J. MATER. CHEM., 1991, 1(3), 365-370 Preparation of Ternary Composite Hydrogels of Agarose, Concanavalin A and a Glycolipid Monolayer, and their Permeation Properties Nobuyuki Higashi, Minoru Takematsu and Masazo Niwa* Department of Applied Chemistrx Faculty of Engineering, Doshisha University, Kamikyo-ku, Kyoto 602, Japan Synthetic glycolipids (2Cl,-de-glu and 2C,,-de-gal) form condensed surface monolayers, which have been successfully deposited on the agar hydrogel by the Langmuir-Blodgett (LB) technique. The monolayer-deposited gel membranes were satisfactory for use in the permeation experiments of ammonium salts such as tetramethyl- ammonium bromide or hexyltrimethylammonium bromide over a wide temperature range. The permeability of the salts through the modified gel membranes depends strongly upon the hydrophobicity of the permeants and could be controlled by the phase transition of the deposited monolayers.Concanavalin A (ConA)-bound agar gels were modified with the 2Cl,-de-glu monolayer, having the a-D-glucopyranosyl head group, but not with the 2C,,-de-gal, having the B-D-galactopyranosyl head group, owing to the specific binding ability of ConA. The permeability of the 2Cl,-de-glu monolayer-deposited ConA-bound agar gel was selectively controlled by the addition of monosaccharides: the permeability was enhanced by the addition of D-mannose, whereas it was hardly changed by the addition of D-galactose, resulting from the high binding constant of D-mannose with ConA. Keywords: Glycolipid; Agar hydrogel; Monolayer; Langmuir-Blodgett film; Permeation The permeation of a hydrogel by a water-soluble solute has received a great deal of attention from the standpoint of a controlled-release system of hydrophilic drugs, owing to the fact that hydrogels provide high biocompatibility based on their large surface free energy.' It is well known that an artificial lipid membrane has a variety of peculiar physico- chemical properties (e.g.crystal-liquid crystal phase transition and phase separation) which can control the substrate per- meability through the membrane in a similar way to biomem- brane function.2 Thus, the combined study of hydrogels and lipid membranes should provide a new promising area of permeability-controllable membranes. The LB technique3 transferring lipid monolayers from an air/water interface is useful for preparing oriented thin films on a substrate.LB multilayer films supported on a porous substrate have been used recently to control the permeability of and gases.g-" The LB deposition of a phospholipid bilayer on a poly(acry1amide) hydrogel has been reported by Arya et al.12 However, in the paper they mention that the experiment has not been successfully repeated. Hongyo et ~1.'~showed the applicability of a phospholipid black lipid membrane (BLM) painted onto an agar hydrogel for conductometric devices. In a previous report,I4 we presented preliminary results of the preparation of a glycolipid monolayer-coated agar hydro- gel and of the substrate permeability through the modified hydrogel membranes.The present paper describes the results of our more intensive study of the substrate permeability through the monolayer-coated hydrogel membrane. Also, we report on the preparation of a novel ternary composite hydrogel membrane composed of agar, ConA and a glycolipid monolayer, and its controlled permeability upon addition of monosaccharides. ConA, isolated from Canaualia ensifomis, is a lectin, and a tetrameric protein with four carbohydrate- binding sites which specifically binds a-D-glucopyranosyl or a-D-mannopyranosyl moieties. ConA has been widely used for model studies of the molecular recognition on cell surfaces; liposomes are agglutinated by addition of ConA if they contain the glycolipid with the a-D-glucopyranosyl moi- ety.15-" These agglutinations are reversed by the addition of an excess of low-molecular-weight saccharides.We employed such glycolipids as 2C18-de-glu or 2Cls-de-gd having the a-D-glucopyranosyl or /?-D-galactopyranosyl unit as hydro- philic head group, respectively, as monolayer components. A schematic illustration of the composite hydrogel membranes is shown in Fig. 1 in which the permeants are also included. Experimental Materials Glycolipid 2C18-de-glu containing a-D-glucopyranosyl-D-gluconamide as a hydrophilic head group was prepared as follows. 