not shown). The failure to fully reconstitute the ␤-subunit also cannot be attributed to the fact that in most experiments the lyases and the apoproteins are from different organisms (M. laminosus and Nostoc PCC7120, respectively), since reconstitution experiments with all proteins from Nostoc PCC7120 resulted in practically identical behavior (an example of such homologous reconstitutions is shown in supplemental Fig. S1).
The above results imply that the lyase actions of CpcT1 and CpcS1 need to be well coordinated. To search for a possible coordinator for CpcT1 and/or CpcS1, the interactions of His-tagged CpcS1 and CpcT1 were tested by Ni 2ϩ affinity chromatography with the following proteins: CpcS1, NblB, CpcS2, CpcT2, CpcE, CpcF, PecE, or PecF. Like CpcS1 (26), CpcT1 did not form a complex with CpcS2, NblB, CpcE, CpcF, or PecF. It does so, however, with CpcS1, CpcT2, or PecE (Fig. 7). These complexes were then tested for their capacity to attach PCB to both sites, Cys-84 and Cys-155, of PecB or CpcB. With none of them, holo-␤-PEC or ␤-CPC could be formed (not shown). These negative results, like those in E. coli (see above), indicate the limits of the heterologous system and suggest other control factor(s) involved in the biosynthesis of holo-␤-PEC or ␤-CPC in Cyanobacteria.

DISCUSSION
Compared with enzymatic chromophore attachment to the ␣-subunits of phycobiliproteins, studies have only recently begun on the catalytic attachment of chromophores to the ␤-subunits, where the process is complicated by the presence of two or even more binding sites (1,2,27,49). Nonenzymatic attachment of PCB to PecB and CpcB from M. laminosus indi- cated that both binding sites can bind the chromophore autocatalytically and that the site specificity can be influenced by the reaction medium; binding of Cys-155 is favored under the influence of the detergent, Triton X-100, and binding to Cys-84 in its absence (30). The autocatalytic reaction, however, is unspecific. Moreover, experiments with multiply transformed E. coli indicate that a spontaneous addition to either site is very low in vivo, at least in E. coli (26,27). The identification of a set of four genes (cpcS, -T, -U, and -V) that catalyze the addition of PCB to phycobiliprotein ␤-subunits (22) initiated a better understanding of the process. Distinct specificities for the two binding sites of ␤-CPC and ␤-PEC recently became obvious for two of these proteins. Under catalysis of CpcS1 from Nostoc PCC7120, PCB is attached regioselectively to cysteine ␤84 (26). Conversely, CpcT from Synechococcus PCC7002 catalyzed the attachment to cysteine 155 of CpcB (21). The site specificity of CpcT is fully supported by our finding that the homologous CpcT1 from Nostoc PCC7120 catalyzes the attachment of PCB to cysteine ␤155 of CpcB and PecB. It is noteworthy that most of the work reported here was done with PecB and CpcB from M. laminosus, and the lyase was from Nostoc PCC7120 but CpcT1 catalyzes equally well the PCB attachment to PecB and CpcB from Nostoc PCC7120. A similar cross-reactivity was shown earlier for the cysteine 84 lyase, CpcS1 (26). Taken together, chromophore attachment to the ␤-subunits of PecB and CpcB is catalyzed by enzymes that are specific for the two binding sites, Cys-␤84 and Cys-␤155, but both lyases have a broader specificity to their substrate proteins than the ␣84 lyases, CpcE/F and PecE/F. Another difference to the latter is that neither CpcS1 nor CpcT1 catalyzed transfer of the chromophore to another apoprotein, viz. PecA (see the supplemental material). Among the biliprotein lyases so far studied, only CpcE/F readily catalyzes both the forward and the backward reactions (10,26). 11 In contrast to Synechococcus PCC7002 (21,22), two copies each of cpeS-and cpeT-like genes are present in Nostoc PCC7120. Only CpcT1 is capable of correctly attaching PCB to cysteine 155 of both PecB and CpcB, whereas its homologue CpcT2 is inactive in vitro and in E. coli. The kinetic constants of CpcT1 (see the supplemental material) are comparable with those of the E/F-type ␣84 lyases (11,18,50) and the CpcS-type ␤84 lyase (26). Similarly, only CpcS1 catalyzes the attachment of PCB to cysteine 84 of both CpcB and PecB, whereas its homologue, CpcS2, is inactive. It should be noted that k cat of CpcT1 is lower than those of CpcS1 (26,27) and the EF-type lyases (11,18,50).
