Conserved Chloroplast Open-reading Frame ycf54 Is Required for Activity of the Magnesium Protoporphyrin Monomethylester Oxidative Cyclase in Synechocystis PCC 6803

Background: The cyclase step in chlorophyll biosynthesis remains uncharacterized. Results: Ycf54 forms a complex with other oxidative cyclase components, and a ycf54 mutant accumulates the cyclase substrate. Conclusion: Ycf54 is essential for cyclase function. Significance: Identification of all of the components of chlorophyll biosynthesis is a step closer. The cyclase step in chlorophyll (Chl) biosynthesis has not been characterized biochemically, although there are some plausible candidates for cyclase subunits. Two of these, Sll1214 and Sll1874 from the cyanobacterium Synechocystis 6803, were FLAG-tagged in vivo and used as bait in separate pulldown experiments. Mass spectrometry identified Ycf54 as an interaction partner in each case, and this interaction was confirmed by a reciprocal pulldown using FLAG-tagged Ycf54 as bait. Inactivation of the ycf54 gene (slr1780) in Synechocystis 6803 resulted in a strain that exhibited significantly reduced Chl levels. A detailed analysis of Chl precursors in the ycf54 mutant revealed accumulation of very high levels of Mg-protoporphyrin IX methyl ester and only traces of protochlorophyllide, the product of the cyclase, were detected. Western blotting demonstrated that levels of the cyclase component Sll1214 and the Chl biosynthesis enzymes Mg-protoporphyrin IX methyltransferase and protochlorophyllide reductase are significantly impaired in the ycf54 mutant. Ycf54 is, therefore, essential for the activity and stability of the oxidative cyclase. We discuss a possible role of Ycf54 as an auxiliary factor essential for the assembly of a cyclase complex or even a large multienzyme catalytic center.

The cyclase step in chlorophyll (Chl) biosynthesis has not been characterized biochemically, although there are some plausible candidates for cyclase subunits. Two of these, Sll1214 and Sll1874 from the cyanobacterium Synechocystis 6803, were FLAG-tagged in vivo and used as bait in separate pulldown experiments. Mass spectrometry identified Ycf54 as an interaction partner in each case, and this interaction was confirmed by a reciprocal pulldown using FLAG-tagged Ycf54 as bait. Inactivation of the ycf54 gene (slr1780) in Synechocystis 6803 resulted in a strain that exhibited significantly reduced Chl levels. A detailed analysis of Chl precursors in the ycf54 mutant revealed accumulation of very high levels of Mg-protoporphyrin IX methyl ester and only traces of protochlorophyllide, the product of the cyclase, were detected. Western blotting demonstrated that levels of the cyclase component Sll1214 and the Chl biosynthesis enzymes Mg-protoporphyrin IX methyltransferase and protochlorophyllide reductase are significantly impaired in the ycf54 mutant. Ycf54 is, therefore, essential for the activity and stability of the oxidative cyclase. We discuss a possible role of Ycf54 as an auxiliary factor essential for the assembly of a cyclase complex or even a large multienzyme catalytic center. Chlorophylls (Chls) 5 are key cofactors of the photosynthetic apparatus, and their biosynthesis results in the formation of the most abundant natural pigments on Earth. Chl is structurally distinguished from other tetrapyrroles such as hemes, bilins, or corrins, by a centrally chelated magnesium atom and by the presence of an isocyclic "fifth" ring. In plants, algae, and cyanobacteria the synthesis of Chl, heme, and bilins shares the same pathway up until protoporphyrin IX (P IX ), which is used as a substrate for two different chelatases. Whereas insertion of Fe 2ϩ results in heme, insertion of Mg 2ϩ by magnesium chelatase leads to Mg-protoporphyrin IX (MgP), the first intermediate on the Chl branch. MgP is converted to Mg-protoporphyrin IX methylester (MgPME) by Mg-protoporphyrin methyltransferase. The isocyclic ring is made in the following step by the oxidative MgPME cyclase (EC 1.14.13.81, hereafter MgPME cyclase), which results in the Chl a precursor protochlorophyllide (PChlide). Conversion of PChlide to chlorophyllide (Chlide) is catalyzed by PChlide oxidoreductase, and Chl a is finally completed by the addition of a polyisoprene tail to Chlide by the Chl-synthase (for review, see Ref. 1).
Although cyclase activity was assayed many years ago using cucumber chloroplasts (2)(3)(4), wheat etioplasts (5) and cell extracts from Chlamydomonas reinhardtii and Synechocystis PCC 6803 (6), the enzyme itself remains enigmatic. Formation of the isocyclic ring is proposed to be a complex enzymatic reaction that consists of three sequential steps: (a) hydroxylation of the ring C to methylpropionate, (b) oxidation to ketopropionate, (c) ligation of an activated methylene group to the ␥-meso carbon of the porphyrin ring (7,8); see supplemental Fig. S1 for a series of reactions proposed for the cyclase step. Biochemical characterization also provided evidence for the incorporation of atmospheric oxygen into synthetized PChlide in plants (9), which implies that the oxidative cyclase differs structurally and mechanistically from the anaerobic MgPME cyclase identified from analysis of bchE mutants of Rhodobacter capsulatus and Rhodobacter sphaeroides (10 -12). In the anaerobic cyclase the 13 1 -oxo group of bacteriochlorophyll is derived from water (13), whereas in aerobic phototrophs such as Roseobacter denitrificans this group arises from molecular oxygen (14).
