Blotechnol. Prog. 1992, 8, 546-552
546
Biomimetic Metal-Sorbing Vesicles: Cd2+ Uptake by Phosphatidylcholine Vesicles Doped with Ionophore A23187 John H. van Zanten and Harold G. Monbouquette’ Chemical Engineering Department, University of California at Los Angeles, Los Angeles, California 90024-1592
Unilamellar phosphatidylcholine vesicles, harboring the ionophore, A23187, in the bilayer and the water-soluble chelating agent, nitrilotriacetate, in the vesicle interior, rapidly sequester and concentrate Cd2+from dilute aqueous solution. Metal-sorbing vesicle permeabilities for cadmium ion at 5 ppm (42.8 pM) ranged from 8.09 X lo-’ to 1.27 X cm/s for surface A23187 concentrations of 0.22-2.27 pmol/cm2 (which correspond to 1ipid:carrier molar ratios of 2000:l to 200:l) and pH’s from 5.5 to 8.5. The Cd2+ permeability shows linear variation with carrier concentration under the conditions studied. As pH is decreased, an increasing fraction of the A23187 becomes protonated, and the permeability exhibits a positive linear relationship with a function related to that for the fraction of unprotonated carrier. These noncovalently assembled, metalsorbing vesicles exhibit shelf lives of several months and remain stable throughout typical metal sorption studies.
Introduction Vesicles are closed-membrane capsules, typically of submicrometer diameters, that consist of single or multiple bilayers of noncovalently assembled surfactants. Since the basic structural unit of a cell membrane also is a surfactant bilayer, vesicles and cell membranes have much in common. In fact, liposomes, which are vesicles composed of natural phospholipid surfactants, have been used extensively as models for study of cell membrane structure and function (Fender, 1982;Johnson andBangham, 1969). Surfactant bilayer membranes not only define cell boundaries and compartmentalize functions within intracellular organelles but also provide the scaffolding for apparatuses involved in energy transduction, signal processing, and separations. The outstanding ability of organisms to sense particular chemical species in their environment, to perform efficient energy conversions, and to selectively sequester nutrients while retaining valuable intracellular metabolites has been well-documented (Neidhardt et al., 1990). Our effort has focused on the development of a vesicular membrane system that mimics both the structure and function of biological cell membranes for the uptake and concentration of heavy metal ions from dilute aqueous solution. The basic principles operable in cell membrane separations closely resemble those of supported or emulsion liquid membranes (Lonsdale, 1984;Borwankar et al., 1987; Noble and Way, 1987; Noble et al., 1989). The lipophilic hydrocarbon core of a surfactant bilayer membrane functions as a virtually impermeable barrier to charged and polar molecules (which are large relative to water). Transport of selected ionic or polar species across the membrane is mediated by membrane-bound carriers or transmembrane pores. Concentration of species on one side of a liquid membrane or within a cell is accomplished by coupling membrane transport with an energetically favorable process such as an opposing transmembrane flux of protons down a steep concentration gradient or a chemical conversion of the speciesupon transport (Cussler, 1971; Lonsdale, 1984; Neidhardt et al., 1990). Although vesicular membranes have much in common with well-studied liquid membranes, they may prove to 8756-7938/92/3008-0546$03.00/0
be more advantageous. Their relatively small size gives rise to a much larger available area for separations on a mass basis. For example, a 1% (vlv) dispersion of 100-nm vesicles provides 3000 m2 of vesicle surface area per liter and 40 mL of encapsulated volume. This high surface to volume ratio makes these vesicle systems ideal candidates for the recovery and concentration of species from dilute solution. The large contact area along with the exceedingly thin bilayer thickness (3-5 nm) makes solute sorption kinetics very rapid. In addition, unlike liquid membranes, vesicular membranes do not leach organic solvents into the aqueous phase. Finally, metal-sorbing vesicles, at about 100 nm in diameter, can be harvested readily for possible regeneration using hollow-fiber filtration cartridges. Other investigators have proposed the use of engineered surfactant vesicles for use in sequestering ions. Radioactive l111n3+has been concentrated in vesicles containing the chelator, nitrilotriacetate, for use in medical imaging (Hwang, 1978; Mauk and Gamble, 1979). Hamilton and Kaler (l987,1990a,b)theorized that vesicles could be used as separation media on the basis of their work on the permeability of synthetic surfactant vesicular membranes to Na+ and K+ in the presence and absence of synthetic, crown ether ionophores. Also, it has been shown that EDTA can be used to concentrate divalent cations in phosphatidylcholine vesicles (Pohl et al., 1980) and to increase the flux of Ca2+ across planar bilayer lipid membranes (Pohl et al., 1990). In this report, we describe the concept of metal-sorbing vesicles as agents for metal ion separations and present an illustration of the ability of an engineered vesicular membrane system to remove and concentrate Cd2+ions from dilute aqueous solution. We have used the naturally occurring membrane-forming lipid, phosphatidylcholine, to form highly stable vesicles using detergent dialysis. The gentle, “equilibrium” detergent dialysis approach also yields vesicles nearly uniform in size (van Zanten and Monbouquette, 19911, whereas sonication and extrusion methods give less stable vesicles with a relatively broad size distribution (Szoka and Papahadjopoulos, 1980). Vesicle-forming protocols based on detergent dialysis can
0 1992 American Chemical Soclety and American Institute of Chemical Englneers
BiotechnoL Rag.., 1992, Vol. 8, No. 6
Figure 1. Metal-sorbing vesicle structure. This drawing is not to scale. Typical vesicle diameters are on the order of 100 nm, while the soybean phosphatidylcholine bilayer thickness is approximately 3.7 nm.
be modified easily to permit incorporation of lipophilic structures in the vesicle wall and to facilitate encapsulation of hydrophilic species. We have incorporated the natural antibiotic, A23187, in the vesicle bilayer to serve as a shuttle for cadmium ions. A23187 (also known as calcimycin) is a highly selectivenatural ionophore for divalent cations (Chapman et al., 1990; Reed and Lardy, 1972). Many reports have appeared in the literature describing the effects of A23187 and other ionophores on biological systems, largely through their ability to effect Ca2+membrane transport (Westley, 1982). The logarithms of stability constants for 1:l complexes between A23187 and various cations in 80% methanol in water at basic pH have been reported to range from 9.77 to 2.36 (Chapman et al., 1990): Cu2+ >> H+, Ni2+,Fe2+,Co2+> Zn2+,Pb2+,Cd2+,Mn2+>> Mg2+,Ca2+ > Sr2+,Ba2+,Li+ > Na+. To provide a driving force for Cd2+uptake by the vesicles, we have encapsulated the water-solublemetal chelatingcompound, nitrilotriacetate, in the vesicle interior (see Figure 1). Our emphasis has been on the recovery of Cd2+from fairlydilute (5ppm and lower) solutionssince conventional methods for metal recovery tend to become impractical a t these concentrations. We have investigated the effect of pH and A23187 concentration on the sorption of Cd2+. We also have monitored the vesicles using static light scattering (van Zanten and Monbouquette, 1991) throughout the metal sorption process to assess the stability of the suspension when the vesicles are brought in contact with a solution bearing divalent metal cations. Materials and Methods Reagents and Solutions. L-a-Phosphatidylcholine from soybeans, sodium cholate, nitrilotriacetic acid (trisodium salt),and the buffers (sodiumsalts) HEPES, MES, and PIPES were purchased from Sigma (St. Louis, MO). Cadmium solutions were prepared from an Aldrich (Milwaukee, WI) cadmium atomic absorption standard solution (Cd in 1w t % HN03). Potassium nitrate (99.999%) also was obtained from Aldrich. Biological-gradesodium chloride and HPLC-grade methanol were purchased from Fisher (Tustin, CA). AntibioticA23187 was obtained from Calbiochem (San Diego, CA). All aqueous solutions were prepared with water purified by a MilliporeMilli-Qsystem. The MES, PIPES, and HEPES buffers were used as 25 mM solutionsadjusted to pH 5.5,7.0, and 8.5, respectively. Preparation of the Mixed Micellar Solution. A total of 225 mg of sodium cholate and 250 mg of L-a-phosphatidylcholine were dissolved in chloroform/methanol to give a solution with a lipid to detergent molar ratio of approximately 0.65. Antibiotic A23187 (ionophore) was
547
added to the solution to produce the desired ionophore: lipid ratio. This detergent, lipid, ionophore, and chloroform/methanol solution was then dried in a 50-mL roundbottomed flask at 40 "C under vacuum for 4 h using a rotary evaporator (Wheaton Instruments). The resultant film was resuspended in 12 mL of unbuffered 50 mM trisodium nitrilotriacetate (NTA) to yield a clear, aqueous solution containing free detergent, mixed lipid/detergent/ ionophore micelles, and NTA. Preparation of Vesicles by Detergent Dialysis. The mixed micellar solution was dialyzed against approximately 1L of unbuffered 50 mM trisodium NTA using a Liposomat (MM Developments,Ottawa, Ontario, Canada), a detergent dialysis device, with dialysis membranes (Diachema,MM Developments)having a molecular weight cutoff of 10 000. A near size monodisperse population of unilamellar vesicles is obtained which harbors A23187 in the lipid bilayer and 50 mM NTA (with 150 mM Na+) in the aqueous vesicle interior. Removal of the External Nitrilotriacetate. Unencapsulated NTA was separated from vesicles by passing the vesicle dispersion through a size exclusion column (1.5 cm X 20 cm) packed with Spectra/Gel AcA 202 (Spectrum Medical Industries, Los Angeles, CA). The mobile phase was unbuffered 150 mM NaCl (pH 5.0-5.1). This separation process yields a suspension of vesicles in 150 mM NaCl at a pH of 5.0-5.1. Determination of the Vesicle Weight Concentration. After the unencapsulated NTA was removed, aliquots of the resulting vesicles in NaCl solution were dried and weighed. The encapsulated NTA and external NaCl dry weights were subtracted from the total to give the lipid dry weight. The ionophore contribution to the dry weight was assumed to be negligible. Typically the concentration of lipid fell between 15 and 20 mg/mL. Size Characterization of the Vesicles. The vesicles were size characterizedby multiangle static light scattering using a Wyatt Technology (Santa Barbara, CA) laser photometer. Rayleigh-Gans-Debye theory and Zimm plots were used to estimate vesicle diameters and weightaveraged molecular weights from multiangle scattered light intensity measurements (van Zanten and Monbouquette, 1991). Given the information gained from analysisof light scattering data and the weight concentration of vesicles in suspension, the surface area and encapsulated volume of the suspension were calculated. Cd2+Uptake Measurements. Cadmium ion uptake by engineered vesicles from a buffered solution of known concentration was followed with a solid-state ion-selective electrode and a double-junction reference electrode containing gelled KC1 in the upper compartment and gelled KN03 in the lower compartment (Gam Rad West, San Juan Capistrano, CA). A 50-mL polypropylene reservoir was filled with approximately20 mL of a solution composed of 100 mM KN03,5 ppm Cd2+,and 25 mM Good's buffer. These buffers contribute 25 mM Na+ (HEPES and MES) or 37.5 mM Na+ (PIPES). A small amount of 1.5-2.0% (w/v)vesicle stock solution, typically on the order of a few hundred microliters, was pipeted into the magnetically stirred reservoir to initiate the Cd2+uptake process. An Orion 720A pH/ISE meter (Boston, MA) connected to a PC was used to capture data from the electrode every 5 s. Since the ion-selective electrode cannot detect metal ions sorbed by the vesicles, this technique allows direct, rapid measurement of the free cadmium ion concentration. Results Effect of pH on Cd2+Sorption Capacity. The effect of pH on the sorption and concentration of cadmium ions
Bbtechnol. Rog., 1902, Vol. 8,
548
F
X
0
Time (sec) Figure 2. Cadmium ion uptake by metal-sorbing vesicles. Dimensionless Cd2+concentration (relative to an initial concentrationof 5 ppm) is plotted versus time at pH 5.5 (A), 7.0 (+), and 8.5 (0). The phosphatidylcholine:A23187 molar ratio was 200:1,and the vesicle dispersionconcentration waa 0.03% (w/v). Also shown is ( 0 )the result from a control study with vesicles harboringencapsulated NTA but having no A23187 in the vesicle wall and (A) the response of the electrode to the addition of excess NTA directly to a 5 ppm solution of Cd2+.
