The heptapeptide Tyr-Gly-Arg-Gly-Asp-Ser-Pro containing the sequenceArg-Gly-Asp(RGD – the essential structure recognised by animal cells in substrate adhesion molecules) was tested on epidermal cells of onion and cultured cells of Arabidopsis upon plasmolysis. Dramatic changes were observed on both types of cells following treatment: on onion cells, Hechtian strands linking the cell wall to the membrane were lost, while Arabidopsis cells changed from concave to convex plasmolysis. A control heptapeptide Tyr-Gly-Asp-Gly-Arg-Ser-Pro had no effect on the shape of plasmolysed cells. Protoplasts isolated from Arabidopsis cells agglutinate in the presence of ProNectinF, a genetically engineered protein of 72 kDa containing 13 RGD sequences: several protoplasts may adhere to a single molecule of ProNectinF. The addition of the RGD-heptapeptide disrupted the adhesion between the protoplasts. Purified plasma membrane from Arabidopsis cells exhibits specific binding sites for the iodinated RGD-heptapeptide. The binding is saturable, reversible, and two types of high affinity sites (Kd1 1 nM, and Kd2 40 nM) can be discerned. Competitive inhibition by several structurally related peptides and proteins noted the specific requirement for the RGD sequence. Thus, the RGD-binding activity of Arabidopsis fulfils the adhesion features of integrins, i.e. peptide specificity, subcellular location, and involvement in plasma membrane-cell wall attachments.
A group of components known from the anchorage-dependent animal cells is the integrins. They are specific plasma membrane receptors that recognise a family of extracellular adhesive glycoproteins (e.g. fibronectin and vitronectin) via the tripeptide sequence Arg-Gly-Asp (RGD), a motif conserved in each of the adhesive proteins (Ruoslahti 1996). On the cytoplasmic side of the plasma membrane, the receptors connect the extracellular matrix to the cytoskeleton. Together they are a specific recognition system between intra-and extracellular compartments of the cell, which has been implicated in bi-directional transmembrane signalling (Dedhar & Hannigan 1996), cell migration, polarity, differentiation and growth (Schwartz et al. 1995).
In fact, synthetic peptides containing the RGD motif specifically block several of the plant processes cited above. It provides functional evidence for similarities in membrane-to-matrix attachments in all eukaryotic cells. For example, the RGD-peptides inhibit fibroblast adhesion by interfering with the binding of integrin to fibronectin or vitronectin (Haas & Plow 1994). In the same way, thigmo-stimulated cell differentiation in Uromyces (Corrêa et al. 1996) and the perception of gravity by Chara cells (Wayne et al. 1992) are inhibited with RGD-peptides, while growth in Glycine max is stimulated (Schindler et al. 1989). There is also immunological evidence for similarities in the molecules involved in the structural continuity between the cytoskeleton and the wall. Antibodies to human vitronectin or fibronectin recognise wall-associated proteins in Fucus (Wagner et al. 1992), Lilium (Sanders et al. 1991) and Nicotiana (Zhu et al. 1993). Antibodies to integrins recognise proteins in Fucus (Quatrano et al. 1991), Glycine max (Schindler et al. 1989), Arabidopsis and Chara (Katembe et al. 1997). However, these studies did not allow any further study with the identification of a protein or a gene homologue to the animal adhesion proteins or integrins. The immunological approaches therefore appear to have been unsuccessful in identifying adhesion molecules in plant cells. Indeed, the tobacco protein supposed to be a plant fibronectin was finally identified as the translation elongation factor 1-α (Zhu et al. 1994). The lily vitronectin-like protein was revealed as a phosphoglycerate mutase (Wang et al. 1996). We also used polyclonal antibodies raised against a human vitronectin receptor (integrin αVβ3/β5) to detect integrin-like protein by immunoselection of a cDNA expression library from Arabidopsis. The isolated clone exhibits structural homologies with VAP-33, a protein involved in membrane trafficking in neuronal tissues (Galaud et al. 1997). These results are intriguing and strongly question whether the homology between the animal and plant proteins exists.