Partial purification of oxidative cyclase from chloroplasts and cyanobacteria suggested that this enzyme requires several proteins (subunits) for activity (6,15). The only known candidate for a catalytic cyclase subunit was identified by mutational analysis of the photosynthetic bacterium Rubrivivax gelatinosus. Although this bacterium also has a BchE homolog functional under anaerobic conditions, inactivation of the acsF gene resulted in accumulation of MgPME under aerobic conditions. Thus, R. gelatinosus possesses two unrelated MgPME cyclases, one oxygen-dependent and one oxygen-independent (16,17). AcsF homologs have since been identified in many photosynthetic eukaryotes including C. reinhardtii (18), Arabidopsis thaliana (19), and barley (15). Two acsF-like genes, sll1214 and sll1874, have been identified in Synechocystis PCC 6803 (hereafter Synechocystis) (20); Sll1214 is essential for growth under aerobic conditions (20) and, to a certain extent, micro-oxic conditions (21), whereas Sll1874 was found to be essential for growth in micro-oxic conditions (20,21).
In this work we performed in vivo FLAG pulldown experiments using Synechocystis strains transformed with genes encoding either the FLAG-Sll1214 or FLAG-Sll1874 protein.
Using mass spectrometry, we identified a Ycf54-like protein (Ycf54) as a putative assembly factor or subunit of the MgPME cyclase that interacts with both Sll1214 and Sll1874. This interaction was confirmed by a "reciprocal" co-purification of FLAG-Ycf54 together with Sll1214. Inactivation of the ycf54 gene (slr1780) results in a significant reduction of cellular Chl and PChlide levels and the accumulation of MgPME. In addition this mutation strongly impairs accumulation of the Sll1214 cyclase component together with other enzymes in the Chl branch of tetrapyrrole pathway, particularly MgP methyltransferase and PChlide oxidoreductase. Thus, Ycf54 is essential for normal levels of MgPME cyclase activity and/or stability, suggesting that Ycf54 plays a critical role in assembly or function of the cyclase complex.

EXPERIMENTAL PROCEDURES
Growth Conditions-The Synechocystis ycf54 strain and the control wild-type (WT) strain were grown photomixotrophically in a rotary shaker under low light conditions (3 mol of photons m Ϫ2 s Ϫ1 ) at 30°C in liquid BG-11 medium (22) supplemented with 5 mM glucose.
Construction of FLAG-tagged Synechocystis Mutants-The plasmid pPD-FLAG was made to allow construction of Synechocystis strains expressing putative subunits of the cyclase with a 3ϫFLAG (hereafter FLAG) tag at the N terminus. pPD-FLAG contains the Synechocystis psbAII promoter, a sequence encoding the 3ϫFLAG tag and flanking sequences for homologous recombination that allows insertion of tagged constructs into the Synechocystis genome in place of the psbAII gene (see supplemental Fig. S2). The sll1214, sll1874, and ycf54 genes were each subcloned into pPD-FLAG, and the resulting plasmids were transformed into the Synechocystis WT. Transformants were selected on a BG-11 agar plate containing 10 g ml Ϫ1 neomycin and fully segregated by increasing the concentration of antibiotic to a final concentration of 80 g ml Ϫ1 .
Preparation of Solubilized Membrane Fraction and Anti-FLAG Pulldown Assay-Synechocystis cells expressing genes for FLAG-tagged proteins were grown photoautotrophically to an OD 750 of 0.5-0.7. Cells were pelleted, washed, and resuspended with buffer A (25 mM sodium phosphate, pH 7.4, 50 mM NaCl, 10 mM MgCl 2 , 10% glycerol, EDTA-free protease inhibitor; Roche Applied Science). Cells were mixed in equal proportions with glass beads broken in a Mini-Beadbeater-16 (Bio-Spec), and the soluble proteins and membranes were separated by centrifugation (65,000 ϫ g, 45 min). The membrane fraction was washed with, then resuspended in, buffer A and solubilized for 1 h at 4°C with 1.5% dodecyl-␤-maltoside (Glycon). Finally, insoluble contaminants were removed by centrifugation (65,000 ϫ g, 25 min).
FLAG-Sll1214, FLAG-Sll1874, and FLAG-Ycf54 complexes were purified from soluble and membrane fractions by batch binding to anti-FLAG-M2-agarose (Sigma) for 1 h at room temperature. To remove contaminants the anti-FLAG-resin was washed with 100 resin volumes of buffer A containing 0.04% dodecyl-␤-maltoside. The FLAG-tagged proteins were eluted with 2.5 resin volumes of buffer A containing 150 g/ml 3ϫFLAG peptide (Sigma) and 0.04% dodecyl-␤-maltoside.
