Eur. J. Biochem. 204, 865-873 (1992) :C; FEBS 1992

Isolation, characterization, and sequence analysis of a cDNA clone encoding L-protein, the dihydrolipoamide dehydrogenase component of the glycine cleavage system from pea-leaf mitochondria Jacques BOURGUIGNON, David MACHEREL, Michel NEUBURGER and Roland DOUCE Laboratoire de Physiologie Cellulaire Vkgktale, Unite AssociCe au Centre National de la Recherche Scientifique No 576, Deparlernent de Biologic Moleculaire et Structurale, Centre dEtudes Nucledires et Universiti Joseph Fourier, Grenoble, France (Received September Z9/Novembcr 21, 1991) - EJB 91 1250

L-protein is the dihydrolipoamide dehydrogenase component of the glycine decarboxylase complex which catalyses, with serine hydroxymethyltransferase, the mitochondrial step of photorespiration. We have isolated and characterized a cDNA from a I gt31 pea library encoding the complete L-protein precursor. The derived amino acid sequence indicates that the protein precursor consists of 501 amino acid residues, including a presequence peptide of 31 amino acid residues. The N-terminal sequence of the first 18 amino acid residues of the purified L-protein confirms the identity of the cDNA. Alignment of the deduced amino acid sequence of L-protein with human, porcine and yeast dihydrolipoamide dehydrogenase sequences reveals high similarity (70% in each case), indicating that this enzyme is highly conserved. Most of the residues located in or near the active sites remain unchanged. The results described in the present paper strongly suggest that, in higher plants, a unique dihydrolipoamide dehydrogenase is a component of different mitochondrial enzyme complexes. Confidence in this conclusion comes from the following considerations. First, after fractionation of a matrix extract of pea-leaf mitochondria by gel-permeation chromatography followed by gel electrophoresis and Western-blot analysis, it was shown that polyclonal antibodies raised against the L-protein of the glycine-cleavage system recognized proteins with an M , of about 60000 in different elution peaks where dihydrolipoamide dehydrogenase activity has been detected. Second, Northernblot analysis of RNA from different tissues such as leaf, stem, root and seed, using L-protein cDNA as a probe, indicates that the mRNA of the dihydrolipoamide dehydrogenase accumulates to high levels in all tissues. In contrast, the H-protein (a specific protein component of the glycine-cleavage system) is known to be expressed primarily in leaves. Third, Southern-blot analysis indicated that the gene coding for L-protein in pea is most likely to be present in a single copy/haploid genome.

Mitochondria from plant leaves are distinct from all other mitochondria in that they are capable of oxidizing glycine at extremely rapid rates (Douce et al., 1977). The oxidation of glycine by mitochondria represents an important step in the metabolic pathway of photorespiration (Lorimer and Andrews, 1981; Husk et al., 1987). Glycine is broken down in the mitochondria1 matrix space by the glycine decarboxylase multienzyme complex (or glycine cleavage system) to produce COz, NH3, NADH and N5,N'o-methy1enetetrahydropteroylL-glutamic acid. The latter compound reacts with a second molecule of glycine to form serine and L- 5,6,7,8-te-

~

-

Correspandencv to J. Bourguignon, DBMSjPCV CEN-Grenoble,

F-38041 Grenoblc-cedex. France Ahhrevicitions. LPD, dihydrolipoamide dehydrogenase; PCRL, polymerase-chain reaction-synthesised L-protein probe. Enzymes. Glutathione reductasc [NAD(P)H] (EC 1.6.4.2); dihydrolipoamide dehydrogenasc (EC 1m . 4 ) ; T-protein (EC 2.1.2.10); P-protein (EC 1.4.4.2); pyruvate dehydrogenase (EC 1.2.4.1); oxoglutarate dehydrogenase (EC 1.2.4.2); dihydrolipoarnide acetyltransferase (EC 2.3.1.12): dihydrolipoamide succinyltransferase (EC 2.3.1.61). Nore. The nucleotide sequence data reported in this paper will appear in the EMBL Genbank and DDBJ Nucleotide sequence Databases under the accession number X63464.

trahydropteroyl-L-glutamic acid in a reaction catalyzed by serine hydroxymethyltransferase (Bourguignon et al., 1988). The glycine cleavage system has been purified from plants (Walker and Oliver, 1986a; Bourguignon et al., 1988), animals (Kikuchi and Hiraga, 1982) and bacteria (Klein and Sagers, 1966 and 1967) and is composed of four different component proteins referred to as P-protein, H-protein, T-protein, and L-protein. The reaction begins with the u-amino group of glycine forming a Schiff base with pyridoxal 5-phosphate at the active site of the P-protein (a homodimer of 97-kDa polypeptides). The 33.9-kDa lipoamide-containing H-protein reacts with the glycine bound to the P-protein to form a methylamine - H-protein intermediate. During the reaction, the a-carboxyl of glycine is lost as C02. The T-protein (a 45kDa monomer) catalyzes the transfer of the carbon of the methylamine group to L-5,6,7,8-tetrahydropteroyl-L-glutamic acid, with the formation of N5,N1'-methylenetetrahydropteroyl-L-glutamic acid and the release of the amino group of glycine as NH3. The reduced lipoamide from the H-protein, resulting from this transfer is reoxidized back to lipoamide by the FAD coenzyme bound to the dihydrolipoamide dehydrogenase (L-protein, a homodimer of 60-kDa polypeptides) with the sequencial reduction of FAD

866 and NAD’ . The recent results of Oliver et al. (1990a) demon- to P-protein, L-protein and T-protein were excised and injectstrated that the component proteins of the glycine-cleavage ed as an emulsion with Freund’s adjuvant into rabbits for system represent about 32% of the soluble protein of the raising antibodies. Injections (300 pg protein each time) were matrix of pea-leaf mitochondria and form a complex with an given four times at 3-week intervals. IgG were purified from approximate subunit ratio of 1 L-protein dimer/2 P-protein rabbit antisera according to Saint Blancard et al. (1981). Antidimers/27 H-protein monomers/9 T-protein monomers. On bodies reacting against L-protein were further immunothe basis of the effect of cycloheximide on the biosynthesis of purified by affinity chromatography using pure L-protein the glycine decarboxylase complex, Walker and Oliver (1986b) (1.2 mg) bound to CNBr-activated Sepharose 4B according to concluded that P-protein, H-protein, T-protein and L-pro- the manufacturer’s instructions (Pharmacia). Immunopurified antibodies were aliquoted and stored at -80’C in 30 mM teins were nuclear encoded. To gain more information about the structure, function Tris/HCI, pH 7.5,150 mM NaCI, 3 mM NaN, and 1YO(mass/ and biogenesis of the glycine-decarboxylase-multienzyme vol.) bovine serum albumin until use. The specificity of anticomplex, we have isolated and sequenced a pea L-protein bodies was checked by Western-blot analysis using a protein cDNA encoding the entire protein. Evidence is also presented extract from pea leaves as described below. For to show that the L-protein is encoded by a single nuclear gene immunoblotting studies of H-protein, we used polyclonal antiand that it is involved in the reoxidation of the dihydrolipoyl bodies raised against H-protein (Macherel et al., 1990). moieties of several dihydrolipoamide-dehydrogenase-containing enzyme complexes. Protein and nucleic-acid extraction

The soluble enzymes were released from the mitochondria (420 mg protein) by three freezelthaw cycles in 5 mM Mops, 5 m M Tris/HCl, 1 mM glycine, 1 mM EDTA, 1 mM octylglucoside, 2 mM 2-mercaptoethanol and 20 pM pyridoxa1 Sphosphate, pH 7.2. After removal of membranes by ultracentrifugation, the resulting crude extract was concentrated on an Amicon XM-10 membrane. This matrix extract contains all the proteins involved in the conversion of glycine to serine (Neuburger et al., 1986).