Whatever the molecules involved, if similarities in the mechanisms of membrane-wall attachments exist in all eukaryotic cells, then the plant receptors at the plasma membrane probably recognise the RGD motif. Indeed, the inhibition of plant cell function or cell adhesion with RGD-containing peptides probably proceeds by masking the extracellular domains of the receptor. In this context, the present communication deals with synthetic peptides and purified membranes from Arabidopsis cells to provide evidence for specific binding sites of the RGD motif. Also presented are the effects of the peptides on membrane-wall attachments, and on the status of protoplast agglutination, both in onion epidermis cells and cultured Arabidopsis cells. The data suggested that the RGD binding sites are at adhesion sites and play a role in the linkages between the plasma membrane and the cell wall.
Effect of RGD peptides on onion and Arabidopsis cells upon plasmolysis
Physical linkages were visualised by plasmolysing onion (monocot) epidermal cells. As the cytoplasmic volume decreased, the plasma membrane pulled away from the wall, except at distinct adhesion sites (Fig. 1a): the points of membrane-wall attachment and the cytoplasmic strands stained brightly (Hechtian strands). When the cells were plasmolysed in the presence of the Tyr-Gly-Arg-Gly-Asp-Ser-Pro heptapeptide, protoplasts separated from the transversal walls with hemispheric forms (Fig. 1b) and no more adhesion sites could be distinguished. Undifferentiated cells from a suspension culture of Arabidopsis thaliana (dicot) also plasmolysed in a concave form (Fig. 1c,d): the plasma membrane separates from the cell wall with some areas of cell wall-membrane still in contact to form concave pockets (Oparka 1994). In contrast, cells plasmolysed in the presence of Tyr-Gly-Arg-Gly-Asp-Ser-Pro present a convex form (Fig. 1e,f): the plasma membrane uniformly separates from the cell-wall within 1–30 min of contact with the peptide solution to give spherical protoplasts. Inversion of the RGD sequence to DGR in the heptapeptide abolished the effect observed for Tyr-Gly-Arg-Gly-Asp-Ser-Pro in both onion and Arabidopsis cells (Fig. 1g,h). The differences in plasmolysis are specific for the RGD sequence since the inversion of only that particular motif prevented any effect. The loss of Hechtian strands in onion, and the change in plasmolysis shape in Arabidopsis because of RGD additions suggest an effect on the areas of contact between the cell wall and the membrane. These peptides may be acting on components of the cell wall or the plasma membrane.
Adhesion of protoplasts isolated from Arabidopsis cells
It was then of interest for us to determine whether the Arabidopsis protoplasts could bind to proteins containing exposed RGD peptides, and we used ProNectinF, a genetically engineered protein of 72 kDa containing 13 RGD sequences. The rationale for this experiment is that protoplasts containing RGD binding sites should adhere to ProNectinF. After purification, the freshly isolated protoplasts dispersed in a homogeneous suspension, and no clumps or aggregates can be observed (Fig. 2a). Within 5 min of contact with a 3.5 μm ProNectinF, the protoplasts extensively agglutinated in one pack (Fig. 2b). This effect should be the consequence of the multiple RGD motifs on ProNectinF allowing the attachment of several protoplasts to a single ProNectinF molecule. The addition of a protein without RGD sequences as BSA did not show any agglutination (Fig. 2c). If the agglutinating effect of ProNectinF is RGD-dependent, then the addition of a 100-fold excess of the Tyr-Gly-Arg-Gly-Asp-Ser-Pro heptapeptide to the agglutinated protoplasts should disrupt the agglutination. This control showed that RGD peptides did interfere with ProNectinF and eliminated the agglutination effect (Fig. 2d), whilst the inverted DGR-peptide had no disrupting effect. These data are consistent with the presence of Arabidopsis plasma membrane receptors that recognise foreign ligands carrying an exposed RGD motif.
Binding of radiolabelled RGD-heptapeptide to Arabidopsis purified plasma membrane
The presence of a tyrosine group at the N-terminal of the active Tyr-Gly-Arg-Gly-Asp-Ser-Pro heptapeptide allowed adding a radiolabel. The radio-iodination of the tyrosine phenoxyl ring was performed with chloramine T as the oxidising agent. A molar ratio of 125I to heptapeptide of 2.5 typically yielded a radio-mono-iodinated heptapeptide with specific radioactivity of approximately 80 TBqmmol–1. Under these conditions, 50% of 125I was incorporated into the mono-iodinated heptapeptide. The presence of an iodo-phenoxyl group at one end of the heptapeptide did not alter its activity significantly (see below, Fig. 5).
