Low phosphorous availability, a common condition of many soils, is known to stimulate phosphatase activity in plants; however, the molecular details of this response remain mostly unknown. We purified and sequenced the N-terminal region of a phosphate starvation induced acid phosphatase (AtACP5) from Arabidopsis thaliana, and cloned its cDNA and the corresponding genomic DNA. The nucleotide sequence of the cDNA predicted that AtACP5 is synthesised as a 338 amino acid-long precursor with a signal peptide. AtACP5 was found to be related to known purple acid phosphatases, especially to mammal type 5 acid phosphatases. Other similarities with purple acid phosphatases, which contain a dinuclear metal centre, include the conservation of all residues involved in metal ligand binding and resistance to tartrate inhibition. In addition, AtACP5, like other type 5 acid phosphatases, displayed peroxidation activity. Northern hybridisation experiments, as well as in situ glucuronidase (GUS) activity assays on transgenic plants harbouring AtACP5:GUS translational fusions, showed that AtACP5 is not only responsive to phosphate starvation but also to ABA and salt stress. It is also expressed in senescent leaves and during oxidative stress induced by H2O2, but not by paraquat or salicylic acid. Given its bifunctionality, as it displays both phosphatase and peroxidation activity, we propose that AtACP5 could be involved in phosphate mobilisation and in the metabolism of reactive oxygen species in stressed or senescent parts of the plant.
Several of the biochemical changes induced in response to phosphate starvation, including the secretion of acids and the stimulation of phosphate transport and phosphatase activities, are common between organisms that acquire their phosphorous from the environment, including bacteria and fungi. Phosphatases have been used successfully as tools to investigate the molecular-genetic mechanisms underlying the adaptative responses of model prokaryotic and eukaryotic microorganisms to phosphate starvation (for reviews, see Lenburg & O’Shea 1996; Torriani 1990; Vogel & Hinnen 1990). In plants, the stimulation of phosphatase activities in response to phosphate starvation is well documented (reviewed by Duff et al. 1994 ; Trull & Deikman 1998), and recently an Arabidopsis mutant lacking one phosphate starvation induced (psi) phosphatase has been reported ( Trull & Deikman 1998). Psi plant phosphatases are acid phosphatases and in most cases do not show strong substrate specificity. The only known psi acid phosphatase displaying certain substrate specificity is a phosphoenol pyruvate (PEP)-phosphatase from Brassica nigra, which is thought to be one of the enzymes that circumvents the adenylate and Pi requiring steps of glycolysis ( Duff et al. 1989 ). Non-specific psi acid phosphatases are thought to be involved in Pi acquisition or recycling, depending on their cellular and subcellular location. Thus, psi phosphatases secreted from roots would function in Pi-scavenging from phosphate esters, present in the soil or released by plant cells. It has been postulated that these secreted phosphatases may participate in Pi transport, given their ability to bind Pi ( Duff et al. 1991 ; Goldstein et al. 1988 ; Lee 1988; Lefebvre et al. 1990 ). On the other hand, psi intracellular phosphatases may play a role in phosphate recycling ( Smyth & Chevalier 1984). Despite the wealth of reports on psi plant phosphatases, little is known about the corresponding proteins or the genes that encode them. One exception is the vegetative storage protein (VSP), a psi protein showing phosphatase activity, whose gene has been cloned ( Berger et al. 1995 ; De Wald et al. 1991 ; Sadka et al. 1994 ). However, given its limited phosphatase activity, the physiological relevance of this function remains to be demonstrated ( De Wald et al. 1991 ; Duff et al. 1994 ; Sadka et al. 1994 ).
In this study, we report on the purification of a psi acid phosphatase, AtACP5, from Arabidopsis thaliana grown under phosphate starvation conditions and on the cloning of its cDNA. AtACP5 shares the characteristics of type 5 acid phosphatases, a class of purple acid phosphatases so far described only in mammals, including sequence similarity, resistance to tartrate inhibition and peroxidation activity. We show that the level of its RNA, and the GUS activity of an AtACP5:GUS translational fusion, is increased in response to phosphate starvation, and to some other conditions that promote phosphate mobilisation and/or oxidative stress. These results are consistent with a role for this bifunctional protein in phosphate mobilisation and in the metabolism of reactive oxygen species.
