The influence of membrane physical properties on microvesicle release in human erythrocytes
- 5.9k Downloads
Exposure of human erythrocytes to elevated intracellular calcium causes fragments of the cell membrane to be shed as microvesicles. This study tested the hypothesis that microvesicle release depends on microscopic membrane physical properties such as lipid order, fluidity, and composition. Membrane properties were manipulated by varying the experimental temperature, membrane cholesterol content, and the activity of the trans-membrane phospholipid transporter, scramblase. Microvesicle release was enhanced by increasing the experimental temperature. Reduction in membrane cholesterol content by treatment with methyl-β-cyclodextrin also facilitated vesicle shedding. Inhibition of scramblase with R5421 impaired vesicle release. These data were interpreted in the context of membrane characteristics assessed previously by fluorescence spectroscopy with environment-sensitive probes such as laurdan, diphenylhexatriene, and merocyanine 540. The observations supported the following conclusions: 1) calcium-induced microvesicle shedding in erythrocytes relates more to membrane properties detected by diphenylhexatriene than by the other probes; 2) loss of trans-membrane phospholipid asymmetry is required for microvesicle release.
PACS Codes: 87.16.dj, 87.16.dt
KeywordsPhosphatidylserine Human Erythrocyte Ionomycin Acetoxymethyl Ester Vesicle Release
We have recently reported the use of fluorescence spectroscopy to generate a pseudo phase map of physical properties of erythrocyte membranes as a function of temperature and membrane cholesterol content . These experiments used three fluorescent probes that differ slightly in the properties they detect based on companion experiments with artificial membranes of defined composition. The result was the equivalent of a phase diagram mapping properties such as lipid order, spacing, and fluidity for intact human erythrocytes. These studies were complemented by a preliminary effort using the enzyme phospholipase A2 to determine whether there was any functional significance to the various regions of membrane behaviors observed by the probes . The success of that initial attempt has prompted us to pursue additional erythrocyte membrane processes to assess their relationship to membrane properties. We report here our investigations of the relationship between microvesicle release and the pseudo phase map.
Microvesicle shedding is stimulated in human erythrocytes by sustained elevation of intracellular calcium . Other events associated with calcium influx in erythrocytes include: opening of calcium-dependent potassium channels, leading to potassium ion efflux [4, 5]; activation of calcium-dependent proteases, including calpain ; activation of the phospholipid transporter scramblase, leading to exposure of phosphatidylserine in the outer leaflet of the bilayer [6, 7]; cytoskeletal alterations [8, 9, 10, 11, 12], a change in erythrocyte morphology from the normal discoid shape to a spherical shape; and an increase in order of membrane lipids as detected by the fluorescent probe laurdan [3, 13].
Apparently, several of these calcium-dependent events are involved in the mechanism of microvesicle production and shedding. For example, convincing evidence argues that activation of both calpain and the calcium-dependent potassium channel is critical for vesicle release [3, 4, 14, 15]. Cytoskeletal and membrane proteins such as synexin, a protein involved in bilayer fusion, are also likely to be involved . Studies in platelets have suggested that tyrosine dephosphorylation may regulate proteins involved in the process [17, 18]. Although several lines of evidence suggest that translocation of phosphatidylserine is also required, other data have implied the opposite conclusion [15, 19, 20, 21]. To our knowledge, a potential role for membrane biophysics in microvesicle release has not yet been explored. Nevertheless, the distinct characteristics of microvesicle membranes, such as enrichment with lipid rafts, support this hypothesis [16, 22].
To investigate the contributions of membrane composition and biophysical properties to calcium-regulated microvesicle release, we altered membrane cholesterol content, temperature, and the activity of scramblase in human erythrocytes treated with ionomycin, a calcium ionophore. We compared the results quantitatively to prior measurements of membrane properties in pseudo phase maps constructed as a function of temperature and cholesterol content . Phosphatidylserine exposure was manipulated using the pharmacological agent R5421, a specific inhibitor of scramblase .
Materials and methods
The intracellular calcium indicator indo-1 (acetoxymethyl ester form), laurdan, and the Amplex® Red cholesterol assay kit were purchased from Invitrogen (Carlsbad, CA). Ionomycin was obtained from Calbiochem (La Jolla, CA) and methyl-β-cyclodextrin (MBCD) and quinine were purchased from Sigma-Aldrich (St. Louis, MO). R5421 was a generous gift from Jeffrey T. Billheimer at Dupont Merck Research Laboratory (Wilmington, DE). All other reagents were obtained from standard sources. Ionomycin, R5421, laurdan, and indo-1 were dissolved in dimethylsulfoxide (DMSO). Quinine was dissolved in ethanol. Stocks of all other reagents were prepared as aqueous solutions.
