Phytochrome Assembly THE STRUCTURE AND BIOLOGICAL ACTIVITY OF 2(R),3(E)-PHYTOCHROMOBILIN DERIVED FROM PHYCOBILIPROTEINS*

The unicellular rhodophyte, Porphyridium cruen- turn, and the filamentous cyanobacterium, Calothrix sp. PCC 7601, contain phycobiliproteins that have co- valently bound phycobilin chromophores. Overnight incubation of solvent-extracted cells at 40 OC with methanol liberates free phycobilins that are derived from the protein-bound bilins by methanolytic cleav- age of the thioether linkages between bilin and apoprotein. Two of the free bilins were identified as 3(E)- phycocyanobilin and 3( E)-phycoerythrobilin by comparative spectrophotometry and high pressure liquid chromatography. Methanolysis also yields a third bilin free acid whose absorption and ‘H NMR spectra sup- port the assignment of the 3(E)-phytochromobilin structure. This novel bilin is the major pigment isolated from cells that are pre-extracted with acetone-contain- ing solvents. Since phytochrome- or phytochromobilin-containing proteins are not present in either organism, the 3(E)-phytochromobilin must arise by oxidation of phycobilin chromophores. This pigment is not obtained by similar treatment of a cyanobacterium

The unicellular rhodophyte, Porphyridium cruenturn, and the filamentous cyanobacterium, Calothrix sp. PCC 7601, contain phycobiliproteins that have covalently bound phycobilin chromophores. Overnight incubation of solvent-extracted cells at 40 O C with methanol liberates free phycobilins that are derived from the protein-bound bilins by methanolytic cleavage of the thioether linkages between bilin and apoprotein. Two of the free bilins were identified as 3(E)phycocyanobilin and 3( E)-phycoerythrobilin by comparative spectrophotometry and high pressure liquid chromatography. Methanolysis also yields a third bilin free acid whose absorption and 'H NMR spectra support the assignment of the 3(E)-phytochromobilin structure. This novel bilin is the major pigment isolated from cells that are pre-extracted with acetone-containing solvents. Since phytochrome-or phytochromobilincontaining proteins are not present in either organism, the 3(E)-phytochromobilin must arise by oxidation of phycobilin chromophores. This pigment is not obtained by similar treatment of a cyanobacterium and a rhodophyte that lack phycoerythrin. Therefore, 3(E)-phytochromobilin appears to be derived from phycoerythrobilin-containing proteins. Comparative CD spectroscopy of 3(E)-phytochromobilin and 3(E)-phycocyanobilin suggests that the two bilins share the R stereochemistry at the 2-position in the reduced pyrrole ring. Incubation of 2(R),3(E)-phytochromobilin with recombinant oat apophytochrome yields a covalent bilin adduct that is photoactive and spectrally indistinguishable from native oat phytochrome isolated from etiolated seedlings. These results establish that the phycobiliprotein-derived 2(R),3(E)-phytochromobilin is a biologically active phytochrome chromophore precursor.
The phytochrome chromophore bears a close structural resemblance to the chromophores of phycocyanin and phycoerythrin, the phycobiliprotein light-harvesting pigments of rhodophyte and cryptophyte algae and cyanobacteria Lagarias et al., 1979;Lagarias and Rapoport, 1980;Crespi et al., 1967;Cole et al., 1967;Chapman et al., 1967;Crespi and Katz, 1969). All of these proteins contain linear tetrapyrrole prosthetic groups ( Fig. 1) that, in their * This work was supported by United States Department of Agriculture Competitive Research Grants 88-37130-3382 and 91-01454 (to S. I. B.) and 89-01162 (to J. C. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom correspondence should be addressed.