We next investigated the presence of RGD-containing heptapeptide binding sites in purified membrane fractions isolated from Arabidopsis cells. We used the free-flow electrophoresis technique to simultaneously obtain a purified plasma membrane fraction (the less electronegative one), a purified tonoplast fraction (the most electronegative one), and a third fraction containing a mixture of ER, Golgi, mitochondria and other membranes (Bardy et al. 1998). The time course of 125I-Tyr-Gly-Arg-Gly-Asp-Ser-Pro binding to plasma membrane is shown in Fig. 3. The association of the labelled RGD-heptapeptide to its binding site reached a maximum after 15 min and remained steady up to 60 min. Specific binding after 15 min averaged 60% of the total binding which represents less than 1% of the total added radioactivity. A rapid dissociation was observed following the addition of unlabelled peptide: 50% dissociation occurred in less than 2 min. The latter data correlated well with the rapid loss of attachments between the plasma membrane and the cell wall observed both in onion and Arabidopsis cells.
On the other hand, the plasma membrane preparation was the only one to bind the radiolabelled peptide (Table 1). Neither the mixture of ER, Golgi and mitochondria membranes, nor the purified tonoplast fraction binds the RGD-heptapeptide. All the specific binding activity co-purifies with the vanadate-sensitive ATPase activity, a plasma membrane marker enzyme. Studies of ATPase latency indicated that these vesicles are cytoplasmic-side-in (Bardy et al. 1998), the right orientation to measure the activity of cell surface receptors. In addition, the binding was strongly inhibited by trypsin treatment (Table 1). These results show the protein nature of the receptor and its location at the plasma membrane.
Table 1. Specific RGD-binding and ATPase activities of different membrane fractions
a Membrane vesicles from Arabidopsis cells were fractionated by preparative free-flow electrophoresis. The purity of membrane fractions was assessed both by the determination of marker enzyme activities and the reactivity of immunological probes (Bardy et al. 1998).
125 I-RGD heptapeptide binding was determined by subtracting non-specific from total binding.
c Inhibition of initial ATPase activity by 100 μm vanadate.
ND: not determined.
6.0 ± 0.4
Trypsinised plasma membrane
0.8 ± 0.5
ER, Golgi, Mitochondria
0.7 ± 0.6
0.3 ± 0.7
Saturability of the binding of radiolabelled RGD-heptapeptide to plasma membrane
The saturability of the binding of RGD-heptapeptide to plasma membrane vesicles was examined. The vesicles were incubated with a constant amount of labelled peptide and various concentrations of unlabelled peptide. The results of this experiment (Fig. 4a) prove that the binding was saturable, although over a wide ligand concentration range. Analysis of the specific binding data by the methods of Scatchard (Fig. 4b) or Woolf (Fig. 4c) indicated the presence of two classes of high affinity binding sites with Kd1≈ 1 nm and Kd2≈ 40 nm, and maximal binding capacities Bmax1≈ 20 fmol mg–1 protein and Bmax2≈ 200 fmol mg–1 protein. The Hill coefficient (Fig. 4d), markedly below one (n = 0.78), confirmed the presence of several binding sites with different affinities.
Specificity of the RGD-heptapeptide binding sites
The specificity of the binding sites was examined by testing several structurally related peptides and proteins for their ability to inhibit the binding of iodinated Tyr-Gly-Arg-Gly-Asp-Ser-Pro peptide. The competition curves (Fig. 5) are clearly biphasic, except for the Tyr-Gly-Arg-Gly-Glu-Ser-Pro peptide, supporting the two-site conclusion from the ligand saturation data. Moreover, the binding sites exhibited strong selectivity towards the RGD motif. ProNectinF required the lowest concentration to achieve 50% inhibition of binding, i.e. 0.4 nm. The RGD-peptides, Tyr-Gly-Arg-Gly-Asp-Ser-Pro and Arg-Gly-Asp-Ser, are equally effective inhibitors of ligand binding requiring 7 nm to achieve 50% inhibition. The cold iodinated RGD-heptapeptide is slightly less efficient, 20 nm to reduce ligand binding by 50%. Fibronectin, an adhesive glycoprotein localised in the extracellular matrix of animal cells containing one RGD motif required higher concentrations (90 nm). Finally, substitution of Asp residue with Glu greatly reduced the ability of the heptapeptide to inhibit ligand binding (1 μm), and the inverted DGR-heptapeptide was not recognised at all.