in vesicles is illustrated in Figure 2. The cadmium ion concentration presented is normalized with respect to the initial concentration of 5 ppm. The pH dependence of Cd2+ uptake is shown for three pH's, 5.5, 7.0, and 8.5, which correspond to the middle three of the five curves exhibited. The lower curve is the electrode response to sudden complexation of Cd2+by addition of excess NTA directly to the reservoir. The rapid electrode response indicates that electrode kinetics and reservoir mixing times do not need to be taken into account in estimating metal sorption parameters from the raw data. The uppermost curve corresponds to a control study where vesicles harboring NTA without any carrier in the vesicle wall were used. Imperceptible metal ion uptake for this control over the time period studied demonstrates that the disappearance of metal ion observed when A23187 is present is due to sorption by the engineered vesicles (rather than complexation with leaked NTA). Since pH influences both kinetics and capacity significantly, it appears that both the carrier and chelator are affected a t the lower p H s studied. The influence of pH on the initial rate of Cd2+uptake, which is due primarily to protonation of the carrier, is discussed later. On the basis of the three pK:s for NTA, 1.9,2.5,and 9.8 (Chaberek and Martell, 1959; Dawson et al., 1986) (defined as -log K1, -log K2, and -log K3, respectively), the ratio of total NTA to total cadmium in the reservoir (1.07 for all cases), and the initial Cd2+ concentration, the free metal ion concentration at equilibrium can be estimated using the equation (Chaberek and Martell, 1959) I
\
where
I , , , , I , , , ,
0 0.5 1 1.5 2 2.5 A23187 Surface Concentration (pmol/cm2)
Figure 3. Initial vesicle permeability to Cd2+versua A23187 vesicle surface concentration in pmol/cm2at pH 5.5 (O),7.0 (A), and 8.5 (m). The four A23187 concentrations shown correspond to phosphatidylcholine:A23187 molar ratios of 2000.1, 1ooO:1, 500:1,and 200:l and total A23187 concentrations (based on the reservoir volume) of 0.32,0.59, 1.07, and 1.97 rM,respectively.
[NTAI) does not interfere with Cd2+chelation. According to eq 1, 78.6%, 97.3%, and 99.9% of the free cadmium should be complexed a t equilibrium at pH 5.5, 7.0, and 8.5, respectively, by NTA in free solution. However, our experimental results with encapsulated NTA where 92 % and 98% of the free Cd2+is sorbed by the vesicles at pH 7.0 and 8.5, respectively, suggest that Donnan effects may influence the capacity of the system since the phosphatidylcholine bilayer essentially is impermeable to all anions present and there is an unequal distribution of ionic species across the membrane (Donnan, 1925; Lakshminarayanaiah, 1984). At pH 5.5, the sorption kinetics is slowed considerably as only 40 % of the free Cd2+is removed in 10 min. Unlike the results a t pH 7.0 and 8.5, the pH 5.5 Cd2+uptake curve exhibits a significant negative slope at 10 min, which indicates that the system still is far from equilibrium. Nevertheless, one should note the rapid overall rate at which cadmium ions were removed from the reservoir and concentrated in the vesicle interior by these engineered biomimetic vesicle systems, at neutral to basic pH. The impressive kinetics was achieved with a vesicular dispersion concentration of only 0.03 w t % ,which corresponds to an encapsulated volume of only 18.6 pL in the 20-mL reservoir. The cadmium was concentrated inside the vesicles by factors of 1040,990, and 430 at the pH's 8.5, 7.0, and 5.5, respectively, over the time period shown. The totalcarrier concentration (based on the reservoir volume) was only 1.97 pM, while the total cadmium ion concentration in the reservoir was 42.8 pM. Thus, the reduction in the concentration of free cadmium ions must be due to their being sequestered by the encapsulated chelating agent, nitrilotriacetate. The rapid system response is due largely to the exceedingly small size of these vesicles, at 84 nm in diameter, and the 17 600 cm2of surface area for separations that they provide in each of the cases shown. Effect of Carrier Concentration on Initial Permeability. The effect of the A23187 carrier concentration on the initial cadmium ion permeability is shown in Figure 3. The permeability was found by fitting the initial points in the free Cd2+concentration traces, where the decrease was exponential with time, to the following equation: [Cd2+l - exp(-PARt) --
and KMAis the binding constant of NTA for Cd2+. This expression includes the assumptions that NTA and Cd2+ form only 1:l complexes and that Na+ (at 3:l [Na+l:
No. 6
(2) [Cd2+lO Here, [Cd2+lorepresents the initial reservoir concentration, P denotes the initial vesicle permeability, and AR is the vesicle area divided by the reservoir volume. By using
649
Biotechnol. Prog., 1992, Vol. 8, No. 6
Table I. A23187 Turnover Numbers for Cd2+Transport (E-')
uH 8.5
7.0 5.5
2000:l 4.3 2.0 0.16
lipid:A23187 molar ratio 1ooo:l 5m1 4.1 2.7 1.5 2.1 0.14 0.14
1.4 1.2
2m1 2.2 1.2
0.17
this expression, we are assuming that the cadmium concentration in the vesicle interior is essentially zero for small t and that the metal ion flux is a linear function of Cd2+concentration at the low concentration of 5 ppm (42.8 pM). As a point of reference, others have shown that the A23187-mediated influx of Ca2+into phosphatidylcholine vesicles is not saturated and is linearly dependent on calcium ion a t concentrations of 3-18 mM (Blau et al., 1984). The permeabilities reported here, ranging from to 1.27 X cmls, are in the expected range 8.09 X based on published Ca2+permeabilities for planar phosphatidylcholine bilayers (Kafka and Holz, 1976;Wulf and Pohl, 1977) and given that Cd2+ transport into phosphatidylcholine vesicles containing EDTA has been found to be 3-fold greater than that of Ca2+(Pohl et al., 1980). A t 5 ppm Cd2+and over the narrow range of carrier concentration studied, P appears to vary linearly with A23187 concentration at a particular pH. However, the permeabilities a t neutral to basic pH may be systematically underestimated due to (1)the limited number of data points gathered early in the sorption run at 5-s intervals; (2) a fraction of a second delay in starting to record the free Cd2+concentration depletion after injection of vesicles at the initiation of the metal sorption process; and (3) loss of up to 25% of the A23187 into the reservoir at pH 8.5 as a consequence of its slight solubility in water. The apparent membrane association constants for A23187, defined as [A23187]mem/([A23187] [PL]), where [A23187]mem is the membrane-bound A23187 concentration and [PL] is the phospholipid concentration, have been measured as 1.9 X lo4 and 7 X lo3 M-l at pH 7.06 and 8.7, respectively (Kolber and Haynes, 1981). Thus, the greatest loss of carrier would be expected at higher pH. (Since the A23187 in the aqueous phase still participates in metal ion transport, through binding to Cd2+ in the aqueous phase and subsequent re-incorporation into the bilayer to facilitate ion transport, we include all A23187 in the calculation of the surface carrier concentrations presented in Figure 3. The corresponding lipid to carrier and total A23187 concentrations, based on the reservoir volume, also are given in the Figure 3 caption.) These systematic errors could be substantial enough to give permeabilities for the higher carrier concentrations at pH 7.0 and 8.5 that would fall on a line through the origin. A23187 turnover numbers, calculated as P[Cd2+1ol [A23187ltotal, provide additional evidence that the permeabilities at higher pH may be underestimated (seeTable I). Since the system does not show saturation behavior with respect to A23187 concentration and the formation of 1:l carrier-ligand complexes is expected to be limiting over the carrier concentration range studied (Kolber and Haynes, 19811, the turnover numbers should remain constant at a given pH. At pH 5.5, the turnover numbers are roughly constant. However, at pH 7.0 and 8.5 where Cd2+uptake is more rapid and the systematic underestimation of the permeability would be expected to be more evident, the turnover number declines with increasing carrier concentration. In agreement with our higher turnover number at 20001 lipidcarrier and pH 8.5, others have estimated a turnover number for A23187 transporting Ca2+at pH 8.0 of about 5 s-l (Caswell and Pressman, 1972;
1 0.8 0.6
0.4 0.2
0
o
0.2
0.6 f(PH)
0.4
0.8
1
Figure 4. Initial vesicle permeability to Cd2+versus f(pH) = 1/[1+10(6.*pH)] at phosphatidylcholine:A23187 molar ratios of 2000:l (+),1ooO:1(~),500:1(~),and200:1 (O),whichcorrespond to surface concentrations of 0.22,0.40,0.86, and 2.27 pmol/cm2, respectively.