Analysis by Nano-LC-MS/MS and Database Searching-The FLAG-eluted proteins were reduced in the presence of 100 mM triethylammonium bicarbonate, pH 8.5, 1% sodium dodecyl sulfate (SDS), and 50 mM dithiothreitol (all from Sigma) at 60°C for 30 min. The proteins were then S-alkylated by incubation with 10 mM iodoacetamide (Sigma) at room temperature in the dark for 30 min. After a 10-fold dilution with 50 mM triethylammonium bicarbonate, the proteins were digested with 4 g of trypsin (proteomics grade, Sigma) at 37°C for 16 h followed by drying in a vacuum centrifuge. The tryptic peptides were redissolved in 0.2 ml of 10 mM potassium phosphate, pH 3, in 25% acetonitrile (loading buffer) with 7 l of 0.5 M phosphoric acid added to acidify the samples and applied to homemade spin columns containing a 20-l bed volume of Poros 20 SP cation exchange medium (Applied Biosystems). The columns were washed with 2ϫ 0.1-ml loading buffer to remove SDS, and the tryptic peptides were eluted with 0.1 ml of loading buffer containing 0.5 M KCl. The samples were desalted using C 18 reverse-phase Spin-Tips (Protea Biosciences) according to the manufacturer's protocol.
Peak lists for database searching in the form of Mascot Generic Files were created with DataAnalysis 4.0 software (Bruker). The Mascot Generic Files were submitted for database searching using Mascot Daemon v. 2.1.3 running with Mascot Server v. 2.2.01 (Matrix Science, London, UK) against the Synechocystis complete proteome database (ExPASy). Search parameters were as follows: enzyme, trypsin; fixed modifications, carbamidomethyl (C); variable modifications, deamidation (NQ), oxidation (M); maximum missed cleavages, 1.
Inactivation of the ycf54 Gene-To prepare the Synechocystis ycf54 strain, chromosomal DNA was isolated from a non-segregated Synechocystis ycf54 mutant obtained as a generous gift from Prof. Teruo Ogawa (Nagoya University, Japan). The mutated allele of the ycf54 is interrupted by an erythromycin resistance cassette inserted at nucleotide 129 of the ycf54 gene. The chromosomal DNA was transformed into Synechocystis WT, and transformants were selected on BG-11 agar plates containing 10 g of ml Ϫ1 erythromycin and then segregated under low light (3 mol of photons m Ϫ2 s Ϫ1 ) on 5 mM glucose and increasing concentrations of erythromycin to a final concentration of 80 g ml Ϫ1 . The mutant is designated as ycf54 in cases where it is partially segregated and ycf54 Ϫ when it has been complemented with the FLAG-ycf54 construct and full segregation was achieved.
Quantification of Chlorophyll and Chlorophyll Precursors-To measure Chl content per cell, pigments were extracted from cell pellets (5 ml, OD 750 ϳ 0.4) with an excess of 100% methanol, and the Chl content was measured spectrophotometrically (23). To compare the contents of all other abundant cellular pigments, the same methanol extract was immediately injected onto an HPLC machine (Agilent-1200), and the pigments were separated using a reverse phase column (Nova-Pak C18, 4 m particle size, 3.9 ϫ 150 mm; Waters) with 30% methanol in 1 M ammonium acetate, pH 6.0, and 40% acetone in methanol as solvents A and B, respectively. Pigments were eluted with a linear gradient of solvent B (45-100% in 18 min) followed by 100% of B at a flow rate of 1 ml min Ϫ1 at 40°C. For quantitative determination of Chl precursors, 75 ml of culture at OD 750 ϭ 0.35-0.4 was filtered through a 4-m cellulose filter to remove precipitated pigments in the growth medium and harvested. Pigments were extracted with an excess of 80% methanol, 20% water, 0.2% NH 4 OH and measured as described previously (24), except in the present work the HPLC was equipped with two fluorescence detectors and a diode array detector. The first fluorescence detector was set to 440/670 nm (excitation/emission wavelengths) for 0 -11 min, 440/640 nm for 11-14 min, and 400/630 nm for 14 -25 min. The second fluorescence detector was set at 416/595 nm throughout the experiment.
Preparation of Antibodies-To prepare an antibody against the Ycf54 protein, the ycf54 gene was amplified from Synechocystis genomic DNA and cloned into pGEX-4-T1 (GE Healthcare). The GST-Ycf54 gene fusion was overexpressed in Escherichia coli BL21 cells followed by purification on GST affinity resin, removal of the GST tag by thrombin cleavage, and removal of thrombin through benzamidine chromatography (GE Healthcare). Purity was further increased using Mono Q ion exchange chromatography. 1 mg ml Ϫ1 Ycf54 protein was used to generate rabbit polyclonal anti-Ycf54 antibody (Bio-Serv). Antibodies against Synechocystis MgP methyltransferase, PChlide oxidoreductase, and geranylgeranyl reductase were prepared using essentially the same protocol. An antibody against the Synechocystis ferrochelatase was kindly provided by Prof. Annegret Wilde (Justus-Liebig University, Giessen, Germany). The Sll1214 antibody (anti-Chl27) was obtained from Agrisera (Sweden).
