Purification of Polymers Used for Fabrication

of an Immunoisolation Barrier“ ALES PROKOPh AND TAYLOR G. WANG‘

hDepartment of Chemical Engineering The Center for Microgravity Research and Applications

School of Engineering Vanderbilt University

Nashville. Tennessee 37232

Immunoisolation by means of capsules based on alginate as a matrix offers a very promising strategy to overcome the host immune response that is observed when implanting xenografts. Experiments in animal models have shown that after implan- tation of alginate-based capsules, loaded with xenografts (cells), foreign-body reac- tions occur, leading to a formation of a thick extracapsular layer of fibrotic and other cells, physically blocking diffusion in and out of capsules and suffocating the embedded xenografts.’ Besides alginates, may other polymers (or their mixtures) are used to produce capsules for the same purpose. Many commercially available polymers, contain, however, impurities which exhibit adverse biological activities and thus contribute to a failure of a xenograftic implant. These impurities are of several kinds, such as monomers, catalysts, initiators, etc., which are present in synthetically derived polymers. These impurities can be mostly removed via dialysis because of their small molecular size. Synthetic polymers, however, are better avoided, if possible, unless they mimic natural components of living cells and of their environment. Pyrogens represent the second kind of impurities. They belong to a group of natural compounds of certain gram-negative bacteria (cell wall) and cause a temperature rise when injected intravenously (hence they were called pyrogens). Chemically, they are represented by a variety of complex lipopolysaccha- rides (LPS) with highly hydrophobic character.* The third group (mitogens) is a rather less defined group of organic compounds which activate many cell types (including lymphocytes, fibroblasts, etc.). Their activation leads to cell proliferation and to subsequent production of lymphokines (cytokines), involved in the end in inflammatory reactions and implant rejection, if mitogens contaminate polymers used to manufacture such implants. This paper deals with removal of the second and third group of polymer contaminants, namely the pyrogens and mitogens.

The removal of the above substances is not a simple procedure. Both groups of impurities are present in traces in the polymers used but elicit powerful biological responses. Both result from contaminants present in the initial raw material as well as due to polymer processing steps. As most of these polymers are extracted and isolated from natural sources (e.g., seaweeds), the starting material is mostly heavily contaminated, unless an axenic or pure culture is used initially (algae, microbial cells). Zimmermann et al.’ and Klock et aL4 have suggested the use of electrophoresis

The financial support from NASA is greatly acknowledged (NASA #NAGW-1707 “Encap- sulation of Living Cells”).

223

224 ANNALS NEW YORK ACADEMY OF SCIENCES

Charcoal treatment -1

Filtration .1

Bead formation -1

Extraction with 1 M Acetic acid (2 times) -1

Water wash -1

Extraction with Citrate (3 times) .1

Water wash .1

Extraction with 50% ethanol/5% acetone (3 times) -1

Extraction with 70% ethanol/5% acetone (3 times) .1

Water wash .1

Dissolution of beads in EDTA .1

Dialysis in 10,000 MWCO dialysis tubes .1

Freeze-drying

FIGURE 1. Summary of extractive purification procedure by Klock et al.' as applied to aigi- nates.

to separate and remove the above contaminants. The size and charge of chemical species involved serves as a basis for polymer purification. However, such processing is only feasible at a lab scale and at a low-gravity environment and does not represent an economic way of purifying these polymers. The same authors have also suggested a multistep chemical extraction procedure for purification of alginate. This method makes use of a gelling capability of alginate. Specifically, small gelled barium alginate beads are successively chemically treated to remove impurities. The advantages of using beads is that they can be easily separated by sedimentation from an extracting agent. Before the barium alginate beads are made, alginate solution is treated with an active carbon (charcoal) and filtered through a microfiltration device to remove the suspended matter. Beads are then produced and treated successively using the following steps (FIG. 1): a) three extractions for 14 h in 1 M acetic acid (pH 2.3) with water wash in between; b) three extractions for 16 h with 50% ethanol (5% acetone) followed by the extractions with 70% ethanol (5% acetone), all interspersed with a water wash (alcohol/acetone step removes partially lipophilic pyrogens); c) dissolution of beads in 0.25 M EDTA solution (pH 10) (by complexing barium and replacing it by sodium ion); and d) dialysis of the polymer solution using 10,OOO MWCO dialysis tube, followed by freeze-drying. Klock et aL4 have demonstrated

PROKOP & WANG: POLYMERS FOR IMMUNOISOLATION 225

that while pyrogens are removed, mitogens are removed as well or reduced to an acceptable level.

