Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Present addresses: *Laboratory for Molecular Plant Physiology, Institut for Plant Biology, University Zurich, Zollikerstrasse 107, Zurich, Switzerland and Recherche et Développement des Plantes, ENS-Lyon, 46, allée d’Italie, 69364 Lyon Cedex 07, France.
In order to cope with phosphate deficiency, white lupin produces bottle-brushed like roots, so-called cluster or proteoid roots which are specialized in malate and citrate excretion. Young, developing cluster roots mainly excrete malate whereas mature cluster roots mainly release citrate. Mature proteoid roots excrete four to six times more carboxylates compared with juvenile proteoid roots. Using a cDNA-amplified restriction fragment length polymorphism (AFLP) approach we identified a gene coding for a putative ATP-citrate lyase (ACL) up-regulated in young cluster roots. Cloning of the lupin ACL revealed that plant ACL is constituted by two polypeptides (ACLA and ACLB) encoded by two different genes. This contrasts with the animal ACL, constituted of one polypeptide which covers ACLA and ACLB. The ACL function of the two lupin gene products has been demonstrated by heterologous expression in yeast. Both subunits are required for ACL activity. In lupin cluster roots, our results suggest that ACL activity could be responsible for the switch between malate and citrate excretion in the different developmental stages of cluster roots. In primary roots of lupin and maize, ACL activity was positively correlated with malate exudation. These results show that ACL is implicated in root exudation of organic acids and hence plays a novel role in addition to lipid synthesis. Our results suggest that in addition to lipid biosynthesis, in plants, ACL is implicated in malate excretion.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Phosphorus (P), is one of the most limiting nutrients in natural and agricultural ecosystems, due to the low concentrations of free inorganic phosphate (Pi) in the soil solution, which is the sole form of phosphorus absorbed by plant roots (Marschner 1995). The strategies of most plants to overcome phosphate starvation are (i) the synthesis of high affinity transporters, which can absorb Pi even if the concentration in the soil solution is very low; (ii) modifications in root morphology and the association with mycorrhiza, mainly improving spatial acquisition of available phosphate; (iii) the secretion of phosphohydrolases which liberate Pi by hydrolysis of organic P forms; (iv) the acidification of the rhizosphere by enhanced net extrusion of protons for mobilization of acid-soluble P fractions in calcareous soils; and (v) the excretion of carboxylates as organic chelators, mobilizing sparingly soluble Fe, Al and Ca phosphates by complexation of the metal cations. Rhizosphere acidification is a widespread response to phosphorus deficiency, particularly in dicotyledonous plants. In contrast, excretion of considerable amounts of carboxylates effective in phosphate solubilization seems to be restricted to a limited number of plant species (Marschner 1995; Neumann & Römheld 1999, Neumann & Martinoia 2002). White lupin and members of the Proteaceae are able to excrete far larger amounts of carboxylates than any other plant species investigated so far (Dinkelaker et al. 1997; Neumann & Martinoia 2002). Compared with rhizosphere acidification, this strategy has the advantage that phosphate is not only liberated from acid-soluble P fractions, but also by anion exchange from Fe–Al–P complexes found in many acidic soils. In these plant species, exudation of organic acid anions occurs at well-defined parts of roots, where short, clustered lateral roots (proteoid or cluster roots), are formed. White lupin is the best-described system forming proteoid roots. Formation of proteoid roots follows a well-defined developmental pattern. Young, growing proteoid roots release mainly malate and only low amounts of citrate whereas immature proteoid roots excrete similar amounts of citrate and malate. In contrast mature proteoid roots excrete far higher amounts of carboxylates, mainly citrate and strongly acidify the rhizosphere. This is associated with a shift from malate to citrate accumulation in the root tissue with increasing age of the clusters (Neumann et al. 1999, 2000). The high demand for carbon used for the synthesis of carboxylates is sustained by the up-regulation of several enzymes, i.e. PEPcarboxylase (Johnson et al. 1996a; Johnson, Vance & Allan 1996b), sucrose synthase, phosphoglucomutase and fructokinase which are involved in the glycolytic pathway (Massonneau et al. 2001).
