Involvement of cytoskeletal proteins in the barrier function of the human erythrocyte membrane. II. Formation of membrane leaks in ghost membranes after limited proteolysis of skeletal proteins by trypsin.

Limited proteolysis of human erythrocyte ghost membranes by low levels of trypsin (10-240 ng/ml) added bilaterally at 0 degrees C together with the proteinase inhibitor, phenylmethylsulfonyl fluoride (PMSF) before resealing at 37 degrees C leads to a graded digestion of spectrin and ankyrin and the disappearance of band 4.1 protein, while band 3 is cleaved only to a very low extent. These alterations are accompanied by an increase of membrane permeability of the resealed ghosts to hydrophilic nonelectrolytes (erythritol to sucrose), taken to reflect impaired resealing. Moreover, the membrane begins to vesiculate. Shedding of vesicles during the efflux measurements can not be responsible for the increased release of test solutes, since the ghosts do not loose hemoglobin and discriminate the nonelectrolytes according to their size. Moreover, the vesiculation site itself does not seem to act as the leak site, since ghosts prepared from erythrocytes pretreated with a carbodiimide which induces membrane rigidification still exhibit a pronounced protein degradation and vesiculation while the permeability enhancement induced by trypsination is markedly suppressed. The trypsin-induced leak has the properties of an aqueous pore as indicated, besides size selectivity, by its inhibition by phloretin and the very low activation energy. In analogy with concepts developed in the preceding paper (Klonk, S. and Deuticke, B. (1992) Biochim. Biophys. Acta 1106, 126-136 (Part I in this series)) the impaired resealing after limited proteolysis is assumed to be related to a perturbation of interactions of membrane skeletal elements with themselves and/or with the bilayer domain constituting the permeability barrier.     

Involvement of cytoskeletal proteins in the barrier function of the human erythrocyte membrane. I. Impairment of resealing and formation of aqueous pores in the ghost membrane after modification of SH groups.

Resealed human erythrocyte ghosts prepared by a two-step procedure were shown to have small residual barrier defects with the properties of aqueous pores, such as size discrimination of hydrophilic nonelectrolytes (erythritol to sucrose), indicative of an apparent pore radius of about 0.7 nm, and a low activation energy (about 12-20 kJ/mol (mannitol, sucrose)) of the leak fluxes. As in other cases (Deuticke et al. (1991) Biochim. Biophys. Acta 1067, 111-122) these leak fluxes can be inhibited by phloretin. Treatment of such resealed ghosts with the mild SH oxidizing agent, diamide, induces additional membrane leaks to the same extent and with the same properties as in native erythrocytes (Deuticke et al. (1983) Biochim. Biophys. Acta 731, 196-210), including reversibility of the leak by SH reducing agents, inhibition by phloretin and stimulation by alkanols. In contrast, resealed ghosts prepared either from diamide-treated erythrocytes or by adding diamide to the 'open' membranes prior to reconstitution of high ionic strength and raising the temperature, exhibit a state of greater leakiness. This leakiness is somewhat different in its origin from the former class of leaks, since it can also be produced by N-ethylmaleimide, which is essentially ineffective when added to the membrane in its 'tight' state. The leaks induced in the 'open' state of the membrane, which can be regarded as a consequence of an impaired resealing, are nevertheless reversible by reducing agents added after resealing and are comparable in many, but not all their characteristics to leaks induced in the 'tight' state of the membrane. Resealing in the presence of the isothiocyanostilbenes DIDS or SITS mimicks the leak forming effect of diamide by modifying a small population of SH groups, while amino groups seem not to be involved. The findings indicate and substantiate an important role of the redox state of membrane skeletal protein sulfhydryls in the maintenance and the re-establishment of the barrier function of the erythrocyte membrane.     

Cross-linking of SH-groups in the erythrocyte membrane enhances transbilayer reorientation of phospholipids. Evidence for a limited access of phospholipids to the reorientation sites

