WO1990006996A1

WO1990006996A1 - Method and apparatus for catalyst containment in multiphase membrane reactor systems

Info

Publication number: WO1990006996A1
Application number: PCT/US1989/005671
Authority: WO
Inventors: Jorge L. Lopez; Stephen L. Matson
Original assignee: Sepracor, Inc.
Priority date: 1988-12-19
Filing date: 1989-12-18
Publication date: 1990-06-28
Also published as: AU4826890A

Abstract

Apparatus for catalysis in multiphase reaction systems is provided in which catalysts are confined within various membranes without the use of covalent coupling. Confinement of the catalysts is achieved by a combination of a small-pore skin on one side of the membrane and the use of a solvent in which the catalysts are not appreciably soluble on the other. Methods are also provided for the preparation of such apparatus and for their regeneration by incorporation of fresh catalyst.

Description

METHOD AND-APPARATUS FOR CATALYST CONTAINMENT IN MULTIPHASE MEMBRANE REACTOR SYSTEMS

TECHNICAL FIELD

₅ The present invention relates to novel apparatus in which enzymes and other catalysts are confined within membranes for use as membrane reactors in multiphase reaction systems. The invention also relates to a variety i ^* . of membranes having different solvent-wetting character and _Q configurations, and to methods for charging such membranes with catalysts and for regenerating the membrane reactors once the catalysts therein confined have become inactivated through use.

5

BACKGROUND OF THE INVENTION

A method and apparatus are disclosed for the _Q confinement or containment of a catalyst either in an asymmetric membrane or in a composite membrane structure, which is subsequently used to conduct a chemical or biochemical reaction in which multiple phases (e.g. organic and aqueous) are involved. 5 "Immobilization" on solid-phase supports of otherwise homogeneous catalysts (including, but not limited to enzymes, whole cells, and non-biological catalysts such as various metal-containing coordination compounds) is useful because immobilization simplifies the separation of reaction products from catalyst and it facilitates the recovery and reuse of catalyst, which frequently is too expensive for one-time use. However, as discussed below, such catalyst immobilization is often accomplished by covalently attaching the catalyst to the support, generally via irreversible, covalent linking chemistry. As a result, when a supported catalyst becomes deactivated, as biocatalysts such as enzymes inevitably do, it is difficult if not impossible to replace the catalyst without at the same time replacing the support matrix. Replacement of the catalyst/support combination can be a considerably more expensive proposition than replacement of the catalyst component alone because of the cost of the immobilization chemistry^"and of the support^"itself.

Typical supports are membrane structures and particύlate media such as microporous and gel-type beads. Membrane supports are attractive because membrane reactors have a number of performance advantages relative to packed- bed reactors employing catalysts bound to particύlate support media. However, they have the significant disadvantage that^"membrane supports are expensive relative to particulate media. Accordingly, the costs associated with periodically replacing membrane-supported^"catalysts can be significantly higher than is the case with particulate supports.

A significant improvement in membrane bioreactor economics would result from the localization of catalysts in a membrane structure in such a way as to (1) provide effective containment of the catalyst in the membrane, (2) permit high effective catalyst loadings to be realized, and (3) make possible simple catalyst replacement by avoiding the covalent attachment of catalyst to the membrane surface. Such a technology would significantly reduce the cost of catalyst replacement in membrane reactors. Additionally, it could have secondary benefits of avoiding the use of immobilization chemistries that can be expensive and dif icult to control and that sometimes can result in disappointing yields and/or activities of immobilized catalyst.

Many approaches exist for the immobilization of enzymes and homogeneous catalysts on solid supports.

Several techniques including covalent bonding, crosslinking, entrapment, adsorption, and icroencapsulation_:have been developed to render many enzymes water-insoluble. See Fig. IA. Reviews of enzyme immobilization procedures have been published., Zaborsky, O.R. , Immobilized Enzymes, CRC press, Cleveland,.,Ohio (1973); Weetal, H. H. , g . , Immobilized ⁵.Enzymes, Antigens, Antibodies, and peptidesj Enzymology, .. Vol. 1, Marcel Dekker, N-.Y. :<1975) ;j_?Gutcho, S.. J. , _:- Immobilized Enzymes — Preparation and Engineering Techniques, Noyes Data Corp., Park Ridge, N.H. (1974). Several industrial processes currently employ immobilized enzymes or immobilized whole cells. Mosbach, K. , "Application of Immobilized Enzymes," pp. 717-858 in Immobilized Enzymes, K. Mosbach, ed. , Methods An Enzymology XLIV. Academic Press, N.Y. (1976).