Assessment of Gene Expression by Northern Blot and RT-PCR-WT and mutant cells were grown photomixotrophically at low light and harvested at an OD 750 ϳ 0.4. For Northern blot total RNA was purified according to Pinto et al. (25). 5 g of purified RNA was than separated on 1.3% agarose formaldehyde gel and blotted onto Hybond-N ϩ membrane (GE Healthcare). A hybridization probe was generated by random prime labeling (Rediprime II labeling kit, GE Healthcare) with [␣-32 P]dCTP (Hartmann Analytic). The membrane was prehybridized for 60 min in 50% (v/v) deionized formamide, 7% SDS, 250 mM NaCl, and 120 mM sodium phosphate, pH 7.2, at 45°C and hybridized overnight with a labeled probe. The membrane was than rinsed in 2ϫ SSC, 1% SDS and washed in 2 subsequent 15-min steps at 48°C in 2ϫ SSC, 0.5% SDS and 0.1ϫ SSC, 0.1% SDS, respectively. Signals were visualized using phosphorimaging. For RT-PCR 4 ϫ 10 8 cells were disrupted by bead beating in the presence of RNAprotect (Qiagen). RNA was purified from the cell extract using the RNeasy kit (Qiagen) and RNase free DNase (Qiagen) according to the application manuals. cDNA was synthesized from 50 or 100 ng of RNA using SuperScript III Reverse Transcriptase (Invitrogen) and random primers (Invitrogen). RT-PCR was performed on cDNA using gene-specific primers, resolved on a 1% agarose gel, and visualized by ethidium bromide.

The Synechocystis Ycf54 Protein Forms a Complex with AcsF
Homologs Sll1214 and Sll1874-To identify interacting partners/subunits of MgPME cyclase, we prepared Synechocystis strains expressing the AcsF homologs Sll1214 and Sll1874 as 3ϫFLAG-tagged proteins under control of the constitutive psbAII promoter. Western blotting with an anti-FLAG antibody showed that both tagged proteins were located exclusively in the membrane fraction (not shown). For pulldown assays, cells of both tagged strains were fractionated, and membrane fractions were solubilized by 1.5% dodecyl-␤-maltoside and incubated with anti-FLAG affinity resin. After extensive washing with 100 column volumes of buffer, the FLAG-Sll1214 and FLAG-Sll1874 bait and prey proteins were eluted using the FLAG peptide. Fig. 1A shows SDS-electrophoresis analysis of elutions from FLAG pulldown assays; in both cases the FLAGtagged Sll1214 or FLAG-Sll1874 proteins were the most prominent bands purified. Eluted proteins were derivatized by S-carbamidomethylation and digested by trypsin in the presence of SDS. The tryptic peptide fragments were prepared for MS analysis by solid phase extraction using a combination of cation exchange and C18 reversed-phase media. The peptides were subsequently analyzed using nano-LC/MS-MS in conjunction with ultra high resolution time of flight mass spectrometry enabling high mass accuracy in both the MS and tandem MS spectra. Table 1 presents a list of the proteins identified using mass spectrometry analysis of the pulldown assays using either FLAG-Sll1214 or FLAG-Sll1874 as bait. Fig. 1, B and C, shows product ion (MS/MS) spectra of the tryptic peptide FLLEEEP-FEEVLK (m/z 811.4, charge state 2ϩ, mass 1620.8 Da) from the Ycf54-like protein from the FLAG-Sll1214 (Fig. 1B) and FLAG-Sll1874 (Fig. 1C) pulldown assays, respectively. These results show that using N-terminal FLAG-tagged Sll1214 and Sll1874 proteins as bait, each identifies Ycf54 as a potential interaction partner in vivo. Other proteins identified in both pulldown assays such as subunits of phycobilisomes and ATP synthase ( Table 1) were presented also in the control WT sample (not shown), so these are regarded as false positives.
To lend further support to the proposed role of Ycf54 in the cyclase step of Chl biosynthesis, a construct encoding a FLAG-Ycf54 protein was expressed in the Synechocystis WT using the same methodology as described for FLAG-Sll1214/Sll1874, and another pulldown experiment was performed. As shown in Fig.  2A, FLAG-Ycf54 was successfully purified under native conditions, and analysis by SDS-PAGE shows a prominent silverstained band. The FLAG-Ycf54 elution was then blotted and probed with anti-Chl27 antibody raised against the AcsF homologue from Arabidopsis. This antibody is known to be reactive to Sll1214, whereas reactivity to Sll1874 is minimal (20). Importantly, a clear signal from the Sll1214 cyclase component was detected (Fig. 2B), and subsequent MS analysis verified the presence of Sll1214 in the FLAG-Ycf54 eluate (not shown) but not in the control WT sample.