Erythrocytes were obtained from blood samples obtained during physical exams at the Brigham Young University Student Health Center. The samples were fresh or stored up to 2 days at 4°C in EDTA vacutainers from which patient identification was removed. Control experiments comparing fresh blood with stored samples demonstrated that these storage conditions did not influence the results .
Erythrocytes were isolated and washed by centrifugation as described previously , resuspended to the original hematocrit (hct) in a balanced salt buffer (NaCl = 134 mM, KCl = 6.2 mM, CaCl2 = 1.6 mM, MgCl2 = 1.2 mM, Hepes = 18.0 mM, glucose = 13.6 mM, pH 7.4, 37°C). To verify that results were due to transport of calcium (and not magnesium) by ionomycin, experiments were repeated in a variant of this medium in which the typical calcium and magnesium salts were replaced with different concentrations of CaCl2. The minimum amount of CaCl2 sufficient to produce a full response was 0.6 mM. Under that condition, the results were indistinguishable regardless of which medium was used; hence, the data obtained with both media were pooled. The ionomycin concentration for all experiments was 300 nM.
Extraction of membrane cholesterol
Washed erythrocytes were suspended in buffer at 0.15% hct and incubated in the presence or absence of varying concentrations of MBCD (0, 0.1, 0.3, 0.6, or 1.0 mM) for 30 min at 37°C. Cells were washed and re-suspended in fresh medium for experiments. The amount of cholesterol extracted from the cells was quantified using the Amplex® Red cholesterol assay kit (Invitrogen, Carlsbad, CA) on methanol/chloroform extracts of samples as described .
Fluorescence spectroscopy and light scattering
Steady-state fluorescence (indo-1) and sample light scatter intensity and noise were monitored using photon-counting spectrofluorometers (Fluoromax-1 and Fluoromax-3 from Jobin Yvon, Edison, NJ, and PC1 from ISS, Champaign, IL). All three instruments maintain sample homogeneity by continuous gentle stirring with a magnetic stir bar. Control experiments (not shown) showed that stir bar speed did not influence light scatter noise. Sample temperature was controlled and maintained using circulating water baths and equilibrated for at least 5 min to achieve temperature stability. Simultaneous assessment of fluorescence intensity at multiple excitation and emission wavelengths was obtained by rapid sluing of monochromator mirrors using control software provided with the instrument. Band pass was set at 4 nm for both monochromators in each case.
Intracellular calcium concentrations at 25°C, 37°C and 45°C were assessed using indo-1 as explained previously . Erythrocyte samples were prepared as described above and further diluted to 1% hct. The acetoxymethyl ester of indo-1 (final concentration = 2.5 μM) was added, and the samples were incubated 60 min at 37°C in a shaking water bath. Cells were washed, resuspended in fresh buffer, placed in cuvette (0.075% hct) and incubated at the indicated temperature in the fluorometer sample chamber. Fluorescence was continuously monitored (excitation = 350 nm, emission = 405 and 480 nm). A baseline was established for 3 min, followed by the addition of ionomycin, and then continued data acquisition for several min.
Phosphatidylserine exposure on the outer leaflet of the cell membrane was assayed by measuring the fluorescence intensity of dansylarginine-N-(3-ethyl-1,5-pentanediyl)amide (3 μM final, excitation = 335 nm, emission = 545 nm) as described [3, 25]. This method detects phosphatidylserine by monitoring the rate at which it converts prothrombin to thrombin. A positive control was obtained by the addition of thrombin (2.7 μM).
For light scatter experiments, cell suspensions (0.075% hct) were incubated at the indicated temperature as for fluorescence. Monochromator settings were 500 nm for excitation and 510 nm for emission. As with indo-1 fluorescence measurements, light scatter data were gathered before and continued after ionomycin addition.
This displacement was standardized to a time frame of 240 s (since vesicle release appears continuous) and normalized to I 0 to account for variations in sample hematocrit and instrument sensitivity. I min and I 240 are defined as shown in Figure 1C; I min is the minimum intensity calculated from the regression to Eq. 1 and I 240 is the intensity 240 s after the time corresponding to I min .