functional state, are covalently linked to specific cysteine residues of the proteins. In addition to having similar structures, these pigments share a common biosynthetic pathway and are derived from protoheme via biliverdin IXa (Beale and Weinstein, 1991). Cleavage of the thioether bilin-protein linkages by heating phycobiliproteins with methanol or strong acids liberates free bilins that have been characterized by spectroscopy and total synthesis Crespi et al., 1967;Crespi and Katz, 1969;Gossauer and Hirsch, 1974;Gossauer and Weller, 1978;Schram and Kroes, 1971). Owing to the low abundance of phytochrome in plant tissue and to the chemical reactivity of the released bilin chromophore, the isolation of phytochromobilin has been considerably more difficult. Although both methanolysis and HBr-trifluoroacetic acid treatment liberate bilin pigments from phytochrome, only trace amounts of phytochromobilin have been obtained (Siegelman et al., 1966;Kroes, 1970;Rudiger et al., 1980). The major pigment isolated from HBr-trifluoroacetic acid treatment of phytochrome was a methanol adduct of phytochromobilin . Cleaved bilin free acids have proven to be useful reagents for the study of phycobiliprotein chromophore biosynthesis as well as the biosynthesis of the free phytochrome chromophore and its assembly with the phytochrome apoprotein. For example, the intermediacy of 3(Z)-phycoerythrobilin and 3(Z)-phycocyanobilin in the enzymatic conversion of biliverdin IXa to 3(E)-phycocyanobilin in Cyanidium caldarium has been established by using the methanolytically released free bilins as standards and enzyme substrates (Beale and Cornejo, 1991b). Because phytochromobilin has not been readily available, the intermediacy of this pigment in phytochrome biosynthesis has been inferred from the ability of phycocyanobilin to substitute for the natural chromophore precursor both in uiuo  and i n vitro Lagarias and Lagarias, 1989;Wahleithner et al., 1991;Deforce et al., 1991). The latter studies have established that the formation of photochemically active holophytochrome proceeds spontaneously in the absence of added enzymes or cofactors. Using holophytochrome assembly as an assay, Terry and Lagarias (1991) have recently presented evidence for phytochromobilin synthase, the enzyme that catalyzes conversion of biliverdin IXa to the phytochrome chromophore precursor, in higher plant plastid preparations. By contrast to the phycocyanobilin-apophytochrome adduct, which is spectrally blue-shifted, the newly synthesized holophytochrome species is spectrophotometrically indistinguishable from native phytochrome isolated from plant tissue. On the basis of this evidence, it was concluded that the plastid enzyme produces the natural phytochrome chromophore pre-FIG. 1. Bilin structures. Structures of biliverdin IXa, the 3(E) isomers of free phycocyanobilin and phycoerythrobilin, the 3(E) and 3(Z) isomers of free phytochromobilin, and the phytochromobilin undecapeptide from oat phytochrome discussed in the text. The conventional carbon numbering system used in the text is shown for biliverdin IXa.
cursor and that this precursor is phytochromobilin (Terry and Lagarias, 1991). Although these earlier studies support the hypothesis that phytochromobilin is the immediate precursor of the phytochrome chromophore, experimental verification requires a source of authentic pigment. Both the 3-E and 3-2 isomers of phytochromobilin have been chemically synthesized as dimethyl esters . Use of the synthetic bilins in biosynthetic studies is problematic because they are racemic, whereas the natural bilins are likely to be chiral. Also, saponification of the esters is expected to afford low yields of the bilin free acids because of isomerization of the ethylidine-bearing reduced pyrrole ring to an ethylpyrrole .
During ongoing studies of phycobilin biosynthesis in algae, a new pigment appeared under certain extraction conditions. This pigment has now been identified as 2(R),3(E)-phytochromobilin. Although the occurrence of this pigment in the algal extracts is probably artifactual and has no biosynthetic significance, it has provided a source of sufficient phytochromobilin for testing as a substrate for phytochrome reconstitution experiments. Results reported here show that this pigment can spontaneously assemble with recombinant oat apophytochrome to form photoactive holophytochrome that contains a covalently bound chromophore and is spectrally indistinguishable from native oat phytochrome.

EXPERIMENTAL PROCEDURES
Growth of Cells-Axenic liquid suspension cultures of Porphyrid i u n cruentum UTEX 637 and C. caldarium strain CPD were grown in liquid medium as described previously (Beale and Chen, 1983;Cornejo, 1991a, 1991b), with illumination (32 pE m-'s-l) provided by equal numbers of cool-white and red fluorescent tubes. Axenic liquid suspension cultures of Calothrix sp. PCC 7601 (previously named Fremyella diplosiphon) and Synechocystis sp. PCC 6803 were grown at 23 "C in BG-11 medium (Rippka et al., 1979) supplemented with 15 mM TES' (pH 8.2) and 50 mM dextrose, under the ' The abbreviations used are: TES, A"tris(hydroxymethy1)methyl-2-aminoethanesulfonic acid HPLC, high pressure liquid chromatography; SDS, sodium dodecyl sulfate; PAGE, polyacrylamide gel electrophoresis. to a density of 8-10 g fresh weight per 500 ml of medium, were harvested by centrifugation. Free pigments and other soluble components were removed by several different solvent extraction treatments (described more fully under "Results"). The solvent-extracted cells were suspended in methanol (40 m1/5 g of cells, fresh weight) containing 1 mg/ml HgC12 and incubated in the dark for 24 h at 40 "C to release bilins from the phycobiliproteins. In some cases (where indicated), HgC1, was omitted from the methanolysis solution. After methanolysis, the cells were centrifugally separated from the bilincontaining methanol solution and washed with methanol (2 m1/5 g of cells, fresh weight). The methanol solutions were combined, and HgCI, (when present) was removed by the addition of P-mercaptoethanol (1 pl/ml) and centrifugal sedimentation of the white precipitate.
