The mobility of intramembrane particles in non-haemolysed human erythrocytes: Factors affecting acridine-orange-induced particle aggregation

Lateral mobility of the integral membrane proteins can be investigated either by freeze-fracture electron microscopy or by measuring the spontaneous lateral diffusion of these proteins by means of physical techniques. Aggregatability of the intramembrane particles has been extensively studied in the erythrocyte membrane (Pinto da Silva, 1972; Speth et al. 1972; Elgsaeter & Branton, 1974; Elgsaeter et al. 1976; Gerritsen et al. 1979). While these experiments were performed on ghost membranes only, recent studies have also pointed out intramembrane particle clustering in non-haemolysed erythrocytes upon treatment with membranepenetrating compounds such as Acridine Orange (Lelkes et al. 1983) and chlorpromazine (Lieber et al. 1984).

The physical techniques for investigating spontaneous lateral diffusion of membrane proteins make it possible to determine long-range lateral motions quantitatively, but cannot be used to study local lateral mobility (Sheetz, 1983). On the other hand, aggregation of the intramembrane particles studied by freeze-fracturing (although rather difficult to quantify) might be used to assess local lateral mobility of membrane proteins in the red cell membrane where the intramembrane particles are known to represent integral membrane proteins.

Bearing this in mind, we decided to study the Acridine-Orange-induced intramembrane particle aggregation in more detail. Here we report that under appropriate conditions rapid and significant particle clustering can be induced by submillimolar concentrations of this dye without agglutination and vesiculation of the erythrocytes. We also describe the effects of both spectrin cross-linking and cytoskeletal destabilization on the aggregability of intramembrane particles in non-haemolysed human erythrocytes.

Acridine Orange (Chroma) was dissolved in 0·15M-phosphate buffer, pH74, 6·5 and 5·5, as well as in phosphate-buffered saline, pH 7·4, and Tris-buffered saline, pH 7·4 and 8·5. At all indicated pH values 0·25mM, 0·5mM, 1·0mM, 2·0mM and 5·0mM solutions of Acridine Orange were prepared. The solutions were filtered and used immediately after adjusting their pH.

Freshly drawn, heparinized human blood samples were washed three times in isotonic NaCl, then incubated in Acridine Orange solution for different periods of time. Incubations were terminated by adding 2-5% glutaraldehyde to the suspensions. The haematocrit of all suspensions was 0·025 ll⁻¹ (50μ1 of red cells in 2-0ml solutions). All incubations were carried out at room temperature.

In order to change the intracellular pH, an approximately 10% red cell suspension in unbuffered saline was slowly titrated with 0·15M-NaOH and 0·15M-HC1, respectively. The alkali and acid were added dropwise to the suspension under constant stirring and pH control. This procedure was continued until the pH of the suspension reached the desired value. The suspension was then centrifuged and partial haemolysis was carried out by adding 1·5 vol. of distilled water to 1 vol. of packed, titrated erythrocytes. Intracellular pH values were estimated by measuring the pH of the haemolysates of the titrated cells. By this method, the intracellular pH values of control cells washed in saline were found to be between 6·8 and 7·0. A 50 μl sample of titrated red cells (intracellular pH 8·0) was added to 2·0 ml of 5·0 mM-Acridine Orange in phosphate buffer, pH 7·4, and fixed with glutaraldehyde after a 10-s incubation. In other experiments 50gl of erythrocytes (intracellular pH 5·8) was added to 2·0 ml of 1·0 mM-Acridine Orange in phosphate buffer, pH 7·4, for 30 s and fixed as described above.

For Diamide treatment red cells were washed three times in isotonic saline, diluted to 10% in phosphate-buffered saline, pH 7·4, containing 2mM and 5 mM-Diamide (Serva), then incubated at 37°C for 30 min. After incubation the erythrocytes were washed twice in saline.

Red cells were washed as described above, resuspended in phosphate-buffered saline, pH 7·4, then heated for 10 min at 48°C and 50°C, respectively. Samples pretreated with Diamide and heating were incubated in 5 mM-Acridine Orange in 0·15 M-phosphate buffer, pH 7·4, for 10s at room temperature and then fixed.

Fixed red cells were washed with 0·1 M-cacodylate buffer, glycerinated and freeze-fractured as described previously (Lelkes et al. 1983). Replicas were investigated in a Philips EM 300 electron microscope at 60 kV accelerating voltage.

Normally, the intramembrane particles show random distribution in the plane of the erythrocyte membrane (Fig. 1A). As described previously (Lelkes et al. 1983), when red cells are incubated in solutions of Acridine Orange, a concentrationdependent, reversible aggregation of the intramembrane particles occurs.

The present experiments show that the rate of Acridine-Orange-induced particle aggregation is not only concentration-but also pH-dependent. While at physiological pH, 5 mM-Acridine Orange dissolved in 0·15 M-phosphate buffer caused marked particle aggregation within 10s (Fig. 1B), no aggregation was observed at pH 6·5 within this short period. At this pH value only slight aggregation was observed after 5 min (Fig. 1C), while at pH 5·5 no aggregation was found even after a 30-min incubation (Fig. 1D).

