Sensitive and high throughput metabolite assays for inorganic pyrophosphate, ADPGlc, nucleotide phosphates, and glycolytic intermediates based on a novel enzymic cycling system

Authors


*For correspondence (fax +49 331 5678101; e-mail mstitt@mpimp-golm.mpg.de)

Summary

Metabolite assays are required to characterise how metabolism changes between genotypes during development and in response to environmental perturbations. They provide a springboard to identify important regulatory sites and investigate the underlying mechanisms. Due to their small size, Arabidopsis seeds pose a technical challenge for such measurements. A set of assays based on a novel enzymic cycling system between glycerol-3-phosphate dehydrogenase and glycerol-3-phosphate oxidase have been developed and optimised for use with growing Arabidopsis seeds. In combination with existing assays they provide a suite of high throughput, sensitive assays for the immediate precursors for starch (adenine diphosphate glucose) and lipid (acetyl coenzyme A, glycerol-3-phosphate) synthesis, as well as pyrophosphate, ATP, ADP and most of the glycolytic intermediates. A method is also presented to rapidly quench intact siliques, lyophilise them and then manually separate seeds for metabolite analysis. These techniques are used to investigate changes in overall seed metabolite levels during development and maturation, and in response to a stepwise decrease of the external oxygen concentration.

Introduction

Arabidopsis is the premier system for the genetic dissection of plant gene function thanks to its small diploid genome, small size and rapid life cycle, the creation of large collections of chemical and tagged mutants and, more recently, access to the complete genome sequence and the development of increasingly sophisticated tools for gene mapping and mutant identification (Kaul et al., 2000). Paradoxically, smallness is often perceived as a handicap for studies of physiology and metabolism.

The power of Arabidopsis for studies of metabolism and physiology is illustrated by numerous studies of mutants and transformants with, for example, lesions in starch (Caspar, 1994; Periappuram et al., 2000), lipid (Browse and Somerville, 1994; Poirier et al., 1999), flavonoid (Albert et al., 1997; Debeaujon et al., 2000, 2001; Shirley, 1998; Wisman et al., 1998), phenylpropanoid (Chapple et al., 1992), amino acid (Galili et al., 2000, 2001) and cell wall (Reiter, 1994) metabolism, or defects in the regulation of these pathways (Albert et al., 1997; Focks and Benning, 1998; Nesi et al., 2000; Walker et al., 1999). Collections of pre-identified Arabidopsis insertion mutants (e.g. http://signal.salk.edu/cgi-bin/tdnaexpress, https://http://www.mpiz-koeln.mpg.de/GABI-Kat/Home/index.html) open the vista of a systematic definition of the function of each individual enzyme, including every member of the large multigene families that have been annotated for many enzymes and enzyme classes. The pre-eminence of Arabidopsis for studies of development, signalling and cell biology provides another compelling argument to use it as a vehicle to investigate metabolism. There are important and poorly understood interactions between development, cell biology and metabolism. For example, mutations that lead to embryo growth arrest (e.g. lec, fus3,Raz et al., 2001; Vicient et al., 2000) or modifiy embryo dormancy (e.g. abi3, Koornneef et al., 1989) also show widespread modifications in the levels of carbohydrates, lipids and storage proteins (Baumlein et al., 1994; Boisson et al., 2001; De Bruiyn et al., 1997; Ooms et al., 1993; Parcy et al., 1997). Similarly, systematic studies of gene families that encode receptors, signal transduction pathway components and transcription factors (Kaul et al., 2000) will lead to new insights into the role and regulation of metabolism.

Measurements of metabolite levels provide information about the physiological status of a tissue. In combination with information about in vivo fluxes and the expression and properties of the enzymes in the pathway, they also pin-point sites at which regulation is acting and aids the formulation and testing of hypotheses about the underlying mechanisms. The small size of Arabidopsis poses a technical challenge for the analysis of metabolites that are present at low concentrations, especially in growing tissues like sink leaves, root tips or developing seeds. Our understanding of metabolism is particularly incomplete in such systems, where important insights into the interaction with cell function and development can be anticipated. There is also a pressing need to increase throughput in order to screen for altered metabolite levels across a wide range of conditions, tissues or genotypes.

Primary metabolism provides the precursors and immediate source of energy for the synthesis of all cellular components. Most of the intermediates and cofactors of primary metabolism are present at low concentrations. Traditional assay methods using coupled enzyme assays or HPLC are slow, tedious and require expensive instrumentation (see, e.g. Stitt et al., 1989). Novel and powerful approaches using GC/MS and LC/MS are becoming available (Fiehn et al., 2000; Roessner et al., 2001) to measure a wide range of metabolites. However, these techniques depend on access to expensive, specialised instrumentation and extensive facilities to de-convolute large amounts of data, they require relatively large amounts of plant material to detect minor components, and they are not readily applicable to some important metabolites including nucleotide phosphates and pyrophosphate.

The sensitivity of coupled enzyme assays can be dramatically extended by a procedure termed ‘enzyme cycling’ (Lowry and Passoneau, 1972). A cycling assay consists of an enzyme that converts metabolite A into metabolite B and a second enzyme that converts B back into A, while at the same time generating an easily measurable metabolite e.g. NAD(P)(H). As the enzymes and cofactors are included in excess, the rate of the cycle depends on the summed concentration of A plus B. Amplification is achieved because each molecule of A or B is repeatedly cycled, leading to the production of a large number of molecules of the metabolite that is detected. The concentration of A or B can be specifically determined, provided that the other metabolite can be selectively destroyed before starting the cycling reaction. The assay can be extended to measure other metabolites that can be converted to A or B, provided a method is available to destroy A, B and any other intervening intermediates. The potential to create novel assays has increased immensely in recent years because large amounts of sequence information and the ready access to systems for functional heterologous overexpression and one-step purification of gene products facilitates the development of tailored assays.

In this article, we present a set of high throughput and high-sensitivity assays for ATP, ADP, PPi, UDPGlc, ADPGlc, 3-phosphoglycerate (3PGA), and glycerol-3-phosphate (Gly3P), based on a novel enzymic cycling system using glycerol-3-phosphate dehydrogenase (Gly3PDH) and glycerol-3-phosphate oxidase (Gly3POX). These have been combined with optimised tests based on established cycling assays for hexose-P, AcCoA and CoA to provide a suite of sensitive high throughput assays for the immediate precursors for starch and lipid synthesis as well as most of the metabolites in the pathways of sucrose breakdown and glycolysis. To illustrate their use we have investigated overall metabolite levels in Arabidopsis seeds at different stages of maturation, and after placing siliques in low [O2].