'H NMR Spectroscopy-Room temperature 'H NMR spectroscopy of bilins dissolved in pyridine-D5 (Sigma Catalogue No. P8141; minimum 99.96% atom D) were obtained using a Bruker AM400-WB spectrometer. Tetramethylsilane was added as an internal standard. Second order spectra of interacting proton signals were analyzed with the aid of the Bruker PANIC NMR simulation and iteration program.
Absorption and CD Spectroscopy-All absorption and CD measurements of bilin pigments were performed at room temperature. Absorption spectra were obtained with Cary 219 (Varian Associates, Inc., Sunnyvale, CA), Aviv 14DS (Aviv Associates, Inc., Lakewood, NJ), and HP 8450A (Hewlett-Packard Co., Palo Alto, CA) UV-visible spectrophotometers. A JASCO 5600 (JASCO Inc., Easton, MD) spectropolarimeter was used for obtaining CD spectra. A scanning rate of 50 nm/min, band pass of 1 nm, and resolution of 1 nm were the parameters used for CD spectroscopy.
Holophytochrome Reconstitution-All reconstitution experiments were performed under green safelight (Litts et al., 1983). The oat phytochrome (phyA3) construct pMphyA3 was expressed in the yeast Saccharomyces cerevisiue strain 29A and apophytochrome-containing extracts were prepared as described previously (Wahleithner et al., 1991). Incubations were initiated by adding 5 pl of bilin stock solution in dimethyl sulfoxide to 600 p1 of incubation medium containing 25 mM Tris-HC1 (pH 7.8), 2 mM EDTA, 2 mM phenylmethanesulfonyl fluoride, 1 mM dithiothreitol, 25% (v/v) ethylene glycol, and 9 pg of apophytochrome. The protein concentration in the final incubation mixture was 2.5 mg/ml. Final bilin concentrations were 4 p~ and 1.7 p~ for phycocyanobilin and phytochromobilin, respectively. After incubation for 30 min at 28 "C, reaction mixtures were clarified by ultracentrifugation for 15 min at 200,000 X g and assayed for holophytochrome as described below.

P. cruentum Phycobiliprotein Methanolysis Products-In
the initial experiments, methods for extracting free pigments and other soluble materials from the cells before methanolysis were adapted from those used previously for C. caldarium (Beale and Cornejo, 1991b). For C. caldarium, it was necessary to pre-extract the cells with dimethyl sulfoxide to disrupt the cell membranes. After the dimethyl sulfoxide treatment, methanol was used to remove dimethyl sulfoxide and complete the cell extraction before methanolysis. In early attempts to apply this procedure to P. cruentum cells, it was found that the cells did not sediment in dimethyl sulfoxide, but that sedimentation could be achieved if the dimethyl sulfoxide solution was diluted 8-fold with acetone. The procedure that was adopted was to extract the cells with dimethyl sulfoxide, dilute the solution 8-fold with acetone, and then extract four more times with dimethyl sulfoxide/acetone (1:8, v/v) and finally three times with methanol.
After extraction by the above procedure, the cells were methanolyzed. The bilins obtained were purified by solvent partitioning and DEAE-Sepharose chromatography and separated by reverse-phase HPLC. Three prominent peaks were apparent in the elution profile (Fig. 2). The middle-and lateeluting pigments were identified as 3(E)-phycocyanobilin and 3( E)-phycoerythrobilin, respectively, by comparative HPLC and spectrophotometry (Table I). The early-eluting peak had an elution position and absorption maxima that differed from those of all bilins previously characterized in methanolysis extracts and enzyme incubations ( Table I). Identification of the early-eluting pigment is described below.
Absorption Spectroscopy of the Early-Eluting Pigment-The visible absorption spectrum of the HPLC-purified pigment was recorded in methanol/36% (w/v) aqueous HC1 (482, v/ v). The spectrum had maxima at 386 and 700 nm, and the peak height ratio was 1.67 ( Fig. 3a; Table I). These values are very close to those reported for chemically synthesized racemic 3( E)-phytochromobilin dimethyl ester dissolved in methanol/HCl (49:1, v/v) . The synthetic pigment has absorption maxima at 386 and 708 nm and a peak height ratio of 1.70 . Differences between the two spectra are attributed to the different esterification states of the two bilins and/or slight differences in the solvent composition. On the basis of spectral similarity, the early-eluting pigment was provisionally identified as phytochromobilin. It could not be determined from the absorption spectrum whether the pigment more closely resembles 3(E)-or 3(Z)-phytochromobilin.