Acid phosphatases from Arabidopsis thaliana responsive to phosphate starvation: purification of AtACP5
To identify acid phosphatases from Arabidopsis thaliana which increase in activity during Pi starvation, we analysed whole Arabidopsis plants grown in liquid medium containing or lacking phosphate and harvested at different time points. We also analysed the acid phosphatase activity using non-denaturing polyacrylamide gel electrophoresis (Experimental procedures). As shown in Fig. 1, the activity of the majority of the acid phosphatase isoforms was increased in phosphate-starved plants with respect to the non-starved controls, although the relative increase in activity of each enzyme was different.
The highest mobility phosphatase, AtACP5, was chosen for further characterisation studies because its activity was one of the most highly induced under Pi starvation stress. In addition, contrary to the case of the lowest mobility phosphatases, AtACP5 displayed an excellent chromatographic behaviour (i.e. eluted as a sharp peak). AtACP5 was purified from plants starved of Pi for 10 days (as described in Experimental procedures) using a relatively simple protocol consisting of one selective precipitation with ammonium sulphate (40–60%), followed by two batch ion-exchange chromatography steps, one hydrophobic interaction HPLC step and finally a DEAE-HPLC step. The follow-up of AtACP5 in all purification steps was done by visualising its acid phosphatase activity after PAGE in the same conditions as in Fig. 1(a). The first two chromatographic steps were carried out in the presence of phosphate-based buffers, resulting in the recovery of the AtACP5 in the exclusion volumes from both the mono Q and mono S columns. When other buffers where tested (50 m m Tris–HCl pH = 7.2, for the mono Q column; 50 m m citrate pH = 5.5, for the mono S column) AtACP5 was retained on the columns, suggesting that there might be some conformational change in the protein in the presence of phosphate. In fact, the DEAE-HPLC step was made using a Tris-based buffer so that AtACP5 was retained in the column. Table 1 summarises the purification of AtACP5. The enzyme was purified 888-fold to a final specific activity of 1333 units mg–1. The purified protein was present as a single major band after SDS-PAGE, with an apparent molecular mass of 34 kDa ( Fig. 1b). Inhibition tests were performed with three compounds, phosphate, molibdate and tartrate. As shown in Fig. 1(c), molibdate and, to a lesser extent, phosphate were inhibitors of AtACP5. In contrast, AtACP5 activity was insensitive to tartrate inhibition, as are purple acid phosphatases, a class of dinuclear metallo phosphatases (reviewed in Doi et al. 1988 ; Que & True 1990; Vincent et al. 1991 ).
Table 1. Purification of AtACP5
Specific activity (units/mg)
AtACP5 is a member of the purple acid phosphatase family
N-terminal sequencing of the purified protein gave a unique sequence (ELQRFIEPAKSDGSVSFIVIGDW, Fig. 2a), with a yield in Edman degradation cycles in agreement with the amount of protein loaded on the protein sequencer (not shown), thus confirming the purity of AtACP5. Comparison with databank sequences using the FASTA program revealed that the known protein with the highest similarity to AtACP5 was a purple acid phosphatase from animals, known as type 5 acid phosphatase. To isolate AtACP5 cDNA and genomic DNA clones, we first prepared a gene-specific probe using a PCR-based approach on cDNA derived from RNA from phosphate starved Arabidopsis plants. For primers, two mixtures of oligonucleotides were used. One was derived from a segment of the determined amino acid sequence of AtACP5 (RFIEPA, Fig. 2a). The other oligonucleotide mixture was derived from a stretch of amino acid residues (GDNFY, Fig. 2a) highly conserved in known mammalian type 5 acid phosphatases and in two related hypothetical plant proteins, represented by ESTs T45273 and D46537 from Arabidopsis and rice, respectively, which were identified searching the plant EST databanks with the sequence of the animal purple acid phosphatase. After cloning and confirming its correspondence with the known AtACP5 sequence, the probe was used to screen