Results

Basis of metabolite determination by a glycerol-3-P oxidase/glycerol-3-P dehydrogenase cycle

The basis for the new assays is the enzymic cycle between Gly3POX and Gly3PDH (Figure 1a). Gly3POX catalyses an O2-mediated conversion of Gly3P into dihydroxyacetone-P (DAP), and Gly3PDH converts DAP back into Gly3P and simultaneously oxidises NADH to NAD+ (Figure 1a). The net reaction (NADH + H+ + O2 = NAD+ + H2O2) is a pseudo zero-order reaction whose rate (– d[NADH]/dt = -d[O2]/dt) depends on the sum of Gly3P and DAP in the assay. Amplification is obtained because each molecule of Gly3P or DAP is cycled many time, leading to the accumulation of NAD+ and H2O2. The decrease of the NADH concentration can be followed directly at 340 nm. In principle, H2O2 formation could be detected by coupling to a peroxidase-based chromogenic reaction (not shown). To assay specific metabolites the cycle must be extended in two ways. First, methods must be developed to feed further metabolites into the cycle, by converting them stoichiometrically into Gly3P or DAP. Second, methods must be developed to destroy Gly3P, DAP and intervening metabolites so that the amount of (Gly3P + DAP) and hence the rate of the cycle depends solely on the metabolite in question.

Figure 1.

Principles of the assays based on the glycerol-3-P cycling.

(a) Enzyme cycling reaction.

(b) Sequences of enzymes to feed further metabolites into the cycling assay.

(c) Arsenolysis to destroy intervening metabolites.

Figure 1(b) summarises potential routes that convert other metabolites into Gly3P or DAP. Inclusion of glycerokinase and excess glycerol provides one important point of entry. As glycerokinase is not specific for ATP, any nucleotide triphosphate can be coupled to Gly3P formation. This facilitates determination of total nucleotide triphosphates and, more importantly, metabolites that participate in reactions in which a nucleotide triphosphate is formed. For example, UDPGlc can be converted to UTP by adding PPi and UDPGlc pyrophosphorylase (UGPase), PPi can be converted to UTP by adding UDPGlc and UGPase, ADPGlc can be converted to ATP by adding PPi and ADPGlc pyrophosphorylase (AGPase), and ADP can be converted to ATP by adding myokinase. Another point of entry is to include triose phosphate isomerase (TPI), glyceraldehyde-3P dehydrogenase (GAPDH) and 3-phosphoglycerate kinase, which allow determination of ATP or 3PGA. Inclusion of TPI and aldolase would in principle allow determination of the individual triose-P and fructose-1,6-bisphosphate.

Two basic strategies were used to detect a specific metabolite in these reaction sequences. One is applicable when the metabolite in question is present at relatively high concentrations relative to other metabolites lying downstream from it in the reaction sequence, for example when 3PGA or ATP are assayed via conversion to DAP. In this case, the cycling assay is initiated in the absence of the key-enzyme (in this case phosphoglycerate kinase), a first constant rate of reaction (V1) is reached which depends on the total level of the intervening metabolites, the critical enzyme (in this case phosphoglycerate kinase) is then added and after a short acceleration the new reaction rate (V2) is recorded. The difference between V2 and V1 gives the amount of the metabolite converted by the key-enzyme with high precision, as the blank and the measurement are run in one well. The other strategy is to remove intervening metabolites before starting the cycling assay, and is the method of choice when the metabolite in question is present at low concentrations relative to intervening metabolites. In some cases, metabolites can be specifically destroyed by chemical treatments. For example, adjusting the extract to pH 8 and heating it before adding it to the cycling assay allows specific determination of Gly3P, because DAP is completely destroyed within 20 min at pH 8 and at 95°C whereas Gly3P is stable (not shown). In other cases the intervening metabolites can be removed using enzymatic reactions. For example (Figure 1c), before assaying PPi or ADPGlc, all the downstream metabolites (i.e. ATP, Gly3P, triose-P) can be converted to 3PGA by adding glycerol, glycerokinase, Gly3POX, TPI, NAD+ dependent GAPDH, NAD+, and arsenate. Arsenate is included to allow an irreversible conversion of triose phosphates through to 3PGA (Bergmeyer, 1987). To measure ADP, pyruvate and pyruvate orthophosphate dikinase can be added to convert ATP into AMP. A method to selectively destroy Gly3P without destroying DAP has not yet been developed.

Development and validation of new assays for PPi and ADPGlc

Based on this principle, optimised assays for use in 96-well microplates were developed for a range of metabolites. The following section describes the optimisation and validation for two examples, PPi and ADPGlc. They presented a particular challenge due to their low concentrations in plant tissues.

The first step was to optimise the conditions for the cycling reactions. The amount of Gly3PDH included in the assay (4 U assay−1) was selected to ensure that contaminating enzymes (aldolase, lactate dehydrogenase) are at such low levels that they do not interfere with the assay (see below for recoveries). The Keq of Gly3PDH ([DAP][NADH][H+]/[Gly3P][NAD+] = 1.10−12 m; Bergmeyer, 1987) strongly favours reduction of DAP to Gly3P, and Gly3PDH has a high affinity for its substrates (Km = 0.45 mm[DAP], Km = 13 µm[NADH]; White and White, 1997) relative to Gly3POX (Km = 3.2 mm[Gly3P]; White and White, 1997). To obtain maximal rates of cycling it is therefore necessary to include Gly3POX in excess of Gly3PDH. Including Gly3POX activity at up to 20-fold excess led to a linear increase in the rate of NAD+ formation. Routinely we used a 6.25-fold excess of Gly3POX, due to the cost of the commercial enzyme. The rate of cycling was optimised at pH 8 (not shown) and MgCl2 was included because it is required for some of the enzymes used to feed metabolites into the cycle (see below). Gly3PDH is fully saturated with NADH at the concentration used in the assay (typically approximately 1.2 mm), indeed the cycling reaction rate was constant until NADH was depleted below 0.3 µm. Gly3POX is saturated by the O2 dissolved in equilibrium with the atmosphere. Catalase was systematically added to prevent accumulation of H2O2, which might inhibit other components of the assay. The rate of NAD+ formation rose in a linear manner as the amount of Gly3P included in the assay was increased from 1 up to 500 pmol (data not shown), which is 5- to 250-fold higher than the levels of the intermediates typically found in the plant extracts assayed in the following experiments.