The present work describes and extends the above chemical extraction procedure to other gel forming polymers as well as simplifies it and suggests new steps to achieve the same. Recently, after our extensive screening of polymers for their capability to form capsules with biocompatible end-products,’ a need for purification of many other polymers other than alginate emerged. New polymer recipes are now available, using new polymers or their blends to generate biocompatible and mechanically stable capsules, suitable for transplantation?

This procedure makes use of the fact that some other polymers can gel in the presence of small ions (cations). Thus kappa and iota carrageenans can make gel beads in the presence of potassium or calcium salts (KCl, CaC12); gellan in presence of calcium salt (CaClJ. It is important that the gelled beads stay intact almost to the end of the whole chemical extraction procedure, when they are dissolved by a chelating agent. This is not the case for cellulose sulfate and pectin. The procedure described in this paper is used to maintain their beads in the beaded form. This is accomplished by employing gelling cations throughout the extraction procedure, otherwise the beads dissolve.

Next, improvement is based on the use of organic extraction solvents to remove pyrogens. The use of solvents is extended also to nongelling polymers. A combina- tion with chemical extraction method results in the best scenario. For gelling poly- mers, chemical treatment prior to bead formation, followed by ethanol and chloro- form extraction, resulted in superior quality of polymers. For nongelling polymers, a combination of chloroform treatment of a solution was also very good in achieving the desired product quality.

EXPERIMENTAL

Polymers

The following polymers were used: LV Keltone alginate (Kelco/Merck, San Diego, CA), HV Keltone alginate (Kelco), HVCR Keltone alginate (Kelco), Manu- gel GHB alginate (Kelco), UP LVG alginate (Pronova Biopolymer, Drammen, Norway), UP MVG alginate (Pronova), cellulose sulfate (Janssen Chimica, Geel, Belgium), Gelcarin GP-911 NF kappa carrageenan (FMC Corp., Newark, CT), Gelcarin GP-379 NF iota carrageenan (FMC), kappa carrageenan (Wako Pure Chemical Ind., Richmond, VA), Gelrite deacylated gellan (Kelco), carboxymethyl- cellulose, medium molecular weight (Sigma, St. Louis, MO) and low-esterified pectin 315 NH ND (Sanofi Bio-Industries, Paris, France).

Chemicals

Ethanol, acetone, chloroform, ethyl ether and phenol were purchased from Sigma.

Other Materials

Polysulfone cartridge (Minicapsule Filter, 0.22 pm, Gelman, Ann Arbor, MI), Acticlean Etox column (Sepragen, Carlsbad, CA).

226 ANNALS NEW YORK ACADEMY OF SCIENCES

Pyrogen Determination

Pyrogen content was determined by a standard gel-clot LAL method (Associates of Cape Cod, Woods Hole, MA) or by a chromogenic LAL method (LAL test kit QCL 1O00, Whitaker Bioproducts, Walkersville, MD).

Mitogen Determination

Mitogens were determined in a bioassay. Klock et aL4 used a splenocyte culture coupled with a haemacytometer count or MTT assay. In our case, we used normal rat kidney cells (NRK) followed by a DNA determination to assay their growth. The culture was treated with 10 nM of EDTA/NaOH (pH 11.3) for 20 min at 37°C. This treatment released DNA. The EDTA extract was subjected to a fluorometric DNA assay. A TNE buffer and Hoechst 33258 dye (Polysciences, Warrington, PA) was used to permit DNA determination in the range 10-400 ng/ml. TNE buffer consisted of 2 M NaCl (Fisher), 10 mM TrislHCl (Sigma) and 1 mM EDTA (Sigma). A TKO100 dedicated minifluorometer (Hoefer Scientific Instruments, San Francisco, CA) was used to measure the fluorescence. Epidermal growth factor (Sigma) was used as a standard mitogen.