Wulf and Pohl, 1977). This data suggests that, at pH 8.5 and a lipidcarrier ratio of 200:1, the permeability, and hence the turnover number, may be underestimated by nearly a factor of 2. Nevertheless, the permeability data presented here, along with reasonable estimates of random and systematic error, could not be used to support an argument that transport of Cd2+is limited by formation of a neutral 2:l (carrier:metal ion) complex; the data simply are not nonlinear enough. The observed relationship between permeability and carrier concentration more likely indicates that the formation of a 1:l A23187-cadmium ion complex or Na+ countertransport from the vesicle interior is limiting the uptake of Cd2+under these experimental conditions. At the reservoir pH of 8.5 when the vesicle is at a lower pH of about 5.0, the rapid efflux of free protons from the unbuffered vesicle interior can account only for an exceedingly small exchange of Cd2+ions with the vesicle exterior. Therefore even at this pH, the efflux of Na+, which binds 4 orders of magnitude less strongly with A23187 (Chapman et al., 19901, may be limiting. Effect of pH on Initial Permeability. As is evident from Figures 2 and 3, the rate of cadmium sorption is reduced at lower pH's, as expected given the reported pKa for A23187 in phosphatidylcholine bilayers of 7.85 (Kauffman et al., 1982). The data are correlated linearly at a given carrier concentration (see Figure 4) by the following function of pH,
which is a function closely related (i.e., the constant, 6.9, substituted for pK,) to that for the fraction of unprotonated A23187 in solution in the absence of other binding cations. This strong correlation is intriguing given that the relevant quantity would seem to be the fraction of vesicle-bound A23187 that is free for chelation of Cd2+. For these experimental conditions, such a quantity must take into account the fraction of carrier that is bound to Na+, K+, and Cd2+as well as protons. Unfortunately, the fraction of vesicle-bound A23187 could not be calculated, as association constant data is available only or 1:lA23187Cd2+complexes in methanol/water (Chapman et al., 1990) and no data exists for K+ and for 2:l A23187-Cd2+ complexes. The substantial initial gradient in proton concentration across the vesicle wall at reservoir pH's of 5.5 and 8.5 does not further complicatethis analysis, since pH equilibration across the vesicle wall likely is rapid. A23187 shows at
550
.
2
Biotechnol. Rog., 1992, Vol. 8, No. 6 1.15
TI
E 0
-.-
1.10
Q)
V
m
1.05
m
-c
I-
t
iTT
Q)
.-0m 1 . 0 0 0
E
50.950
!i
p
' ' '
0
"
' ' ' '
50
I ' ' ' ' I ' '
100
150
I
' '
200
*
250
Time (min)
Figure 5. Metal-sorbing vesicle size during the sorption of Cd2+ at pH 8.5. Dimensionless vesicle diameter (relative to initial diameter), as determined by applying Rayleigh-Gans-Debye theory to static light scattering data, is plotted versus sorption time in minutes.
least an order of magnitude higher affinity for protons than for any of the other cations present (Chapman et al., 1990), and given the small vesicle-encapsulated volume, an exceedingly small number of protons needs to be transported for pH equilibration. In addition, it is known that the rate constant for translocation of the 1:l H+A23187 complex across a lipid bilayer is 3 orders of magnitude higher than that for 2 1 A23187-M2+ complexes (Kolber and Haynes, 1981). This comparison serves as additional indirect evidence that pH equilibration under these conditions is rapid relative to Cd2+ transport and Na+ countertransport. Metal-Sorbing Vesicle Stability. These metal-sorbing vesicles are stable for several months on the shelf. In addition, exposure for 1day to a temperature of 90 "C or to 1N acid or base causes no deterioration in stability of the suspension. Vesicle dispersion stability is indicated by no evidence of visual aggregates or precipitates and by the continued ability of the suspension to scatter light as size-homogeneous, unilamellar vesicles. The metal-sorbing vesicles appeared to be stable throughout the sorption process. Typical vesicle diameter measurements based on light scattering data are shown in Figure 5 for the pH 8.5 case over the first 50 min of a sorption run. Throughout this time period, the vesicles scattered light as isotropic hollow spheres within the Rayleigh-Gans-Debye approximation (van Zanten and Monbouquette, 1991)with no visual evidence of aggregates or precipitates. This indicates that the vesicles retain their spherical, capsular geometry during the metal sorption process. Discussion In the accepted models for A23187-mediated transport of divalent cations (Caswell and Pressman, 1972; Kolber and Haynes, 1981; Chapman et al., 1987), an A23187 molecule first binds a single cation at the membrane/ aqueous solution interface. Subsequently, a second A23187 combines with the 1:l complex to form a neutral 2:l carrier-cation complex that can translocate to the opposite membrane interface where the complexation process may reverse. Using fluorescence techniques, Kolber and Haynes (1981) obtained convincing evidence that divalent cations are transported across lipid bilayer membranes predominantly in 2: 1neutral complexes. This observation does not conflict with other published reports indicating that the permeability of lipid bilayers doped with A23187 varies linearly with A23187 concentration (Hunt et al., 1978; Kafka and Holz, 1976; Wulf and Pohl,
1977) or varies with the square of A23187 concentration (Blau et al., 1984; Pohl et al., 19801, as any step in the complexation and translocation process can be limiting (Caswell and Pressman, 1972; Kolber and Haynes, 1981). Our study was conducted at total carrier concentrations (based on the reservoir volume) above 0.1 pM where a linear relationship would be expected between permeability and A23187 concentration (Kolber and Haynes, 1981). There are difficultiesin comparing our results with those in the literature, however. Although a number of studies have appeared that discuss aspects of A23187-mediated transport of multivalent cations across the walls of lipid bilayervesicles (Blau et al., 1984;Hunt et al., 1978;Kolber and Haynes, 1981;Pohl et al., 1980;Pohl et al., 1990;Shastri et al., 19871, initial permeabilities of the type reported here have never been published previously for vesicles. Ion transport data typically were gathered over at least severalminutes for the calculation of time-averaged kinetic properties, and the influence of a changing transmembrane cation concentration gradient or electropotential could not always be ruled out. In the two studies most closely related to our work, where metal ion uptake by A23187doped vesicles was inferred from an efflux of protons (Pohl et al., 1980) and where metal ion efflux from A23187doped vesicles was mirrored using a change in the absorbance of a metal-complexing dye (Blau et al., 1984; Pohl et al., 19801, insufficient information was presented to calculate initial permeabilities. These permeability calculationsrequire knowledge of vesicle surface area and encapsulated volume as well as trustworthy data on the initial uptake or release of metal ions from vesicles. Vesicle size information is frequently absent from these studies. Further, the method used to follow metal ion concentration often is not characterized fully in terms of the response time or the relationship between the metal ion concentration and the measured quantity, e.g., metal ion/Arsenazo I11 dye complex absorbance (Blau et al., 1984). A popular approach has been to analyze ion leakage from vesicles separated from a surrounding buffered reservoir by a dialysis bag (Johnson and Bangham, 1969; Hamilton and Kaler, 1987). Using this method, Hamilton and Kaler (1987)have reported Na+and K+permeabilities ranging from 5.1 X 10-locm/s for Na+ and a surfactant: metal carrier ratio of 1OOO:l to 2.75 X 10+ cm/s for K+ and a surfactant:metal carrier ratio of 1OO:l for vesicles composed of a synthetic surfactant doped with a crown ether ionophore. In order to conduct these experiments, the vesicles first must be loaded with metal ions and then passed through a desalting gel column to remove residual metal ions external to the vesicles before the vesicles are charged to the dialysis bag. Due to the cumbersome, timeconsuming nature of this process, the first minute or so of metal ion leakage from the vesicles, which is critical for the determination of initial permeabilities, may be missed. The dialysisbag also complicates the analysisby presenting another mass transfer resistance that can obscure rapid ion leakage kinetics. These comments may partially explain, in addition to the differences in the chemical composition of the systems, the difference of several orders of magnitude between the vesicle permeabilities reported here and those measured by Hamilton and Kaler (1987) for their system. We report P as an initial permeability in this study because the simple exponential expression given by eq 2 cannot be used to fit sorption data for later times, despite the fact that at neutral to basic pH's encapsulated NTA effectively maintains the free cadmium ion concentration
551
Biotechnol. Prog.. 1992, Vol. 8, NO. 6
in the vesicle interior a t or near zero for much of the sorption process. The slower efflux of Naf from the vesicles may limit cadmium ion transport at intermediate times when the internal sodium ion concentration is reduced from its relatively high initial concentration of 150mM. At relatively large t, when the encapsulated NTA approaches saturation with cadmium, a reduced chelation rate likely influences the overall kinetics as well. An effort currently is being made to include these effects in a model of metal sorption by these vesicles. On the basis of the data shown, the rapid initial kinetics as well as the capacity of this system could be improved further by increasing the vesicle concentration by 4 orders of magnitude to about 10% (w/v), which would give an encapsulated volume of about 50%. We have not worked a t higher vesicle loadings because the very rapid kinetics of metal uptake would then be obscured by the slower electrode response and we could not obtain useful data on vesicle permeability. However, we expect that, in addition to increasing the vesicle suspension concentration to about 10% (w/v), we can, if desired, increase the internal NTA concentration to 0.5 M and improve the A23187 to phosphatidylcholine mole ratio to about 150. These changes would increase by orders of magnitude both the metal loading capacity and the kinetics of metal uptake. Since the driving force for the separation is the encapsulated chelating agent and NTA has a very high binding constant, metal ion concentrations in the surrounding solution can be driven very low. As a point of reference, the NTA binding constant for Cd2+is 6.3 X lo9 (Chaberek and Martell, 1959). Therefore at equilibrium with sufficient NTA, the concentration of cadmium in the chelated state is many orders of magnitude higher than that of the free ion. In addition, since transport of divalent cations by A23187 occurs by way of a 2:l A23187:M2+ neutral complex, a diffusion potential does not arise to inhibit metal ion uptake by our metal-sorbing vesicles. Obviously this system can be used to drive metal ion concentrations in a waste stream to extremely low levels, or it can be exploited for the harvesting of metal from very dilute solutions. When a chelating agent such as NTA is used to provide the driving force for metal ion uptake by ionophore-doped vesicles, the overall selectivity of the system is dependent on the selectivity profile of both the carrier and chelator. Since the chelator provides the driving force for metal uptake by depleting the free metal ion concentration in the vesicle interior, it provides a greater driving force for those ions for which it has the greatest affinity. If the metal carrier in the vesicle wall is characterized by a high affinity for the same ion(s) as the chelator, the overall metal ion selectivity of the vesicle system will be greater than that of the carrier or chelator alone. In a crude sense, if the probability that both carrier and chelator will bind an undesired cation over the target metal ion is one out of a hundred (and these are roughly independent events), then the overall probability that the vesicle will sequester the undesired cation would be on the order of one in ten thousand. In such a way, these vesicular metal-sequestering media can be engineered for outstanding selectivity. Although the selectivity of metal-sorbing vesicles has not been demonstrated as yet, the versatility in the design of these systems both a t the level of the membrane-bound carrier and the encapsulated water-soluble chelator provides for this potential. A large number of natural and synthetic carriers have been identified with various selectivityprofiles that would be suitable for use in our system (Lamb et al., 1981;
Hilgenfeld and Saenger, 1982; Izatt et al., 1987; Noble et al., 1989; Fyles, 1990). Relatively expensive carrier chemistries could be used in our system since the capacity for metal ion uptake is not dependent on the carrier concentration. The concentration of internal water-soluble chelating agent provides the loading capacity, and chelators, such as NTA ($22.85/kg, Sigma, 19921,are relatively inexpensive. Charged carriers such as A23187 also can be used in metal recovery schemes that do not depend on encapsulated chelating agents, but utilize an opposing pH gradient to supply the energy to concentrate metal ions within the capsular membrane. Additional charged carriers that would be suitable include the Henkel L E carriers (Lonsdale, 1984), which have proven useful for copper, nickel, and zinc recovery, the compounds of Marshall et al. (1988), and macrocycles derivatized with proton ionizable groups (Izatt et al., 1987; Fyles, 1990). Useful carriers not only must participate in cationtmetal ion exchange reactions and exhibit selectivity but also must be sufficiently lipophilic to remain immobilized in the vesicle wall. A number of practical considerations ultimately will determine the utility of metal-sorbing vesicles including the development of truly selective systems with adequate shelf life and physicochemical stability. Issues such as these related to technologicalfeasibilitycurrently are under intensive investigation.
