The glycine decarboxylase system: a fascinating complex

Review 49 Lerouge, P. et al. (1998) N-glycoprotein biosynthesis in plants: recent developments and future trends. Plant Mol. Biol. 38, 31–48 50 von S...
Author: Nancy Perkins
5 downloads 0 Views 458KB Size
Review

49 Lerouge, P. et al. (1998) N-glycoprotein biosynthesis in plants: recent developments and future trends. Plant Mol. Biol. 38, 31–48 50 von Schaewen, A. et al. (1993) Isolation of a mutant Arabidopsis plant that lacks N-acetyl glucosaminyl transferase I and is unable to synthesize Golgi-modified complex N-linked glycans. Plant Physiol. 102, 1109–1118 51 Strasser, R. et al. (1999) Molecular cloning and characterization of cDNA coding for β1,2N-acetylglucosaminyltransferase I (GlcNAcTI) from Nicotiana tabacum. Glycobiology 9, 779–785 52 Strasser, R. et al. (2000) Molecular cloning and functional expression of β1,2-xylosyltransferase cDNA from Arabidopsis thaliana. FEBS Lett. 472, 105–108

TRENDS in Plant Science Vol.6 No.4 April 2001

53 Strasser, R. et al. Molecular cloning of cDNA encoding N-acetylglucosaminyltransferase II from Arabidopsis thaliana. Glycoconjugate J. (in press) 54 Essl, D. et al. (1999) The N-terminal 77 amino acids from tobacco Nacetylglucosaminyltransferase I are sufficient to retain a reporter protein in the Golgi apparatus of Nicotiana benthamina cells. FEBS Lett. 453, 169–173 55 Carpita, N. and McCann, M. (2000) The Cell Wall. In Biochemistry and Molecular Biology of Plants (Buchanan, B.B. et al., eds), pp. 52–109, American Society of Plant Physiologists 56 Levy, S. et al. (1991) Simulations of the static and dynamic molecular conformations of xylogucan. The role of the fucosylated sidechain in surfacespecific sidechain folding. Plant J. 1, 195–215

167

57 Faik, A. et al. (2000) Biochemical characterization and molecular cloning of an α-1,2fucosyltransferase that catalyzes the last step of cell wall xyloglucan biosynthesis in pea. J. Biol. Chem. 275, 15082–15089 58 Perrin, R.M. et al. (1999) Xyloglucan fucosyltransferase, an enzyme involved in plant cell wall biosynthesis. Science 284, 1976–1979 59 Edwards, M.E. et al. (1999) Molecular characterization of a membrane-bound galactosyltransferase of plant cell wall matrix polysaccharide biosynthesis. Plant J. 19, 681–697 60 Wulff, C. et al. (2000) GDP-fucose uptake into the Golgi apparatus during xyloglucan biosynthesis requires the activity of a transporter-like protein other than the UDP-glucose transporter. Plant Physiol. 122, 867–878

The glycine decarboxylase system: a fascinating complex Roland Douce, Jacques Bourguignon, Michel Neuburger and Fabrice Rébeillé The mitochondrial glycine decarboxylase multienzyme system, connected to serine hydroxymethyltransferase through a soluble pool of tetrahydrofolate, consists of four different component enzymes, the P-, H- , T- and L-proteins. In a multi-step reaction, it catalyses the rapid destruction of glycine molecules flooding out of the peroxisomes during the course of photorespiration. In green leaves, this multienzyme system is present at tremendously high concentrations within the mitochondrial matrix. The structure, mechanism and biogenesis of glycine decarboxylase are discussed. In the catalytic cycle of glycine decarboxylase, emphasis is given to the lipoate-dependent H-protein that plays a pivotal role, acting as a mobile substrate that commutes successively between the other three proteins. Plant mitochondria possess all the necessary enzymatic equipment for de novo synthesis of tetrahydrofolate and lipoic acid, serving as cofactors for glycine decarboxylase and serine hydroxymethyltransferase functioning.

Roland Douce* Jacques Bourguignon Michel Neuburger Fabrice Rébeillé Département de Biologie Moléculaire et Structurale, Physiologie cellulaire végétale, CEA Grenoble, CNRS et Université Joseph Fourier, 17 rue des martyrs, F 38054 Grenoble, Cedex 9, France. *e-mail: [email protected]

The prime function of the oxidative photosynthetic carbon cycle or C2 cycle – inappropriately named ‘photorespiration’ – is to salvage the glycolate-2-P that is produced continuously in the light by the oxygenase activity of ribulose-1,5-bisphosphate carboxylase/oxygenase (Rubisco)1–3. The C2 cycle initiated by Rubisco4 requires a large machinery consisting of 16 enzymes and more than five translocators, distributed over the chloroplast, peroxisome and mitochondrion, in close proximity to each other3. The most interesting reaction in the C2 cycle occurs within the mitochondrial matrix, where glycine molecules formed in the peroxisomes are broken down by a set of proteins (glycine decarboxylase, GDC). By acting in concert, these proteins catalyse the oxidative decarboxylation and deamination of glycine, with the formation of CO2, NH3 and the

concomitant reduction of NAD+ to NADH (Refs 5–8). The remaining carbon, the methylene carbon of glycine, is then transferred to 5,6,7,8tetrahydropteroylpolyglutamate (tetrahydrofolate, THF) to form N 5,N 10-methylene-5,6,7,8tetrahydropteroylpolyglutamate (CH2–THF). CH2–THF reacts with a second molecule of glycine in a reaction catalysed by serine hydroxymethyltransferase (SHMT) to form serine. The net reaction catalysed is (Eqn 1): 2 Glycine + NAD+ + H2O → Serine + CO2 + NH3 + NADH + H+

[1]

Rates of glycine metabolism by isolated leaf mitochondria from C3 plants can exceed 1200 nmol of glycine converted to serine per mg of protein, per minute. The rate of CO2 release from glycine decarboxylation is as much as five times the rate of normal tricarboxylic acid cycle activity. To accomplish these rapid rates of glycine oxidation, the glycine cleavage system and SHMT are present at tremendously high concentrations within the mitochondrial matrix, where they comprise about half of the proteins in mitochondria from pea or spinach leaves9. With matrix concentrations approaching 0.3 g/ml, it actually alters the density of the organelles10. Conversely, in mitochondria from nongreen tissues, such as potato tubers, glycine decarboxylase is present at low concentrations11. From a mere trickle in the dark reflecting C1 metabolism, glycine metabolism becomes the major metabolic reaction in mitochondria of illuminated leaves5.

http://plants.trends.com 1360-1385/01/$ – see front matter © 2001 Elsevier Science Ltd. All rights reserved. PII: S1360-1385(01)01892-1