RESULTS

Rationale for Polymer Purification

Several disadvantages of Klock’s procedure can be noted: 1) the considerable length of the procedure lends itself to the possibility of polymer degradation. Based on our observations, a polymer viscosity is often reduced to some degree (10-30%), supporting this notion; 2) our experience shows that in addition to polymer degrada- tion (lower molecular weight) some chemical extraction steps may lead to polymer modification (e.g., partial esterification of carboxyl groups of alginate chain by ethanol, as detected by NMR, note results are not presented in detail), leading to a nonfunctional behavior of alginate. Our procedure attempts to simplify Klock’s procedure in terms of number of steps involved and time (and to extend this procedure to other gelling polymers) and, in addition, incorporates several new steps, namely extraction with organic solvents as well as pyrogen removal by a proprietary ligand coupled to a chromatographic carrier (or by a positively charged cartridge). The latter step also allows for extension to nongelling polymers.

Simplijication and Extension of Klock’s Procedure

TABLE 1 presents a comparison of original Klock’s procedure with some other treatments. Column (1) presents results for Klock’s procedure as applied to a particular alginate. The efficiency of pyrogen removal was high. Procedures (2)-(7) are simpler, omitting some steps (and adding some others instead). Discussion on TABLE 1 can be summarized as follows: 1) Klock’s extractive procedure yields high purity (high pyrogen removal efficiency) but often to a lower molecular mass product (not documented here); 2) extraction times and number of repetitive steps can be often reduced (not documented); 3) single-step charcoal or chloroform treatment

PROKOP & WANG POLYMERS FOR IMMUNOISOLATION 227

TABLE 1. ComDarison of Klock’s Procedure with Other Treatments on Aleinates Pyrogen Level in EUlgram Dry Weight

1“ 2 3 4 5 6 7 Raw material Charcoal

once twice

Filtration Beads/BaC12 Extraction

A-C-E Chloroform

0.1 N NaOH Etox column Dialysis Freeze-drying Final value 9’0 Removal

32,700 6,400 17,600 15,400 15,600 15,000 15,200

.h

640 1,92(Y 1,800 3,590 2,050 1,000 2,000 98 70 90 77 87 93 88

HV Keltone, (2)-(7)-HVCR Keltone. Polysulfone filter. 625 by gel-clot method.

yields satisfactory levels of impurities; 4) “residual” pyrogens (as derived from alkali-treated samples) may be partially due to other polysaccharides present in algae extracts, reactive in the LAL gel-clot detection system (glucans); and 5) glucanase treatment showed (not presented here) that pyrogen “levels” can be further lowered by this enzymatic step. Note, both alkali and enzymatic steps don’t represent a practical purification procedure as product is partially destroyed (polysaccharide) or procedure is expensive. They are considered merely as an analytical procedure.

Different organic solvents are often used to isolate LPS from bacteria.’ It is thus not surprising that chloroform can remove pyrogens. As pyrogens are in part hydrophobic, having a lipid moiety, suitable organic solvent and their mixtures have been tested by us for removal of pyrogens: chloroform and phenolkhlorofod ethyl ether (5/5/8 mixture by volume). Both appear to be effective when applied to dried polymer powder or in a mixture with water. The former method is more advantageous as the removal of solvent is simply by vacuum drying. The use of solvent in a mixture with water is also feasible. However, in some cases, a dispersion is formed, which is difficult to separate.