Conclusions Cadmium ions in dilute aqueous solution can be sequestered and concentrated several hundred-fold in a few minutes time by metal-sorbing vesicles. As expected, the vesicle permeability increases with carrier concentration but declines with pH. The actual permeabilities measured, 8.09 X lO-'to 1.27 X lo4 cm/s, are several orders of magnitude higher than those estimated for Na+ and K+ transport across the wall of synthetic surfactant vesicles doped with a crown ether ionophore (Hamilton and Kaler, 1987). The stability of our noncovalently assembled, biomimetic, metal-sorbing vesicles combined with their probable ease of recovery using standard hollow-fiber filtration membranes may make them useful for precious or toxic metal ion recovery from dilute solutions.
Acknowledgment This work was supported by the NSF (CTS-8910103 and the UCLA Engineering Research Center for Hazardous Substances Control, CDR-8622184), the California Department of Health Services (88-T0336), the Universit: 7 of California Toxic Substances Research and Teaching Program, the DOE Environmental Restoration and Waste Management Young Faculty Award Program administered by Oak Ridge Associated Universities, the DOE Innovative Concepts Program administered by Battelle Pacific Northwest Laboratories, and a RAND/UCLA fellowship for J.H.v.Z.
Literature Cited Blau, L.; Stern, R. B.; Bittman, R. The Stoichiometry of A23187and X537A-Mediated Calcium Ion Transport across Lipid Bilayers. Biochim. Biophys. Acta 1984, 778, 219-223. Borwankar, R. P.; Wasan, D. T.; Li, N. L. Liquid Membranes for Metal Ion Removal from Aqueous Media-An Overview. In Metals Speciation, Separation, and Recovery; Patterson, J. W., Passino, R., Eds.; Lewis Publishers: Chelsea, MI, 1987;pp 289-306. Caswell,A. H.; Pressman, B. C. Kinetics of Transport of Divalent Cations across Sarcoplasmic Reticulum Vesicles Induced by Ionophores. Biochem. Biophys. Res. Commun. 1972,49,292298.
Biotechnol. Prog., 1992, Vol. 8, No. 6
552
Chaberek, S.; Martell, A. E. Organic Sequestering Agents; John Wiley & Sons: New York, 1959. Chapman, C. J.; Puri, A. K.; Taylor, R. W.; Pfeiffer, D. R. Equilibria between Ionophore A23187 and Divalent Cations: Stability of 1:l Complexes in Solutions of 80% Methanol/ Water. Biochemistry 1987,26, 5009-5018. Chapman, C. J.; Puri, A. K.; Taylor, R. W.; Pfeiffer, D. R. General Features in the Stoichiometry and Stability of Ionophore A23187-Cation Complexes in Homogeneous Solution. Arch. Biochem. Biophys. 1990,281,44-57. Cussler, E. L. Membranes Which Pump. AIChE J. 1971, 17, 130-1303. Dawson, R. M. C.; Elliott, D. C.;Elliott, W. H.; Jones, K. M. Data for Biochemical Research, 3rd ed.; Clarendon Press: Oxford, 1986. Donnan, F. G. The Theory of Membrane Equilibria. Chem. Rev. 1926,1, 73-90. Fendler, J. H. Membrane Mimetic Chemistry; John Wiley: New York, 1982. Fyles, T. M. Biomimetic Ion Transport with Synthetic Transporters. Bioorg. Chem. Front. 1990,1, 71-113. Hamilton, R. T.; Kaler, E. W. Ionic Permeability of Synthetic Vesicles. J. Colloid Interface Sci. 1987, 116, 248-255. Hamilton, R. T.;Kaler, E. W. Alkali Metal Ion Transport through Thin Bilayers. J. Phys. Chem. 1990a, 94, 2560-2566. Hamilton, R. T.; Kaler, E. W. Facilitated Ion Transport through Thin Bilayers. J. Membr. Sei. 1990b, 54, 259-269. Hilgenfeld, R.; Saenger, W. Structural Chemistry of Natural and Synthetic Ionophores and their Complexeswith Cations. Top. Cum. Chem. 1982,101,l-82. Hunt, G. R. A.; Tipping, L. R. H.; Belmont, M. R. RateDetermining Processes in the Transport of P9+ Ions by the Ionophore A23187 across Phospholipid Vesicular Membranes. A 'H-NMR and Theoretical Study. Biophys. Chem. 1978,8, 341-355. Hwang, K. J. Modes of Interaction of (In3+)-8-Hydroxyquinoline with Membrane Bilayer. J. Nucl. Med. 1978,19,1162-1170. Izatt, R. M.; Lindh, G. C.; Clark, G. A.; Nakatsuji, Y.; Bradshaw, J. S.; Lamb, J. D.; Christensen, J. J. Proton-Ionizable Crown Compounds: 5. Macrocycle-MediatedProton-Coupled Transport of Alkali Metal Cations in HzO-CHzClS-HzO Liquid Membrane Systems. J . Membr. Sei. 1987, 31, 1-13. Johnson, S. M.; Bangham, A. D. Potassium Permeability of Single Compartment Liposomes with and Without Valinomycin. Biochim. Biophys. Acta 1969,193,82-91. Kafka, M. S.; Holz,R. W. Ionophores X537A andA23187: Effects on the Permeability of Lipid Bimolecular Membranes to Dopamine and Calcium. Biochim. Biophys. Acta 1976,426, 31-37. Kauffman, R. F.; Taylor, R. W.; Pfeiffer, D. R. Acid-Base Properties of Ionophore A23187 in Methanol-Water Solutions and Bound to Unilamellar Vesicles of Dimyristoylphosphatidylcholine. Biochemistry 1982,21, 2426-2435. Kolber, M. A.; Haynes, D. H. Fluorescence Study of the Divalent Cation-Transport Mechanism of Ionophore A23187 in Phospholipid Membranes. Biophys. J. 1981, 36, 369-391.
Lakshminarayanaiah, N. Equations of Membrane Biophysics; Academic: New York, 1984; pp 107-114. Lamb, J. D.; Izatt, R. M.; Garrick, D. G.; Bradshaw, J. S.; Christensen, J. J. The Influence of Macrocyclic Ligand Structure on Carrier-Facilitated Cation Transport Rates and SelectivitiesThrough Liquid Membranes. J.Membr. Sei. 1981, 9,83-107. Lonsdale, H. K. Membrane Research and Applications in the U.S. In Synthetic Membranes; Chenoweth, M. B., Ed.; MMI Press: Midland, MI, 1984; p 191. Marshall, L.; Parris, K.; Rebek, J. A New Class of Chelating Agents. Science 1988,110, 5192-5193. Mauk, M. R.; Gamble, R. C. Preparation of Lipid Vesicles Containing High Levels of Entrapped Radioactive Cations. Anal. Biochem. 1979,94, 302-307. Neidhardt, F. C.; Ingraham, J. L.; Schaechter, M. Physiology of the Bacterial Cell: A Molecular Approach; Sinauer Associates: Sunderland, MA, 1990. Noble, R. D.; Way, J. D. Applications in Liquid Membrane Technology. In Liquid Membranes: Theory and Applications; Noble, R. D., Way, J. D., Eds.; ACS Symposium Series No. 347; American Chemical Society: Washington, DC, 1987; pp 110-112.
Noble, R. D.; Koval, C. A.; Pellegrino, J. J. Facilitated Transport Membrane Systems. Chem. Eng. h o g . 1989,85 (3),58-70. Pohl, P.; Antonenko, Y. N.; Yaguzhinsky, L. S. Kinetic Properties of Cation/H+-Exchange: Calcimycin (A23187)-mediatedCa2+/ 2H+-Exchange on the Bilayer Lipid Membrane. Biochim. Biophys. Acta 1990,1027, 295-300. Pohl, W. G.; Kreikenbohm, R.; Seuwen, K. The Specificity of Ionophore A23187 in Cation Transport across Lipid Membranes: Studies with Lecithin Vesicles. 2.Naturforsch. 1980, 35c, 562-568. Reed, P. W.; Lardy, H. A. A23187: A Divalent Cation Ionophore. J. Biol. Chem. 1972,247, 6970-6977. Shastri, B. P.; Sankaram, M. B.; Easwaran, K. R. K. Carboxylic Ionophore (Lasalocid A and A231871 Mediated Lanthanide Ion Transport across Phospholipid Vesicles. Biochemistry 1987,26,4925-4930. Szoka, F.; Papahadjopoulos, D. Comparative Properties and Methods of Preparation of Lipid Vesicles. Annu. Rev. Biophys. Bioeng. 1980, 9,467-508. van Zanten, J. H.; Monbouquette, H. G. Characterization of Vesicles by Classical Light Scattering. J. Colloid Interface Sei. 1991, 146, 330-336. Westley, J. W. The Polyether Antibiotics: Naturally Occurring Acid Zonophores; Dekker: New York, 1982; Vol. 1. Wulf, J.;Pohl, W. G. CalciumIon Flux across Phosphatidylcholine Membranes Mediated by Ionophore A23187. Biochim. Biophys. Acta 1977,465,471-485. Accepted August 24, 1992. Registry No. A23187, 52665-69-7; Cd, 7440-43-9.