The next step was to develop methods to feed PPi or ADPGlc into the cycling assay. This required removal of interfering metabolites, determination of the blank, and determination of the metabolite in question. (i) To destroy ATP, Gly3P and triose-P, we added an excess of glycerol and glycerokinase, Gly3POX, TPI and GAPDH, arsenate (see above) and catalase. Destruction was incomplete if catalase was omitted, presumably due to some deleterious effects of H2O2. A 20-min treatment led to complete removal of 1 nmol of ATP, both in the absence and in the presence of a range of plant extracts (data not shown). This is at least 10-fold higher than the total amount of ATP, Gly3P and DAP in the amount of extract typically added to the assay. If for some reason arsenolysis is incomplete, this is revealed by an increased assay blank for that particular sample (see below). After 20 min, other metabolites needed for the reaction (i.e. UDPGlc for determination of PPi, or PPi for determination of ADPGlc) and Gly3PDH and NADH are added. This blocks further arsenolysis, probably because the high NADH to NAD+ ratio inhibits the forward reaction of the GAPDH. (ii) The assay blank (V1) is determined after 20 min. During the optimisation of the assays, we added Gly3P at this stage to check that the arsenolysis was blocked. We found that a new and constant steady state rate was achieved (not shown). If for some reason arsenolysis had not been completely blocked, the amount of Gly3P and hence the rate of the cycling reaction would decline with time. (iii) After registering V1 for a period of time, the final enzyme (UGPase or AGPase, respectively) is added to allow conversion of PPi or ADPGlc into Gly3P. The resulting increase of the Gly3P concentration in the assay leads to an increase in rate of the enzymatic cycle, which reaches a new and constant rate once all the PPi or ADPGlc has been converted to Gly3P (V2). It is important that the new steady-state rate V2 is maintained for reasonable period of time. A rapid decline provides a further alert that arsenolysis has not been completely blocked. The amount of PPi or ADPGlc is given by the difference (V2 − V1). As AGPase cannot be obtained commercially, GLGC was cloned from E. coli, overexpressed in E. coli as a His-tagged fusion protein, and purified.

Quantification is achieved by comparison with a standard curve, in which pure PPi or ADPGlc are added in the presence of pseudoextract. The rate of the cycling reaction was unaffected by including pseudoextract at a 5-fold excess of the extract concentration in a typical assay. Compounds in the extracts do not significantly inhibit the cycling reaction, because the extracts are highly diluted before addition to the cycling assay. The signal (V2-V1) is linearly dependent on the amount of extract included until at least 3-fold more extract is added than in a typical assay. Inhibitory effects in individual extracts can be routinely excluded by checking that the background reaction (V1) is similar for assays containing water, pseudoextract and each individual sample.

Figure 2 documents details of the assays for PPI and ADPGlc. In both, V1 and V2 are linear with time for at least 30 min. For the PPi assay, the blank in the absence of added PPi is very low, and the V2 rises in a linear manner as the amount of PPi in the assay is increased from 1 to 20 pmol (Figure 2a). For ADPGlc the signal was linearly related to the amount of ADPGlc in the assay, but the blank was higher (Figure 2b). The high blank was due to ATP in the AGPase preparation. If the extract contains low levels of ADPGlc, the blank can be decreased by using less AGPase. To check the reliability of the assay procedure, different amounts of pure PPi or ADPGlc were added to an Arabidopsis seed extract. Recovery of PPi and ADPGlc was 95% and 102%, respectively, and sensitivity lay in the 1–2 pmol range. Similar recoveries and sensitivities were found for Arabidopsis leaf and potato tuber extracts (not shown). In initial experiments, recovery of PPi was lower, due to traces of PPase in the coupling enzymes. Fluoride, which is an inhibitor of pyrophosphatase (Bergmeyer, 1987), was added to prevent the hydrolysis of PPi. In our conditions 1.5 mm were sufficient to block pyrophosphatase activity (not shown) and allow the recoveries shown in Figure 2.

Figure 2.

Validation of the assays for ADPGlc and PPi.

(a) Standard curve for the assay of PPi in the 1–20 pmol range.

(b) recovery of PPi standards (1–20 pmol) from Arabidopsis seed extracts.

(c) Standard curve for the assay of ADPGlc for the pmol range 1–20.

(d) recovery of ADPGlc standards (1–20 pmol) from Arabidopsis seeds extracts. The slope of the line gives the fraction of the metabolite recovered, the sd were used to estimate the sensitivity of the assays. Bars, sd(n = 3).

Analytic platform to measure a wide range of metabolites in primary energy and carbon metabolism

Table 1 lists metabolites for which assays have been developed based on the Gly3PDH/Gly3POX cycle, and summarises information about the assay systems, recoveries, sensitivity and the minimum amount of seed DW needed for a determination. Table 1 also lists further metabolites for which existing high sensitivity assays have been optimised for assay of Arabidopsis seed extracts. The details of the assay procedures are given in the Experimental procedures section. As the limit of detection for the assays lies in range of 1–2 pmol assay−1 very small amounts of tissue are needed, even for metabolites that are present at very low levels. For example, the determination of ADPGlc, which is present at levels of only about 100 nmoles/g DW, requires less than 20 µg of dried seed material, which is equivalent to about 5 seeds. An extract from 20 to 25 seeds would suffice to allow all of these metabolites to be assayed, although it is advantageous to work with larger amounts where possible.

Table 1.  Summary of the individual tests in the assay platform. Recovery and sensitivity were estimated as described for ADPGlc and PPi in Figure 2. The amount of seed DW used for each assay was calculated from the observation that at stage 17, 10 siliques contained 1.5 ± 0.3 mg of dry seeds. The DW of one seed was estimated by 2–5 µg in the 17 developmental stage of the fruit
MetaboliteCycling
assay
Recovery
(%)
Sensitivity
(pmol)
Seed DW
corresponding to
sensitivity (µg)
Gly3P 10223
ATP 10622
ADP 97515
PPiGly3POX/Gly3PDH951–26
UDPGlc 9922
ADPGlc 1021–220
3PGA 10323
Glc6PGlc6PDH10121
Glc1P/(Phenazine 10327
Fru6Pmethosulfate + MTT)10023
AcCoACitrate synthase/n.d.n.d.n.d.
CoASHPhospho-n.d.n.d.n.d.
transacetylase   

To allow high throughput, the assays were adapted to a 96-well microplate format. Throughput depends on the time required for the measurement in the ELISA reader(s). For example the photometric step of the procedure for the assay of Glc1P only takes 10 min while the work upstream takes about 80 min, allowing 500 determinations pro hour with one microplate reader if the preceding operations are efficiently organised. The ADPGlc assay has a throughput of one microplate every 50 min, which is 5-fold lower, but is easier to automate since there is no heating, and thus no centrifugation, step (see materials and methods). Although the assays can be carried out with normal laboratory equipment, throughput and accuracy are enhanced if a programmed pipetting robot is available (data not shown).