The original Klock’s procedure was then applied to other gelling polymers (polysaccharides gelled by small molecular ionic species, ionotropic gelation): cellu- lose sulfate (gelled by potassium ions; this is considered a novel observation as there is no literature entry on this subject), kappa and iota carrageenans (by potassium or calcium ions), gelan (by calcium ions), carboxymethylcellulose (CMC) (by aluminum ions) and low-esterified pectin (by calcium ions). TABLE 2 presents some results. It can be concluded that Klock’s procedure works with other gelling polymers and purification efficiency is satisfactory. It was surprising to observe very high initial levels of pyrogens in cellulose sulfate (CS) as the starting material for synthesis (cellulose) should be relatively free of pyrogens because of the cellulose processing conditions. This finding could be due to the CS interference in the assay or due to

228 ANNALS NEW YORK ACADEMY OF SCIENCES

TABLE 2. Extension of Klock's Procedure to Other Polymers (Example) EUlgram Dry Weight

8 9 10 Cellulose Sulfate kappa-Carrageenan Gelrite Gellan

Janssen Fluka Kelco Raw material 63,000 1,250 2,500 Char c o a 1

once twice

Filtration Beads Potassium Potassium Calcium Extraction

Acetic acid-A Citrate-C Ethanol-E

Dialysis Freeze-drying Final value 690 200 250 % Removal 90 84 90

a high pyrogen content in this particular source of CS (no other is available on the market).

A special note is necessary in terms of gelling of CS and low-esterified pectin: the beads are not stable throughout the whole extraction procedure. In order to preserve beads, it was essential to maintain at least 1% (wt) potassium chloride concentration in all chemical extraction steps till the end of purification (for CS) and 0.5% (wt) calcium chloride (for pectate).

TABLE 3 presents pyrogen content in some polymers as measured by us in comparison with some published data. It can be observed that ultra pure alginates are now readily available (since this work was concluded, Kelco has launched an ultra pure product) and that other polymers should be purified, if considered for biomedical applications.

TABLE 4 presents a comparison between pyrogen and mitogen content for two commercial polymers. A Kelco sodium alginate has been purified by a combined chloroform-charcoal treatment and product compared to nonpurified one. Two

TABLE 3. Pyrogen Content in Polymers and Comparison with Published Data (in EUlgram)

Reference LV Keltone Algin (Kelco) 6,300 our data Manugel DPB (Kelco) 99,800 our data UP LVG Alginate (Pronova) 910 our data UP MVG Alginate (Pronova) 310 our data kappa carrageenan (Wako) 3,900 our data Low-esterified pectin 315 NH ND (SBI) 2,400 our data Carboxymethylcellulose, medium M W (Sigma) 1,080 our data Manugel GHB Algin (Kelco) 140,000 Zimmermann eta[.'

Note that all measurements were made on freeze-dried samples or on samples from a manufac- turer and all, perhaps, contain some moisture (around 10%). No correction has been applied.

PROKOP & W A N G POLYMERS FOR IMMUNOISOLATION 229

TABLE 4 Pyrogen and Mitogen Content of Three Commercial Alginate Samples (compared at 1% wt level of polymer)

Pyrogens Mitogens Samde E U l d % Inhibition vs. Control

~ ~~

Purified HV Keltone alginate 22 8 (Kelco) as above) Ultrapure UP LVG alginate (Pronova) 9 8 Nonpurifled HV Keltone alginate (Kelco) 327 15

triplicate runs were used for the mitogen assay. In this assay, higher inhibition of a test culture vs. control signifies higher content of mitogens, as observed for nonpurified alginates.

DISCUSSION

Possible mechanisms involved in Klock’s purification procedure can be summa- rized as follows: 1) pyrogen adsorption in the charcoal capillaries of convenient size. Endotoxin removal by charcoal has been documented in the literat~re.~,’ We have observed that different charcoals behave differently and it is the relative capillary size to LPS molecular size which may govern the effective pyrogen removal. The LPS aggregation and heterogeneity is another factor; 2) extractive removal of low molecular mass species and polymer breakdown products, as well as their adsorption on charcoal; 3) inactivation and extraction of pyrogens via alkalic1° and hydrophilic (in case of ethanol) interactions; 4) removal of heavy metals by a chelating agent (citrate); and 5) simultaneous removal of mitogens, a group of low molecular species acting on many cell types.

