J Biol Chem, Vol. 274, Issue 31, 22114-22121, July 30, 1999

Plant Riboflavin Biosynthesis
CLONING, CHLOROPLAST LOCALIZATION, EXPRESSION, PURIFICATION, AND PARTIAL CHARACTERIZATION OF SPINACH LUMAZINE SYNTHASE^*

Douglas B. Jordan, Karen O. Bacot^§, Thomas J. Carlson^¶, Martin Kessel, and Paul V. Viitanen^§**

From the  E. I. DuPont de Nemours Agricultural Products, Stine-Haskell Research Center, Newark, Delaware 19714, the ^§ E. I. DuPont de Nemours Life Sciences, Experimental Station, Wilmington, Delaware 19880-0402, ^¶ E. I. DuPont de Nemours Agricultural Products, Experimental Station, Wilmington, Delaware 19880-0402, and the  Laboratory of Structural Biology, National Institutes of Health, Bethesda, Maryland 20892

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Lumazine synthase, which catalyzes the penultimate step of riboflavin biosynthesis, has been cloned from three higher plants (spinach, tobacco, and arabidopsis) through functional complementation of an Escherichia coli auxotroph. Whereas the three plant proteins exhibit some structural similarities to known microbial homologs, they uniquely possess N-terminal polypeptide extensions that resemble typical chloroplast transit peptides. In vitro protein import assays with intact chloroplasts and immunolocalization experiments verify that higher plant lumazine synthase is synthesized in the cytosol as a larger molecular weight precursor protein, which is post-translationally imported into chloroplasts where it is proteolytically cleaved to its mature size. The authentic spinach enzyme is estimated to constitute <0.02% of the total chloroplast protein. Recombinant "mature" spinach lumazine synthase is expressed in E. coli at levels exceeding 30% of the total soluble protein and is readily purified to homogeneity using a simple two-step procedure. Apparent V_max and K_m values obtained with the purified plant protein are similar to those reported for microbial lumazine synthases. Electron microscopy and hydrodynamic studies reveal that native plant lumazine synthase is a hollow capsid-like structure comprised of 60 identical 16.5-kDa subunits, resembling its icosahedral counterparts in E. coli and Bacillus subtilis.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Riboflavin, vitamin B₂, is the precursor of flavin mononucleotide and flavin adenine dinucleotide, essential cofactors for a multitude of mainstream metabolic enzymes that mediate hydride, oxygen, and electron transfer reactions. Consequently, critical cellular processes as diverse as the citric acid cycle, fatty acid oxidation, photosynthesis, mitochondrial electron transport, and de novo pyrimidine biosynthesis are fundamentally dependent on riboflavin availability. Despite its essentiality, however, only plants and certain microorganisms can synthesize vitamin B₂, whereas higher animals, including man, must obtain it through their diet. In contrast, flavokinase and FAD pyrophosphorylase, the enzymes that convert riboflavin to FMN and FAD, respectively, are widely distributed in nature. Our current knowledge of riboflavin biosynthesis is largely restricted to bacteria and yeast (reviewed in Ref. 1). In both cases, the synthetic pathway consists of seven distinct enzyme-catalyzed reactions, with GTP and ribulose 5-phosphate the ultimate, noncommitted precursors. Although the sequence of events that are catalyzed in the second and third steps occur in opposite order in bacteria and fungi, the remaining pathway intermediates are identical in both types of microorganisms.

The last two enzymes of riboflavin biosynthesis, lumazine synthase (LS)¹ and riboflavin synthase (RS), are the best characterized, both structurally and mechanistically. In Bacillus subtilis, these proteins physically interact with each other to form a huge spherical particle with a combined molecular mass of ~1 MDa (2); the x-ray structure of this complex has been determined at 3.3 Å (3). Through an elegant series of experiments, Bacher and co-workers (4, 5) have shown that the outer shell of the B. subtilis LS-RS complex consists of 60 LS subunits that are organized as 12 densely packed pentamers to form a hollow icosahedral capsid. Encaged within the central core of the sphere resides a single molecule of RS, a trimer of three identical subunits. It has been proposed that compartmentation of the two terminal enzymes in the LS·RS complex improves the overall catalytic efficiency of riboflavin production through "substrate channeling," especially at low substrate concentrations (6). Although a bifunctional LS·RS complex has only been reported for certain strains of Bacillus and Clostridium (2), it is important to note that Escherichia coli LS also exists in vivo as a hollow icosahedral capsid of 60 identical subunits (7). The functional significance of this single protein complex, if any, remains to be elucidated.

LS catalyzes the penultimate step of riboflavin biosynthesis, namely the condensation of 3,4-dihydroxy-2-butanone 4-phosphate (DHBP) with 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione (ARAPD) to yield one molecule each of orthophosphate and 6,7-dimethyl-8-ribityllumazine (DMRL) (Fig. 1). The latter is the immediate precursor of riboflavin. LS genes have been cloned from numerous microorganisms, including E. coli (8), Actinobacillus pleuropneumoniae (9), Photobacterium phosphoreum (10), B. subtilis (7, 11), and Saccharomyces cerevisiae (12). In all cases, the subunit molecular mass of the encoded protein is small, ranging in size from ~16 to 17 kDa. Although the various microbial lumazine synthases share certain structural features, their overall homology is rather poor. For example, the proteins from yeast and E. coli are only 36% identical at the amino acid sequence level.