Review

168

TRENDS in Plant Science Vol.6 No.4 April 2001

Glycine COO– S

CH2

Hox

NH3+

S

NADH

H

P

L

CO2

NAD+

Hred

S-CH2–NH3+ Hmet

H

SH NH3

H

SH

SH

T CH2 −THF

THF

Serine

SHMT

Glycine H2O TRENDS in Plant Science

Fig. 1. Outline of the reactions involved in oxidative decarboxylation and deamination of glycine in plant mitochondria. P-, H-, T- and L- are the protein components of the glycine-decarboxylase multienzyme system. The pivotal enzyme in the entire sequence of reactions is the 14 000 Mr lipoamide-containing H-protein, which undergoes a cycle of reductive methylamination (catalysed by the P-protein), methylamine transfer (catalysed by the T-protein) and electron transfer (catalysed by the L-protein). The lipoyl moiety in the H-protein is attached by an amide linkage to the ε-amino group of a lysine residue. This linkage provides a rather flexible arm, ~14 Å in length, conveying the reactive dithiolane ring from one catalytic centre to another. SHMT: serine hydroxymethyltransferase involved in the conversion of CH2–THF to THF at the expense of a second molecule of glycine. Note that the methylamine moiety deriving from glycine is passed to the distal sulfur of the dithiolane ring. Hmet, Hred and Hox: methylaminated, reduced and oxidized forms of the H protein, respectively.

Structure of the glycine decarboxylase complex

The glycine decarboxylase multi-enzyme complex catalyses in a multi-step reaction the rapid ‘cracking’ of glycine molecules flooding out of the peroxisomes during the course of photorespiration (Fig. 1). As in its mammalian counterpart13, the glycine decarboxylase multienzyme complex of green-leaf mitochondria consists of four different component enzymes, designated as the P-protein (a homodimer containing pyridoxal phosphate, 200 kDa), the H-protein (a monomeric lipoamide-containing protein, 14 kDa), the T-protein (a monomeric protein requiring THF cofactor, 41 kDa), and the L-protein, a dihydrolipoamide dehydrogenase (a homodimer containing FAD and a redox active cysteine residue, 100 kDa)6,8. The primary sequence of all these proteins from different sources, including pea, were determined by cDNA and genomic cloning. Precise measurements (using ELISA) of the amount of each component protein within the matrix of pea mitochondria indicate a subunit stoichiometry of 4 P-protein:27 H-protein:9 T-protein:2 L-protein (the matrix extract contained ~9.5% P-protein, http://plants.trends.com

~9% H-protein, ~10% T-protein and ~3.5% L-protein)9. Following mitochondrial inner-membrane rupture after several cycles of freezing and thawing14 or precipitation of the soluble enzymes by cold acetone15, all the components of the glycine decarboxylase system dissociate easily and behave as independent proteins. The overall activity of the glycine decarboxylase complex, as well as each catalytic step, can thus be recovered in vitro by association of the purified proteins, permitting the study of the different interaction mechanisms between the various partners14,15. However, once isolated the free T-protein exhibits a strong propensity to form high-molecular mass aggregates and is unstable in solution. Formation of these aggregates is prevented in the presence of H-protein16. Under these conditions, one molecule of T-protein associates with one molecule of H-protein to form a complex characterized by small angle X-ray scattering17. In the catalytic cycle of the glycine decarboxylase, the lipoate-dependent H-protein plays a pivotal role, acting as a mobile substrate (its final concentration in the matrix space reaches millimolar concentrations when plants are grown in light conditions) that commutes successively between the other three proteins (Fig. 1). During the course of this catalytic cycle the lipoate group binds covalently to a lysine through an amide linkage, making a 14 Å long lipoamide arm, and appears to ‘visit’ all three active sites. It undergoes a cycle of reductive methylamination (catalysed by the P-protein), methylamine transfer (catalysed by the T-protein) and electron transfer (catalysed by the L-protein). Thus during catalysis the lipoate arm is successively methylamined, reduced and oxidized. A synthetic gene encoding the entire mature H-protein from pea has been constructed and expressed in E. coli18. Activity measurements and mass-spectrometry analysis of the purified recombinant protein have shown that in the presence or absence of lipoic acid in the culture medium, recombinant H-protein can be produced on a large scale as the unlipoylated apoform, or as the lipoylated form, respectively. The X-ray crystal structure of all forms of the Hprotein (apoform; oxidized, Hox; methylaminated, Hmet; and reduced, Hred) have been determined18–21. The core of all these structures consists of two antiparallel β-sheets forming a hybrid barrel-sandwich structure, one with four strands and one with three strands and two adjacent antiparallel strands joined by a loop (hairpin β-motif) (Fig. 2a). The lipoylated lysine residue (in pea Lys63) is situated at the tip of this loop. Confirmation of the proper lipoic acid position was achieved by combined liquid chromatography and tandem mass spectrometry analyses of the peptides obtained after digestion of Hprotein with endoproteinases22. As in the oxidized form, the reduced lipoate arm is localized at the surface of the protein and is able to move in the solvent (Fig. 2a,b). By contrast, it is locked in the

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

169

Fig. 2. Structure of the H- and L-protein from the pea-leaf glycine decarboxylase system. (a) Schematic ribbon representation of the overall folding of the H-protein (oxidized form, Hox). Note that the conformation of the core of the H-protein consists of seven β-strands in a sandwich structure made of two antiparallel β-sheets and that the lipoate cofactor is attached by an amide linkage to a lysine side chain located in the loop of a hairpin configuration. (Figure produced using ‘Molscript’61); (b) and (c) Surface representation of (b) the reduced (Hred) and (c) methylaminated (Hmet) forms of H-protein. Note that when the lipoamide is charged with methylamine it rotates like an arm to come into contact with hydrophobic residues of a cavity well opened at the surface of the H-protein (c). The methylaminated lipoate arm is also locked and stabilized by hydrogen bonds between the methylamine group and the protein. By contrast, as in the oxidized form (a), the reduced lipoate arm (b) appeared freely exposed to the solvent. This freedom is required to allow its targeting inside the hollow active site of L-protein, see (e). (Figure produced using ‘O’62). (d) Schematic ribbon representation of the overall folding of the Lprotein dimer (monomer a in shades of red; monomer b in shades of blue). The different domains of the protein are the FAD domain (lightviolet, monomer b), the NAD binding domain (blue, monomer b), the central domain (blue–green, monomer b) and the interface domain (dark violet, monomer b). The two FAD molecules are represented as yellow spheres. (Figure produced using ‘Molscript’61); (e) Potential surface of the L-protein and the entrance of the dihydrolipoamide binding site (arrow), see (b). The surface is coloured according to the charges: dark-violet indicates positively charged surfaces, red indicates negatively charged surfaces. Yellow–green indicates hydrophobic surfaces. (Figure produced using ‘Grasp’63).