Additional mechanisms can be ascribed to our extension of Klock’s procedure: 6) extractive removal of LPS by organic solvents via hydrophobic interactions; and 7) pyrogen removal by a ligand attached to a sorbent/filter. We have not investigated different chemistries of ligands. Mitzner et al.” reported on cellulosic beads with immobilized polyethylene ligand and Hanasawa et ~ 1 . ’ ~ on fibrous carrier with cou- pled polymyxin B. Both ligands exhibit polycationic functionalities. A polymer contamination by a ligand leakage products may be an issue. In our case, we tested a polysulfone filter, exhibiting a stable polycationic surface chemistry and thus devoid of ligand leakage. Nonspecific adsorption of polyanionic polymers may be a problem in this case (as well as with other ligands mentioned above). Although Bommer et ~ 1 . ’ ~ reported on a total capture of LPS in a dialysis system with a closed loop having a polysulfone filter, our data show that some “residual” LPS remains in the polymer solution even after such a filtration step (TABLE 1).

It is well established that some polymers can interfere with the LAL gel-clot method. Among them, notably glucan (laminarin, etc.) can give false readings.I4-l6 Some residual readings observed after purification by several researchers can thus be explained on the basis of apparent pyrogenicity. Adam et a1.” observed residual pyrogen readings even after multiple detergent aided extractions. Although glucans can interfere in the LAL assay, their pyrogenicity in uiuo is very small. Glucans can originate from other algae or yeasts while certain algae species are harvested from the sea for polymer extraction and manufacturing. During this process bacterial

230 ANNALS NEW YORK ACADEMY OF SCIENCES

pyrogens and mitogens can contaminate the product because of the raw material used and because of nonaseptic manipulation of the product.

The following techniques have been developed: 1) a simplified method of purifi- cation of gelling polymers by chemical extraction, particularly suitable for alginates, carrageenans, gelan, pectin and cellulose sulfate; 2) a method of chemical purifica- tion of cellulose sulfate and pectin, featuring the presence of low molecular gelling ions throughout the extractive procedure; 3) a method of purification of gelling and nongelling polymers by means of chloroform extraction of dry polymers, followed by a chemical treatment of their solutions; and 4) “residual” pyrogens may be an analytical artifact due to other interfering polymers.

A simple calculation of safety limits for the assumed device is presented in the APPENDIX.

SUMMARY

A multistep extraction procedure has been tested for purification of natural and semi-synthetic polymers used for fabrication of an immunoisolation barrier for implanting animal cells. This procedure, originally described by Klock et al. for alginates, has been adapted for other gelling polymers to remove pyrogens (endotox- ins) and mitogens. Several other steps have also been tested, resulting in a new and simple procedure for polymer purification, giving satisfactory levels of contami- nation. Endotoxin levels have been quantified by means of chromogenic and gel- clot LAL methods. A simple calculation of the endotoxin permissible levels shows that the quality of purified polymers exceeds FDA specifications for implantable polymers.

ACKNOWLEDGMENTS

Authors appreciate an input of I. Lacik of the analysis of polymer solutions by means of NMR and of S. DiMari in the mitogen assay.

REFERENCES

1. ZIMMERMANN, U. et al. 1992. Production of mitogen-contamination free alginates with variable ratios of mannuronic acid to glucuronic acid by free Row electrophoresis. Electrophoresis 13: 269-274.

2. NOWOTNY, A. 1987. Review of the molecular requirements of endotoxin actions. Revs. Infect. Dis. B(Supp1. 5): S503-S511.

3. ZIMMERMANN, U. et al., Inventors. 1993. Mitogen-free substance, its preparation and its use. World patent 93116111. Date of application: August 19.

4. KLOCK, G. et al. 1994. Production of purified alginate suitable for use in immunoisolated implantation. Appl. Microbiol. Biotechnol. 40: 638-643.

5. PROKOP, A. etal. 1997. Water soluble polymers for immunoisolation. I. Complex coacerva- tion and cytotoxicity. Advances in Polymer Science. In press.