Separation of freeze quenched and lyophilised Arabidopsis seeds from the silique wall

Metabolites turn over rapidly, with half-lives in the order of seconds or less (see, e.g. ap Rees et al., 1977; Stitt et al., 1980). Metabolism must therefore be rapidly quenched to prevent artefacts due to changes during harvesting or killing. To achieve this, siliques were transferred rapidly under the prevailing light and atmospheric conditions into liquid nitrogen. To separate the seeds from the silique walls, the intact siliques were lyophilised at − 60°C, and opened under a binocular and the seeds removed using a custom built microaspirator.

Age-dependent changes in overall seed metabolite levels

A first set of experiments investigated overall metabolite levels in Arabidopsis seeds at different stages of development (Figure 3). As flowers develop sequentially along an inflorescence stem, the siliques form a developmental progression with the youngest at the tip (Meinke, 1994). Several inflorescences containing about 50 siliques were divided into 5 pools of 10 consecutive siliques. Seeds in pool 1 were white to pale green, seeds from fractions 2 and 3 were green and had a translucent coat, seeds from fraction 4 were brownish, and seeds from fraction 5 were mature (data not shown), corresponding to stages 17 A, 17B, 18 and 19, respectively, of the scale proposed by Smyth et al. (1990) and modified by Ferrandiz et al. (1999).

Figure 3.

Metabolite levels in growing siliques of Arabidopsis as related to the developmental stage.

Inflorescences were harvested in the middle of the photoperiod and immediately quenched in liquid nitrogen. The siliques were lyophilised and the seeds (•) separated from the silique walls (○).

(a) ATP (b) PPi (c) UDPGlc (d) 3PGA (e) ADPGlc (f) Gly3P (g) AcCoA (h) CoASH. Bars, se (n = 5).

Compared to seeds in the youngest siliques, ATP decreased by 30 and 60% in seeds from siliques between position 10–20 and 20–30 and was very low in the oldest seeds (Figure 3a). PPi was highest in seeds from the youngest siliques, decreased by 35% in seeds from siliques in positions 10–20, and remained at this level except in the oldest seeds (Figure 3b). PPi was relatively high compared to ATP in the youngest seeds, and rose relative to ATP as the seeds matured. UDPGlc (Figure 3c) remained high in seeds across silique positions 1–30 and fell to low levels in old seeds. The glycolytic intermediate 3PGA declined progressively as seeds became older (Figure 3d). ADPGlc, the immediate precursor for starch synthesis, decreased in parallel with ATP (Figure 3e). Gly3P (Figure 3f) decreased only slightly and AcCoA/Figure 3g peaked in seeds from siliques in the position 10–20 and remained high in positions 20–30. The AcCoA:CoA ratio (compare Figure 3g,h) rose between positions 1–10 and 21–30. The silique wall contained less PPi relative to ATP, less AcCoA relative to ADPGlc, a much lower AcCoA/CoA ratio, and the general age-dependent decrease of metabolites was less marked than in seeds (Figure 3a-h).

Although these measurements do not distinguish between metabolites in the testa, endosperm and embryo, they already point to some distinctive features of seed metabolism. These include (i) the high level of PPi relative to ATP; (ii) high levels of UDPGlc; and (iii) the high levels of the precursors for lipid synthesis at stages 17B and 18. PPi is generated during the synthesis of many biopolymers and plays a unique role in the cytosol of plant cells as an alternative energy donor to ATP (Stitt, 1998). UDPGlc is the product of sucrose breakdown via sucrose synthase and the precursor for cell wall synthesis. The shift from ADPGlc to Gly3P and AcCoA corresponds with a shift towards lipid accumulation as the embryo grows and matures (Rawsthorne et al., 1999). Further studies of metabolism during seed development will require a further refinement of the harvesting technique to allow separation of these different tissues under conditions that prevent changes in the levels of the metabolites, or the adaptation of the assay techniques to allow them to be employed in situ on tissue sections.

Influence of external oxygen on metabolite levels and fluxes in developing seeds

A second application investigates the response of seed metabolism to decreasing external [O2]. In contrast to animals, plants have not evolved a circulatory system or oxygen-carrying proteins, with the result that bulky or metabolically active tissues often become hypoxic (see Geigenberger et al., 2000). This could have far-reaching implications for the regulation of plant metabolism and growth. [O2] falls below 5% in the growing seeds of several species (Geigenberger et al., 2000) including Arabidopsis (Kuang et al., 1998; Porterfield et al., 1999), even when the pods are in air. It has also been long known that low external [O2] seriously impairs seed yield (Quebedeaux and Hardy, 1975, 1976). Their smallness makes Arabidopsis siliques an ideal system to investigate if O2 entry is a general constraint for plant metabolism.

Stalks sections containing siliques between positions 10–30 (corresponding to stage 17B) were prepared from 20 plants, placed in 50 ml Falcon tubes in the dark and supplied with 21%, 12%, 8%, 4%, 1% or zero [O2] for 2 h. Metabolism was then quenched by rapidly filling the Falcon tubes with liquid N2 (Figure 4). For each sample, 5 samples of 10 siliques were collected, lyophilised and separated into seeds and silique walls (yielding approximately 1.5 and 2.4 mg DW, respectively). To check whether detachment of the stalk leads to significant changes in metabolism, we compared siliques on detached stalks that were incubated for 2 h in 21% [O2] in the dark before harvest with siliques that were harvested directly from intact plants in the dark. Their seeds contained identical levels of adenine nucleotides, hexose phosphates, 3PGA, ADPGlc, Gly3P and AcCoA (data not shown). To provide information about metabolic fluxes a parallel incubation was carried out in which small amounts of 2 mm[U14C]-sucrose were injected through the silique coat into the internal airspaces of the siliques (Figure 5).

Figure 4.

Metabolite levels in siliques of Arabidopsis incubated at different oxygen tensions.

Inflorescences were harvested in the middle of the photoperiod, incubated for 2 h at 23°C in the dark at 20, 12, 8, 4, 1 and 0% oxygen, quenched in liquid nitrogen, lyophilised and the seeds (•) separated from the silique wall (○).

