Amphiphilic Gels with Controlled Mesh Dimensions for Insulin Delivery

desirable well-defined molecules, must have controlled pore sizes, must be ... hydrophilic/hydrophobic content of our network is 50/50, and the diamet...
0 downloads 0 Views 842KB Size
Chapter 19

Amphiphilic Gels with Controlled Mesh Dimensions for Insulin Delivery 1

1

1

1

2

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

J. P. Kennedy , G. Fenyvesi , S. Na , B. Keszler , and K. S. Rosenthal 1

Institute o f P o l y m e r S c i e n c e , The U n i v e r s i t y o f

Akron,

Akron,

OH

44325-3909

2

Northeastern Ohio Universities College of Medicine, Box 95, Rootstown, OH 44272

Introduction Our ultimate goal is the development of a functioning bioartificial pancreas to provide long term correction of blood glucose levels in individuals with Type 1 diabetes. Our device consists of porcine islets or islet fragments enclosed in a fully synthetic precision designed polymer membrane that is flexible, biocompatible and permeable to glucose, insulin, and oxygen, but impermeable to immune cells and proteins greater than ~ 60,000 g/mol, e.g. immunoglobulins. The membranes are designed such that upon implantation, the encapsulated porcine islets are able to sense and regulate the glucose level in the blood of diabetic individuals and at the same time are isolated from the onslaught of the immune system. In this manner, the bioartificial device can remain functional for an extended period of time and then be readily removed and replaced. Encapsulation of islets has been investigated by various groups of researchers employing a variety of materials. A thorough analysis of the requirements discussed in the scientific and patent literature leads us to suggest that the optimum membrane of an implantable biohybrid device for irnmunoisolating islets should exhibit the following characteristics: 1. 2.

290

Biocompatible, biostable, sturdy, rubbery with properties maintained for long periods (~ 1 year) of implantation Nonthrombogenic, hemocompatible, avascular, nonclogging, smooth, immunologically "invisible" surfaces

© 2003 American Chemical Society Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

291

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

3.

Membranes with designed pore sizes (i.e., molecular weight cut-off ranges) for controlled permeability 4. Physiologically satisfactory bi-directional fluxes of glucose, insulin, nutrients and metabolites, with exclusion of immunoglobulins 5. Thin membrane walls to ensure rapid oxygen diffusion 6. Good mechanical properties (strength, modulus, elongation) maintained for long implantation times 7. Easily manufactured into sealable transparent tubules of well-defined volumes 8. Easily retrievable after long implantation 9. Sterilizable (with steam or ethylene oxide) 10. Reasonable cost Available implantable devices for the correction of diabetes have several but not all of these properties. Many researchers utilize alginate gels to microencapsulate individual islets (1-3). The fundamental disadvantage of microencapsulation, however, is that the microcapsules enveloping the islets cannot be completely and reliably retrieved. Our research started some 10 years ago after the development of designed controlled release devices when we recognized that the material requirements for a drug delivery device and that of an implantable macro-irnmunoisolatory membrane are quite similar. For both applications the devices must be biocompatible ("bioinvisible") and biostable, must have smooth non-clogging surfaces even after long implantation times, must allow the diffusion of desirable well-defined molecules, must have controlled pore sizes, must be sterilizable, and must have robust mechanical properties for implantation, use, and easy retrieval. Upon analysis of the requirements for an ideal immunobarrier membrane we concluded that the synthesis of such a membrane was feasible by our recently discovered/patented "living" polymerization technique (4,5). By this technique we have synthesized polyisobutylene-based, novel, hydrophobic crosslinking agents, e.g., methacrylate-ditelechelic polyisobutylene (MA-PIB-MA), which has become a key ingredient for our membranes:

CH =C(CH )—CO—O 2

3

polyisobutylene

O—CO—(CH )C=CH 3

2

(MA-PIB-MA) Immunoisolatory amphiphilic networks have been prepared by crosslinking water-soluble acrylate (e.g., N,N-dimethyl acrylamide) chains with MA-PIBMA (6). Figure 1 helps to visualize the microarchitecture of a network that changes its microstructure depending upon the medium with which it is in

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

292

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

contact, i.e. water, hydrocarbon (HC), or amphiphilic solvent e.g., tetrahydrofuran (THF) ("smart" networks) (7,8). This conformational isomerization upon contact with various media is virtually instantaneous and may account for the biocompatibility of our networks. The hydrophilic polyacrylate moiety in our networks give rise to slipperiness while the hydrophobic PIB moiety ensures elasticity, softness, and strength. The physical-chemical-mechanical-biological properties of our amphiphilic networks have been characterized by a battery of various techniques (9-16). The hydrophilic/hydrophobic content of our network is 50/50, and the diameter of the hydrophobic domains is 2-8 nm diameter (12). These membranes can be formed into rubbery 3-5 cm long, 0.5 cm diameter tubules with very thin walls of defined thickness (0.15-0.2 mm). The tubules can be sealed with methyl cyanoacrylate ("superglue") by crimping or by inserting a plug of the same amphiphilic material. The tubules are optically clear, heat resistant and can be sterilized by autoclaving at 120°C.