The possibility of immobilizing non-biological, ionic homogeneous catalysts as the counterions in ion exchange resins has been recognized for over thirty years. Helfferich, F., Ion Exchange, McGraw-Hill, N.Y. (1971). More recently, homogeneous catalyst complexes have been tied to polymeric and ceramic supports via bifunctional ligands which are simultaneously coordinated with the active metal center and anchored to the solid support. Pittman, C. U. , and Evans, G. 0. Chemtech, 3, 560 (1975); Michalska, Z. M. , and Webster, D. E., Chemtech, 5, 117 (1975); Grubbs, R. H. , Chemtech, 7, 512 (1977); Bailar, J. C. , Jr., Cat. Rev. — sci. Eng. , 10(1) 17 (1974). Examples are shown in Figs. IB and 1C.

Enzymes have been immobilized in membranes (as opposed to particles) in several different fashions. They have been covalently bound or crosslinked within porous membranes (Thomas, D. , "Artificial enzyme membranes: transport, memory, and oscillatory phenomena," pp. 115-150 in Analysis and Control of Immobilized Enzyme Systems, D. Thomas and J. P. Kernevez, eds., American Elsevier, N.Y.

(1976); Thomas, D. , and Caplan, S. R. , "Enzyme Membranes," pp. 351-398 in Membrane Separation Processes , P. Meares, ed., Elsevier, Amsterdam (1976) ; Fernandes, P.M., Constanides, A., Vieth, W. R. , and Vendatasubramanian, K. , Chemtech, .5, 438 (1975); Goldman,-. R., Kedem, 0., and Katchalski, E., Bioche , 1 , 4518,(1968)), attached to membrane surfaces (Emery, A. , Sorenson, J., Kolarik, M. , Swanson, S., and Lim, H., Biotechnol. Bioeng. , XS , ,1359 _, (1974)), entrapped in. membrane gels (Blaedel, W. J. , Kissel, T. R., .and Bogulaski, R. C, Anal. Chem., 44, 2030 (1972)* Blaedel, W. J. , and Kissel, T. R. , Anal. Chem., 47, 1602 (1975)), encapsulated by polymeric or liquid surfactant membrane microcapsules, (Chang, T. M. S., Artificial Cells, Charles C. Thomas, Springfield, IL (1972); Chang, T. M. S., and Kuntarian, N., pp. 193-7 in Enzyme Engineering 4, . G.

B..Brown, G. Manecke, and L. B. Wingard, eds.,.Plenum Press, NY (1978); May, S. W. , and Landgraff, L. M., Biochem. Biophys. Res. Commun., 68,786 (1976); Mohan, R. R. , and Li, N. N., Biotechnol. Bioeng., 16, 513 (1974).) and confined to reaction vessels by ultrafiltration membranes (Porter, M.

C, "Applications of Membranes to Enzyme Isolation and - Purification," pp. 115-144 in Enzyme Engineering 3, L. B. wingard, ed. , Interscience, N.Y. (1972); Closset, G. P., Cobb, J. T., and Shah, Y. T. , Biotechnol. Bioeng., 16, 345 (1974); Madgavkar, A. M. , Shah, Y. T. , and Cobb, J. T. , Biotechnol. Bioeng., 19, 1719 (1977)). The latter type of containment with membranes has been called "figurative immobilization" by Weetal (Messing, R. A., ed., Immobilized Enzymes for Industrial Reactors, Academic Press, NY (1975)), a term which also applies to the localization of an enzyme solution by hollow fibers (Rony, P. R. , J. Am. Chem. Soc. , 94, 8247 (1972); Davis, J. C., Biotechnol. Bioeng., 16, 1113 (1974) ^• Leis, W., and Middleman, S., AICHE J. , 20, 1012

(1974); Waterland, L. R., Robertson, C. R. , and Michaels, A. S., Chem. Eng. Commun., 2, 37 (1975)); U.S. Patent No. 4,266,026 to Breslau and 4,440,853 to Michaels (both all aqueous systems) . Enzyme entrapment outside the fiber (i.e., within the-"shell") , within the porous matrix, and in the fiber lumen have all been demonstrated in fully aqueous systems where reactants and products have been supplied and withdrawn, respectively, in aqueous process, streams (see Figp. 2A, 2B, 2C) . -. . ,, ._

