Preparation of thermo-responsive polymer membranes. I Iwao Nozawa,* Yosuke Suzuki, and Shuji Sat0 Applied Research Laboratory Ill, Lion Co., Ltd., 7-13-12 Hirai, Edogawa-ku, Tokyo 132, Japan Kenji Sugibayashi and Yasunori Morimoto Faculty of Pharmaceutical Sciences, Josai University, 1-1 Keyakidai, Sakado, Saitama 350-02, Japan Two types of liquid crystal (LC)entrapped membranes, (a) polymer alloyed membranes and (b) LC-adsorbed membranes, were investigated for the purpose of developing the drug delivery systems (DDS) with thermal stimuli responsing. Polymer alloyed membranes were obtained by polymerizing acrylic monomers in presence of LC and LC-adsorbed membrane were obtained by adsorbing LC into porous hydrophobic polymer membrane. It was made clear from the indomethacin permeation experi-
ments below and above the gel-liquid crystal phase transition temperature of the LC that the extent of thermo-sensitivity for LC-adsorbed membranes was greater than that for the alloyed membrane. The permeability ratio (38°C vs. 32°C) was found to be about 120 with the LCadsorbed membrane. It was suggested from the results that the LC-adsorbed membrane was one of the useful candidates as a thermo-responsive system for DDS.
INTRODUCTION
“Intelligent materials” have been a focus for therapeutic and diagnostic uses. They may be fabricated in drug delivery systems (DDS) which respond to stimuli such as temperature, pH, photoirradiation, and chemicals. Insulin-maltose complex adsorbed to concanavalin A-immobilized cellulose membrane was developed to establish the insulin release in response to blood glucose level.’ Kim et al. have studied the self-regulating system to improve the side effect of concanavalin A.’ Horbett et al. have investigated the insulin release system consisting of pH-sensitive polymer membrane.3Heller et al. have established a triggered DDS of naltrexone in response to pH change^.^ Okahata and his coworkers have developed nylon capsules whose pores were soaked with synthetic lipid bilayer~.~ The release of entrapped small molecules from the nylon capsules was controlled by thermal stimuli6-’ as well as photoirradiation’ and pH.” Other DDS and materials responding *To whom correspondence should be addressed at Faculty of Pharmaceutical Sciences, Josai University, 1-1Keyakidai, Sakado, Saitama 350-02, Japan. Journal of Biomedical Materials Research, Vol. 25, 243-254 (1991) CCC 0021-9304/91/020243-12$04.00 0 1991 John Wiley & Sons, Inc.
NOZAWA ET AL.
244
to thermal stimuli have been evaluated as follows. Hoffman et al. have been studying the synthesis and application of thermally reversible polymers such as N-isopropylacrylamide to therapeutics and diagnostics'' and Cussler et al. have used temperature-sensitive hydrogels as an extraction solvent.'' Bae et al. have reported the insulin permeation system using thermo-sensitive hydr~gels'~ and the on-off regulation of indomethacin transport utilizing modified N-isopropylacrylamide network^.'^ They claimed that such a regulation can be achieved within a few degrees fluctuations of temperature (even within one degree).I5A complex membrane containing liquid crystal for on-off thermoregulation of ionic solute has been reported by Shinkai et a1.I6 The purpose of the present study is to develop thermo-responsive membranes by using liquid crystals (LC). The membranes containing LC were prepared to have following advantages: (a) the membranes are electrically neutral and therefore compatible with ionized drugs; (b) thermo-response efficacy is so sharp that the membrane can control the drug release in response to minute temperature change. In order to develop such thermo-sensitive system, two kinds of LCentrapped polymer membranes were prepared: One is polymer-alloyed membrane synthesized from radical polymerization of acrylic monomers in presence of LC, and the other is LC-adsorbed membrane prepared by adsorption of nonionic LC molecules into porous hydrophobic membranes, such as polytetrafluoroethylene (PTFE). In uitro permeability of indomethacin (IND) as a model drug through these membranes was evaluated and the effect of temperature on the drug permeation was investigated.
EXPERIMENTAL
Materials The monomers used to prepare polymer alloyed membranes were 2hydroxyethyl methacrylate (HEMA),2-hydroxypropyl methacrylate (HPMA), ethyl acrylate (EA), ethyl methacrylate (EMA), and 2-ethylhexyl acrylate (EHA). They were reagent grade and purchased from Shin-Nakamura Chemical Co., Ltd. (Wakayama, Japan).Azobis(isobutyronitri1e) (AIBN) as an initiator and tetradecaoxyethylene diacrylate (TDODA) as a crosslinking agent were reagent grade and purchased from Tokyo Kasei Co., Ltd. (Tokyo, Japan). The liquid crystals (LC) used in this study were polyoxyethylene trimethylolpropane distearate (PTDS), monooxyethylene trimethylolpropane tristearate (MTTS), polyoxyethylene glyceryl distearate (PGDS), polyoxyethylene distearates (PDS), polyoxyethylene stearyl ether stearates (PSES) and polyoxyethylene lauryl ether stearate (PLES) were purchased from Nihon Emulsion Co., Ltd. (Tokyo). The structural formulas of these LC were described in Figure 1. Millipore membrane filter GS (pore size: 0.22 pm, void volume: 75%), VS (pore size: 0.025 pm, void volume: 70%)and Toyo Roshi membrane filter PT-
THERMO-RESPONSIVE POLYMER MEMBRANES. I
f
a3m2c
a2(%?a2'M4%!
'a-l-lF35
a2m2a2(m25)b-1-l?35 c3320cH2'H2(a
a
'c-loH
Folyoxyethylene Rimethylolpmpme Distearate a + b + c = 5 PTDS
a3M2C
t
245
1' F35-2m2
(m2a2
?35
In- 1 -
Folyoxyethylene Distearate P D S 300 P D S 600
n - 1 = 3 0 n - 1 = 6 0
a2m12a2-1F35 M2a2a2-17%3
C12H25m2a2(m2M2
M2m2m2-l?35
~lyoxyethyleneLsuryl Ether Steurate
Mnooxyethylene W i m t h y l o l p m p m e Wintearate HTTS ;)12a2%!(m2a2 -2M2(a2m2 a,Oga,(m,a2
n - 1 = 1 0
?35
PLES
'a- 1Oarl7%5 'b-l-lF35
I
'n-1-1
' C - P
mlyoxyethylene Glyceryl Distearate a + b + c = 4 PGDS
c18H37m2a2(m2a2
'"-1-1
?35
Ether Strearate m i ~ t h y Stearyl l ~ n - l = 4 P S E S 4 n - 1 = 1 0 PSES10
Figure 1. Structural formulas of liquid crystals.
