Biliprotein Chromophore Attachment

Biliproteins are post-translationally modified by chromophore addition. In phycoerythrocyanin, the heterodimeric lyase PecE/F covalently attaches phycocyanobilin (PCB) to cysteine-α84 of the apoprotein PecA, with concomitant isomerization to phycoviolobilin. We found that: (a) PecA adds autocatalytically PCB, yielding a low absorbance, low fluorescence PCB·PecA adduct, termed P645 according to its absorption maximum; (b) In the presence of PecE, a high absorbance, high fluorescence PCB·PecA adduct is formed, termed P641; (c) PecE is capable of transforming P645 to P641; (d) When in stop-flow experiments, PecA and PecE were preincubated before chromophore addition, a red-shifted intermediate (P650, τ = 32 ms) was observed followed by a second, which was blue-shifted (P605, τ = 0.5 s), and finally a third (P638, τ = 14 s) that yielded the adduct (P641, τ = 20 min); (e) The reaction was slower, and P605 was missing, if PecA and PecE were not preincubated; (f) Gel filtration gave no evidence of a stable complex between PecA and PecE; however, complex formation is induced by adding PCB; and (g) A red-shifted intermediate was also formed, but more slowly, with phycoerythrobilin, and denaturation showed that this is not yet covalently bound. We conclude, therefore, that PecA and PecE form a weak complex that is stabilized by PCB, that the first reaction step involves a conformational change and/or protonation of PCB, and that PecE has a chaperone-like function on the chromoprotein.

Phycobiliproteins are a family of light-harvesting proteins of cyanobacteria, rhodophytes, and cryptophytes (1)(2)(3)(4). The apoproteins are post-translationally modified by regio-and stereoselective chromophore attachment to cysteines. Addition to the cysteine-␣84 of cyanobacterial phycocyanins and phycoerythrins is catalyzed by the heterodimeric E/F-type lyases (5). The first enzyme of this type studied, phycocyanobilin:cysteine-␣84-phycocyanin lyase, from the cyanobacterium, Syn-echococcus sp. PCC 7002, attaches phycocyanobilin (PCB) 3 to apo-␣-phycocyanin (CpcA) (6). It is encoded by two genes, cpcE and cpcF, located on the phycocyanin (CPC) operon downstream of the two apo-phycocyanin and two linker genes. A similar operon structure has been found in several other organisms, but the respective genes can also be located in other positions (5,7). The functions of the two subunits are still unclear. Chromophore attachment to the phylogenetically related site on the ␤-subunit, cysteine-␤82, is catalyzed by a single-subunit protein, or mixtures of similar proteins (8,9), whereas the chromophores are attached autocatalytically in phytochromes (10,11), allophycocyanin, ApcA (12), and the core-membrane linker, ApcE (13). The situation is complicated by the capacity of apo-phycobiliproteins like CpcA/B to also add chromophores autocatalytically but only slowly and in an error-prone fashion (14 -17).
The isomerizing phycoviolobilin:cysteine-␣84-phycoerythrocyanin lyase is a member of the E/F-type lyases, which, in addition to chromophore attachment, has an isomerase activity thereby converting PCB to phycoviolobilin (PVB). This isomerase activity is related to the ␤-subunit of the enzyme, PecF, where a short, central motif has been identified that is related to this function (18). If only PecE is present, PCB is attached to PecA, and the correct thioether bond is formed to cysteine-␣84, but the chromophore is not isomerized to PVB (19,20). To better clarify the mechanism of the lyase, we have now studied the spontaneous addition of PCB to PecA, and the same reaction in the presence of PecE. Rapid kinetics was investigated by stopped-flow methods, and protein-protein interactions were analyzed by gel filtration. The results support an autocatalytic reactivity of PecA, which is modified by a chaperone-like action of PecE.

