Individual leaflets of a membrane bilayer can independently regulate permeability.

Water rapidly crosses most membranes, but only slowly crosses apical membranes of barrier epithelia such as bladder and kidney collecting duct, a feature essential to barrier function. How apical membrane structure reduces permeabilities remains unclear. Cell plasma membranes contain two leaflets of distinct lipid composition; the role of this bilayer asymmetry in membrane permeability is unclear. To determine how asymmetry of leaflet composition affects membrane permeability, effects on bilayer permeation of reducing single leaflet permeability were determined using two approaches: formation of asymmetric bilayers in an Ussing chamber, with only one of two leaflets containing cholesterol sulfate, and stabilization of the external leaflet of unilamellar vesicles with praeseodymium (Pr). In both systems, permeability measurements showed that each leaflet acts as an independent resistor of water permeation. These results show that a single bilayer leaflet can act as the barrier to permeation and provide direct evidence that segregation of lipids to create a low permeability exofacial leaflet may act to reduce the permeability of barrier epithelial apical membranes.

Epithelial cells generate and maintain apical membrane bilayers made up of leaflets of distinct composition by mechanisms involving asymmetric biosynthesis in the Golgi, oriented insertion into the plasma membrane, and the activity of ATPdriven phospholipid flippases (1)(2)(3)(4)(5)(6). In several epithelia, such as those of the stomach, kidney collecting duct and thick ascending limb, and mammalian bladder, these apical membranes exhibit exceptionally low permeabilities to water, small nonelectrolytes, protons, and ammonia. These low permeabilities are critical to the barrier function of these epithelia (7)(8)(9)(10)(11). The structural features responsible for the low permeabilities of these apical membranes remain poorly defined as is the physiological role of the bilayer asymmetry observed in these membranes. The present studies examine the role of bilayer asymmetry in governing the permeability of membranes to water by measuring permeabilities across artificial symmetric and asymmetric membranes. The results show that a single leaflet of a membrane bilayer can act as a barrier to water flux and provide strong evidence that each leaflet acts as an independent resistor to permeation.

Symmetric and Asymmetric Planar Bilayer Preparations and Flux
Measurements--All lipids were obtained from Avanti Polar Lipids (Alabaster, AL). Bilayers were formed by the folded method as described (14) in an Ussing chamber with an orifice for the bilayer of 150 m diameter cut into 10-m thick sheets of Saran Wrap. Pure DPhPC 1 alone or mixed with 25% w/w cholesterol sulfate (CS) were dissolved at 1% w/v in hexane and layered onto the top of aqueous buffer. Bilayers were formed in the orifice of the chamber by sequentially raising the level of each aqueous reservoir (12)(13)(14). Membrane formation was monitored using a dissecting microscope and by a marked increase in capacitance as described (14). Once the membrane was formed and achieved electrical stability, 30 Ci of 3 H 2 O or 2.5 l of [ 14 C] butanol was added to the rear hemichamber, and both hemichambers were sealed with Parafilm to prevent distillation of tracer into the front hemichamber. Samples were taken from each hemichamber over time, and radioactivity was detected by liquid scintillation counting. Fluxes were linear over the time course of the experiment (1.5-4 h). Permeabilities were calculated using the flux equation: P d ϭ ⌽/(A)(⌬C), where P d is the diffusive water permeability, ⌽ is the flux of the tracer across the membrane (calculated from the slope of the concentration of tracer appearing as a function of time in the front hemichamber, in cpm/s), A is the surface area of the orifice (1.77 ϫ 10 Ϫ4 cm 2 ), and ⌬C is the concentration gradient for isotope across the membrane (in cpm/cm 3 ). Over the course of the experiment, the aggregate flux of 3 H 2 O was small enough so that ⌬C remained essentially constant throughout the experiment. Each of the three measured fluxes was statistically different from the other values by analysis of variance using the Neuman-Keuls test. All values reported represent mean Ϯ S.D.