020 (pore size: 0.2 pm, void volume: 75%) were used as model porous hydrophobic membranes made from PTFE to support LC molecule. The diameters of these membranes were 47 mm. IND was commercial products of Japanese Pharmacopoeia grade. All other chemicals were reagent grade products.
Preparation of LC-entrapped membranes Two kinds of LC-entrapped membranes, (a) polymer alloyed membrane and (b) LC-adsorption membrane, were prepared. The polymer-alloyed membranes were prepared through radical polymerization of predetermined amount of HEMA, HPMA, EMA, EA, and/or EHA in presence of PTDS as a LC, with or without TDODA as a crosslinking agent at 60°C for 15 min in a mold (15 x 15 cm') with a gasket (silicone gum, 1mm of thickness) in presence of AIBN (1%against total monomer weight). The concentrations of PTDS and TDODA were 30 and 2%, respectively, against total weight. The mixture was initially clear (homogeneous), but became cloudy (heterogeneous) with progress of polymerization. The membranes synthesized were removed from the mold and purified by immersing into distilled water for several days. The distilled water was changed everyday. The thickness of polymer alloyed membranes obtained was approximately 1 mm. The LCadsorbed membrane was prepared by soaking a PT-020 membrane filter in 100 mL of 20% LC chloroform solution for 1h and then drying in air for 24 h at room temperature. When studying the influence of the pore size and LC content of the membranes on the thermo-response efficacy, on the other hand, the LC-adsorbed membranes were prepared by a soaking Millipore membrane filter GS and VS into 100 mL of various concentrations (0, 5, 10, 20%)of LC chloroform solution. LC contents in LC-adsorbed membranes pre-
NOZAWA ET AL.
246
pared by using 5,10,20% LC chloroform solution were 30,38,48% per membrane weight, respectively.
Measurements of phase transition temperature The gel-liquid crystal phase-transition behavior of LC in 40% EtOH aqueous solution, 50% N-methyl-2-pyrrolidone aqueous solution or phosphate buffered saline (PBS)(pH 7.4) was examined by a differential scanning calorimeter (DSC, Daini Seikousha SSC/560, Tokyo). The solvents alone were used as standard substances. Temperature increasing rate was adjusted to l"C/min.
Measurement of drug permeability Two-chamber glass diffusion cell17was used for the measurement of membrane permeation of IND. Permeation experiments were performed at 32 and 38°C in thermostatic water bath. The compartments (100 mL of each volume) were separated by a test membrane (effective area, 7.7 cm'). Donor chamber was filled with 100 mL of 0.05% IND in 40% EtOH, 50% N-methyl-2pyrrolidone or PBS (pH 7.4), and receiver with 100 mL of the same solution without IND. The solution of each compartment was stirred at 500 rpm by magnetic stirrer to neglect the boundary layer effect. The permeation of the drug was pursued by sampling 50 p L of aliquot from the receiver chamber and assaying with HPLC. The permeability coefficient, P, was calculated with the following equation: ln(1 - 2Ct/Co) = -2PAt/V
(1) where Co, C,, A, and V are the initial drug concentration in the donor chamber, the drug concentration in the receiver side at time f, the effective area for permeation, and the volume of each chamber, respectively. RESULTS A N D DISCUSSION
Thermo-responsive property of polymer alloyed membrane DSC measurements of LC were performed to select the suitable LC. LC was selected in view of position and sharpness of the endothermic peak of DSC. Figure 2 shows DSC peaks of LC in PBS (pH 7.4). Most LC examined had two endothermic peaks in narrow range. The lower temperature peak may be ascribed to the gel-liquid phase transition due to microscopic movements of alkyl chains of LC. The gel-liquid phase transition temperature (T,) was influenced by the basic structure, the number of ethylene oxide units, and the length of alkyl chain of each LC. Among the LC examined, PTDS showed a sharp endothermic peak near the body temperature. Therefore, PTDS was selected for the system in PBS.
THERMO-RESPONSIVE POLYMER MEMBRANES. I
247
Temperature ( " C ) 20 3040506007080
Y I
v-
F -pLEslo
O0
Figure 2. DSC charts of liquid crystals.
Table I shows the permeability coefficients of IND through the crosslinked and non-crosslinked polymer alloyed membranes (heterogeneous) at 32°C and 38°C. The membrane permeabilities of IND without LC alloyed were very high at both temperatures, 32°C and 38"C, and few difference in the permeabilities was observed between two temperatures (data not shown). Although the thermo-response efficacy evaluated by the ratio of the permeability coefficients at 38°C to 32"C, P,,/P,,, showed a tendency to increase with increasing alkyl group length of the polymer, the influence of polymer structure on the thermo-responsive efficacy was not remarkable in this range examined. TABLE I Permeability Coefficients* of IND through the Polymer Alloyed Membranes and Thermo-Responsive Efficacies (P38/P32) Permeability Coefficient x 106(cm/s) Polymer
P38
PB/P32
(a) Crosslinked polymer alloyed membranes P(HEMA/EA) (7/3) 1.44 f 0.09 1.13 f 0.09 P(HEMA/EA) (5/5) P(HEMA/EHA) (7/3) 0.97 2 0.11
3.19 k 0.14 3.14 f 0.09 2.99 f 0.23
2.2 2.8 3.1
(b) Non-crosslinked polymer alloyed membranes PHEMA 1.86 f 0.18 PHPMA 2.18 f 0.13 P(HEMA/EA) (5/5) 1.50 f 0.16 P(HEMA/EMA) (5/5) 1.11f 0.11
4.34 k 0.07 4.25 f 0.17 4.87 f 0.23 3.76 f 0.17
2.3 2.0 3.2 3.4
P32
*Values are means f SE of results for 3 experiments.
248
NOZAWA ET AL.