Although up-regulation of these enzymatic activities can explain increased exudation of carboxylates from cluster roots in general, the reasons for the differential exudation of malate or citrate in different root zones (juvenile and mature clusters) still remain an open question. Using the cDNA-amplified restriction fragment length polymorphism (AFLP) technique, we identified a large number of genes with putative differential expression in young and mature proteoid roots. Among these genes, we also identified a putative ATP-dependent citrate lyase (ACL). ACL (EC 184.108.40.206) catalyses the formation of acetyl-CoA and oxaloacetate from citrate and CoA with a concomitant hydrolysis of ATP. Due to its instability, the study and cloning of ACL has been delayed and it was only in 1990 that the first animal ACL was cloned (Elshourbagy et al. 1990). Several reports on the plant ACL have been published and its activity has been found mainly in young growing parts and during seed maturation (Kaethner & ap Rees 1985; Ratledge, Bowater & Taylor 1997). In developing seeds of Brassica napus, Ratledge et al. (1997) found that most of the ACL activity was localized in chloroplasts and established a correlation between ACL activity and the rate of lipid synthesis. In other plants, cell fractionation studies showed that ACL was exclusively localized in the cytosol (Kaethner & ap Rees 1985; Ma et al. 2001) and a role in terpenoid synthesis in response to pathogens was postulated (Suh et al. 2001).
Despite the importance of ACL, to our knowledge, plant ACL has yet not been cloned. Molecular identification of ACL will provide the opportunity to learn more about the role of the corresponding gene product in plant metabolism. Indeed, as shown in this report, cloning of lupin ACL revealed that in contrast to animals, plant ACL is constituted by two different subunits and that in addition to its well-established role in lipid synthesis, it may also play a role in malate exudation in roots.
MATERIALS and METHODS
Growth of white lupin (Lupinus albus L. cv. Amiga; Südwestdeutsche Saatzucht, Rastatt, Germany) in presence (+P) or absence (–P) of P source has been previously described (Massonneau et al. 2001). Maize (Zea mais L. cv. Delprim; Delley Semences et Plantes SA, Yverdon, Switzerland) was pre-germinated for 4 d and grown hydroponically in complete (+P) medium. Plants were grown at 22 °C and 65% relative humidity with a light period of 16 h at 200 µmol m−2 s−1.
Harvest of different root parts
The different stages of proteoid roots were harvested as described by Massonneau et al. (2001). The apex (1 cm) of fast-growing secondary roots (named N2 in our previous studies) are harvested on secondary roots holding clusters. In order to differentiate the developmental stages of root clusters, the root system was immersed in a pH-indicator solution, which indicates acidification in mature cluster regions (Neumann et al. 1999). In experiments in which organic acid excretion was measured, root segments were washed twice for 1 min in the bath solution. Demonstration that organic acid contents measured in the bath solution corresponds to excretion and not to organic acid release through the cut surface has been described in Neumann et al. (1999) for proteoid roots and by Ryan, Delhaize & Randall (1995) for normal root tips.
Collection and analysis of exudates
Excised roots were rinsed in water and incubated for 1 h in water containing penicillin (Massonneau et al. 2001). Malate contents were determined using the Malate Test Kit (Boehringer, Mannheim, Germany).
Heterologous expression of the ACL in yeast
Saccharomyces cerevisiae W303 (MAT-a ade 2–1 can 1–100 hi 3–11 leu 2–3 trp 1–1 ura 3–1) were grown on appropriate selective media and transformed via the Li-acetate protocol (Lundblad 1997). La ACLA and La ACLB full-length cDNAs in pBluescript were excised with XhoI and NotI and ligated in the pNEV vectors (Sauer N. et Scholtz J. 1994) containing Ura, respectively, Leu as markers. This system allowed the constitutive expression by the pma-1 promoter (Villalba et al. 1992). Transformed yeast cells were grown to OD 4–5 in selective media and subsequently transferred to YPD medium for 1 h 30 min before preparing spheroplasts (Lundblad 1997).
Extraction of soluble proteins and ACL assay
Frozen plant tissues were ground in liquid nitrogen and homogenized with 3 vol. of extraction buffer (0·1 m Hepes-KOH pH 7·5, 5 mm MgCl2, 2·5 mm dithiothreitol, 3 mm Na-DEDTC (diethyldithiocarbamate), 1 mm eethylenediaminetetraacetic acid, 1 mm benzamidine, 1 mm phenylmethanesulfonylfluoride, 3% polyvinylpolypyrrolidone K30).