546
Biomimetic Metal-Sorbing Vesicles: Cd2+ Uptake by Phosphatidylcholine Vesicles Doped with Ionophore A23187 John H. van Zanten and Harold G. Monbouquette’ Chemical Engineering Department, University of California at Los Angeles, Los Angeles, California 90024-1592
Unilamellar phosphatidylcholine vesicles, harboring the ionophore, A23187, in the bilayer and the water-soluble chelating agent, nitrilotriacetate, in the vesicle interior, rapidly sequester and concentrate Cd2+from dilute aqueous solution. Metal-sorbing vesicle permeabilities for cadmium ion at 5 ppm (42.8 pM) ranged from 8.09 X lo-’ to 1.27 X cm/s for surface A23187 concentrations of 0.22-2.27 pmol/cm2 (which correspond to 1ipid:carrier molar ratios of 2000:l to 200:l) and pH’s from 5.5 to 8.5. The Cd2+ permeability shows linear variation with carrier concentration under the conditions studied. As pH is decreased, an increasing fraction of the A23187 becomes protonated, and the permeability exhibits a positive linear relationship with a function related to that for the fraction of unprotonated carrier. These noncovalently assembled, metalsorbing vesicles exhibit shelf lives of several months and remain stable throughout typical metal sorption studies.
Introduction Vesicles are closed-membrane capsules, typically of submicrometer diameters, that consist of single or multiple bilayers of noncovalently assembled surfactants. Since the basic structural unit of a cell membrane also is a surfactant bilayer, vesicles and cell membranes have much in common. In fact, liposomes, which are vesicles composed of natural phospholipid surfactants, have been used extensively as models for study of cell membrane structure and function (Fender, 1982;Johnson andBangham, 1969). Surfactant bilayer membranes not only define cell boundaries and compartmentalize functions within intracellular organelles but also provide the scaffolding for apparatuses involved in energy transduction, signal processing, and separations. The outstanding ability of organisms to sense particular chemical species in their environment, to perform efficient energy conversions, and to selectively sequester nutrients while retaining valuable intracellular metabolites has been well-documented (Neidhardt et al., 1990). Our effort has focused on the development of a vesicular membrane system that mimics both the structure and function of biological cell membranes for the uptake and concentration of heavy metal ions from dilute aqueous solution. The basic principles operable in cell membrane separations closely resemble those of supported or emulsion liquid membranes (Lonsdale, 1984;Borwankar et al., 1987; Noble and Way, 1987; Noble et al., 1989). The lipophilic hydrocarbon core of a surfactant bilayer membrane functions as a virtually impermeable barrier to charged and polar molecules (which are large relative to water). Transport of selected ionic or polar species across the membrane is mediated by membrane-bound carriers or transmembrane pores. Concentration of species on one side of a liquid membrane or within a cell is accomplished by coupling membrane transport with an energetically favorable process such as an opposing transmembrane flux of protons down a steep concentration gradient or a chemical conversion of the speciesupon transport (Cussler, 1971; Lonsdale, 1984; Neidhardt et al., 1990). Although vesicular membranes have much in common with well-studied liquid membranes, they may prove to 8756-7938/92/3008-0546$03.00/0
be more advantageous. Their relatively small size gives rise to a much larger available area for separations on a mass basis. For example, a 1% (vlv) dispersion of 100-nm vesicles provides 3000 m2 of vesicle surface area per liter and 40 mL of encapsulated volume. This high surface to volume ratio makes these vesicle systems ideal candidates for the recovery and concentration of species from dilute solution. The large contact area along with the exceedingly thin bilayer thickness (3-5 nm) makes solute sorption kinetics very rapid. In addition, unlike liquid membranes, vesicular membranes do not leach organic solvents into the aqueous phase. Finally, metal-sorbing vesicles, at about 100 nm in diameter, can be harvested readily for possible regeneration using hollow-fiber filtration cartridges. Other investigators have proposed the use of engineered surfactant vesicles for use in sequestering ions. Radioactive l111n3+has been concentrated in vesicles containing the chelator, nitrilotriacetate, for use in medical imaging (Hwang, 1978; Mauk and Gamble, 1979). Hamilton and Kaler (l987,1990a,b)theorized that vesicles could be used as separation media on the basis of their work on the permeability of synthetic surfactant vesicular membranes to Na+ and K+ in the presence and absence of synthetic, crown ether ionophores. Also, it has been shown that EDTA can be used to concentrate divalent cations in phosphatidylcholine vesicles (Pohl et al., 1980) and to increase the flux of Ca2+ across planar bilayer lipid membranes (Pohl et al., 1990). In this report, we describe the concept of metal-sorbing vesicles as agents for metal ion separations and present an illustration of the ability of an engineered vesicular membrane system to remove and concentrate Cd2+ions from dilute aqueous solution. We have used the naturally occurring membrane-forming lipid, phosphatidylcholine, to form highly stable vesicles using detergent dialysis. The gentle, “equilibrium” detergent dialysis approach also yields vesicles nearly uniform in size (van Zanten and Monbouquette, 19911, whereas sonication and extrusion methods give less stable vesicles with a relatively broad size distribution (Szoka and Papahadjopoulos, 1980). Vesicle-forming protocols based on detergent dialysis can
0 1992 American Chemical Soclety and American Institute of Chemical Englneers
BiotechnoL Rag.., 1992, Vol. 8, No. 6
Figure 1. Metal-sorbing vesicle structure. This drawing is not to scale. Typical vesicle diameters are on the order of 100 nm, while the soybean phosphatidylcholine bilayer thickness is approximately 3.7 nm.
be modified easily to permit incorporation of lipophilic structures in the vesicle wall and to facilitate encapsulation of hydrophilic species. We have incorporated the natural antibiotic, A23187, in the vesicle bilayer to serve as a shuttle for cadmium ions. A23187 (also known as calcimycin) is a highly selectivenatural ionophore for divalent cations (Chapman et al., 1990; Reed and Lardy, 1972). Many reports have appeared in the literature describing the effects of A23187 and other ionophores on biological systems, largely through their ability to effect Ca2+membrane transport (Westley, 1982). The logarithms of stability constants for 1:l complexes between A23187 and various cations in 80% methanol in water at basic pH have been reported to range from 9.77 to 2.36 (Chapman et al., 1990): Cu2+ >> H+, Ni2+,Fe2+,Co2+> Zn2+,Pb2+,Cd2+,Mn2+>> Mg2+,Ca2+ > Sr2+,Ba2+,Li+ > Na+. To provide a driving force for Cd2+uptake by the vesicles, we have encapsulated the water-solublemetal chelatingcompound, nitrilotriacetate, in the vesicle interior (see Figure 1). Our emphasis has been on the recovery of Cd2+from fairlydilute (5ppm and lower) solutionssince conventional methods for metal recovery tend to become impractical a t these concentrations. We have investigated the effect of pH and A23187 concentration on the sorption of Cd2+. We also have monitored the vesicles using static light scattering (van Zanten and Monbouquette, 1991) throughout the metal sorption process to assess the stability of the suspension when the vesicles are brought in contact with a solution bearing divalent metal cations. Materials and Methods Reagents and Solutions. L-a-Phosphatidylcholine from soybeans, sodium cholate, nitrilotriacetic acid (trisodium salt),and the buffers (sodiumsalts) HEPES, MES, and PIPES were purchased from Sigma (St. Louis, MO). Cadmium solutions were prepared from an Aldrich (Milwaukee, WI) cadmium atomic absorption standard solution (Cd in 1w t % HN03). Potassium nitrate (99.999%) also was obtained from Aldrich. Biological-gradesodium chloride and HPLC-grade methanol were purchased from Fisher (Tustin, CA). AntibioticA23187 was obtained from Calbiochem (San Diego, CA). All aqueous solutions were prepared with water purified by a MilliporeMilli-Qsystem. The MES, PIPES, and HEPES buffers were used as 25 mM solutionsadjusted to pH 5.5,7.0, and 8.5, respectively. Preparation of the Mixed Micellar Solution. A total of 225 mg of sodium cholate and 250 mg of L-a-phosphatidylcholine were dissolved in chloroform/methanol to give a solution with a lipid to detergent molar ratio of approximately 0.65. Antibiotic A23187 (ionophore) was
547
added to the solution to produce the desired ionophore: lipid ratio. This detergent, lipid, ionophore, and chloroform/methanol solution was then dried in a 50-mL roundbottomed flask at 40 "C under vacuum for 4 h using a rotary evaporator (Wheaton Instruments). The resultant film was resuspended in 12 mL of unbuffered 50 mM trisodium nitrilotriacetate (NTA) to yield a clear, aqueous solution containing free detergent, mixed lipid/detergent/ ionophore micelles, and NTA. Preparation of Vesicles by Detergent Dialysis. The mixed micellar solution was dialyzed against approximately 1L of unbuffered 50 mM trisodium NTA using a Liposomat (MM Developments,Ottawa, Ontario, Canada), a detergent dialysis device, with dialysis membranes (Diachema,MM Developments)having a molecular weight cutoff of 10 000. A near size monodisperse population of unilamellar vesicles is obtained which harbors A23187 in the lipid bilayer and 50 mM NTA (with 150 mM Na+) in the aqueous vesicle interior. Removal of the External Nitrilotriacetate. Unencapsulated NTA was separated from vesicles by passing the vesicle dispersion through a size exclusion column (1.5 cm X 20 cm) packed with Spectra/Gel AcA 202 (Spectrum Medical Industries, Los Angeles, CA). The mobile phase was unbuffered 150 mM NaCl (pH 5.0-5.1). This separation process yields a suspension of vesicles in 150 mM NaCl at a pH of 5.0-5.1. Determination of the Vesicle Weight Concentration. After the unencapsulated NTA was removed, aliquots of the resulting vesicles in NaCl solution were dried and weighed. The encapsulated NTA and external NaCl dry weights were subtracted from the total to give the lipid dry weight. The ionophore contribution to the dry weight was assumed to be negligible. Typically the concentration of lipid fell between 15 and 20 mg/mL. Size Characterization of the Vesicles. The vesicles were size characterizedby multiangle static light scattering using a Wyatt Technology (Santa Barbara, CA) laser photometer. Rayleigh-Gans-Debye theory and Zimm plots were used to estimate vesicle diameters and weightaveraged molecular weights from multiangle scattered light intensity measurements (van Zanten and Monbouquette, 1991). Given the information gained from analysisof light scattering data and the weight concentration of vesicles in suspension, the surface area and encapsulated volume of the suspension were calculated. Cd2+Uptake Measurements. Cadmium ion uptake by engineered vesicles from a buffered solution of known concentration was followed with a solid-state ion-selective electrode and a double-junction reference electrode containing gelled KC1 in the upper compartment and gelled KN03 in the lower compartment (Gam Rad West, San Juan Capistrano, CA). A 50-mL polypropylene reservoir was filled with approximately20 mL of a solution composed of 100 mM KN03,5 ppm Cd2+,and 25 mM Good's buffer. These buffers contribute 25 mM Na+ (HEPES and MES) or 37.5 mM Na+ (PIPES). A small amount of 1.5-2.0% (w/v)vesicle stock solution, typically on the order of a few hundred microliters, was pipeted into the magnetically stirred reservoir to initiate the Cd2+uptake process. An Orion 720A pH/ISE meter (Boston, MA) connected to a PC was used to capture data from the electrode every 5 s. Since the ion-selective electrode cannot detect metal ions sorbed by the vesicles, this technique allows direct, rapid measurement of the free cadmium ion concentration. Results Effect of pH on Cd2+Sorption Capacity. The effect of pH on the sorption and concentration of cadmium ions
Bbtechnol. Rog., 1902, Vol. 8,
548
F
X
0
Time (sec) Figure 2. Cadmium ion uptake by metal-sorbing vesicles. Dimensionless Cd2+concentration (relative to an initial concentrationof 5 ppm) is plotted versus time at pH 5.5 (A), 7.0 (+), and 8.5 (0). The phosphatidylcholine:A23187 molar ratio was 200:1,and the vesicle dispersionconcentration waa 0.03% (w/v). Also shown is ( 0 )the result from a control study with vesicles harboringencapsulated NTA but having no A23187 in the vesicle wall and (A) the response of the electrode to the addition of excess NTA directly to a 5 ppm solution of Cd2+.