The visual observations in Figures 5A–C were confirmed by correlation analysis (Figures 5D–F). Diphenylhexatriene anisotropy was the fluorescence parameter showing the highest correlation with and least systematic deviation from the level of microvesicle release (Figure 5F; r2 = 0.63 for merocyanine 540, 0.56 for laurdan, and 0.75 for diphenylhexatriene). Multiple regression yielded similar results. When all three pseudo phase maps were considered simultaneously, only laurdan and diphenylhexatriene fluorescence were significant predictors of the rate of microvesicle shedding (p < 0.008 in each case). Nevertheless, diphenylhexatriene explained a greater proportion of the variation in vesicle release rates than laurdan (41% versus 19%). These results suggest that membrane fluidity is the property upon which the ability of ionomycin to induce microvesicle release depends most.
Our purpose was to characterize microvesicle release from ionophore-treated erythrocytes in terms of microscopic membrane physical properties such as lipid order, fluidity, and composition. Previous studies had shown that these properties could be manipulated in human erythrocytes by altering temperature , membrane cholesterol content [1, 29], and by using the scramblase inhibitor R5421 . Our observations support two conclusions. First, calcium-stimulated microvesicle shedding depends on the level of membrane fluidity. This conclusion is based on the strong two-dimensional correlation between microvesicle release and diphenylhexatriene fluorescence (Figure 5). The temperature dependence of microvesicle release observed here (Figure 2) also corresponds well with other reports of erythrocyte physical behavior in the range of 30 to 40°C [1, 13, 29, 35, 36, 37]. The second conclusion is that the loss of membrane phospholipid asymmetry that accompanies calcium loading is necessary for microvesicle release.
As always, one must exercise caution in interpreting data with MBCD. Although the compound is used to manipulate membrane cholesterol content, it is also capable of extracting other lipids . For this reason, we use relatively mild extraction conditions (30 min incubation, ≤ 1 mM MBCD) [1, 2]. Nevertheless, we cannot exclude the possibility that some of the observations shown in Figures 4 and 5 involve lipids other than cholesterol. For this reason, our interpretations are based on relationships between the observed physical properties and membrane behavior rather than trying to draw direct conclusions about cholesterol per se. We were able to eliminate the possibility of non-specific direct artifacts of MBCD by including a control experiment in which we pre-loaded the MBCD molecules with cholesterol (see Ref.  for experimental details). Treatment of cells with pre-loaded MBCD had a significantly smaller effect on microvesicle shedding compared to treatment with virgin MBCD (not shown).
An alternative explanation for the temperature effect in Figure 2 is that it reflects direct thermal sensitivity of protein behavior independent of lipid properties. This possibility has been considered previously for the enzymes such as scramblase that mediate phosphatidylserine flip-flop . In that case, no temperature sensitivity was identified that could account for the data of Figure 2. Moreover, direct effects of temperature on protein activity or conformation alone cannot explain the effects of MBCD shown in Figures 4 and 5. Finally, the data of Figure 3 exclude the possibility that the temperature dependence was due to variation in the ability of ionomycin to transport calcium.
How would the fluidity of the lipid bilayer affect the ability of the membrane to release vesicles? This question is further complicated by an apparent paradox. On the one hand, microvesicle release appears dependent on a membrane that is fluid and/or disordered at the time calcium is introduced to the cell interior (Figures 4 and 5). On the other hand, influx of calcium causes the membrane to become more ordered as the vesicles are shed . Some evidence suggests that the quandary may be resolved by considering the spatial distribution of physical properties along the membrane surface. A study by Salzer et al. concluded that membrane liquid ordered domains (termed "lipid rafts" in the paper) are involved in the vesiculation process because the microvesicles are enriched in lipids and proteins typical of these domains . Two-photon microscopy of laurdan-labeled erythrocytes suggests that the increased membrane order associated with calcium ionophore treatment is distributed unevenly across the membrane, perhaps representing an expansion of liquid-ordered domains [1, 3, 13, 22]. Therefore, as temperature is increased or cholesterol is removed, the ordered domains propagated by ionophore treatment would be superimposed on a background of greater fluidity thus creating greater contrasts between these domains and the surrounding lipids (as observed in Ref. ). Perhaps, then, enhanced microvesicle shedding at high temperature or low cholesterol is explained by a more extreme diversity in lateral bilayer structure. Whether the enhanced membrane fluidity promotes microvesicle release directly or indirectly by altering the activity of membrane-bound enzymes cannot be distinguished by the experiments of this study.