The morphology of the alloyed membrane which is an important factor in the drug transport is not yet cleared. Because the polymer membrane itself have no thermosensitivity, high efficacy can not be expected if the LC is not a continuous phase. In order to examine the effect of morphology on the IND permeation behavior, membranes with higher LC-content than 30% were synthesized. The membranes obtained, however, were too low mechanical strength to do permeation experiments. The low thermo-responsive efficacy of LC alloyed membrane may be explained as follows. In general, the drug molecules permeate both by ”pore mechanism” through solvent channel in polymer matrix and by “partition mechanism” through the continuous polymer phase. In case of the polymer alloyed membranes, IND permeation at 38°C may be dominated by the former mechanism because the polymer matrix is relatively hydrophilic and also the state of LC is disordered at the T,. On the other hand, at 32”C,LC should be rigid and then IND permeation through LC part should be mainly governed by ”partitioning mechanism.” On the contrary, there is little difference between the permeation mechanisms in the polymer matrix at 32°C and 38°C. The similar permeability of polymer matrix itself at both temperatures, 32°C and 38”C, may lower the thermo-responsive efficacy of total LC alloyed membranes. It can be expected from the above speculations that impermeable polymer matrices will be needed to obtain the high thermo-responsive effect of LC on the drug permeation through LC-entrapped membrane.
Thermo-responsive property of LC-adsorbed membranes On the hypothesis that impermeable polymer matrices improve the thermosensitivity of LC-entrapped membrane, hydrophobic membrane filters made from PTFE were chosen and examined as LC-supporting membranes. The permeability of IND through the PTDS-adsorbed membranes which have various pore size and PTDS content was measured at several temperatures in order to examine the influence of membrane’s pore size and LC content on the thermo-response efficacy. Figure 3 represents the effect of pore size and PTDS content in the membrane on the membrane permeability of IND. The flux for two membranes containing 0% PTDS was strikingly similar in spite of the difference in pore size. This similarity of the flux between two membranes may be explained by the similar void volume. It was clear that the membrane with the smaller pore and higher concentration of PTDS showed the higher thermo-response efficacy. The membrane prepared with 20% LC chloroform solution indicated the highest efficacy. The thermoresponsive efficacy presented as a ratio of IND permeability coefficients at 38°C to 32°C (P38/P32)was nearly 120 (Fig. 3b). From these results, it is concluded that the content of LC in membrane-in other words, the extent of LC filled up into pores of the membrane-is one of the key factors for the thermo-response efficacy. The void volume of the membrane may influence the extent of LC filled up. There is no significant
THERMO-RESPONSIVE POLYMER MEMBRANES. I
T
40
-12
36
("C
249
1
32
28
-12
-13
a
a -13
-14
C
ri
r(
y. -14
-15 -16
-17 3.20
3.28
I/K x 1 ~ 3( K - ~ )
3.20 I/K
3.28
x 1 ~ 3( K - ~ )
Figure 3. The relationship between temperature and the permeation coefficient of IND through the PTDS-adsorbed membrane, (a) pore size: 0.22 pm; (b) pore size: 0.025 pm. Each number in percentage means the concentration of LC chloroform solution.
difference in the void volume between the small pore (0.025 pm) and large pore (0.22 pm) membranes (70 and 75%, respectively). The amounts of LC adsorbed on the both membranes prepared by immersing into 10% LC chloroform solution were the same (38%).The degree of orientation and mobility of the adsorbed LC in the different pore size were estimated by DSC. However, there was no difference in the DSC-thermogram for both membranes (data not shown). The LC-contents in both LC-adsorbed membranes (38%)were so small in comparison with the void volume (70-78%) that the pores in both membranes were not completely filled with LC. As mentioned above, the difference in thermo-sensitivity between two membranes was not explained by the differences in the mobility of absorbed LC nor the void volume of supporting membranes. The thermo-response efficacy of the polymer-alloyed membrane was less than that of the LC-adsorbed membrane. This higher thermo-response efficacy of the LC-adsorbed membrane can be explained not only by less permeability through matrix part of the LC-supporting PTFE membrane but also by higher mobility of the LC molecules due to the phase transition below and above the T,.
The reversible thermo-control of LC-adsorbed membrane The reversible thermo-control of IND permeation through the LCadsorbed membrane was studied. Figure 4 shows a pulsatile permeation of IND through PTDS-adsorbed membrane, in which PTDS content was 48% per membrane weight, in response to the temperature change between 32°C and 38°C with the interval of 2 h in PBS (pH 7.4). The onset of response to the temperature change was very fast and no clear lag time was observed. In most cases, transdermal therapeutic system (TTS) contains organic solvents as solubilizers and/or penetration-enhancers, so estimation of the
-
250
.. .
...
2
4
...
...
6
8
NOZAWA ET AL.
" 0
T i m e
( h )
Figure 4. The pulsatile release of IND through the PTDS-adsorbed membrane in response to temperature changes in PBS.
influence of organic solvents on the T , of LC is very important. The temperature-dependent permeation of IND was investigated in various solution systems. IND permeation experiments through the PTDS-adsorbed membrane were carried out in 40% EtOH aqueous solution (Fig. 5) and 50% N-methyl-2-pyrrolidone aqueous solution (Fig. 6). The dotted lines in Figures 5b and 6b show the permeation coefficients of IND calculated from the steady state regions of the permeation profiles at constant temperatures (32°C and 38°C). The pulsatile release function was reduced by EtOH and IND permeation was not decreased at 32°C as expected. LC was not desorbed from the membrane to 40% EtOH aqueous solution at 40°C for 6 months (data not shown). LC was not dissolved to these organic solution systems. The reduction of the thermo-response efficacy by N-methyl-2-pyrrolidonewas not observed so clearly as expected, which could be due to the difference in hydrophobicity of solvents. W
3
W
14 12
10 8
6 4 %
2
2
0
0
u.l
0
I
0
2
4
6
T i m e ( h )
8
0
2
4
6
8
T i m e ( h )
Figure 5. The influence of EtOH on the pulsatile release of IND through the PTDS-adsorbed membrane in 40% EtOH aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
THERMO-RESPONSIVE POLYMER MEMBRANES. I
251
38°C
0
2
4
6
T i m e ( h )
8
P
0
2
4
6
8
T i m e ( h 1
Figure 6. The influence of N-methyl-2-pyrrolidone on the pulsatile release of IND through the PTDS-adsorbed membrane in 50% N-methyl-2pyrrolidone aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
The influence of EtOH and N-methyl-2-pyrrolidone on the T, of PTDS and MTTS with a higher T, than PTDS was examined by DSC (Fig. 7). The shift of T, to a few centidegrees lower side was observed for 40% EtOH aqueous solution, which might be due to the plasticizing effects of EtOH on alkyl chains of LC. The plasticizing effects was yielded from the hydrophobic interaction between alkyl chains of LC and organic solvents which controls the gel-liquid phase transition behavior of LC. Besides plasticizing effects, a decrease in activity coefficient of IND in membrane might affect the relatively high permeability of IND through the PTDS-adsorbed membrane in 40% EtOH aqueous solution. EtOH immersed into the membranes may increase the membrane permeability of IND. In Figure 7, the T, shift of PTDS with Nmethyl-2-pyrrolidone aqueous solution seems to be nearly the same degree with that of EtOH aqueous solution. However, the thermo-responsive efficacy with N-methyl-2-pyrrolidone solution seems to be higher than that in PBS when compared with Figures 4 and 6. The difference in influences on the thermo-responsive efficacy by N-methyl-2-pyrrolidone and EtOH is not yet cleared. It may be impossible to evaluate the difference only from the DSC data. Since MTTS has a T , near 37-38°C in the organic solutions, it is suitable in the organic solution system. MTTS-adsorbed membrane was prepared by the same method as LC-adsorbed membrane described in experimental part. The MTTS content in MTTS-adsorbed membrane was 48% per membrane weight. Figures 8 and 9 show the IND permeation through MTTS-adsorbed membrane in response to temperature change in 50% N-methyl-2-pyrrolidone and 40% EtOH aqueous solution, respectively. Compared to PTDS adsorbed membranes in PBS (Fig. 5), almost no permeation below T , and low permeation at 38°C were shown with MTTS-adsorbed membranes in N-methyl-2pyrrolidone aqueous solution (Fig. 9). There were two order differences
NOZAWA ET AL.