Figure 1. Microarchitecture ofAmphiphilic Networks in Various Media

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

293

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

The biocompatibility of the membranes was demonstrated by in vivo implantation in rats for longer than 120 days. The absence of inflammation and fibrous cell growth at the site of implantation, and lack of adherence or abnormal vascularization on or near the device was demonstrated by histology and scanning electron microscopy of explants. These membranes adsorb less fibrinogen, alburnin and Hageman factor (factor XII) than glass, polyethylene or silicone rubber (12). Reduced protein adsorption and cell adhesion also indicated hemocompatibility at blood contacting surfaces (9,14). The present research concerned investigations to elucidate the molecular weight cut-off (MWCO) ranges of the membranes intended for use in our bioartificial pancreas and the diffusional characteristics of glucose and insulin through the membranes. The MWCO ranges were compared to the hydrodynamic (or Stokes) radii (R ) of the various protein markers capable of permeating through the membranes as descriptors of membrane permeability characteristics. s

Experimental Network Synthesis: The synthesis of methacrylate ditelechelic polyisobutylene (MA-PIB-MA) has been described (6,10,17-19). The hydrophilic monomer N,N-dimethylacrylamide, (DMMAm) (Aldrich)) was copolymerized/crosslinked with MA-PIB-MA in THF solution. Active copolymerizing charges together with azo-bisisobutyronitrile (AIBN) were rotated in 20-25 cm long and -4.0 mm inner diameter glass tubes at ~60°C. The centrifugal force moves the active polymerizing charges to the wall of the revolving tube and thus tubular membranes can be prepared. The tubes were removedfromthe reactor, washed sequentially by hexanes, ethanol and distilled water, and conditioned with a physiological buffer solution. The inner diameter (~0.25 cm) and wall thickness (-0.02 cm) of the tubules were determined with a hemocytometer using a light microscope. The hydrophilic/hydrophobic ratios of the tubules were 40/60, 50/50, and 60/40 wt%. The amphiphilic networks are identified by a three character name, e.g. A-10-50 in which the letter refers to the hydrophilic component; A: N,N-chmethyl acrylamide; the 10 refers to the molecular weight of the PIB component; and the 50 refers to the weight percentage of hydrophobic component in the membrane. Pore Size Diameter: The molecular weight cut-off (MWCO) and pore size ranges of tubules were studied by deterrruning the permeability of the membranes by the use of a series of commercially available protein markers of known molecular weight and size, i.e., aprotinin (MW=6,500 g/mol, Rs = 1.5 nm), cytochrome C (MW = 12,400 g/mol, Rs = 1.63 nm), carbonic anhydrase (MW = 29,000 g/mol, Rs = 2.01 nm), and albumin (MW = 66,000 g/mol, Rs = 3.62). Tubules were loaded with a mixture of four different proteins, and were sealed by crimping and cementing with cyano aerylate. The protein-filled tubules were incubated in 5 mL of phosphate buffered saline at 37 °C with

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

294

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

shaking, and aliquots of the surrounding buffer were removed at different times. The aliquots were then analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) to identify the proteins that had permeated the tubes. The proteins were electrophoresed in tris-glycine 4-20% bispolyacrylamide gels (20) and then stained with coomassie blue. Permeability and Diffusion Studies: Tubules of different composition were filled with sterile glucose (40mg/mL) and insulin (40mg/mL) solutions in phospate buffered saline, pH 7.4 (PBS). The tubules were sealed and then incubated with shaking at 37° C in 5 mL of PBS. Aliquots were removed at different time intervals and the concentration of glucose determined colorimetrically by the glucose oxidase method (21) and insulin by the protein dye binding assay (22).