methylaminated form and stabilized in a cavity at the surface of the protein (Fig. 2c). The reaction catalysed by GDC commences with the amino group of glycine forming a Schiff base with the pyridoxal phosphate of the P-protein. The carboxyl group of glycine is lost as CO2 and the remaining methylamine moiety is passed onto the distal sulfur of the lipoamide cofactor of the H-protein19 (Fig. 1). The rapid methylamination of the H-protein (Hmet) catalysed by the P-protein is half-saturated at micromolar concentrations of H-protein Km H-protein~9 µM; Vmax~5 µmol/mg protein per min)23. http://plants.trends.com

A structural domain of the H-protein is recognized by the P-protein because the analogues of H-protein (apoform, Hred and Hmet) behave as competitive inhibitors, increasing the apparent Km for the H-protein. Compounds such as carboxymethoxylamine24 that react with the pyridoxal phosphate of the P-protein strongly inhibit GDC. During the course of the reductive methylamination catalysed by the P-protein, the lipoamide arm loaded with methylamine readily rotates to interact with several specific amino acid residues located within a cleft at the surface of the H-protein. The NH3+ moiety of the methylamine group is tightly linked by an ionic bond to the lateral chain carbonyl group of Glu14 (pea), and by hydrogen bonds to the carbonyl group of Ser12 and Asp67. The carbon atoms of the lipoamide arm interact through van der Waals contacts with hydrophobic residues19. This locks (the ‘snapfastener system’) the methylamine group into a highly stable conformation, preventing the non-enzymatic release of NH3 and formaldehyde, which would otherwise take place as a result of nucleophilic attack by OH− of the carbon atom bearing the NH3+ group (Fig. 3a), until the reaction with tetrahydrofolate and T-protein takes place. At 30°C and at pH 7 the full demethylamination of Hmet in the absence of T-protein requires several days. The replacement of Glu14 by Ala (site-directed mutagenesis) within the hydrophobic cleft weakens the snap-fastener system25. Indeed, the complete methylamine loading of this mutant in the presence of glycine and P-protein, in contrast to the wild type, can never be reached because it cannot properly protect the methylamine group against uncontrolled nucleophilic attacks by OH−, leading to the continuous production of formaldehyde. Molecular dynamic calculations show that the presence of THF alone does not allow direct transfer of the methylene group, suggesting that the

170

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

(a) H

H

HN

O

NH3

HS 3

CH2 S –

(b)

H

HN

HN

O

+NH

H HN O

HCHO

HS

H+

H O CH2 S

O

HS

HS

–S

HS

OH

THF (H4FGlun) 5

CH2H4FGlun

H

N

10

NH R

H

H

+

N CH2 S

CH2 S

HS

HS

HS

NH R H2 C H+

NH3

NH H

O

NH

O

H

N

NH R

H

NH H

H

N

N NH R

H H3+N CH2 S

H

H

+

O

H 2C S

HS

HS

HS

NH H

O

NH

N

R

O

H

TRENDS in Plant Science

Fig. 3. Proposed model for the reaction catalysed by the T-protein. (a) In the absence of 5,6,7,8-tetrahydropteroylpolyglutamate (THF) in the incubation medium, the T-protein causes a change in the overall conformation of the H-protein (methylaminated form, Figs 1,2), leading to the release of the lipoamide methylamine arm from the cleft at the surface of the H-protein (Fig. 2c). This situation favours nucleophilic attack by OH− of the carbon atom bearing the NH3+ group. Such an attack leads to a slow release of NH3 and formaldehyde, which is accompanied by the full reduction of the lipoamide arm. (b) In the presence of THF (H4FGIun), the methylamine group undergoes a preferential nucleophilic attack by the N-5 atom of the pterin ring of THF. CH2–THF is therefore rapidly formed in place of formaldehyde, concomitantly with the reduction of the lipoamide arm and the release of NH3.

reactive nucleophilic N5 (or N10 ) atom of the pteridine ring could not sufficiently approach the C atom of the methylamine group16,26. It appears, therefore, that the network of hydrogen bonding and hydrophobic contacts stabilizing the methylamine group of the lipoamide arm must be ruptured to allow an interaction between the C atom of the methylamine group and the THF nitrogen. In fact, differences in the 1H and 15N chemical shifts (observed with NMR spectroscopy) of the H-protein in its isolated form and in the complex with the T-protein show that T-protein induces a small structural perturbation of Hmet: the interaction surface on the H-protein is localized on one side (β-sheet structure) of the cleft where the lipoate arm is positioned16. Although the rate of demethylamination (with the concomitant formation of Hred, formaldehyde and http://plants.trends.com

NH3) (Fig. 3a) is rapid in the presence of stoichiometric amounts of T-protein, it is far below that observed in the additional presence of THF (with the concomitant formation of Hred, CH2–THF and NH3) (Fig. 3b). This might be explained by the low nucleophilicity of OH−, whose concentration at pH 8 is close to 1 µM. At this concentration the THF, which binds the T-protein27, accelerates the reaction by a factor of nearly 104. Only the 6 S stereoisomer of THF, the naturally occurring form, is the substrate for this reaction in vitro28. It is clear, therefore, that the T-protein has a double function in the reaction mechanism: first, in modifying the stability of the Hmet, leading to a release of the methylamine moiety, and second in locating the THF in a position favourable to a nucleophilic attack by the methylene carbon. As the T-protein interacts with all the different forms of H-protein in a one-to-one molar ratio it would therefore be of interest to study the dependence of the reaction kinetics on the respective concentrations of Hmet, Hox and Hred and their different association constants. From a mechanistic point of view, the observed competitive binding of Hox, Hmet and Hred for T-protein must be crucial in vivo. The dihydrolipoamide dehydrogenase (L-protein) catalysing the last step of the GDC cycle has a redox disulfide and a tightly, but not covalently, bound FAD cofactor, both of which participate in the rapid electron transfer from dihydrolipoamide to NAD+. This enzyme