Protein extracts were prepared by the method of Hurkman and Tanaka (1986). Tissues were ground to a fine powder in liquid N2 with a mortar and a pestle. Powder (2.5 g) was adjusted to 5 ml with 0.7 M sucrose, 0.5 M Tris/HCl, pH 7.6, 50 mM EDTA, 0.1 M KCI, 2% (by vol.) 2-mercaptoethanol and 2 mM phenylmethylsulphonyl fluoride (buffer A). Phenol (1 vol.) was added and the mixture was gently shaken on a rotary wheel at room temperature for 20 min. The phases were separated by centrifugation and the phenol phase re-extracted twice with buffer A. Finally, the proteins were precipitated from the phenol phase by adding 5 vol. 0.1 M ammonium acetate in methanol at -20°C. The precipitate was washed three times with methanolic 0.1 M ammonium acetate and once with acetone. The proteins were solubilized in 50 mM Tris/HCl, pH 6.8, 100 mM dithiothreitol, 2 % SDS, 0.1O/o Bromophenol blue and 10% glycerol and stored at -80°C. RNA extraction was prepared as described by Macherel et al. (1990). The polyadenylated RNA fraction was prepared from the total RNA with the use of oligo(dT)-cellulose as described by Maniatis et al., 1982. Nuclear DNA was prepared from purified leaf nuclei (Watson and Thompson, 1986).

Matrix-extract fractionation

Construction of a pea-leaf cDNA library

The matrix extract (150 mg protein), supplemented with 4 pM leupeptin, was applied to a 2.5 cm x 90 cm column of Sephacryl S-300 (superfine grade, Pharmacia) equilibrated in 20 mM KCl, 5 mM Mops, 5 mM Tris, pH 7.5, 2 m M 2mercaptoethanol, 1 mM EDTA and 1 mM glycine. The column, connected to a Pharmacia F.P.L.C. system, was eluted with the same buffer at 4°C (flow rate 0.3 ml/min; fraction size 3 ml). The different proteins of the glycine-cleavage system were eluted in relation to their molecular masses as described earlier (Bourguignon et al., 1988). P-protein is eluted first (‘heavy’ fraction), followed by L-protein (‘intermediary’ fraction), then by T-protein and H-protein (‘light’ fraction).

cDNA (double-stranded, blunt-ended cDNA) was synthesized from pea-leafpolyadenylated mRNA (5 pg) using the Pharmacia cDNA synthesis kit according to the manufacturer’s instructions. Blunt-ended cDNA was ligated to EcoRI/ Not1 adaptors (Pharmacia). cDNA species were phosphorylated before ligation to 1 g t l l DNA that had been cut and dephosphorylated at the EcoRI restriction site. Ligation products were packaged with A head and tail proteins to produce viable bacteriophage particles using a Gigapack I1 Plus kit according to the manufacturer’s instructions (Stratagene). The resulting library consisted of approximately 3.5 x lo6 bacteriophage particles containing 96% recombinants.

Preparation of antibodies to P-protein, L-protein and T-protein

cDNA cloning

P-protein, L-protein and T-protein were purified from ‘heavy’, ‘intermediary’ and ‘light’ fractions, respectively, as described by Bourguignon et al. (1988). Purified proteins were checked for purity by SDSjPAGE in gels containing a 1015% (mass/vol.) acrylamide gradient. Bands corresponding

Approximately lo5 phage were subjected to screening with immunopurified antibodies raised against L-protein according to the method of Young and Davis (1983). One positive clone was purified, subcloned in Bluescript (pBS-SK, Stratagene) and sequenced. Several restriction fragments were

EXPERIMENTAL PROCEDURES Isolation of mitochondria Pea (Pisum sativurn) leaf mitochondria were isolated from 12 d-old plants by differential centrifugation and purified on Percoll/polyvinylpyrrolidone gradients (Douce et al., 1987). The purified mitochondria were suspended in a medium containing 0.3 M mannitol, 10mM phosphate, pH 7.2, 1 mM EDTA and 1 mM 2-mercaptoethanol, at 100 mg protein/ml. Preparation of matrix extract

867 also subcloned into plasmid Bluescript for sequencing. Recombinant pBS plasmids were used to transform competent Escherichia coli strain DHScl. Plasmid DNA was prepared by the alkaline lysis method (Birnboim and Doly, 1979). DNA sequencing

DNA sequencing was performed on both strands by the dideoxy-chain-termination method (Sanger et al., 1977) on double-stranded templates using 35S-labelled 2‘-desoxyadenosine 5’-[a-thio] triphosphate (Amersham France) and Sequenase version 2.0 (United States Biochemical Corporation). Plasmid DNA was denaturated by the alkaline method according to the procedures of the United States Biochemical Corporation. T3 and T7 primers were used to enable sequencing from each end of the insert. Oligonucleotides complementary to the sequence of the pea Lprotein cDNA were synthesized using a DNA synthesizer, model 381 A (Applied Biosystems Inc.) and used as primers. Immunoblotting analysis

Protein extracts were separated by SDS/PAGE in the buffer system described by Laemmli (1970). The proteins were electrophoreticdlly transferred to nitrocellulose sheets (Biordd). The membrane was blocked with 2% (mass/vol.) Tween 20 in 20 mM Tris/HCI, pH 8.0 and 150 mM NaCl (buffer B) for 5 min. The membrane was then incubated with rabbit antibodies against H-protein, T-protein, L-protein or P-protein in 0.05% (mass/vol.) Tween 20 in buffer B (solution C) during 2 h or overnight. After washing with solution C (6 x 5 min), antibody binding was determined using alkaline-phosphatase-conjugated goat[anti-(rabbit IgG)] (Boehringer), 1 :2000 diluted in solution C and staining for alkaline phosphatase activity. The colour reaction was performed by incubation of the membrane in 0.1 mM Tris, pH 9.5,4 mM MgC12, nitroblue tetrazolium chloride (0.1 mgfml) and 5-bromo-4chloro-3-indolyl-phosphate (0.06 mgfml). Northern-blot and Southern-blot analysis Northern-blot transfer of RNA, denaturated with glyoxal, to Hybond-N+ membranes (Amersham) was carried out as described by Macherel et al. (1990). A DNA probe corresponding to the complete cDNA was prepared by polymerasechain reaction amplification of the /z gtll insert with d forward-sequencing and 1-reverse-sequencing primers (PCRL, Innis and Gelfand, 1990). The PCRL probe was 32Plabelled with the random-primed labelling DNA kit (Boehringer) and hybridization was allowed to proceed overnight at 42 “C. Membranes were autoradiographed at - 80°C. High molecular-mass nuclear DNA was digested overnight with restriction endonucleases and concentrated by ethanol precipitation. DNA was fractionated by agarose-gel (0.8%) electrophoresis and Southern-blot transfer was performed as described by Macherel et al. (1990). DNA molecular-mass weight markers (1 kbp ladder, Gibco BRL) were 32P labelled by filling in the 3’-recessed ends with the Klenow fragment of E. coli DNA polymerase I (Maniatis et al., 1982). Amino acid sequencing Portions of protein (100 pmol) were used for direct sequence analysis by automated serial Edman degradation

with an Applied Biosystem 477 gaslliquid-phase protein Sequenator. Assay of dihydrolipoamide dehydrogenase