'H NMR Spectroscopy of the Early Eluting Pigment-HPLC-purified bilin was dried in uacm and dissolved in anhydrous pyridine-D6 containing a trace of tetramethylsilane. The 400-MHz 'H NMR spectrum was recorded at room temperature. Because of the low concentration of bilin (approximately 0.25 mM) and concerns about sample stability at room temperature during the NMR spectroscopy, the sensitivity was maximized by using 10-ps (90") pulses and 2.26-s data acquisition times. Three thousand scans, collected over a period of approximately 2 h, were signal-averaged. The 0.0-7.5-ppm region of the 'H NMR spectrum is shown in Fig. 4, and the properties of the relevant NMR signals are listed in Table 11. Also tabulated are the 'H NMR signals of racemic 3(E)-phycocyanobilin dimethyl ester (Gossauer and Hirsch, 1974) and the phytochromobilin undecapeptide from oat phytochrome (Lagarias and Rapoport, 1980) in pyridine-D5 and racemic 3(E)-phytochromobilin dimethyl ester in CDC13 . Most of the proton signals correspond very closely to those of equivalent protons of the other pigments. It was shown previously that the chemical shifts of several biliverdin IXa proton signals are significantly influenced by the solvent (Beale and Cornejo, 1991b). This is particularly true for the downfield signals assigned to the 18vinyl protons and the 10-and 15-methylene protons. This solvent effect accounts for the differences between the chemical shifts for these protons measured in pyridine-D6 and the values reported for 3( E)-phytochromobilin dimethyl ester recorded in CDCls . All of the 'H NMR signals assigned to protons on the free bilin compared well with those that also occur on the peptidebound phytochrome chromophore (Lagarias and Rapoport, 1980), except for the proton at the 2-position and those attributed to the 3-ethylidine group. These differences are consistent with the replacement of the ethylidine group of the free pigment by a thioethyl moiety in the phytochromobilin undecapeptide. Analysis of the signals assigned to the 18vinyl group was complicated by the closeness of the chemical shifts of two of the protons, leading to a second order spectrum with distorted peak heights and anomalous shifts in peak Reverse-phase HPLC was on a 4.6-mm-diameter X 25-cm-long column of octadecylsilane-coated 5-fim-diameter spherical silica particles (Altex). The elution solvent was ethanol/acetone/water/acetic acid (4834:17:1, v/v) flowing at 4.0 ml/min. Visible absorption spectra of the HPLC-purified Digments were recorded in methanol/36% (w/v) aqueous HC1 (49:1, v/v).  ' Killilea et al. (1980).
positions (Jackman and Sternhell, 1969). A very similar complication was encountered previously with the spectrum of phytochromobilin undecapeptide recorded in pyridine-D6 (Lagarias and Rapoport, 1980). Resolution of the ABX spincoupled system and accurate assignment of the signals was facilitated by the Bruker PANIC NMR simulation and iteration program. This program generates a simulated spectrum based on input of the observed signals and optimizes the spectral match by adjustment of arbitrary chemical shifts and coupling constants. the recorded spectrum very closely (Fig. 5 ) . The calculated chemical shifts and couplings are very close to those reported for the 18-vinyl group protons of biliverdin IXa (Beale and Cornejo, 1991b) and the phytochrome chromopeptide recorded in the same solvent (Table 11).
From the chemical shifts of the signals assigned to the 3ethylidine protons, it was determined that the ethylidine group has the E configuration. Specifically, the 3'-proton signal at 1.72 ppm lies within the range of 1.63-1.91 ppm reported for several 3(E)-bilins in various solvents, and out of the range of 2.08-2.18 ppm for 3(Z)-bilins (Table 111). Similarly, the 3l-proton signal at 6.35 ppm is within the range of 6.23-6.68 ppm reported for several 3(E)-bilins and out of the range of 5.77-5.90 ppm for 3(Z)-bilins.
The occurrence of small signals close to those assigned to the 5-and 15-methine protons (Figs. 4 and 5 ) suggests that other bilins were generated while the pigment was standing at room temperature in pyridine during the NMR data acquisition period. After the 'H NMR spectrum was recorded, the solvent was evaporated in vacuo and the residue was analyzed for degradation by reverse-phase HPLC. Some degradation products were detected; the total A370 of these products was approximately 5% of that of the major component.
CD Spectroscopy of Methanolysis-derived 3(E)-Phytochrornobilin-The CD spectra of methanolysis-derived 3(E)-phytochromobilin and 3(E)-phycocyanobilin were recorded in 36% (w/v) HCl/methanol (2:48, v/v). In this solvent, the bilins are fully protonated and free of helical conformations that can complicate interpretation of the spectra. Under these conditions, CD signals only reflect information about chiral carbons. Each of these bilins contains only one asymmetric carbon, at the 2-position (Fig. 1). The CD spectra of the two bilins are strikingly similar (Fig. 3b). The small differences in the positions of the peak wavelengths are expected from the similar differences in the absorption maxima (Fig. 3a).