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

belongs to the disulfide oxidoreductase family whose redox centre is formed by a disulfide bridge coupled to a flavin ring29–31. It has been shown in pea-leaf mitochondria that the L-protein is shared by several complexes (GDC and 2-oxoacid, dehydrogenases, a family of multienzyme complexes that catalyse the irreversible oxidative decarboxylation of 2-oxoacids including pyruvate and 2-oxoglutarate)32–34. However, it seems that the chloroplastic pyruvate dehydrogenase complex involved in acetyl-CoA formation during lipid synthesis35 has its own distinct dihydrolipoamide dehydrogenase isoform, different to that of its mitochondrial counterpart36. The crystal structure of L-protein from pea mitochondria has been determined by X-ray crystallography21 (Fig. 2d) and is similar to the other dihydrolipoamide dehydrogenases. The functional molecule is a dimer with an internal twofold axis connecting the two subunits. Each monomer is composed of four domains: the FADbinding domain (residues 1–147), the NAD-binding domain (148–279), the central domain (280–348) and the interface domain (349–470) (Fig. 2d). The catalytic site includes some residues in the interface domain (this could explain why a dimer association is necessary for the activity of the protein) and consists of the isoalloxazine ring of FAD in an extended conformation, a disulfide bridge (residues 45 and 50) in one monomer and the residue His449 in the other. The isoalloxazine ring separates the binding site of the dihydrolipoamide from the binding site of NAD+. The binding site of the dihydrolipoamide associated with Hred forms an 11 Å-deep canal with a narrow aperture facing the solvent (Fig. 2e). Inside the tunnel in the vicinity of FAD, the residues including Cys45 and Cys50 (monomer A) and His449 (monomer B) are mainly polar or charged. His449 interacts with the disulfide bridge during the reaction. By contrast, the region of the cavity accessible to the solvent forms an uncharged crevice with several hydrophobic residues that can interact with the aliphatic chain of the dihydrolipoamide21. The position of the dihydrolipoate arm at the surface of Hred, in a flexible conformation21 (Fig. 2b), allows it to rotate through 90° and penetrate the hydrophobic canal up to the active site to discharge the electrons on the cysteine bridge. In this case, there would be no strong interaction between the L- and H-proteins. In support of this suggestion, the oxidation of Hred is not affected by the presence of structurally related analogues (Hapoform; Hmet)23. In addition, free hydrophilic lipoate analogues as well as proteolytic fragments of the H-protein containing the lipoamide moiety in their reduced forms have Km values for the L-protein close to that found for Hred. Conversely, the L-protein has a much lower affinity for the free hydrophobic lipoamide than for Hred (Km values 250 versus 27 µM). Therefore, during the course of Hred oxidation, the main function of the H-protein is to maintain a high concentration of the hydrophobic lipoate molecules in a non-micellar state that would http://plants.trends.com

171

be accessible to the catalytic site of the dihydrolipoamide dehydrogenase. Glycine decarboxylase-serine hydroxymethyl transferase coupled reactions

The glycine decarboxylase complex is connected to SHMT through a soluble pool of THF (Fig. 1). SHMT in pea leaf mitochondria is a 220 kDa homotetramer with a subunit molecular weight of 53 kDa (Refs 14,37). Each subunit contains a single pyridoxal-phosphate bound as a Schiff base to an ε-amino group of a lysine residue. If the C2-cycle is strictly cyclic, then all CH2–THF formed upon glycine oxidation must be used for serine synthesis and does not gain access to the general C1 pool. SHMT is present in the three main cellular compartments (cytosol, mitochondria and chloroplast)38. In fact, it is the cytosolic form that plays a key role in the source of C1 units for the synthesis of methionine, which is in turn, via S-adenosyl methionine formation, the donor of methyl groups in numerous reactions11. In other words, cytosolic CH2–THF is mainly directed towards generating methyl groups, whereas CH2–THF generated from glycine in mitochondria is directed more towards serine formation. Analysis of plant mitochondrial THF has revealed a heterogeneous population of molecules that differ from one another in their glutamate chain lengths27,28. This pool of polyglutamates polyglutamates (~1 nmol/mg matrix protein) is dominated by tetra and pentaglutamates, which account for ~25% and 55% of the total pool, respectively. This chain could have a key role in controlling the velocity and coordination of the two reactions catalysed by GDC and SHMT. Indeed, glutamate chain-length influences the affinity constant for THF displayed by SHMT and T-protein28. The polyglutamates of leaf mitochondria are probably bound loosely to the active sites of T-protein and tetrameric SHMT (Ref. 27). THF binding to folate-dependent proteins contributes significantly to the protection of this readily oxidizable cofactor14,27. A comparison of the known primary amino acid sequence of the T-protein39 with that of SHMT (Ref. 37) did not reveal conserved folylpolyglutamatebinding consensus sequences. Each folate-dependent enzyme might have its own specific folate-binding site, or the binding sites might rely on secondary or tertiary structures. The recycling of THF, necessary to sustain glycine oxidation during photorespiration, implies that the mitochondrial SHMT reaction must be permanently pushed out of equilibrium towards the production of serine and THF (i.e. in the thermodynamically unfavourable direction)14,28, to allow the whole process to take place. This is apparently the case because in an in vitro system containing soluble enzymes from leaf mitochondria containing GDC and SHMT, the high CH2–THF:THF ratio observed during the steady-state course of glycine oxidation strongly supports the idea that GDC activity overwhelms SHMT activity27. This last observation also suggests that during the course

172

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

of steady-state glycine oxidation, the folate cofactor was not channeled between the T-protein and SHMT (Ref. 27). However, the in vivo situation might be different because the high concentrations of T-protein and SHMT in the matrix space could lead to a situation in which the diffusional pathway of folate compounds is considerably reduced12. Metabolic control of glycine oxidation and substrate transport

Metabolic regulation of the glycine decarboxylase system might be affected by feedback inhibition by two of the final reaction products, serine and NADH (Refs 6,40). Serine inhibits in a competitive manner with glycine [Ki(serine) = 4 mM; Km(glycine) = 6 mM] the reaction catalysed by the P-protein41. NADH competes with NAD+ for the active site on the L-protein [Km(NAD+) = 75 µM; Ki(NADH) = 15 µM]14. This means that increasing the ratio of NADH to NAD+ in the matrix space should result in a logarithmic increase in inhibition of GDC. Serine and NADH inhibition is prevented by the unique substrate transporters found in plant mitochondria. The competition between serine and glycine with P-protein is prevented by a glycine–serine exchange reaction. The rate of glycine oxidation demands that green-leaf mitochondria support a phenomenal rate of glycine transport. Although inhibitor studies with isolated mitochondria show the existence of this transporter, the details of glycine and serine transport in green-leaf mitochondria remain unknown. Thus, the question of whether both glycine and serine are transported by a single protein or two different ones cannot be answered yet. Competition between NAD+ and NADH with L-protein is prevented by a powerful phthalonatesensitive oxaloacetate carrier42–44 catalysing a rapid malate–oxaloacetate exchange reaction. This transporter is half saturated at micromolar concentrations of oxaloacetate and has a high Vmax (at least 2000 nmol min−1 per mg protein). When combined with NAD+-linked malate dehydrogenase in the mitochondria and peroxisomes, this transporter can potentially export NADH from the matrix with great efficacy for use in the reduction of hydroxypyruvate in the peroxisomes3. The purification and functional reconstitution, as well as the completion of detailed kinetic analyses, of this specific oxaloacetate transporter should be undertaken. At the molecular level, we do not currently know how any of these transporters involved in the photorespiratory cycle function in establishing the co-ordinated function of the C3–C2-cycle for maximum efficiency. Glycine decarboxylase biogenesis