The activity of dihydrolipoamide dehydrogenase was assayed by measuring the formation of NADH that was dependent upon the presence of dihydrolipoic acid. The standard reaction mixture contained 20 mM Tris/HCl, pH 7.3, 15 mM 2-mercaptoethanol, 2 mM NAD’, 2 mM dihydrolipoate and enzyme in a total volume of 500 pl. The reaction was initiated by addition of dihydrolipoate. The rate of NADH formation was measured at 340 nm by using a Kontron (Uvikon 810) spectrophotometer. Dihydrolipoic acid was prepared from lipoic acid (~,~-6&thioctic acid; Sigma) as described by Kochi and Kikuchl(l976). Assay of pyruvate-dehydrogenase complex

The pyruvate-dehydrogenase complex was assayed at 340 nm using a Kontron (Uvikon 810) spectrophotometer. The standard assay mixture contained 20 mM Tris/HCl, pH 7.5, 0.2 mM thiamine pyrophosphate, 5 mM MgC12, 2 mM NAD’, 300 pM CoASH, 10 mM potassium pyruvate and enzyme in a total volume of 500 pl.

RESULTS Characterization of L-protein

A two-step protocol was devised to purify L-protein from pea-leaf mitochondria. The first step of purification was a fractionation of the soluble mitochondria1 proteins which was performed by loading a matrix extract containing all the proteins involved in glycine oxidation (Neuburger et al., 1986) on a Sephacryl S-300 gel-filtration column in the presence of 20 mM KC1. Under these conditions, the enzyme was eluted from the filtration column as three distinct peaks of activity. The distribution of activity between the three peaks was very stable from one experiment to another. About 3.5% of the total activity recovered was eluted with peak 1, 10% was eluted with peak 2 and 82Y0 was eluted with peak 3 (Fig. 1A). The proteins in all the fractions eluted from the column are shown in Fig. 1B. After gel filtration on Sephacryl S-300, the glycine-decarboxylase complex appeared to be unstable because all the components of the glycine-decarboxylase complex were clearly resolved (Bourguignon et al., 1988); on the basis of SDS/PAGE the P-protein was eluted first (‘heavy’ fraction), followed by the L-protein (‘intermediary’ fraction), T-protein and H-protein (‘light’ fraction; Fig. 1B). L-protein, present in the ‘intermediary’ fraction and associated with the major peak of activity (see Fig. IA), was then purified to homogeneity by anion-exchange chromatography on Mono Q using an F.P.L.C. system (see Bourguignon et al., 1988). On the basis of SDSjPAGE analysis, the L-protein exhibited a subunit molecular mass of 61 & 3 kDa. The molecular mass of the native enzyme was 120 5 kDa upon gel filtration and PAGE under non-denaturing conditions. This would suggest that the native L-protein from pea-leaf mitochondria is a dimer composed of two subunits with identical M, of 61 000. The N-terminal sequence of the purified L-protein was determined using a gasfliquid-phase sequencer and gave a single sequence of 18 amino acid residues (Ala-Ser-Gly-Ser-Asp-

868

.

h

I

v

A

1.6 n

I

U

l2

0

4

a

0.8

0.4

0

1

5

10

15

20

25

30

35

40

45

50 54

Fraction number

Fig. 1. Isolation of dihydrolipamidedehydrogenasefrom pea-leaf mitochondria. (A) Fractionation of thc matrix protcins of pea-leaf mitochondria by gel-filtration chromatography on a Sephacryl S-300 column. Approximately 5 ml (150 mg) of the matrix extract (see Experimental Procedures) was applied to the Sephacryl S-300 column (2.5 cm x 90 cm; flow rate: 0.3 ml/min) equilibrated in 20 mM KCI, 5 rnM Mops. 5mM Tris/HCl, pH 7.5, 2 mM, mercaptoethanol, 1 mM EDTA and 1 mM glycine. Fractions of 3 ml were collected. 30 PI cach fraction was used to assay the activity of dihydrolipoamide dehydrogenase by measuring the formation of NADH in the presence of dihydrolipoic acid as described in experimental procedures. (B) SDSjPAGE analysis of the eluted fractions. 10 pl each fraction were analysed on a SDSi7.5- 15% gradient polyacrylamide slab gel stained with Coomassie brilliant blue R-250. Proteins P, L, T, and H showing by arrows have been characterized by immunodetection. (C) Immunodetection of the dihydrolipoamide dehydrogenase using the immunopurified antibodies raised against the L-protein. Proteins (10 p1 of each fraction) were subjected to SDS/7.5- 15% PAGE and transferred to nitroccllulosc.

Glu-Asn-Asp-Val-Val-Ile-Ile-Gly-Gly-Gly-Pro-Gly-Gly). antibodies raised against the L-protein exhibited an identical Antibodies were raised against the purified L-protein. The proteins eluted from the Sephacryl S-300 column (see Fig. 1B) were then transferred to nitrocellulose sheets and analyzed by Western blotting. The results presented in Fig. 1 C show severa1 signals corresponding to the three distinct peaks of dihydrolipoarnide dehydrogenase activity reported above (Fig. 1A). In addition, all the proteins recognized by the

M , of 61 000. Isolation and characterization of cDNA encoding Lprotein of pea-leafmitochondria Plaques (lo5plaque-forming units) produced from a pealeaf cDNA 1-g t l l library were screened with immunopurified

869

MAAAGWAGCGTTMGTGATCGAATCGAAMGEGCTATGGCG 60

7

-------)

EcoR 1

Hind 111

M A M A N L A R R 9 I AAGGGTI'ACTCTClTCTCTCATCGGAGACTCTCCGCTACTCTTTlTCTCTCAGGTCAAGA 120 X G Y S L L S S E T L R Y S F S L R S R 29

Spei

EcoRI

GCT'ITCGCCTCCGGATC'EATGMAACGACGTCGTCATCATCGGTGGTGGTCCCGGCGGT 180 A F A S G S D E N D V V I I G G G ~ G G 49 TACGTCGCCGCCAWAMGCCGCTCAGCTTGGTITCMMCTACTTGCATCGAGAAGCGT Y V A A I X A A Q L G P K T T C I E K R

240 69

GGCGCCCTCGGTGGTA~TCTCAACGTI'GGATGCATCCCITCCAAGGCACTTTTGCAT 300 G A L G G T C L N V G C I P S K A L L H 89

I

s

200 bp I Fig. 2. Restriction endonuclease map and sequencing strategy of the pea L-protein cDNA. Arrows indicate thc direction and extent of the sequencing reactions. Regions indicated by dashed arrows were sequenced using either the T3or T, primer. Solid arrows indicate regions that were sequenced using 17-oligomer primers, complementary to the L-protein cDNA.