Oxidation of erythrocyte membrane SH-groups and concomitant cross-linking of spectrin, which induce a partial loss of phospholipid asymmetry (Haest, C.W.M., Plasa, G., Kamp, D. and Deuticke, B. (1978) Biochim. Biophys. Acta 509, 21-32) are now shown to result in a remarkable increase of the rates of transbilayer reorientation of exogenously incorporated lysophospholipids. Reorientation of both, neutral lysophosphatidylcholine and of negatively charged lysophosphatidylserine is enhanced. A decrease of the activation energy of the reorientation process as well as quantitative changes of the dependence of reorientation on the lysophosphatidylcholine and cholesterol content of the membrane indicate formation of new reorientation sites or modification of existing sites. A common mechanism may underly the formation of reorientation sites and the occurrence of leaks for small solutes (Deuticke, B., Poser, B., Lütkemeier, P. and Haest, C.W.M. (1983) Biochim. Biophys. Acta 731, 196-210) subsequent to oxidation of membrane SH-groups. Whereas exogenous lysophospholipids completely equilibrate between the two lipid layers regardless of the extent of oxidation of SH-groups, endogenous inner layer phospholipids become available for reorientation in a graded way. Native phospholipid asymmetry is therefore not the result of a low transbilayer mobility of phospholipids, but probably due to a lack of access of inner layer phospholipids to the reorientation sites.     

Oxidative stress of human erythrocytes by iodate and periodate. Reversible formation of aqueous membrane pores due to SH-group oxidation

Human erythrocytes were exposed to oxidative stress by iodate and periodate. Oxidation causes a time- and concentration-dependent increase in membrane permeability for hydrophilic molecules and ions. The induced leak discriminates nonelectrolytes on the basis of molecular size and exhibits a very low activation energy (Ea = 1-4 kcal.mol-1). These results are reconcilable with the formation of aqueous pores. The pore size was approximated to be between 0.45 and 0.6 nm. This increase in permeability is reversible upon treatment with dithioerythritol. Blocking of membrane thiol groups with N-ethylmaleimide protects the membranes against leak formation. The oxidation causes dithioerythritol-reversible modification of membrane proteins as indicated by the gel electrophoretic behavior. These modifications can also be suppressed by blocking the membrane thiol groups with N-ethylmaleimide. About half of the membrane methionine is oxidized to acid hydrolysis-stable derivatives. A fast saturating increase in diene conjugation was observed in whole cells but not in isolated membranes, with only minor degradation of fatty acid chains. The oxidation of cell membrane lipids as well as oxidation of cell surface carbohydrates are not involved in leak formation. Taken together with earlier data (Deuticke, B., Poser, B., Lütkemeier, P. and Haest, C.W.M. (1983) Biochim. Biophys. Acta 731, 196-210), these findings indicate that formation of disulfide bonds by different oxidative mechanisms results in leaks with similar properties.     

Changes of nonelectrolyte permeability in cholesterol-loaded erythrocytes.

Membrane cholesterol in porcine and bovine erythrocytes was elevated up to 165% of its normal value by incubation of the cells in cholesterol/phosphatidylcholine dispersions with or without serum. This alteration of membrane lipid composition brought about only a minor (10-40%) decrease of the permeability to glycerol, erythritol and to organic acids penetrating by non-ionic diffusion, although additional cholesterol had actually been incorporated into the lipid bilayer, as indicated by determinations of cell surface area from the critical hemolytic volume, in combination with quantitative evaluation of freeze-etch electron micrographs. On the basis of this finding and of the previously demonstrated (Grunze, M. and Deuticke, B. (1974) Biochim. Biophys. Acta 356, 125-130) considerable increase of permeability in cholesterol-depleted cells, it is proposed that in the erythrocyte membrane a pronounced "specific" reduction of permeability by cholesterol occurs only up to a molar ratio cholesterol/polar lipid of 0.6. At higher ratios cholesterol affects permeability only slightly, owing to an "unspecific" rigidifying effect on the membrane lipid phase.     

Formation of aqueous pores in the human erythrocyte membrane after oxidative cross-linking of spectrin by diamide

Oxidation of erythrocyte membrane SH-groups by diamide and tetrathionate induces cross-linking of spectrin (Haest, C.W.M., Kamp, D., Plasa, G. and Deuticke, B. (1977) Biochim. Biophys. Acta 469, 226–230). This cross-linking was now shown to go along with a concentration- and time-dependent enhancement of membrane permeability for hydrophilic nonelectrolytes and ions. The enhancement is specific for oxidative SH-group modifications, is reversible by reduction of the induced disulfides, can be suppressed by a very brief pre-treatment of the cells with low concentrations of $\text{N-}\text{ethylmaleimide}$ and is strongly temperature-dependent. The pathway of the induced permeability discriminates nonelectrolytes on the basis of molecular size and exhibits a very low activation energy ($\text{E}{}_{\mathrm{<mtext>a</mtext>}}\text{3–8}\text{kcal/mol}$). These findings are reconcilable with the formation of a somewhat inhomogeneous population of aqueous pores with radii probably $\text{? 0.65}\text{nm}$. Estimated pore numbers vary with the size of the probe molecule. Assuming a diffusion coefficient as in bulk water within the pore, at least 20 pores per cell have to be postulated; more realistic lower diffusion coefficients increase that number. Alterations of the lipid domain by changes of cholesterol contents and insertion of hexanol or nonionic detergents alter the number or size of the pores. Since aggregation of skeletal and intrinsic membrane proteins also occurs after the SH-oxidation, in parallel to the formation of membrane leaks, one may consider (a) defects in the disturbed bilayer interface, (b) a mismatch between lipid and intrinsic proteins or (c) channels inbetween aggregated intrinsic proteins as structures forming the pores induced by diamide treatment.     