Yeast spheroplasts were lysed osmotically and gently homogenized at 4 °C in 1·5 vol. of extraction buffer (complemented with Na-Citrate to a final concentration of 5 mm) using a syringe and needle. After centrifugation (25 min, 12 000 × g, 4 °C) the supernatant was rapidly used to determine the ACL activities and protein concentrations (DC Protein Assay kit; Bio-Rad).
ACL (EC 4·1.3·8) activity was determined spectrophotometrically at room temperature, using the malate dehydrogenase coupled assay. The assay mixture contained 0·2 m Tris pH 8·4, 10 mm MgCl2, 10 mm 2-mercaptoethanol, 20 mm Na3-Citrate, 0·2 mm Coenzyme A, 10 mm ATP (omitted in blanks), 0·2 mm NADH, 0·4 U mL−1 malate dehydrogenase. Values were taken after 30 min and 10 min for plants and yeast, respectively. These time points were chosen since prolonged incubation resulted in the inactivation of the enzyme. Blanks were performed by omitting ATP or Coenzyme A, resulting in similar values.
The library was constructed from mRNA isolated from juvenile cluster roots. cDNAs synthesis and cloning were conducted using the ZAP-cDNA Synthesis Kit and ZAP-cDNA Gigapack III Gold Cloning Kit (catalogue no. 200400 and catalogue no. 200450; Stratagene, Amsterdam, The Netherlands). The screening of La ACLA was performed using a polymerase chain reaction (PCR) approach. Primers were deduced from the sequence of the cDNA-AFLP clone showing homology with the C-terminal part of the rat ACL. Primers 5′GTGCCATTGATGATGCTGCT3′ and 5′TGTTTGCCTTCGTGAGAGTG3′ amplified a 249 bp fragment. Six independent clones have been completely sequenced and showed approximately the same size, the longest being 2222 bp.
To screen La ACLB, degenerated primers have been designed from consensus regions between cotton, alfalfa, soybean, tomato and Arabidopsis thaliana. 5′TGAGYTA GTARASARRGARCCNTGG3′ and 5′ACCICCICC NGCNACCATNGTCCA3′ amplified a fragment from which specific primers were deduced to screen the library. Six independent clones were isolated and completely sequenced. They showed approximately the same size, the longest was 1618 bp.
Reverse transcriptase-polymerase chain reaction
One million copies of pAW109 mRNA (Perkin Elmer, Applied Biosystems, Foster City, CA, USA) was mixed with 1 µg of DNA-free total RNA. Reverse transcriptase (RT)-PCR was performed as described by Massonneau et al. (2001) using (alpha)33P-labelled dATP. One-hundredth of the RT reaction product was used to perform PCR. Two sets of primers were used in the same tube, one specific for pAW (AW112 and AW113, amplifying a 301 bp fragment) and one for La ACLA (5′GCTGGAGCTAAG AGTGGTGG3′ and 5′CTGTCACGAGCATCCTTGAA3′ amplifying a 600 bp fragment) or La ACLB (5′CGGGTC CATAAACCTCAATG3′ and 5′TGGTCGATGGAAAG CCTTAT3′ amplifying a 603 bp fragment). The annealing temperature was 60 °C. After 24 cycles, when strict proportionality between cycle number and amplification was observed for all tissues studied, the PCR fragments were separated on an agarose gel and blotted onto a nylon membrane for autoradiography. An additional control was performed using the ribosomal protein 60S as probe.
One of our interests is to understand the mechanisms leading to the exudation of the huge amounts of carboxylates by mature proteoid roots and how the cellular metabolism changes during the conversion of young, growing proteoid roots to mature proteoid roots. In order to identify stage-specific expression of mRNA we performed a cDNA-AFLP analysis (Bachem et al. 1996) with juvenile, mature and senescent root clusters. In a former contribution we showed that several enzymes implicated in glycolysis, namely fructokinase, sucrose synthase and phosphoglucomutase were up-regulated in juvenile and mature cluster roots (Massonneau et al. 2001). Among the genes identified to be differentially expressed, we also identified a putative ATP-dependent citrate lyase. ACL catalyses the reaction which forms acetyl-CoA and oxaloacetate from citrate, CoA and ATP.