a cDNA library also prepared from RNA from phosphate-starved Arabidopsis plants. In parallel, we screened an Arabidopsis genomic DNA library. The sequence of the largest cDNA insert and of a 3.9 kb fragment of the genomic clone (spanning from 1.9 kb upstream of the initiation codon to 0.7 kb downstream of the termination codon) were determined, revealing the presence of an open reading frame of 338 amino acid residues interrupted by two introns ( Fig. 2a). This protein contained an N-terminal extension of 31 amino acid residues relative to the mature AtACP5 protein, with the characteristics of a signal peptide. The signal peptide processing site predicted by statistical analysis, using the PSORT program ( http://psort.nibb.ac.jp:8800) ( vonHeijne 1986), corresponded precisely to the one determined experimentally. No other known targeting sequence was detected in AtACP5. On the other hand, the molecular weight predicted for the mature protein (307 aa, 35 kDa) matched very well with that determined by SDS-PAGE (34 kDa). A computer-assisted search with the complete aa sequence of AtACP5 in databases, using the FASTA and BLAST programs ( Altschul et al. 1990 ; Pearson & Lipman 1988), showed that this protein is a member of the purple acid phosphatase (PAP) family. In particular, all residues implicated in metal ligand binding ( Klabunde et al. 1996 ) are conserved in AtACP5 ( Fig. 2a). A phylogram of PAP proteins was constructed using the neighbour-joining method of the CLUSTALW program ( Thompson et al. 1994 ). Two monophyletic groups (I and II) could be distinguished ( Fig. 2b), and both plant and animals had members in the two groups. AtACP5 is in the same group as type 5 acid phosphatases (group I).
AtACP5 responsiveness to phosphate starvation
The effect of phosphate starvation on AtACP5 RNA accumulation was investigated at the whole plant level. Arabidopsis plants were grown in liquid medium under Pi-sufficient conditions for 5 days, transferred to Pi-deficient or sufficient medium and harvested at different times. Northern hybridisation analysis of AtACP5 RNA showed that this transcript starts to accumulate after 2 days of Pi deprivation and increases continuously for at least 10 days ( Fig. 3a). Similar results were obtained with plants grown in solid medium as shown in Fig. 3(b). In this case, AtACP5 RNA was analysed separately in roots and aerial parts showing that AtACP5 is induced in both parts of the plant. To evaluate whether a genetically induced Pi deficiency would also stimulate AtACP5 expression, its RNA was also analysed in the pho1 mutant which has impaired xylem loading of Pi, thereby having a reduced concentration of Pi in its leaves ( Poirier et al. 1991 ). As shown in Fig. 3(c), AtACP5 was expressed in this mutant in Pi-sufficient conditions.
The Pi control of AtACP5 expression was also examined by evaluating transcript levels following re-supplying Pi for different times to Pi-starved plants. Northern hybridisation experiments showed that Pi-starved plants had negligible levels of AtACP5 RNA after 8 days of growth under Pi-sufficient conditions ( Fig. 3d). Therefore, the inducibility of AtACP5 under Pi-starvation is reversible.
AtACP5 response to other types of environmental/developmental signals
To evaluate the specificity of the response of AtACP5 to phosphate starvation, we assessed the effect of other types of stress and phytohormonal molecules on AtACP5 RNA levels. As shown in Fig. 4, none of the other types of nutritional starvation (K- and N-starvation) induced AtACP5 RNA accumulation. In fact, N-starvation reduced AtACP5 RNA levels. However, AtACP5 was responsive to abscisic acid and to salt stress, as well as to H2O2, although not to wounding, paraquat, cysteine or other phytohormone or signalling molecules other than ABA.
We also investigated the expression of AtACP5 during normal development in different organs, roots, leaves (both senescing and non-senescing), inflorescence stems, flowers and siliques ( Fig. 4c). AtACP5 RNA was most highly abundant in senescent leaves, but expression was also detected in flowers, resembling the phosphate starvation induced RNS1 gene ( Bariola et al. 1994 ).