Ion selectivity of aqueous leaks induced in the erythrocyte membrane by crosslinking of membrane proteins

The aqueous leak induced in the human erythrocyte membrane by crosslinking of spectrin via disulfide bridges formed in the presence of diamide (Deuticke, B., Poser, B., Lütkemeier, P. and Haest, C.W.M. (1983) Biochim. Biophys. Acta 731, 196-210) was further characterized with respect to its ion selectivity by means of (a) measurements of cell volume changes or hemolysis, (b) determination of membrane potentials and (c) analysis of potential-driven ion fluxes. The leak turned out to be slightly cation-selective (PK:PCl approximately equal to 4:1). It discriminates mono- from divalent ions (PNa:PMg greater than 100:1, PCl:PSO4 greater than 10:1) and to a much lesser extent monovalent ions among each other. The selectivities for monovalent ions follow the sequence of free solution mobilities, increasing in the order Li+ less than or equal to Na+ less than K+ less than or equal to Rb+ less than Cs+ and F- less than Cl- less than Br- less than I-. Polyatomic anions also fit into that order. Quantitatively, the ratios of permeabilities of the leak are larger than those of the ion mobilities in free solution. The ion permeability of the leak is concentration-independent up to at least 150 mM. The ion milieu, however, has marked effects on leak permeability, most pronounced for chaotropic ions (guanidinium, nitrate, thiocyanate), which increase leak fluxes of charged and uncharged solutes. The results support the view that, besides geometric constraints, weak coulombic or dipolar interactions between penetrating ions and structural elements of the leak determine permselectivity.     

Effect of L-carnitine and acetyl-L-carnitine on the human erythrocyte membrane stability and deformability

In this study we examined the effect of carnitine and acetylcarnitine on the human erythrocyte membrane stability and membrane deformability. Since erythrocyte membranes are impermeable to these compounds, we resealed erythrocyte ghosts in the presence of different concentrations of carnitine or acetylcarnitine. Resealed ghosts can be adequately studied in their cellular deformability and membrane stability properties by means of ektacytometry. Both carnitine and acetylcarnitine alter the membrane stability but not membrane deformability of the red cell membrane. Resealed ghosts containing 20, 50, 150, and 300 microM carnitine had 1.1, 1.6, 0.9, and 0.7 times the normal stability. While resealed ghosts containing 20, 50, 150, and 300 microM acetylcarnitine had 1.1, 1.5, 1.3, and 1.2 times the normal stability. Such changes were found to be reversible. We also conducted SDS PAGE of cytoskeletal membrane proteins from membrane fragments and residual membranes produced during membrane stability analysis, and unsheared resealed membranes in those samples where we observed an increase or a decrease of membrane stability. No changes in the cytoskeletal membrane proteins were noticed, even when the samples, prior SDS PAGE analysis, were treated with or without dithiothreitol. In addition, fluorescence steady state anisotropy of DPH in the erythrocyte membrane treated with carnitine or acetylcarnitine shows no modification of the lipid order parameter. Our results would suggest that both carnitine and its acetyl-ester, at physiological concentrations, may increase membrane stability in mature erythrocytes, most likely via a specific interaction with one or more cytoskeletal proteins, and that this effect would manifest when the erythrocytes are subjected to high shear stress.     