252 (a)
P TD
(b) M T T S
S
('1
'rewerat-
%mp=.rature ( ' )
203004050
20304080
aq. solution
Figure 7. DSC charts of (a) PTDS and (b) MTTS in PBS, 40% EtOH and 50% N-methyl-2-pyrrolidone aqueous solutions.
38'C~3ZoCi 3 8 ~ ~ 3 2I ° C
ul ,
. .
. .
... . ...
... .. ..
.
.
.
.
-
i 0
2
4
T i m e
6 ( h )
8
k
0O
... .....
,.. 2.
......
.....4 .....6.;u: 8
T i m e
( h )
Figure 8. The influence of EtOH on the pulsatile release of IND through the MTTS-adsorbed membrane in 40% EtOH aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
between the two cases. These differences in permeability might be due to the difference in the chemical structure of LC, the number of alkyl chains. It can be seen that MTTS with three alkyl chains may have higher resistance to IND permeation than PTDS with two alkyl chains. Thus, the thermo-responsive system can be prepared by adsorbing LC, selected by DSC measurement
THERMO-RESPONSIVE POLYMER MEMBRANES. I
253
30°C ,.....
E
f
0
T2 i m 4e
( 6h ) 0
8
O
2
4
T i m e
6
8
( h )
Figure 9. The inf hence of N-methyl-2-pyrrolidone on the pulsatile release of IND through the MTTS-adsorbed membrane in 50% N-methyl-2pyrrolidone aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
in a given composition of the solution, to the porous hydrophobic polymer membrane. In conclusion, the present study demonstrated that the drug permeation could be sharply on-off controlled in response to minute temperature change by using the LC-adsorbed membrane. The mechanism for drug permeation through LC-adsorbed membrane may be estimated from Figures 3 and 7 and by the microscopic movement of LC-alkyl chain due to thermal stimuli. This thermo-responsive system is applicable to DDS which responds to the change of body temperature and/or the external thermal stimuli. References 1.
2. 3. 4. 5.
6.
7. 8.
M. Brownlee and A. Cerami, “A glucose-controlled insulin-delivery system: Semisynthetic insulin bound to lectin,” Science, 206,1190-1191 (1979). J. M. Anderson and S. W. Kim, in Recent Advances in Drug Delivery Systems, Plenum Press, New York, 1984, pp. 123-136. T. A. Horbell, B. D. Ratner, J. Kost and M. Singh, in Recent Advances in Drug Delivery Systems, Plenum Press, New York, 1984, pp. 209-220. J. Heller, S. H. Pangburn, and D. W. Penhale, Proceed. Intern. Symp. Control. Rel. Bioact. Muter., 14, 107-108 (1987). Y. Okahata, “Lipid bilayer-corked capsule membrane. Reversible signal-receptive permeation control,” Acc. Chem. Res., 19, 57-63 (1986). Y. Okahata, Han-Jim Lim, G. Nakamura, and S. Hachiya, “A large nylon capsule coated with a synthetic bilayer membrane. Permeability control of NaCl by phase transition of the dialkylammonium bilayer coating,” J. Am. Chem. SOC., 105, 48554859 (1983). Y. Okahata, N. Iizuka, G. Nakamura, and T. Seki, “Permeability of fluorescent probes at phase transitions from bilayer-coated capsule membranes,” f. Chern. SOC.,Perkin Truns. 11,1591-1600 (1985). C. Yanagawa, Y. Tanaka, M. Endo, Y. Nakajima, and Y. Okahata, ”Drug delivery system in combination with hyperthermia,” Drug Delivery System, 3(1), 255-261 (1988).
NOZAWA ET AL.
254 9.
10. 11. 12. 13. 14. 15. 16.
17.
Y. Okahata, Han-Jim Lim, and S. Hachiya, "Bilayer coated capsule membrane. Part 2. Photo-responsive permeability control of sodium chloride across a capsule membranes," J. Chem. SOC.,Perkin Trans. 11, 989-994 (1984). Y. Okahata, H. Noguchi, and T. Seki, "Functional capsule membrane. 26. Permeability control of polymer-grafted capsule membranes responding to ambient pH changes," Macromolecules, 20, 15-21 (1987). A.S. Hoffman, "Application of thermally reversible polymers and hydrogels in therapeutics and diagnostics," J. Controlled Rel., 6, 297-305 (1987). S. Freitas and E. L. Cussler, "Temperature sensitive gels as extraction solvents," Chem. Eng. Sci., 42, 79-86 (1987). Y.H. Bae, T. Okano, and S.W. Kim, "Insulin permeation through thermo-sensitive hydrogels," J. Controlled Rel., 9, 271-279 (1989). Y. H. Bae, T. Okano, R. Hsu, and S. W. Kim, "Thermo-sensitive polymers as on-off switches for drug release," Makromol. Chem., Rapid Commun., 8, 481-485 (1987). Y. H. Bae, K. Mukae, T. Okano, and S. W. Kim, "On-off transport regulation through thermosensitive hydrogels," Proceed. Intern. Symp. Contr. Rel. Bioact. Mater., 15, 182 (1988). S. Shinkai, S. Nakamura, K. Ohara, S. Tachiki, 0. Manabe, and T. Kajiyama, "Complete thermocontrol of ion permeation through ternary composite membranes composed of polymerAiquid crystayamphiphilic crown ethers," Macromolecules, 20, 21-28 (1987). Y. Morimoto, K. Sugibayashi, K. Hosoya and W. I. Higuchi, "Penetration enhancing effect of Azone on the transport of 5-fluorouracil across the hairless rat skin," Inf. J. Pharm., 32, 31-38 (1986).