Activity of an AtACP5:GUS translational fusion in transgenic Arabidopsis in response to different stresses
To obtain more information on the expression of AtACP5 in response to different stimuli, we prepared an AtACP5:GUS translational fusion and transformed Arabidopsis plants. Towards this, a 2 kb DNA fragment containing the promoter and part of the transcribed region (up to nucleotide 129 of cDNA, Fig. 2) of AtACP5 was fused in frame to the coding region of the GUS reporter gene present in the PBI101.3 plasmid ( Jefferson et al. 1987 ).The resulting plasmid was used to transform Arabidopsis plants, using the Agrobacterium infiltration technique ( Bechtold et al. 1993 ). V1 progeny of 15 Arabidopsis transgenic plants were grown under phosphate starvation or under phosphate sufficient conditions and GUS activity was determined. In 13 transgenic plants, a clear increase of GUS activity in response to phosphate starvation was observed. Homozygous lines from two transgenic Arabidopsis plants were obtained for in situ analysis of GUS activity in response to different stimuli ( Fig. 5). The four stimuli tested that induced the accumulation of AtACP5 mRNA (phosphate starvation, salt stress, H2O2 and ABA) also induced the expression of the AtACP5:GUS translational fusion. The three stimuli tested that did not induce AtACP5 (nitrogen, potassium starvation and paraquat) also did not induce the reporter gene. In fact, the level of GUS activity was reduced in nitrogen-starved transgenic plants in the same way that the steady state transcript level of AtACP5 was reduced (see Fig. 4). These results indicated that expression of the reporter gene mimicked AtACP5 steady state transcript levels. Given that the reporter gene was constructed as a translational fusion, it appears that there is not a major translational control mechanism influencing AtACP5 activity. In situ analysis of GUS activity using histochemical staining showed that the spatial expression of the AtACP5 reporter gene was not the same for all stimuli. Thus, phosphate starvation and abscisic acid stimulated the activity of the reporter gene both in roots and aerial parts, whereas the H2O2 and salt stresses only resulted in induction of GUS activity in the aerial parts of the transgenic plants.
Peroxidation activity of AtACP5
The presence of a di-iron-oxo cluster in animal purple acid phosphatases has been shown to endow these proteins with peroxidation activity ( Hayman & Cox 1994; Sibille et al. 1987 ). Given the responsiveness of AtACP5 gene to some conditions causing oxidative stress, such as H2O2 and phosphate starvation, we were interested to test whether the AtACP5 phosphatase also had this activity. Using a chemiluminescence assay ( Hayman & Cox 1994) we found that, in the presence of luminol and H2O2, nanomolar amounts of the purified AtACP5 phosphatase induced a striking chemiluminescence. As shown in Fig. 6, AtACP5 produces higher chemiluminescence than equimolar amounts of bovine serum albumin ( Fig. 6). In fact, photon emission induced by the phosphatase was proportional to the amount of protein in the assay, whereas increasing the amounts of BSA hardly had an effect on chemiluminescence. Therefore, like its animal counterparts, AtACP5 displays peroxidation activity.
Induction of phosphatase activity in response to phosphate starvation is a common phenomenon among organisms acquiring phosphorous from the environment. These enzymes have been successfully used as markers to investigate the molecular mechanisms underlying the adaptive responses to phosphate starvation of bacteria and fungi (for reviews, see Lenburg & O’Shea 1996; Torriani 1990; Vogel & Hinnen 1990). This led us to clone an Arabidopsis thaliana psi phosphatase gene as a first step towards performing a similar analysis on plants. In this study, we have purified an acid phosphatase, AtACP5, induced by phosphate starvation, cloned its cDNA and prepared an AtACP5:GUS translational fusion whose expression mimics AtACP5 steady state transcript levels. In addition, we have shown that AtACP5 also displays peroxidation activity, suggesting that this protein is not exclusively involved in the phosphate mobilisation role typical of known psi phosphatases.