It has been demonstrated that interactions between the wall and the plasma membrane are essential for plant development. A growing body of research supports the notion that the cytoskeleton-plasma membrane-cell wall continuum forms an interactive structure for perception and transduction of signals in plants (reviewed by Reuzeau & Pont-Lezica 1995;Roberts 1989;Wyatt & Carpita 1993). Several studies pointed out the biological activity of RGD-containing peptides in disrupting the attachments between plasma membrane and cell wall, or in inhibiting different plant physiological processes. For instance, the adhesion of salt-adapted tobacco protoplasts was noticeably disrupted with 1 mm of the Gly-Arg-Gly-Asp-Ser-Pro hexapeptide (Zhu et al. 1993). Twenty-five per cent of the membrane-wall adhesions in Pelvetia zygotes were disrupted with 50 μm of the Arg-Gly-Asp-Ser tetrapeptide (Henry et al. 1996). In the same way, the development of soybean root cells grown in suspension culture was significantly enhanced with 0.9 mm of the Gly Arg-Gly-Asp-Ser-Pro hexapeptide (Schindler et al. 1989). The tetrapeptide Arg-Gly-Asp-Ser (0.4 mm) completely inhibits gravi-sensing in Chara (Wayne et al. 1992). Finally, cell differentiation (appressorium formation) in Uromyces germlings was inhibited with four different peptides, e.g. Arg-Gly-Asp, Arg-Gly-Asp-Ser, Gly-Arg-Gly-Asp and Gly-Arg-Gly-Asp-Gly-Ser-Pro-Lys (0.5–2.0 mm) (Corrêa et al. 1996). All these studies push forward a parallel with the integrin–ligand interactions, although the biological activity of ligands (proteins or peptides) was obtained at considerably lower concentrations. For instance, soluble fibronectin (0.1 μm) or Gly-Arg-Gly-Asp-Ser-Cys hexapeptide (3.0 μm) cause the inhibition of mammalian cell attachment (Pierschbacher & Ruoslahti 1984). The RGD recognition system at the cell surface has an essential role in establishing attachments, and the plant plasma membrane likely houses receptors recognising the RGD motif.
We provide the first evidence that Arabidopsis membranes possess binding sites for RGD-containing peptides. The binding site, which is detectable only in the plasma membrane, appears to be a protein. Binding of iodinated RGD-heptapeptide is saturable over a wide range of ligand concentration (1 nm to 50 nm). This is comparable with the range of fibrinogen concentrations (1 nm to 80 nm) required to saturate binding to platelet integrin αIIb-β3 (Wippler et al. 1994). Two classes of binding sites, with a very high affinity for the radiolabelled heptapeptide (apparent Kd1≈ 1 nm, Kd2≈ 40 nm), were present in Arabidopsis plasma membrane. For fibrinogen binding to purified integrin αIIb-β3, two classes of binding sites were also found: the dissociation constants were 1.0 nm for the higher affinity and 41.7 nm for the lower affinity (Wippler et al. 1994). The higher affinity binding site in Arabidopsis contributed for 10% to the maximal binding capacity. Binding was reversible suggesting that the active heptapeptide does not become covalently attached to the binding protein. Thus, these binding sites have properties characteristic of a physiologically effective receptor.
The Tyr-Gly-Arg-Gly-Asp-Ser-Pro heptapeptide was biologically active in modifying the shape of the plasmolysed cells, presumably by disrupting the membrane-wall attachments in both Arabidopsis and onion epidermal cells. The effective concentration of the peptide (4.5 μm) was considerably lower than in other plant studies, and very close to that of the hexapeptide causing the inhibition of mammalian cell attachment (3.0 μm). The non-RGD peptides had no effect. The agglutination behaviour of the Arabidopsis protoplasts equally revealed the specific requirement for the RGD sequence. Contrary to the tobacco protoplasts adapted to high-salt concentrations that agglutinates after their isolation (Zhu et al. 1993), the agglutination of Arabidopsis protoplasts was induced by ProNectinF (3.5 μm). Disruption of the agglutination was obtained by a 100-fold excess of RGD-peptide (400 μm necessary to compete with the RGD sequences of ProNectinF). Again, the non-RGD peptides were inefficient. The specific requirement for the RGD sequence, which is the main feature of the integrin–ligand interactions in mammalian cells, was also noted in plants. Indeed, the non-RGD peptides were inefficient in inhibiting gravi-sensing in Chara (Wayne et al. 1992) and cell differentiation in Uromyces (Corrêa et al. 1996), or in stimulating growth in soybean (Schindler et al. 1989). In Arabidopsis, the non-RGD peptides Tyr-Gly-Arg-Gly-Glu-Ser-Pro or Tyr-Gly-Asp-Gly-Arg-Ser-Pro, that have no biological activity, are also several-hundredfold less efficient or are not competitive inhibitors for the binding sites studied at the plasma membrane.