(a) ATP (b) ADP (c) ADP/ATP ratio (d)aspartate (e) glutamate (f) alanine (g) Gly3P (h) lactate (i) GABA (j) ADPGlc (k) AcCoA (l) PPi (m) PPi/ATP ratio (m) UDPGlc (o) Glc1P (p) Glc6P (q) Fru6P (r) 3PGA. Data are means ± se (n= 5).

Figure 5.

Metabolism of 14C-sucrose in growing Arabidopsis seeds.

In the middle of the light period, selected siliques were injected with 5 µl of 2 mm[U-14C] sucrose and then incubated for 2 h at 23°C in the dark with 21, 12, 8, 4, 2, 1 or 0% oxygen, quenched in liquid nitrogen, lyophilised, the seeds separated, and the distribution of the absorbed [14C] determined in

(a) Total sugars (b) Starch (○) and lipids (▪) (c) Amino acids (•) and organic acids (○) (d) Cell wall (○) and protein (•) was calculated by dividing the label determined in the respective fractions by the total label in the seed. Data are means ± se (n = 5 injections on separate plants).

Decreased external [O2] led to a dramatic decrease of ATP (Figure 4a), a small decrease of ADP (Figure 4b) and a marked decrease of the ATP/ADP ratio (Figure 4c) in seeds. These changes were already marked when external [O2] was in the 8–12% range. This contrasts with the silique wall, where the ATP/ADP ratio did not decrease until the external [O2] was very low. The Km(O2) of cytochrome oxidase is about 14 µm, which is equivalent to the oxygen concentration in equilibrium with a gas phase containing 0.013% oxygen at 20°C (Drew, 1997). Clearly, a very large O2 concentration gradient is required to drive O2 entry through the silique wall and into the seed at the rate required to maintain energy metabolism in the developing seed.

Independent evidence that seed metabolism becomes O2-limited at a relatively high external [O2] is provided by the response of metabolites produced in pathways that recycle NADH. A decrease of external [O2] in the range between 21% and 8% led to a decrease of aspartate (Figure 4d) and glutamate (Figure 4e) and an increase of alanine (Figure 4f). Alanine is formed in a pathway in which the amino group is transferred to pyruvate from asparate, releasing oxaloacetate that is sequentially reduced to succinate. This pathway leads to less acidification than lactate formation (Menegus et al., 1989). Alanine also accumulates at relatively high external [O2] in potato tuber discs (K. Bologa and P. Geigenberger, unpublished results) and the phloem of Ricinus seedlings (P. Geigenberger and U. Schurr, unpublished data). Reduction of DAP to Gly3P provides another way of accumulating a finite amount of reducing equivalents without acidification. There was a marked increase of Gly3P (Figure 4g) when external [O2] was decreased from 21% to 12%. Accumulation of lactate (Figure 4h) and GABA (Figure 4i) commenced when the external [O2] fell below about 8%. The relatively large amounts of lactate at the start of the experiment (identical to the siliques incubated in 21% [O2], data not shown) indicate that the seeds may have been previously exposed to periods of low [O2]. Ethanol could not be measured, because it was lost during the lyophilisation. In contrast to the seeds, alanine, lactate and GABA did not increase in the silique walls until external [O2] was below 1%.

Based on studies in growing potato tubers, Geigenberger et al., 2000) concluded that falling [O2] leads to (i) a restriction of glycolysis and respiration which decreases the adenylate energy status; (ii) a widespread decrease in biosynthetic activity which decreases ATP consumption; and (iii) a switch to pathways which consume less ATP. They proposed that this represents a metabolic adaptation to decrease O2 consumption and prevent the tissue from driving itself into anoxia. It was clearly separated from the inhibition of cytochrome oxidase and switch to fermentation, which does not occur until much lower [O2] concentrations. Our results provide preliminary evidence for a similar protective response in Arabidopsis seeds.

First, the changes of metabolites and fluxes reveal a progressive restriction of respiration and biosynthetic activity in the seeds as the [O2] surrounding the silique is decreased. The proportion of label in the seeds that remained in sugars rose (Figure 5a), revealing a progressive inhibition of metabolism. Although it was impossible to separate 14CO2 released from the seeds and the silique wall, the progressive decrease of ATP in the seeds (see above) provides indirect evidence that seed respiration starts to decline at relatively high external [O2]. Label decreased in starch (Figure 5b) when external [O2] decreased below 21%, in lipids (Figure 5c) when external [O2] fell below 12%, in organic acids (Figure 5d) when external [O2] fell below 21%, and in the cell wall fraction (Figure 5e) when external [O2] decreased below 12%. ADPGlc (Figure 4j) decreased sharply when [O2] decreased below 4% and AcCoA (Figure 4k) decreased 50% when external [O2] was decreased below 12%. These changes were much less marked in silique walls, where ADPGlc did not decrease until [O2] was decreased below 1% and AcCoA remained high even in zero O2. Labelling of protein did not decrease until [O2] decreased below 4%, and labelling of amino acids (Figure 5d) rose to a slight maximum in 8% O2 and fell in zero O2. The increase of label in amino acids at intermediate [O2] may reflect the accumulation of alanine (see above). The sharp decrease of protein and amino acids when extenal [O2] is decreased under 4% corresponds with rapid lactate accumulate, and may reflect a switch to anoxic metabolism. Direct measurements of the internal [O2] will be needed to confirm this interpretation.

Second, PPi levels were high in low external [O2] (Figure 4l), both in absolute terms and (Figure 4m) relative to ATP. PPi is formed during the synthesis of proteins, lipid and carbohydrates and is used in the cytosol of plant cells as an alternative energy donor to ATP during sucrose degradation, in glycolysis and for tonoplast energisation (Stitt, 1998). Recycling of PPi might allow O2 consumption to be decreased. PPi also remains high relative to ATP when pea roots, Arum maculatum spadices (Dancer and ap Rees, 1989) and potato tubers (Geigenberger et al., 2000) become hypoxic. UDPGlc (Figure 4n), which is produced by sucrose synthase, remained high until external [O2] fell below 8% and then fell sharply. Glc1P, Glc6P and Fru6P (Figure 4o-q) and 3PGA (Figure 4r) rose when external [O2] was decreased to 12%, decreased at lower external [O2], and increased slightly in zero O2. This is consistent with respiration being inhibited already at relatively high [O2] (see also above), with superimposed changes due to an inhibition of carbohydrate mobilisation [O2] decreases further.