View larger version (10K): [in this window] [in a new window]   Fig. 1.   The physiological reaction catalyzed by LS.

The ultimate step of riboflavin biosynthesis is mediated by RS, which catalyzes the dismutation of two molecules of DMRL to yield one molecule of riboflavin and one molecule of ARAPD. RS has been cloned from yeast and bacteria, and all of the species examined thus far exhibit marked internal homology in their N-terminal and C-terminal domains (13, 14). Consequently, it has been proposed that the two "halves" of the RS protomer have arisen through gene duplication and that each contains a substrate-binding site for DMRL. Although the atomic structure of RS remains to be determined, it is known that the B. subtilis (2, 13) and S. cerevisiae (14, 15) homologs are both trimers in the native state, each consisting of three identical ~25-kDa subunits.

It is frequently stated in the literature that plants can synthesize riboflavin (1). However, virtually nothing is known about the metabolic machinery that is responsible for this process. Apart from an arabidopsis GTP cyclohydrolase II that was recently cloned (16), there are no reports of either homogeneous proteins or full-length cDNA sequences for any other enzyme that is involved in plant riboflavin biosynthesis. RS from spinach has been partially purified (17), but there is no information on its primary structure. The arabidopsis GTP cyclohydrolase II is only 39-58% identical to its counterparts in E. coli, B. subtilis, and P. phosphoreum at the amino acid sequence level. Assuming that the other pathway enzymes have diverged to a similar degree, it is extremely doubtful that strategies that only rely on structural similarities to the known microbial homologs will be very useful in cloning the corresponding plant genes.

In the present work, we have cloned LS genes from three plant species (spinach, tobacco, and arabidopsis) through their abilities to rescue an E. coli mutant whose endogenous LS gene was disrupted by insertional inactivation. Since this strategy relies only on functional similarities between the disrupted host gene and the target gene of interest, it is ideally suited for cloning heterologous proteins that catalyze the same reaction. Indeed, several bacterial (18, 19) and fungal (12, 14) riboflavin biosynthetic enzymes, as well as plant GTP cyclohydrolase II (16), were identified through functional complementation of microbial riboflavin auxotrophs. Unexpectedly, higher plant LS is synthesized as a precursor protein with a cleavable N-terminal chloroplast targeting sequence. Upon post-translational import into the chloroplast stroma, the transit peptide is removed giving rise to the mature polypeptide. Mature spinach LS has been expressed in E. coli, purified to homogeneity, and characterized, both structurally and kinetically. Here we present the results of these studies.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Materials-- The spinach, tobacco, and arabidopsis  cDNA expression libraries were obtained from Stratagene (catalog numbers 936005, 936002, and 937010, respectively).

Cloning and Gene Disruption of E. coli LS-- The E. coli LS gene (GenBank^TM accession number X64395) was amplified from genomic DNA of strain W3110 (20) using two PCR primers.² Primer 1 (5'-CGA AGG AAG Acc atg gCC ATT ATT GAA GCT AAC GTT GC-3') introduces an NcoI site at the initiation codon, and primer 2 (5'-ATC TTA CTg tcg acT TCA GGC CTT GAT GGC TTT C-3') provides a SalI site just past the stop codon. The PCR fragment was cut with NcoI and SalI and ligated into similarly digested pGEM-5Zf(+) (Promega, Madison, WI). The ligation mixture was used to transform E. coli DH5, and growth was selected on LB media that contained ampicillin (100 µg/ml). A representative plasmid with an insert of the correct size (referred to below as "pGEM-LS") was sequenced completely to confirm the absence of PCR errors.

Next, the E. coli LS gene was insertionally inactivated. Briefly, a unique NotI site was introduced into the middle of its coding region, and a DNA fragment conferring kanamycin resistance was ligated into the engineered site. The selectable marker was provided by the Kan^r GenBlock cartridge (Amersham Pharmacia Biotech), which was modified by PCR to add a NotI cleavage site to both of its ends. This was accomplished using primer 3 (5'-AAC TAG ATC Agc ggc cgc AGC CAC GTT GTG TCT CAA A-3') and primer 4 (5'-GAC AAA CAT Agc ggc cgc TGA GGT CTG CCT CGT GAA-3') in a standard PCR reaction. The resulting fragment was cleaved with NotI and purified by agarose gel electrophoresis.