The length of the putative ACL clone was 278 bp and exhibited 71% similarity with the rat ACL over a stretch of 92 amino acids. Transcript analysis using semi-quantitative RT-PCR confirmed that the sequence identified was indeed highly expressed in juvenile proteoid roots and down-regulated in mature proteoid roots (see below).
Since, to our knowledge, no plant ACL has been cloned and functionally identified, we were interested to verify whether the cDNA-AFLP fragment corresponds to the ACL gene. PCR-based screening, using primers deduced from the putative ACL, of a lupin library made from juvenile proteoid roots, resulted in six independent clones, the longest being 2222 bp. Alignment of the lupin clone with the animal ACL (3303 bp) revealed that our clone was highly similar to the C-terminal half of the rat ACL. Comparison of the lupin sequence with the Arabidopsis database revealed two genes of similar size also corresponding to the C-terminal part of rat ACL (Table 1). Since these clones were nearly 1500 bases shorter than the full-length cDNA reported for rat, we performed a Northern blot. The size of the transcripts hybridizing with our probe was 2·3 kb, a size similar to the cDNA cloned for the lupin (not shown). A similarity search using the rat ACL gene revealed three putative Arabidopsis genes exhibiting a high homology to the N-terminal part of the rat ACL. These putative genes have no overlap with the genes coding for the C-terminal part mentioned above. ESTs from other plants corresponding to the N-terminal part of ACL are also present in the database. Using degenerated oligonucleotides corresponding to conserved parts of this putative subunit of ACL, we screened the lupin library and identified several clones of approximately 1600 bp. An alignment of the lupin cDNA with the Arabidopsis and rat genes shows that the deduced gene products are highly similar (Table 1). As we supposed that in plants the ACL activity is dependent on two gene products, we called the N-terminal ACLA and the C-terminal part ACLB (Table 1, Fig. 1). As in their animal counterparts, we have identified within the subunit A the putative CoA binding domain and a conserved histidine, which is phosphorylated during the enzymatic reaction, two ATP-dependent citrate/succinate-CoA ligase signatures in this subunit and one in subunit B (Fig. 1). Interestingly, both in fungi and bacteria, ACL exhibit a similar organization pattern as in plants (Nowrousian et al. 2000; Kanao et al. 2001). In comparison with the animal ACL, Chlorobium limicola and Lupinus albus miss a region of about 60 amino acids located between subunit B and A in animal ACLs.
Table 1. Comparison of Lupinus albus ATP citrate lyase (ACL) with ACL of Rattus norvegicus and putative ACLs of Arabiopsis thaliana
In order to prove that the two genes identified indeed code for the plant ACL, we performed expression studies. In a first attempt, we expressed the two gene products in Escherichia coli. On a sodium dodecyl sulphate-polyacrylamide gel electrophoresis gel, we observed high levels of the corresponding proteins, however, no activity could be detected, most probably due to the fact that the ACL gene products were only present in inclusion bodies and hence insoluble. Therefore we decided to use yeast as expression system. The genome of Saccharomyces cerevisae does not contain genes corresponding to the ACL and in fact no ACL activity could be detected in wild-type yeasts (Fig. 2). In contrast, when the two ACL subunits were expressed in yeast under the control of a constitutive promoter, ACL activity could be detected. Expression of the A or B subunit alone resulted in negligible ACL activities, demonstrating that both subunits are required for ACL activity.
After the demonstration that the heterologously expressed clones code for the ACL, we were interested to correlate the expression of ACL with the enzymatic activity of its gene product in the different root types and to investigate whether ACL influences carboxylate exudation. Both subunits showed a very similar expression pattern, exhibiting high transcript levels in young, growing parts of roots and very low in senescent roots (Fig. 3). Quantification using a phosphoimager confirmed that the expression patterns of both subunits approximately paralleled the ACL activity in the different root types (Fig. 4b). The slight differences from one experiment to the other, mainly observed between juvenile and immature cluster roots, could be due to the difficulty in clearly separating both tissues. ACL activity was highest in juvenile proteoid roots and decreased in immature, mature and senescent proteoid roots. Extracts of root apexes exhibited ACL activities comparable with those of immature proteoid roots. Since our results suggest that ACL activity is highest in young, growing tissues, we measured this activity in lupin leaves. Young leaves showed a similar activity to root apex, whereas the ACL activity of adult leaves was of the same order of magnitude as senescent proteoid roots (not shown). This result confirms former observations, that ACL activity is mainly present in young tissues (Kaethner & ap Rees 1985; Ratledge et al. 1997).