Every conceivable membrane geometry .-- planar 5 films (Kay, T.., j.lly,, M.^jD.,_s Sh^rp,_^ A.^K.,..an(i,/Wilsonj,_t.R. J. H.,.,Nature, 217, 641 (1968); Wilson, R. J. H. , Kay, G. , and Lilly, M. D. , Biochem. J., 108, 845 (1968a); Wilson, R.J.H., Kay, G. , and Lilly, M.D., Biochem. J. , 109,137 (1968b)) and spiral-wrapped membranes (Vieth, W. R. , Wang,

10 s. S., Bernath, F. R. , and Mogensen, A. 0., "Enzyme Polymer Membrane Systems," pp. 175_202 in Recent Developments in Separation Science, Vol. l, N. N. Li, ed. , CRC Press, Cleveland, Ohio (1972); Broun, G. , Thomas, D., Gellf, G. , Domurado, D. , Berjonneau, A. M. , and Buillon, C,

15 Biotechnol. Bioeng., 15, ,359 (1973); Gautheron, D. C. , and Coulet, P. R. , pp. 123-7 in Enzyme Engineering 4, G. B. Broun, G. Manecke, and L. B. Wingard, eds., Plenum Press, NY (1978)), tubular membranes, (Madgavkar, A. M. Shah, Y. T. , and Cobb, J. T. , Biotechnol. Bioeng., 19, 1719 (1977);.

20 Tachauer, E., Cobb, J. T. , and Shah, Y. T. , Biotechnol. Bioeng. , 16, 545 (1974)) and hollow fibers, asymmetric hollow fibers having a single shell layer to retain a component, such as a catalyst, within the hollow fiber on one side of the process stream (as in U.S. Pat. No.

254,266,026 and U.S. Pat. No. 4,440,853) and microcapsules — and nearly all membrane types — porous and nonporous, electrically charged and neutral — ave been considered in connection with enzyme immobilization.

³⁰ SUMMARY OF THE INVENTION

Briefly stated, the present invention operates by trapping a catalyst between two catalyst-impermeable boundaries that it cannot cross under normal membrane

3 ^5 reactor operating conditions. These two barriers to catalyst transport are, generally speaking, (1) a. "skin" or surface layer of said support membrane structure, which contaiiis pores that are sufficiently small so as to prevent the^transport and leakage of catalyst (which will often be either macromolecular or particulate in nature) , and (2 a

⁵ ligtiid-liqtfϊd phafee- boundary'(e.g'r, between -Uquebus^-⁵- solution entrapped in the poέes of the membrane and an organic solvent residing just outside of it) that is located at"the^opposite surface of tHe catalyst-containing-membrane structure. On one surface of the membrane structure, the 0 size of the catalytic species prevents it from diffusing across the skin or surface layer of the asymmetric or composite membrane, while on the other surface the poor solubility of the catalyst in the immiscible liquid phase residing just outside of the membrane prevents loss of the

15 catalytic species from that surface.

As an example, water-soluble enzymes used in two-phase and/or extractive membrane bioreactors could readily be contained in asymmetric, ultrafiltration-type membranes prepared from suitably hydrophilic polymers/"where 0 the membrane skin and the aqueous/organic phase boundary at opposite surfaces of the membrane would serve to confine the biocatalyst to the interior region of the water-wet porous membrane. Alternatively, a two-layer composite structure consisting of a gel-type diffusion membrane (as used, for 5 example, in dialysis) atop a microporous membrane support could also be employed for catalyst containment.

In cases where catalyst lifetime is short compared to that of the membrane support, the present invention makes possible the removal of deactivated catalyst

³⁰ and economical replacement thereof with active catalyst.

Few of the above-cited prior-art immobilization techniques bear much similarity to the present invention, either in structure or in function. Perhaps microencapsulation comes closest to the present invention, 3 *5 involving as it does a selective membrane barrier that prevents loss of catalyst from the interior; generally, selectivity is based on the size of the catalyst relative to the .diameter of the pores in the microcapsule wall. ^*?' However, microcapsules have but a single interface with the process stream (i.e., the microcapsule wall), and as a result the encapsulated catalyst is in contact with **onl a single process stream.^- in contrast, it is su*-* urpose of the present invention to immobilize catalysts in membrane structures that permit close contact of the catalyst with multiple (and often immiscible) process streams.