A unique NotI site was also introduced into the middle of the E. coli LS gene using "inverse PCR" (21). The target for amplification was pGEM-LS, and the 100-µl reaction contained ~0.5 ng of plasmid DNA and primer 5 (5'-AAC TAG ATC Agc ggc cgc GGT ACG GTT ATT CGT GGT-3') and primer 6 (5'-GAC AAA CAT Agc ggc cgc GTC GTA TTT ACC GGT-3') each at 0.5 µM. The primers hybridize to the E. coli LS gene (GenBank^TM accession number X64395) at nt 2273-2290 and nt 2243-2261, respectively, and each contains a NotI cleavage site at its 5' end. The PCR product was cleaved with NotI and subjected to a standard ligation reaction with the modified Kan^r GenBlock. The ligation mixture was used to transform E. coli DH5, and growth was selected on LB media that contained ampicillin (100 µg/ml) and kanamycin (30 µg/ml). A plasmid harboring the disrupted E. coli LS gene (referred to "pLSKan") was sequenced completely to confirm that the selectable marker was inserted at the correct location.

Creation of an E. coli LS Knockout Mutant-- pLSKan was digested with NcoI and SalI to liberate the insertionally inactivated E. coli LS gene. The purified fragment was electroporated into E. coli JC7623, and growth was selected on LB media with riboflavin (400 µg/ml) (18) and kanamycin (30 µg/ml). JC7623 was chosen as the initial recipient for gene replacement since it undergoes double-crossover homologous recombination at high frequency (22-24). The correct chromosomal integration event was detected phenotypically. Mutants resulting from homologous recombination of the disrupted E. coli LS gene were resistant to kanamycin and could only grow in the presence of added riboflavin.

Due to mutations in the recBCD loci, E. coli JC7623 is unable to propagate ColE1-type plasmids (Ref. 24 and references therein). It was therefore necessary to move the disrupted E. coli LS gene from the chromosome of JC7623 to a suitable strain that could be used for screening cDNA libraries. This was accomplished through phage transduction using P1_vir and well established methodologies (25). The recipient strain was E. coli W3110 (20), and growth was selected on LB media with sodium citrate (7.5 mM), magnesium sulfate (1.5 mM), kanamycin (35 µg/ml), and riboflavin (400 µg/ml). A stable transductant, in which the disrupted E. coli LS gene had replaced the normal chromosomal copy, was identified phenotypically and was used for complementation cloning as described below; the mutant was resistant to kanamycin and could only grow in the presence of added riboflavin.

Complementation Cloning-- cDNA expression libraries were subjected to mass excision according to the vendor's protocol. Upon excision, the cDNA inserts are contained in the plasmid vector, pBluescript, which confers ampicillin resistance. The plasmid mixture was then introduced into the E. coli LS null mutant using electroporation, and transformants were plated on LB media that contained sodium citrate (7.5 mM), magnesium sulfate (1.5 mM), kanamycin (35 µg/ml), ampicillin (100 µg/ml), and 0.6 mM isopropyl-1-thio--D-galactopyranoside. After a 48-h growth period at 37 °C, plasmids were isolated from representative colonies, and their cDNA inserts were sequenced completely.

Overexpression of Mature Spinach LS-- PCR primers were used to remove the chloroplast transit peptide from the cloned spinach LS precursor. Primer 7 (5'-CTA CTC ATT Tca tat gAA CGA GCT TGA AGG TTA TGT CAC-3') hybridizes to nt 205-224 of the DNA sequence shown in Fig. 2 and replaces Val-67 (the predicted start of the mature protein) with an initiator Met residue. It also provides a unique NdeI site for cloning purposes and changes the second residue of the mature protein from Arg to Asn. It was anticipated that these changes would not affect enzyme activity and might actually improve bacterial expression of spinach LS, since the E. coli and B. subtilis homologs both start with Met-Asn. Primer 8 (5'-CAT CTT ACT gga tcc ATC AGG CCT TCA AAT GAT GTT CG-3') hybridizes at the other end of the gene to nt 648-667 (Fig. 2) and provides a unique BamHI site just past the stop codon. The PCR fragment was digested with NdeI and BamHI and ligated into the T7 expression vector, pET-24a(+) (Novagen), cleaved with the same enzymes. The construct was introduced into E. coli BL21(DE3), and plasmid DNA from a selected transformant was sequenced completely to check for PCR errors; none were found.

View larger version (57K): [in this window] [in a new window]   Fig. 2.   Nucleotide sequence and deduced amino acid sequence of the spinach LS precursor. The coding region for the spinach LS precursor starts at nucleotide 1. The sequence of the putative N-terminal chloroplast transit peptide is underlined. Numbers in parentheses refer to amino acid residues.

For protein production, the recombinant strain described above was grown at 37 °C in LB media containing kanamycin (50 µg/ml). Induction was with isopropyl-1-thio--D-galactopyranoside (1 mM) at an A_600 nm of 0.7, and the cells were harvested 3 h later by centrifugation. A cell-free extract was prepared as described previously (26), and stored at 80 °C at a protein concentration of 31 mg/ml (27).