Cluster roots are specialized in excreting high amounts of carboxylates and a peak is observed for mature cluster roots (Massonneau et al. 2001). Comparing the different stages of proteoid roots we observed that similar amounts of malate were excreted in juvenile, immature and mature proteoid roots (Fig. 4a). In contrast, a strong increase in citrate release can be observed during proteoid root development. The ratio of excreted malate to citrate was by far the highest in juvenile roots, which exhibit the highest ACL activity (Fig. 4c). Our results indicated that ACL could be implicated in malate excretion.
We have therefore used a second experimental system to test our hypothesis. In this case we have used primary roots of lupins (not containing proteoid roots) and maize. The first 30 mm of the roots were cut in four sections, two apical sections of 5 mm and the next two sections of 10 mm. In lupin as well as in maize the largest amounts of malate are excreted in the 5–10 mm section. This corresponds also to the part of the root exhibiting the highest ACL activity and has been shown to be an actively growing zone under similar culture condition in maize (Peters & Felle 1999). Root sections behind the 5–10 mm section excrete reduced amounts of malate and exhibit lower ACL activities. This result indeed suggests that ACL could be implicated in malate excretion (Fig. 5).
A detailed correlation analysis between ACL activity and malate exudation gives further confirmation of this hypothesis (Fig. 6). The calculation results in a highly significant regression coefficient (r2 = 0·729) and P-value (0·0001 calculated on 15–2-values). The statistical approach demonstrates a very good correlation between ACL activity and malate exudation under the conditions used and in the range of values of ACL activity and malate exudation measured. It can therefore be concluded that increased ACL activity and malate exudation are highly correlated.
Root exudation of organic acids plays an important role in a large number of processes, such as deficiency of mineral nutrients, particularly phosphate, exposure to toxic metals, such as aluminium or lead and to hypoxia (Neumann & Römheld 1999; Ryan & Delhaize 2001; Neumann & Martinoia 2002). In addition, organic acid exudation can also selectively stimulate microbial activities in the rhizosphere with impact on the availability of nutrients. At least in some cases, exudation of carboxylates implicates the activation of a carboxylate channel (Kollmeier et al. 2001) and changes in cellular metabolism. In lupin, proteoid roots formed during phosphate starvation exhibit increased PEP carboxylase and citrate synthase activity and concomitant decreased aconitase activity (Neumann et al. 1999). In addition, increased activity of glycolytic enzymes suggests a high carbohydrates demand for biosynthesis of carboxylates (Massonneau et al. 2001). During the development of proteoid roots the pattern of exuded organic acids changes. In young, developing proteoid roots malate is the major carboxylic acid excreted. In contrast, mature proteoid roots excrete mainly citrate, which is more efficient in desorbing phosphate from Al–Fe–P complexes. In a cDNA-AFLP approach we have identified a cDNA coding for a putative ACL highly expressed in juvenile proteoid roots. Since the products of the ACL reaction are acetyl-CoA and oxaloacetate, which is readily reduced to malate, we have investigated the role of ACL in organic acid exudation.
Cloning of the full-length cDNA coding for the putative ACL revealed that in plants the ACL was not constituted by a single but by two different polypeptides corresponding to the N- and C-terminus of the animal ACL. A similar structure of the ACL has been postulated for fungi. However, the functional activity of these two subunits still remained to be proven. A very recent publication demonstrated that C. limicola exhibits a similar structure and that both subunits are required for ACL function. Our results demonstrate that this is also true for plants.
ACL activity measurements paralleled the expression pattern observed for both genes coding for the ACL, indicating that the transcript levels reflect the activity of the enzyme and that the enzyme is probably regulated at the transcriptional level. The fact that both subunits follow the same expression pattern implies a common regulation. However, the fact that ACL is encoded by two genes (ACLA and ACLB), may indicate that in plants additional regulatory mechanisms are involved compared with animals.