3-Azapentamethylene bis(octadecan0ate) was syn-thesized by condensation of octadecanoic acid (14.7 g, 52 mmol) with bis(2-hydroxyethy1)amine(2.4 g, 22 mmol) in toluene in the presence of toluene-p-sulphonic acid (10.7 g, 56 mmol): pale-yellow powder (8.0 g, 56%), m.p.57 "C from acetone. Maltose [O-a-D-ghcopyranosyl-(1-+4)-~-glucopy-ranose] (12.0 g, 33 mmol) was oxidized in methanol in the presence of 17.1 g of iodine (67 mmol) to maltonolactone (8.2 g, 73 YO)according to the literature.20 Maltonolactone (1.7 g, 5 mmol) was allowed to react with 4.1 g of 3-azapenta- methylene bis(octadecan0ate) (5 mmol) in refluxing methanol for 4 h. After cooling the mixture to room temperature, a pale-yellow powder precipitated. The powder was filtered and recrystallized from methanol (2.5 g, 52%), m.p. 73.5 "C (Found: C, 63.55; H, 10.09; N, 1.49. Calc. for C52H99015N: C, 63.84; H, 10.20; N, 1.43%); TLC (CHC13) RfO single spot; v,,,/cm-' 1600 (C=O, amide), 1740 (C=O, ester); dH (C,D,) 0.90 (6 H, t, CH3), 3.08-4.22 (21 H, m, glucopyranose).Glycolipid 2C1 ,-de-gal containing /?-D-galactopyranosyl- D-gluconamide as head group was prepared from 4.1 g of 3-azapentamethylene bis(octadecan0ate) (5 mmol) and the oxi- dized lactose [0-B-D-galactopyranosyl-(1+4)-~-glucopy-ranose] (1.7 g, 5 mmol) in the same way as above: pale-yellow powder (2.7 g, 55%), m.p. 75.5 "C from methanol (Found: C, 63.63; H, 10.00; N, 1.39. Calc. for C52H99015N: C, 63.84; H, agar glycoli pid acJ+aT---ConA glycolipid .glycolipid per meant I+ -I+ -Et+ -N-Br , -N-Br QN-Et Ci I I Et CgN+ C1 N+ BTAC Fig. 1 A schematic illustration of glycolipid-deposited agar gel membranes 10.20; N, 1.43%); TLC (CHC13) RfO single spot; v,,,/cm-' 1600 (C=O, amide), 1740 (C=O, ester).ConA was purchased from Sigma. Other reagents were analytical grade used without further purification. Spreading Experiments The monolayers were obtained by spreading benzene-ethanol (8:2 in volume) solutions of glycolipids on purified water ('Milli-Q' system, Millipore). The concentration of the spread- ing solution was 1.0mgcm-3. 10 min after spreading the gaseous monolayer was compressed continuously. The com- pressional velocity was 1.20 cm2 s-'. Below this value, the effect of compression rate on the monolayer area was within experimental error. Wilhelmy's plate (filter paper plate) method and a PTFE-coated trough with a microprocessor- controlled film balance (FSD-20, San-Esu Keisoku), having a precision 0.01 mN m- ', were used for surface pressure measurements.Measurements of the surface pressure(n)- area(A) curves for all samples were repeated several times to check their reproducibility. Preparation of Composite Hydrogel Membranes A 4%-agar gel was loaded into a glass plate of 2 mm thickness with a pore size of 6 mm diameter as described previou~ly.'~ Deposition of the surface monolayer onto one side of the 4%- agar gel-loaded glass plate thus prepared was performed at a constant surface pressure with a microprocessor-controlled film balance (FSD-21, San Esu Keisoku). The bare gel sub- strate had been immersed in the subphase before spreading the monolayer and was withdrawn at a speed of 15 mm min-' at a constant surface pressure of 30 mN m- '.The ternary composite hydrogels were prepared as follows. J. MATER. CHEM., 1991, VOL. 1 The 4%-agar gel-loaded glass plates were immersed in a ConA aqueous solution (1 mg ~m-~) for 12 h to give ConA- modified agar gels. Subsequently, the glycolipid monolayers were deposited onto one side of the ConA-modified agar gels in the same way as mentioned above. Permeation Measurements Permeation experiments were performed between pure water (14 cm3) and aqueous CIN+, CsNf, or BTAC (50 cm3) with a conventional, thermostatically controlled H-shaped cell. The membrane area was 0.28 cm2, and both sides of the cell were stirred at a constant speed.Permeation of the salts (C,N+, C6N+, and BTAC) was followed by detecting increases in the electrical conductance in the water side, since good linear correlations between the conductance and the concentration of salts were obtained in the range 0.1-10mmol dm-3. Portions of the solutions of D-mannose or D-galactose were directly added to the salt solution side ([monosaccharide] = 1.5 x mol dm-3). Apparent permeation rates, P (cm s-I), were calculated from P =kv/aC, (1) where k, v, a, and Co are the initial slope