A novel interaction partner for the C-terminus of Arabidopsis thaliana plasma membrane H+-ATPase (AHA1 isoform): site and mechanism of action on H+-ATPase activity differ from those of 14-3-3 proteins#

Authors

  • Piero Morandini,

    Corresponding authorSearch for more papers by this author
  • Marco Valera,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Cristina Albumi,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Maria Cristina Bonza,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Sonia Giacometti,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Giuseppe Ravera,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Irene Murgia,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Carlo Soave,

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author
  • Maria Ida De Michelis

    1. Dipartimento di Biologia ‘L. Gorini’, Sezione di Fisiologia e Biochimica delle Piante, Centro di Studio CNR-Biologia Cellulare e Molecolare delle Piante, c/o Dip. di Biologia, Via Celoria 26, 20133 Milan, Italy
    Search for more papers by this author

  • #

    Dedicated to Antonio Graniti on the occasion of his 75th birthday.

Corresponding author (fax +39-02-5031-4815; e-mail piero.morandini@unimi.it)

Summary

Using the two-hybrid technique we identified a novel protein whose N-terminal 88 amino acids (aa) interact with the C-terminal regulatory domain of the plasma membrane (PM) H+-ATPase from Arabidopsis thaliana (aa 847–949 of isoform AHA1). The corresponding gene has been named Ppi1 for Protonpumpinteractor 1. The encoded protein is 612 aa long and rich in charged and polar residues, except for the extreme C-terminus, where it presents a hydrophobic stretch of 24 aa. Several genes in the A. thaliana genome and many ESTs from different plant species share significant similarity (50–70% at the aa level over stretches of 200–600 aa) to Ppi1. The PPI1 N-terminus, expressed in bacteria as a fusion protein with either GST or a His-tag, binds the PM H+-ATPase in overlay experiments. The same fusion proteins and the entire coding region fused to GST stimulate H+-ATPase activity. The effect of the His-tagged peptide is synergistic with that of fusicoccin (FC) and of tryptic removal of a C-terminal 10 kDa fragment. The His-tagged peptide binds also the trypsinised H+-ATPase. Altogether these results indicate that PPI1 N-terminus is able to modulate the PM H+-ATPase activity by binding to a site different from the 14-3-3 binding site and is located upstream of the trypsin cleavage site.

Introduction

The electrochemical proton gradient established by the action of the PM proton pump across the PM influences many processes such as solute import and export, pH homeostasis and cell growth (Palmgren, 2001). All these processes can be very energy consuming: one or more protons may be required per imported solute molecule and therefore the cell must be able to support large solute fluxes when required (e.g. phloem loading) with a corresponding high proton pumping activity. On the other hand, pump activity must be tightly regulated in order to avoid extreme pH and electrical gradient or excessive ATP consumption when cellular activities, such as solute transport, slow down.

The activity of the PM proton pump is influenced by several signals such as hormones (auxin, ABA) (Palmgren, 1998), light (Kinoshita and Shimazaki, 1999), water potential (Curti et al., 1993), acid load (Hager and Moser, 1985), toxins like FC (Rasi-Caldogno et al., 1986) and pathogens (Vera Estrella et al., 1994), but a detailed molecular description of the mediators involved is missing for most of these signals. A major regulatory site of PM H+-ATPase has been identified in the autoinhibitory C-terminal domain, which comprises around 100 aa, located in the cytoplasm. This domain can be removed by trypsin or displaced by binding of 14-3-3 proteins: both events cause a shift in pH optimum and alter kinetic parameters, leading to an increase in pump activity (Baunsgaard et al., 1998; Johansson et al., 1993; Oecking and Hagemann, 1999; Olivari et al., 1998; Palmgren et al., 1991; Rasi-Caldogno et al., 1993). This domain is not required for pump activity, as complete removal via deletions at the gene level generates a constitutively active enzyme (Regenberg et al., 1995). Interaction of the C-terminus with the 14-3-3 proteins is dependent upon its phosphorylation, but this requirement can be bypassed, both in vivo and in vitro, by FC treatment (Fuglsang et al., 1999; Maudoux et al., 2000; Svennelid et al., 1999). An alanine scanning mutagenesis study on the C-terminus of the PM H+-ATPase (isoform AHA2 of A. thaliana) revealed that most of the residues (64/87) can be mutated to alanine without significant effect (Axelsen et al., 1999). An activated phenotype correlated with increased binding of 14-3-3 proteins was observed only upon mutation of residues clustered into two regions (region I: K863-L885 and II: S904-L919). The authors suggest that the mutation makes a pre-existing site more accessible either to 14-3-3 or to a kinase that phosphorylates it, allowing subsequent 14-3-3 binding. Other studies demonstrated that 14-3-3 binding occurs at a region distant from these two stretches and namely at the extreme C-terminus (HYTV, highly conserved among plant PM H+-ATPases), with phosphorylation on the Thr residue required for binding (Fuglsang et al., 1999; Maudoux et al., 2000; Olsson et al., 1998; Svennelid et al., 1999). The relevance of this phosphorylation for the transduction of a physiological stimulus such as blue light is stressed by the work of Kinoshita and Shimazaki (1999, 2001) which demonstrates that the in vivo phosphorylation level correlates with the degree of H+-ATPase activation.