Mitochondrial SHMT (Ref. 37) and all four proteins of the GDC complex are encoded by nuclear genes32,33,39,45–49. The proteins of GDC are synthesized most abundantly in leaves, and at detectable but much lower levels in non-green tissues. The http://plants.trends.com

spectacular accumulation of glycine decarboxylase proteins in mitochondria from mesophyll cells of C3 plant leaves is probably attributable to a lightdependent transcriptional control of the genes encoding these proteins. Studies of transgenic Arabidopsis have shown that a region between –376 and –117 bp of the H-protein gene, containing GT boxes, directed the light-induced expression of a reporter gene50. Similarly, the gene encoding the T-protein ( gdcT ) carries in its upstream region cisacting elements known to direct the light activation of photosynthetic genes such as rbcS or cab (Ref. 49). The authors also suggest that more control elements might be responsible for the constitutive low levels of gene expression noted in all non-photosynthetic tissues. In support of this suggestion, two main transcription sites have been detected in the gene encoding H-protein ( gdcH )51. Alternative splicing leads to the synthesis of two different H-proteins in all advanced Flaveria C4 species52. However, the expression of the gene encoding the L-protein appears to be regulated in a different manner than that of the three other proteins (P-, T- and H-) of the GDC (Refs 10,53). This difference is explained by the fact that the same isoform of L-protein is shared between different multienzyme complexes. In monocots, the P-, T- and H-proteins of glycine decarboxylase accumulate along the developmental gradient from the young tissue at the leaf base to the older tissue at the tip54. Photosynthetic and photorespiratory enzymes, including GDC, are both located in the matured region of the leaf and follow the same gradient of differentiation and a similar spatial expression. However, during the course of pea (dicot) leaf development, the appearance of glycine oxidation capacity seems to be closely related to the opening of the leaflets, which occurs after the sevenday stage (L3-stage), an event that allows the leaf to function as a morphologically efficient solar captor (Fig. 4). This coincides with the dramatic accumulation of the GDC proteins in the (preexisting?) mitochondria of mesophyll cells, producing a marked increase in their density10. There is a pronounced delay between the appearance of Rubisco and glycine decarboxylase, possibly because of a translational control that is lifted when the leaflets open (Fig. 4). Indeed, although the mRNAs for both enzymes follow similar patterns during the development of pea leaves10,41, the biosynthesis of Rubisco starts several days before that of GDC. This implies the existence of post-transcriptional control of gene expression. This could result from some as yet unidentified metabolic signals from several cell compartments, including chloroplasts (e.g. glycolate), peroxisomes (e.g. glycine) or the cytosol [CH2–THF (this compound is central to one-carbon metabolism)]. Biosynthesis of cofactors

Obviously, the formation of active GDC and SHMT is strictly dependent on an endogenous supply of

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

(c) L4

L2 L5

Relative amount of messengers

L1 L2 L3 L4 L5

100 80 60 40 20 0

Relative amount of messengers

L3

100 80 60 40 20 0

Relative amount of messengers

P-mRNA

100 80 60 40 20 0

Relative amount of messengers

(a)

L1

173

100 80 60 40 20 0

H-mRNA L1 L2 L3 L4 L5

T-mRNA L1 L2 L3 L4 L5

Activity (at state 3, in nmol of O2 consumed/min/mg of protein)

(b)

400 300 200

L-mRNA NADH Malate Pyruvate Glycine

100 0 L1

L2

L3

L4

L5

L1 L2 L3 L4 L5

L1 L2 L3 L4 L5 Developmental stages

L1 L2 L3 L4 L5 Developmental stages

L1 L2 L3 L4 L5 Developmental stages

L1 L2 L3 L4 L5 Developmental stages TRENDS in Plant Science

Fig. 4. Biogenesis of glycine decarboxylase during the course of pea leaf development. (a) The different stages of pea leaf development used to study the biogenesis of GDC. L1, L2, L3, L4 and L5 correspond to the developmental stages of the primary leaf of 4-, 5-, 7-, 9- and 12-day-old plants, respectively. Arrows indicate leaves harvested. (b) Substrate (glycine, pyruvate, malate, NADH) oxidation by purified mitochondria extracted from pea leaves at different developmental stages. (c) Northern blot analysis of mRNAs encoding the protein components of GDC during development of the pea primary leaf. The right side of the figure shows the quantification of signals generated using phosphor imaging apparatus. The amounts of mRNA are expressed relative to the maximum quantity of radioactivity detected in the experiment. Note that the appearance of glycine oxidation capacity seems to be closely related to the opening of the leaflets, which occurs after the seven-day stage.

various cofactors, including lipoic acid and THF. This raises the important problem of the origin of these cofactors3. Plants and microorganisms, in contrast with animals (folate supply in these organisms is dependent on feeding), are able to synthesize THF from 6-hydroxymethyl-7,8-dihydropterin and pamino benzoic acid (p-ABA) de novo. This pathway requires the sequential operation of five enzymes: a 6-hydroxymethyl-7,8-dihydropterin pyrophosphate kinase (HPPK), a 7,8-dihydropteroate synthase (DHPS), a dihydrofolate synthetase (DHFS), a dihydrofolate reductase (DHFR), and a folylpolyglutamate synthetase (FPGS) (Fig. 5). The http://plants.trends.com

second step of this pathway is the target of antimicrobial sulfonamide drugs and herbicides (such as Azulam), which are p-ABA analogues recognized by DHPS as alternative substrates. In pea leaves, all the enzymes involved in THF synthesis are present in the mitochondria55 and the maximal capacity of this biosynthetic pathway is sufficient to keep pace with the rapid accumulation of SHMT and GDC during the greening of the leaves. In plants, the HPPK enzyme is bifunctional and also supports DHPS activity56. This enzyme (515 residues) is synthesized with a mitochondrial transit peptide and Southern blot experiments suggest that a single-copy gene codes for the enzyme. Taken together these observations strongly suggest that the plant mitochondria are the unique site of 7,8dihydropteroate leading to THF (Fig. 5). It is possible that the binding of THF molecules to the T-protein and SHMT during the course of their accumulation would increase the rate of THF formation. Indeed, the inhibition of DHPS-catalysed reaction is tightly controlled because it is feedback inhibited by intermediates, including THF, synthesized downstream in the folate biosynthetic pathway. It is clear that the synthesis of THF in the matrix space of mitochondria raises the problem of the origin of the

174

Fig. 5. The biosynthetic pathway of tetrahydrofolate (5,6,7,8-tetrahydropteroylpolyglutamate, H4PteGlun) synthesis from 6-hydroxymethyl dihydropterin (H2pterin) and para-aminobenzoate (pABA) in plant mitochondria. (1) 6-hydroxymethyl dihydropterin pyrophosphokinase, (2) dihydropteroate synthase, (3) dihydrofolate synthetase, (4) dihydrofolate reductase, (5) folylpolyglutamate synthetase. H4PteGlun is a key cofactor in glycine decarboxylase (6) and serine hydroxymethyltransferase (7) functioning. Note that H2pterin and pABA are probably synthesized in an extra-mitochondrial compartment. If this assumption is verified, specific carriers should be present in the inner mitochondrial membrane.