Symmetric and Asymmetric Liposome Bilayers and Flux Measurements-DPPC was dissolved in buffer (150 mM KCl and 20 mM HEPES, pH 7.4, 300 mosm/kg) and LUV were formed by sonication in buffer in the absence or presence of 20 mM CF (Molecular Probes) followed by overnight incubation at 4°C to permit equilibration of diameter to approximately 100 nm. LUV measured 105 Ϯ 5 nm in diameter by quasielastic light scattering (11,15); a similar value is obtained in the absence of CF or with CF entrapped at lower concentrations. Prior to flux measurements, extravesicular CF was removed by passing the liposomes through Sephadex G-50 columns, where the liposomes emerged with the void volume and the extravesicular CF remained entrapped in the column (11,15). The minimal amount of extravesicular CF remaining was quenched with anti-fluorescein antibody (11,15). P f was measured by exposing the liposomes to an osmotic gradient by rapid mixing with an equal volume of buffer plus sufficient sucrose to increase osmolality in the mixed solutions to a value twice that in the original solution. P f measurements were performed using a stoppedflow fluorimeter (Applied Photophysics model SF.17mv) with incident light set at 490 Ϯ 1 nm and emission light Ͼ510 nm using a cut-on filter (11,(15)(16)(17). As water effluxed from the vesicles, they shrank, increasing * This research was supported by the National Institutes of Health and development funds from the University of Pittsburgh Medical Center as well as facilities support from the National Science Foundation Science and Technology Centers Program. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. the concentration and self-quenching of entrapped CF, so that fluorescence decreased (11,(15)(16)(17). Fluorescence data from 8 -10 individual determinations were averaged and fit to a single exponential curve using software supplied by Applied Photophysics. The software utilizes a nonlinear regression (Marquardt) algorithm calculated from the time course using the "Curfit" routine. P f was calculated from the time course of relative fluorescence by comparing single-exponential time constants fitted to simulated curves in which P f was varied (11,(15)(16)(17). Simulated curves were calculated using a commercially available software package (MathCad) from the osmotic permeability equation: where V(t) is the relative volume of the vesicles at time t, P f is osmotic water permeability, SAV is the vesicle surface area-to-volume ratio, MVW is the molar volume of water (18 cm 3 /mol), and C in and C out are the initial concentrations of total solute inside and outside the vesicle, respectively. Since the volume within the vesicle was small compared with the volume outside, it was assumed that C out remained constant throughout the experiment. Parameters from the exponential fit (amplitude and end point) were used to relate relative fluorescence to relative volume using boundary assumptions that relative fluorescence and volume are 1.0 at time zero and that relative volume reaches a known value (if at time zero the osmolality outside is double that inside, the relative volume reaches 0.5 at the end of the experiment) (11,(15)(16)(17).

RESULTS AND DISCUSSION
To determine the effects of single leaflet structure on bilayer permeability, P d was measured using tritiated water and values corrected for unstirred layer effects are shown in Table I. To correct for unstirred layer effects, [ 14 C]butanol fluxes were performed in a manner identical with water fluxes. Because butanol is highly permeable across membranes, its flux measures the thickness of the unstirred layer (18). P d of butanol averaged 18.2 Ϯ 1.8 ϫ 10 Ϫ4 cm/s (n ϭ 3). Using this value and the known diffusion coefficient for butanol in water of 1.0 ϫ 10 Ϫ5 cm 2 /s (18), the unstirred layer thickness was 55 m. The accuracy of the calculated permeability of DPhPC bilayers using this correction for unstirred layers was determined by measuring P f of DPhPC liposomes. Importantly, the P f values obtained at similar temperatures were 16.7 ϫ 10 Ϫ4 and 15.0 ϫ 10 Ϫ4 cm/s in two different liposome preparations, in good agreement with the calculated P d from the chamber experiments of 14.9 Ϯ 1.7 ϫ 10 Ϫ4 cm/s. Since artificial bilayers in chambers or liposomes do not contain water channels, P d and P f are expected to be equal. Therefore, the agreement of the P f values with the P d values corrected for unstirred layers confirms the validity of the calculation of the thickness of the unstirred layers used for all of Table I.