Purification of Mature Spinach LS-- Aliquots (5 ml) of the E. coli cell-free extract were diluted 1:1 with water and applied to a Mono-Q HR 10/10 column (Amersham Pharmacia Biotech) pre-equilibrated at 25 °C with Buffer Q (50 mM Tris-HCl, pH 7.6, 10 mM sodium sulfite, 1 mM EDTA). The column was developed at 4.0 ml/min with a linear gradient (120 ml) of 0-0.5 M NaCl (in Buffer Q). Four more aliquots of the extract were processed in the same manner, and fractions eluting between 0.35 and 0.38 M NaCl were pooled, supplemented with 5% glycerol, and concentrated to 22 mg/ml of protein; the recovery of protein was ~135 mg.

The entire sample was then fractionated on a TSK G3000SW gel filtration column (21 × 600 mm) in 2-ml aliquots. The column was developed at 4 ml/min with 50 mM Tris-HCl, pH 7.7, 0.3 M NaCl (25 °C). Purified mature spinach LS emerged from the column as a single peak immediately after the void volume. Fractions eluting between 19.6 and 20.7 min were supplemented with 5% glycerol, concentrated to ~20 mg of protein per ml, and stored at 80 °C. The final yield was ~73 mg of purified protein.³

Chloroplast Import Assays-- The full-length spinach LS precursor (original cDNA clone) was modified for insertion into the MscI and BamHI sites of the in vitro transcription/translation vector, pCITE-4a(+) (Novagen) using PCR primers 8 and 9 (5'-TCA TTT CAT Atg gcc aGT TTT GCA GCT TCT CAA ACT TGT-3'). As noted above, the former adds a unique BamHI site just after the stop codon, whereas the latter introduces an MscI site at the initiator Met residue; neither primer alters the primary sequence of the spinach LS precursor. The resulting construct was subjected to in vitro transcription/translation using [³⁵S]methionine and the "Single Tube Protein System 2, T7" kit (Novagen) according to the vendor's protocol. Reactions were terminated with 2× import buffer containing 60 mM unlabeled methionine (28) and stored at 80 °C for subsequent use.

Chloroplasts were isolated from 13-day-old pea seedlings (Pisum sativum) and subjected to in vitro import assays (28) using the radiolabeled spinach LS precursor. Protease post-treatment was used to distinguish between bound and imported polypeptides (28, 29). Intact plastids were then repurified by centrifugation through Percoll cushions, resuspended in 100 µl of 2× gel sample buffer, and analyzed by SDS-PAGE/fluorography as described previously (28).

Lumazine Synthase Activity-- Initial rates of lumazine formation were monitored spectrophotometrically at 408 nm, using an extinction coefficient of 10,000 M¹cm¹ (30). The 1-ml reactions contained 50 mM Tris-HCl, pH 7.5, 0.5 mM ARAPD, 1.0 mM DHBP, 0.1 mM dithiothreitol, and purified mature spinach LS at 37 °C; reactions were initiated with enzyme. Unless otherwise indicated, kinetic constants were obtained at 25 °C by holding ARAPD or DHBP at a fixed saturating concentration (0.5 mM), while varying the concentration of the other substrate.

DHBP was synthesized enzymatically in a reaction containing 50 mM Tris-HCl, pH 7.5, 20 mM MgCl₂, 20 mM D-ribose 5-phosphate, 10 units/ml phosphoriboisomerase, and 0.3 units/ml purified E. coli DHBPS at 25 °C (18). ARAPD was prepared from 4-ribitylamino-5-nitroso-2,6-dihydroxypyrimidine (30) by catalytic hydrogenation. The 40-ml reaction contained 0.4 mmol of the latter compound (in 10 mM acetic acid) and 20 mg of 10% palladium on carbon. After a 15-h incubation period at 25 °C (under 50 pounds/square inch H₂), the catalyst was removed by filtration and the ARAPD-containing filtrate was stored at 80 °C. ARAPD and DHBP concentrations were determined by enzymatic conversion to DMRL, with one of the substrates limiting and the other in excess.

Other Methods-- SDS-PAGE was performed with 15% gels (31) that were aged overnight. For Western blots, proteins were transferred to nitrocellulose, reacted with antisera against the purified recombinant spinach LS, and sequentially probed with biotinylated anti-rabbit IgG and streptavidin-conjugated horseradish peroxidase, which were both from Vector Laboratories. Electrospray ionization mass spectrometry was performed on a Fisons VG Quattro 11, calibrated with myoglobin; the sample was analyzed at a concentration of 20 µM protomer in acetonitrile/water/formic acid (50/50/0.5). PCR amplification was with the GeneAmp PCR Reagent kit using a DNA Thermocycler 480 (Perkin-Elmer). Automated DNA sequencing was performed on an Applied Biosystems Inc. 377 using custom-designed primers. Amino acid sequence comparisons and alignments were performed using the GAP and PILEUP programs of the Genetics Computer Group (Wisconsin package, version 9, Genetics Computer Group, Madison, WI). For sedimentation equilibrium studies, spinach LS (33 µM protomer) in 100 mM potassium phosphate buffer at pH 7.0 was placed into two separate sectors of a six-sectored cell and centrifuged at 4 °C and 4500 rpm in an XL-A Optima (Beckman) analytical centrifuge; the reference sectors contained the same buffer. After 28, 47, and 72 h of centrifugation absorbances at 280 nm of the six-sectored cell were scanned stepwise at 0.02-mm intervals with 10 readings per step. The absorbance data were fit to a model for an ideal monodisperse system having a partial specific volume of 0.724 as calculated from Ref. 32. For electron microscopy, purified recombinant mature spinach LS was diluted with 50 mM Tris-HCl, pH 7.7, 0.3 M NaCl, negatively stained with 1% uranyl acetate, and examined in a Zeiss 902 transmission electron microscope operating at 80 kV. Micrographs were recorded at a nominal magnification of 50,000; bacteriophage T4 was used for calibration (33).