6. PROKOP, A. er al. 1997. Water soluble polymers for immunoisolation. 11. Evaluation of multicomponent systems. Advances in Polymer Science. In press.

7. GI-ANOS, C. et al. 1969. A new method for the extraction of R lipopolysaccharides. European J. Biochem. 9: 245-249.

8. NOLAN, J. P. et al. 1975. Endotoxin binding by charged and noncharged resins. Proc. SOC. Exptl. Biol. Med. 149: 766-770.

PROKOP 8i WANG: POLYMERS FOR IMMUNOISOLATION 231

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

PEGUES, A. S. er al. 1979. The removal of 14C labelled endotoxin by activated charcoal. Int. J. Artif. Organs 2: 153-158.

SOMLYO, B. et al. 1992. Molecular requirements of endotoxin (ET) actions: Changes in the immune adjuvant, the liberating and toxic properties of endotoxin during alkaline hydrolysis. Int. J. Immunopharmac. 14: 131-142.

MITZNER, S. er al. 1993. Extracorporeal endotoxin removal by immobilized polyethylene- imine. Artif. Organs 7: 775-781.

HANASAWA, K. et al. 1989. New approach to endotoxic and septic shock by means of polymyxin B immobilized fiber. Surg. Gynecol. Obstet. 168: 323-331.

BOMMER, J. et al. 1987. No evidence for endotoxin transfer across high flux polysulfone membranes. Clin. Nephrol. 27: 278-282.

MIKAMI, T. etal. 1982. Gelation of Limulus amoebocyte lysate by simple polysaccharides. Microbiol. Immunol. 26: 403-409.

IKEMURA, K. etal. 1989. False-positive result in Limulus test caused by Limulus amebocyte lysate-reactive material in immunoglubulin products. J. Clin. Microbiol. 27: 1965-1968.

ROSLANSKY, P. F. & T. J. NOWITSKY. 1991. Sensitivity of Limulus amebocyte lysate (LAL) to LAL-reactive glucans. J. Clin. Microbiol. 29: 2477-2483.

ADAM, 0. et al. 1995. A nondegradative route for the removal of endotoxin from polysac- charides. Analyt. Biochem. 225: 321-327.

Anonymous. 1987. Guideline on validation of the limulus amebocyte lysate test as end- product endotoxin test for human and animal parenteral drugs, biological products, and medical devices. Food and Drug Administration, Rockville, MD.

APPENDIX

Calculation of endotoxin permissible limits (1) Dosage of encapsulated pancreatic islets per capsule It is assumed that the number of xenogeneic islets needed for implantation to alleviate the need for insulin is on the average 15,000 islets/kg of human body weight. Assuming body weight of 70 kg on the average, the total requirement is 1.05 X l@ islets per patient. Assuming the loading of 2 islets/capsule and capsule size about 0.7 mm, the volume of one capsule is B d3/6 = 0.18 mm3 and the total volume of capsules required is 0.18 X 1.05 X 1,00012 = 47 ml (or grams). Note that the total requirement is very much sensitive to the capsule size. (2) Limit calculation based on the maximum dose Endotoxin limit dose allowed by FDA is 5 EU/kg. 5 EUlkg X 70 kg / 47 ml = 7.4 EU/ml of polymer solution, which is easily attainable. Note that when 7.4 EU/ml is divided by the quantity of a given polymer in the capsule, say 6 mg/ml for alginate, the result is 1.22 EU/mg per dry weight of polymer or 1220 EU/g of dry weight of alginate. Our purified alginate reads 600-3600 EU/g dry weight. Compare this number with the specifications provided for “pyro- gen free” UP Pronova alginates: 300-910 EU/g. This clearly shows that our purifica- tion procedure, if properly selected, is adequate. (3) Limit calculation based on device limits FDA permits 20 EU/device, thus the limit in EU/ml is as follows: 20 EU/device divided by 47 mudevice (one application) = 0.1 EUlml (or 0.05 EU/ ml for one islet per capsule).