Changes of nonelectrolyte permeability in cholesterol-loaded erythrocytes

Membrane cholesterol in porcine and bovine erythrocytes was elevated up to 165% of its normal value by incubation of the cells in cholesterol/phosphatidylcholine dispersions with or without serum. This alteration of membrane lipid composition brought about only a minor (10–40%) decrease of the permeability to glycerol, erythritol and to organic acids penetrating by non-ionic diffusion, although additional cholesterol had actually been incorporated into the lipid bilayer, as indicated by determinations of cell surface area from the critical hemolytic volume, in combination with quantitative evaluation of freeze-etch electron micrographs.On the basis of this finding and of the previously demonstrated (Grunze, M. and Deuticke, B. (1974) Biochim. Biophys. Acta 356, 125–130) considerable increase of permeability in cholesterol-depleted cells, it is proposed that in the erythrocyte membrane a pronounced “specific” reduction of permeability by cholesterol occurs only up to a molar ratio cholesterol/polar lipid of 0.6. At higher ratios cholesterol affects permeability only slightly, owing to an “unspecific” rigidifying effect on the membrane lipid phase.     

Spectrin as a stabilizer of the phospholipid asymmetry in the human erythrocyte membrane

After treatment of intact human erythrocytes with SH-oxidizing agents (e.g. tetrathionate and diamide) phospholipase A₂ cleaves approx. 30% of the phosphatidylserine and 50% of the phosphatidylethanolamine without causing hemolysis (Haest, C.W.M. and Deuticke, B. (1976) Biochim. Biophys. Acta 436, 353–365). These phospholipids are scarcely hydrolysed in fresh erythrocytes and are assumed to be located in the inner lipid layer of the membrane (Verkleij, A.J., Zwaal, R.F.A., Roelofsen, B., Comfurius, P., Kastelijn, D. and van Deenen, L.L.M. (1973) Biochim. Biophys. Acta 323, 178–193). The enhancement of the phospholipid cleavage is now shown to be accompanied by a 50% decrease of the membrane SH-groups and a cross-linking of spectrin, located at the inner surface of the membrane, to oligomers of < 10⁶ dalton.Blocking approx. 10% of the membrane SH groups with N-ethylmaleimide suppresses both the polymerization of spectrin and the enhancement of the phospholipid cleavage. N-Ethylmaleimide, under these conditions, reacts with three SH groups per molecule of spectrin, 0.7 SH groups per major intrinsic 100 000 dalton protein (band 3) and 1.1 SH groups per molecule of an extrinsic protein of 72 000 daltons (band 4.2). Blocking studies with iodoacetamide demonstrate that the SH groups of the 100 000-dalton protein are not involved in the effects of the SH-oxidizing agents.It is suggested that a release of constraints imposed by spectrin enables phosphatidylserine and phosphatidylethanolamine to move from the inner to the outer lipid layer of the erythrocyte membrane and that spectrin, in the native erythrocyte, stabilizes the orientation of these phospholipids to the inner surface of the membrane.     

ATP does not regulate the reconstituted glucose transporter

ATP has been reported to affect glucose transport in human erythrocytes and resealed erythrocyte ghosts [Jacquez, J. A. (1983) Biochim. Biophys. Acta 727, 367-378; Jensen, M. R., & Brahm, J. (1987) Biochim. Biophys. Acta 900, 282-290]. In more detailed studies, effects of micromolar levels of ATP on transport in ghosts and inside-out vesicles, and on the fluorescence of ghosts and the purified glucose transporter [Carruthers, A. (1986) Biochemistry 25, 3592-3602; Hebert, D. N., & Carruthers, A. (1986) J. Biol. Chem. 261, 10093-10099; Carruthers, A. (1986) J. Biol. Chem. 261, 11028-11037], have been interpreted as supporting a model in which ATP regulates the catalytic properties of the transporter. Both allosteric and covalent effects of ATP were proposed; among the allosteric effects was a 60% reduction in the Km for zero-trans uptake. In order to test whether allosteric ATP regulation of the transporter occurs, we reconstituted glucose transport activity into liposomes using erythrocyte membranes without detergent treatment. The effects of ATP, present either outside, inside, or both inside and outside the liposomes, on the transport activity were examined. Effects of ATP on trypsin-treated liposomes, which have only a single orientation of active transporters, were also tested. While the model predicts activation by ATP, only inhibition was observed. This was significant only at millimolar concentrations of ATP, in contrast to the previously reported effects at micromolar levels, and was primarily on the extracellular surface of the transporter. In addition, the ATP effects on reconstituted transport were nonspecific, with similar effects produced by tripolyphosphate.(ABSTRACT TRUNCATED AT 250 WORDS)     

Calcium effects on human erythrocyte membrane proteins

The effects of Ca²⁺ on human erythrocyte membrane proteins were examined by sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis. Ca²⁺ had several effects on normal human erythrocyte membrane proteins. It affected the binding of cytoplasmic proteins to the membrane, produced a non-reversible aggregation of several membrane proteins and activated apparent proteolysis of membrane proteins. The Ca²⁺ effect could be obtained with isolated, washed membranes when the erythrocyte cytoplasm was added. These studies indicate that the Ca²⁺-induced membrane proteolysis and aggregation effects are not due simply to its presence at the time of hemolysis as previously suggested (Carraway, K.L., Triplett, R.B. and Anderson, D.R. (1975) Biochim. Biophys. Acta 379, 571–581), but are the result of more complex interactions between the erythrocyte membrane and cytoplasmic factors.     