Much less is known about in vitro and in vivo activation by other effectors and about the molecular players involved: auxin effects, for instance, seem to require a factor that is lost during membrane isolation (Jahn et al., 1996; Kim et al., 1998) and indeed a soluble auxin receptor seems to bind and activate the H+-ATPase (Kim et al., 2001). Here we report the identification, the cDNA sequence and a preliminary characterisation of a protein, isolated with the two-hybrid technique, containing a domain which interacts with the C-terminal regulatory domain of the PM proton pump. This novel interactor, named PPI1, acts on a site different from the already identified 14-3-3 binding site.

Results

Identification of a novel interactor

A DNA fragment encoding the C-terminal 103 aa of A. thaliana PM H+-ATPase isoform AHA1 (aa 847–949) was cloned into vector pAS-2 (Durfee et al., 1993) in frame with the GAL4 DNA binding domain. We chose this C-terminal peptide as a bait in the two-hybrid screen, because it has been shown to contain a regulatory domain (Palmgren et al., 1991; Regenberg et al., 1995). The fusion was expressed in yeast cells and showed the predicted molecular weight, as judged by Western blotting (data not shown). The yeast strain was then transformed with a two-hybrid A. thaliana cDNA library: out of 1.3 × 106 primary transformants, several thousands were able to grow in the absence of added histidine and a few hundred stained blue in the presence of X-GAL. Bait plasmid loss and subsequent mating of transformants against a panel of unrelated baits (Lamin, CDK and SNF1, see Harper et al., 1993; data not shown), allowed identification of colonies containing preys specific for the C-terminal tail of the H+-ATPase; these colonies were examined further by recovering and sequencing the prey plasmids.

Several yeast clones isolated in independent transformations contained the same prey, coding for a 102 aa peptide, having no homology to any protein of known function; the gene identified by its sequence was named Ppi1 (for Protonpumpinteractor 1). Upon retransformation into a bait-expressing yeast strain, the plasmid was able to reproduce both phenotypes (histidine prototrophy and β-galactosidase activity) confirming that the isolated plasmid coded for a putative interactor. It conferred ability to interact with AHA1 C-terminus (Figure 1), but neither with a construct bearing a frameshift around the fusion point (due to a single nucleotide deletion) nor with unrelated proteins (such as SNF4, SNF1), itself or the empty vector. The interaction was retained when the AHA1 and PPI1 coding fragments were exchanged on the bait and prey plasmids (Figure 1).

Figure 1.

Two-hybrid analysis for validation of a novel interactor.

Yeast diploids bearing different pairs of bait and prey were stained with X-Gal. Bait constructs in pAS vector (columns): AHA1, in frame fusion with the 103 C-terminal aa of isoform AHA1; STOP, same construct as AHA1, but with a single bp deletion causing a frameshift around the fusion point and a premature translation stop codon; SNF1, positive control bait vector that shows interaction with SNF4; PPI1, the 306 bp insert from the prey vector isolated in the two-hybrid screen was cloned in frame, together with flanking sequences, into the bait vector. Prey constructs in pACT vector (rows): VECT, empty vector pACT; PPI1, clone isolated in the two hybrid screen; AHA1, in frame fusion with the 103 C-terminal aa of isoform AHA1; SNF4, positive control prey vector that shows interaction with SNF1.

A new, unexpected interaction of AHA1 C-terminal tail with itself was revealed by the assay, suggesting that the last 103 aa may contain a homodimerization domain for the H+-ATPase. Several reports support the idea that the H+-ATPase functions as an oligomer in vivo (Briskin, 1990). The significance and the mechanism of such an interaction will not be discussed here but will be the subject of a following paper (P. Morandini et al., unpublished results).

Gene structure and similarity

A PCR approach was used to recover the sequence of the whole cDNA (see Experimental procedures). The sequence (Figure 2) was extended for a few nucleotides on the 5′ end, while almost 2 Kb were added on the 3′ end. This allowed the identification in the database of ESTs or genomic sequences identical to our sequence. The genomic sequence was localised on BAC F27G19 (chromosome IV, coincident with marker mi123 at kb 12 709).

Figure 2.

cDNA sequence, deduced protein sequence and gene structure of Ppi1.

(a) Complete cDNA sequence (Accession n. AJ002020) was assembled from sequencing runs on cDNA clones isolated from the library (see Experimental procedures), genomic clones and GenBank data from various EST and genomic clones. The deduced protein sequence starting from the first ATG is presented below the cDNA sequence. The N-terminal region originally identified in the two-hybrid screen is boxed.

(b) Schematic representation of gene structure with position and size of introns, exon size and ORF; the C-terminal hydrophobic stretch is highlighted in black.

The whole cDNA comprises 2117 bp and contains a single ORF coding for a protein of 612 aa with a predicted molecular mass of 68.8 kDa. The putative protein is quite rich in charged residues (120 positive, 108 negative), especially towards the C-terminal part, even though the last 24 aa are completely devoid of charged or polar residues, suggesting the presence of a transmembrane domain. The protein lacks a canonical leader peptide, but this is not uncommon. For example, in the so-called tail-anchored proteins, membrane insertion occurs post translationally and depends only upon the terminal hydrophobic tail and a few surrounding aa (Wattenberg and Lithgow, 2001).

The nucleotide and the deduced protein sequence did not show significant similarity to any gene or protein with known function present in the database. Beside a large number of cDNA and genomic clones from A. thaliana that were identical to Ppi1, we found, in the same species, several similar sequences defining a family of proteins in the A. thaliana genome. At the protein level, the translation of at least 5 different genomic sequences could be aligned with PPI1 on the basis of a partially conserved stretch (residues 344–379, see Figure 3), with 4 out of 5 showing 26–36% identity over at least 380 aa (see Table S1). Interestingly, all but possibly one of the genomic sequences from A. thaliana do show the presence of the C-terminal hydrophobic stretch. The great majority of ESTs from A. thaliana were derived from Ppi1, meaning that it is the family member most represented in cDNA libraries.