Results and Discussion Pore Size Determination: The permeability of tubules with networks ranging from A-10-50 to A-2.5-60 were compared (Figure 1 and Table 1). A molecular weight cut-off range (MWCO) was obtained by deterrnining which of a series of globular proteins was capable of permeating the membrane of the tubule. This is the conventional parameter which is used to describe the pore size of membranes. In this study we also assessed the permeability of the membranes with regard to the Stokes radius (R ) of the proteins capable of passing into the surrounding medium. This parameter provides a better description of the shape and size of the molecules which permeate the network membrane and hence, its porosity. s

In Figure 1, the M for a set of membranes was plotted with respect to either MWCO (Figure 2A) or Stokes radius (Figure 2B). The M i is the molecular weight of the hydrophilic segment. The lines are drawn to indicate the ranges of proteins which were capable of permeating through the membrane such that the area between the lines includes the proteins which were capable of permeating the membrane. The membranes are transparent to proteins with molecular weights or R below the upper line. The upper line is indicative of the cut-off range for permeability of proteins and hence, opacity. The MWCO or R cut-off are represented as ranges because these membranes do not have well defined pores. C ) h y d r o p h i l i c

C}hydrophi

ic

s

s

The M of the membranes provides the best correlation with the permeability cut-off ranges of the membranes. This parameter can therefore be used to select the appropriate membrane for an insulin delivery biodevice. The membrane should provide optimal permeability to glucose (too small to be of concern) and insulin (~5,700g/mol) but restrict the entry of immunoglobulins (>60,000g/mol). With this criterion, the A-10-50, A-10-60 or A-4.5-40 C 5 h y d r o p h i l i c

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

295

membranes should be appropriate to immunoisolate encapsulated islet cells but still allow optimal permeability of glucose and insulin. Permeability and diffusion studies: Amphiphilic network tubules varying in composition and M drophiiic were evaluated further for permeability and diffusion coefficients for glucose, insulin and albumin (Table II). There was some variability in thickness of the membranes due to experimental conditions but this was factored into the calculations. Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

Cjhy

As expected, glucose was freely permeable through the membranes while insulin was somewhat less permeable and albumin was impermeable. The membranes exhibited superior permeability characteristics. Membranes with higher M c ^ o p ^ c were more permeable to insulin. In comparison, the diffusion coefficients of glucose and insulin through membranes prepared with poly(hydroxyethylmethacrylate) as the hydrophilic segment were only 0.042 and 0.009 x 10 (cm /sec) respectively (15). Earlier diffusion studies performed with different MA-PIB-MA membranes formed into flat membranes (15,18) gave very similar results as described herein for tubules. Our results for insulin permeability and diffusion indicate that the membranes and the tubules are satisfactory for use in an artificial pancreas biodevice. On the basis of these results the A-10-40 network was chosen for further in vitro and in vivo experiments. 6

2

T A B L E 1. Molecular Weight Cut-Offs and Pore Size Ranges

Amphiphilic Network A-10-50 A-10-60 A-4.5-40 A-4.5-50 A-4.5-60 A-2.5-40 A-2.5-50 A-2.5-60

M hydrophilic C5

g/mol 5000 3333 3375 2250 1500 1875 1250 833

Molecular Weight Range g/mol 29,000-66,000 29,000-66,000 29,000 - 66,000 12,400 - 29,000 12,400 - 29,000 12,400-29,000 6,500-12,400 6,500- 12,400

Stokes Radii (R ) Range nm 2.01 - 3.62 2.01 - 3.62 2.01 - 3.62 1.63-2.01 1.63-2.01 1.63-2.01 1.50-1.63 1.50-1.63

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

s

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

3000 4000 5000

c > h y d r

C

4>i

m

" 1

11

C f h

200

3030

H

4000

5DCD

• i

Transparent Zone (Diffusion)

MJiydrophilic

Opaque Zone (No Diffusion)

Figure 2. Correlation between M ophiii and MW (A), and M y d r o p h i i i c and Stokes radius (B). The symbols • and , respectively indicate molecular weights and radii of the protein markers that have or have not diffused through the membrane.

M,4»ydrophilic

1000 2000

B

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

COO

-"1

OS

vo

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

5500 3333 3375 2250 1500 1875 1250 833

g/mol

•^c,hydrophilic

6

2

Glucose xl0 cm /s 1.34±0.03 1.18±0.02 1.01±0.01 1.01±0.01 1.03±0.03 1.06±0.02 1.25±0.03 1.34±0.02 6

2

Insulin xl0 cm /s 0.19±0.04 0.16±0.03 0.14±0.03 0.10±0.01 0.187±0.01 0.076±0.02 0.057±0.01 0.047±0.01

Diffusion Coefficient

7

2

7

2

Insulin xl0 cm /s 0.26±0.02 0.10±0.01 0.14±0.03 0.065±0.02 0.026±0.01 0.036±0.01 0.022±0.01 0.015±0.01

Permeability Glucose xl0 cm /s 1.03±0.02 1.06±0.02 1.71±0.02 1.66±0.02 1.56±0.02 1.13±0.01 1.12±0.02 1.12±0.02

*Wall thickness of tubule was approximately 0.02 cm.