Errata

Human erythrocytes were exposed to oxidative stress by iodate and periodate. Oxidation causes a time- and concentration-dependent increase in membrane permeability for hydrophilic molecules and ions. The induced leak discriminates nonelectrolytes on the basis of molecular size and exhibits a very low activation energy (E_a = 1–4 kcal · mol^?1). These results are reconcilable with the formation of aqeous pores. The pore size was approximated to be between 0.45 and 0.6 nm. This increase in permeability is reversible upon treatment with dithioerythritol. Blocking of membrane thiol groups with N-ethylmaleimide protects the membranes against leak formation. The oxidation causes dithioerythritol-reversible modification of membrane proteins as indicated by the gel electrophoretic behavior. These modifications can also be suppressed by blocking the membrane thiol groups with N-ethylmaleimide. About half of the membrane methionine is oxidized to acid hydrolysis-stable derivatives. A fast saturating increase in diene conjugation was observed in whole cells but not in isolated membranes, with only minor degradation of fatty acid chains. The oxidation of cell membrane lipids as well as oxidation of cell surface carbohydrates are not involved in leak formation. Taken together with earlier data (Deuticke, B., Poster, B., Lütkemeier, P., and Haest, C.W.M. (1983) Biochim. Biophys. Acta 731, 196–210), these findings indicate that formation of disulfide bonds by different oxidative mechanisms results in leaks with similar properties.     

Mechanism of membrane damage induced by the amphipathic peptides gramicidin S and melittin

The action of gramicidin S and melittin on human erythrocytes, Staphylococcus aureus and Escherichia coli was studied as an extension of the previous study (Katsu, T., Ninomiya, C., Kuroko, M., Kobayashi, H., Hirota, T. and Fujita, Y. (1988) Biochim. Biophys. Acta 939, 57-63). These amphipathic peptides stimulated the release of membrane phospholipids outside cells in a concentration range causing permeability change. The shape change of erythrocytes from normal discoid to spiculate form was observed just prior to the release of membrane components. We have proposed the following action mechanism of gramicidin S and melittin. The peptide molecules were predominantly accumulated in the outer half of the bilayer, deforming the erythrocyte cell into crenature. A large accumulation made the membrane structure unstable, resulting in the release of membrane fragments and the simultaneous enhancement of permeability. The action mechanism of these peptides was compared with that of simple surfactants.     

Effects of gamma-irradiation on the erythrocyte membrane: ESR, NMR and biochemical studies

The effects of gamma-irradiation on resealed erythrocyte ghosts have been examined with different techniques. Phospholipid analysis reveals peroxidative damage on the polyunsaturated chains of phosphatidylethanolamine. Gel electrophoresis and ESR measurements indicate modifications of the cytoskeletal proteins. 31P Nuclear magnetic resonance data show bilayer modifications that can be interpreted as changes in lipid-protein interactions. The overall picture from the present results favours interaction between lipids and proteins in the inner monolayer of the membrane.     

Inhibition of the binding of cytochrome b5 to phosphatidylcholine vesicles by cholesterol

The binding of cytochrome b₅ to single-walled liposomes of egg phosphatidylcholine was inhibited by the presence of cholesterol in the lipid bilayer under conditions where a limited amount of liposomes was incubated with the cytochrome. Since similar conditions seem to apply for the binding of cytochrome b₅ to erythrocyte ghosts, this observation supports the conclusion of Enomoto and Sato (Enomoto, K. and Sato, R. (1977) Biochim. Biophys. Acta 466, 136–147) that the localization of cholesterol on the outer surface of the ghost membrane prevents the binding of cytochrome b₅ to this surface. The finding reported by Roseman et al. (Roseman, M.A., Holloway, P.W. and Calabro, M.A. (1978) Biochim. Biophys. Acta 507, 552–556) that cholesterol did not prevent the cytochrome binding to phosphatidylcholine liposomes in the presence of a large excess of liposomes could be confirmed in the present study, but this does not contradict the abovementioned conclusion.     