of a permeant transport, the volume of the water side of the cell (14 cm3), the membrane area (0.28 cm2) and the concentration of per- meants in the salt solution side (50 mmol dm-3), respectively. Results and Discussion Glycolipid Monolayer-deposited Agar Gels Fig.2 shows surface pressure(n)-area(A) isotherms of mono- layers of 2C18-de-glu and 2C18-de-gal amphiphiles on pure water at 20 "C. Both monolayers showed a similar shape in their n-A isotherms and a condensed solid phase alone with a limiting area per molecule extrapolated at zero pressure (A,) of 0.64 nm2 (2C18-de-glU) or 0.66 nm2 (2C18-de-gal). These values are relatively large compared with those for other dialkyl amphiphiles such as phospholipids (0.40- 0.50 nm2),'l owing to the steric effect of the large sugar head group. In order to obtain information about the phase change of monolayers, n-A isotherms of 2C1p-de-glU monolayers were measured at various temperatures (10-40 "C), and the limiting area per molecule (A,) is shown plotted against temperature in Fig.3. The value of A, increased drastically near 30°C. At the melting point the isobar of a three-area/nm2 per molecule Fig. 2 Surface pressure (71)-area (A) isotherms of (a) 2Cl,-de-gal and (b) 2Cl,-de-glu on pure water at 20 "C J. MATER. CHEM., 1991, VOL. 1 layer-deposited gel membrane was apparently enhanced relative to that across the bare gel membrane when the -Q, o-68t I I I I I I I 10 20 30 40 Tl "C Fig. 3 A typical temperature dependence of the limiting area (A,) for the 2C18-de-glu monolayer dimensional system (crystal) shows an increase in volume. Similarly an increase in area is displayed in a two-dimensional monolayer system when an increase in temperature causes a phase change from the solid analogue to the liquid analogue orientation.Thus, the drastic increase in the A. value near 30 "C (T,)observed in Fig. 3 is ascribed to the phase transition from the gel state to the liquid-crystalline state of the 2C18- de-glu monolayer. It is important to estimate the phase- transition behaviour in an aqueous bilayer state of the lipid and to compare it with that of its monolayer state. However, the lipids prepared in this study could not be dispersed at all in water by sonication. Deposition of the 2Cl8-de-glU monolayer onto one side of a 4%-agar gel-loaded glass plate was performed at a surface pressure of 30 mN m-' with the monolayer in condensed phase.The permeation experiment through the monolayer- deposited agar gel thus obtained was carried out between pure water and aqueous ammonium salt. Fig. 4 shows typical time courses of ammonium salts [(a) C6N+; (b) CIN+] through the monolayer-deposited gel membranes at 20 "C. The membranes were set in two ways: the monolayer faced either the pure-water side or the salt-solution side as shown in Fig.4 (inset). For the permeation of the relatively hydro- phobic permeant C6N+, the permeability across the mono- monolayer was set to face the salt-solution side. In contrast, when the monolayer was set to face the water side, the permeability was markedly reduced, and was close to that of the bare gel membrane. The reason why the membrane is more permeable when the monolayer faces the salt-solution side is probably because the hydrophobicity of C6N+ causes it to penetrate and concentrate in the monolayer, so that the concentration gradient across the hydrogel layer becomes greater than that for the bare hydrogel layer.In the permeation of the hydrophilic probe C,N+ [Fig. 4(b)],the permeability was suppressed compared with that of the bare gel membrane, in contrast to the results of C6N+. When the monolayer faced the pure-water side, the permeability was close to that of the bare gel membrane. These results suggest that the surface monolayer of 2C18-de-gh was definitely deposited onto the gel membrane and its hydrophobic surface covered with alkyl chain would provide a higher permeability for C6N+ than CIN+.In order to confirm that such a stable deposition of the 2Cl,-de-glu monolayer onto the agar gel is due to a sugar-sugar interaction between the &D-glUCOpyranOSyl head group of 2Cl,-de-glu and the sugar residues of agar, a lipid that has a trimethylammonium group in place of the sugar residue of 2C18-de-glU was deposited on the agar gel and used for the same permeation experiment. The permeation rate of C6Nf was initially the same