The ycf54 gene Is Required for Accumulation of Enzymes Catalyzing Several Steps of Chlorophyll Biosynthesis-To investigate the function of Ycf54, the cognate sll1780 gene was inactivated by insertion of an erythromycin cassette. PCR analysis of genomic DNA confirmed the near-complete segregation of the mutant allele (supplemental Fig. 3A), but full segregation could not be achieved despite the various growth conditions tested combined with high concentrations of erythromycin (not shown). Nevertheless, the level of the ycf54 transcript in the ycf54 strain was clearly lowered (Fig. 3A), and a Western blot probed with the anti-Ycf54 antibody showed a significantly reduced level of Ycf54 compared with the WT strain (Fig. 3C). This antibody also allowed us to localize Ycf54 to both the sol-uble and membrane fractions (Fig. 3C). A lowered amount of Ycf54 was accompanied by a significant decrease in the level of Sll1214 and also in the amounts of the enzymes that precede and follow the cyclase step, MgP methyltransferase, and PChlide oxidoreductase. The level of geranylgeranyl reductase was also reduced to some extent. Interestingly, the level of MgP methyltransferase in the soluble fraction is similar in WT and mutant strains despite very different amounts bound to membranes. The association of MgP methyltransferase with the membrane is apparently weaker in the ycf54 mutant or perhaps there is a specific degradation of this enzyme. In addition, we analyzed the level of ferrochelatase and found it comparable in the WT and the mutant, implying that the ycf54 mutation specifically affects enzymes in Chl branch of the tetrapyrrole pathway (Fig. 3C).
To confirm that the observed changes in enzyme levels are not due to a positional effect of the ycf54 mutation, the ycf54 mutant was transformed by the FLAG-ycf54 construct. Importantly, this complementation experiment made it possible to completely segregate the mutated ycf54 gene (supplemental Fig. 3B). Fig. 3C shows that levels of all Chl biosynthesis enzymes examined were restored to WT levels in the complemented FLAG-ycf54/ycf54 Ϫ strain. Thus, the reduction in the

Proteins identified by nano-LC-MS/MS and database searching in FLAG-Sll1214 and FLAG-Sll1874 pulldown assays
The tryptic peptides identified by database searching are shown with their neighboring residues in the sequence separated by periods. We further examined levels of transcripts for sll1214 in the ycf54 mutant, the WT, and in the complemented strain. Northern blotting shows that the level of the sll1214 transcript is a bit reduced in the ycf54 mutant, but no such decrease was observed in the complemented strain (Fig. 3B). A lowered level of the sll1214 transcript in the mutant was detected also by RT-PCR, and a somewhat reduced expression was also found for the por gene (Fig. 3A).

Inactivation of the ycf54 Gene Impairs the Cyclase
Step of Chlorophyll Biosynthesis-The phenotypic effects of inactivating the sll1780 gene were investigated in cells grown under low light (3 mol of photons m Ϫ2 s Ϫ1 ) to minimize any effects of photo-oxidative damage arising from disrupted tetrapyrrole metabolism. To maintain the ycf54 mutation, the growth medium was supplemented by 5 g ml Ϫ1 of erythromycin; this level of antibiotic has no effect on a control WT strain harboring a genomic insertion of the erythromycin cassette in place of the psbAII gene (not shown). The ycf54 cells obtained were markedly pale, suggesting a deficiency in photosynthetic pigments (see supplemental Fig. 3C for pictures of cultures of WT and ycf54 strains). The Chl content per cell was reduced by 70% in the ycf54 mutant relative to the WT (Fig. 4A). The complemented FLAG-ycf54/ycf54 Ϫ strain has a much improved Chl content reaching almost 80% of the WT level (Fig. 4A), which is in line with the improved accumulation of enzymes in this strain (Fig. 3C). The appearance of an absorbance peak at 420 nm hints at the accumulation of a "new"' pigment produced by the ycf54 mutation (Fig. 4A). To identify this component of the ycf54 cell spectra, methanol extracts were prepared from equal quantities of ycf54 and WT cells, then cellular pigments were separated and analyzed by HPLC. The ycf54 chromatogram contains a distinct peak at 11 min that is completely absent in the WT sample (Fig. 4B). The elution time and absorbance and fluorescence spectra of this pigment identified it as MgPME (supplemental Fig. 4). Indeed, the 416-nm absorbance maximum of MgPME in solvent is consistent with the observed absorbance of ycf54 cells at 420 nm.
The high content of MgPME in the ycf54 mutant suggests that the pathway is specifically blocked at the cyclase step. To examine the effects of the ycf54 mutation further, we quantified levels of Chl precursors and performed more detailed spectroscopic analyses. Methanol extracts from low light-grown cells were purified by phase partitioning, concentrated, and separated by HPLC. It should be noted that some P IX is released by ycf54 cells into the growth medium and that all such precipitated pigments were removed by filtration before cell extraction and HPLC analysis. Thus, the data in Fig. 5 reflect the intracellular accumulation of Chl biosynthetic intermediates. The fluorescence elution profiles for ycf54 extracts in Fig. 5A show enhanced levels of P IX (a 3.2-fold increase) and, as expected, a very high level of MgPME; this was quantified using the absorbance data in Fig. 5B as an ϳ100-fold increase. Use of the second fluorescence detector enabled quantification of MgP, which increased 4-fold in the mutant (results not shown). These large rises in Chl biosynthetic intermediates before the cyclase step in the pathway were accompanied by a dramatic decrease in the level of the product of the cyclase, PChlide, to 5% of the WT level. Interestingly, levels of Chlide are not altered significantly in the ycf54 mutant in comparison with the WT (Fig. 5A), which suggests that this pool of Chlide results from Chl recycling rather than de novo synthesis (see "Discussion"). In summary, the accumulation of Chl precursors shows that although the ycf54 mutation strongly decreases the amount of the MgP methyltransferase, there is no impairment of the pathway at this step, in contrast with the very low cyclase activity.