The role of phosphatidylserine exposure in microvesicle release has been controversial. It is well known that the two events are coincident when erythrocytes are loaded with calcium [3, 19, 39]. Similar coincidence is also observed when erythrocytes are stimulated with lysophosphatidic acid . Furthermore, blood cells from patients suffering from a bleeding disorder known as Scott syndrome neither translocate phosphatidylserine across the cell membrane nor release microvesicles in response to calcium ionophore .
Naturally, these observations of coincidence do not designate cause and effect. It is in the attempts to establish causal relationships that the controversy has emerged. On the pro side, microvesicle release in platelets appears to require translocation of phosphatidylserine . A similar claim has been made for erythrocytes . One study used dithioerythritol to suppress phospholipid scrambling and discovered that microvesicle release was likewise suppressed . In contrast, in erythrocyte ghosts loaded with spermine, calcium induced microvesicle release successfully even though phosphatidylserine translocation was severely reduced . Similarly, spermine-loaded ghosts from a Scott syndrome patient also released microvesicles in the complete absence of phosphatidylserine flip-flop .
Inhibition of phosphatidylserine translocation with spermine or dithioerythritol is unlikely to be specific. Hence, use of the specific scramblase inhibitor R5421 in the present study may simplify interpretations. It also allowed experiments to be done on intact erythrocytes rather than relying on ghosts, in which cytoskeletal and cytosolic components are altered or depleted. The data shown in Figure 6 seem to provide an unambiguous interpretation with respect to cause and effect. Therefore, phosphatidylserine translocation appeared to be required for microvesicle release.
In summary, two conclusions are supported by the results of this study. First, calcium-stimulated microvesicle release depends on the level of membrane fluidity. Second, loss of trans-membrane phospholipid asymmetry is required for microvesicle shedding.
Physiologically, the results presented in this study may relate to the process of membrane blebbing during apoptosis. It has been suggested that the microvesiculation of erythrocyte membranes in response to calcium loading is a model for the blebbing that occurs as part of the apoptotic process . If this concept is true, it is logical to ask whether apoptotic membrane blebbing requires increased membrane fluidity as described in this report for microvesicle release. The possibility has not yet been examined experimentally, but investigations demonstrating that apoptotic cell membranes are more fluid or disordered than those of healthy cells support the viability of the idea [42, 43, 44, 45, 46, 47]. These bilayer changes occur early during apoptosis and could be the result of phosphatidylserine exposure on the external membrane face [42, 48, 49, 50, 51, 52] or other apoptotic events . Based on these observations and the data presented in this report, we propose the hypothesis that increased fluidity of the cell membrane during apoptosis may be a necessary step in preparation for cell membrane blebbing.
This work was supported by the National Institutes of Health (GM073997).
- 9.Takakuwa Y: Int J Hematol. 2000, 72: 298-309.Google Scholar
- 10.Sikorski AF, Hanus-Lorenz B, Jezierski A, Dluzewski AR: Acta Biochim Pol. 2000, 47: 565-578.Google Scholar
- 15.Zwaal RF, Schroit AJ: Blood. 1997, 89: 1121-1132.Google Scholar
- 19.Bevers EM, Wiedmer T, Comfurius P, Shattil SJ, Weiss HJ, Zwaal RF, Sims PJ: Blood. 1992, 79: 380-388.Google Scholar
- 23.Dekkers DW, Comfurius P, Vuist WM, Billheimer JT, Dicker I, Weiss HJ, Zwaal RF, Bevers EM: Blood. 1998, 91: 2133-2138.Google Scholar
- 25.Krishnaswamy S, Church WR, Nesheim ME, Mann KG: J Biol Chem. 1987, 262: 3291-3299.Google Scholar
- 26.Hoffman JF: Blood Cells. 1987, 12: 565-588.Google Scholar
- 33.Dale RE, Chen LA, Brand L: J Biol Chem. 1977, 252: 7500-7510.Google Scholar
- 41.Chukhlovin AB: Scanning Microsc. 1996, 10: 795-803.Google Scholar
- 48.Fadok VA, Voelker DR, Campbell PA, Cohen JJ, Bratton DL, Henson PM: J Immunol. 1992, 148: 2207-2216.Google Scholar
- 49.Mower DA, Peckham DW, Illera VA, Fishbaugh JK, Stunz LL, Ashman RF: J Immunol. 1994, 152: 4832-4842.Google Scholar
- 51.Hess KL, Johnson JD, Cook-Mills JM: J Leukoc Biol. 2001, 70: 405-412.Google Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.