If the binding sites in Arabidopsis show a remarkable specificity for RGD-peptides, two proteins were also competitive inhibitors, namely ProNectinF and fibronectin. ProNectinF is the best competitive inhibitor in good agreement with the 13 RGD sequences contained in that engineered 72 kDa protein. Interestingly, fibronectin, which contains one RGD sequence, is 10 times less efficient as a competitive inhibitor in the binding assays than the synthetic RGD-peptides. Therefore, the situation found in our plant material is exactly opposite to the situation found in mammalian cells (see above, Pierschbacher & Ruoslahti 1984). It suggests that the ligands of the RGD receptor in plants are different from fibronectin, as the plant cell wall is different in structure from the animal extracellular matrix (Roberts 1989). The plant natural ligands should be, at least, equally effective as competitive inhibitors in the binding assays than the synthetic peptides. Among the 2600 sequenced proteins that contain the RGD sequence, only a minority of them are the natural ligands of integrins (Ruoslahti 1996). About 100 plant proteins contain the RGD motif (per Swiss Prot database 10/30/97). As examples, a 4.4 kDa soybean protein rich in aspartic acid (Odani et al. 1987) and an in planta-induced product gene of the potato late blight fungus, Phytophtora infestans (Pieterse et al. 1994), have been reported. In the latter work, the gene is highly expressed during the pathogenic interaction of P. infestans with its host plant potato and the protein secreted, suggesting that the gene product has a function in the infection process. It is worth noting that both proteins have no homology with animal adhesive proteins: it may be one reason for the unsuccessful immunological approach to identify plant adhesive proteins until now. On the other hand, the yeast genome is now fully sequenced, and no protein with obvious homology to an integrin has been uncovered. However, RGD peptides were active in yeast physiological processes (Bendel & Hostetter 1993). The nature and the number of proteins participating in membrane-wall linkages in plants remain to be discovered.
In conclusion, we have found that the plasma membrane possess RGD-binding sites together with RGD-dependent linkages between the plasma membrane and the cell wall. In this way, the RGD-binding activity of Arabidopsis fulfils the adhesion features of integrins. The RGD-containing peptides have been instrumental in the identification of the integrin receptors. We hope that the high affinity of the plant plasma membrane proteins for the RGD sequence will lead to the dissection of the RGD-recognition system in plants.
Bulbs of sweet red onion (Allium cepa) were obtained from producers. M. Axelos (UMR 215, CNRS-INRA, Castanet-Tolosan, France) provided cell suspension cultures of Arabidopsis thaliana. Arabidopsis cells were grown on Gamborg liquid medium (Axelos et al. 1992). Fifteen ml cell suspension were routinely transferred to 300 ml new medium in 1000 ml Erlenmeyer every 2 weeks and shaken (150 r.p.m., New Brunswick orbital shaker model G 10–21) in continuous light (60 W m–2 Philips fluorescent TL 65 W) at 26°C. Before use for analysis, cells were transferred to fresh culture medium and maintained in the dark with shaking prior to harvesting. Seven-day-old cells (exponential phase of growth) were used in all experiments.
Microscopy and protoplast isolation
Onion epidermal or Arabidopsis cells were incubated for 5 min in 0.01% fluorescein diacetate in fresh culture medium. Cells were plasmolysed with the addition of 0.5 m CaCl2. Stock solutions of peptides were applied (4.5 μm final concentration) to the cells for 5–30 min at room temperature then plasmolysed. Cells were observed with a Leitz DB-IRBE inverted microscope in either bright field or epifluorescence using a Multiscan 2 AMKO monochromator at 490 nm excitation light. Images were acquired with a Photonic Science Colour Coolview CCD camera, and Image-Pro Plus image analysis software (Media Cybernetics, USA).