At pH 7·4, 1·0mM-Acridine Orange dissolved in phosphate buffer caused no significant aggregation during a 30-s incubation (Fig. 2A). In order to estimate the role of the intracellular pH on Acridine-Orange-induced particle aggregation, we titrated an unbuffered red cell suspension with 0·15 M-HC1 to lower the intracellular pH to 5·8. The titrated red cells were then incubated in 1·0 mM-Acridine Orange, pH 7·4, for 30s. Relative to the extracellular pH, changes in the intracellular pH were found to have the opposite effect on the rate of particle clustering because the decreased intracellular pH significantly enhanced particle aggregation (Fig. 2B). Moreover, raising the intracellular pH to 8·0 by titrating the cells with 0·15 M-NaOH strongly decreased particle clustering compared to control when the titrated cells were incubated in 5·0 mM-Acridine Orange, pH 7·4, for 10s (Fig. 2C,D). It should be emphasized that titration of the erythrocyte suspension either by alkali or acid did not on its own cause particle clustering.

When red cells were incubated in solutions of Acridine Orange dissolved in Trisbuffered saline, a more significant particle clustering was obtained than in identical concentrations of Acridine Orange dissolved in phosphate. For example, 1·0 mM-Acridine Orange in Tris-buffered saline, pH 7·4, caused almost as intense aggregation as that obtained in 5·0 mM-Acridine Orange in phosphate at the same pH value (Fig. 3A). Acridine Orange at 0·25mM in Tris-buffered saline, pH7·4, still caused significant particle aggregation within 30s (Fig. 3B), although, at this concentration, only discocyte-stomatocyte transformation occurred; neither agglutination nor vésiculation of the cells was seen (Fig. 4A,B).

In Diamide-treated cells Acridine Orange caused the same rate of particle aggregation as in the control (Fig. 5A,B).

Heat treatment caused a dramatic increase in the aggregatability of the particles in response to Acridine Orange, especially in erythrocytes heated to 50°C. All fracture faces showed an almost maximal rate of particle clustering (Fig. 5C), but no haemolysis accompanied the aggregation of the particles.

Like titration of the erythrocytes, none of the pretreatments described above led, on its own, to particle aggregation.

According to Sheetz & Singer (1974) amphipathic, tertiary amine compounds can diffuse across the red cell membrane only in discharged form and regain a positive charge inside the cell. In an aqueous medium, Acridine Orange (3,6-dimethylamino-acridine hydrochloride) dissociates to charged acridine base and chloride, while the former can reversibly release a proton and form a discharged acridine base. This latter dissociation depends on the pH of the medium: it is increased at alkaline and decreased at acidic pH. It follows, therefore, that alkaline pH increases and acidic pH decreases the penetrating ability of Acridine Orange. Since we demonstrated previously (Lelkes et al. 1983) that this dye in all probability causes particle aggregation by perturbing the inner surface of the red cell membrane, we decided to study the effect of both extracellular and intracellular pH on Acridine-Orange-induced particle clustering. We presumed that changes in the extracellular and intracellular pH can modify particle aggregation by altering the ratio of discharged acridine base to the charged one.

We have found, in fact, that the Acridine-Orange-induced particle aggregation strongly depends on the pH of the dye. Lowering the pH of the Acridine Orange solutions gradually decreased and finally abolished aggregation (Fig. 1). Furthermore, we observed that changes in the intracellular pH had the opposite effect on the intensity of particle aggregation, i.e. low intracellular pH increased and elevated intracellular pH decreased aggregation (Fig. 2). Since acidic intracellular pH increases and alkaline intracellular pH decreases reprotonation of the discharged acridine base inside the cells, the latter observations indicate that it is only the charged acridine base that causes particle aggregation by perturbing the inner surface of the membranes of non-haemolysed erythrocytes.

While we found previously (Lelkes et al. 1983) that the aggregation begins at a concentration of about 2-5 mM-Acridine Orange, the present experiments show that much lower concentrations of Acridine Orange can induce very rapid and significant particle clustering. This discrepancy can be explained by two facts. First, it is known that millimolar concentrations of Acridine Orange can acidify an unbuffered or only slightly buffered medium. Since in our previous experiments Acridine Orange was dissolved in isotonic saline buffered with 5–10mM-phosphate only, acidification of the medium, which leads to a less intense and slowly occurring particle aggregation at a given Acridine Orange concentration, must have occurred. Second, the rate of aggregation depends not only on the concentration and pH of the Acridine Orange solution, but also on the buffers in which Acridine Orange is dissolved. In Trisbuffered saline this dye caused significant particle aggregation at 0-25 mM.within 30 s (Fig. 3), which is about one order of magnitude less than that which caused some aggregation in our previous experiments. Tris, like Acridine Orange, is also a membrane-penetrating tertiary amine compound. Although it cannot induce particle aggregation by itself, it is conceivable that Tris may to some extent neutralize the negative charges on the inner surface of the red cell membrane. If this is so, in the presence of Tris a lower concentration of Acridine Orange is needed to bring about particle aggregation.