Use and advantages of the enzyme cycling-based metabolite assay platform

We have established an analytic platform that allows key metabolites of primary carbon metabolism to be assayed with high sensitivity and throughput. The assays are a major advance on previous detection methods. For example, the previous assay for ADPGlc by anion exchange chromatography followed by UV detection (Geigenberger et al., 1997) required a HPLC system with cooled autosampler and sensitive UV detection systems, lasted 30 min per individual sample, and the levels found in plant extracts usually lay close to the limit of detection and in many tissues including leaves were obscured by other compounds absorbing in the same spectral range. The previous assay of PPi involved an enzymatic assay in which Fru6P was converted to Fru1,6bisP by pyrophosphate-dependent 6-phosphofructose-1-kinase, and the Fru1,6bisP formation then coupled to NADH oxidation by sequential action of aldolase, TPI and Gly3PDH (Merlo et al., 1993), and was tedious, operated close to the limits of detection and contamination, and required a double wavelength photometer. Other metabolites like Gly3P, various phosphorylated intermediates are usually present at higher concentrations and do not pose such technical problems, but our new assays nevertheless allow them to be measured without needing an expensive high performance spectrophotometer, and permit analysis in smaller samples than previously. Even in cases where a sensitive assay was previously available, the new assays have some advantages. For example, ATP can be assayed by highly sensitive luciferase assay, but this has the disadvantage that ADP can only be measured by subtraction from ATP, which leads to errors in tissues where the ATP/ADP ratio is high, whereas in the enzymatic cycling assays ATP and ADP are assayed independently of each other.

To avoid artefacts, precautions are needed when applying these techniques. First, it is essential to use an appropriate method to harvest the material and quench metabolism. Due to the rapid turnover of intermediates in primary metabolism (see above), serious artefacts can occur if the external conditions are transiently altered during the harvesting procedure or if the quenching process is too slow. Second, an adequate extraction technique is required. The reliability of the extraction procedure should always be checked by spiking samples of plant material with small representative amounts of each metabolite. Third, the reliability of each individual assay must be assessed. In initial experiments, aliquots of extracts should be spiked with representative amounts of each metabolite to provide a quantitative check on the reliability of the assay. Once the assay is established, it is essential to include a standard curve with each batch of assays. A dilution series should be included for each extract where possible, as this provides an excellent check that the various steps of the assay have not been saturated. Possible errors in individual assays can be detected by checking that the assay blank lies in the normal range and the reactions are linear.

Experimental procedures

Plant material

Arabidopsis thaliana (var. Columbia) was grown in a growth chamber (average 250 µmol photons m−2 s−1 light intensity, 8 h light/16 h dark, 23/20°C, 50/60% humidity day/night regime). Siliques were used from 12- to 14-week-old-plants. To vary external [O2], portions of the inflorescence containing siliques from positions 10–30 from top were collected into 50 ml Falcon tubes (approximately 200 siliques per tube) and incubated for 2 h at 23°C in the dark under continuous aeration using premixed gases containing 0, 1, 4, 8, 12 and 21% oxygen. The gas mix was pre-saturated with water by bubbling through water. To quench metabolism, the lid was removed and the Falcon tube filled within 1 sec with liquid nitrogen. Samples were stored at − 80°C.

Separation of seeds from silique walls

Intact siliques were lyophilised at − 60°C and dissected into seeds and envelopes under a binocular, using a custom made vacuum sampler consisting of a fine pipette tip attached to a microtube, attached to a vacuum. Tweezers were used to hold the silique wall while the pipette tip was used to open the silique along the replum and concomitantly to aspire the seeds. The narrow aperture of the pipette tip allowed seeds but not silique walls to be collected. Cross contamination was not visible.

Preparation of samples for the analysis of metabolites

The fractions (1–2 mg DW) were extracted with 500 µl 16% v/v trichloracetic acid (Geigenberger et al., 1998). Recovery of metabolites through the extraction and storage procedures has been documented (Farréet al., 2000; Geigenberger et al., 1994; Hajirezaei et al., 1994; Jellito et al., 1992; Merlo et al., 1993).

Reagents for metabolite analysis

Enzymes for biochemical analysis were purchased as suspensions in 3.2 m ammonium sulfate (when available), from Roche (Mannheim, Germany) except UGPase (Sigma). Enzymes for molecular biology were from Roche, Sigma-Aldrich (Taufkirchen, Germany), MBI Fermentas (St. Leon-Rot, Germany) and Stratagene (Amsterdam, The Netherlands). Chemicals were purchased from Sigma, except NAD+, NADH, NADP+, NADPH and phosphoenolpyruvate (Roche).

Expression and purification of ADPGlc pyrophosphorylase. Specific primers (Forward 5′-ACGCGTCGACATGCTTAGTTTAGAGAAGAACGATCAC and Reverse 5′-CCATTTCTGCAGTTATCGCTCCTGTTTATGCCCTAAC) were designed for the GLGC gene of E. coli coding for AGPase. The primers contained in each end a SalI and a PstI restriction site to clone the amplification product (1300 bp) of the PCR reaction (Pfu Polymerase, Stratagene) into the pQE-9 expression vector (Qiagen). The E. coli strain M15:pREP4 (Qiagen) was transformed with the recombinant vector and was grown at 37°C in batch cultures of 500 ml LB (amp, kan) medium to an OD of 0.6. Induction of protein expression (0.8 mm IPTG) was then carried out for further 2 h and the cells were then harvested by centrifugation (4000 g, 5 min). The cells were re-suspended in 8 ml breaking buffer (20 mm Hepes-KOH pH 7.6 5 mm MgCl2, 5 mm Imidazole, 10% (w/v) sucrose) containing one tablet of protease inhibitors (Roche Catalog no. 1836170). The cells were homogenised with a French press, centrifuged (50000 g, 30 min, 4°C) to release all soluble proteins, the supernatant loaded into a pre-equilibrated nickel-agarose column (Qiagen) coupled to a UV detector (260 nm) to monitor protein binding and elution, enabling the precise collection of one sharp protein peak. Elution was performed with 20 mm Hepes-KOH pH 7.6, 5 mm MgCl2, 25 mm imidazole, 10% w/v sucrose and elution of the His-tagged AGPase enzyme with 20 mm Hepes-KOH pH 7.6, 5 mm MgCl2, 250 mm imidazole, 10% w/v sucrose. The eluted fraction was dialysed for 10 h at 4°C in 20 mm Hepes-KOH pH 7.6, 5 mm MgCl2 20% w/v sucrose. The recombinant AGPase enzyme precipitated during dialysis and was collected by centrifugation (5000 g, 10 min, 4°C) and the pellet was re-suspended in refolding buffer (100 mm Tricine pH 7.8, 10 mm MgCl2, 1 m betaine). The activity of the AGPase was adjusted to 20 units ml−1 and the enzyme was frozen in liquid nitrogen and stored at − 80°C until use in the assays. The purified enzyme migrated in a denaturing SDS gel as a single 50 kDa band and was 98% pure as judged by Coomassie stain (not shown). The enzyme was free from other contaminating activities (e.g. UGPase) as judged in activity assays (not shown).