I __-_-______8

TCTTCGCATATGTACCATGAGGCTAMCATTCATTTGCCAACCATGGTGTTAAAGTTTCA 360 S S H M Y H E A K H S F A N H G V K V S 109

MTGTGGAGATTGACTTGGCTGCCATGATGGGGCMAMGATAAAGCTGTTTCTAATCTT 420 N V E I D L A A H M G Q X D X A V S N L 129 ACCCGGGGTATTGAAGGCCTATI'TMGAAGAATAAGGTAACCTATGTTAMGGATATGGA T R G I E G L F K K N K V T Y V K G Y G

AAATIYG~GCCGTCTGAMTCTCTGTAGACACCATFGMGGTGMAATACTGTGGTT 540 X F V S P S E I S V D T I E G E N T V V 169 AAGGGA?AGCGCTACFX"CAGATGTCAMTCTCTCCCCGGTGTCACT

X

antibodies raised against the L-protein component of the glycine-decarboxylase complex. Among several positive clones, 1 LD1 was shown to contain the largest cDNA insert (1.6 kb). It was subcloned into the Bluescript vector (clone pBS-L). The restriction map of the cDNA, together with the sequencing strategies employed, are shown in Fig. 2. The nucleotide sequence of the cDNA and the deduced amino acid sequence of the protein are shown in Fig. 3 . The first ATG triplet (nucleotides 34- 36) encountered downstream from the 5' end is designated as the translation start site because the surrounding sequence ( A A A A m G C ) agrees well with the favoured sequence ( A A C A a G G C ) flanking the consensus functional plant-initiator codon (Lutcke et al., 1987). The synthesis of the L lipoamide dehydrogenase polypeptide will therefore begin with the dipeptide methionyl-alanine, as do the majority of plant proteins. Interestingly, the sequence of the L-protein contains two adjacent methionyl-alanine. Starting with the first in-frame methionine residue, the open reading frame codes for a protein of 503 amino acid residues. The identity of the encoded protein was confirmed by comparison of the deduced amino acid sequence with the N-terminal sequence of the purified L-protein. The N-terminus of the mature protein corresponds to Ah32 in the deduced sequence, indicating that the protein is synthesized with a 31-amino acid-residue N-terminal presequence. The mature L-protein is predicted to consist of 470 amino acid residues, giving a protein of M , 49721 (or M , 50441 if the FAD cofactor is included), which is lower than the value of 61 000 obtained by SDS/PAGE. The presequence exhibits a typical enrichment for alanine (13%), leucine (l6%), arginine (I 6%) and serine (19Y0)and contains no isoleucine, one lysine (0.03%) and only one negatively charged residue (glutamic acid). This feature is characteristic of presequences of mitochondrial proteins (Von Heijne et al., 1989).

480 149

G

K

H

I

I

I

A

T

G

S

D

V

K

S

L

P

G

V

T

ATTGATGAAAAGMMTTGTATCATCGACTGGGGCTCTCGCATTGTCTGAMTCCCTMG I D E X K I V S S T G A L A L S E I P K

600 189

660 209

AAGCTTGTAGTTATTGGGGCAGGlTACATTGGGCTGGAAATGGGCTCAGTGTGGGGCCGA 720 229 X L V V I G A G Y I G L E M G S V W G R

ATTGGGTCTGAGGTMCTGTTGTTWLGmOWITCAGAGAGATTGTTCCMCCATGGACGCA I G S E V T V V E F A S E I V P T M D A

780 249

GAAATCCGGAAGCAGTTTCCGTTCTCTTGMAAGCAAGGCATGAMTTCAAACTGAAG 840 269 E I R X Q F Q R S L E K Q G M X F X L K ACCAAGGTGGTTGGAGTTGATACTTCTGGGGATGGTG'XAAGCTMCCGTTGMCCAWC 900 T K V V G V D T S G D G V K L T V E P S 289 G C T G G C G G T G A A C A G A C C A T A A T T G T C 960 309 A G G E Q T I I E A D V V L V S A G R T

CCATTCACTTCTGGAC?TMTTTCGATMGATAAGATAGGAGTTGA~CTGA~G~A~ACGG 1020 P F T S G L N L D X I G V E T D X L G R 329 ATTTTGGTAAATGAMGATTTTCAATGTCTCTGTCTCTGGTGTCTATGCAATCGGAGATGTG LO80 I L V N E R F S T N V S G V Y A I G D V 349 ATTCCAGGTCCMTGTTGGCACACMGGCAGMGMGATGGAGTTGCTTGTGTCGAGTAC I P G P M L A H X A E E D G V A C V E Y

1140 369

T T A G C C G G T M G G T T G G C C A T G T G G A C T A T G A C A A A C T A C 1200 389 L A G K V G H V D Y D K V P G V V Y T N

1260 CCTGAAGTTGCATCTGTAGGGMGACAGAGGAG~GGTTMGG~CTGGAGTTGAATAC 409 P E V A S V G K T E E Q V K E T G V E Y

CGTGTTGGAAAGTTCCCCATGGCTAATAGCAGGGCAAAGGCMTTGATAACGCTGM 1320 R V G K F P F W A N S R A X A I D N A E 429 GGACTAGTCAAGATAATTGCTGAMAGGAGACAGACAAAATATTGGGAGTACATATTATG G L V X I I A E K E T D K I L G V H I M

1380 449

GCACCTAATCCAGGAGMCTCATTCATGMGCAGCCATAGCATTGCAGTATGATGCATCA 1440

A

P

N

A

G

E

L

I

H

E

A

A

I

A

L

Q

Y

D

A

S

469

IIGTGAGGACATAGCACGTGTGTGCCATGCCAATCCAACAATGAGCGAAGCTATTAAAGAA 1500 S E D I A R V C H A N P T M S E A I K E 489

GCTGCAATGGCAACATATGACAAGCCCATTCACATTTGAAGAGCTGGTTGCTTCTTTCTT1560 A A M A T Y D K P I H I * 501 C T C T T G T T T T C A T I T ' T T G T C C M G G A - A C T T G G I

1620

AMCTTA

1627

Fig. 3. Nucleotide and predicted amino acid sequence of cDNA encoding Gprotein of the glycinedecarboxylase complex. Nucleotides and amino acids are numbered on the right of the sequence. The deduced amino acid sequence of the protein is depicted as single-letter codes positioned below the nucleotide sequence. The amino acid sequence which was determined by protein sequencing is underlined. The important disulfide active site is also represented between cysteine residues 76 and 81. The stop codon is marked with an asterisk.