EXPERIMENTAL PROCEDURES
Expression and Purification of the Proteins-PecA and PecE from Mastigocladus laminosus (PCC7603) were expressed in Escherichia coli as described (20). Both products carry N-terminal His 6 and S tags and were purified by Ni 2ϩ affinity chromatography (21). Chromatography was repeated twice in potassium phosphate buffer, which was then replaced by Tris-HCl (1 M, pH 6.0) by repeated dialysis, thereby improving the solubility of PecE.
UV-visible Spectra-UV-visible spectra were recorded with a model UV-2401 PC spectrophotometer (Shimadzu, Japan). Enzymatic reactions were studied in the dark at room temperature and were initiated by adding the pigment (stock solution: 1-4 mM in Me 2 SO) to a final concentration of 10 -50 M in the respective cuvette.
Fluorescence Spectra-Fluorescence spectra were recorded with an LS55 luminescence spectrometer (PerkinElmer Life Sciences) at room temperature.
Stopped-Flow Measurements-These were performed with a model SFM-300 machine equipped with two 10-ml syringes and a third 1.9-ml syringe, all controlled by Bio-Kine software (V4.21a), using a TC-100/10T cuvette with 10-mm light path (BioLogic France), and interfaced with a diode array detector (TIDAS MMS 100-1, controlled by TIDASAQ software V2.15, J&M, Germany). The reaction was followed at 20°C with a time resolution of 10 ms. Data were collected in the spectral range 450 -800 nm with data points every 1 nm, thus recording the Q-bands of the chromophore. Solutions were adjusted so that after mixing, there were equimolar amounts of pigment and apoprotein (see "Results" for details). The resulting data were imported to the SpecFit fitting program (SPECFIT/32, Spectrum Software Associates) on a 3-decade logarithmic time base with additional compression (ϫ30) and averaged to 44 wavelength points (8 nm spaced within the range from 450 to 800 nm). This increased the signal-to-noise ratio and reduced the computational burden with negligible loss of spectrophotometric data and optimized detection of early changes.

RESULTS
Reactivity of PecA-It had been shown before that incubation of PecA with PCB ( Fig. 1) leads to an autocatalytic covalent attachment and formation of a fluorescent product absorbing ϳ645 nm (23). This reaction is inhibited by Triton X-100 (TX) and cannot be restored by the addition of Mn 2ϩ . Chromophore attachment is greatly accelerated by the addition of PecE, and further by Mn 2ϩ , which in the presence of PecE can even partially compensate for the TX inhibition ( Fig. 2 and Table 1). Under optimum conditions, the PecE-catalyzed reaction is complete in Ͻ30 min, compared with several hours for the autocatalytic process (Table 1). Moreover, there are distinctly different products formed depending on the presence or absence of PecE. The spontaneous addition product, which we designate P645 due to its absorption maximum at 645 nm, is relatively weak and comparable to that in the near-UV (Q vis/uv Ϸ 1, Table 2), it also possesses low fluorescence yield (supplemental Fig. S1). The product, P641, formed in the presence of PecE has blue-shifted absorption ( max ϭ 641 nm, Fig. 2 and Table 1). The absorbance intensity of P641 is double (Q vis/uv Ϸ 2, Table 2), that of P645, and it also has a much more intense fluorescence (supplemental Fig. S1); its properties, therefore, are approaching those of native PC ␣-subunits (24,25). The chromophores of the two species are nonetheless identical, when judged from denaturation experiments. Addition of acidic urea (8 M and pH 1.5) leads to products with identical spectra ( max ϭ 662 nm, Fig. 2B) that are also identical to that of denatured CPC, but blue-shifted compared with the free chromophore, PCB, that contains an additional double bond. Such properties are characteristic for thioether-bound PCB (Fig. 1); P645 and P641 therefore only differ in the non-covalent interactions between protein and chromophore. Apparently, the chromophorylated protein is misfolded in the P645 state, at least in the region of attachment to the chromophore. This is strikingly supported by the fact, that the low absorbance, low fluorescence P645 is transformed, by incubation with PecE, to the high absorbance, high fluorescence P641 (Fig. 2C). As a sub-stoichiometric amount of PecE is sufficient to induce this change, we accept this as the first direct evidence for a previously suspected (5) chaperone-like activity for PecE, by which the metastable 645 nm product is (allowed to) conformationally rearrange to the more stable 641 nm product. We observed that the reaction is independent of ATP or GTP (21) and is therefore different from that of chaperones (26), and so we use the term "chaperone-like." The action also proceeds with the post-translationally modified chromoprotein; that is, with the chromophore already bound to the apoprotein. In this respect, it resembles, e.g. calnexin, which acts on glycosylated proteins (27).