Figure 3.

Alignment of PPI1 with several plant putative proteins in the region 336–388.

Sequences were translated and aligned with ClustalW (http://www.ch.embnet.org/software/ClustalW.html); shading was generated with Boxshade 3.21 (http://www.ch.embnet.org/software/BOX_form.html). Accession n. Secale cereale (BE587621), Hordeum vulgare (BF257616), Triticum aestivum (BE517746), Sorghum bicolor (BE358754), Solanum tuberosum (Access. n. BE343923), Nicotiana sylvestris (X71905), Gossypium hirsutum (AW186860), Ricinus communis (T24303); genomic sequences from A. thaliana Ppi1–6 are Ppi1: AL078467; Ppi2: AC023839; Ppi3: AP002029; Ppi4: AC022520; Ppi5: AC007369; Ppi6: AC005698. Further details about the sequences are available in the Supplementary material (Table S-I and S-II)

ESTs from several other plants, both mono and dicots showed significant similarity (more than 50% identity over stretches of 200–600 aa, see Table S2). Many more ESTs from other plant species also showed significant similarity, although over shorter stretches, due to limited sequence information. The alignment of some ESTs within the partially conserved stretch (residues 336–388) is also presented in Figure 3. A more extensive alignment with several dicot species is available in the supplementary material (Figure S1).

PPI1 binds to and stimulates the H+-ATPase

The Ppi1 clone encoding the first 88 aa of the protein was recovered in a two-hybrid screen using the isolated C-terminus of isoform AHA1 of the PM H+-ATPase as bait. The N-terminus of PPI1 was expressed as a fusion with both His tag (His6Ppi) and glutathione S-transferase (GST-Ppi + 88) in the E. coli strain BL21(DE3) pLysS allowing purification of large quantities (2–4 mg per 100 ml culture) of fusion proteins (see Experimental procedures). The His6-Ppi fusion showed an apparent molecular mass of 16 kDa (14.6 predicted), while for the GST fusion, the apparent molecular mass was 40 kDa (37.5 predicted).

To verify whether PPI1 N-terminus was also able to bind the native/entire enzyme, purified His6-Ppi and GST-Ppi + 88 were used as probes in overlay experiments. The H+-ATPase, immunoprecipitated from PM purified from A. thaliana cultured cells with a polyclonal antibody against its N-terminal domain (Olivari et al., 1998), was subjected to SDS PAGE, blotted and probed with either His6-Ppi or GST-Ppi + 88 (Figure 4, lanes B and D): both proteins bound to a 100-kDa band, whose identity with the PM H+-ATPase was confirmed by immunodecoration with a monoclonal antibody (Baur et al., 1996) (Figure 4, lane A). Control fusion proteins containing either GST or a His-tag bound to unrelated proteins were unable to bind the PM H+-ATPase (Figure 4, lanes C and E), indicating that the binding of PPI1 fusion proteins is specifically due to the PPI1 N-terminus. When His6-Ppi overlay was performed on the whole PM fraction, after SDS PAGE and blotting, several bands appeared, but a strong signal was observed in correspondence of the 100 kDa band (Figure 4, lane G), which was again confirmed as the PM H+-ATPase (lane F). Most of the other bands (but not the 100 kDa band) are labelled by the reagent employed to detect the HisTag (lane H) also in the absence of His6-Ppi.

Figure 4.

PPI1 binds to native and immunoprecipitated H+-ATPase.

Native PM and immunoprecipitated PM H+-ATPase (see Experimental procedures), were subjected to SDS-PAGE and Western blot.

(a) Immunoprecipitate corresponding to 15 µg of total PM proteins was immunodecorated with an anti-C-terminal H+-ATPase monoclonal antibody.

(b-e) Immunoprecipitates corresponding to 30 µg of total PM proteins were blotted onto nitrocellulose and incubated overnight at 4°C with: His6-Ppi (0.5 mg ml−1) (b) or His6-thioredoxin (0.5 mg ml−1) (c) and then probed with India HisProbe-HRP; GST-Ppi + 88 (1.5 mg ml−1) (d) or GST-Ca+± ATPase (1.5 mg ml−1) (e) and then immunodecorated with an anti-GST polyclonal antibody.

(f-h) native PM (50 µg total protein) blotted onto nitrocellulose was incubated with (g) His6-Ppi (100 µg ml−1) overnight at 4°C and then probed with (h) India HisProbe-HRP or (f) immunodecorated with an anti-C-terminal H+-ATPase monoclonal antibody. Samples (a) to (e) were separated with a 4–20% acrylamide gel; samples (f) to (h) were run on a 8% acrylamide gel. The position and the molecular mass (in kDa) of protein markers is shown for each gel type.

Since the H+-ATPase and PPI1 N-terminus are able to recognise each other under the above-described conditions, we checked whether PPI1 could affect enzyme activity. Addition of His6-Ppi to purified PM fractions stimulated the H+-ATPase activity (Figure 5), while a protein of similar size with the same His-tag (His6-Yc, a DNA binding protein fragment, Bellorini et al., 1997) was not (data not shown). His6-Ppi stimulated H+-ATPase activity both at pH 6.4 and at pH 7.3 (Figure 5). The pH of the assay medium strongly influenced the concentration dependence of His6-Ppi effect: at pH 6.4 stimulation of the H+-ATPase was nearly maximal at 20 µm His6-Ppi, while at pH 7.3 it increased steadily by increasing His6-Ppi concentration up to 60 µm. Also the GST-Ppi + 88 fusion protein was able to stimulate the H+-ATPase activity (Table 1), but the effect did not show the same pH dependence, being similar at the two different pH values.