Gene expression of L-protein When the polypeptides from the matrix of mitochondria from dark-grown or light-grown leaves are separated by SDS/ PAGE, the accumulation of H-protein, P-protein and T-protein of the glycine-cleavage system, as well as the serine hydroxymethyltransferase in the light-grown tissue, is striking (Macherel et al., 1990). However, the accumulation of Lprotein in the light-grown tissue is less pronounced than that observed for the other protein components of the glycinecleavage system. To determine whether the difference in the accumulation of the L-protein component of the glycine-

cleavage system corresponds to a change in the amount of mRNA, a Northern-blot analysis of the mRNA for L-protein was carried out using a 32P-labelledcDNA as probe. Poly(A)rich RNA species were prepared from leaves of 8-d old pea seedlings, which were grown in total darkness on a light/ dark program, or which were illuminated for 20 h after 8 d darkness. The result of Northern-blot analysis, performed under stringent conditions, are shown in Fig. 4. A mRNA of about 1.8 kb was easily detected in the three samples after an overnight exposure at - 80 "C. Clearly, the light-grown leaves

870

Fig. 4. Effect of light on the abundance of L-protein mRNA. Carefully calibrated amounts (2 pg) of poly (A)-rich RNA were size fractionated and hybridized with the 'ZP-labelled PCRL probe. (A) mRNA from green leaves, 8 d old; (B) mRNA from dark-grown leaves, 8 d old; (C) mRNA from leaves grown for 8 d in darkness and then exposed to 20 h light. The size of mRNA was determined by comparison with RN A molecular-mass markers (Boehringer). Autoradiography was for 48 h at -80°C.

Fig. 6. Occurrence of a single gene of pea L-protein. High-molecularmass DNA from pea-leaf nuclei (10 pg) was digested with BamHI (lane B), EcoRI (lane E), CZuI (lane C), Ssp1 (lane S), Dral (lane D), XbuI (lane X) or HindIII (lane H). Digested genomic DNA was fractionated on an 0.8% agarose gel, transferred to a Hybond N f membrane and hybridized with the 32P-labelled PCR, probe. The final wash was carried out in 2 x NaCi/Cit. containing 0.1 % SDS at 55 "C. Autoradiography was for 10 d a t - 80 "C.

Southern-blot analysis of nuclear DNA Southern-blot analysis was used to estimate the number of genes encoding L-protein in pea. Indeed, dihydrolipoamide dehydrogenase is also part of the three enzyme complexes catalyzing the oxidative decarboxylation of a-oxoacids (pyruvate, a-oxoglutarate and perhaps a-oxoisocaproate). This raises the question of whether a unique protein is involved in the reoxidation of the dihydrolipoyl moieties of all dihydrolipoamide - dehydrogenase-containing enzyme complexes or whether isoenzymes are involved. Pea genomic DNA digested with seven restriction endonucleases was fractionated on a 0.8% (massivol.) agarose gel and Southern blots were probed with 32P-labelledcDNA. The results, shown in Fig. 6, Fig. 5. Expression anaIysis of L-protein mRNA in different tissues. reveal that in six digestions, a single genomic DNA restriction Total RNA (1 5 pg) was size fractionated and hybridized with the 32P- fragment hybridizes to this probe, even after a low-stringency labelled PCRL probe. Autoradiography was processed after 5 d at wash. In Fig. 6, lane H, where Hind111 was used, two bands -8O'C. were observed. The appearance of these two bands is due to the presence of an HindIII site in the cDNA (Fig. 2). These results strongly suggest that pea contains a single L-protein contain more mRNA coding for L-protein than the dark- gene. grown leaves, and after a 20-h light exposure an increase in the mRNA level was observed. These results may indicate that light could play a role in the induction of L-protein mRNA. DISCUSSION However, the steady-state level of L-protein mRNA in etioThis is the first report of the isolation and the characterizalated pea leaves was far from negligible, suggesting that L- tion of a cDNA corresponding to the mRNA for plant Lprotein could be involved in some other complexes not en- protein, a constituent of the glycine-cleavage system involved gaged in a light-dependent process, such as photorespiration. in the photorespiratory pathway (Lorimer and Andrews, To study the expression of L-protein in different tissues of the 1981; Husic et al., 1987). Confidence in its identity comes plant, we also performed a Northern-blot analysis of total from the consideration that the predicted N-terminal sequence RNA from leaf, root, stem and seed (Fig. 5). The Northern- matches the sequence derived from Edman degradation of the blot analysis shows that the amount of L-protein mRNA is intact protein. This research follows previous work of this comparable in all the tissues analyzed, except in green leaves laboratory dealing with the isolation and biochemical characwhere the signal is lower. This may be due to the dilution of terization of the glycine-cleavage system from pea-leaf mitothc L-protein mRNA by chloroplastic ribosomal RNA. The chondria (Neuburger et al., 1986; Bourguignon et al., 1988; abundance of L-protein mRNA in all the tissues contrasts Oliver et al., 1990a; Oliver et al., 1990b) and the characterizawith the H-protein mRNA which accumulates primarily in tion of cDNA species for H-protein, another component of the glycine-cleavage system (Macherel et al., 1990). the leaf tissue (Macherel et al., 1990).

The cDNA clone for the L-protein was sequenced and revealed a 1503-nucleotide open reading frame that encoded a 501-amino-acid protein. We assigned the ATG (nucleotides 34 - 36) as the initiation codon because it agrees perfectly with the proposed plant initiation consensus sequence (Joshi, 1987; Lutcke et al., 1987). However, further experiments and genesequence data are required in order to characterize the initiation codon. The first 31 residues of the precursor have been found to bear some of the hallmarks of a mitochondria1 transit peptide, notably a ‘positive hydrophobic-polar design’ in the primary structure (Hart1 et al., 1989; von Heijne et al., 1989). The AATAAA element signalling polyadenylation in higher eukaryotes was not found at the 3’ end of pea L-protein cDNA. The predicted M,, calculated from the cDNA sequence, is 49 721. When the Mr of FAD is included, the total M , is 50441, which is lower than the value of 61 000 based on its mobility in SDSjPAGE (Bourguignon et al., 1988). The larger size estimated by SDS/PAGE analysis probably resulted from the relatively high density of positively charged amino acids (41 lysine, 12 histidine and 11 arginine) present in the L-protein (the deduced isoelectric point is 5.96). Consequently, the large number of SDS molecules bound to the protein may slow down its migration in SDS/PAGE. We compared the amino acid sequence of mature L-protein with other dihydrolipoamide dehydrogenases using the Palign program of PCGENE (Intelligenetics Inc.) according to the method of Myers and Miller (1988). The overall sequence similarity at the amino-acid level between the L-protein from pea (470 amino acids) and human dihydrolipoamide dehydrogenase (474 amino acids; Otulakowski and Robinson, 1987) is very high (70.2%; Fig. 7 and Table 1). Therefore, it is anticipated that the plant enzyme would not be different from human dihydrolipoamide dehydrogenase in secondary and tertiary structure. Table 1 shows the percentage similarity of the plant enzyme with several dihydrolipoamide dehydrogenases on the basis of identical residues (YOidentity) and when similar residues are taken into account (% similarity). Similar residues are those which can be interchanged without modifying the biological activity of protein. The following groups of residues are defined to be ‘similar’: A, S and T; D and E; N and Q; R and K; I, L, M and V; F, Y and W. It appears that the L-protein is more similar to the mammalian and yeast dihydrolipoamide dehydrogenases than those from bacteria. From crystallographic studies, the stereochemical description of the catalytic mechanism of human glutathione rcductase has been proposed (Pai and Schulz, 1983) and the involvement of four domains in the catalytic mechanism of the enzyme have been elucidated (Untucht-Grau et al., 1981; Krauth-Siege1 et al., 1982). All lipoamide dehydrogenases whose sequences have been reported have the same domain structure as glutathione reductase (for review see Carothers et al., 1989). Human glutathione reductase, a dimeric enzyme that catalyses the reduction of disulphides by pyridine nucleotide through an enzyme disulphide and a flavin, contains four structural domains within each molecule; the FAD domain, the NADP domain, the central domain and the interface domain [the interface domain binds the two identical subunits to each other (Williams et al., 1982)l. Sequence comparisons show that the important sites for enzyme activity are highly conserved. For example, the amino acid sequence of the redox-active thiol center or the disulfideactive site (the redox-active cysteines are labeled in Fig. 7 by an asterix) of flavin-nucleotide disulfide oxidoreductases, which is located in the FAD-binding domain of human glutathione reductase, show considerable similarity (Figs 7