Received December 1988 Accepted August 30,1990
ments below and above the gel-liquid crystal phase transition temperature of the LC that the extent of thermo-sensitivity for LC-adsorbed membranes was greater than that for the alloyed membrane. The permeability ratio (38°C vs. 32°C) was found to be about 120 with the LCadsorbed membrane. It was suggested from the results that the LC-adsorbed membrane was one of the useful candidates as a thermo-responsive system for DDS.
INTRODUCTION
“Intelligent materials” have been a focus for therapeutic and diagnostic uses. They may be fabricated in drug delivery systems (DDS) which respond to stimuli such as temperature, pH, photoirradiation, and chemicals. Insulin-maltose complex adsorbed to concanavalin A-immobilized cellulose membrane was developed to establish the insulin release in response to blood glucose level.’ Kim et al. have studied the self-regulating system to improve the side effect of concanavalin A.’ Horbett et al. have investigated the insulin release system consisting of pH-sensitive polymer membrane.3Heller et al. have established a triggered DDS of naltrexone in response to pH change^.^ Okahata and his coworkers have developed nylon capsules whose pores were soaked with synthetic lipid bilayer~.~ The release of entrapped small molecules from the nylon capsules was controlled by thermal stimuli6-’ as well as photoirradiation’ and pH.” Other DDS and materials responding *To whom correspondence should be addressed at Faculty of Pharmaceutical Sciences, Josai University, 1-1Keyakidai, Sakado, Saitama 350-02, Japan. Journal of Biomedical Materials Research, Vol. 25, 243-254 (1991) CCC 0021-9304/91/020243-12$04.00 0 1991 John Wiley & Sons, Inc.
NOZAWA ET AL.
244
to thermal stimuli have been evaluated as follows. Hoffman et al. have been studying the synthesis and application of thermally reversible polymers such as N-isopropylacrylamide to therapeutics and diagnostics'' and Cussler et al. have used temperature-sensitive hydrogels as an extraction solvent.'' Bae et al. have reported the insulin permeation system using thermo-sensitive hydr~gels'~ and the on-off regulation of indomethacin transport utilizing modified N-isopropylacrylamide network^.'^ They claimed that such a regulation can be achieved within a few degrees fluctuations of temperature (even within one degree).I5A complex membrane containing liquid crystal for on-off thermoregulation of ionic solute has been reported by Shinkai et a1.I6 The purpose of the present study is to develop thermo-responsive membranes by using liquid crystals (LC). The membranes containing LC were prepared to have following advantages: (a) the membranes are electrically neutral and therefore compatible with ionized drugs; (b) thermo-response efficacy is so sharp that the membrane can control the drug release in response to minute temperature change. In order to develop such thermo-sensitive system, two kinds of LCentrapped polymer membranes were prepared: One is polymer-alloyed membrane synthesized from radical polymerization of acrylic monomers in presence of LC, and the other is LC-adsorbed membrane prepared by adsorption of nonionic LC molecules into porous hydrophobic membranes, such as polytetrafluoroethylene (PTFE). In uitro permeability of indomethacin (IND) as a model drug through these membranes was evaluated and the effect of temperature on the drug permeation was investigated.
EXPERIMENTAL
Materials The monomers used to prepare polymer alloyed membranes were 2hydroxyethyl methacrylate (HEMA),2-hydroxypropyl methacrylate (HPMA), ethyl acrylate (EA), ethyl methacrylate (EMA), and 2-ethylhexyl acrylate (EHA). They were reagent grade and purchased from Shin-Nakamura Chemical Co., Ltd. (Wakayama, Japan).Azobis(isobutyronitri1e) (AIBN) as an initiator and tetradecaoxyethylene diacrylate (TDODA) as a crosslinking agent were reagent grade and purchased from Tokyo Kasei Co., Ltd. (Tokyo, Japan). The liquid crystals (LC) used in this study were polyoxyethylene trimethylolpropane distearate (PTDS), monooxyethylene trimethylolpropane tristearate (MTTS), polyoxyethylene glyceryl distearate (PGDS), polyoxyethylene distearates (PDS), polyoxyethylene stearyl ether stearates (PSES) and polyoxyethylene lauryl ether stearate (PLES) were purchased from Nihon Emulsion Co., Ltd. (Tokyo). The structural formulas of these LC were described in Figure 1. Millipore membrane filter GS (pore size: 0.22 pm, void volume: 75%), VS (pore size: 0.025 pm, void volume: 70%)and Toyo Roshi membrane filter PT-
THERMO-RESPONSIVE POLYMER MEMBRANES. I
f
a3m2c
a2(%?a2'M4%!
'a-l-lF35
a2m2a2(m25)b-1-l?35 c3320cH2'H2(a
a
'c-loH
Folyoxyethylene Rimethylolpmpme Distearate a + b + c = 5 PTDS
a3M2C
t
245
1' F35-2m2
(m2a2
?35
In- 1 -
Folyoxyethylene Distearate P D S 300 P D S 600
n - 1 = 3 0 n - 1 = 6 0
a2m12a2-1F35 M2a2a2-17%3
C12H25m2a2(m2M2
M2m2m2-l?35
~lyoxyethyleneLsuryl Ether Steurate
Mnooxyethylene W i m t h y l o l p m p m e Wintearate HTTS ;)12a2%!(m2a2 -2M2(a2m2 a,Oga,(m,a2
n - 1 = 1 0
?35
PLES
'a- 1Oarl7%5 'b-l-lF35
I
'n-1-1
' C - P
mlyoxyethylene Glyceryl Distearate a + b + c = 4 PGDS
c18H37m2a2(m2a2
'"-1-1
?35
Ether Strearate m i ~ t h y Stearyl l ~ n - l = 4 P S E S 4 n - 1 = 1 0 PSES10
Figure 1. Structural formulas of liquid crystals.
020 (pore size: 0.2 pm, void volume: 75%) were used as model porous hydrophobic membranes made from PTFE to support LC molecule. The diameters of these membranes were 47 mm. IND was commercial products of Japanese Pharmacopoeia grade. All other chemicals were reagent grade products.