Protoplasts were generated by enzymatic digestion of the cell wall in a medium containing 0.55 m sorbitol buffered to pH 5.5 (Mes) and a mixture of 1.0% Caylase (Cayla, Toulouse, France) and 0.3% Pectolyase Y23 (Seishin Pharmaceuticals, Japan) for 30 min at 35°C with gentle shaking. After filtering the digestion through nylon net (60 μm pore-width), the protoplasts were purified by centrifugation on a density gradient at 100 g for 5 min. The protoplasts floated in 10 mm Hepes pH 7.3 buffer containing 0.55 m sorbitol. A droplet (50 μl) of the protoplast suspension was used for visualisation under the microscope. Stock solutions of peptides and proteins were applied to the adhesion assay to give appropriate final concentrations. ProNectinF, in particular, was solubilised according to the manufacturer’s specifications (Protein Polymer Technologies, San Diego, USA). ProNectinF is a protein polymer produced from a synthetic gene via bacterial fermentation. It incorporates 13 copies of the RGD cell attachment ligand of human fibronectin interspersed between repeated structural peptide segments. The RGDS peptide was obtained from Bachem Biochimie (France) and the human fibronectin from Chemicon (Temecula, USA).
Peptide synthesis and labelling
The synthesis of three peptides, Tyr-Gly-Arg-Gly-Asp-Ser-Pro, Tyr-Gly-Arg-Gly-Glu-Ser-Pro and Tyr-Gly-Asp-Gly-Arg-Ser-Pro, was done automatically by stepwise Fmoc-t-butyl solid phase synthesis (Fields & Noble 1990) in a Synergy Applied Biosystems peptide synthesiser. Crude synthetic peptides were purified by reverse-phase high-pressure liquid chromatography (HPLC). The purified peptides were characterised by amino acid composition, sequence analysis (Ferrara et al. 1987) and laser desorption time of flight by mass spectrometry on a Lasermat spectrometer (Finnigan) as described previously (Hillenkamp et al. 1991). Iodination of the Tyr-Gly-Arg-Gly-Asp-Ser-Pro peptide was carried out at room temperature in a glove box equipped with charcoal filter. In a microfuge tube, 18 MBq of Na[125I] (0.23 nmol) in 10 μl of solution at pH 8 (125I – 629 GBq mg–1, from Du Pont de Nemours) were added with 10 μl of 0.3 m phosphate buffer pH 7.4 containing 1 nmol of the peptide. The reaction was started by the addition of 3 μl chloramine T (37 μg ml–1) in phosphate buffer; another addition of 3 μl chloramine T was performed 1 min later. The reaction was allowed to proceed for 5 min and was stopped with 10 μl of tyrosine (1 mg ml–1) in phosphate buffer. The iodinated peptide was purified by reverse-phase HPLC. Routinely, radioactivity incorporation of 50% into the mono-iodinated heptapeptide was obtained.
Purification of plasma membrane and binding assays
The purified plasma membrane vesicles were isolated from microsomes of Arabidopsis thaliana by preparative free-flow electrophoresis using a Tris-borate-based electrophoresis medium (Canut et al. 1991). The purity of the membrane fractions was assessed both by the determination of marker enzyme activities and by the reactivity of immunological probes (Bardy et al. 1998). Based on the measurement of ATPase latency, the plasma membrane fraction appeared essentially to be a population of cytoplasmic-side-in vesicles. Proteolytic digestion of plasma membrane vesicles was performed with trypsin (0.05 mg per mg of membrane protein) in binding buffer (see below) for 5 min at 20°C. The reaction was stopped by pelleting membrane vesicles and three subsequent washes with binding buffer.
The binding of radiolabelled heptapeptide to plasma membrane vesicles was carried out in binding buffer (50 mm BTP-HCl, pH 7.0, containing 0.25 m sucrose) as follows. Unless otherwise indicated, incubation mixtures contained plasma membrane vesicles (50 μg of protein in binding buffer) and 125I-labelled heptapeptide (7. 105 dpm, 0.15 pmol) in a total volume of 0.1 ml of binding buffer. Incubations were terminated by rapid filtration with vacuum on Millipore GVWP filters (0.22 μm) in a filtration manifold (sample collector 1225, Millipore). The filters were washed six times with 1 ml of washing buffer (5 mm BTP-HCl, pH 7.0, containing 0.25 m sucrose), and the amount of radioactivity remaining on the filters was measured directly using a γ counter.
This research was supported by the Université Paul Sabatier, Toulouse, by the Centre National de la Recherche Scientifique (grants 96037 and 96N88/0010), and by the Région Midi-Pyrénées.