L-protein

-

Human LPD

-

-

871

* .::

A D Q P I D A D V T V I G S G P G G Y I K A A Q L G F K T V C I E K G C -50 FAD

L-protein Human LPD L-protein Human LPD L-protein Human L P D L-protein Human L P D L-protein Xuman LPD

Domain

- I. P.S .K A.L.U S. S.H .M Y.H.E A. K.H.S -. F.A .N H.G.V K. V.S .N V.E.I D. L.A.A M M - 9 9 ..... .... .... ..... .. - 1PSKALL”SHYYHMAHGTDFASRGIEMSEVRLNLDKMMEQKSTAVKALT -100 - RGIEGLFKXNKVTYVKGYGKFVSPSEISVDTIEGENTWKGKHIIIATGS -149 ............. . . . . .. .. .. ..... .. .. .. .. - GGIAHLFKQNKWHVNGYGKITGKNQVTATKADGGTQVIDTWILIATGS -150 - DVKSLPGVTIDEKKIVSSTGALALSEIPKKLWIGAGYIGLEMGSVWGRI -199 ................................ ........................................ -

EVTPFPGITIDEDTIVSSTGALSLKKVPEXMWIGAGVIGVELGSVWQRL

-

N M + Domain GSEVTWEFASSIVPTM-DAEIRKQFQRSLEKQGMKFXLXTWGVDTSG

Human L P D

L-protein Ruuan LPD

-200

-248 . . . . . .. . . . ..................... .................. - G A D V T A V E F L G H V G G V G I D M E I S K N F Q R I L Q K Q G F K F X -250

- DG-VKLTVEPSAGGEQTITEADWLVSAGRTPFTSGLNLDKIGVETDKLG .. . . . . . ... . ... . . .. .. . . ... . ... . ... .. . . ... . ... . . -297 -

DGKIDVSIEAASGGKAEV1TCDVLLVCIGRRPFTKNU;LEELGIELDPRG- 3 0 0

*

Central Domain

L-protein

*

ASGSDF.NDWIIGGGPGGYVAAIKAAQLGFKTTCIEI~RGALGTCLNVGC -30 : . :: :::::::::::::::::: :::: .::::::::::

-347 - RILVNERFSTNVSGVYAIGDVIPGPNWKAEE~VACVEYLAGKVGHVD . ..... . ..... . . . ........ . . . ........... . . . ............ .

-

RIPVNTRFQTKIPNIYAIGDWAGPMLAHKAEDEGIICVEGMAGGAVHID -350

-

- YDKVPGWYTNPEVASVGKTEEQVKETGVEYRVGKFPFMANSRAKAIDNA -J97 : :: :.:: :::: :::.:::.:: :.::.:::::: ::::::. . - YNCVPSVIYTHPEVAWVGIEEQLKEEGIEYKVGKFPFAANSRAITTNADT -400 Interface Domain

L-protein Human LPD

- EGLVKTIAEKETDKILGWIMAPNAGELIHEAAIAIQYDASSEDIARVCH -447 ............................... ........................................ - DGIWKILGQKSTDRVLGAHILGPGAGEMVNEAALALEYGASCEDIARVCH -450

- ANPTMSEAIKEAA-MATYDKIHI . ... .. . ... . ... .. . ... . -470 c

L-protein Human LPD

-

AHPTLSEAFREANLAASFGKSINF -474

Fig. 7. Alignment of amino-acid sequencesof L-protein of pea-leaf mitochondria and human-liver dihydrolipoamide dehydrogenase. Human LPD sequence is from Otulakowski and Robinson (1987). The different domains (FAD, NAD’ ,central and interface domains) are located by the similarity of their amino acid sequences to human crythrocyte glutathione reductase (Krauth-Siege1 et al., 1982). The asterices indicate the redox-active cysteines.

Table 1. Percentage of amino-acid identity and similarity of pea mature L-protein with other dihydrolipoamide dehydrogenases. Dihydrolipoamide dehydrogenase

Identity Similarity

References

YO Human

56.8

70.2

Pig

56.4

70.0

Yeast P.,fluorexens A . vinelundii

56.2 46.8 45.7

70.7 60.4 60.6

E. coli P.putidu (LPD-vat)

40.4 37.3

55.9 52.6

Otulakowski and Robinson, 1987 Otulakowski and Robinson, 1987 Ross et al., 1988 Benen et al., 1989 Westphal and de Kok, 1988 Stephens et at., 1983 Burns et al., 198Ya

and 8). Likewise, two regions in the mature L-protein are very similar to the ‘fingerprint’ structural motif in the adeninebinding domain of either FAD (amino acid residues 13- 22) or NAD (amino acid residues 184-193, Fig. 7) of many enzymes, including all the dihydrolipoamide dehydrogenases ; and is manifested as a Pap fold centered around a highly conserved sequence Gly-Xaa-Gly-Xaa-Xaa-Ala/Gly-XaaXaa-Xaa-Ala/Gly (Scrutton et al., 1990; Wierenga et al., 1985). The presence of several clustered regions of high se-