Since the 2-carbon of natural phycocyanobilin is known to have the R configuration (Brockmann and Knobloch, 1973), the methanolysis-derived phytochromobilin also appears to have a 2R configuration.

FIG. 4. 'H NMR spectrum of 3(E)-
phytochromobilin obtained by methanolysis of solvent-extracted P. cruenturn cells. The spectrum of the bilin dissolved in anhydrous pyridine-Db at approximately 0.25 mM was recorded at room temperature. The signal at 0 ppm is that of the tetramethylsilane internal standard. Residual water in the sample and solvent produced the large signal at 5 ppm. The signal at 7.2 ppm was produced by residual pyridine-H g in the solvent. Signals below 1.5 ppm are attributed to hydrocarbon contaminants. Bilin resonances were assigned as described under "Results" and are listed in Table 11.   Gossauer and Hirsch (1974).
In conclusion, the early-eluting pigment obtained from P. bly experiments were performed using recombinant oat apocruentum was identified as 2(R),3(E)-phytochromobilin by phytochrome prepared from yeast. For comparative purposes, comparative spectrophotometry, CD spectroscopy, and 'H a parallel experiment was performed using authentic 3 ( E ) -NMR spectroscopy. phycocyanobilin, derived by methanolysis of C-phycocyanin Holophytochrome Assembly Using P. cruentum-derived from Synechococcus sp. PCC 7002 cells. Covalent attachment Phytochromobilin-To determine whether the 2 ( R ) , 3 ( E )of bilins to apophytochrome was first determined using a zinc phytochromobilin is biologically active, phytochrome assem-blot assay. In this assay, bilin-linked polypeptides are visual- to the 18-vinyl protons match the synthetic spectrum, the 5-and 15-methylene singlets are a t 5.91 and 6.21 ppm, respectively, and the 3l-ethylidine quartet of doublets is a t 6.35 ppm.

Compound
Chemical shift Solvent Ref.

3(E)-Phytochromobilin
Pyridine-D6 1.72 6.35 Our data 3(E)-Phytochromobilin DME CDC4 -' Weller and Gossauer (1980). 'Chemical shifts, which were originally reported relative to that of hexamethyldisiloxane, have been adjusted by adding 0.06 ppm to the published values to facilitate comparison with the other tabulated values, which are relative to that of tetramethylsilane.
ized as orange fluorescent bands on blots of proteins resolved by SDS-PAGE (Berkelman and Lagarias, 1986;Wahleithner et al., 1991). A biliprotein of the same molecular size as native oat phytochrome (124 kDa) was produced by incubation of recombinant oat apophytochrome with either of the bilins (Fig. 6). No fluorescent 124-kDa protein was detected in the lane containing apophytochrome only, where no bilin was added.
To determine whether the bilin-apoprotein adducts were photoreversible, difference photoassay measurements were obtained. Both bilin-apophytochrome adducts were photoactive and the yield of photoactive holophytochrome for both bilins was similar (Fig. 7). Based on spectrophotometric and Apophytochrome-containing extracts from yeast cultures expressing the plasmid pMphyA3 were incubated with 3(E)-phytochromobilin ( l a n e 3 ) , 3(E)-phycocyanobilin (lane 4 ) , or no bilin as a control (lane 2 ) , then resolved by 7.5% T SDS-PAGE (1.5-mm thickness), transblotted to a poly(viny1idine difluoride) membrane, and assayed for zinc-dependent fluorescence as described under "Experimental Procedures.'' Lune 1 contains partially purified oat phytochrome extracted from etiolated seedlings as described previously . Sample loads were 200 ng of phytochrome protein per lane. For photography, the blot was exposed for 75 s using a red cutoff filter as described previously (Wahleithner et al., 1991). quantitative immunoblot analyses (not shown), reconstitution yields in excess of 90% were estimated for both bilins in this experiment. The difference maximum and minimum of the phytochromobilin adduct at 666 and 730 nm, respectively, are nearly identical with the values of 668 and 732 nm reported for native oat phytochrome, within the resolution limits of the instruments used (Lagarias et aL, 1987). By contrast, the difference maximum and minimum of the phycocyanobilin adduct were both blue-shifted as has been reported previously Wahleithner et al., 1991). The

A A , , , l A A m i ,
ratio of the phytochromobilin adduct was 1.09, which is also in good agreement with the value of 1.14 reported previously for native oat phytochrome (Lagarias et al., 1987). All of the above spectrophotometric measurements were obtained with the HP 8450A diode-array spectrophotometer, which has a fixed 2 nm resolution in the visible region. For greater wavelength accuracy, an Aviv 14DS spectrophotome- ter, set at 0.5 nm resolution, was used to obtain a difference spectrum of the phytochromobilin adduct. In this case, values of 668 and 729 nm were determined for the difference maximum and minimum, respectively (data not shown). Taken together with the spectroscopic measurements which establish the structure of the early-eluting pigment from P. cruentum to be 2(R),3(E)-phytochromobilin, these reconstitution experiments demonstrate that this pigment is a biologically active phytochrome chromophore precursor.