Further evidence for an essential role of Ycf54 in the cyclase step comes from inspection of the small elution peaks identified with an arrow in Fig. 5A, also identified as peaks at 7.6 and 8.0 min using detection of absorption (Fig. 5B). Fig. 6 shows a more detailed spectroscopic analysis of each elution component, neither of which corresponds to the product of the cyclase, PChlide, or of the cyclase substrate, MgPME (see the inset in Fig. 6). Instead, the Soret band at 432 nm, normalized to a value of 1.0 at this wavelength, is situated between that of MgPME at 416 nm and PChlide at 440 nm (see inset to Fig. 6A). Similar behavior is observed for the red-most bands, which are located at 614 nm for the 7.6-min component and 616-nm for the 8.0 min peak; the corresponding MgPME absorbance band is at 588 nm, and for PChlide the peak is at 630 nm. This pattern is also found in the fluorescence emission spectra in Fig. 6B, which clearly show the same progressive red shift, starting with MgPME at 595 nm, the two putative cyclase reaction intermediates at 620 and 628 nm, and ending with the cyclase product, FIGURE 2. Affinity purification of FLAG-Ycf54 and identification of Sll1214/Sll1874 by Western blot. A, FLAG-Ycf54 was purified from a Synechocystis cytoplasmic fraction (S) and a dodecyl-␤-maltoside-solubilized membrane fraction (M) using anti-FLAG-agarose and eluted with 3xFLAG peptide. Eluted proteins were resolved by SDS-PAGE and visualized with silver stain. B, Eluted proteins from the FLAG-Ycf54 pulldown assay were resolved by SDS-PAGE and transferred by Western blot to a nitrocellulose membrane. The amount of protein loaded for each sample corresponded to 1 ⁄10 of the total eluate. The membrane was then probed with anti-Chl27 (Agrisera) raised against the AcsF homolog in Arabidopsis. The antibody is reactive to Sll1214 but not to Sll1874 (16). A WT control consisting of a Synechocystis whole cell lysate probed with anti-Chl27 is included for comparison.

Ycf54 Is Required for Protochlorophyllide Synthesis
PChlide, at 642 nm. On the basis of these spectra, it appears that Synechocystis cells harboring the ycf54 mutation synthesize pigments with the absorption properties of possible cyclase reaction intermediates.

DISCUSSION
Based on available biochemical data and also with regard to the complexity of the enzymatic reaction required to create the fifth Chl ring (supplemental Fig. 1), the MgPME cyclase is expected to be active as a multisubunit complex. However, our knowledge of the individual components of this enzyme is very limited. In cucumber, barley, C. reinhardtii, and Synechocystis the MgPME cyclase was resolved into membrane and soluble components (6,15,26). In addition, the membrane component appears to be formed from at least two different proteins; the first one is a homolog of AcsF, and the second one is an as yet unknown protein deficient in the viridis-k mutant of barley (15). AcsF contains a putative diiron site and, thus, is viewed as a true catalytic subunit of MgPME cyclase (16).
Previous work had identified two genes in Synechocystis, sll1214 and sll1874, as acsF homologs that encode the membrane component of the MgPME cyclase (20,21). To identify their protein partners, we tagged the N termini of Sll1214 and Sll1874 in vivo with 3ϫFLAG and purified both proteins from detergent-solubilized membrane extracts. These experiments identified Ycf54 as a possible partner for Sll1214 and Sll1874. A reciprocal pulldown assay using FLAG-Ycf54 showed that Sll1214 is trapped as prey. The absence of Sll1874 in the FLAG-Ycf54 eluate is not so surprising bearing in mind the aerobic growth conditions used to prepare the culture of FLAG-ycf54 strain; Sll1874 is specifically expressed under micro-oxic conditions, and its level is probably minimal under atmospheric oxygen level (20,21). In contrast to the membrane-bound Sll1214/Sll1874, Ycf54 appears to be a rather hydrophilic protein (Fig. 3C); it is likely that the association of Ycf54 with membrane is provided via an interaction with AcsF homologs.