as that of the 2Cl,-de-glu- deposited gel and then approached that of the bare gel within 1 h, suggesting that the ammonium monolayer flaked off into the aqueous solution from the gel surface. Therefore, the sugar-sugar interaction between the monolayer and the agar surface can be concluded to play an important role in stabiliz- ation and facilitated deposition of the 2C18-de-gh monolayer on the agar gel.Fig.4 shows that when the monolayer- deposited gel membranes were set to face the pure water side, permeation data for neither permeant showed any significant difference between the monolayer-deposited gel membranes and the bare gel membranes. The monolayer was thus set to face the salt-solution side in the following permeation experiments. Effect of the Phase Transition of Monolayer-deposited Gels Arrhenius plots of apparent permeation rate P are shown in Fig. 5. The original bare agar gel membrane gave straight Arrhenius plots for both permeants, although the permeability 20 40 60 80 100 20 40 60 80 100 timelmin timelmin Fig.4 Time courses of (a) C6N+ or (b) CIN+ permeation through the 2CI8-de-glu monolayer-deposited agar gel membranes at 20°C.The membranes were set in two ways: the monolayer faced either pure-water side or salt-solution side. Dashed lines show data for the bare gel membranes 368 J. MATER. CHEM., 1991, VOL. 1 TI "C TI "C 40 30 20 40 30 20 I I -8.5 h -I v)E -9.0 0c 0)--9.5 -lo/ I I I I I -3.1 3.2 3.3 3.4 3.5 3.1 3.2 3.3 3.4 3.5 103 KIT 103 KIT Fig. 5 Arrhenius plots of permeation coefficient (P)for (a)C6Nf and (b)CIN+ across the 2C1,-de-glu monolayer-deposited agar gel membranes (a)and the bare agar gel membrane (0).The membranes were set to face salt-solution side.Arrows indicate the phase-transition temperature (T,)of the monolayer determined by the temperature dependence of the n-A curve for CIN+ was always larger than that for C6N+, probably owing to the difference in steric bulkiness and/or hydro- phobicity between the permeant molecules. In the permeation through the monolayer-deposited gel membrane, two types of Arrhenius curve were obtained, which were different from those of the bare membrane. For the permeation of the hydrophilic CINf, the Arrhenius plot was lower than that of the bare gel membrane in the whole temperature range, and gave a discontinuous inflection at ca.30 "C, which corresponds to the T, obtained from the temperature dependence of 7~-A curves (Fig. 3). At temperatures above T,, the permeability was suppressed compared with that for below T,.We do not have any conclusive evidence so far to explain this phenom- enon, but one possibility, which will require further explo- ration, is as follows.The monolayer deposited on the hydrogel seems to produce defective pores in the gel state, and the permeation of a smaller and more hydrophilic molecule such as C,N+ relative to C6Nf below T, is not suppressed greatly compared with that of the bare hydrogel. Since these pores may disappear in the liquid-crystalline state of the deposited monolayer above T,, CIN+ permeation is drastically decreased above T,. Similar situations were observed in the substrate release from a non-ionic lipid-coated capsule membrane." In the case of the hydrophobic permeant C6N+, the per- meability was remarkably enhanced.The Arrhenius plot was always higher than that of the bare gel membrane and was clearly inflected at T,. Such an enhancement of the per- meability for C6N+ may be considered as due to a significant increase in hydrophobicity of the gel surface by deposition of the monolayer. The slope of the Arrhenius plot below T, was steeper than that above T,. Activation energies (E,) for the permeation were calculated on the basis of Arrhenius plots and are listed in Table 1. In the crystalline monolayer below T,,the permeant must diffuse with a high activation energy (90 kJ mol-'). At temperatures above T,,the permeant could easily pass through the liquid-crystalline monolayer with a relatively small activation energy (8 kJ mol- I).These results clearly demonstrate that even on the agar gel the glycolipid monolayer shows a phase transition from gel to liquid crystal, which can control the substrate permeability. Table 1 Comparison of activation energy (E,) for the permeation through the monolayer-deposited hydrogel membranes between permeants E,/kJ mol-' monolayer-deposited gel permeant bare gel T< T, DT, 27 32 90 8 To demonstrate further that the glycolipid monolayer located at the gel surface plays a key role in controlling the substrate permeability, we prepared an agar gel membrane containing 2C18-de-glu in bulk and examined the permeability of C6N + through it. The glycolipid-included gel membrane was prepared by mixing the 4%-agar solution and 2Cl,-de- glu (1 wt.