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

pABA ? H2Pterin

ADP + Pi

PPi

ATP

Mg2+

Mg2+ H2Pteroate

H2PteGlu1 3

AMP

pABA

Glu

2

?

ATP Mg2+

H2PterinPPi

H2Pterin 1

NADPH 4 NADP H4PteGlu1 (THFGlu1)

Serine

Glycine

nGlu H4PteGlun + 1 (THFGlun + 1)

Mg2+ 5 nATP nADP + nPi

6

CH2−H4PteGlun (CH2−THF)

7 Glycine CO + NH 2 3

TRENDS in Plant Science

Acknowledgements Work on glycine decarboxylase is supported by the Centre National de la Recherche Scientifique (CNRS), the Commissariat à l’Energie Atomique (CEA) and the Ministère de la recherche. We would like to thank Claudine Cohen-Addad and David Macherel for their active collaboration during the course of this work.

substrates specifically involved in folate synthesis, such as 6-hydroxymethyldihydropterin and p-ABA. The synthesis of these metabolites and their eventual transport across the inner mitochondrial membrane are probably factors of regulation for folate synthesis. Although many organisms including plants are capable of synthesizing lipoate, the biosynthetic pathway is not fully described in any system. Octanoic acid probably serves as a specific precursor of lipoic acid. Plant mitochondria seem to contain all the enzymatic machinery involved in lipoic acid biosynthesis57,58 including lipoic acid synthase59 (Fig. 6). Fatty acid biosynthesis is initiated in the matrix space by malonate, which can be activated in malonyl-acyl carrier protein (malonyl-ACP) because of the presence of a malonyl-CoA synthetase coupled to a malonyl-CoA:ACP transacylase58 (Fig. 6). In contrast with the situation observed in plastids, plant mitochondria cannot use acetyl-CoA or acetate as the sole precursor for fatty acid synthesis because they lack acetyl-CoA carboxylase60. Malonyl-ACP thus formed is then used via the elongation reaction catalysed by a set of enzymes (the fatty-acid synthase multienzyme system consists of four different component enzymes designated as β-ketoacyl synthase, β-keto-acyl-ACP reductase, β-hydroxyacylACP dehydratase and enoyl-ACP reductase) to yield octanoyl-ACP. The octanoyl-ACP is probably the substrate used by the lipoate synthase to form the lipoyl-ACP. However, we cannot exclude the possibility that octanoyl-H protein is also the substrate of the http://plants.trends.com

lipoate synthase. Indeed, when present, H-apoprotein behaves as an eight-carbon fatty acid sink, with a flux of fatty acid synthesis being engaged towards the production of octanoylated H-protein58. One problem with this is the nature of the sulfur donor needed to drive the lipoate synthase reaction. A second problem, associated with the conversion of octanoic acid into lipoic acid, concerns the mechanism of the sulfur introduction process. The lipoyl and/or the octanoyl group are attached to the H-protein by the lipoate protein ligase. High specificity is required to attach lipoate and/or octanoate to the appropriate lysine residues of the lipoate-dependent enzyme (H-protein). In fact, this attachment occurs once the H-protein is fully folded: that is, after its transport in the form of an uncoiled precursor through the inner mitochondrial membrane18,58. Mutagenesis studies of the hairpin region around the lipoylated lysine (Val62 and Ala64) indicate that its primary structure is not sufficient for the selective recognition process by the lipoate ligase. Rather these proximal amino acids play an important role in the molecular events that govern the reaction between P- and H-protein25. Conclusions

Over the past several years, our understanding of the enzymes that impact upon glycine metabolism in plants has advanced dramatically. All the protein components (P-, H-, T- and L-proteins) of the glycine decarboxylase complex present in large amounts in the matrix space of green-leaf mitochondria have

Review

TRENDS in Plant Science Vol.6 No.4 April 2001

Malonate

CoASH

ACP

2 1

Malonyl-CoA 3

ACP CoASH

4? Malonyl-ACP

CO2

Acetyl-ACP CO2, ACP

4, 5, 6, 7 Octadecanoyl-ACP

Butyryl-ACP Hexadecanoyl-ACP

Malonyl-ACP

4, 5, 6, 7

CO2, ACP Hexanoyl-ACP 4, 5, 6, 7 Octanoyl-ACP

Substrate

9

8

9

ACP

ACP

Octanoyl-H protein

Lipoyl-ACP H apoprotein

Malonyl-ACP CO2, ACP H apoprotein

8? H protein (lipoylated)

Substrate

TRENDS in Plant Science

Fig. 6. Biosynthesis of fatty acids and lipoic acid in plant mitochondria. Fatty acid biosynthesis is initiated by malonate (plant mitochondria are devoid of acetyl-CoA carboxylase), which can be activated in malonyl-acyl carrier protein (malonyl-ACP) by two pathways: one catalysed by malonyl-ACP synthetase (1) and the other catalysed by malonyl-CoA synthetase (2) coupled to malonyl-CoA: ACP transacylase (3). Fatty acid synthase (FAS) catalyses the transformation of malonyl-ACP thus formed to octanoyl (C8)-, hexadecanoyl (C16)- and octadecanoyl (C18)-ACP. Mitochondrial FAS is made of a set of enzymes: the βketoacyl synthase (4), the β-ketoacyl-ACP reductase (5), the β-hydroxyacyl-ACP dehydratase (6) and the enoyl-ACP reductase (7). The octanoyl-ACP is probably the substrate used by the lipoate synthase (8) to form lipoyl-ACP. However, we cannot exclude the possibility that octanoyl-H-protein is also the substrate of the lipoate synthase. The electron donor for the sulfur incorporation is not known. The lipoyl and/or the octanoyl group are attached to the H-apoprotein by the lipoate transferase (9).