Figure 5.

A tagged version of PPI1 N-terminus stimulates H+-ATPase activity.

Effect of increasing concentrations of His6-Ppi on the PM H+-ATPase activity assayed at pH 6.4 (○) and at pH 7.3 (□). PM H+-ATPase activity is expressed as per cent stimulation calculated respect to the activity measured in the absence of His6-Ppi (606 nmol min−1 mg−1 protein at pH 6.4; 235 nmol min−1 mg−1 protein at pH 7.3).

Table 1.  H+-ATPase activitya in the presence and in the absence of two different GST-Ppi fusion proteins
 ControlGST-Ppi + 88b% stimulationControlGST-Ppi + 612b% stimulation
  1. a Activity is expressed as nmol min−1 mg−1 protein; assays were performed as described in the legend to Figure 5.

  2. b In the presence of 60 µm for GST-Ppi + 88 or 1.0 µm for GST-Ppi + 612 fusion protein.

pH 6.47011210+ 735501084+ 97
pH 7.3207395+ 91196273+ 39

By using a commercial strain possessing extra tRNA genes for rare E. coli codons (see Experimental procedures), we were able to express the entire PPI1 protein in fusion with GST (GST-Ppi + 612), in sufficient amounts for testing its effects on the H+-ATPase activity. The entire protein was clearly able to stimulate PM H+-ATPase, being effective at a concentration as low as 1 µm (Table 1); as in the case of His6-Ppi, the stimulus was stronger at lower pH. The different pH dependence behaviour of the fusion proteins might reflect changes in conformation due to the presence of the tags. A detailed characterisation of the effect of PPI1 on the PM H+-ATPase activity and of its binding will of course require larger amounts of PPI1 protein devoid of tag sequences.

PPI1 activates the PM H+-ATPase activated by fusicoccin or by tryptic cleavage of the C-terminus

The PM H+-ATPase can be activated by tryptic cleavage of a 7–10 kDa C-terminal fragment (Johansson et al., 1993; Palmgren et al., 1991; Rasi-Caldogno et al., 1993) or by displacement of the C-terminus upon FC-induced binding of 14-3-3 proteins (Baunsgaard et al., 1998; Olivari et al., 1998; Piotrowski et al., 1998). We checked the effect of His6-Ppi on the activity of the PM H+-ATPase activated either by controlled tryptic treatment or by treating cells with FC prior to PM purification. Figure 6 shows that controlled proteolysis or FC only marginally affected the PM H+-ATPase activity assayed at pH 6.4 (Palmgren et al., 1991; Rasi-Caldogno et al., 1993): the observed difference is within the variability of independent PM preparations.

Figure 6.

FC and trypsin synergise with PPI1 in the stimulation of H+-ATPase.

Stimulatory effect of His6-Ppi on the PM H+-ATPase activity assayed at pH 6.4 (a) or at pH 7.3 (b). PM H+-ATPase activity was measured in the absence (white bars) or in the presence of 70 µm His6-Ppi (grey bars) in control PM, in PM pretreated in vitro with trypsin (100 µg mg−1 protein) or in PM isolated from A. thaliana cultured cells treated in vivo with 10 µm fusicoccin (FC).

At this pH (Figure 6a), PPI1 stimulated the PM H+-ATPase activity of controls, trypsin- and FC-treated samples to similar extents.

At pH 7.3, in agreement with previous observations (Rasi-Caldogno et al., 1993), both FC and controlled tryptic treatment of the PM strongly stimulated the PM H+-ATPase activity (around 140% and 120%, respectively, Figure 6b). His6-Ppi substantially stimulated the H+-ATPase activity not only in control PM (around 60%), but also in trypsin- and FC-treated samples (60 and 90%, respectively, Figure 6b): the effects were more than additive, indicating that PPI1 synergises with trypsin and FC.

PPI1 binds to the H+-ATPase of trypsin treated PM

The above results suggest that binding of PPI1 to the C-terminus of the PM H+-ATPase occurs at a site different from the 14-3-3- protein binding site (localised at the extreme C-terminus, Fuglsang et al., 1999; Svennelid et al., 1999) and upstream of the trypsin cleavage site. To further examine this issue, PM fractions were treated with or without trypsin, resolved by SDS PAGE, blotted and probed with His6-Ppi (Figure 7). Immunodecoration of the blot with a polyclonal antibody against the PM H+-ATPase N-terminus (Figure 7a) revealed a strong band at 100 kDa and a vary faint one at 90 kDa (probably due to endogenous proteases) in the control; after trypsin treatment we observed an almost quantitative conversion of the immunoreactive protein in the upper band to the lower one. Since the cleavage is known to occur at the C-terminal end (Palmgren et al., 1991), Figure 7a indicates that proteolysis had been effective in quantitatively removing a 10-kDa fragment from the C-terminus of the H+-ATPase. When the blot was probed with His6-Ppi (Figure 7b), a major prominent band at 100 kDa and a faint one at 90 kDa were labelled in the controls, while in trypsin-treated samples the opposite was true, with marked labelling of the 90 kDa band. The results indicate that the extreme 10 kDa of the C-terminus of the H+-ATPase are not required for the interaction to take place.