AtACP5 is a member of the purple acid phosphatase family, specifically of the type 5 acid phosphatases, and is also induced by some conditions other than phosphate starvation. Purple acid phosphatases are a family of enzymes which contain a dinuclear centre in their active site (for reviews Doi et al. 1988 ; Que & True 1990; Vincent et al. 1991 ). Their purple colour, visible when present in sufficient concentration, results from a tyrosinate–FeIII charge transfer transition. The presence of proteins of the PAP family has been detected in a wide range of species, from humans to bacteria ( Fig. 2b), although genes encoding these proteins have not been detected in some organisms for which the whole genome sequence is known, such as Sacharomyces cerevisiae. The assignation of AtACP5 to this family, and specifically as a type 5 acid phosphatase, is based on the following criteria. (i) It shows significant sequence similarity to other members of the PAP family; (ii) it shares all the residues implicated in ligand binding, including the tyrosinate residue specific of PAPs ( Klabunde et al. 1996 ); (iii) its phosphatase activity is resistant to tartrate; and (iv) like its mammalian type 5 acid phosphatase counterparts, AtACP5 displays peroxidation activity ( Hayman & Cox 1994). In addition, phylogenetic analysis has shown that purple acid phosphatases fall into two monophyletic groups (I and II) and, like type 5 acid phosphatases, AtACP5 belongs to group I. The two purple acid phosphatase groups differ in the size of their members (group I, 300–400 aa-long; group II, 450–650 aa-long). Other possible differences between these groups is the oligomerisation status of the active form and the composition of the binuclear metallic centre. Known mammal PAPs, which belong to group I, are monomeric and contain Fe3+– Fe2+, and kidney bean purple acid phosphatase, belonging to group II, is homodimeric and Zn2+ substitutes for the ferrous ion ( Klabunde et al. 1996 ). However, it remains to be shown whether these properties are shared among all the members within each group.
Both groups of phosphatases are represented at least in some animal and plant species, such as C. elegans and Arabidopsis, respectively, indicating that the divergence between the two classes of phosphatases probably occurred before the plant/animal diversification. In agreement with this is that within each of the two phylogenetic groups, distances among phosphatases is in consonance with the phylogenetic distances among the taxa; for example, animals being closer to fungi than to plants, and closer to plants than to bacteria ( Kumar & Rzhetsky 1996).
The two activities displayed by AtACP5, hydrolysis of phosphate and peroxide formation, might insinuate that this protein performs two distinct physiological roles. Thus, the phosphatase activity probably reflects a role in phosphate mobilisation. This is suggested by the induction of this gene by phosphate starvation and senescence, the latter being a condition which involves the mobilisation of nutrients, including phosphate from the oldest to the growing parts of the plant. Other stimuli inducing AtACP5 expression, such as salt stress, oxidative stress and abscisic acid, which induce senescence-related processes, would also promote nutrient mobilisation ( Kelly & Davis 1988). In all these cases in which AtACP5 is expressed, its role would probably concern the recycling of the phosphate from the phosphate ester pool of the plant. Note that phosphate starvation induces AtACP5 not only in roots but also in aerial parts of the plant ( Figs 3 and 5). Nevertheless, its induction in roots during phosphate starvation raises the possibility that AtACP5 also participates in the scavenging of phosphate from the soil. This possibility would require an extracellular location of the protein consistent with the fact that it is synthesised as a precursor with an N-terminal extension with the characteristics of a signal peptide. We have not been able to extract AtACP5 or other acid phosphatases in the apoplastic fluid, even in the presence of high salt, which would indicate that they are probably tightly anchored to the cell wall or the plasma membrane. In line with this, a psi PAP from Spirodela oligorrhiza (belonging to group II) has been found to be anchored to the outside surface of the plasma membrane by a covalently linked glycosylphosphatidylinositol (GPI) moiety ( Nakazato et al. 1998 ).