in vesicles is illustrated in Figure 2. The cadmium ion concentration presented is normalized with respect to the initial concentration of 5 ppm. The pH dependence of Cd2+ uptake is shown for three pH's, 5.5, 7.0, and 8.5, which correspond to the middle three of the five curves exhibited. The lower curve is the electrode response to sudden complexation of Cd2+by addition of excess NTA directly to the reservoir. The rapid electrode response indicates that electrode kinetics and reservoir mixing times do not need to be taken into account in estimating metal sorption parameters from the raw data. The uppermost curve corresponds to a control study where vesicles harboring NTA without any carrier in the vesicle wall were used. Imperceptible metal ion uptake for this control over the time period studied demonstrates that the disappearance of metal ion observed when A23187 is present is due to sorption by the engineered vesicles (rather than complexation with leaked NTA). Since pH influences both kinetics and capacity significantly, it appears that both the carrier and chelator are affected a t the lower p H s studied. The influence of pH on the initial rate of Cd2+uptake, which is due primarily to protonation of the carrier, is discussed later. On the basis of the three pK:s for NTA, 1.9,2.5,and 9.8 (Chaberek and Martell, 1959; Dawson et al., 1986) (defined as -log K1, -log K2, and -log K3, respectively), the ratio of total NTA to total cadmium in the reservoir (1.07 for all cases), and the initial Cd2+ concentration, the free metal ion concentration at equilibrium can be estimated using the equation (Chaberek and Martell, 1959) I
\
where
I , , , , I , , , ,
0 0.5 1 1.5 2 2.5 A23187 Surface Concentration (pmol/cm2)
Figure 3. Initial vesicle permeability to Cd2+versua A23187 vesicle surface concentration in pmol/cm2at pH 5.5 (O),7.0 (A), and 8.5 (m). The four A23187 concentrations shown correspond to phosphatidylcholine:A23187 molar ratios of 2000.1, 1ooO:1, 500:1,and 200:l and total A23187 concentrations (based on the reservoir volume) of 0.32,0.59, 1.07, and 1.97 rM,respectively.
[NTAI) does not interfere with Cd2+chelation. According to eq 1, 78.6%, 97.3%, and 99.9% of the free cadmium should be complexed a t equilibrium at pH 5.5, 7.0, and 8.5, respectively, by NTA in free solution. However, our experimental results with encapsulated NTA where 92 % and 98% of the free Cd2+is sorbed by the vesicles at pH 7.0 and 8.5, respectively, suggest that Donnan effects may influence the capacity of the system since the phosphatidylcholine bilayer essentially is impermeable to all anions present and there is an unequal distribution of ionic species across the membrane (Donnan, 1925; Lakshminarayanaiah, 1984). At pH 5.5, the sorption kinetics is slowed considerably as only 40 % of the free Cd2+is removed in 10 min. Unlike the results a t pH 7.0 and 8.5, the pH 5.5 Cd2+uptake curve exhibits a significant negative slope at 10 min, which indicates that the system still is far from equilibrium. Nevertheless, one should note the rapid overall rate at which cadmium ions were removed from the reservoir and concentrated in the vesicle interior by these engineered biomimetic vesicle systems, at neutral to basic pH. The impressive kinetics was achieved with a vesicular dispersion concentration of only 0.03 w t % ,which corresponds to an encapsulated volume of only 18.6 pL in the 20-mL reservoir. The cadmium was concentrated inside the vesicles by factors of 1040,990, and 430 at the pH's 8.5, 7.0, and 5.5, respectively, over the time period shown. The totalcarrier concentration (based on the reservoir volume) was only 1.97 pM, while the total cadmium ion concentration in the reservoir was 42.8 pM. Thus, the reduction in the concentration of free cadmium ions must be due to their being sequestered by the encapsulated chelating agent, nitrilotriacetate. The rapid system response is due largely to the exceedingly small size of these vesicles, at 84 nm in diameter, and the 17 600 cm2of surface area for separations that they provide in each of the cases shown. Effect of Carrier Concentration on Initial Permeability. The effect of the A23187 carrier concentration on the initial cadmium ion permeability is shown in Figure 3. The permeability was found by fitting the initial points in the free Cd2+concentration traces, where the decrease was exponential with time, to the following equation: [Cd2+l - exp(-PARt) --
and KMAis the binding constant of NTA for Cd2+. This expression includes the assumptions that NTA and Cd2+ form only 1:l complexes and that Na+ (at 3:l [Na+l:
No. 6
(2) [Cd2+lO Here, [Cd2+lorepresents the initial reservoir concentration, P denotes the initial vesicle permeability, and AR is the vesicle area divided by the reservoir volume. By using
649
Biotechnol. Prog., 1992, Vol. 8, No. 6
Table I. A23187 Turnover Numbers for Cd2+Transport (E-')
uH 8.5
7.0 5.5
2000:l 4.3 2.0 0.16
lipid:A23187 molar ratio 1ooo:l 5m1 4.1 2.7 1.5 2.1 0.14 0.14
1.4 1.2
2m1 2.2 1.2
0.17
this expression, we are assuming that the cadmium concentration in the vesicle interior is essentially zero for small t and that the metal ion flux is a linear function of Cd2+concentration at the low concentration of 5 ppm (42.8 pM). As a point of reference, others have shown that the A23187-mediated influx of Ca2+into phosphatidylcholine vesicles is not saturated and is linearly dependent on calcium ion a t concentrations of 3-18 mM (Blau et al., 1984). The permeabilities reported here, ranging from to 1.27 X cmls, are in the expected range 8.09 X based on published Ca2+permeabilities for planar phosphatidylcholine bilayers (Kafka and Holz, 1976;Wulf and Pohl, 1977) and given that Cd2+ transport into phosphatidylcholine vesicles containing EDTA has been found to be 3-fold greater than that of Ca2+(Pohl et al., 1980). A t 5 ppm Cd2+and over the narrow range of carrier concentration studied, P appears to vary linearly with A23187 concentration at a particular pH. However, the permeabilities a t neutral to basic pH may be systematically underestimated due to (1)the limited number of data points gathered early in the sorption run at 5-s intervals; (2) a fraction of a second delay in starting to record the free Cd2+concentration depletion after injection of vesicles at the initiation of the metal sorption process; and (3) loss of up to 25% of the A23187 into the reservoir at pH 8.5 as a consequence of its slight solubility in water. The apparent membrane association constants for A23187, defined as [A23187]mem/([A23187] [PL]), where [A23187]mem is the membrane-bound A23187 concentration and [PL] is the phospholipid concentration, have been measured as 1.9 X lo4 and 7 X lo3 M-l at pH 7.06 and 8.7, respectively (Kolber and Haynes, 1981). Thus, the greatest loss of carrier would be expected at higher pH. (Since the A23187 in the aqueous phase still participates in metal ion transport, through binding to Cd2+ in the aqueous phase and subsequent re-incorporation into the bilayer to facilitate ion transport, we include all A23187 in the calculation of the surface carrier concentrations presented in Figure 3. The corresponding lipid to carrier and total A23187 concentrations, based on the reservoir volume, also are given in the Figure 3 caption.) These systematic errors could be substantial enough to give permeabilities for the higher carrier concentrations at pH 7.0 and 8.5 that would fall on a line through the origin. A23187 turnover numbers, calculated as P[Cd2+1ol [A23187ltotal, provide additional evidence that the permeabilities at higher pH may be underestimated (seeTable I). Since the system does not show saturation behavior with respect to A23187 concentration and the formation of 1:l carrier-ligand complexes is expected to be limiting over the carrier concentration range studied (Kolber and Haynes, 19811, the turnover numbers should remain constant at a given pH. At pH 5.5, the turnover numbers are roughly constant. However, at pH 7.0 and 8.5 where Cd2+uptake is more rapid and the systematic underestimation of the permeability would be expected to be more evident, the turnover number declines with increasing carrier concentration. In agreement with our higher turnover number at 20001 lipidcarrier and pH 8.5, others have estimated a turnover number for A23187 transporting Ca2+at pH 8.0 of about 5 s-l (Caswell and Pressman, 1972;
1 0.8 0.6
0.4 0.2
0
o
0.2
0.6 f(PH)
0.4
0.8
1
Figure 4. Initial vesicle permeability to Cd2+versus f(pH) = 1/[1+10(6.*pH)] at phosphatidylcholine:A23187 molar ratios of 2000:l (+),1ooO:1(~),500:1(~),and200:1 (O),whichcorrespond to surface concentrations of 0.22,0.40,0.86, and 2.27 pmol/cm2, respectively.