Human erythrocyte membrane acid proteinase (EMAP): sidedness and relation to cathepsin D

An acid proteinase purified from human erythrocyte membranes (Yamamoto, K. & Marchesi, V.T. (1984) Biochem. Biophys. Acta 790, 208-218), now termed "EMAP," was further characterized with respect to its localization and relation to cathepsin D. The membrane-associated form of EMAP was shown to be latent by demonstrating that no activity was detectable in both resealed (right-side-out) ghosts and inside-out vesicles in the absence of detergents. The enzyme associated with the inside-out vesicles was unstable when exposured to acidic pH between 4.0 and 4.5, whereas the enzyme associated with the resealed ghosts was stable in the wide pH range of 3.7 to 9.0. Tryptic digestion produced the loss of activity for the enzyme associated with the inside-out vesicles but not the resealed ghosts. The antibody to rat spleen cathepsin D, which cross-reacted weakly but detectably with EMAP, selectively bound to the inside-out vesicles. These results indicate the location of EMAP on th inner surface of the membranes. Comparison of a number of enzymatic properties of EMAP with rat cathepsin D showed significant differences between these two enzymes. EMAP was less stable in the pH range of 3.5 to 6.0 than cathepsin D. The enzymes were distinguished from each other by differences in their elution profiles on DEAE-Sephacel and chromatofocusing columns and by differences in the extent of inhibition by a few specific inhibitors. Both enzymes revealed significant differences in the amino acid composition and specific activity towards bovine hemoglobin. The immunological relationship between these two enzymes is discussed.     

A calcium-activated polyphosphoinositide phosphodiesterase in the plasma membrane of human and rabbit erythrocytes.

Haemoglobin-free human erythrocyte ghosts that were prepared in the presence of EDTA and were then exposed to Ca2+ showed a substantial loss of phosphatidylinositol phosphate and phosphatidylinositol diphosphate, measured either chemically or by loss of 32P from the lipids of prelabelled membranes. At the same time there was, as reported previously (Allan, D. and Michell, R.H., (1976) Biochim. Biophys. Acta 455, 824--830), and approximately equivalent rise in the diacylglycerol content of the membranes. Analysis of the 32P-labelled water-soluble material released during this process showed that the major products were inositol diphosphate and inositol triphosphate. No change was seen in the phosphatidylinositol or phosphatidate content of the membranes, and there was no Ca2+-activated loss of 32P from the phosphatidate of prelabelled membranes: this suggests that Ca2+ did not activate phosphoinositide phosphomonoesterases or phosphatidate phosphomonoesterase in human erythrocyte membranes. It is concluded that human erythrocyte membranes contain at their cytoplasmic surface a Ca2+-activated phosphodiesterase that is active against both phosphatidylinositol phosphate and phosphatidylinositol diphosphate. Rabbit erythrocytes also contained this enzyme, but in these cells there was also evidence for the presence of a Ca2+-activated phosphatidate phosphomonoesterase.     

Elastic properties of the erythrocyte membrane and the critical cell volume of erythrocytes

The results on elastic membrane area extension during hemolysis, reported by Richieri and Mel (Richieri, G.V. and Mel, H.C. (1985) Biochim. Biophys. Acta 813, 41-50), are discussed. Careful analysis of their data leads to the conclusion, that the differences in osmolarity, as found in the experiment, were insufficient to cause the reported values of elastic changes in erythrocyte volume (17-22%) and of membrane area extension (11-14%). The recalculated values of the elastic extensions of membrane area are not different from those measured by the micropipet method (i.e. 3-4%).     

Reflection coefficients of permeant nonelectrolytes for dog and beef red cell membranes.

The reflection coefficient, sigma, for several small permeant nonelectrolytes was determined for dog and beef red blood cell membranes. Our sigma values were considerably higher than those previously reported for dog cells; e.g., out sigma urea was 87% higher than the sigma urea of Rich, Sha'afi, Barton and Solomon (J. Gen. Physiol. 50: 2391, 1967). Our sigma values for urea were only slightly greater in beef cells than previously reported by Farmer and Macey (Biochim. Biophys. Acta 290: 290, 1972). We found that a trend exists when (1 - sigma) is plotted against the log of the permeability coefficient, omega. This observation is also consistent with our previously reported sigma data for human red cell membranes (Owen & Eyring, J. Gen. Physiol. 66: 241, 1972). This trend suggests that small hydrophilic molecules interact highly with cell membrane water. The exceptions to this trend were lipophilic molecules, indicating they do not interact with water while penetrating the red cell membrane.