872 which is only abundant in green-leaf mitochondria. Third, Southern-blot analysis indicated that the L-protein gene in pea Human LPD is most likely to be present as a single copy/haploid genome. Pig LPD Interestingly E. coli also contains a single diYeast LPD hydrolipoamide dehydrogenase for pyruvate and 2-0x0g k v a L G G T C L N V G C I P s k P. fluorescens LPD glutarate dehydrogenases encoded by the Ipd gene which is g k t a L G G T C L N V G C I P S k A . vinelandii LPD adjacent to the structural genes for the El (pyruvate dehydror y n t L G G m C L N V G C I P s k B . coli LPD genase; aceE) and E2 (dihydrolipoamide acetyltransferase; e q k a L G G T C L N m G C I P s k P. putida LPD aceF) components of pyruvate dehydrogenase (Stephens et al., 1983). Mutants of E. coliin the dihydrolipoamide dehydroHuman erythrocyte glutathione reductase genase structural gene lpd, lack both pyruvate and 2Fig. 8. Primary structures around the disulfide-active site in several oxoglutarate dehydrogenase (Alwine et al., 1973). Furtherdihydrolipoamide dehydrogenases and in human erythrocyte glutathione more, Spencer and Guest (1985) have shown that all three reductase. Human LPD and pig LPD from Otulakowski and genes (aceE, aceF and Ipd) can be transcribed in a single Robinson (1987); yeast LPD from Ross et al. (1988); P.fluorescens transcription unit from a promoter upstream from the first LPD from Benen et al. (1989); A. vinelundii LPD from Westphal and gene in the operon encoding pyruvate dehydrogenase. Howde Kok (1988); E. coli LPD from Stephens et al. (1983); P. putida ever, the lpd gene can be transcribed from its own promoter LPD-Val from Burns et al. (1989a); human erythrocyte glutathione separately from the other two genes, thus allowing excess reductase from Krauth-Siege1 et al. (1982). The asteriks indicate the production of dihydrolipoamide dehydrogenase which can be redox-active cysteines. used to complement the 2-oxoglutarate dehydrogenase complex. This also holds true for Saccharomyces cerevisiae. In that case, Dickinson et al. (1986) have shown that a single mutation quence similarity suggests that all these proteins, including L- affecting the dihydrolipoamide dehydrogenase gene (Ipdl) protein, have evolved from a common ancestor. abolishes the activity of both the 2-oxoglutarate and pyruvate The dihydrolipoamide dehydrogenase is also a component dehydrogenase complexes. S. cerevisiae is similar to E. coli in of the multienzyme composed of pyruvate, 2-oxoglutarate and having a single dihydrolipoamide dehydrogenase structural branched-chain 2-oxoacid dehydrogenase complexes (Reed, gene (Ipdl) which has been cloned and sequenced (Ross et al., 1974; Williams, 1976). In that case, the dihydrolipoamide 1988; Browning et al., 1988). In contrast, Pseudomonusputidu dehydrogenase is called E3. The results described in the present contain different lipoamide dehydrogenases, called LPD-Val paper strongly suggest that, in higher plants, a unique ( M , 49000) and LPD-Glc ( M , 56000) (Sokatch et al., 1981; dihydrolipoamide dehydrogenase is involved in different Sokatch and Burns, 1984). LPD-Val is the specific E3 mitochondria1 complexes. Confidence in this conclusion component of branched-chain oxoacid dehydrogenase, and comes from the following considerations. First, after fraction- LPD-Glc is the E3 component of 2-oxoglutarate and pyruvate ation of a matrix extract of pea-leaf mitochondria by gel- dehydrogenase and the L-factor of the glycine oxidation syspermeation chromatography and Western-blot analysis, it has tem (Burns et al., 1989a, Palmer et al., 1991). Burns et al. been shown that polyclonal antibodies raised against the L- (1989b) have described the isolation of a third lipoamide deprotein of the glycine-cleavage system recognized proteins of hydrogenase from P. putida. This enzyme, provisionally M , about 60000 in each elution peak associated with named LPD-3, differs in molecular mass, amino acid compodihydrolipoamide dehydrogenase activity (Fig. 1 C). Exper- sition and N-terminal amino acid sequence from LPD-Glc iments carried out by Walker and Oliver (1986a) indicate and LPD-Val. LPD-3 seems to be associated with an unknown that a monoclonal antibody against the L-protein produced multienzyme complex. In mammals, the situation is not yet indistinguishable inhibition profiles for both glycine and clarified. An interaction between the glycine-cleavage system pyruvate decarboxylase complexes. During the course of this and the three 2-oxoacid dehydrogenase complexes has been work, the activity of the pyruvate-dehydrogenase complex postulated. Pyruvate, 2-oxoglutarate and the branched-chain was also measured (result not shown) and this complex, which 2-oxoacids all inhibit the glycine-cleavage system (Kochi et has a very large size ( M r 5-6 x lo6; Randall and Miernyk, al., 1986). However, decarboxylation of the 2-oxoacids in the 1990), was eluted just after the void volume (Bourguignon et absence of NADH stimulates glycine synthesis. These authors al., 1988). Interestingly, the smallest peak ofdihydrolipoamide speculated, therefore, that this effect is produced by the prodehydrogenase activity was associated with pyruvate dehydro- vision of reducing equivalents to the glycine-cleavage system, genase. It is worthwhile to note that during the fractionation possibly through a shared common E3. Otulakowski and by gel filtration, the pyruvate dehydrogenase does not exhibit Robinson (1987) have published the sequence of the human the stability previously reported (Randall and Miernyk, 1990) dihydrolipoamide dehydrogenase as being the sequence of the because addition of L-protein to the fraction containing L-protein of the glycine cleavage system and E3 of the 2pyruvate dehydrogenase activity stimulates the activity of oxoacid dehydrogenase complexes. However, there is pyruvate decarboxylation (data not shown). Secondly, North- immunological evidence for two dihydrolipoamide dehydroern-blot analysis of RNA from different tissues such as leaf, genases in rat-liver mitochondria (Carothers et al., 1987), one stem, root and seed using L-protein cDNA as a probe, indi- of which may be the L-protein of the glycine cleavage system. The role of dihydrolipoamide dehydrogenase in several cates that the mRNA of the dihydrolipoamide dehydrogenase accumulates at high level in all the tissues (Fig. 4). In contrast, multienzyme complexes in higher plants raises some interestthe H-protein (a specific protein component of the glycine- ing questions concerning the regulation of synthesis and concleavage system) mRNA level was found to be negligible in trol of distribution of a unique enzyme associated with differthe root, stem and seed tissues (Macherel et al., 1990). Such ent complexes. Cloning of the gene is needed to define the results strongly suggest, that in higher plant mitochondria, primary control of dihydrolipoamide dehydrogenase gene exlipoamide dehydrogenase is probably involved in some other pression. In this manner, it will eventually be possible to protein complexes distinct from the glycine-cleavage system understand precisely the mechanisms involved in the assembly Pea L-protein

873 of protein complexes containing dihydrolipoamide dehydrogenase. Furthermore, it would be possible to understand how light exerts an additional effect by increasing the level of transcripts in leaves, allowing the synthesis of dihydrolipoamide dehydrogenase involved in glycine oxidation.

REFERENCES Alwine, J. C., Russel, F. M. & Murray K. N. (1973) J . Bacteriol. 115, 1-8.