Expression and purification of pyruvate phosphate dikinase (PPDK)

The clone for a 6× His-tagged PPDK (Saavedra-Lira et al., 1998) (BBA 1382 p47–54) was kindly provided by Ruy Perez-Montfort (UNAM, Institute Fisiol. Celular, Mexico City, DF, Mexico). PPDK from E. histolytica was cloned into PET23b expression vector (Novagen) as a NheI fragment and introduced into E. coli strain BL21DE3pLysS. Expression and purification was carried similarly to AGPase except that dialysis was omitted and 10% glycerol was included in the refolding buffer to make the enzyme freezing stable. The purified enzyme migrated in a denaturing SDS gel as a single 100 kDa band and was 90% pure as judged by Coomassie stain (not shown). The enzyme was free from other contaminating activities (e.g. pyruvate kinase) as judged in activity assays (not shown).

Determinations based on the glycerol-3-phosphate cycling assay

All assays were carried out in tricine/KOH (pH 8) containing MgCl2 (10 mm in the 200 mm buffer stock solution). Assays were prepared using standard hand pipettes or a Multiprobe II pipetting robot (Packard, Dreieich). Multichannel hand pipettes often gave unacceptable variation. Reactions were run at 30°C, and absorbance monitored at 340 nm in an Anthos htII microplate reader. Reaction rates were calculated using Biolise software.

Glycerol-3-phosphate. Aliquots of extract (10 µl) or Gly3P standard (0–20 pmol) were added to 40 µl 200 mm buffer, heated (20 min, 95°C) in closed microtubes to destroy DAP, cooled on ice, centrifuged (30 sec), the supernatants transferred to a 96-well microplate, mixed with 40 µl 50 mm buffer containing 2 units Gly3POX, 130 units catalase, 0.4 unit Gly3PDH and 0.12 µmol NADH, and absorbance read for 20 min.

ATP and 3-phosphoglycerate. Aliquots of extracts (5 µl) or standards (0–50 pmol) were dispensed directly into a microplate, followed by 95 µl 100 mm buffer containing 2 units Gly3POX, 130 units catalase, 0.4 unit Gly3PDH, 0.05 unit TPI, 0.12 µmol NADH and 1 µmol 3PGA or 0.5 µmol ATP, absorbance followed for 20 min, 0.05 unit of 3-phosphoglycerate kinase and 0.05 unit of GAPDH added (both in 1 µl 100 mm Tricine buffer), and absorbance monitored for another 20 min.

ADP. In 1.5 ml microtubes, 60 µl Tricine buffer containing 0.1 unit PPDK, 0.05 unit pyrophosphatase, 1 µmol inorganic phosphate and 2 µmol pyruvate was added to 20 µl aliquots of extracts or ADP standard (0–50 pmol), incubated 40 min at room temperature, heated (5 min, 95°C), cooled, centrifuged, the supernatants transferred to a microplate, 20 µl of 100 mm buffer containing 0.05 units glycerokinase, 2 units Gly3POX, 0.4 unit Gly3PDH, 130 units catalase, 0.12 µmol NADH, 0.25 µmol glycerol and 50 pmol ADP added, absorbance read for 20 min, 1 unit myokinase (in 1 µl 100 mm Tricine buffer) added, and absorbance monitored for a further 20 min ADP was added because the conversion of ADP to Gly3P was not linear until about 50 pmol ADP was present in the assay.

UDP-glucose. Aliquots of extracts (5 µl) or UDPGlc standards (0–50 pmol) were dispensed directly into a microplate, followed by 95 µl of 100 mm buffer containing 0.05 unit glycerokinase, 2 units Gly3POX, 130 units catalase, 0.4 unit Gly3PDH, 0.12 µmol NADH, 0.15 µmol NaF and 0.02 µmol PPi, absorbance read for 20 min, 0.05 unit of UDP-glucose pyrophosphorylase added (in 1 µl 100 mm Tricine buffer), and absorbance monitored for a further 20 min.

PPi. Aliquots of 10 µl extracts, as well as PPi standards (ranging from 0 to 20 pmol) were dispensed into a microplate, 70 µl of a 100-mm Tricine buffer containing 0.05 unit glycerokinase, 2.5 units Gly3POX, 130 units catalase, 0.05 unit TPI, 0.1 unit GAPDH, 0.15 µmol NaF, 0.05 µmol EDTA, 0.05 µmol EGTA, 0.25 µmol glycerol, 0.01 µmol NAD+ and 0.1 µmol sodium arsenate added, incubated for 40 min incubation at room temperature, 20 µl of 100 mm Tricine buffer containing 0.4 unit Gly3PDH and 0.12 µmol NADH added, absorbance read for 20 min, 0.05 unit UGPase and 1 mmol UDPGlc added (both in 1 µl 100 mm Tricine buffer), and absorbance then monitored for a further 20 min.

ADP-glucose.  Aliquots of 20 µl extracts or ADPGlc standards (0–20 pmol) were dispensed into a microplate, 60 µl Tricine buffer containing 0.05 unit UGPase, 0.05 unit glycerokinase, 2.5 units Gly3POX, 130 units catalase, 0.05 unit TPI, 0.1 unit GAPDH, 0.02 µmol PPi, 0.15 µmol NaF, 0.25 µmol glycerol, 0.01 µmol NAD+ and 0.1 µmol sodium arsenate added, incubated for 40 min at room temperature, 20 µl of a solution of glycerol-3-phosphate dehydrogenase and NADH added (final concentrations of, respectively, 4 units ml−1 and 1.2 mm), absorbance read for 20 min, 0.03 unit AGPase added (in 1 µl 100 mm Tricine buffer), and absorbance monitored for another 20 min

Determinations based on the NADP+ cycling assay

Hexose phosphates were assayed by the method of Lowry and Passonneau, 1972), in which they are stoichiometrically converted into NADPH in the presence of saturating NADP+ and glucose-6-phosphate dehydrogenase, plus phosphoglucomutase or phosphoglucose isomerase for the determination of glucose-1-phosphate or fructose-6-phosphate, respectively. The remaining NADP+ is destroyed with alkali before determining NADPH by a cycling assay. The NADPH cycling assay was adapted from Nisselbaum and Green (1969), where NADPH is alternatively reduced by glucose-6-phosphate dehydrogenase and oxidised by phenazine methosulfate (PMS) in the presence of thiazolyl blue (MTT), yielding formazan. The buffer used in all procedures was tricine/KOH pH 9 containing MgCl2 (10 mm in the 200 mm buffer stock solution).