Protein Complex Formation-Long-lived interactions between PecA and PecE, and the influence of Mn 2ϩ and PCB on complex formation among proteins and chromophore, were investigated by gel filtration and subsequent SDS-PAGE. Samples were routinely incubated for 2.5 h. They were then concentrated by Ni 2ϩ -affinity chromatography, dialyzed, concentrated again by centrifugal filter devices, and subjected to gel filtration (see "Experimental Procedures" for details). Individual fractions were then analyzed by SDS-PAGE. In the absence of PCB, the gel-filtration experiments showed no formation of PecA⅐PecE complexes that are stable enough to survive workup and chromatography. There are only homo-dimers and homo-trimers formed from PecE, whereas PecA is monomeric under the conditions applied (Fig.  3, A and B). There is no significant influence of Mn 2ϩ on this behavior (not shown).
The situation changes when PCB (30 M) is present in otherwise identical conditions. Four major species can now be discerned in the chromatogram of the incubation mixture (Fig.  3C). Most chromophore absorption co-purifies with the last (and most intense) protein fraction, which subsequent SDS-PAGE proves to possess only PecA. This band shows a strong Zn 2ϩ -induced fluorescence (Fig. 3D) and is assigned, by its absorption, to the covalent PCB⅐PecA adduct, P641. All earlier eluting bands contain PecA and PecE in an approximately equimolar ratio, as judged from Coomassie staining. Based on their retention times, the major aggregates are assigned to heterodimers (PecA⅐PecE) 1 , heterohexamers (PecA⅐PecE) 3 , and high molecular complexes close to the exclusion volume that fits best to (PecA⅐PecE) 9 . All these bands contain PCB (absorption at 641 nm). Obviously, the presence of PCB promotes the interaction of the apoprotein, PecA, with the lyase ␣-subunit,

TABLE 1 Influence of cofactors on formation of PCB⅐PecA adduct
Concentrations: PecA ϭ 10 M, PecE ϭ 10 M, PCB ϭ 10 M, TX ϭ 0.1%, and Mn 2ϩ ϭ 1.5 mM. For calculation of the relative yields, the spectrum at time zero is subtracted from the spectrum obtained after 3 h of incubation, and the integral of this band is normalized to that obtained from the reaction of PCB with PecA in the presence of PecE (compare supplemental Fig. S4).  PecE. The amount of PCB present is reduced with increasing size of the complexes, indicating that it even promotes complex formation if only few of the PecA molecules in the complex contain PCB. The aggregation is analytically fortunate, because it attests to the interaction between the lyase subunit and the substrates, PCB and PecA. Catalytically, it seems unreasonable for the product to remain associated with the enzyme. Importantly, however, P641 is not the end-product of the holo-lyase because, in the presence of the PecF subunit, the chromophore is isomerized to phycoviolobilin and the properly chromophorylated PVB-PecA ( ϭ ␣-PEC) is then released from the lyase. Rapid Kinetics of PecE-catalyzed PCB⅐PecA Formation-First evidence for interaction of PecA with PecE was obtained from experiments, in which PecA was preincubated with PecE prior to adding PCB. Increasing preincubation periods result in the higher fluorescence signals of the PecA-PCB-adduct immediately after adding PCB (supplemental Fig. S5); at the end of the reaction there was no difference in the yield of the PecA-PCB-adduct. This indicated that PecA and PecE, in less than a minute, formed a complex that is competent for rapid PCB addition. Therefore, the PecE-catalyzed attachment of PCB to PecA was studied by stopped-flow kinetics with diode-array detection using two different mixing modes. The results are summarized in Fig. 4. In the first set of experiments (panels A and B), three syringes were employed to mix solutions of PecA and Mn 2ϩ , of PecE and Mn 2ϩ , and of PCB, to a final 1:1:1 ratio of chromophore and the two proteins (30 M each), and a final Mn 2ϩ concentration of 1.5 mM. These conditions are referred to as type I. In the second set of experiments, a solution containing PecA, PecE, and Mn 2ϩ was mixed with a solution containing PCB (panels C and D) to the same concentrations as before, and this scheme is referred to as type II.  