Preparation of LC-entrapped membranes Two kinds of LC-entrapped membranes, (a) polymer alloyed membrane and (b) LC-adsorption membrane, were prepared. The polymer-alloyed membranes were prepared through radical polymerization of predetermined amount of HEMA, HPMA, EMA, EA, and/or EHA in presence of PTDS as a LC, with or without TDODA as a crosslinking agent at 60°C for 15 min in a mold (15 x 15 cm') with a gasket (silicone gum, 1mm of thickness) in presence of AIBN (1%against total monomer weight). The concentrations of PTDS and TDODA were 30 and 2%, respectively, against total weight. The mixture was initially clear (homogeneous), but became cloudy (heterogeneous) with progress of polymerization. The membranes synthesized were removed from the mold and purified by immersing into distilled water for several days. The distilled water was changed everyday. The thickness of polymer alloyed membranes obtained was approximately 1 mm. The LCadsorbed membrane was prepared by soaking a PT-020 membrane filter in 100 mL of 20% LC chloroform solution for 1h and then drying in air for 24 h at room temperature. When studying the influence of the pore size and LC content of the membranes on the thermo-response efficacy, on the other hand, the LC-adsorbed membranes were prepared by a soaking Millipore membrane filter GS and VS into 100 mL of various concentrations (0, 5, 10, 20%)of LC chloroform solution. LC contents in LC-adsorbed membranes pre-
NOZAWA ET AL.
246
pared by using 5,10,20% LC chloroform solution were 30,38,48% per membrane weight, respectively.
Measurements of phase transition temperature The gel-liquid crystal phase-transition behavior of LC in 40% EtOH aqueous solution, 50% N-methyl-2-pyrrolidone aqueous solution or phosphate buffered saline (PBS)(pH 7.4) was examined by a differential scanning calorimeter (DSC, Daini Seikousha SSC/560, Tokyo). The solvents alone were used as standard substances. Temperature increasing rate was adjusted to l"C/min.
Measurement of drug permeability Two-chamber glass diffusion cell17was used for the measurement of membrane permeation of IND. Permeation experiments were performed at 32 and 38°C in thermostatic water bath. The compartments (100 mL of each volume) were separated by a test membrane (effective area, 7.7 cm'). Donor chamber was filled with 100 mL of 0.05% IND in 40% EtOH, 50% N-methyl-2pyrrolidone or PBS (pH 7.4), and receiver with 100 mL of the same solution without IND. The solution of each compartment was stirred at 500 rpm by magnetic stirrer to neglect the boundary layer effect. The permeation of the drug was pursued by sampling 50 p L of aliquot from the receiver chamber and assaying with HPLC. The permeability coefficient, P, was calculated with the following equation: ln(1 - 2Ct/Co) = -2PAt/V
(1) where Co, C,, A, and V are the initial drug concentration in the donor chamber, the drug concentration in the receiver side at time f, the effective area for permeation, and the volume of each chamber, respectively. RESULTS A N D DISCUSSION
Thermo-responsive property of polymer alloyed membrane DSC measurements of LC were performed to select the suitable LC. LC was selected in view of position and sharpness of the endothermic peak of DSC. Figure 2 shows DSC peaks of LC in PBS (pH 7.4). Most LC examined had two endothermic peaks in narrow range. The lower temperature peak may be ascribed to the gel-liquid phase transition due to microscopic movements of alkyl chains of LC. The gel-liquid phase transition temperature (T,) was influenced by the basic structure, the number of ethylene oxide units, and the length of alkyl chain of each LC. Among the LC examined, PTDS showed a sharp endothermic peak near the body temperature. Therefore, PTDS was selected for the system in PBS.
THERMO-RESPONSIVE POLYMER MEMBRANES. I
247
Temperature ( " C ) 20 3040506007080
Y I
v-
F -pLEslo
O0
Figure 2. DSC charts of liquid crystals.
Table I shows the permeability coefficients of IND through the crosslinked and non-crosslinked polymer alloyed membranes (heterogeneous) at 32°C and 38°C. The membrane permeabilities of IND without LC alloyed were very high at both temperatures, 32°C and 38"C, and few difference in the permeabilities was observed between two temperatures (data not shown). Although the thermo-response efficacy evaluated by the ratio of the permeability coefficients at 38°C to 32"C, P,,/P,,, showed a tendency to increase with increasing alkyl group length of the polymer, the influence of polymer structure on the thermo-responsive efficacy was not remarkable in this range examined. TABLE I Permeability Coefficients* of IND through the Polymer Alloyed Membranes and Thermo-Responsive Efficacies (P38/P32) Permeability Coefficient x 106(cm/s) Polymer
P38
PB/P32
(a) Crosslinked polymer alloyed membranes P(HEMA/EA) (7/3) 1.44 f 0.09 1.13 f 0.09 P(HEMA/EA) (5/5) P(HEMA/EHA) (7/3) 0.97 2 0.11
3.19 k 0.14 3.14 f 0.09 2.99 f 0.23
2.2 2.8 3.1
(b) Non-crosslinked polymer alloyed membranes PHEMA 1.86 f 0.18 PHPMA 2.18 f 0.13 P(HEMA/EA) (5/5) 1.50 f 0.16 P(HEMA/EMA) (5/5) 1.11f 0.11
4.34 k 0.07 4.25 f 0.17 4.87 f 0.23 3.76 f 0.17
2.3 2.0 3.2 3.4
P32
*Values are means f SE of results for 3 experiments.
248
NOZAWA ET AL.
The morphology of the alloyed membrane which is an important factor in the drug transport is not yet cleared. Because the polymer membrane itself have no thermosensitivity, high efficacy can not be expected if the LC is not a continuous phase. In order to examine the effect of morphology on the IND permeation behavior, membranes with higher LC-content than 30% were synthesized. The membranes obtained, however, were too low mechanical strength to do permeation experiments. The low thermo-responsive efficacy of LC alloyed membrane may be explained as follows. In general, the drug molecules permeate both by ”pore mechanism” through solvent channel in polymer matrix and by “partition mechanism” through the continuous polymer phase. In case of the polymer alloyed membranes, IND permeation at 38°C may be dominated by the former mechanism because the polymer matrix is relatively hydrophilic and also the state of LC is disordered at the T,. On the other hand, at 32”C,LC should be rigid and then IND permeation through LC part should be mainly governed by ”partitioning mechanism.” On the contrary, there is little difference between the permeation mechanisms in the polymer matrix at 32°C and 38°C. The similar permeability of polymer matrix itself at both temperatures, 32°C and 38”C, may lower the thermo-responsive efficacy of total LC alloyed membranes. It can be expected from the above speculations that impermeable polymer matrices will be needed to obtain the high thermo-responsive effect of LC on the drug permeation through LC-entrapped membrane.