Wulf and Pohl, 1977). This data suggests that, at pH 8.5 and a lipidcarrier ratio of 200:1, the permeability, and hence the turnover number, may be underestimated by nearly a factor of 2. Nevertheless, the permeability data presented here, along with reasonable estimates of random and systematic error, could not be used to support an argument that transport of Cd2+is limited by formation of a neutral 2:l (carrier:metal ion) complex; the data simply are not nonlinear enough. The observed relationship between permeability and carrier concentration more likely indicates that the formation of a 1:l A23187-cadmium ion complex or Na+ countertransport from the vesicle interior is limiting the uptake of Cd2+under these experimental conditions. At the reservoir pH of 8.5 when the vesicle is at a lower pH of about 5.0, the rapid efflux of free protons from the unbuffered vesicle interior can account only for an exceedingly small exchange of Cd2+ions with the vesicle exterior. Therefore even at this pH, the efflux of Na+, which binds 4 orders of magnitude less strongly with A23187 (Chapman et al., 19901, may be limiting. Effect of pH on Initial Permeability. As is evident from Figures 2 and 3, the rate of cadmium sorption is reduced at lower pH's, as expected given the reported pKa for A23187 in phosphatidylcholine bilayers of 7.85 (Kauffman et al., 1982). The data are correlated linearly at a given carrier concentration (see Figure 4) by the following function of pH,
which is a function closely related (i.e., the constant, 6.9, substituted for pK,) to that for the fraction of unprotonated A23187 in solution in the absence of other binding cations. This strong correlation is intriguing given that the relevant quantity would seem to be the fraction of vesicle-bound A23187 that is free for chelation of Cd2+. For these experimental conditions, such a quantity must take into account the fraction of carrier that is bound to Na+, K+, and Cd2+as well as protons. Unfortunately, the fraction of vesicle-bound A23187 could not be calculated, as association constant data is available only or 1:lA23187Cd2+complexes in methanol/water (Chapman et al., 1990) and no data exists for K+ and for 2:l A23187-Cd2+ complexes. The substantial initial gradient in proton concentration across the vesicle wall at reservoir pH's of 5.5 and 8.5 does not further complicatethis analysis, since pH equilibration across the vesicle wall likely is rapid. A23187 shows at
550
.
2
Biotechnol. Rog., 1992, Vol. 8, No. 6 1.15
TI
E 0
-.-
1.10
Q)
V
m
1.05
m
-c
I-
t
iTT
Q)
.-0m 1 . 0 0 0
E
50.950
!i
p
' ' '
0
"
' ' ' '
50
I ' ' ' ' I ' '
100
150
I
' '
200
*
250
Time (min)
Figure 5. Metal-sorbing vesicle size during the sorption of Cd2+ at pH 8.5. Dimensionless vesicle diameter (relative to initial diameter), as determined by applying Rayleigh-Gans-Debye theory to static light scattering data, is plotted versus sorption time in minutes.
least an order of magnitude higher affinity for protons than for any of the other cations present (Chapman et al., 1990), and given the small vesicle-encapsulated volume, an exceedingly small number of protons needs to be transported for pH equilibration. In addition, it is known that the rate constant for translocation of the 1:l H+A23187 complex across a lipid bilayer is 3 orders of magnitude higher than that for 2 1 A23187-M2+ complexes (Kolber and Haynes, 1981). This comparison serves as additional indirect evidence that pH equilibration under these conditions is rapid relative to Cd2+ transport and Na+ countertransport. Metal-Sorbing Vesicle Stability. These metal-sorbing vesicles are stable for several months on the shelf. In addition, exposure for 1day to a temperature of 90 "C or to 1N acid or base causes no deterioration in stability of the suspension. Vesicle dispersion stability is indicated by no evidence of visual aggregates or precipitates and by the continued ability of the suspension to scatter light as size-homogeneous, unilamellar vesicles. The metal-sorbing vesicles appeared to be stable throughout the sorption process. Typical vesicle diameter measurements based on light scattering data are shown in Figure 5 for the pH 8.5 case over the first 50 min of a sorption run. Throughout this time period, the vesicles scattered light as isotropic hollow spheres within the Rayleigh-Gans-Debye approximation (van Zanten and Monbouquette, 1991)with no visual evidence of aggregates or precipitates. This indicates that the vesicles retain their spherical, capsular geometry during the metal sorption process. Discussion In the accepted models for A23187-mediated transport of divalent cations (Caswell and Pressman, 1972; Kolber and Haynes, 1981; Chapman et al., 1987), an A23187 molecule first binds a single cation at the membrane/ aqueous solution interface. Subsequently, a second A23187 combines with the 1:l complex to form a neutral 2:l carrier-cation complex that can translocate to the opposite membrane interface where the complexation process may reverse. Using fluorescence techniques, Kolber and Haynes (1981) obtained convincing evidence that divalent cations are transported across lipid bilayer membranes predominantly in 2: 1neutral complexes. This observation does not conflict with other published reports indicating that the permeability of lipid bilayers doped with A23187 varies linearly with A23187 concentration (Hunt et al., 1978; Kafka and Holz, 1976; Wulf and Pohl,
1977) or varies with the square of A23187 concentration (Blau et al., 1984; Pohl et al., 19801, as any step in the complexation and translocation process can be limiting (Caswell and Pressman, 1972; Kolber and Haynes, 1981). Our study was conducted at total carrier concentrations (based on the reservoir volume) above 0.1 pM where a linear relationship would be expected between permeability and A23187 concentration (Kolber and Haynes, 1981). There are difficultiesin comparing our results with those in the literature, however. Although a number of studies have appeared that discuss aspects of A23187-mediated transport of multivalent cations across the walls of lipid bilayervesicles (Blau et al., 1984;Hunt et al., 1978;Kolber and Haynes, 1981;Pohl et al., 1980;Pohl et al., 1990;Shastri et al., 19871, initial permeabilities of the type reported here have never been published previously for vesicles. Ion transport data typically were gathered over at least severalminutes for the calculation of time-averaged kinetic properties, and the influence of a changing transmembrane cation concentration gradient or electropotential could not always be ruled out. In the two studies most closely related to our work, where metal ion uptake by A23187doped vesicles was inferred from an efflux of protons (Pohl et al., 1980) and where metal ion efflux from A23187doped vesicles was mirrored using a change in the absorbance of a metal-complexing dye (Blau et al., 1984; Pohl et al., 19801, insufficient information was presented to calculate initial permeabilities. These permeability calculationsrequire knowledge of vesicle surface area and encapsulated volume as well as trustworthy data on the initial uptake or release of metal ions from vesicles. Vesicle size information is frequently absent from these studies. Further, the method used to follow metal ion concentration often is not characterized fully in terms of the response time or the relationship between the metal ion concentration and the measured quantity, e.g., metal ion/Arsenazo I11 dye complex absorbance (Blau et al., 1984). A popular approach has been to analyze ion leakage from vesicles separated from a surrounding buffered reservoir by a dialysis bag (Johnson and Bangham, 1969; Hamilton and Kaler, 1987). Using this method, Hamilton and Kaler (1987)have reported Na+and K+permeabilities ranging from 5.1 X 10-locm/s for Na+ and a surfactant: metal carrier ratio of 1OOO:l to 2.75 X 10+ cm/s for K+ and a surfactant:metal carrier ratio of 1OO:l for vesicles composed of a synthetic surfactant doped with a crown ether ionophore. In order to conduct these experiments, the vesicles first must be loaded with metal ions and then passed through a desalting gel column to remove residual metal ions external to the vesicles before the vesicles are charged to the dialysis bag. Due to the cumbersome, timeconsuming nature of this process, the first minute or so of metal ion leakage from the vesicles, which is critical for the determination of initial permeabilities, may be missed. The dialysisbag also complicates the analysisby presenting another mass transfer resistance that can obscure rapid ion leakage kinetics. These comments may partially explain, in addition to the differences in the chemical composition of the systems, the difference of several orders of magnitude between the vesicle permeabilities reported here and those measured by Hamilton and Kaler (1987) for their system. We report P as an initial permeability in this study because the simple exponential expression given by eq 2 cannot be used to fit sorption data for later times, despite the fact that at neutral to basic pH's encapsulated NTA effectively maintains the free cadmium ion concentration
551
Biotechnol. Prog.. 1992, Vol. 8, NO. 6
in the vesicle interior a t or near zero for much of the sorption process. The slower efflux of Naf from the vesicles may limit cadmium ion transport at intermediate times when the internal sodium ion concentration is reduced from its relatively high initial concentration of 150mM. At relatively large t, when the encapsulated NTA approaches saturation with cadmium, a reduced chelation rate likely influences the overall kinetics as well. An effort currently is being made to include these effects in a model of metal sorption by these vesicles. On the basis of the data shown, the rapid initial kinetics as well as the capacity of this system could be improved further by increasing the vesicle concentration by 4 orders of magnitude to about 10% (w/v), which would give an encapsulated volume of about 50%. We have not worked a t higher vesicle loadings because the very rapid kinetics of metal uptake would then be obscured by the slower electrode response and we could not obtain useful data on vesicle permeability. However, we expect that, in addition to increasing the vesicle suspension concentration to about 10% (w/v), we can, if desired, increase the internal NTA concentration to 0.5 M and improve the A23187 to phosphatidylcholine mole ratio to about 150. These changes would increase by orders of magnitude both the metal loading capacity and the kinetics of metal uptake. Since the driving force for the separation is the encapsulated chelating agent and NTA has a very high binding constant, metal ion concentrations in the surrounding solution can be driven very low. As a point of reference, the NTA binding constant for Cd2+is 6.3 X lo9 (Chaberek and Martell, 1959). Therefore at equilibrium with sufficient NTA, the concentration of cadmium in the chelated state is many orders of magnitude higher than that of the free ion. In addition, since transport of divalent cations by A23187 occurs by way of a 2:l A23187:M2+ neutral complex, a diffusion potential does not arise to inhibit metal ion uptake by our metal-sorbing vesicles. Obviously this system can be used to drive metal ion concentrations in a waste stream to extremely low levels, or it can be exploited for the harvesting of metal from very dilute solutions. When a chelating agent such as NTA is used to provide the driving force for metal ion uptake by ionophore-doped vesicles, the overall selectivity of the system is dependent on the selectivity profile of both the carrier and chelator. Since the chelator provides the driving force for metal uptake by depleting the free metal ion concentration in the vesicle interior, it provides a greater driving force for those ions for which it has the greatest affinity. If the metal carrier in the vesicle wall is characterized by a high affinity for the same ion(s) as the chelator, the overall metal ion selectivity of the vesicle system will be greater than that of the carrier or chelator alone. In a crude sense, if the probability that both carrier and chelator will bind an undesired cation over the target metal ion is one out of a hundred (and these are roughly independent events), then the overall probability that the vesicle will sequester the undesired cation would be on the order of one in ten thousand. In such a way, these vesicular metal-sequestering media can be engineered for outstanding selectivity. Although the selectivity of metal-sorbing vesicles has not been demonstrated as yet, the versatility in the design of these systems both a t the level of the membrane-bound carrier and the encapsulated water-soluble chelator provides for this potential. A large number of natural and synthetic carriers have been identified with various selectivityprofiles that would be suitable for use in our system (Lamb et al., 1981;
Hilgenfeld and Saenger, 1982; Izatt et al., 1987; Noble et al., 1989; Fyles, 1990). Relatively expensive carrier chemistries could be used in our system since the capacity for metal ion uptake is not dependent on the carrier concentration. The concentration of internal water-soluble chelating agent provides the loading capacity, and chelators, such as NTA ($22.85/kg, Sigma, 19921,are relatively inexpensive. Charged carriers such as A23187 also can be used in metal recovery schemes that do not depend on encapsulated chelating agents, but utilize an opposing pH gradient to supply the energy to concentrate metal ions within the capsular membrane. Additional charged carriers that would be suitable include the Henkel L E carriers (Lonsdale, 1984), which have proven useful for copper, nickel, and zinc recovery, the compounds of Marshall et al. (1988), and macrocycles derivatized with proton ionizable groups (Izatt et al., 1987; Fyles, 1990). Useful carriers not only must participate in cationtmetal ion exchange reactions and exhibit selectivity but also must be sufficiently lipophilic to remain immobilized in the vesicle wall. A number of practical considerations ultimately will determine the utility of metal-sorbing vesicles including the development of truly selective systems with adequate shelf life and physicochemical stability. Issues such as these related to technologicalfeasibilitycurrently are under intensive investigation.