In both cases, there is early development of a red shoulder (ϳ670 nm), followed by an increase around 641 nm. However, the kinetics and the details of the absorption changes differ; which becomes more obvious in the difference spectra (panels B and D). In the type I experiments, where PecA and PecE come into contact only during rapid mixing, the red shoulder appears as a prominent, red-shifted difference band, which at periods Ͼ60 s is replaced by the difference band ( max ϭ 638 nm), which is characteristic of the product, P641. The type II reaction, in which PecA and PecE are already in contact before mixing, shows far more rapid formation of P641. Here, the redshifted component shows at very early times, together with a blue-shifted component.

Components in reaction
The data of both reactions were analyzed using a multiexponential, irreversible linear reaction model. Single value decompositions of the data from the type II reactions yielded five significant temporal eigenvectors, indicating that four kinetic processes occur within the observed time window (5.5 ms to 150 s). These data sets were fitted to a linear four-step model. Starting with rate constants for the slow phase estimated from  5 s), forming a P605 species named according to its blue-shifted maximum. The last intermediate, P638 ( ϭ 14 s), which is not obvious from visual inspection of the data shown in Fig. 5 (C and D), absorbs maximally near the final product. Like all previous intermediates, however, it still has an extinction coefficient of only ϳ20% larger than that of free PCB. The large absorption increase is therefore confined to the final, slow step ( Ϸ 20 min) related to formation of the final product, P641.
Global analyses of type I reactions, where PecA and PecE come together only during rapid mixing, converged more slowly and less reproducibly, but resulted consistently in only four significant components. The spectral signatures of the initial component (PCB), the final component (P641), and of the two intermediates are very similar to the respective ones of the type II reaction, but the intermediate, P605, is completely missing. The rate constants also differ considerably from those of the type II reaction; in particular, the formation of P650 is two orders of magnitude slower. Formation of P638 is only decreased by a factor of 2. No accurate estimate was possible for the rate constant of the slowest process: it was fixed at a value ( 3 ϭ 33 min) that results in the correct extinction coefficient of P641. No rapidly formed intermediates were observed when PCB was incubated individually with PecA or PecE (not shown).
Addition of Phycoerythrobilin-PecE is also capable of catalyzing the addition of phycoerythrobilin (Fig. 1) to PecA forming a red-shifted intermediate, but the reaction is so much slower that it could be investigated by conventional spectroscopy (Fig. 6). These slower kinetics enabled us to test whether the chromophore is already covalently bound in the red-shifted intermediate. Samples taken early during the reaction were subjected to denaturation by treatment with acidic urea (pH 1.5, 8 M), which uncouples chromophoreprotein interactions (28,29); this resulted in spectra ( max ϭ 586 nm) that were identical to those of unreacted free PEB. Because the conjugation system would be shortened by cysteine addition, this indicates that the PEB is non-covalently adsorbed to the apoprotein binding pocket but not yet covalently attached (Fig. 6B). Only later after the formation of the red-shifted intermediate is the chromophore covalently bound, as shown by the blue-shifted absorption after denaturation treatment.