% to agar) and then by loading the mixture into the glass plate in the same way as described in the Experimen- tal.The resulting gel membrane included a larger amount of 2C1 ,-de-glu than that of the 2C18-de-glu monolayer-modified gel membrane. Arrhenius plots of the apparent permeation rate are shown in Fig. 6. The glycolipid-included gel mem- brane gave only the straight Arrhenius plot without an inflection point as had been observed in case of the glycolipid monolayer-deposited gel membranes. Furthermore, the Arrhenius plots fell on the same line as the bare gel membrane. These results imply that the permeability of the glycolipid- included gel membrane is not affected by the phase transition; in other words, the glycolipid molecule is homogeneously dispersed in the gel matrices, or the domain size of the glycolipid would be too small to cause a phase change even if the glycolipid could form a domain in the gel matrices.The above considerations lead to the following conclusions. The glycolipid molecules are stably immobilized at the surface of the agar gel in a monolayer state (mainly by a sugar-sugar interaction) and the monolayer on the gel has a physicochemi- cal property (phase transition) similar to the monolayer on water, which results in controlling the substrate permeability. J. MATER. CHEM., 1991, VOL. 1 -8.5 --v) E2 -9.0 !5 0) '-9.5 3.1 3.2 3.3 3.4 3.5 lo3 KIT Fig.6 Arrhenius plot of permeation coefficient (P) for C6N+ across the 2C1,-de-glu-included (1 wt.% to agar) agar gel membrane (e)and the bare agar gel membrane (0) Ternary Composite Hydrogels Fig. 7 shows time courses of BTAC transport from the salt- solution side, when the bare gel was treated with ConA solution and then the ConA-modified gels were coated with a 2CI8-de-gal or 2C18-de-glu monolayer. These modified-gel surfaces were set to face the salt-solution side. BTAC was employed as a permeant; it has a relatively bulky benzyl group but not a long alkyl chain. The permeation of BTAC through the ConA-modified gel membrane was slightly reduced compared with the bare-gel membrane; this results from an increase of the barrier capability to BTAC permeation due to the binding of ConA with sugar residues in the agar surface.When the monolayer of 2Cl,-de-gal having the B-D-galactopyranosyl moiety was deposited on the ConA-modified gel membrane, the BTAC permeability was hardly changed. This result strongly suggests that the 2C18-de-gal monolayer could not be deposited on the ConA-bound agar surface owing to its poor binding ability with ConA, whereas it could be stably deposited on the pure agar gel (without ConA). In contrast, when the monolayer of 2C1,-de-glu, carrying 3.4 -/ ,(a) / / r U l-I I I 1 1 1 a-D-glucopyranosyl moiety was deposited on the ConA-modi- fied gel membrane, the BTAC permeability was markedly suppressed by a factor of ca.10 compared with that of the bare gel membrane. This means that vacant binding sites of the ConA immobilized on the agar gel surface recognize and strongly bind with the a-D-glucopyranosyl head group of 2C1,-de-glu monolayer. These results show good correlation with the specific bind- ing ability of ConA: ConA binds specifically with a-D-glucopy- ranosyl, but not with /3-D-galactopyranosyl, in glycolipids and polysaccharides on the cell ~urface.'~-'~ It is well known that when ConA is added to an aqueous dispersion of liposomes containing a-D-glucopyranosyl lipids, liposomes are agglutin- ated with each other owing to the specific binding between ConA and the E-D-glucopyranosyl head group of the bilayer surface^.'