Conclusions Cadmium ions in dilute aqueous solution can be sequestered and concentrated several hundred-fold in a few minutes time by metal-sorbing vesicles. As expected, the vesicle permeability increases with carrier concentration but declines with pH. The actual permeabilities measured, 8.09 X lO-'to 1.27 X lo4 cm/s, are several orders of magnitude higher than those estimated for Na+ and K+ transport across the wall of synthetic surfactant vesicles doped with a crown ether ionophore (Hamilton and Kaler, 1987). The stability of our noncovalently assembled, biomimetic, metal-sorbing vesicles combined with their probable ease of recovery using standard hollow-fiber filtration membranes may make them useful for precious or toxic metal ion recovery from dilute solutions.
Acknowledgment This work was supported by the NSF (CTS-8910103 and the UCLA Engineering Research Center for Hazardous Substances Control, CDR-8622184), the California Department of Health Services (88-T0336), the Universit: 7 of California Toxic Substances Research and Teaching Program, the DOE Environmental Restoration and Waste Management Young Faculty Award Program administered by Oak Ridge Associated Universities, the DOE Innovative Concepts Program administered by Battelle Pacific Northwest Laboratories, and a RAND/UCLA fellowship for J.H.v.Z.
Literature Cited Blau, L.; Stern, R. B.; Bittman, R. The Stoichiometry of A23187and X537A-Mediated Calcium Ion Transport across Lipid Bilayers. Biochim. Biophys. Acta 1984, 778, 219-223. Borwankar, R. P.; Wasan, D. T.; Li, N. L. Liquid Membranes for Metal Ion Removal from Aqueous Media-An Overview. In Metals Speciation, Separation, and Recovery; Patterson, J. W., Passino, R., Eds.; Lewis Publishers: Chelsea, MI, 1987;pp 289-306. Caswell,A. H.; Pressman, B. C. Kinetics of Transport of Divalent Cations across Sarcoplasmic Reticulum Vesicles Induced by Ionophores. Biochem. Biophys. Res. Commun. 1972,49,292298.
Biotechnol. Prog., 1992, Vol. 8, No. 6
552
Chaberek, S.; Martell, A. E. Organic Sequestering Agents; John Wiley & Sons: New York, 1959. Chapman, C. J.; Puri, A. K.; Taylor, R. W.; Pfeiffer, D. R. Equilibria between Ionophore A23187 and Divalent Cations: Stability of 1:l Complexes in Solutions of 80% Methanol/ Water. Biochemistry 1987,26, 5009-5018. Chapman, C. J.; Puri, A. K.; Taylor, R. W.; Pfeiffer, D. R. General Features in the Stoichiometry and Stability of Ionophore A23187-Cation Complexes in Homogeneous Solution. Arch. Biochem. Biophys. 1990,281,44-57. Cussler, E. L. Membranes Which Pump. AIChE J. 1971, 17, 130-1303. Dawson, R. M. C.; Elliott, D. C.;Elliott, W. H.; Jones, K. M. Data for Biochemical Research, 3rd ed.; Clarendon Press: Oxford, 1986. Donnan, F. G. The Theory of Membrane Equilibria. Chem. Rev. 1926,1, 73-90. Fendler, J. H. Membrane Mimetic Chemistry; John Wiley: New York, 1982. Fyles, T. M. Biomimetic Ion Transport with Synthetic Transporters. Bioorg. Chem. Front. 1990,1, 71-113. Hamilton, R. T.; Kaler, E. W. Ionic Permeability of Synthetic Vesicles. J. Colloid Interface Sci. 1987, 116, 248-255. Hamilton, R. T.;Kaler, E. W. Alkali Metal Ion Transport through Thin Bilayers. J. Phys. Chem. 1990a, 94, 2560-2566. Hamilton, R. T.; Kaler, E. W. Facilitated Ion Transport through Thin Bilayers. J. Membr. Sei. 1990b, 54, 259-269. Hilgenfeld, R.; Saenger, W. Structural Chemistry of Natural and Synthetic Ionophores and their Complexeswith Cations. Top. Cum. Chem. 1982,101,l-82. Hunt, G. R. A.; Tipping, L. R. H.; Belmont, M. R. RateDetermining Processes in the Transport of P9+ Ions by the Ionophore A23187 across Phospholipid Vesicular Membranes. A 'H-NMR and Theoretical Study. Biophys. Chem. 1978,8, 341-355. Hwang, K. J. Modes of Interaction of (In3+)-8-Hydroxyquinoline with Membrane Bilayer. J. Nucl. Med. 1978,19,1162-1170. Izatt, R. M.; Lindh, G. C.; Clark, G. A.; Nakatsuji, Y.; Bradshaw, J. S.; Lamb, J. D.; Christensen, J. J. Proton-Ionizable Crown Compounds: 5. Macrocycle-MediatedProton-Coupled Transport of Alkali Metal Cations in HzO-CHzClS-HzO Liquid Membrane Systems. J . Membr. Sei. 1987, 31, 1-13. Johnson, S. M.; Bangham, A. D. Potassium Permeability of Single Compartment Liposomes with and Without Valinomycin. Biochim. Biophys. Acta 1969,193,82-91. Kafka, M. S.; Holz,R. W. Ionophores X537A andA23187: Effects on the Permeability of Lipid Bimolecular Membranes to Dopamine and Calcium. Biochim. Biophys. Acta 1976,426, 31-37. Kauffman, R. F.; Taylor, R. W.; Pfeiffer, D. R. Acid-Base Properties of Ionophore A23187 in Methanol-Water Solutions and Bound to Unilamellar Vesicles of Dimyristoylphosphatidylcholine. Biochemistry 1982,21, 2426-2435. Kolber, M. A.; Haynes, D. H. Fluorescence Study of the Divalent Cation-Transport Mechanism of Ionophore A23187 in Phospholipid Membranes. Biophys. J. 1981, 36, 369-391.
Lakshminarayanaiah, N. Equations of Membrane Biophysics; Academic: New York, 1984; pp 107-114. Lamb, J. D.; Izatt, R. M.; Garrick, D. G.; Bradshaw, J. S.; Christensen, J. J. The Influence of Macrocyclic Ligand Structure on Carrier-Facilitated Cation Transport Rates and SelectivitiesThrough Liquid Membranes. J.Membr. Sei. 1981, 9,83-107. Lonsdale, H. K. Membrane Research and Applications in the U.S. In Synthetic Membranes; Chenoweth, M. B., Ed.; MMI Press: Midland, MI, 1984; p 191. Marshall, L.; Parris, K.; Rebek, J. A New Class of Chelating Agents. Science 1988,110, 5192-5193. Mauk, M. R.; Gamble, R. C. Preparation of Lipid Vesicles Containing High Levels of Entrapped Radioactive Cations. Anal. Biochem. 1979,94, 302-307. Neidhardt, F. C.; Ingraham, J. L.; Schaechter, M. Physiology of the Bacterial Cell: A Molecular Approach; Sinauer Associates: Sunderland, MA, 1990. Noble, R. D.; Way, J. D. Applications in Liquid Membrane Technology. In Liquid Membranes: Theory and Applications; Noble, R. D., Way, J. D., Eds.; ACS Symposium Series No. 347; American Chemical Society: Washington, DC, 1987; pp 110-112.
Noble, R. D.; Koval, C. A.; Pellegrino, J. J. Facilitated Transport Membrane Systems. Chem. Eng. h o g . 1989,85 (3),58-70. Pohl, P.; Antonenko, Y. N.; Yaguzhinsky, L. S. Kinetic Properties of Cation/H+-Exchange: Calcimycin (A23187)-mediatedCa2+/ 2H+-Exchange on the Bilayer Lipid Membrane. Biochim. Biophys. Acta 1990,1027, 295-300. Pohl, W. G.; Kreikenbohm, R.; Seuwen, K. The Specificity of Ionophore A23187 in Cation Transport across Lipid Membranes: Studies with Lecithin Vesicles. 2.Naturforsch. 1980, 35c, 562-568. Reed, P. W.; Lardy, H. A. A23187: A Divalent Cation Ionophore. J. Biol. Chem. 1972,247, 6970-6977. Shastri, B. P.; Sankaram, M. B.; Easwaran, K. R. K. Carboxylic Ionophore (Lasalocid A and A231871 Mediated Lanthanide Ion Transport across Phospholipid Vesicles. Biochemistry 1987,26,4925-4930. Szoka, F.; Papahadjopoulos, D. Comparative Properties and Methods of Preparation of Lipid Vesicles. Annu. Rev. Biophys. Bioeng. 1980, 9,467-508. van Zanten, J. H.; Monbouquette, H. G. Characterization of Vesicles by Classical Light Scattering. J. Colloid Interface Sei. 1991, 146, 330-336. Westley, J. W. The Polyether Antibiotics: Naturally Occurring Acid Zonophores; Dekker: New York, 1982; Vol. 1. Wulf, J.;Pohl, W. G. CalciumIon Flux across Phosphatidylcholine Membranes Mediated by Ionophore A23187. Biochim. Biophys. Acta 1977,465,471-485. Accepted August 24, 1992. Registry No. A23187, 52665-69-7; Cd, 7440-43-9.