Figure 7.

Binding of His6-Ppi to the PM H+-ATPase from control or trypsin-treated PM.

(a) PM (13 µg of total protein) untreated (C) or treated with trypsin (100 µg/mg protein) (T) was subjected to SDS-PAGE and Western blot. Immunodecoration was performed using an anti-N-terminal H+-ATPase polyclonal antibody.

(b) PM (50 µg total protein) untreated (C) or treated with trypsin (100 µg/mg protein) (T) was subjected to SDS-PAGE and Western blot. The nitrocellulose membrane was incubated with His6-Ppi (100 µg ml−1) overnight at 4°C and then probed with India HisProbe-HRP.

Discussion

We have identified a novel interaction partner of the PM H+-ATPase. Its novelty concerns both the protein, as it does not resemble previously identified protein interactors, and the site of action. The N-terminal fragment of PPI1 is able to interact in the two-hybrid system with the C-terminal tail of the H+-ATPase in both configurations (as a bait and as a prey), a result usually regarded as symptomatic of a ‘true interaction’ (Bai and Elledge, 1997). The same fragment, purified as a His-tag or a GST fusion protein, binds to the whole H+-ATPase protein in overlay experiments. Moreover, these fusion proteins and a further fusion containing the entire coding region are able to stimulate the H+-ATPase activity of purified plasma membranes. Both the experiments with the fusion proteins and those from the two hybrid underline the presence of an interacting domain at the PPI1 N-terminus. A detailed characterisation of the interaction and of its effect on the H+-ATPase activity will require larger amounts of PPI1 devoid of any tag.

Several other proteins bind to the C-terminus of the H+-ATPase and have an effect on pumping activity. First and foremost, proteins belonging to the 14-3-3 family bind to the C-terminus and stimulate proton extrusion (Baunsgaard et al., 1998; Jahn et al., 1997). This interaction in vitro requires the phosphorylation of a specific threonine residue and there is strong evidence that the same happens in vivo: work by Kinoshita and Shimazaki (1999, 2001) demonstrates that the H+-ATPase is phosphorylated in vivo upon blue light treatment and that the phosphorylation level correlates with enzyme activity and 14-3-3 binding. The interaction is inhibited by 5′-AMP (Camoni et al., 2001) and the requirement for the phosphorylation can be bypassed by FC, which allows the formation of the complex between 14 and 3-3 protein and the H+-ATPase C-terminus in the absence of phosphorylation (Fuglsang et al., 1999). An implication of these data is that there must be at least two more proteins interacting with the C-terminal domain of the H+-ATPase, namely a kinase and a phosphatase responsible, respectively, for the addition and the removal of the phosphate. A protein phosphatase able to dephosphorylate the H+-ATPase in vitro has been purified from maize roots and the dephosphorylation inhibits the association of 14-3-3 proteins to the H+-ATPase (Camoni et al., 2000).

The interactor we have identified does not resemble the proteins acting on the C-terminus because there is no sequence similarity and the biochemical evidence (Figures 5 and 6) localises the interaction at a position far from the phosphorylated threonine and close to the last transmembrane domain. We cannot rule out the possibility that PPI1 has some kinase activity and might thus be able to phosphorylate the H+-ATPase using the ATP present in the assay mixture, but this should take place on a site different from that involved in 14-3-3 binding. Since trypsin treatment removes a 7–10 kDa fragment from the C-terminus (Johansson et al., 1993; Palmgren et al., 1991; Rasi-Caldogno et al., 1993), but does not abolish PPI1 binding, the interaction PPI1/H+-ATPase must take place upstream of the major trypsin cleavage site. Since Ppi1 cDNA was first isolated using the last 103 aa of AHA1 as bait, the site of interaction must involve the first few tens of aa of the last cytosolic domain of the pump.

Present evidence suggests that it is the displacement of the autoinhibitory domain rather than 14-3-3 binding per se which mediates the activation (Jahn et al., 2002). In fact, either removal of the last 3–10 residues or mutagenesis of the last 3 residues of the H+-ATPase, abolishes 14-3-3 binding (Fuglsang et al., 1999), but does not activate the enzyme. Changes in the kinetic parameters were recorded only for the larger deletions (at least 38 aa, Regenberg et al., 1995), and this was sufficient for the activation in vitro and, for deletions of at least 51 aa, also in vivo. So, despite the 14-3-3 binding site is absent in the shorter deletions, the H+-ATPase is not activated, while deletions completely removing region II (Axelsen et al., 1999; see Introduction), generate an active enzyme. Other studies (Baunsgaard et al., 1996; Morsomme et al., 1998) suggest that altering the structure by point mutation or short deletions in regions far away from the 14-3-3 binding site (but possibly coupled to it by intramolecular bonding), might expose the 14-3-3 binding site allowing the interaction to take place and the autoinhibitory domain to be displaced. Binding of 14-3-3 proteins also requires residues within region II (Jelich-Ottmann et al., 2001) possibly involved in the activation process. Since part of the PM H+-ATPase molecules heterologously expressed in yeast are phosphorylated (Fuglsang et al., 1999) and phosphorylation levels correlate with 14-3-3 binding (Dambly and Boutry, 2001), it is possible that these mutations expose the 14-3-3 binding site to the activity of an endogenous kinase; activation will then be a consequence of yeast 14-3-3 proteins binding to the phosphorylated site. The finding that mutations most activating cause a larger fraction of 14-3-3 proteins to associate with plasma membrane (Axelsen et al., 1999) and the fact that ‘activating’ mutations, i.e. causing an activation similarly to FC (or phosphorylation and 14-3-3 proteins), are no longer prone to activation by lysophosphatidylcholine (Morsomme et al., 1996; Regenberg et al., 1995) are consistent with this hypothesis. In this context, it is possible that N-terminal domain of PPI1 partially displaces the C-terminus facilitating further displacement by 14-3-3 proteins.