Protein-Protein Interactions
The PecA Protein-PecA alone is competent for PCB addition (Fig. 2D), which indicates that it has an enzymatic capacity to covalently add free PCB to its Cys-84 residue. This process requires no cofactors and results in a covalent bond, as shown by SDS-PAGE with subsequent Zn 2ϩ treatment. This autocatalytic PCB addition to form P645 is a slow process taking over 5 h for completion. Stopped-flow monitoring of PecA incubated with PCB gave no evidence for any interaction of the two within the monitored time scale of 100 s after mixing (not shown). These two findings suggest that PecA alone has an (autocatalytic) lyase activity but reacts only slowly with free PCB, without kinetically discernible intermediates. In addition, the product has lower absorption and fluorescence than biliproteins like ␣-CPC containing the PCB-chromophore in the native state (24,25); the chromophore is, therefore, bound in a sub-optimal state.
The PecE Protein-The role of PecE is to assist the autocatalytic function of PecA, which is further enhanced by the presence of Mn 2ϩ . Rather than an intrinsic enzymatic activity of PecE, this suggests a chaperone-like role that involves activation of PecA, most likely brought about by conformational changes. The strongest evidence for this function comes from the capacity of PecE to catalyze transformation of the low absorbance, low fluorescence P645 to the high absorbance, high fluorescence PCB-adduct, P641 ( Fig. 2C and supplemental Fig.  S1). We again emphasize that this role differs from that of native chaperones, because it does not require ATP or other high energy cofactors. Apparently, thermal energy is sufficient to overcome the transition energy of the chromoprotein between P645 and P641.
In principle, the chaperone-like action could be an interaction between PecE and PCB by which the latter delivered in the correct conformation to PecA; a (mechanistically different, see below) preparatory action is also discussed for the "chaperone," CcmE in cytochrome c formation (30). Several lines of evidence, however, suggest that prior formation of complexes between PecA and PecE accelerates this action. Such evidence comes from comparing the courses of type I and type II reactions of PCB adduct formation, which involve separate storage (type I), or preincubation (type II) of PecA and PecE prior to initiating the reaction by adding PCB with rapid mixing. After co-incubation (type II), PCB⅐PecA formation is 10-fold faster than separate storage before mixing (type I). Thus, we conclude that PecA and PecE form a complex during co-incubation that can bind PCB the moment it becomes available (supplemental Fig. S5). Furthermore, the spectral changes derived for the first kinetic step in the course of a type I fitting procedure were absent during incubation of PCB with either PecA or PecE alone.
Preincubation of PecA and PecE affects both the first and later kinetic steps, indicating that the ternary PCB⅐PecA⅐PecE complex remains stable for at least part of the reaction. This conclusion was supported and extended by gel filtration experiments. No PecA⅐PecE complexes were seen in the presence or absence of Mn 2ϩ , whereas there was no PCB present in the reaction mixture; they do form, however, when PCB is present. No interaction energies can be derived from these experiments, but it can be stated that only weak complexes are formed by PecA and PecE alone that do not survive gel chromatography, but are stabilized sufficiently in the presence of PCB. Although Mn 2ϩ was included in all assays, we made no attempts to analyze it in the chromatographically separated protein complexes or in the final PCB adduct. It is intriguing that much of the PecA⅐PecE complex does not dissociate after PCB⅐PecA (ϭ P641) formation but is stabilized by the chromophore. P641 is not, however, the final product of the ␣-PEC⅐PVB lyase reaction, but rather it is a non-natural product that is formed only in the absence of PecF.
Although PecA without PCB is largely monomeric, PecE forms aggregates that may contribute to the difficulties in analyzing kinetic data of the type I reaction causing the slow convergence of the fits and the much larger scatter of the data, which may relate to the presence of different size aggregates (Fig. 3A) prior to mixing, thus resulting in heterogeneous kinetics that vary critically with aggregate composition.  that the detergent affects the protein, but the effect is most pronounced on the chromophore. Conformational changes of bilins in solution can also be caused by amphiphiles (32). Conformation is a factor known to influence the reactivity of chromophores (33), including the site selection in chromophorylation of PecB (17). Similarly, interaction of the detergent with the chromophore could modify its conformation to restrict its access to PecA (17).