Yield of 2(R),3(E)-Phytochromobilin from P. cruentum-
The relative concentrations of the three major bilins obtained from P. cruentum were approximated as follows. The relative absorption peak areas for phytochromobilin, phycocyanobilin, and phycoerythrobilin, in the HPLC elution profile monitored at 370 nm (Fig. 2), were 3.0, 1.8, and 1.0, respectively. The wavelengths of the absorption maxima of the pigments in HPLC elution solvent (not shown) are similar to values reported for the pigments in neutral organic solvents such as methanol and CHC13. The molar extinction coefficient of 57,500 M" cm" a t 372 nm was reported for 3(E)-phytochromobilin dimethyl ester in CHC13 . For 3(E)-phycocyanobilin dimethyl ester in neutral methanol, the molar extinction coefficient of 41,700 M" cm" a t 368 nm was reported (Gossauer and Hirsch, 1974). The molar extinction coefficient of 6,600 M" cm" a t 370 nm for 3(E)-phycoerythrobilin in CHCl, was obtained from the published spectrum of aplysioviolin (Rudiger, 1967a), which is 3(E)phycoerythrobilin-12-monomethyl ester (Rudiger, 1967b). From these extinction coefficients (used without correction for differences in wavelengths and solvents) and the relative peak areas of the HPLC elution profile, the molar ratio of phytochromobilin/phycocyanobilin/phycoerythrobilin in the methanolysis solution was calculated to be 1.0:0.8:2.9. Phytochromobilin was 27% of the total. Under the extraction and methanolysis conditions described above, each gram (fresh weight) of P. cruentum cells yielded approximately 10-15 nmol of purified 2(R),3(E)-phytochromobilin.
Requirements for Obtaining Phytochromobilin-Since P. cruentum cells do not contain phytochromobilin or phyto-chromobilin-bearing phycobiliproteins, this bilin must have arisen by chemical modification of phycobiliprotein chromophores. To determine which specific step in the cell extraction and methanolysis procedures were required for the production of phytochromobilin, these steps were varied systematically. During the course of the studies discussed above, it was found that P. cruentum cells, unlike C. caldarium, do not require dimethyl sulfoxide for the initial cell permeabilization; acetone alone is sufficient. Cells were extracted several times with acetone, until the extraction supernatants were colorless, and then methanolyzed. These cells yielded pigment mixtures in which phytochromobilin was a major component (data not shown). Therefore, exposure of the cells to dimethyl sulfoxide is not required to obtain phytochromobilin. It was not possible to directly examine the dependence on acetone for obtaining phytochromobilin from P. cruentum cells, because acetone is required either for permeabilizing the cells or for sedimenting cells that are permeabilized with dimethyl sulfoxide. When cells that had been permeabilized with acetone were washed extensively with methanol to remove residual acetone before methanolysis, they aggregated into clumps that were resistant to methanolysis.
Cells of the filamentous cyanobacterium, Culothrix sp. PCC 7601, unlike P. cruentum and C. caldarium, can be permeabilized with methanol. After extraction of the cells several times with methanol, until the extraction supernatants were colorless, methanolysis yielded phycocyanobilin and phycoerythrobilin, but no phytochromobilin. To test the effect of acetone, methanol-extracted cells were washed with acetone, and the acetone was then removed by centrifugation. In this case, methanolysis yielded all three bilins (data not shown). All three bilins were also produced when methanol-extracted cells were methanolyzed in the presence of controlled concentrations of acetone. The yield of phytochromobilin was approximately equal when the methanolysis solution contained acetone within the range of 0.5-10% (v/v) (data not shown). To test whether the effect of acetone was due to the presence of a contaminant in commercial reagent grade acetone, a methanolysis was done with methanol containing freshly redistilled acetone. This methanolysis yielded about the same amount of phytochromobilin as those containing commercial acetone (data not shown).