Experimental procedures

Strains, media and general techniques

E. coli XL10 (Stratagene, La Jolla, CA, USA) or DH5α were used for recombinant DNA work while BL21(DE3)pLysS (Novagen, Madison, WI, USA) was employed as a host for protein expression. All strains were grown in Lennox broth base (#12780; Gibco BRL, Rockville, MD, USA). Yeast strains Y187 and Y190 for the two-hybrid screen have been described (Harper et al., 1993). YPAD (as a rich medium), and YNB + glucose (for the selection of transformants) were from Gibco BRL (#15680 and #25685), the latter being supplemented with the relevant components (adenine, Trp, Leu).

Bacterial transformation was according to the protocol of Inoue et al. (1990), while yeast transformation was according to Gietz et al. (1995). Soluble proteins were assayed with the Bio-Rad protein assay (cat. #500–0001; Bio-Rad, Hercules, CA, USA), while membrane proteins were assayed according to Markwell et al. (1978).

Plasmid construction

A DNA fragment coding for the C-terminal 103 aa from the Arabidopsis PM H+-ATPase (isoform AHA1) was amplified from EST clones 49E5 from Arabidopsis Biological Resource Center (ABRC, Ohio State University, OH, USA) using the following primers: GGTCCATGGSYGGAAVRGCITGG annealing at the sequence coding for 847SGKAW and CAGGATCCTYAMASR GTRTAGTG annealing at 946HYTVZ. The degeneration allowed amplification of other PM H+-ATPase isoforms. The PCR product was cloned into pAS-2 (Harper et al., 1993) using NcoI and BamHI restriction sites, resulting in plasmid pAS2-CODA. The frame and the identity of the cloned fragment were verified by sequencing with an upstream primer.

Two-hybrid screen

The bait plasmid was transformed into the yeast strain Y190 (Harper et al., 1993) and the expression of the fusion tested by Western blot. The vector bearing the fusion was slightly more transactivating than the control vector, requiring a 3-aminotriazole concentration slightly higher (30 versus 25 mm) to suppress residual growth on plates lacking histidine. The transformed strain (Y190 pAS2-CODA) was grown on selective medium up to a density of 2–3 × 106 ml−1, diluted 1 : 5 into rich medium and grown for 2–3 more generations before harvesting for transformation with a two-hybrid library kindly provided by the ABRC (cat. #CD4-10). This was done in several independent experiments with variable transformation efficiency ranging from 3 × 104 to 3 × 105 transformants µg−1 of DNA. Bait loss, plasmid recovery and mating was done essentially as described (Durfee et al., 1993; Harper et al., 1993).

CDNA isolation

The following strategy was devised to gain sequence information for regions outside the interaction domain. An Arabidopsis cDNA plasmid library (Minet et al., 1992) was subjected at first to 5 rounds of amplification using single Ppi1 specific primers (ExpF, gaggatccgggTCTTCAGATCTCCATTTCACGGATC, in the reaction for cloning the 3′-end and ExpR, ctggatccACCAGCGTAA GAACGGTATTTAACAAGG, in the reaction for the 5′-end; capital letters highlight sequences complementary to Ppi1 cDNA) and then 25 more cycles after adding another primer annealing to the vector in proximity of the cDNA insertion site. In this way we amplify the cDNA ends of Ppi1 clones exponentially and only linearly any other library clone. An aliquot of both reaction was then used for a nested PCR reaction containing the same vector primer and another Ppi1 specific primer (SP1: TCACGGATCA ATCCTGTG for the 3′ end and SP2: GAGTGGATCTGCTTAGC for the 5′ end). Both reaction products were cloned in pCR2.1 (Invitrogen, San Diego, CA, USA) and sent for sequence determination to either Primm (Milan, Italy) or Biostrands (Trieste, Italy).

Protein expression and purification

The Ppi1 fragment coding for the interacting domain (bp 64–370 of the cDNA, see Figure 2a) was amplified from plasmid DNA isolated from a positive yeast clone, with primers ExpF and ExpR (see above) and inserted into the BamHI site of pET15b (Novagen). The resulting plasmid was transformed into strain BL21(DE3)pLysS and expression induced in liquid cultures at 37°C starting at OD595 0.6–0.7 with 1 mm IPTG. After 2–3 h of induction, cells were cooled on ice, centrifuged and stored at − 80°C. Cell pellets were lysed in the presence of 0.7% N-lauroil sarcosine and 1.4% Triton X-100 and sonicated until a clear, non-viscous solution was obtained (Frangioni and Neel, 1993). Particulate matters were removed by centrifugation (15 min at 12 000 g) and the soluble fraction loaded onto a Ni2+-nitriloacetic acid (Ni-NTA) agarose affinity column. Protein was purified essentially as described by the Ni-NTA supplier (Qiagene, Milan, Italy). Eluted fractions were monitored by SDS PAGE, pooled and dialysed against 5 mm Bis-Tris propane (BTP, #4679; Sigma, St Louis, MO, USA)-HEPES pH 7.0, 10% (w/v) glycerol. Upon further isolation of cDNA clones, we located the first ATG at bp 106, suggesting that the first 14 aa originally present in the two-hybrid clone are in a 5′ untranslated region (and were therefore not included in subsequent fusions). The lines of evidence supporting the localisation of the ATG at bp 106 are: (i) PCR amplifications performed on the cDNA library yielded only slightly larger fragments on the 5′ end; (ii) sequence comparison against the database does not reveal larger clones, despite the presence of almost 30 ESTs (see TC115447 at http://www.tigr.org/) for A. thaliana alone; (iii) Northern analysis (unpublished data) suggests that clone length is close to the estimated size of the mRNA; (iv) several stop codons and no other plausible translation start sites were found upon examination of the genomic region preceding the first putative exon; any earlier translation start site would require the presence of additional intron/exon; (v) all gene predictions do not detect consensus splice sites for additional intron/exons.