Synergistic effects have been reported for the new type of lyases when used in combination, but the specificities toward the different subunits depends on the particular mixture of lyases used (22). The combining of lyases also affected the activities of the individual lyases studied in this work. Under heterologous (E. coli) conditions, the inactive homologues influenced the activities of the "active" homologues, but generally only negatively and to a limited degree (Table 1). Because neither CpcS2 or CpcT2 nor their mixtures with the active CpcS1, CpcT1, CpcE/F, PecE/F, and NblB could remove the chromophore from the respective chromoproteins and allow transfer (by PecE/F) to PecA, it is also unlikely that CpcS2 or CpcT2 are involved in chromophore detachment. They may be involved in regulating the active lyases, CpcS1 and CpcT1, as indicated by the complex formation between CpcT1 with CpcS1, CpcT2, or PecE (Fig. 7). It is unclear, however, under which conditions they may be co-expressed in Nostoc PCC7120; preliminary experiments indicate that cpcS1 and cpcT1 are constitutively expressed, whereas cpcS2 and cpcT2 are expressed during N-starvation, but these observations require more detailed investigations because it is conceivable that both homologues are present during transients or under other physiological conditions. The evolution of site-specific enzymes may relate to the different binding situations of the two chromophores on the ␤-subunits. The conformation of PCB bound to cysteine ␤155 differs from that of PCB bound to cysteines ␤84 (and ␣84) (51) as does the stereochemistry of the asymmetric C-3 1 of PCB that is generated together with the asymmetric C-3 during the addition reaction. In the high resolution structures, C-3 1 is R-configured at PCB bound to cysteine 84, whereas it is S-configured in PCB bound to cysteine 155 (52)(53)(54)(55). Conceivably, this different binding requires a different conformation of the chromophore when it is presented by the lyase to the binding site, as indicated by a conformation-dependent site selectivity of the autocatalytic addition reaction (30).
An intriguing problem of phycobiliproteins that has not been experimentally addressed until now is the coordination of chromophore attachment to subunits with multiple binding sites. Assembly of ␣ and ␤ subunits to holo-CPC is affected in mutants lacking either chromophore, which may relate to pleiotropic effects in lyase mutants (9,21,23,(55)(56)(57)(58). The results obtained here by using a combination of CpcS1 and CpcT1 indicate for the first time that a specific attachment sequence may be required for multichromophoric biliprotein subunits. Both in E. coli and in vitro, a chromophore at cysteine 84 inhibits the subsequent attachment to cysteine 155, whereas a chromophore bound to cysteine 155 allows attachment to cysteine 84, even if only incompletely. The inhibitory effect of PCB at cysteine 84 is observed most clearly by the sequential binding experiments, whereas in experiments where both lyases were present, additional modulating effects may occur 11 Chromophore transfer to PecA is a purely operational criterion, as pointed out by one of the reviewers. No microreversibility was tested in the current context nor was the intermediary covalent binding of PCB to the lyase. simultaneously due to interactions among the lyases. In part, the failure to obtain doubly chromophorylated proteins when both lyases are present could relate to the relatively low k cat of CpcT1 as compared with CpcS1. If this situation were reversed in Nostoc by whichever mechanism and/or factor, then the correct sequence could be maintained in the presence of both enzymes. It is intriguing that even when the attachments were performed in the "correct" order, viz. first at cysteine ␤155 and then at cysteine ␤84, best results are obtained if the monochromophorylated product is not purified; that is, when the supernatant from E. coli producing PCB-Cys-155-CpcB was treated directly with CpcS1 (Fig. 5F). In this extract CpcS1 is present as well, but the co-incubation data argue against the requirement of a ternary complex. Rather, some factors in E. coli extracts, possibly chaperone-type proteins, seem to help in the process. Although the need for additional factors may relate to the heterologous system, there are reports (59) indicating that such interactions may also be relevant in Cyanobacteria. However, it is possible that a cyanobacterial factor(s) may be missing in the E. coli system and thereby interfere with chromophore binding. The only ones tested so far are the complementary apoprotein subunit, CpcA, and the ␣-lyases CpcE/F and PecE/F. Neither of these affected the reaction, but there are at least two other such candidates that need further attention, viz. linker proteins (49) and factors like NblA (60) involved in phycobilisome degradation. The identification and characterization of the CpcS-and CpcT-like lyases in several organisms and of their interactions and also the possibility for the co-expression of the various components in E. coli now opens a new approach to study the assembly of phycobilisomes, the structures responsible for a major fraction of global photosynthesis.