Glucose-6-phosphate.  Aliquots of 5 µl of extracts or Glc6P standards (0–50 pmol) were dispensed into 1.5 ml microtubes, 50 mm buffer containing 0.02 unit and 6.5 nmol NADP+ added, incubated for 20 min at room temperature, 20 µl 0.5 m NaOH added, heated at 95°C for 5 min, cooled on ice, centrifuged for 30 sec and the contents transferred to a 96-well microplate containing 20 µl of 0.5 m HCl in buffer. After mixing, 50 µl buffer containing 1.2 units glucose-6-phosphate dehydrogenase, 0.3 µmol glucose-6-phosphate, 0.5 µmol EDTA, 0.04 µmol PMS and 0.1 µmol MTT was added and absorbance read at 570 nm for 10 min.

Glucose-1-phosphate.  Aliquots of 20 µl of extracts or Glc1P standards (0–20 pmol) were disposed into 1.5 ml microtubes containing 20 µl 50 mm buffer containing 0.2 unit glucose-6-phosphate dehydrogenase and 6.5 nmol NADP+, incubated 20 min incubation at room temperature, 10 µl of 0.25 m HCl was added and incubated for 10 min at room temperature. This incubation removed Glc6P and destroyed the resulting NADPH. After neutralising with 10 µl of 0.25 m NaOH in buffer, 20 µl of 50 mm buffer containing 0.2 unit glucose-6-phosphate dehydrogenase, 0.04 unit phosphoglucomutase and 0.5 nmol glucose-1,6-bisphosphate were added, incubated 20 min at room temperature, 20 µl of NaOH 0.5 m was added and the tubes were heated at 95°C for 5 min. After cooling on ice the tubes were centrifuged for 30 sec and transferred to a 96-well microplate containing 20 µl of HCl 0.5 m in buffer. After mixing, 50 µl 200 mm buffer containing 1.2 units glucose-6-phosphate dehydrogenase, 0.3 µmol glucose-6-phosphate, 0.5 µmol EDTA, 0.04 µmol PMS and 0.1 µmol MTT were added and absorbance read at 570 nm for 10 min.

Fructose-6-phosphate. Aliquots of 10 µl extracts or Fru6P standards (0–50 pmol) were disposed into 1.5 ml micro tubes containing 20 µl 50 mm buffer containing 0.2 unit glucose-6-phosphate dehydrogenase and 6.5 nmol NADP+, incubated 20 min at room temperature, 10 µl 0.25 m HCl added, and incubated 10 min at room temperature to remove Glc6P and destroy the resulting NADPH. After neutralising with 10 µl of 0.25 m NaOH in buffer, 20 µl of 50 mm buffer containing 0.2 unit glucose-6-phosphate dehydrogenase and 0.04 unit phosphoglucose isomerase were added, incubated 20 min at room temperature, 20 µl of NaOH 0.5 m added, the tubes heated at 95°C for 5 min, cooled on ice, centrifuged for 30 sec and transferred to a 96-well microplate containing 20 µl of HCl 0.5 m in buffer. After mixing, 50 µl of 200 mm buffer containing 1.2 units glucose-6-phosphate dehydrogenase, 0.3 mol glucose-6-phosphate, 0.5 µmol EDTA, 0.04 µmol PMS and 0.1 µmol MTT were added, and absorbance read at 570 nm for 10 min.

Determination of acetyl coenzyme A and coenzyme A

AcCoA and CoASH were determined as described in Bergmeyer (1987) except that the volumes were scaled down. Aliquots of 20 µl extract (40 µl when determining also CoASH), as well as AcCoA standards (ranging from 0.5 to 10 pmol) were disposed into a microplate and 0 or 1 µmol of N-ethylmaleimide was added (20 µl) in order to remove CoASH. After 10 min incubation, 100 µl of 200 mm Tris/HCl pH 7.4 containing 0 or 1 µmol DTT and 20 µmol malate were added and the plate was incubated for 15 min at room temperature. Finally 50 µl of a mixture containing 0.5 µmol DTT, 0.92 µmol acetyl phosphate, 0.28 µmol NAD+, 2.8 units phosphotransacetylase, 1 unit citrate synthase and 2 units malate dehydrogenase in Tris/HCl 100 mm were added and absorbance was read at 340 nm for 10 min.

Determination of lactate

Aliquots of 50 µl extract or standards (ranging from 0 to 20 nmol) were disposed into a microplate and 100 µl of 100 mm Tricine/KOH pH 8.5 containing 0.1 µmol MTT, 0.04 µmol PMS, 0.5% v/v Triton X100 were added. After reading the absorbance at 570 nm for 5 min, 50 munits of lactate oxidase were added and the absorbance was red for 5 more min.

Determination of amino acids

Amino acids were determined by HPLC as described in Geigenberger et al. (1996).

Labelling experiments

Each silique was injected with 5 µl [U-14C] sucrose 2 mm (specific activity 740 kBq µmol−1) with a 5-µl Hamilton syringe and then incubated for 2 h in the conditions described above, except that a vial containing 5% w/v KOH was added in order to trap the radioactive respiratory CO2. The siliques were then harvested in liquid N2, lyophilised and dissected as described above. The ground seeds (1.5 mg DW) were extracted twice with 1 ml 80% v/v ethanol at 80°C for 30 min, and re-extracted with 1 ml 50% v/v ethanol at 80°C for 30 min. Then 1 ml of chloroform and 1 ml water were added to the combined supernatants and the extracts were vigorously shaken. The final supernatants (ethanol/water) were transferred into a small tube and dried under an air stream at 45°C, taken up in 1 ml H2O (soluble fraction), and separated into neutral, anionic, and basic fractions by ion-exchange chromatography (Geigenberger et al., 1997). The chloroform fraction was dried and counted for total lipids. The insoluble material (starch, proteins, cell wall) left after ethanol extraction was analysed for label in starch, protein and cell walls as in Merlo et al. (1993).

Acknowledgements

Thanks are due to Tobias Haas for help in the expression and purification of ADPGlc pyrophosphorylase.

Ancillary