References 1 Leegood, R.C. et al. (1995) The regulation and control of photorespiration. J. Exp. Bot. 46, 1397–1414 2 Tolbert, N.E. (1997) The C2 oxidative photosynthetic carbon cycle. Annu. Rev. Plant Physiol. Plant Mol. Biol. 48, 1–25 3 Douce, R. and Heldt, H.W. (2000) Photorespiration. In Advances in Photosynthesis (Vol. 9) (Leegood, R.C. et al., eds), pp. 115–136, Kluwer 4 Cleland, W.W. et al. (1998) Mechanism of Rubisco: the carbamate as general base. Chem. Rev. 2, 549–561 http://plants.trends.com

175

been purified and their genes cloned. This should help to further our understanding of the self recognition of these proteins in the association process. The structural and mechanistic heart of this complex is provided by the lipoic acid-containing H-protein exerting its activity with the other partners in a temporally co-ordinated manner. This protein, which undergoes a cycle of reductive methylamination, methylamine transfer and electron transfer, acts as a mobile co-substrate that commutes between the other three proteins. The availability of folate to glycine decarboxylase and its recycling through SHMT is a crucial step for glycine oxidation during photorespiration. In spite of several impressive advances we remain unsure of how THF interacts with T-protein and SHMT to allow the continuous operation of the glycine oxidation reaction. It would be interesting to study the dependence of the reaction kinetics on the respective concentrations of THF and CH2–THF and their different association constants, to understand the shuttling of this cofactor between SHMT and T-protein. Developmental regulation of GDC and SHMT has been well documented in leaves. In monocot leaves, the protein component of GDC accumulates along the developmental gradient from the young tissue at the leaf base to the older tissue at the tip. However, during the course of pea leaf development the appearance of glycine oxidation capacity seems to be closely related to the opening of the leaflets, an event that allows the leaf to function as a morphologically efficient solar captor. This event is correlated with the huge accumulation of GDC proteins and SHMT in mitochondria, producing an increase in their density. Light directs the transcription of GDC-specific proteins in the same way as for photosynthetic genes such as rbcS or cab. However, the GDC and Rubisco expression differ at the translational level because Rubisco translation clearly precedes that of GDC. The synthesis of various cofactors including THF and lipoic acid should keep pace with the rapid accumulation of GDC and SHMT observed within the mitochondrial matrix during the course of green-leaf development. Without a full understanding of the enzymatic machinery that mediates THF and lipoic acid synthesis, the regulatory properties of the process that are needed to meet a demand for GDC and SHMT accumulation can only be surmised.

5 Oliver, D.J. et al. (1990) Glycine metabolism by plant mitochondria. Physiol. Plant. 80, 487–491 6 Oliver, D.J. (1994) The glycine decarboxylase complex from plant mitochondria. Annu. Rev. Plant Physiol. Plant Mol. Biol. 45, 323–338 7 Rawsthorne, S. et al. (1995) The glycine decarboxylase complex in higher plant mitochondria: structure, function and biogenesis. In Amino Acids and Their Derivatives in Higher Plants (Wallsgrove, R.M., ed.), pp. 87–109, Academic Press 8 Douce, R. and Neuburger, M. (1999) Biochemical dissection of photorespiration. Curr. Opin. Plant Biol. 2, 214–222

9 Oliver, D.J. et al. (1990) Interaction between the component enzymes of the glycine decarboxylase multi-enzyme complex. Plant Physiol. 94, 833–839 10 Vauclare, P. et al. (1996) Regulation of the expression of the glycine decarboxylase during pea leaf development. Plant Physiol. 112, 1523–1530 11 Mouillon, J.M. et al. (1999) Glycine and serine catabolism in non-photosynthetic higher plant cells: their role in C1 metabolism. Plant J. 20, 197–205 12 Prabhu, V. et al. (1996) 13C nuclear magnetic resonance detection of interactions of serinehydroxymethyltransferase with C1tetrahydrofolate synthase and glycine

176

13

14

15

16

17

18

19

20

21

22

23

24

25

26

27

28

Review

decarboxylase complex activities in Arabidopsis. Plant Physiol. 112, 207–216 Kikuchi, G. and Hiraga, K. (1982) The mitochondrial glycine cleavage system – unique features of the glycine decarboxylation. Mol. Cell Biochem. 45, 137–149 Bourguignon, J. et al. (1988) Resolution and characterization of the glycine cleavage reaction in pea leaf mitochondria. Properties of the forward reaction catalyzed by glycine decarboxylase and serine hydroxymethyltransferase. Biochem. J. 255, 169–178 Walker, J.L. and Oliver, D. (1986) Glycine decarboxylase multienzyme complex. Purification and partial characterization from pea leaf mitochondria. J. Biol. Chem. 261, 2214–2221 Guilhaudis, L. et al. (2000) Combined structural and biochemical analysis of the H–T complex in the glycine decarboxylase cycle: evidence for a destabilization mechanism of the H-protein. Biochemistry 39, 4259–4266 Cohen-Addad, C. et al. (1997) Structural studies of the glycine decarboxylase complex from pea leaf mitochondria. Biochimie 79, 637–644 Macherel, D. et al. (1996) Expression, lipoylation and structure determination of recombinant pea H-protein in Escherichia coli. Eur. J. Biochem. 236, 27–33 Pares, S. et al. (1994) X-ray structure determination at 2.6 Å-resolution of a lipoate containing protein: the H-protein of the glycine decarboxylase from pea leaves. Proc. Natl. Acad. Sci. U. S. A. 91, 4850–4853 Cohen-Addad, C. et al. (1995) The lipoamide arm in the glycine decarboxylase complex is not freely swinging. Nat. Struct. Biol. 2, 63–68 Faure, M. et al. (2000) Interaction between the lipoamide-containing H-protein and the lipoamide dehydrogenase (L-protein) of the glycine decarboxylase multienzyme system. 2. Crystal structures of H- and L-proteins. Eur J. Biochem. 267, 2890–2898 Mérand, V. et al. (1993) Characterization of the primary structure of H-protein from Pisum sativum and location of a lipoic acid residue by combined LC-MS and LC-MS-MS. Biol. Mass Spectrom. 22, 447–456 Neuburger, M. et al. (2000) Interaction between the lipoamide-containing H-protein and the lipoamide dehydrogenase (L-protein) of the glycine decarboxylase multienzyme system. 1. Biochemical studies. Eur. J. Biochem. 267, 2882–2889 Sarojini, G. and Oliver D. (1985) Inhibition of glycine oxidation by carboxymethoxylamine, methoxylamine, and acethydrazide. Plant Physiol. 77, 786–789 Gueguen, V. et al. (1999) Structural and functional characterization of H protein mutants of the glycine decarboxylase complex. J. Biol. Chem. 274, 26344–26352 Guilhaudis, L. et al. (1999) Investigation of the local structure and dynamics of the H subunit of the mitochondrial glycine decarboxylase using heteronuclear NMR spectroscopy. Biochemistry 38, 8334–8346 Rébeillé, F. et al. (1994) Interaction between glycine decarboxylase, serine hydroxymethyltransferase and tetrahydrofolate polyglutamates in pea leaf mitochondria. Biochem. J. 302, 223–228 Besson, V. et al. (1993) Effects of tetrahydrofolate polyglutamates on the kinetic parameters of serine hydroxymethyltransferase from pea leaf mitochondria. Biochem. J. 292, 425–430 http://plants.trends.com