Summing up so far, we have shown: firstly, that the ␣-subunit of the isomerizing ␣-PEC:phycoviobilin lyase, PecE, accelerates and rectifies PCB binding to PecA (without isomerization to PVB); secondly, that it forms a complex with the apoprotein, PecA, which is stabilized by PCB; and, thirdly, that it probably functions in a chaperone-like fashion the initially formed chromoprotein, and the chromophore. It should be noted, however, that more involved reactions, including transient covalent binding to PecE, cannot currently be excluded.

Assignment of Intermediates
General Considerations-Of the five and four species obtained by global analysis (of type I and II reactions, respectively), on the basis of a linear model, the first four and three, respectively, show shifts of the absorption maxima but have very similar and relatively low extinction coefficients. However, very similar absorbing components of global fits can be artifacts of computation, caused by poor signal resolution and/or deficiencies in the proposed model. The reproducibility over many experiments, however, supports the existence of a series of intermediates with very similar extinction coefficients, which, in bilins, mainly depends on differences in molecular geometry (29,33,34). Free PCB, which has a relatively weak long-wavelength absorption (⑀ Ϸ 17 ϫ 10 3 M Ϫ1 cm Ϫ1 , Table 2), predominantly assumes a cyclic-helical conformation, whereas more extended conformations of PCB result in much more intense absorptions (29,(35)(36)(37)(38)(39). Thus, the relatively weak absorbencies of P650, P605, and P638 suggest that they all retain the cyclic conformation similar to that of free PCB and that the major conformational change occurs only during the last step. Nonetheless, the small differences of the first four species indicate distinct changes on the molecular level, caused by interactions with the proteins.
The P650 Intermediate-Both the type I and II reactions show an early red-shifted intermediate. Two reasons for such red shifts are known. The first is an increased length of the conjugation system. Conjugation of ring D is most important in rationalizing the extreme red shift of the ZZE-PCB (or -P⌽B) chromophore in phytochrome, when compared with the blue shift of the PVB chromophore following the ZZZ3 ZZE conversion in ␣-PEC (31). The results with PEB, however, do not support the involvement of ring D in the formation of the redshifted intermediate: in PEB, Ring D is no longer conjugated to rings A-C but, nonetheless, a red-shifted intermediate is formed during the addition reaction to PecA (Fig. 6A).
A second reason for a red shift is protonation of the chromophore. A conserved feature in the crystal structures of biliproteins, including PEC (38), is the aspartate (Asp-87) carboxyl group, near the pyrrole nitrogens of bilin rings B and C, that protonates the chromophore in the native structures (40 -43).
Thus, protonation of the non-covalently bound chromophore could account for the red-shifted absorption of P650. Protonation of free PCB, however, results in a red shift of ϳ47 nm with doubling of oscillator strength (36), whereas the red shift in P650 amounts only to ϳ25 nm with only a slight hyperchromism (ϳ20%). These lesser effects may be related to electrostatic interactions rather than a full protonation of the chromophore; possibly, arginine Arg-86 modulates the acidity of the neighboring Asp-87. Alternatively, protonation could be accompanied by conformational changes that can also greatly affect the absorption of bilins (29,33,44).
Judged from the experiments with PEB as substrate, the chromophore of the early, red-shifted intermediate is not yet covalently bound to the protein. Covalent attachment of the chromophore shortens the conjugation system which, in the absence of substantial conformational changes, would result in a blue shift. Non-covalent binding of the chromophore in P650 is also supported by the subsequent blue shift in both type I and type II reactions of PCB.
Three lines of evidence suggest that PecA and PecE are both involved in the formation of P650: (i) neither PecA nor PecE alone show red-shifted species when incubated with PCB; (ii) formation of P650 is two orders of magnitude faster when PecA and PecE are preincubated before the addition of PCB; and (iii) PCB so strongly stabilized the complex between PecA and PecE that it survived gel-chromatography.