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES

Cloning of the Spinach LS-- As detailed under "Experimental Procedures," a spinach cDNA library was screened for functional complementation of an E. coli LS knockout mutant. The library was introduced by electroporation, and transformants were plated in the absence of added riboflavin. Reversion to riboflavin prototrophy occurred at a frequency of ~4 × 10⁶, and one of the complementing plasmids (designated pPVLS22) was selected for further analysis; the latter contained a cDNA insert of ~1.3 kilobase pairs. Shown in Fig. 2 is the complete nucleotide sequence and deduced amino acid sequence of the spinach protein that rescued the E. coli LS null mutant. The open reading frame encodes a protein of 222 amino acid residues with a predicted molecular mass of 23,426.46 Da. Although there are unmistakable similarities between the higher plant protein and known microbial LS homologs, there are also some important differences. The most obvious discrepancy is that spinach LS is a much larger protein (Fig. 3). Fused to its N terminus is a stretch of ~65 additional amino acid residues that are not present in known bacterial and fungal counterparts. That this polypeptide extension is relatively basic and rich in Ser and Thr residues suggests that it could be a chloroplast targeting sequence (34). According to this scenario, spinach LS is synthesized in the plant cytosol as a larger molecular weight precursor and is subsequently taken up by chloroplasts where it is cleaved to its mature size. Unfortunately, it is difficult to predict with certainty the exact site where processing occurs, since even among the known microbial LS homologs the first 15-20 N-terminal amino acid residues are poorly conserved. However, based on the sequence alignment shown in Fig. 3, it is likely that the critical cleavage event occurs between Ala-66 and Val-67, giving rise to a polypeptide with a protomer molecular mass of ~16.5 kDa for the mature spinach LS. Note that even without its chloroplast targeting sequence, spinach LS is only ~29-49% identical to its counterparts in E. coli, B. subtilis, A. pleuropneumoniae, P. phosphoreum, and S. cerevisiae.

View larger version (81K): [in this window] [in a new window]   Fig. 3.   Sequence alignment of LS homologs. The amino acid sequence of the spinach LS precursor is aligned to the corresponding sequences from E. coli (8), B. subtilis (11), A. pleuropneumoniae (9), P. phosphoreum (10), and S. cerevisiae (12). Residues that are identical in at least four of the sequences are indicated by open squares, and those that are identical in all six sequences are indicated by filled black circles. The predicted cleavage site for maturation of the spinach LS precursor is also indicated (inverted black triangle).

Chloroplast Localization-- To test the hypothesis that spinach LS is a chloroplast protein, the cloned full-length precursor was labeled with [³⁵S]methionine and subjected to in vitro import assays. Although the experiment shown in Fig. 4 A was performed with pea chloroplasts, virtually identical results have been obtained with spinach plastids. In vitro transcription/translation of the spinach LS precursor resulted in the synthesis of a radioactive polypeptide with an apparent molecular mass of ~23 kDa (lane T), consistent with the value predicted from its DNA sequence. In the presence of ATP (lane 3), this polypeptide was efficiently taken up by chloroplasts and processed to a smaller size. Classical protease protection experiments (28, 29) established that the radioactive polypeptide that was recovered with intact chloroplasts following import assays had actually been internalized (lane 4). Based on migration during SDS-PAGE, the subunit molecular mass of mature spinach LS is ~15.5 kDa, similar to the value predicted above after removal of the chloroplast transit peptide. Although not shown, imported spinach LS is exclusively located in the chloroplast stroma (e.g. the soluble fraction of chloroplasts).