The presence of CS in both leaflets markedly reduced P d . P d of asymmetrical bilayers were determined by adding CS to one leaflet and not the other. Although cholesterol rapidly flips from one leaflet to the other, CS exhibits a half-time for flipping of 14 h or more, a value far higher than the 4 h required for these measurements (19). P d of the asymmetric bilayer fell between the values obtained with symmetric DPhPC and DPhPC-CS bilayers. This result suggests that the resistance to permeation of a bilayer (the reciprocal of its permeability P BL ) is the sum of the resistances to permeation of its two leaflets (the reciprocal of permeabilities of each leaflet, P A and P B ): 1/P BL ϭ 1/P A ϩ 1/P B , where P BL is the measured permeability of the bilayer, and P A and P B are the permeabilities of the individual leaflets, A and B. In all artificial bilayer systems to date, both leaflets have been identical (11,20), and P A ϭ P B . If it is assumed that a leaflet resists permeation to the same degree whether it forms part of an asymmetric or a symmetric bilayer, then it is possible to calculate (as shown in Table I) the permeabilities of individual single leaflets from symmetric bilayers. Using these, we predict a permeability of the asymmetric bilayer of 7.9 ϫ 10 Ϫ4 cm/s, a value indistinguishable from the measured value.
The measurements of P d in bilayers indicate that individual leaflets can act as independent resistors to permeation in an asymmetric bilayer. To determine whether this finding applies generally to asymmetric bilayers, an entirely different asymmetric bilayer system was generated in large unilamellar vesicles (LUV) using Pr 3ϩ to reduce the fluidity of the external leaflet. Pr 3ϩ complexes with the phosphate head groups of phospholipid molecules, reducing their mobility and that of the adjacent hydrocarbon chains. This stabilization has been detected previously as an increase in the phase transition temperature of the outer leaflet of vesicles using NMR measurements (21-23). To determine whether, in our experiments, Pr 3ϩ was increasing the phase transition temperature of the outer leaflet, we monitored fluorescence anisotropy as an estimate of membrane fluidity. To examine the fluidity of the outer leaflet, we introduced a phospholipid-bound probe of anisotropy, 2-(3-(diphenylhexatrienyl)propanyl)-1-hexadecanoyl-snglycero-3-phosphocholine (DPH HPC) into the outer leaflet, and monitored fluorescence anisotropy as a function of temperature (24,25). Because phospholipids added to the outside of the liposomes enter the outer leaflet rapidly but cross to the inner leaflet slowly, this probe monitored exclusively the outer leaflet of the bilayer. As shown in Fig. 1A, addition of Pr 3ϩ raised the transition temperature for this probe by 2-3°C, in good agreement with the magnetic resonance studies (21-23). Fig. 1, B and C, shows the effect of Pr 3ϩ on water permeability at varying temperatures. In the absence of Pr 3ϩ , P f rose abruptly at 42°C, and leveled off at values of 0.03 cm/s, corresponding to the known phase transition temperature of DPPC. Following addition of Pr 3ϩ to the same LUV, P f was similar to control values at temperatures well below and above the phase transition temperature. However, the phase transition temperature for P f was raised by 2-3°C, so that Pr 3ϩ markedly reduced P f at temperatures in the vicinity of the transition temperature for control LUV. These results indicate that reducing the fluidity of a single leaflet reduces the permeability of water across the entire bilayer.
To determine whether the resistance equation applies as well to the LUV exposed to Pr 3ϩ , leaflet A (DPPC without Pr 3ϩ ) permeabilities were calculated from the values for control LUV at temperatures above the phase transition. Leaflet B (DPPC ϩ Pr 3ϩ ) permeabilities (P B ) were estimated by taking the anisotropy value obtained with Pr 3ϩ and applying that value to a standard curve relating permeability to anisotropy in control LUV. Because we have previously shown that DPH anisotropy is directly related to measured water permeability (11), this approach should give an accurate estimate of the water permeability of leaflet B. Using these P A and P B values, we calculated expected values for P BL , and these expected values are shown in Fig. 1C (as filled triangles) in comparison with the measured values (filled squares). It is apparent that this equation applies equally well to diffusive water permeability in bilayers made asymmetric with CS and to osmotic water permeability in liposomes made asymmetric with Pr 3ϩ . Previous studies had provided evidence that individual leaflets of membrane bilayers could alter their physical properties independent of the other leaflet (23,24). Our results demonstrate for the first time that individual leaflets of the membrane bilayer can independently regulate permeation. If the permeabilities of both leaflets are similar, then both will contribute similarly to membrane permeability. By contrast, if one of the leaflets has a very low permeability, the permeability of this leaflet will predominate, so that the permeability of the entire bilayer will be close to that of the low permeability leaflet. This conclusion has several important implications for our current understanding of the role of bilayer asymmetry in epithelial cell biology.