A partial inactivation of the ycf54 gene produced a strong phenotype, greatly impeding cell growth and resulting in accumulation of Chl biosynthetic intermediates for the steps preceding the cyclase step in Chl biosynthesis. Given the essential nature of the cyclase step in the Chl biosynthetic pathway, it is perhaps not surprising that, despite repeated efforts, full segregation of the ycf54 mutation could not be achieved. The importance of Ycf54 is underlined by the effects of the ycf54 mutation, which include impaired accumulation of several enzymes of Chl biosynthesis including the AcsF homolog Sll1214 and significantly reduced Chl levels. This decrease in Sll1214 protein to the observed extent (Fig. 3C) would itself explain the observed 70% decrease in Chl in the ycf54 mutant. The sll1214 Ϫ Synechocystis mutant possessing about 50% of WT Sll1214 levels displays similar decreases in Chl content (21). In contrast, some other enzymes in tetrapyrrole biosynthesis appear to be present in apparent excess. For example, Synechocystis mutants FIGURE 3. A, RT-PCR analysis of ycf54, sll1214, and por expression in the WT and ycf54 mutant was performed using specific primers; the ftsZ transcript was included as a control. B, expression of the sll1214 gene in the WT, ycf54 mutant, and the FLAG-ycf54/ycf54 Ϫ strain was analyzed using northern blotting. A DNA probe specific for sll1214 was used for hybridization; RNA transcripts were used as a loading control. C, shown are levels and localization of Ycf54, FLAG-Ycf54, and enzymes of Chl biosynthesis in the WT, ycf54 mutant, and the complemented FLAG-ycf54/ycf54 Ϫ strain grown mixotrophically under low light conditions. The same amounts of protein of soluble (Sol) and membrane (Mem) fractions were separated by SDS-PAGE and blotted onto nitrocellulose. The membrane was probed using specific antibodies, which are listed on the left-hand side of the figure. The mobility of the FLAG-Ycf54 protein (top row) differs substantially from that of Ycf54, and the band for this protein is not shown. In the second row, the FLAG-Ycf54 protein was detected by the anti-FLAG antibody (Sigma). The lower panel shows blotted proteins stained with Ponceau S as a loading control. The band for the Photosystem I core PsaA/B subunits is indicated. The abbreviations used are: MgPMT, Mg-protoporphyrin methyltransferase; POR, protochlorophyllide oxidoreductase; GGR, geranylgeranyl reductase; FeCH, ferrochelatase.
with 5-10% of WT levels of ferrochelatase are not deficient in heme and have no discernable phenotype (27,24), whereas Synechocystis strains with reduced levels of Chl synthase have no growth or pigmentation defects. 6 This seems be also the case of MgP methyltransferase, where strongly reduced levels still produce plenty of MgPME in the ycf54 mutant (Figs. 3C and 4B). Although it appears that the level of Sll1214 is important for synthesis of PChlide and Ycf54 is essential for the accumulation of Sll1214, we cannot exclude the possibility that Ycf54 is directly involved in the catalytic function of the cyclase. Regarding the phenotype of the ycf54 mutant together with our evidence showing that Ycf54 physically interacts with AcsF homologs, it is not unreasonable to consider Ycf54 as a new cyclase subunit. Given the presence of Ycf54 in the soluble cellular fraction, we do not expect that this protein is related to the viridis-k mutation that affects a membrane component (15). The elusive soluble component of MgPME cyclase was reported to have a mass above 30 KDa (26), which is apparently at odds with the 12.5 KDa mass of Ycf54. In relation to a catalytic role, the structure of Ycf54 from Thermosynechococcus elongatus (PDB code 3HZE, see supplemental Fig. S5) shows that this protein does not contain any apparent redox or electron transfer sites that would support its role in catalysis. Taking into account the very low accumulation of Sll1214 in the ycf54 mutant, it looks more probable that Ycf54 plays a critical role in AcsF synthesis/maturation or in the process of cyclase assembly rather than being a subunit of the cyclase enzyme complex.
There is, however, another possibility that Ycf54 facilitates formation of a catalytic complex between cyclase and preceding or after enzymes, and this interaction is required for cyclase 6 R. Sobotka, unpublished data. B, shown is pigment content in the WT and ycf54 strains grown mixotrophically under low light conditions. The same amount of cell material was extracted by methanol and separated by HPLC. The ycf54 strain has a low Chl content but contains very high levels of the Chl precursor MgPME; this pigment is not detectable in methanol extracts of the WT (see also supplemental Fig. 4). Myxo, myxoxanthophyll; Zea, zeaxanthin; Echi, echinenon; ␤-Car, ␤-carotene. . Analysis of Chl precursors in the WT and ycf54 strains grown mixotrophically under low light conditions. Chl precursors were extracted with 80% methanol, 20% water, 0.2% NH 4 OH from cells at OD 750 ϭ ϳ0.35 and analyzed by HPLC equipped with a diode array detector and a pair of fluorescence detectors (for detector settings see "Experimental Procedures"). A, shown is separation of precursors as detected by fluorescence detector 1; values indicate relative content of each precursor in the mutant when compared with WT. The concentration of MgPME in the ycf54 strain exceeded the detection limit of the fluorescence detector. B, an identical run was recorded by the diode array detector at 435 nm. Two unusual pigments detected in the mutant at 7.6 and 8.0 min are also highlighted by an arrow (see Fig. 6 for absorbance spectra). The use of the diode array detector allowed an estimate of MgPME level that was approximately100 times higher in the ycf54 strain relative to the WT. a.u., absorbance units.