View larger version (25K): [in this window] [in a new window]   Fig. 4.   Import and localization of spinach LS into isolated chloroplasts. A, import assays were performed in the light, in the presence of 5 mM ATP (lanes 3 and 4) or in the dark, without ATP (lanes 1 and 2). After import, chloroplasts were treated with protease as indicated and were analyzed by SDS-PAGE/fluorography as described under "Experimental Procedures." A photograph of the fluorogram is shown. The locations of the spinach LS precursor (pLS) and mature polypeptide (LS) are indicated. Lane T, the 35S-labeled translation product that was used for import experiments. B, localization of authentic mature spinach LS. Intact spinach chloroplasts, prepared as described in Ref. 28, were osmotically lysed at a chlorophyll concentration of 0.25 mg/ml, by resuspension in ice-cold 10 mM Hepes/KOH, pH 8, 5 mM EDTA. An aliquot of the lysate was removed for subsequent analysis, and the remainder was centrifuged (15,000 × g, 10 min) to yield membrane and soluble fractions. The samples were analyzed by SDS-PAGE, transferred to nitrocellulose, and probed with a 1:200 dilution of antisera against the recombinant mature spinach LS; each lane received an equivalent amount of the original lysate (~1.5 µg of chlorophyll). Lane 1, total lysate; lane 2, membrane fraction; lane 3, soluble fraction; lane 4, same as lane 3, plus 7 ng of purified recombinant mature spinach LS.

When chloroplasts were incubated with the spinach LS precursor in the dark without ATP, uptake and processing were not observed (Fig. 4A). Under non-energized conditions, the only radioactive band recovered with intact plastids was the full-length precursor (lane 1). Moreover, the latter completely disappeared after treatment with protease (lane 2), demonstrating that it had not been imported but was merely bound to the outer membrane. Thus, translocation of the spinach LS precursor is an energy-dependent process similar to the situation reported for other nuclear-encoded chloroplast proteins.

That plant LS is a chloroplast protein is also supported by immunological data (Fig. 4B). In these experiments, spinach plastids containing the authentic protein were osmotically lysed and subjected to centrifugation to yield membrane and soluble fractions. Following SDS-PAGE and transfer to nitrocellulose, the proteins in both fractions were probed with antisera directed against recombinant spinach LS (see below). Consistent with the in vitro import experiments, spinach LS was only found in the soluble fraction (lane 3) and was not detected in membranes (lane 2). Moreover, the resident protein detected by the antibodies co-migrates precisely with recombinant mature spinach LS (e.g. compare lanes 3 and 4). Finally, it should be noted that spinach LS is not an abundant protein. Based on quantitative experiments with antibodies and authentic standards, it constitutes <0.02% of the total chloroplast protein.

Expression, Purification, and Characterization of Mature Spinach LS-- In order to obtain chemical quantities of mature spinach LS for kinetic and structural studies, the chloroplast targeting sequence of the precursor protein was removed by PCR, and the modified gene was cloned behind a T7 promoter. Assignment of the initiator Met residue was based on the predicted cleavage site for precursor maturation (Fig. 3). As shown in Fig. 5A, the recombinant protein is well expressed in E. coli (lane 1), and easily purified to homogeneity using anion exchange and gel filtration chromatography (lane 2). The nucleotide sequence of mature spinach LS predicts a polypeptide of 16,534.71 Da, which is virtually identical to the value that was determined by electrospray ionization mass spectrometry (16,536.3 Da). Assuming that the higher plant protein forms an icosahedral capsid of 60 identical subunits similar to the E. coli (7) and B. subtilis (2) homologs, its native molecular mass should be ~992 kDa. Consistent with this prediction, recombinant mature spinach LS emerged from an analytical gel filtration column as a huge complex with an apparent molecular mass of ~850 kDa (Fig. 5B). Similar results were obtained when a spinach chloroplast extract was applied to the same sizing column, and fractions were subjected to Western blot analysis using anti-spinach LS antisera to visualize the authentic chloroplast protein (data not shown). A better estimate of the molecular mass of the spinach LS complex was obtained through sedimentation equilibrium studies with the purified recombinant protein. The four values determined after 47 and 72 h of centrifugation (e.g. at equilibrium) ranged from 973 ± 1 to 991 ± 1 kDa, very close to the value predicted for a 60-mer.

View larger version (13K): [in this window] [in a new window]   Fig. 5.   Purification and characterization of mature spinach LS. A, expression and purification of mature spinach LS. Lane 1, E. coli cell-free extract (24 µg of protein); lane 2, purified recombinant protein (8 µg). Analysis was by SDS-PAGE and Coomassie Blue staining. B, size analysis of mature spinach LS. Purified recombinant spinach LS (~220 µg) was applied to a TSK G3000SW gel filtration column (7.5 × 600 mm). The latter was developed at 1.0 ml/min with 50 mM Tris-HCl, pH 7.7, 0.3 M NaCl, 1 mM EDTA, 10 mM sodium sulfite (25 °C), and A280 nm was monitored. Inset, molecular mass analysis of mature spinach LS. The column was calibrated with the following standards: thyroglobulin (670 kDa), gamma globulin (158 kDa), ovalbumin (44 kDa), myoglobin (17 kDa), and vitamin B-12 (1.35 kDa).