We have previously reconstituted lipids quantitatively extracted from gastric apical membrane vesicles and shown that the reconstituted lipids newly arranged in artificial liposomes do not reconstitute low water permeability (15). The failure of the extracted lipids when reconstituted into artificial liposomes to duplicate the low permeabilities of the intact membrane may be due to the influence on permeability of two factors in the intact membrane: the presence of integral membrane proteins or the arrangement of the lipid components of the membrane into asymmetrical leaflets (15). Our new results show directly that segregation of lipid of low fluidity in a single leaflet of the bilayer reduces the permeability of the entire bilayer. Therefore, we can anticipate that cells create apical membranes of low permeability by segregating phospholipid molecules with long saturated hydrocarbon chains and cholesterol in the outer leaflet of the bilayer. Indeed, where the composition of individual bilayer leaflets has been examined, this segregation has been observed (26 -28). Such segregation of lipids is important in determining the low permeability properties of the apical as opposed to the basolateral membrane domain, because, in epithelial cells, lipids of the exofacial but not cytoplasmic leaflet of the apical membrane are prevented from mixing with those of the corresponding basolateral membrane leaflets by the tight junctions (3,6). These considerations plus the evidence from the current study indicate that the lower permeabilities of apical as compared with basolateral membrane domains are due to the lipid structure of the exofacial leaflet.
Cells invest a great deal of effort to generate and maintain bilayers with distinct leaflet compositions. Membrane biosynthesis in the Golgi apparatus occurs in an asymmetric fashion, with distinct lipids going into the different membrane leaflets (1-3). Moreover, cells maintain phospholipid "flippases," in their plasma membranes (4,5). These transporters couple ATP degradation to the movement of phospholipid molecules from one leaflet to the other, a process which is otherwise energetically unfavorable and occurs at a very slow rate. Our studies provide direct biophysical evidence that the generation and maintenance of asymmetric apical membranes can result in effective barrier function.
The present studies emphasize the importance of segregation of low fluidity lipids in the exofacial leaflet for maintenance of low permeability. Any disease process which disrupts the ability of cells to create and maintain membranes with distinct cytoplasmic and exofacial leaflets may result in failure of apical membrane barrier function, with resulting damage to subepithelial structures or loss of homeostatic function. Candidate diseases which may disrupt the generation and maintenance of asymmetric bilayer structure include ulcer disease in the stomach, cystitis in the bladder, and renal tubular acidosis and inability to concentrate or dilute the urine in the kidney collecting duct.
FIG. 1. Effect of external Pr 3؉ on phase transition temperature for outer leaflet fluidity and for osmotic water permeability (P f ) of DPPC liposomes. A, fluidity of the outer leaflet was monitored using DPH HPC (excitation 360 nm, emission 430 nm) at a probe:lipid ratio of 1:400 at varying temperatures in the absence (filled circles) and presence (filled squares) of 10 mM extravesicular Pr 3ϩ on a SPEX spectrofluorimeter using standard methods (24,25). B, measurements of water flux at 44°C in the absence (DPPC) and presence (DPPC/Pr 3ϩ ) of 10 mM extravesicular Pr 3ϩ . Data from 6 -10 curves were averaged and fitted as described; averaged data and fitted curves are shown. C, effect of extravesicular Pr 3ϩ on P f at varying temperatures. P f was calculated from curves similar to those of B at varying temperatures as described in the absence and presence of 10 mM Pr 3ϩ . Data from 6 different experiments are shown. Values obtained with Pr 3ϩ differ significantly from those obtained in the absence of Pr 3ϩ at 42-46°C, by t test.