A-10-50 A-10-60 A-4.5-40 A-4.5-50 A-4.5-60 A-2.4-40 A-2.4-50 A-2.4-60

1

Amphiphilic Network

Table 2. Diffusion and Permeability of Glucose and Insulin Through Tubular Amphiphilic Networks

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

298

Summary A series of amphiphilic networks, poly(N,N-dimethylacrylamide)-lpolyisobutylene, with different hydrophilic to hydrophobic ratios were synthesized and thin-walled tubular membranes were prepared. The molecular weight cut-off ranges and the insulin and glucose diffusion coefficients and permeabilities were determined. Our results show a direct correlation between the M h and the permeability of the membrane. According to these studies, M i is the significant parameter that correlates with permeability, allowing selection of the appropriate membrane for development of a bioartificial pancreas. Cj

ydrophilic

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

Cjhydrophi

ic

Acknowledgements This research was supported by grants from the National Science Foundation (DMR-99-88808), the Summa Health System Foundation and a Research Challenge GrantfromNortheastern Ohio Universities College of Medicine. We thank Drs. R.P.Levy and A. Sokolov for stimulating discussions.

References 1.

C.F. Chen, H.T. Chern, F.J. Leu, T.M. Chang, L.R. Shian, A . M . Sun, Artif. Organs, 18, 193, 1994. 2. P.Brunetti, G. Basta., A. Falorni, F. Calcinaro, M . Pietropaolo, R. Calafiore, Int. J. Artif. Organs. 14, 789, 1991. 3. R.P. Lanza, W.M. Kuhtreiber, D. Ecker, J.E. Staruk, W. Chick, Transplantation, 59, 1377, 1995. 4. J.P. Kennedy, and R.Faust, U.S. Patent 5,122,572, June 16, 1992. 5. J.P. Kennedy, B. Ivan, Designed Polymers by Carbocationic Macromolecular Engineering, Theory and Practice, Hanser Publishers, Munich Vienna, New York, Barcelona, p. 35, 1992. 6. D. Chen, J.P. Kennedy, A.J. Allen, J. Macromol. Sci. Chem., A25(4), 389, 1988. 7. B. Keszler, J.P. Kennedy, P.W. Mackey, J. Controlled Release, 25, 115, 1993. 8. D. Park, B. Keszler, V. Galiatsatos, J.P. Kennedy, Macromolecules, 28, 101, 1995. 9. D. Chen, J.P. Kennedy, M.M. Kory, D.L. Ely, J. Biomed. Mat. Res., 23, 1327, 1989. 10. B. Ivan, J.P. Kennedy, P.W. Mackey, Polymeric Drugs and Delivery Systems, R.L. Dunn and R.M. Ottenbrie, eds. ACS Symposium Series No. 469, Chapter 18, 194, 1991.

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.

Downloaded by EAST CAROLINA UNIV on March 27, 2017 | http://pubs.acs.org Publication Date: October 15, 2002 | doi: 10.1021/bk-2002-0833.ch019

299 11. B. Ivan, J.P. Kennedy, P.W. Mackey, Polymeric Drugs and Delivery Systems, R.L. Dunn and R.M. Ottenbrie, eds., ACS Symposium Series No. 469, Chapter 19, 203, 1991. 12. B. Keszler, J.P. Kennedy, N.P. Ziats, M.R. Brunstedt, S. Stack, J.K. Yun, J.M. Anderson, Polym. Bull.,29, 681, 1992. 13. B. Keszler, J.P. Kennedy, J. Polym. Sci. Part A, Polym. Chem.., 32, 3153, 1994. 14. R. Blezer, T. Lindhout, B. Keszler, J.P. Kennedy, Polym. Bull., 34, 101, 1995. 15. S. Shamlou, J.P. Kennedy, R.P. Levy, J. Biomed. Mat. Res., 35, 157, 1997. 16. D. Park, B. Keszler, V. Galiatsatos, J.P. Kennedy, Macromolecules, 66, 901, 1997. 17. P.W. Mackey, Ph.D. Thesis, University of Akron, 1991. 18. S. Shamlou, Ph.D. Thesis, University of Akron, 1996. 19. S.Na, Thesis, The University of Akron 1999. 20. B.D. Hames., B.D. Hames, in: Gel Electrophoresis of Protein, Rickwood, D. ed., 1990, Ch.1. 21. E.Raabo, T.C. Terkildsen, Scand.J.Clin.Lab. Invest. 12, 402, 1960. 22. M.M.Bradford, Analytical Biochemistry, 72, 248-254. 1976.

Bohidar et al.; Polymer Gels ACS Symposium Series; American Chemical Society: Washington, DC, 2002.