While it seems to be established that the dye causes aggregation by perturbing the inner red cell surface it is not yet clear which membrane components are involved in the phenomenon. In this respect, two components, spectrin and anionic phospholipids, can be taken into account. Elgsaeter et al. (1976) found that low pH, Ca²⁺ and basic proteins, which are known to induce particle aggregation in ghosts, also cause precipitation of spectrin in vitro. On the basis of these results, the authors suggested that it is the microprecipitation of spectrin on the inner surface of ghost membranes that causes particle aggregation. Gerritsen et al. (1979), however, also found particle aggregation in cytoskeleton-free ghosts upon treatment with low pH and Ca²⁺. Their results were explained by the lateral phase separation of anionic phospholipids caused by these agents.

Acridine dyes are known to cause protein precipitation (Steinbuch, 1980). It might be said, therefore, that Acridine Orange acts by a mechanism similar to the one proposed by Elgsaeter et al. (1976). Our preliminary experiments, however, show that Acridine Orange causes maximal particle aggregation in large, haemoglobincontaining, cytoskeleton-free erythrocyte vesicles (Lelkes et al. 1984), indicating the role of lipid phase separation in the aggregation. The involvement of such a mechanism in the Acridine-Orange-induced aggregation is also supported by our preliminary calorimetric experiments made on phosphatidylcholine-phosphatidyl-serine liposomes. These results will be published in detail elsewhere.

It is known that the erythrocyte cytoskeleton strongly affects the lateral mobility of integral membrane proteins as well as the aggregatability of the intramembrane particles. Partial removal of spectrin increased both glycoprotein lateral mobility (Golan & Veatch, 1980) and particle aggregatability (Elgsaeter & Branton, 1974) in ghosts. Therefore, we decided to examine the possible effects of both spectrin cross-linking and cytoskeletal destabilization on the Acridine-Orange-induced intramembrane particle aggregation.

Diamide, which cross-links proteins via disulphide bridges, was reported to cause slight particle aggregation in ghost membranes (Kurantsin-Mills & Lessin, 1981) as well as a decrease in the lateral diffusion of band 3 proteins in fused pairs of erythrocytes (Smith & Palek, 1982). This latter observation was made on erythrocytes in which Diamide cross-linked spectrin specifically without affecting integral membrane proteins. In our experiments no aggregation of intramembrane particles could be observed upon treatment with both 2mM and 5 mM-Diamide, nor could we detect any decrease in the Acridine-Orange-induced particle aggregation (Fig. 5B). It seems likely, therefore, that inside the interstices of the cytoskeletal meshwork the particles can diffuse or aggregate freely, even if covalent stabilization of the transitory cytoskeletal connections decreases lateral diffusion over large distances.

Heating the erythrocytes to 50°C causes thermal unfolding of spectrin (Brandts et al. 1977). This results in vésiculation and fragmentation of the erythrocytes, suggesting a profound decrease in the mechanical stability of the erythrocyte membrane (Coakley & Deeley, 1980). In heat-treated red cells Acridine Orange induced an almost maximal particle aggregation (Fig. 5C). It is known that the large vesicles derived from the heat-treated cells are deficient in spectrin (Herrmann et al. 1985). This raises the possibility that this reduced spectrin content is responsible for the increased particle aggregatability observed. However, investigating fracture faces, even large vesicles could be distinguished from the erythrocytes by their different sizes, and no significant difference could be established in particle aggregation between vesicles and erythrocytes’. We believe that the strongly increased aggregatability is due to disruption of the normal cytoskeletal structure and aggregation of the skeletal elements (Palek et al. 1981). This redistribution may lead to the formation of extended skeleton-free areas beneath the bilayer. Since the remaining skeletal filaments are not covalently associated with each other, the particles that are able to diffuse freely inside these large areas can also move from one area to another. This may result in the highly increased mobility of the intramembrane particles in the heat-treated cells.

The results obtained in this study indicate that the Acridine Orange method may be a suitable means of assessing the lateral mobility of the integral membrane proteins in different cytoskeletal disorders, such as hereditary spherocytosis (Burke & Shotton, 1983; Agre et al. 1985) as well as hereditary elliptocytosis and pyro-poikilocytosis (Cohen & Branton, 1981). Moreover, this aggregating procedure may be useful in detecting possible changes in the local mobility of the integral membrane proteins in cases of haemoglobinopathies with altered haemoglobin-membrane interaction, such as Heinz-body anaemias and sickle cell disease. The combination of this procedure with the so-called label-fracture technique (Pinto da Silva & Kan, 1984) is under way.