The peroxidation activity displayed by AtACP5, indicated by the chemiluminescence experiments ( Fig. 6), suggests that in addition to its likely role in phosphate mobilisation, AtACP5 could also play a role in the metabolism of reactive oxygen species. In humans, it has been suggested that PAP has a central role in the generation of reactive oxygen species in macrophages and osteoclasts, which are associated with microbial killing and bone resorption ( Hayman & Cox 1994; Hayman et al. 1996 ). In kidney bean, its purple acid phosphatase has been suggested to play an antioxidant role to prevent the formation of oxygen radicals in the seed ( Klabunde et al. 1995 ). We cannot exclude either of these possibilities for AtACP5 (i.e. generation or scavenging of free radicals) as they are not mutually exclusive. Thus, during senescence-related processes, AtACP5 could contribute to the generation of reactive oxygen species accompanying these phenomena ( del Rio et al. 1998 ). In response to oxidative stress, it is likely that the enzyme co-operates with other protection mechanisms in the scavenging of active oxygen species.
The fact that AtACP5 is induced by different stimuli sharing in common senescence related aspects does not indicate that AtACP5 is a general stress response gene which is triggered by a single signal transduction mechanism. AtACP5 neither responds to all types of nutritional deficiency (as indicated by non-responsiveness to nitrogen and potassium starvation), nor to all conditions inducing oxidative stress (since it is not induced by salicylic acid, paraquat or cysteine). On the other hand, its response to phosphate starvation was detected after 2 days of stress when no visible symptoms of senescence could be observed. It is also constitutively expressed in healthy, non-senescent leaves of the pho1 mutant. In addition, despite the fact that abscisic acid induces AtACP5 expression, mutants impaired in ABA synthesis/signal transduction do not affect phosphatase activity in phosphate starved Arabidopsis plants ( Trull et al. 1997 ). Moreover, the AtACP5:GUS reporter gene analysed in this study was differentially expressed in response to the various types of stress since transgenic plants harbouring the reporter gene displayed GUS activity in the roots only in response to phosphate starvation and ABA. Therefore, we consider it most likely that several signal transduction pathways influence AtACP5 activity. The isolation of mutants with altered AtACP5 expression will shed light on the number and on the relations among these pathways as well as on the degree of conservation of the phosphate starvation signal transduction pathway(s) between plants and yeasts, the eukaryote in which this response has been the most well characterised ( Lenburg & O’Shea 1996; Vogel & Hinnen 1990). In this context, the finding that an AtACP5:GUS translational fusion mimics AtACP5 expression opens the possibility to use a reporter gene-based strategy approach to search for mutants.
Plant material and growth conditions
Arabidopsis thaliana (L), ecotype Landsberg erecta, was grown in soil in a growth chamber under conditions of 16 h light/8 h dark at 20°C. For organ analysis, samples of roots, leaves, senescing leaves, stems and flowers of 4–5-week-old plants were collected. Senescing leaves in stage 1 (sen1) were considered to be leaves starting to show some wilting. Senescent leaves in stage 2 (sen2) were those showing visible signs of chlorosis and wilting. Plants grown in the same conditions were used for wounding experiments. Leaves were wounded using forceps and collected 24 h later, according to Titarenko et al. (1997) . For nutrient starvation analysis, plants were first germinated and grown for 5 days either on continuously shaken liquid MS medium ( Murrashige & Skoog 1962) (modified because sucrose was reduced to 0.5% and KH2PO4 was raised to 2 m m and 50 m m MES pH 5.5 was included), or in solid medium, with the same composition as the liquid medium except that sucrose was kept at 1% and bactoagar (1%) was included. Subsequently, plants were subjected to the nutritional starvation stress by transferring them to a medium lacking the appropriate nutrient ( Bariola et al. 1994 ) and allowed to grow for different periods of time. In Pi deficient medium, KH2PO4 was omitted. In N-deficient medium, equimolar amounts of KCl (1.4 g l–1) were substituted for the KNO3. In K-deficient medium, NaNO3 was substituted for KNO3 and NaH2PO4 was substituted for KH2PO4. Growth in these media was for 2–10 days. To analyse the effects of different phytohormones or other molecules, plants were first grown in liquid medium for 7 days prior to the treatment and then kept under treatment for 12–24 h. The treatments were as follows: ABA (abscisic acid), 100 μm; (sodium chloride) NaCl, 250 m m; JA (jasmonic acid), 50 μm; ACC (1-aminocyclopropane-1-carboxylic acid), 50 μm; SA (salicylic acid), 100 μm; IAA (Indolacetic acid), 10 μm; Kin (Kinetin), 10 μm; GA3 (giberellic acid), 100 μm; Bras (Brasinolide), 2 μm; H2O2, 4 m m; Cys (cysteine) 1 m m; PQ (paraquat), 100 μm. All samples were kept at –80°C until used.