Thermo-responsive property of LC-adsorbed membranes On the hypothesis that impermeable polymer matrices improve the thermosensitivity of LC-entrapped membrane, hydrophobic membrane filters made from PTFE were chosen and examined as LC-supporting membranes. The permeability of IND through the PTDS-adsorbed membranes which have various pore size and PTDS content was measured at several temperatures in order to examine the influence of membrane’s pore size and LC content on the thermo-response efficacy. Figure 3 represents the effect of pore size and PTDS content in the membrane on the membrane permeability of IND. The flux for two membranes containing 0% PTDS was strikingly similar in spite of the difference in pore size. This similarity of the flux between two membranes may be explained by the similar void volume. It was clear that the membrane with the smaller pore and higher concentration of PTDS showed the higher thermo-response efficacy. The membrane prepared with 20% LC chloroform solution indicated the highest efficacy. The thermoresponsive efficacy presented as a ratio of IND permeability coefficients at 38°C to 32°C (P38/P32)was nearly 120 (Fig. 3b). From these results, it is concluded that the content of LC in membrane-in other words, the extent of LC filled up into pores of the membrane-is one of the key factors for the thermo-response efficacy. The void volume of the membrane may influence the extent of LC filled up. There is no significant
THERMO-RESPONSIVE POLYMER MEMBRANES. I
T
40
-12
36
("C
249
1
32
28
-12
-13
a
a -13
-14
C
ri
r(
y. -14
-15 -16
-17 3.20
3.28
I/K x 1 ~ 3( K - ~ )
3.20 I/K
3.28
x 1 ~ 3( K - ~ )
Figure 3. The relationship between temperature and the permeation coefficient of IND through the PTDS-adsorbed membrane, (a) pore size: 0.22 pm; (b) pore size: 0.025 pm. Each number in percentage means the concentration of LC chloroform solution.
difference in the void volume between the small pore (0.025 pm) and large pore (0.22 pm) membranes (70 and 75%, respectively). The amounts of LC adsorbed on the both membranes prepared by immersing into 10% LC chloroform solution were the same (38%).The degree of orientation and mobility of the adsorbed LC in the different pore size were estimated by DSC. However, there was no difference in the DSC-thermogram for both membranes (data not shown). The LC-contents in both LC-adsorbed membranes (38%)were so small in comparison with the void volume (70-78%) that the pores in both membranes were not completely filled with LC. As mentioned above, the difference in thermo-sensitivity between two membranes was not explained by the differences in the mobility of absorbed LC nor the void volume of supporting membranes. The thermo-response efficacy of the polymer-alloyed membrane was less than that of the LC-adsorbed membrane. This higher thermo-response efficacy of the LC-adsorbed membrane can be explained not only by less permeability through matrix part of the LC-supporting PTFE membrane but also by higher mobility of the LC molecules due to the phase transition below and above the T,.
The reversible thermo-control of LC-adsorbed membrane The reversible thermo-control of IND permeation through the LCadsorbed membrane was studied. Figure 4 shows a pulsatile permeation of IND through PTDS-adsorbed membrane, in which PTDS content was 48% per membrane weight, in response to the temperature change between 32°C and 38°C with the interval of 2 h in PBS (pH 7.4). The onset of response to the temperature change was very fast and no clear lag time was observed. In most cases, transdermal therapeutic system (TTS) contains organic solvents as solubilizers and/or penetration-enhancers, so estimation of the
-
250
.. .
...
2
4
...
...
6
8
NOZAWA ET AL.
" 0
T i m e
( h )
Figure 4. The pulsatile release of IND through the PTDS-adsorbed membrane in response to temperature changes in PBS.
influence of organic solvents on the T , of LC is very important. The temperature-dependent permeation of IND was investigated in various solution systems. IND permeation experiments through the PTDS-adsorbed membrane were carried out in 40% EtOH aqueous solution (Fig. 5) and 50% N-methyl-2-pyrrolidone aqueous solution (Fig. 6). The dotted lines in Figures 5b and 6b show the permeation coefficients of IND calculated from the steady state regions of the permeation profiles at constant temperatures (32°C and 38°C). The pulsatile release function was reduced by EtOH and IND permeation was not decreased at 32°C as expected. LC was not desorbed from the membrane to 40% EtOH aqueous solution at 40°C for 6 months (data not shown). LC was not dissolved to these organic solution systems. The reduction of the thermo-response efficacy by N-methyl-2-pyrrolidonewas not observed so clearly as expected, which could be due to the difference in hydrophobicity of solvents. W
3
W
14 12
10 8
6 4 %
2
2
0
0
u.l
0
I
0
2
4
6
T i m e ( h )
8
0
2
4
6
8
T i m e ( h )
Figure 5. The influence of EtOH on the pulsatile release of IND through the PTDS-adsorbed membrane in 40% EtOH aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
THERMO-RESPONSIVE POLYMER MEMBRANES. I
251
38°C
0
2
4
6
T i m e ( h )
8
P
0
2
4
6
8
T i m e ( h 1
Figure 6. The influence of N-methyl-2-pyrrolidone on the pulsatile release of IND through the PTDS-adsorbed membrane in 50% N-methyl-2pyrrolidone aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
The influence of EtOH and N-methyl-2-pyrrolidone on the T, of PTDS and MTTS with a higher T, than PTDS was examined by DSC (Fig. 7). The shift of T, to a few centidegrees lower side was observed for 40% EtOH aqueous solution, which might be due to the plasticizing effects of EtOH on alkyl chains of LC. The plasticizing effects was yielded from the hydrophobic interaction between alkyl chains of LC and organic solvents which controls the gel-liquid phase transition behavior of LC. Besides plasticizing effects, a decrease in activity coefficient of IND in membrane might affect the relatively high permeability of IND through the PTDS-adsorbed membrane in 40% EtOH aqueous solution. EtOH immersed into the membranes may increase the membrane permeability of IND. In Figure 7, the T, shift of PTDS with Nmethyl-2-pyrrolidone aqueous solution seems to be nearly the same degree with that of EtOH aqueous solution. However, the thermo-responsive efficacy with N-methyl-2-pyrrolidone solution seems to be higher than that in PBS when compared with Figures 4 and 6. The difference in influences on the thermo-responsive efficacy by N-methyl-2-pyrrolidone and EtOH is not yet cleared. It may be impossible to evaluate the difference only from the DSC data. Since MTTS has a T , near 37-38°C in the organic solutions, it is suitable in the organic solution system. MTTS-adsorbed membrane was prepared by the same method as LC-adsorbed membrane described in experimental part. The MTTS content in MTTS-adsorbed membrane was 48% per membrane weight. Figures 8 and 9 show the IND permeation through MTTS-adsorbed membrane in response to temperature change in 50% N-methyl-2-pyrrolidone and 40% EtOH aqueous solution, respectively. Compared to PTDS adsorbed membranes in PBS (Fig. 5), almost no permeation below T , and low permeation at 38°C were shown with MTTS-adsorbed membranes in N-methyl-2pyrrolidone aqueous solution (Fig. 9). There were two order differences
NOZAWA ET AL.