The GST-Ppi + 88 and GST-Ppi + 612 constructs were assembled with a similar strategy, involving primer Ppi-ATG (gatggatcccatATGGGTGTTGAAGTTGTA, annealing at bases 106–123) and primer ExpR or Ppi-CIFY (gcaggatccTACTCGT ATGTAAAAGATGC, annealing at bases 2045–65), respectively: the amplified fragments (corresponding to bp 106–369 or 106–1960 and coding for peptides of 88 and 612 aa, respectively), were cloned into pGEX-2TK. Expression and lysis were as for the His6-Ppi, except that the GST-Ppi + 612 construct (whole cDNA) required the use of BL21(DE3) Codon plusTM pRIL strain (Stratagene) to obtain full-length protein. In contrast, use of the parent strain allowed expression of a protein ladder (presumably premature translational termination products) between 90 and 30 kDa.

Purification was done with GSH-Agarose (Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, UK) according to the supplier with elution in 100 mm TRIS, pH 8.0, 10 mm GSH; elution of GST-Ppi + 612 was performed in the presence of 0.2% (w/v) sarcosine in the buffer. Eluted fractions were monitored on SDS gel, pooled and dialysed as before. In the case of GST-Ppi + 612, eluted fractions were concentrated by centrifugation with Ultrafree (cat. #UFV5BCG00, cutoff 10 kDa; Milipore, Bedford, MA, USA). Sarcosine was removed by repeated cycles of concentration-dilution with HEPES pH 7.0, 5 mm glycerol 10% (w/v). Brij58 (cat. #23,599–7; Aldrich, Milan, Italy) was added to the sample (final conc. 0.1%) after the first concentration cycle.

Plant material and isolation of PM vesicles

Cell suspension cultures of A. thaliana ecotype Landsberg were grown as described in Curti et al. (1993).

In vivo treatment with FC was performed for 120 min by adding the phytotoxin to the culture medium at the final concentration of 10 µm; cells were harvested by a double centrifugation at 1000 g for 5 min. A highly purified PM fraction was obtained by a two-step aqueous two-phase partitioning system as described in Olivari et al. (2000).

Trypsin treatment

PM (1 mg ml−1 protein, 0.1 ml) was incubated for 5 min on ice in 0.75 mm DTT, 5% glycerol, 1 mm ATP, 0.25 mm CaCl2, 375 µg ml−1 Brij 58, BTP-HEPES pH 7.5 10 mm in the presence of 100 µg ml−1 trypsin. The reaction was blocked by addition of a 100-fold excess of soybean trypsin inhibitor.

PM H+-ATPase activity

The PM H+-ATPase activity was assayed in 0.2 mm EGTA, 50 mm KNO3, 2.3 mm MgSO4, 5 mm (NH4)2SO4, 0.1 mm ammonium molybdate, 1 µg ml−1 oligomycin, 100 µg ml−1 Brij 58, 5 µm FCCP, 40 mm BTP-HEPES pH 7.3 or 40 mm BTP-Mes pH 6.4, 2 units ml−1 pyruvate kinase, 2 mm PEP and 0.3 mm ATP. Plasma membranes (1–2 µg protein) were incubated with His6-Ppi or GST-Ppi fusions at the concentrations specified in figure or table legends in 50 µl of assay medium without ATP, PK and PEP at 0°C for 15 min. The volume was then adjusted to 100 µl with assay medium with added ATP, PK and PEP and the reaction was carried out for 60 min at 30°C. Released Pi was determined as described in De Michelis and Spanswick (1986). The PM H+-ATPase activity was evaluated as the difference between total activity and that measured in the presence of 100 µm vanadate (less then 10% of total activity at pH 7.3; less than 5% at pH 6.4). Data reported are the results from one experiment run in triplicate, representative of at least two experiments; se of the assays did not exceed 3% of the measured value.