Benen, J . A. E., Van Berkel, W. J. H., Van Dongen, W. M. A. M., Miiller, F. & de Kok, A. (1989) J . Gen. Microbiol. 135, 17871797.

Birnboim, H. C. & Doly, J. (1979) Nucleic Acids Res. 7, 1513-1523. Bourguignon, J., Neuburger, M. & Douce, R. (1988) Biochem. J . 255, 169- 178.

Browning, K. S., Uhlinger, D. J. & Reed, L. J. (1988) Proc. Natl Acad. Sci. USA 85,1831 - 1834. Burns, G . , Brown, T., Hatter, K. & Sokatch, J. R. (1989a) Eur. J . Biochem. 179, 61 -69. Burns, G., Sykes, P.J., Hatter, K. &Sokatch, J. R. (1989b) J . Bacteriol. 171,665-668.

Carothers, D. J., Pons, G. & Patel, M. S. (1989) Arch. Biochem. Biophys. 268,409 -425. Carothers, D. I., Raefsky-Estrin, C., Pons, G. & Patel, M. S. (1987) Arch. Biochem. Biophys. 256, 597-605. Dickinson, J. R., Roy, D. J . & Dawes, I. W. (1986) J. Gen. Genet. 204, 103 - 107. Douce, R., Bourguignon, J., Brouquisse, R. & Neuburger, M. (1987) Methods Enzymol. 148,403-425. Douce, R., Moore, A. L. & Neuburger, M. (1977) Plant Physiol. 60, 625 - 628. Hartl, F., Pfanner, N., Nicholson, D. W. & Neupert, W. (1989) Biochim. Biophys. Acta 988, 1-45. Kurkman, W. J . &Tanaka, C . K. (1986) Plant Physiol. 82,802-806. Husic, D. W., Husic, H. D. & Tolbert, N. E. (1987) Crit. Rev. Plant Sci. 5 , 45- 100. Innis, M. A. & Gelfand, D. H. (1990) in PCRprotocofs (Innis, M. A., Gelfand, D. H., Sninsky, J. J. & White, T. J., eds) pp. 3-12, Academic Press, Inc. Joshi, C. P. (1987) Mucleic Acids Res. 15, 9627-9640. Kikuchi, H. & Hiraga, K. (1982) Mol. Cell. Biochem. 45, 137- 149. Klein, S. M. & Sagers, R. D. (1966) J. Biol. Chem. 241, 197-205. Klein, S. M. & Sagers, R. D. (1967) J. Biol. Chem. 242,297-300. Kochi, H. & Kikuchi, G. (1976) Arch. Biochem. Biophys. 173, 71 81.

Kochi, H., Seino, H. & Ono, K. (1986) Arch. Biochem. Biophys. 249, 263 -272.

Krauth-Siegel, R. L., Blatterspiel, R., Saleh, M., Schiltz, E., Schirmer, R. H. & Untucht-Grau R. (1982) Eur. J . Biochem. 121,259-267. Laemmli, U. K . (2970) Mature (Lond.) 227, 680-685. Lorimer, G. H. & Andrews, T. J. (1981) in The biochemistry ofplants, a comprehensive treatise: photosynthesis (Hatch, M. D. & Boardman, N. K., eds) vol. 8, pp. 329- 374, Academic Press, New York.

Liitcke, H. A,, Chow, K. C., Mickel, F. S., Moss, K. A., Kern, H. F. & Scheele G. A. (1987) EMBO J . 6,43 -48. Macherel, D., Lebrun. M., Gagnon, J., Neuburger, M. & Douce, R . (1990) Biochem. 1.268,783-789. Maniatis, T., Fritsh, E. F. & Sambrook (1982) Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY. Myers, E. W. & Miller, W. (1988) CABZOS 4 , 11- 17. Neuburger, M., Bourguignon, J. & Douce, R. (1986) FEBS Lett. 207, 18-22.

Oliver, D. J., Neuburger, M., Bourguignon, J. & Douce, R. (1990a) Plant Physiol. 94, 833 - 839. Oliver, D. J., Neuburger, M., Bourguignon, .I. & Douce, R. (1990b) Physiol. Plant. 80,487 -491. Otulakowski, G. & Robinson, B. H. (1987) J . Biol. Chem. 262, 17313- 17318.

Pai, E. F. & Schulz, G. E. (1983) J. Biol. Chem. 258, 1752-1757. Palmer, J. A., Hatter, K. & Sokatch, J. R. (1991) J . Bacteriol. 173, 3109 - 3116.

Randall, D. D. & Miernyk, J. A. (1990) in Methods inplant biochemistry (Dey, P. M. & Harborne J. B., eds) vol. 3, pp. 175- 192, Academic Press. Reed, L. J. (1974) Ace. Chem. Res. 7,40-46. Ross, J., Reid, G. A. & Dawes I. W. (1988) J . Gen. Microbiol. 134, 1131 -1139.

Saint-Blancard, J., Fourcart, J., Limonne, F., Girot, P. & Boschetti (1981) Ann. Pharm. Fr. 39,403-409. Sanger, F., Nicklen, S. & Coulson, A. (1977) Proc. Mat1 Acad. Sci. USA 74, 5463 - 5467. Scrutton, N. S., Berry, A. & Perham, R. N. (1990) Nature (Lond.) 343, 38-43.

Sokatch, J. R. & Burns, G. (1984) Arch. Biochem. Biophys. 228,660666.

Sokatch, J. R., McCully, V., Gebrosky, J. & Sokatch, D. J. (1981) J . Bacteriol. 148,639 - 646. Spencer, M. E. & Guest, J. R. (1985) Mol. Gen. Genet. 200,145 - 154. Stephens, P. E., Lewis, H. M., Darlison, M. G. & Guest, J. R. (1983) Eur. J . Biochem. 135,519-527. Untucht-Grau, R., Schirmer, R. H., Schirmer, I. & Krauth-Siegel, R. L. (1981) Eur. J . Biochem. 120,407-419. Von Heijne, G., Steppuhn, J. & Herrmann, R. G. (1989) Eur. J. Biochem. 180, 535-545. Walker, J. L. &Oliver, D. J. (1986a) J. Biol. Chem. 261, 2214-2221. Walker, J. L. & Oliver, D. J. (1986b) Arch. Biochem. Biophys. 248, 626 - 638. Watson, J. C. & Thompson, W. F. (1986) Methods Enzymol. 118, 57 - 75. Westphal, A. H. & de Kok, A. (1988) Eur. J. Biochem. 272,299-305. Wierenga, R. K.,De Maeyer, M. C. H. &Hol, W. G . (1985) Biochemistry 24, 1346- 1357. Williams, C . H. Jr. (1976) in The enzymes (Boyer, P. D., ed.) vol. 13, pp. 89- 173, Academic Press, New York. Williams, C. J. Jr., Arscott, L. D. & Schulz, G. E. (1982) Proc. Nut/ Acad. Sci. USA 79,2199 - 2201. Young, R. A. & Davis, R. W. (1983) Proc. Nail Acad. Sci. USA 80. 1194 - 1198.