activity. The Chl intermediates that accumulate in the ycf54 mutant, P IX , MgP, and MgPME, all precede the cyclase step in the Chl biosynthetic pathway. If the enzymes for the first three steps of the Chl pathway all act independently of each other, one might expect that the rise in pigment levels would be confined to just MgPME, the substrate for the cyclase. The observed increase in pigments all the way back to P IX , the "branch point" substrate for both Mg-chelatase and ferrochelatase, argues for some coupling between the Mg-chelatase, MgP methyltransferase, and cyclase steps, so that the effects of MgPME accumulation are transmitted back along the pathway. Indeed, mechanistic coupling between the Mg-chelatase and methyltransferase steps has been demonstrated (28). Disrupted channeling of intermediates between MgP methyltransferase and the cyclase by a deficiency in Ycf54 could impair the stability of enzymes along the Chl pathway. Inactivation of the cyclase in tobacco has already been observed to have a negative effect on the level of PChlide oxidoreductase (29). We observed lowered levels of sll1214 and por transcripts (Fig. 3, A and B), so it appears that the proposed assembly of a catalytic complex could involve coupled transcription, translation, and protein folding.
Interestingly, the structure of Ycf54 exhibits a similarity with the structure of the Psb28 protein (see supplemental Fig. 5). Deletion of the gene encoding Psb28 also affects PChlide synthesis in Synechocystis (30), although not to the extent observed in the ycf54 mutant. Given that Psb28 binds CP47 during the process of photosystem II assembly (30), this protein might be a good candidate for a factor harmonizing Chl biosynthesis and photosystem II biogenesis.
There is a sharp contrast between the large decrease in PChlide and the unaffected level of Chlide. It might be expected that the demands of assembling photosynthetic complexes in a mutant that can synthesize only 5% of the normal amount of PChlide would exhaust the Chlide pool. This precursor is, however, also an intermediate in the constant process of recycling Chl (31), which might be accelerated in the ycf54 mutant, resulting in replenished levels of Chlide. The fact that an apparently unaffected Chlide level can exist in a cell severely depleted in photosynthetic complexes suggests the existence of two separate Chlide pools, one filled by PChlide reductases and destined for de novo synthesis of photosystems and the other arising from de-phytolation of Chls. It seems that de novo production of Chl is essential to build new photosystems despite the existence of a pool of Chlide that probably originates from recycled Chl. It is also possible that there is a pool of Chlide-binding proteins that releases the pigment only when more Chlide is supplied by de novo production.
Analysis of the absorption and fluorescence emission spectra (Fig. 6) revealed some intriguing spectroscopic properties for the 7.6-and 8-min elution peaks identified in Fig. 5, A and B. The absorbance features of these pigments extracted from the ycf54 mutant resemble those found in earlier work when greening cucumber cotyledons from etiolated plants were studied using fluorometry (32). In these experiments, MgPME-enriched cotyledons were illuminated for 5 h, and emission features were observed at ϳ596, ϳ603, and at 614 -617 nm, with 420-nm excitation. The excitation spectra for these various emission peaks had a single maximum, at 434 -436 nm (32), that matches well with the 432-nm absorbance maximum of a FIGURE 6. Absorbance and fluorescence spectra of two unusual pigments found in the ycf54 strain. A, shown are absorbance spectra recorded by HPLC diode array detector. Pigments are described according to the time of elution from the HPLC column (7.6 and 8 min; see Fig. 5) and normalized to an absorbance at 432 nm ϭ 1. The inset shows absorbance spectra of MgPME and PChlide obtained using the same detector. B, shown are fluorescence emission spectra recorded by the HPLC fluorescence detector and normalized to the same maximum of fluorescence. The 7.6-and 8.0-min peaks and PChlide were excited at 435 nm; MgPME was excited at 416 nm. pigment (Fig. 6A) produced by the ycf54 mutant. These pigments, called "longer wavelength metalloporphyrins," were proposed as biosynthetic intermediates between MgPME and PChlide (32). However, the unusual pigments we observed in the ycf54 strain and probably also the pigments identified in (32) do not correspond to the chemically synthesized cyclase intermediates used by Wong et al. (8); the spectroscopic properties of 6-methyl-␤-hydroxy-MgPME are very similar to those of MgPME (33), and the 6-methyl-␤-keto derivative has a similar absorbance spectrum red-shifted by just a few nanometers (34). Early studies of the cyclase employed enzyme assays based on fractionated extracts prepared from chloroplasts and could not benefit from current methods using the recombinant production of proteins. The lack of any mechanistic data on the MgPME cyclase is in sharp contrast to recent progress in the kinetic analysis of other enzymes in the Chl biosynthetic pathway, for example with Mg chelatase, MgP methyltransferase, and PChlide oxidoreductase (35)(36)(37). The identification of Ycf54 as a factor critical for the activity of MgPME cyclase in vivo should allow further progress with the biochemical analysis of this step in chlorophyll biosynthesis.