^-'^ ConA-induced agglutination is inhibited in the presence of an excess of low-molecular-weight sugars such as D-mannose, because the monosaccharide, which has a high binding constant with ConA, expels the glucopyranosyl resi- due of lipids from the binding sites resulting in the dissociation of liposome agglutination.To test the effect of the addition of monosaccharides to our ternary composite gel systems, we employed the 2C18-de-glu monolayer-deposited, ConA-agar gel membrane which had been found to have a barrier to BTAC permeation. Fig. 8 displays the time course of BTAC transport from the salt- solution side. When a large excess of D-galactose (1.5 x 10-mol dm -3, was directly added to the salt-solution side, BTAC permeability did not change, which is as expected since ConA cannot bind D-galactose.On the other hand, in the case of the addition of D-mannose, which has an extremely high binding constant with ConA, BTAC permeability was apparently enhanced. This means that D-mannose expels either the glucopyranosyl residues of the lipid or the sugar residues of agar from the binding sites of ConA, resulting in the formation of a defect at the gel surface as illustrated schematically in Fig. 9. Hence, the permeability increases. Concluding Remarks The glycolipid (2C ,-de-glu and 2C ,-de-gal) monolayer films were successfully formed on the agar hydrogel by the LB technique. The monolayer-deposited gel membranes were satisfactory for the permeation experiments over a wide temperature range.The permeability of the ammonium salts through these modified gel membranes depended strongly -D-mannose 0 ____ __--------30 min w time + Fig. 8 Permeation changes of BTAC through the 2C18-de-glu mono- layer-deposited, ConA-modified agar gel membrane by the addition of D-galactose or D-mannose (1.5 x mol dm-7 agai D-galactose Fig. 9 A schematic representation for effect of the addition of mono- saccharides on the substrate permeability through the ternary com- posite gel membrane upon the hydrophobicity of permeants and could be controlled by the phase transition of the deposited monolayers. The ConA-bound agar gels were modified only with the 2CI8-de- glu monolayer having the a-D-glucopyranosyl head group but not with the 2C18-de-gal having the /3-D-galactopyranosyl head group owing to the specific binding ability of ConA.The permeability of the 2C18-de-glu monolayer-deposited ConA-agar gel was selectively controlled by the addition of monosaccharides: the permeability was enhanced by the addition of D-mannose, whereas it was hardly changed by the addition of D-galactose which resulted from the high binding constant of D-mannose with ConA. The combined properties of hydrogels and lipid membranes could be interesting, not only for a variety of biomedical studies, including the development of new controlled drug- release systems but also for an applicability to conductometric devices. J. MATER. CHEM., 1991, VOL. 1 References 1 Hydrogels for Medical and Related Application, ed. J.D. Andrade, American Chemical Society, Washington DC, 1976. 2 For a recent review, see H. Ringsdorf, B. Schlarb and J. Venzmer, Angew. Chem., 1988, 100, 117 and references cited therein. 3 K. B. Blodgett and I. Langmuir, Phys. Rev., 1937, 51, 964. 4 T. Kajiyama, A. Kumano, M. Takayanagi and T. Kunitake, Chem. Lett., 1984, 915. 5 K. Hechmann, C. Strobl and S. Bauer, Thin Solid Films, 1983, 99, 265. 6 Y. Okahata, K. Ariga, H. Nakahara and K. Fukuda, J. Chem. SOC., Chem. Commun, 1986, 1069. 7 T. Kunitake, N. Higashi, M. Kunitake and Y. Fukushige, Mucro-molecules, 1989, 22, 485. 8 M. Niwa, E. Matsuyoshi and N. Higashi, Langmuir, 1989, 5, 1256. 9 G. D. Rose and J. A. Quinn, J. Colloid Interface Sci., 1968, 27, 193. 10 0. Albrecht, A. Laschewsky and H. Ringsdorf, Macromolecules, 1984, 17, 937. 11 N. Higashi, T. Kunitake and Y. Kajiyama, Polym. J., 1987, 19, 289. 12 A. Arya, U. J. Krull, M. Thompson and H. E. Wong, Anal. Chim. Acta, 1985, 173, 331. 13 K. Hongyo, J. Joseph, R. J. Huber and J. Janata, Langmuir, 1987, 3, 827. 14 N. Higashi, M. Takematsu and M. Niwa, Chem. Lett., 1990,675. 15 H. Bader, H. Ringsdorf and J. Skura, Angew. Chem. Znt. Ed. Engl., 1981, 20, 305. 16 T. Williams, N. R. Plessas and I. J. Goldstein, Arch. Biochem. Biophys., 1979, 19, 145. 17 J. Slama and R. R. Rando, Carbohydr. Rex, 1981, 88, 213. 18 G. A. Orr, R. R. Rando and F. W. Bangerster, J. Biol. Chem., 1979,254,4721. 19 R. Y. Hampton, R. W. Halz and I. J. Goldstein, J. Biol. Chem., 1980,255,6766. 20 K. Kobayashi, H. Sumitomo and Y.Ina, Polym. J., 1985, 17, 567. 21 M. C. Philips and D. Chapman, Biochim. Biophys. Acta, 1968, 163, 301. 22 Y. Okahata, H-J. Lim and G. Nakamura, J. Membrane Sci., 1984, 19, 237. Paper 0/04935B; Received 2nd November, 1990