TRENDS in Plant Science Vol.6 No.4 April 2001

29 Williams, C.H. (1992) Lipoamide dehydrogenase, glutathione reductase, thioredoxine reductase and mercuric reductase family of flavoenzyme transhydrogenases. In Chemistry and Biochemistry of Flavoproteins (Müller, F., ed.), pp. 121–211, CRC Press 30 Reed, L.J. and Hackert, M.L. (1990) Structure– function relationships in dihydrolipoamide acyltransferases. J. Biol. Chem. 265, 8971–8974 31 Berg, A. and Kok, A. (1997) 2-oxo acid dehydrogenase multienzyme complexes. The central role of the lipoyl domain. Biol. Chem. 378, 617–634 32 Turner, S.R. et al. (1992) The cloning and characterization of a cDNA encoding the Lsubunit of glycine decarboxylase from pea (Pisum sativum). J. Biol. Chem. 267, 7745–7750 33 Bourguignon, J. et al. (1992) Isolation, characterization, and sequence analysis of a cDNA clone encoding L-protein, the dihydrolipoamide dehydrogenase component of the glycine decarboxylase system from pea leaf mitochondria. Eur. J. Biochem. 204, 865–873 34 Bourguignon, J. et al. (1996) Glycine decarboxylase and pyruvate dehydrogenase complexes share the same dihydrolipoamide dehydrogenase in pea leaf mitochondria: evidence from mass spectrometry and primary structure analysis. Biochem. J. 313, 229–234 35 Ke, J. et al. (2000) The role of pyruvate dehydrogenase and acetyl-coenzyme A synthetase in fatty acid synthesis in developing Arabidopsis seed. Plant Physiol. 123, 497–508 36 Lutziger, I. and Oliver, D.J. (2000) Molecular evidence of a unique lipoamide dehydrogenase in plastids. FEBS Lett. 484, 12–16 37 Turner, S.R. et al. (1992) Identification and localization of multiple forms of serine hydroxymethyltransferase in pea (Pisum sativum) and characterization of a cDNA encoding a mitochondrial isoform. J. Biol. Chem. 267, 13528–13534 38 Besson, V. et al. (1995) Evidence for three serine hydroxymethyltransferase in green leaf cells. Purification and characterization of the mitochondrial and chloroplastic isoforms. Plant Physiol. Biochem. 33, 665–673 39 Bourguignon, J. et al. (1993) Glycine decarboxylase complex from higher plants. Molecular cloning, tissue distribution and mass spectrometry analysis of the T-protein. Eur. J. Biochem. 217, 377–386 40 Douce, R. and Neuburger, M. (1989) The uniqueness of plant mitochondria. Annu. Rev. Plant Physiol. Plant Mol. Biol. 40, 371–414 41 Oliver, D.J. and Raman, R. (1995) Glycine decarboxylase: protein chemistry and molecular biology of the major protein in leaf mitochondria. J. Bioenerg. Biomembr. 27, 407–414 42 Douce, R. and Bonner, W.D. (1972) Oxaloacetate control of Krebs cycle oxidation in purified plant mitochondria. Biochem. Biophys. Res. Commun. 47, 619–624 43 Day, D.A. and Wiskich, J.T. (1981) Glycine metabolism and oxaloacetate transport by pea leaf mitochondria. Plant Physiol. 68, 425–429 44 Ebbighausen, H. et al. (1985) Oxaloacetate translocator in plant mitochondria. Biochim. Biophys. Acta 810, 184–199 45 Kim, Y. and Oliver, D. (1990) Molecular cloning, transcriptional characterization and sequencing of the cDNA encoding the H-protein of the glycine decarboxylase complex in peas. J. Biol. Chem. 265, 848–853

46 Macherel, D. et al. (1990) Primary structure and expression of the H-protein, a component of the glycine cleavage system of pea leaf mitochondria. Biochem. J. 268, 783–789 47 Turner, S.R. et al. (1992) Cloning and characterization of the P-subunit of glycine decarboxylase from pea. J. Biol. Chem. 267, 5355–5360 48 Bauwe, H. et al. (1995) Structure and expression analysis of the gdcsPA and gdcsPB genes encoding two P-isoproteins of the glycine-cleavage system from Flaveria pringlei. Eur. J. Biochem. 234, 116–124 49 Vauclare, P. et al. (1998) The gene encoding T protein of the glycine decarboxylase complex involved in the mitochondrial step of the photorespiratory pathway in plants exhibits features of light-induced genes. Plant Mol. Biol. 37, 309–318 50 Srinivasan, R. and Oliver, D.J. (1995) Lightdependent and tissue-specific expression of the H-protein of the glycine decarboxylase complex. Plant Physiol. 109, 161–168 51 Macherel, D. et al. (1992) Cloning of the gene (gdcH) encoding H-protein, a component of the glycine decarboxylase complex of pea (Pisum sativum L.). Biochem. J. 286, 627–630 52 Kopriva, S. et al. (1996) H-protein of the glycine cleavage system in Flaveria: alternative splicing of the pre-mRNA occurs exclusively in advanced C4 species of the genus. Plant J. 10, 369–373 53 Rogers, W.J. et al. (1991) Changes to the stoichiometry of glycine decarboxylase subunits during wheat (Triticum aestivum L.) and pea (Pisum sativum L.) leaf development. Plant Physiol. 96, 952–956 54 Thompson, P. et al. (1998) Heterogeneity of mitochondrial protein biogenesis during primary leaf development in barley. Plant Physiol. 118, 1089–1099 55 Neuburger, M. et al. (1996) Mitochondria are a major site for folate and thimidylate synthesis in plants. J. Biol. Chem. 271, 9466–9472 56 Rébeillé, F. et al. (1997) Folate biosynthesis in higher plants: purification and molecular cloning of a bifunctional 6-hydroxymethyl-7,8dihydropterin pyrophosphokinase/7,8dihydropteroate synthase localized in mitochondria. EMBO J. 16, 947–957 57 Wada, H. et al. (1997) Why do mitochondria synthesize fatty acids? Evidence for involvement in lipoic acid production. Proc. Natl. Acad. Sci. U. S. A. 94, 1591–1596 58 Gueguen, V. et al. (2000) Fatty acid and lipoic acid synthesis in higher plant mitochondria. J. Biol. Chem. 275, 5016–5025 59 Yasumo, R. and Wada, H. (1998) Biosynthesis of lipoic acid in Arabidopsis: cloning and characterization of the cDNA for lipoic acid synthase. Plant Physiol. 118, 935–943 60 Alban, C. et al. (2000) Biotin metabolism in plants. Annu. Rev. Plant Physiol. Plant Mol. Biol. 51, 17–47 61 Kraulis, P.J. (1991) MOLSCRIPT: a program to produce both detailed and schematic plots of protein. J. Appl. Crystallogr. 24, 946–950 62 Jones, T.C. et al. (1991) Improved methods for building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A47, 110–119 63 Nicholls, A. et al. (1993) GRASP: graphical representation and analysis of surface properties. Biophys. J. 64, 166–170