Purification of AtACP5
To isolate AtACP5 phosphatase, plants were grown in phosphate sufficient liquid medium for 5 days and subsequently transferred to phosphate depleted medium, where they were grown for a further 10 days. All steps for protein purification were carried out at 4°C. About 500 g of seedlings were homogenised in a mortar in the presence of liquid N2, and proteins were extracted with 1.5 l of buffer A (50 m m citrate-phosphate, pH 7.6, 10 m m EDTA, 1 m m PMSF). After stirring for 1 h, the slurry was centrifuged at 8000 g for 30 min. Soluble proteins were fractionated by selective precipitation with ammonium sulphate, which was slowly added to the supernatant. The fraction precipitated between 40% and 60% saturation was collected for further processing after centrifugation at 10 000 g for 30 min. This fraction was redissolved in 30 ml of buffer A and dialysed for 16 h against 4 × 5L of buffer B (50 m m citrate-phosphate pH 7.6). The desalted solution was loaded onto an Econo-Pac Q cartridge equilibrated with buffer B. AtACP5 activity was recovered in the exclusion volume. This eluate was dialysed for 6 h against 3 × 5 l of buffer C (50 m m citrate-phosphate, pH 5) and loaded onto an Econo-Pac S column. Again, AtACP5 was recovered in the exclusion volume. This eluate was dialysed for 16 h against 4 × 4 l of buffer D (50 m m Tris–HCl, pH 7.2) and concentrated to 4 ml in a centriprep-10 cartridge. This concentrate was mixed with 6 ml of Buffer D plus 2.8 m (NH4)2SO4 and loaded onto a Supercosil hydrophobic HPLC column equilibrated in the same buffer as the sample. The column was eluted with a linear gradient (1.7 m to 0 m; 200 ml) of (NH4)2SO4 in buffer D at 25°C. AtACP5 was recovered in the 500–300 m m range of (NH4)2SO4. This fraction was dialysed for 16 h against 4 × 5 l of buffer D, and loaded into a DEAE-HPLC column equilibrated in the same buffer. The column was eluted with a linear gradient (0 m m to 250 m m; 100 ml) of NaCl in buffer D. Peak activity eluted at about 150 m m NaCl. The fractions containing AtACP5 were dialysed for 16 h against distilled water and lyophilised.
Gel electrophoresis of proteins
Denaturing polyacrylamide gel electrophoresis (SDS-PAGE) was performed using the discontinuous system of Laemmli (1970) in a Bio-Rad minigel apparatus with a 4–20% polyacrylamide gradient. Non-denaturing PAGE was carried out in the same conditions, except that SDS was omitted in all buffers and the samples were not pre-incubated at 100°C before loading.
Protein determination and phosphatase activity assays
Protein concentration was determined by the Coomassie Protein Assay Reagent (PIERCE) using bovine serum albumin as a standard. Phosphatase activity in solution was determined by measuring the release of nitrophenol from 7.5 m m p-nitrophenylphosphate in 100 m m citrate pH 5.8. Reactions (1 ml) were incubated at 37°C for 10 min and stopped with 2 ml of 2N NaOH, after which absorbance at 410 nm was measured. One unit of enzyme was defined as that releasing 1 m m of nitrophenol per min at 37°C. Staining for acid phosphatase activity in gels was carried out as described by Lefebvre et al. (1990) .
Determination of peroxidation activity
To evaluate the capacity of AtACP5 to catalyse the peroxidation of luminol a chemiluminescence assay was performed essentially according to Hayman & Cox (1994). Chemiluminescence was recorded in a Lumat 630 luminometer. The reaction was started by the addition of 0.2 ml of hydrogen peroxide (4.4 m m) to a 0.2 ml solution of 0.2 m Tris–HCl pH 7.6, and 300 m m 5-aminonaphtalide (luminol, Sigma). Emission of photons was measured for 10 sec after the addition of the H2O2 solution.