Therefore, we interpret the first observed kinetic step as the non-covalent binding of PCB to a PecA⅐PecE complex, by which the cyclic-helical conformation is largely retained, and the chromophore is either protonated, or the central pyrrole nitrogens come close to a negatively charged residue. The rapid covalent binding in the subsequent step suggests that the chromophore is located very close to the binding site in PecA.
The P605 Intermediate-This blue-shifted intermediate is only obvious in the type II reaction (see below). However, based on the similar spectra of the next intermediate, P638, it is possible that this intermediate is also formed in the type I reaction, but never accumulates enough to be seen as a transient. Formation of the covalent bond is expected to cause a blue shift, because addition to cysteine results in the loss of the ⌬3,3 1 double bond. Surprisingly, the absorption changes and rate constant seen here during the second kinetic step of the type II reaction are reminiscent of those found in the rapid kinetics of PecE⅐PCB incubation in the presence of Mn 2ϩ (supplemental Fig. S3). This is remarkable, because no such blue shift is seen with PecE and PCB alone, even during prolonged incubation. However, weak, reversible binding of PCB to PecE has recently been proposed by Zhao et al. (18); therefore, it cannot be excluded that in P605 (part of) the chromophore is bound to PecE, rather than to PecA.
Formation of the thioether linkage may not be the only process displayed in the spectrum of P605, because the observed blue shift from 650 to 605 nm is larger than expected for loss of just one double bond. Shifts can also be induced by conformational changes that twist the double bonds, because delocalization of electrons is optimal only when the involved orbitals overlap maximally. Torsion of the bilin scaffold in the course of its interaction with protein components is a likely scenario. In all x-ray structures, the chromophores are non-planar, including the conjugated parts of the molecule (35,37,39,(45)(46)(47)(48)(49)(50). Although the exact location of the distortions is probably only reliable at a resolution Ͻ2 Å (46,50,51), pronounced deviations from planarity are obvious in all structures. Therefore, we assign the second step in the type II reaction to the covalent bond formation, associated with a conformational change brought about by the interaction of PCB with PecA and PecE.
A blue shift, after formation of a red-shifted intermediate, has also been observed in phytochrome chromophore attachment, where it has been ascribed to formation of P r (52). The absorbance ratio Q vis/uv of the red and blue bands, however, is relatively low compared with that of native P r. It is therefore not clear if this is already the end-product of the reaction or still an intermediate in which the covalent bond has already been formed, but the chromophore has not yet attained its native conformation.
The P638 Intermediate-The identity of P638 remains unclear. The oscillator strength, and the wavelengths of the absorption peaks of the spectrum are reminiscent of free PCB and of PCB⅐PecA. The low extinction coefficient of P638 suggests that it still maintains a cyclic-helical conformation. Due to potential computational artifacts posed by components with similar spectra in the reaction model, 4 we are presently unable to definitely identify this intermediate.
The Final Product, P641-The last kinetic step is the formation of PCB⅐PecA, which involves a large increase in oscillator strength while there is only little change in the position of the absorption maximum. The final, non-natural product has all characteristics of a covalently bound, protonated chromophore in an extended conformation. Because the chromophore in the preceding intermediate, P638, is protonated and covalently bound, but still in a cyclic-helical conformation, the last step must involve the conformational change of the chromophore from the cyclic, porphyrin-like geometry (typical for free bilins) to the extended geometry that is characteristic of the bilin chromophores in their native binding sites.
Influence of PecA⅐PecE Complex Formation on Kinetics-Coincubation of PecA and PecE in the presence of Mn 2ϩ before adding PCB results in a very different reaction sequence. In particular, the reaction is faster and an additional intermediate is indicated by global analysis. This indicates that preincubation leads to an interaction between PecA and PecE, such that the pre-formed complexes react more quickly with small, rapidly diffusing PCB than under conditions where PecA and PecE are kept separately before mixing, and have to interact with each other before the reacting with PCB.