Edman degradation of the purified recombinant protein confirmed that the first two N-terminal amino acid residues had been correctly altered through PCR from Val-Arg to Met-Asn as described under "Experimental Procedures." Despite these substitutions and removal of the chloroplast targeting sequence, the purified recombinant protein exhibited significant catalytic activity. At 37 °C, in the presence of saturating concentrations of DHBP and ARAPD, the apparent V_max of mature spinach LS was ~16,500 nmol mg¹ h¹. This value compares favorably to those reported for the homologous proteins from E. coli, B. subtilis, and S. cerevisiae (7), which range from 11,800 to 15,400 nmol mg¹ h¹. The low turnover numbers that are observed for all four proteins (e.g. ~ 3 per min) suggests that the overall rate of catalysis is hindered by a rather slow step or steps. The apparent K_m values for ARAPD and DHBP, determined with purified recombinant spinach LS, were 20 ± 2 and 26 ± 3 µM, respectively. Although the authentic plant protein is not available for comparison, the results suggest that purified recombinant spinach LS is probably fully active and that processing of the precursor protein in chloroplasts occurs at or very close to the predicted cleavage site.

Quaternary Structure of Spinach LS-- The results above open up the possibility that native spinach LS is a spherical 60-mer similar to the E. coli and B. subtilis homologs. However, two independent groups have recently reported that the yeast enzyme is a pentamer in the native state (7, 12), which represents the minimal functional unit. To explain this anomaly, it was proposed that a small insertion (four amino acids) near the C terminus of yeast LS precludes the assembly of pentamers into 60-mers (7). The Brucella abortus homolog also contains a small insertion at the exact same position (three amino acids), and it too behaves like a pentamer in solution. However, preliminary crystallographic analysis of this protein indicates that it can assemble into icosahedral particles, at least under certain conditions (35). It was therefore of interest to examine in greater detail the quaternary structure of higher plant LS.

Shown in Fig. 6 is an electron micrograph of purified recombinant spinach LS negatively stained with uranyl acetate. The predominant images observed are spherical particles with an average apparent diameter of about 17.5 nm. At this low level of resolution they superficially resemble the 17-nm particles that were obtained with recombinant E. coli LS, which were shown to be icosahedral capsids of 60 subunits each through hydrodynamic studies and metal decoration experiments (7). That the spinach LS particles are also hollow spheres is suggested by the accumulation of stain in the central portion of the molecules. Although the size distribution of spinach LS particles was fairly uniform, ranging from 17.5 to 18.5 nm in diameter, a few larger particles were evident in certain electron micrographs (data not shown). The latter could correspond to the heterogeneous population of spherical aggregates that are observed when B. subtilis LS is exposed to alkaline pH or its dissociated subunits reassemble in the absence of ligands. The true nature of the spinach LS complex remains to be determined by more exacting methodologies. Taken together, however, the above results suggest that the quaternary structure of lumazine synthase has been highly conserved in the evolution from bacteria to plants.

View larger version (89K): [in this window] [in a new window]   Fig. 6.   Electron micrograph of negatively stained recombinant mature spinach LS. Purified recombinant spinach LS was prepared for electron microscopy as described under "Experimental Procedures." A, a representative field showing spherical spinach LS particles with stain accumulation in their centers. B, a gallery of individual particles. The bar represents 50 nm.

Other Plant LS Genes-- During the present study, cDNAs for tobacco and arabidopsis LS were also cloned by functional complementation. The amino acid sequences of these proteins are shown in Fig. 7 aligned to spinach LS. It is clear that all three plant proteins are synthesized as larger precursors with N-terminal chloroplast targeting sequences. Although the transit peptides are of comparable length, they are highly divergent and bear little resemblance to each other. Excluding the transit peptides, however, the three plant proteins are 72-76% identical at the amino acid sequence level. This observation is reminiscent of other nuclear-encoded chloroplast proteins that typically exhibit much greater plant-to-plant variation in the targeting sequence than in the mature portion of the molecule (36). When the amino acid sequences of spinach, arabidopsis, and tobacco LS were scanned against the data base, a large number of bacterial and fungal counterparts were identified as being related. However, none of these proteins exhibited >50% identity to any of the plant proteins. The only plant LS homolog that was identified during this search was an arabidopsis protein with accession number AC004005 whose sequence is identical to the protein shown in Fig. 7.

View larger version (55K): [in this window] [in a new window]   Fig. 7.   Sequence alignment of three plant LS homologs. An amino acid sequence alignment of the spinach, tobacco, and arabidopsis LS homologs that were cloned by complementation is displayed. Residues that are identical in all three proteins are indicated by filled black circles, and those that are identical in only two of the proteins are indicated by open squares. The boxed residues at the N termini correspond to the putative chloroplast targeting sequences. Note: the tobacco cDNA clone is not full-length. Based on the alignment shown, the tobacco LS precursor protein is missing the first 3 amino acid residues.

Apart from their overall similarity to one another, the three plant LS homologs each possess a unique stretch of amino acids at their C terminus (i.e. ASLFEHHLK), a region of the protein that is not highly conserved in yeast or bacteria. That these nine residues are identical in three different species suggests that they have been selected for and that they might be of unique importance to the functionality of higher plant LS. According to the crystal structure of B. subtilis LS complexed with a substrate-like inhibitor (4), this region of the polypeptide is not directly involved with substrate binding or catalysis. Assuming this is also true for spinach LS, the nine plant-conserved residues must serve another function.