Standard molecular procedures
All methods, including screening of cDNA- or genomic DNA-libraries, RNA and genomic DNA isolation, labelling of DNA and oligonucleotides, etc., were performed as described previously ( Avila et al. 1993 ; Sambrook et al. 1889 ), except where indicated. The vectors for cloning were pBluescriptII ( Alting-Mees & Short 1989).
To isolate AtACP5 cDNA clones, we first prepared a gene-specific probe using a PCR-based approach with cDNA derived from RNA from phosphate starved Arabidopsis. For primers, two mixtures of oligonucleotides were used. One was derived from a segment of the determined amino acid sequence of AtACP5 (RFIEPA, Fig. 2) preceded by a linker sequence: GGAATTCAGNTTYATHGARCCNGC (N = A, G, C or T; H = A, C or T; R = A or G; Y = C or T). The other oligonucleotide mixture was derived from a stretch of amino acids highly conserved in known mammalian type 5 acid phosphatases and in two related hypothetical plant proteins, represented by ESTs T45273 and D46537 from Arabidopsis and rice, respectively, which were identified searching the plant EST databanks with the sequence of the animal purple acid phosphatase residues (GDNFY, Fig. 2), also preceded by a linker sequence: GCGGATCCRTAR- AARTTRTCNCC. The cDNA used in the PCR reactions was derived from poly A+ RNA prepared from plants first grown for 5 days under phosphate sufficient conditions and subsequently under Pi-starving conditions for 7 days. PCR amplification was performed as follows: the DNA (20–200 pg ml–1) was amplified for 30 cycles using polymerase (0.025 U ml–1). Each cycle of amplification consisted of 1 min at 94°C, 90 sec at 55°C and 2 min at 72°C, except the first two cycles in which the annealing temperature was 40°C instead of 55°C. A cDNA clone encoding the full-size AtACP5 protein was isolated by screening an Arabidopsis cDNA library in vector λNM1149 (105 Pfu., M. Sanchez, unpublished results), representing RNA from a mixture of plants collected after growing under phosphate starvation regimens for 1, 2, 4, 7 or 10 days. An AtACP5 genomic clone, to be used to prepare an AtACP5:GUS translational fusion was isolated from a library of Arabidopsis thaliana genomic DNA in the vector λDashII (kindly provided by Jeff Dangl). Both libraries were screened using the above isolated AtACP5 probe.
Preparation of an AtACP5:GUS translational fusion and transformation of Arabidopsis plants
An AtACP5:GUS translational fusion was made by in-frame cloning a 2 kb HindIII-EcoRI fragment of the AtACP5 genomic clone, containing the promoter region and part of the transcribed region (until nucleotide +5 from the AtACP5 translation initiation codon), with the GUS coding region present in the binary vector pBI101.2 ( Jefferson et al. 1987 ). Agrobacterium tumefaciens harbouring these constructs (individually) were used to transform Arabidopsis ecotype Col-0 with the in planta vacuum infiltration method ( Bechtold et al. 1993 ). Kanamycin-resistant T1 plants were selected by plating seeds on MS medium supplemented with 1% sucrose and 50 μg ml–1 kanamycin and transferred to soil.
Computer programs for protein and nucleic acid analysis
We are grateful to Dr Carlos Lopez-Otin for performing the N-terminal sequence of AtACP5; to José Ignacio Leguina for help with the tree construction; and to Drs Yves Poirier, Csaba Koncz and Jeff Dangl for providing us with the pho1 mutant, the RBP4 probe and the Arabidopsis genomic DNA library, respectively. We also thank Dr Cathie Martin for her critical review of the manuscript. The excellent technical assistance of Maria Jesus Benito is acknowledged. J.C.d.P. was the recipient of a PhD fellowship from the Universidad Complutense-Madrid. This research was financed by the EU (Biotech Program, contract numbers BIO-CT93–0400 and BIO4-CT96–0770).
EMBL data library accession numbers AJ133747 (AtACP5 cDNA) and AJ243527( AtACP5 gene).