In particular, the first reaction step, when the two proteins first come into contact during rapid mixing, is retarded by almost two orders of magnitude. So the same intermediate, P650, is formed but more slowly. Because no P650 intermediate is seen when incubating PCB with PecE alone, and because the autocatalytic reaction with PecA is much slower, it appears that the PecA⅐PecE complex alone provides the proper environment for P650 formation. Only if such complexes are already present before mixing is P650 formation rapid enough so that the subsequent covalent binding step (formation of P605) is kinetically discernible. In the kinetics of the type I reaction, the formation of P650 is 10-fold slower (4.7 s) than that of P605 from P650 in the type II reaction. Assuming similar kinetics for P605 formation in the type I and type II reactions, only negligible accumulations of P650 are expected at all times. This is not the case, however, rather P605 is missing in the fits of the type I reaction. This indicates that the binding step is slowed down even more than P650 formation in the type I reaction.
Comparison with Other Systems-Phycobiliproteins are one of three classes of proteins that are post-translationally modified by covalent attachment of tetrapyrrole chromophores to cysteines: the other two are the phytochromes and the c-type cytochromes. Binding is autocatalytic in the phytochromes where it involves the cooperation of two domains that are ϳ250 amino acid residues distant (10,11,(52)(53)(54)56). Either of the two can host the binding site, the one near amino acid 250 in the "classical" phytochromes binding PCB or phytochromobilin, the other near the N terminus in the bacterial phytochromes binding biliverdin. In both cases, the other domain is required for chromophore attachment (54,56,57). To our knowledge, only one report has focused on the mechanisms of this binding reaction (52), and it has some similarity to the chromophore binding reported here: in particular, the formation of a redshifted non-covalent intermediate, followed by a blue-shifted intermediate in which the covalent bond has already been formed. There are similar chromophores and similar chromophore conformations in the bound state and also distant homologies of the binding sites of the classical phytochromes with those of phycobiliproteins. However, a common lyase motif for phytochromes and phycobiliproteins has so far not been detected.
In the c-type cytochromes, the apoprotein is also principally capable of attaching the chromophore autocatalytically (58). However, this requires that both the apoprotein and the heme are pre-reduced. Three systems, some of considerable complexity, have evolved for the in vivo process of maturation that involves not only maintenance of the required redox states of the substrates but is further complicated by membrane transport (59 -62). System III involves the CchL-type heme-lyases to doubly link the heme chromophore to the apo-cytochromes (63)(64)(65)(66). The other systems involve several other proteins, including heme-transporters ("chaperones"), and, only recently, has the lyase function been assigned to a pair of gene products, CcsA and CcsB (30). Again, no obvious homologies are so far apparent for either the phytochrome lyase domains or the phycobiliproteins lyases, even though the apoproteins of cytochromes are phylogenetically and structurally related to those of the biliproteins (35). Further, the heme-chromophore of the cytochromes is the biosynthetic precursor of the bilin chromophores of the biliproteins (5,67). The most complex system I involves several Ccm proteins for the maturation process of cytochrome c. One of these proteins, CcmE, has a chaperone-like role and can bind heme covalently and non-covalently to present it in an appropriate form for covalent binding to the apo-cytochrome (68).
Conceptually, the lyases of both the phycobiliproteins and the cytochromes then share a chaperone-like function by which the chromophore and the protein are brought to appropriate states, whereas the final covalent binding is autocatalytic. These states are different: in the cytochromes, the redox states of both the chromophore and the apoprotein have to be controlled. In the biliproteins, however, it is probably the conformation of the flexible open-chain tetrapyrrole (17), which mainly requires controlling. Phylogenetically, it remains unclear to what extent the post-translational binding of the different tetrapyrroles to cysteines is related in phycobiliproteins, phytochromes and c-type cytochromes. Phycobiliprotein lyases are members of a large and varied group of repeat proteins that are involved in protein-protein interactions but share little homology and are of diverse function (69 -73). The results reported liken the catalytic function of PecE to a chaperone-like role for the bilin chromophore, and, therefore, add further to the multitude of functions for these proteins beyond protein-protein interactions.