Immunoprecipitation

PM preparations (120 µg protein in 50 µl) were mixed with an equal volume of immunoprecipitation buffer (20 mm Na-phosphate pH 7.2, 300 mm NaCl, 2 mm EDTA, dodecyl-β-d-maltoside (detergent : protein, 2 : 1) and 0.1 mg ml−1 leupeptin) and incubated for 15 min on ice. After the addition of 0.2% (w/v) SDS, PM was incubated at 30°C for 15 min and then centrifuged for 10 min at 20 000 g at room temperature. The supernatant was diluted 5-fold with the immunoprecipitation buffer, added with 20 µl of an anti-N-terminal H+-ATPase polyclonal antibody (Olivari et al., 1998) and a protease inhibitor cocktail (0.7 mg ml−1 tosyl-lysine, 0.1 mg ml−1 leupeptin, 2.8 mmp-aminobenzamidine, 3.5 µg ml−1 soybean trypsin inhibitor, final concentrations) and incubated overnight at 4°C. Protein A (Sigma, cat. #P-1052) immobilised on 250 µm acrylic beads at a final concentration of 1.5% (w/v) was added and the sample was incubated at room temperature for 2 h before centrifugation for 10 min at 20 000 g at room temperature. The supernatant was discarded and the pellet was washed three times with water, added with the protease inhibitor cocktail and an equal volume of 8 m urea in an acidic buffer (40 mm H3PO4, 2 mm EDTA, 2% mercaptoethanol, 4% Triton X-100, pH 3.0) and incubated for 30 min at 60°C. The sample was then centrifuged for 10 min at 20 000 g and the supernatant was resuspended in SDS-PAGE solubilization buffer.

SDS-PAGE and Western blot

Protein samples were treated as reported in Rasi-Caldogno et al. (1993) and SDS-PAGE was performed according to Laemmli (1970). Briefly, solubilized PM proteins (13–50 µg) and immunoprecipitates (corresponding to 15–30 µg of total PM protein) were loaded onto Tris-gly polyacrylamide 4%-20% gels (cat. #EC6028; Novex, San Diego, USA) or 8% gels (Novex, cat. #EC6018), subjected to electrophoresis under standard conditions and transferred to a 0.2-µm nitrocellulose membrane (cat. #401 391; Schleicher & Schuell, Legnamo, Italy). Immunodetection of H+-ATPase in PM fractions was performed using an anti-N-terminal H+-ATPase polyclonal antibody (Olivari et al., 1998) followed by an antirabbit IgG conjugated with alkaline phosphatase. In the case of immunoprecipitates, an anti-H+-ATPase monoclonal antibody (Baur et al., 1996) was used, followed by an antimouse IgG conjugated with alkaline phosphatase. Signal detection was obtained using BCIP-NBT alkaline phosphatase substrate (Sigma cat. #B-5655).

Overlay experiments

After SDS-PAGE and Western blot, the nitrocellulose membrane was saturated for 1 h at room temperature in TBS-T buffer (20 mm Tris–HCl pH 7.4, 150 mm NaCl, 0.05% (v/v) Tween (20) added with 5% (w/v) fatty acid milk. Overlay with His6-Ppi (0.5 mg ml−1), His-thioredoxin (0.5 mg ml−1), GST-Ppi + 88 (1.5 mg ml−1) or GST-Ca4+-ATPase (1.5 mg ml−1) (Bonza et al., 2000) were performed overnight at 4°C in a buffer containing 20 mm Tris-HCl pH 7.5, 75 mm KCl, 0.1 mm EDTA, 1 mm DTT, 5 mm MgSO4, 0.04% Tween 20 and 2% fatty acid milk (Fullone et al., 1998). After the overlay with His-tagged proteins the membrane was washed 3 times in TBS-T buffer and incubated in the same buffer containing India His probe-HRP (Pierce, cat. #15165) for 1 h and signal detection was obtained using the enhanced chemiluminescence system (cat. #RPM 2209; Amersham, Rockford, IL, USA). In the case of GST fusion proteins, the membrane was washed 3 times in TBS-T buffer and incubated in the same buffer containing an anti-GST polyclonal antibody (Amersham, cat. #27-4577); signal detection was performed using an antigoat IgG conjugated with horseradish peroxidase and the enhanced chemiluminescence system.

Acknowledgements

We thank the ABRC and NSF/DOE/USDA Collaborative Research in Plant Biology Program, Research Collaboration Group in Plant Protein Phosphorylation (USDA 92-37105-7675) for providing the two-hybrid library and several cDNA and genomic clones. Several control plasmids for the two-hybrid screen were kindly supplied by S. Elledge (Baylor College of Medicine, Houston, T10, USA). We greatly acknowledge help from Anna Moroni for cDNA amplification and extensive moral support, from Sara Mazzetti for ATPase assays and from Prof H. K. MacWilliams for critical reading of the manuscript. Help was also given by M. Foiani (Milan University) and laboratory members on the handling of yeast. R. Mantovani (Parma University) kindly supplied a plasmid for the production of a control fusion protein. We thank also W. Michalke for the monoclonal antibody against H+-ATPase and Luca Mizzi for constant advice on file conversions and software. Supported by grants from Ministero per le Politiche Agricole e Forestali (framework ‘Piano nazionale per le Bioecnologie vegetali’) and by the Ministero per l'Università e la Ricerca Scientifica e tecnologica (Cofin 2000 framework) to M.I.D.M.; P.M. was supported in part by a CNR fellowship.

Supplementary Material

The following material is available from http://www.blackwell-science.com/products/journals/suppmat/TPJ/TPJ1373/TPJ1373sm.htm

Figure S1. Alignment of PPI1 with putative proteins from other plant species. Sequences from Solanum tuberosum (Access. n. BE343923), Lycopersicon esculentum (AW217967), Nicotiana sylvestris (X71905), Ppi1 (AJ002020), Medicago truncatula (AW691646) were translated and aligned with ClustalX; numbering is according to PPI1 protein sequence. The accession number for S.tuberosum, L.esculentum and M.truncatula cover only part of the deduced protein sequence; a consensus was achieved by using other EST clones with significant overlap.

Table S1. Ppi gene family: genes of A.thaliana showing significant similarity to Ppi1.

Table S2. Genes from other plant species showing significant similarity to Ppi1.

Ancillary