Conclusions-- Lumazine synthase from spinach, tobacco, and arabidopsis have been cloned as the first such examples from higher plants. Functional complementation was key to obtaining the cDNAs for these proteins, since they exhibit very little homology to known bacterial and fungal counterparts. Surprisingly, plant LS is a nuclear-encoded protein that is synthesized in the cytosol as a larger precursor with an N-terminal chloroplast targeting sequence. Following import into chloroplasts, the transit peptide is removed to yield "mature" LS, which assembles into a large spherical particle that superficially resembles the icosahedral 60-mers that have been described for E. coli and B. subtilis (5, 7). In contrast, fungal LS (the only other eucaryotic LS homologs for which sequence information is available) is synthesized in its mature form without an obvious organellar targeting sequence and upon folding does not assemble into a higher order structure larger than a pentamer (12).

That plant LS is a chloroplast protein opens up the possibility that other plant riboflavin biosynthetic enzymes might be similarly localized. A likely candidate is riboflavin synthase, the last enzyme in the pathway, which converts 2 mols of lumazine to 1 mol of riboflavin and 1 mol of ARAPD. Not only does this enzyme use lumazine, but one of its products, ARAPD, is recycled as a substrate by LS. As already noted, LS and RS are physically associated with each other in B. subtilis, to form a bifunctional "comparticle" that operates more efficiently than the sum of its parts (6). Whether a similar situation occurs in plants remains to be determined. However, given the above observations, it is difficult to imagine that plant LS and RS would physically be separated from each other in different cellular compartments.

Other considerations support the notion that riboflavin biosynthesis is largely, if not entirely, confined to chloroplasts. For example, GTP and ribulose 5-phosphate, the precursors for the two converging pathways that lead to riboflavin, are both abundant in this organelle. The enzymes that catalyze the formation of the first committed intermediates that result from these compounds are GTP-cyclohydrolase II (GTP-CHII) and DHBP synthase (DHBPS), respectively. In most bacteria and fungi, these two proteins are distinct enzymes that are encoded for by separate genes. However, other microorganisms, like B. subtilis and Photobacterium leiognathi, produce a single chimeric protein that appears to catalyze both enzyme activities (17, 18). In the bifunctional protein, DHBPS and GTP-CHII are literally linked together "head-to-tail" to form a species that resembles a fusion protein; the DHBPS moiety is at the N terminus.

Two sequences have recently appeared in the data base that clearly indicate that tomato and arabidopsis also have a bifunctional DHBPS/GTP-CHII; the accession numbers of these proteins are AJ002298 and AJ000053, respectively. Compared with the B. subtilis and P. leiognathi homologs, however, both plant proteins have about 125 extra amino acids at their N terminus. That these polypeptide extensions are rich in Ser and Thr residues suggests that they could be chloroplast targeting sequences. Although additional experiments are required to test this hypothesis, with two DNA sequences now available for higher plant DHBPS/GTP-CHII, the cellular location and biochemical properties of this interesting protein will soon be known.

	ACKNOWLEDGEMENTS

We thank Mike Picollelli, Rand Schwartz, Sylvia Stack, Tom Miller, and Mary Bailey for excellent technical assistance; Dana Smulski for providing bacterial strains and useful advice on the bacterial strains and P1 transduction; James Doughty for help in electrospray mass spectrometry measurements; and Daniel Camac and Joel Schneider for help with the analytical centrifugation studies.

	FOOTNOTES

^* The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

The nucleotide sequence(s) reported in this paper has been submitted to the GenBank^TM/EMBL Data Bank with accession number(s) AF147203 (spinach), AF148648 (tobacco), and AF148649 (arabidopsis).

^** To whom correspondence should be addressed: E. I. DuPont de Nemours Life Sciences, Experimental Station, Wilmington, DE 19880-0402. Tel.: 302-695-7032; Fax: 302-695-4509; E-mail: paul.v.viitanen@usa.dupont.com.

² For all PCR primers that appear in the text, the underlined bases hybridize to the target gene, and the unique restriction sites that were added for subcloning purposes are indicated in lowercase letters.

³ An extinction coefficient at 280 nm of 12,270 M¹ cm¹ (as calculated from the GCG Peptidesort program) was used to determine protein concentrations.

	ABBREVIATIONS

The abbreviations used are: LS, lumazine synthase (6,7-dimethyl-8-ribityllumazine synthase); RS, riboflavin synthase; DHBP, 3,4-dihydroxy-2-butanone 4-phosphate; ARAPD, 5-amino-6-ribitylamino-2,4(1H,3H)-pyrimidinedione; DMRL, 6,7-dimethyl-8-ribityllumazine; GTP-CHII, GTP-cyclohydrolase II; DHBPS, 3,4-dihydroxy-2-butanone 4-phosphate synthase; PAGE, polyacrylamide gel electrophoresis; PCR, polymerase chain reaction; nt, nucleotide(s).

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS AND DISCUSSION REFERENCES