252 (a)
P TD
(b) M T T S
S
('1
'rewerat-
%mp=.rature ( ' )
203004050
20304080
aq. solution
Figure 7. DSC charts of (a) PTDS and (b) MTTS in PBS, 40% EtOH and 50% N-methyl-2-pyrrolidone aqueous solutions.
38'C~3ZoCi 3 8 ~ ~ 3 2I ° C
ul ,
. .
. .
... . ...
... .. ..
.
.
.
.
-
i 0
2
4
T i m e
6 ( h )
8
k
0O
... .....
,.. 2.
......
.....4 .....6.;u: 8
T i m e
( h )
Figure 8. The influence of EtOH on the pulsatile release of IND through the MTTS-adsorbed membrane in 40% EtOH aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
between the two cases. These differences in permeability might be due to the difference in the chemical structure of LC, the number of alkyl chains. It can be seen that MTTS with three alkyl chains may have higher resistance to IND permeation than PTDS with two alkyl chains. Thus, the thermo-responsive system can be prepared by adsorbing LC, selected by DSC measurement
THERMO-RESPONSIVE POLYMER MEMBRANES. I
253
30°C ,.....
E
f
0
T2 i m 4e
( 6h ) 0
8
O
2
4
T i m e
6
8
( h )
Figure 9. The inf hence of N-methyl-2-pyrrolidone on the pulsatile release of IND through the MTTS-adsorbed membrane in 50% N-methyl-2pyrrolidone aqueous solution. (a) Cumulative amount of IND released, (b) permeation coefficients calculated from the slopes of release profile.
in a given composition of the solution, to the porous hydrophobic polymer membrane. In conclusion, the present study demonstrated that the drug permeation could be sharply on-off controlled in response to minute temperature change by using the LC-adsorbed membrane. The mechanism for drug permeation through LC-adsorbed membrane may be estimated from Figures 3 and 7 and by the microscopic movement of LC-alkyl chain due to thermal stimuli. This thermo-responsive system is applicable to DDS which responds to the change of body temperature and/or the external thermal stimuli. References 1.
2. 3. 4. 5.
6.
7. 8.
M. Brownlee and A. Cerami, “A glucose-controlled insulin-delivery system: Semisynthetic insulin bound to lectin,” Science, 206,1190-1191 (1979). J. M. Anderson and S. W. Kim, in Recent Advances in Drug Delivery Systems, Plenum Press, New York, 1984, pp. 123-136. T. A. Horbell, B. D. Ratner, J. Kost and M. Singh, in Recent Advances in Drug Delivery Systems, Plenum Press, New York, 1984, pp. 209-220. J. Heller, S. H. Pangburn, and D. W. Penhale, Proceed. Intern. Symp. Control. Rel. Bioact. Muter., 14, 107-108 (1987). Y. Okahata, “Lipid bilayer-corked capsule membrane. Reversible signal-receptive permeation control,” Acc. Chem. Res., 19, 57-63 (1986). Y. Okahata, Han-Jim Lim, G. Nakamura, and S. Hachiya, “A large nylon capsule coated with a synthetic bilayer membrane. Permeability control of NaCl by phase transition of the dialkylammonium bilayer coating,” J. Am. Chem. SOC., 105, 48554859 (1983). Y. Okahata, N. Iizuka, G. Nakamura, and T. Seki, “Permeability of fluorescent probes at phase transitions from bilayer-coated capsule membranes,” f. Chern. SOC.,Perkin Truns. 11,1591-1600 (1985). C. Yanagawa, Y. Tanaka, M. Endo, Y. Nakajima, and Y. Okahata, ”Drug delivery system in combination with hyperthermia,” Drug Delivery System, 3(1), 255-261 (1988).
NOZAWA ET AL.
254 9.
10. 11. 12. 13. 14. 15. 16.
17.
Y. Okahata, Han-Jim Lim, and S. Hachiya, "Bilayer coated capsule membrane. Part 2. Photo-responsive permeability control of sodium chloride across a capsule membranes," J. Chem. SOC.,Perkin Trans. 11, 989-994 (1984). Y. Okahata, H. Noguchi, and T. Seki, "Functional capsule membrane. 26. Permeability control of polymer-grafted capsule membranes responding to ambient pH changes," Macromolecules, 20, 15-21 (1987). A.S. Hoffman, "Application of thermally reversible polymers and hydrogels in therapeutics and diagnostics," J. Controlled Rel., 6, 297-305 (1987). S. Freitas and E. L. Cussler, "Temperature sensitive gels as extraction solvents," Chem. Eng. Sci., 42, 79-86 (1987). Y.H. Bae, T. Okano, and S.W. Kim, "Insulin permeation through thermo-sensitive hydrogels," J. Controlled Rel., 9, 271-279 (1989). Y. H. Bae, T. Okano, R. Hsu, and S. W. Kim, "Thermo-sensitive polymers as on-off switches for drug release," Makromol. Chem., Rapid Commun., 8, 481-485 (1987). Y. H. Bae, K. Mukae, T. Okano, and S. W. Kim, "On-off transport regulation through thermosensitive hydrogels," Proceed. Intern. Symp. Contr. Rel. Bioact. Mater., 15, 182 (1988). S. Shinkai, S. Nakamura, K. Ohara, S. Tachiki, 0. Manabe, and T. Kajiyama, "Complete thermocontrol of ion permeation through ternary composite membranes composed of polymerAiquid crystayamphiphilic crown ethers," Macromolecules, 20, 21-28 (1987). Y. Morimoto, K. Sugibayashi, K. Hosoya and W. I. Higuchi, "Penetration enhancing effect of Azone on the transport of 5-fluorouracil across the hairless rat skin," Inf. J. Pharm., 32, 31-38 (1986).
Received December 1988 Accepted August 30,1990