T o determine whether the effect of acetone is exerted before or after the bilins are released from the phycobiliproteins by methanolysis, Calothrix cells were extracted with methanol and then divided into two portions for methanolysis. One portion was methanolyzed for 12 h at 40 "C in methanol/ HgClz and other in methanol/HgC12 containing 2% (v/v) acetone. After separation of the solvent, the cell residues from both methanolysis reactions were washed extensively with methanol. The cells from the acetone-free first methanolysis were methanolyzed for 16 h at 40 "C with methanol/HgC12 containing 2% (v/v) acetone, and the cells from the acetonecontaining first methanolysis were methanolyzed for 16 h with methanol/HgCI2 in the absence of acetone. Finally, portions of the methanol solution from the first acetone-free methanolysis were held at 40 "C for 16 h in the presence and absence of 2% (v/v) acetone. Phytochromobilin was detected in the products of both the first and second methanolysis incubations of the cells that were first methanolyzed in the presence of acetone (data not shown). Phytochromobilin was detected only in the products of the second methanolysis (in the presence of acetone) of the cells that were first methanolyzed in the absence of acetone. Addition of 2% (v/v) acetone to the methanolysis products of cells that were methanolyzed in the absence of acetone did not yield phytochromobilin upon further incubation for 16 h at methanolysis temperature.
To test whether phytochromobilin is derived from phycocyanin or phycoerythrin, similar experiments were performed with Synechocystis sp. PCC 6803, a unicellular cyanobacterium whose cells contain only phycocyanobilin-bearing phycobiliproteins. Synechocystis cells that were extracted and methanolyzed identically with the conditions that produced phytochromobilin from Calothrix cells yielded only phycocyanobilin; neither phycoerythrobilin nor phytochromobilin was detected as methanolysis products. Similar experiments with C. caldarium, which also lacks phycoerythrin, produced no phytochromobilin recovery, regardless of the presence of acetone in the extraction and methanolysis solutions (data not shown).
From these results, it appears that acetone acts on proteinbound phycoerythrobilin to facilitate oxidation at the 15,16position to form protein-bound phytochromobilin, which is subsequently cleaved from the protein by methanol to yield free phytochromobilin. Formation of phytochromobilin is also not specifically dependent on HgC12, but the presence of HgClz in the methanolysis medium increases the yield of all free bilins (data not shown).

DISCUSSION
These studies have established that methanolysis of solvent-extracted P. cruenturn and Calothrix cells releases phytochromobilin in addition to the expected phycobilins, phycocyanobilin and phycoerythrobilin. Comparative spectroscopic measurements have confirmed that the released phytochromobilin has the 3E configuration of the ethylidine double bond and an R stereochemistry at the %carbon. Evidence for the correct assembly of 2(R),3(E)-phytochromobilin with recombinant oat apophytochrome to form native holophytochrome includes formation of a covalent bond between bilin and apoprotein and the similarity of the difference spectrum of the reconstituted phytochrome with that of native oat phytochrome. Previous reconstitution experiments using apophytochrome and phycocyanobilin have also yielded a covalently bonded adduct that underwent photoreversible spectral changes, but the difference spectrum differed from that of native phytochrome, reflecting the substitution of phycocyanobilin for the natural phytochrome chromophore precursor Lagarias and Lagarias, 1989;Wahleithner et al., 1991;Deforce et al., 1991). The generation of the native photoreversible difference spectrum in the reconstitutions with P(R),S(E)-phytochromobilin strongly suggests that this bilin is the natural in uivo phytochrome chromophore precursor.
Previous predictions of the absolute configuration of the phytochrome chromophore have been based on analogy with those of phycobilins. The existence of a strong CD signal for methanolysis-derived 3(E)-phytochromobilin indicates that the 2-carbon of the bilin was not racemized during the oxidation of protein-bound phycoerythrobilin and methanolysis. The CD spectrum shows that the configuration at the 2position of 3(E)-phytochromobilin is the same as that of 3(E)-phycocyanobilin from methanolytically cleaved C-phycocyanin. The absolute configurations at C-2 of phycobiliprotein-derived 3(E)-phycocyanobilin and 3(E)-phycoerythrobilin have previously been shown to be R (Brockmann and Knobloch, 1973;Gossauer and Weller, 1978). The magnitude of coupling constants between the C-2 and (2-3 protons in the 'H NMR spectrum of oat phytochromobilin undecapeptide showed that the C-2 and C-3 substituents are trans, indicating that the absolute configuration is either 2R,3R or 2S,3S (Lagarias and Rapoport, 1980). The coupling constants among the 2-, 3-, and 3l-protons of phytochromobilin undecapeptide are also very similar to those of the C-phycocyanin PI chromophore (Lagarias et al., 1979). The latter has recently been shown by x-ray crystallography to have the 2R,3R,3'R configuration (Schirmer et al., 1987;Duerring et al., 1991), confirming earlier predictions based on chemical degradation results (Klein and Rudiger, 1978). It was also predicted from chemical degradation that the configuration of the phytochrome chromophore is 2R,3R,3'R, but the 2S,3S,3'S configuration could not be excluded (Klein et al., 1977). From our direct determination, by comparative CD spectroscopy, that the configuration at C-2 of the 3(E)-phytochromobilin used for reconstitution is R, together with the foregoing conclusions, it follows that the configuration at C-3 is also R and that the reconstituted phytochrome chromophore is 2R,3R93'R. In view of the high reconstitution efficiency and the spectral indistinguishability of the reconstituted adduct from natural phytochrome, it is highly unlikely that their chromophores differ.