ACL activity has been extensively studied in animals and it has been shown to be a key step for lipid biosynthesis providing acetyl-CoA. A similar function has been postulated in rape seeds (Ratledge et al. 1997). In this case, ACL activity was localized in plastids, and the activity was correlated with lipid accumulation. In contrast, Kaethner & ap Rees (1985) localized pea ACL in the cytosol. A hint that ACL may be localized differentially according to the plant species was given by Rangasamy & Ratledge (2000), who found ACL activity predominantly in chloroplasts in rape and spinach, whereas in pea and tobacco, distribution was mainly cytosolic. The sequences of the lupin and Arabidopsis genes have no obvious transit peptide (search with PSORT, version 6·4) indicating that at least in these plants, ACL could be localized in the cytosol. Preliminary results using a GFP fusion protein also suggest a cytoplasmic localization. As membranes are considered to be impermeable to acetyl-CoA, it has to be used at its production place in the cytosol, for example, for synthesis of cytosolic terpenoids such as sterols or synthesis of malonylCoA used for flavonoid or ACC production, or transported to another site via an acetyl-CoA translocator. An acetyl-CoA translocator has already been cloned in animals (Kanamori et al. 1997; Bora, Kanamori & Hirabayashi 1999). However, an acetyl-CoA transporter has not been identified so far in plants and no homologue of animal transporters can be found in plant databases.
In lupin and maize, the highest ACL activities were found in growing tissues such as young leaves and elongation zones of roots. In maize, the root elongation rate is maximal 6 d after germination and the elongation zone was identified between 3 and 8 mm from the apex (Muller, Stosser & Tardieu 1998). The correlation of the highest ACL activities with the elongation zone suggests that one role of ACL may indeed be linked to lipid biosynthesis.
Investigations of ACL function have focused on the acetyl-CoA production and lipid biosynthesis. In contrast, little attention has been paid to the second compound produced during this reaction, oxaloacetate, which is readily reduced to malate. This is probably due to the fact that in animals the role of malate is not as complex as in plants (Martinoia & Rentsch 1994). In lupin proteoid roots PEPCase is highly expressed and thus constitutes a second pathway to produce malate (Fig. 7). Part of malate will be used for respiration, while the excess can be stored within the vacuole (Martinoia & Rentsch 1994). This strategy has been demonstrated mainly for leaves. In roots, both vacuolar storage as well as excretion have often been observed (Ryan & Delhaize 2001). Our results on primary roots of lupin and maize show that root parts exhibiting a high ACL activity excrete more malate. This may be due to the fact that in fast-growing zones a high demand for acetylCoA exists and that not all malate is used for respiration. A slightly different situation is present in proteoid roots, which have a specialized metabolism allowing them to excrete large amounts of carboxylates. In young, growing cluster roots the main organic acid excreted is malate. In a later fully developed stage, excretion is strongly increased and the main carboxylate released is citrate. This switch between malate and citrate excretion is illustrated by the change in the exuded malate/citrate ratio. A good correlation can be found between ACL activity and the ratio between the two organic acids, indicating that in proteoid roots, ACL plays a role as a metabolic switch. In mature proteoid roots ACL activity is decreased and the citrate removed from the Krebs cycle is no longer readily converted to malate. The cells produce large amounts of citrate, which cannot be used in metabolic pathways and are therefore forced to excrete citrate to maintain cellular homeostasis. This is reflected by a shift from preferential malate production in juvenile clusters to almost exclusive citrate production in mature and senescent clusters (Neumann et al. 1999, 2000). ACL activity, which links malate and citrate (Fig. 7), is therefore likely to be responsible for the switch in the organic acid excreted. The reduced ACL activity in mature proteoid roots favours the production and the release of citrate.
This may have important consequences from the ecological point of view. Consulting the complex formation constants with metals or mobilization of rock phosphate shows that citrate is much more efficient than malate (Ryan & Delhaize 2001). Exudation of citrate instead of malate may therefore improve P acquisition and the ability to cope with adverse soil chemical conditions such as Al toxicity. However, a mechanism of detoxification to avoid over-accumulation of organic acids in the root tissue may be regarded as the original function of intense root exudation of malate and citrate.
We would like to thank Jacqueline Moret from the Institut de Mathématiques, Université de Neuchâtel for her help in the statistical approach. This work was partially funded by the EU Biotechnology project ‘Phosphate and Crop Productivity’, EU Nr. BIO4CT960770 and BBW Nr. 96·0023-3 and ‘Metallophytes’ EU Nr QLK3-CT- 2000-00479, BBW Nr. 00·0413
Received 6 May 2002; received in revised form 17 June 2002; accepted for publication 17 June 2002