The spontaneous assembly of phytochromobilin and phycocyanobilin with apophytochrome to form spectrally active covalent adducts is in distinct contrast to the inability of phycocyanobilin to correctly assemble with apophycocyanin. With apophycocyanin, although covalent adducts were formed, they were not equivalent to native holophycocyanin but differed with respect to spectral properties, peptide chromatographic behavior, and structure of the bound chromophores (Arciero et al., 1988a(Arciero et al., , 1988b. These results suggest that phycobiliprotein assembly requires additional enzymes and/or cofactors.
Phytochromobilin is probably not naturally present in P.
cruentum or Calothrix cells. Its occurrence is unlikely to have been missed in earlier detailed studies of the phycobiliprotein chromophore content of these organisms, especially if it is present at the 27% mole fractional content that was estimated for the P. cruentum methanolysis products. In addition, neither of these organisms has been reported to contain phytochrome or have phytochrome responses. Moreover, the pigment was obtained only when cells were exposed to acetone before or during methanolysis. Therefore, it seems probable that phytochromobilin is formed by conversion of a phycobiliprotein chromophore during methanolysis. Our studies point to phycoerythrin as the source of this pigment. First, phytochromobilin was obtained only from phycoerythrin-containing cells. Second, the transformation required for conversion of the phycoerythrin chromophore to phytochromobilin involves the oxidation of a methylene bridge (Fig. l), a reaction that is known to occur readily in bilins. For example, bilirubin (10,ll-dihydrobiliverdin) is readily oxidized to biliverdin, both chemically (McDonagh, 1979;McDonagh and Palma, 1980) and enzymatically (Murao and Tanaka, 1982). By contrast, the required chemical transformation necessary for converting phycocyanin chromophore to phytochromobilin entails the more difficult oxidation of an ethyl group to a vinyl group.
Together with the observation that acetone treatment prior to methanolysis is required for phytochromobilin production, these results support the hypothesis that phytochromobilin is derived from the bilin prosthetic group(s) of phycoerythrin. In P. cruentum, phycoerythrin contains a third phycobilin chromophore, phycourobilin, in addition to phycocyanobilin and phycoerythrobilin (Glazer and Hixson, 1977), which raises the possibility that phytochromobilin is derived from the protein-linked phycourobilin chromophore. This possibility is excluded because Calothrix phycoerythrin contains no phycourobilin (Beguin et al., 1985;Glazer, 1988) but yields phytochromobilin upon methanolysis.
It is therefore concluded that phytochromobilin is derived from the phycoerythrobilin chromophore of phycoerythrin. At this time, neither the identity of the proposed oxidant nor the role of acetone in facilitating the oxidation is known. I t must be stressed that methanolysis was performed with whole cells. Although soluble materials were extracted with organic solvents before methanolysis, materials that could serve as the oxidant may have remained in the cells. The role of HgClz in increasing the yield of bilins may be to react with sulfhydryl groups on proteins and thereby prevent them from condensing with the liberated bilins (Manitto and Monti, 1979). Another role for HgC12 could be to promote denaturation and unfolding of the phycobiliproteins, thereby increasing the accessibility of the bilin-cysteine thioether bonds to attack by methanol. A third possible role for HgClZ may be to inactivate chemical and enzymatic bilin degradation during methanolysis.
The availability of relatively large quantities of phytochromobilin now enables us to address a number of important questions related to phytochrome chromophore biosynthesis and holophytochrome assembly. For example, HPLC studies have already revealed that (E)-phytochromobilin is enzymatically produced from biliverdin IXa by isolated cucumber cotyledon plastids.* Comparative kinetic analyses of the assembly of apophytochrome with phytochromobilin and other ligatable bilins, such as phycocyanobilin, should also help define specific structural features of the bilin moiety that contribute to the assembly process. In conjunction with experiments to determine the roles of specific amino acid residues of the apoprotein in the assembly of holophytochrome, bilin specificity studies of this type provide a powerful approach to the molecular dissection of phytochrome assembly in vitro.