BRIEF DESCRIPTION OF THE FIGURES

The present invention may be more readily understood by reference to the following figures, wherein: Figs. IA, IB, and 1C are schematic representations of commonly used catalyst immobilization techniques, with conventional methods for enzyme and for non-biological catalyst immobilization shown in Fig. IA and Figs. IB and 1C, respectively; Figs. 2A, 2B and 2C are schematic representations of some conventional hollow fiber membrane/enzyme reactors as have been investigated for fully aqueous systems, with enzyme outside a fiber in a shell, enzyme in the porous matrix of a fiber, and enzyme in the lumen of a fiber shown in Figs. 2A, 2B and 2C, respectively;

Fig. 3 is a schematic representation of an illustrative embodiment of the invention in which a biocatalyst is contained within an inside-skinned, hydrophilic hollow fiber; Fig. 4 is a schematic representation of an illustrative embodiment of the invention in which a non- biological homogeneous catalyst is contained within an outside-skinned, hydrophobic hollow fiber; and

Fig. 5 is a schematic representation of an illustrative embodiment of the invention in which a biocatalyst is contained within a composite, hydrophilic hollow fiber.

DETAILED DESCRIPTION OF THE INVENTION

- Fig. 3 shows a preferred embodiment of the

**^*- *-. '». *^•* *-.^**«₌ r* f_*s^* _^~ "< -'-'• present invention based on the use of a single asymmetric membrane with appropriate surface properties and wetting characteristics. Asymmetric membranes suitable"for the practice of this invention are chosen from the group of anisotropic ultrafiltration (UF) and microfiltration (MF) membranes. These are characterized by a more-or-less thin "skin" layer, which in the case of UF membranes is on the order of 0.1-0.2 μ in thickness, supported atop a much thicker (100 200 μm) and highly porous substrate region. The skin of appropriate asymmetric UF-and MF-type membranes is characterized by sufficiently small pores (10's of Angstroms to perhaps 100 Angstroms in diameter) that macromolecular catalysts such as enzymes and colloidal or particulate catalysts are prevented from diffusing across to be lost to a process stream. Thus, the skin or surface region of the asymmetric membrane forms one catalyst- impermeable boundary.

The required characteristics of the "skin" layer will, of course, depend strongly on the size and other properties of the catalyst that is to be retained. In addition to the ultraporous (or even finely microporous) skin structures contemplated in the above paragraph, it may be advantageous in other situations to employ asymmetric membrane structures characterized by surface layers that resemble swollen gel-type membranes such as the type used in dialysis, or even to employ relatively "tight" membrane materials such as those used in thin-film-composite reverse osmosis membranes (generally for non-biologically catalyzed reactions) .

The pores in the highly porous substrate region underlying the ^i,skin" region of the membrane support can be and preferably are much larger (θ.02 μm to several μm,s in diameter) than those in th& skin.^1* The dl&meter of these substrate pores is chosen with two constraints in mind: (1) the pores must be large enough to accommodate the catalyst, which, in the case of whole cells, may be several microns in diameter and (2) the pores must be sufficiently small (a few microns at most) that capillary forces within them are significant. The latter consideration is important because the porous membrane substructure must be "wet" or impregnated by the "correct" liquid phase (e.g., usually the aqueous phase in the case of an enzyme-catalyzed conversion) , and hence it is important that the intrusion pressure (the pressure at which the "incorrect" fluid can be forced into the pores of the substrate) not be exceeded during operation of the reactor. The intrusion pressure A P is inversely related to pore radius r _r by the Young- LaPlace equation:

^ΔP ⁽2 ₇ /^r _pore>^cos - M

where γ is the interfacial tension between the organic and aqueous phases and θ is the contact angle between the membrane material and the liquid phase contained therein. Typically, substrate pore sizes will be chosen such that the intrusion pressure is at least several psi, to ensure stable operation of the membrane contained catalyst.

The second catalyst-impermeable boundary that defines the catalyst-containing region is defined by the liquid-liquid interface (typically, an aqueous/organic phase boundary) that is located at the surface of the membrane furthest removed from the "skin" layer. Capillarity acts to confine the desired liquid phase to the highly porous membrane matrix, and to exclude the other, immiscible liquid phase. Shown in Fig. 3 is the situation wherein the catalyst is water-soluble £>r hydrophilic (i.e., preferentially wet by water) ,÷-andr'÷ he. membrane material is chosen also to be hydrophilic. In this case, the aqueous/organic phase boundary will reside essentially at the.iOuter, nskinned surface of the membrane, assuming that the pressure difference across the membrane is not in the direction so as to cause ultrafiltration of aqueous solution across the membrane or not so large as to cause intrusion of organic solvent into it. Under these circumstances, a water-soluble or hydrophilic catalyst will be confined to the aqueous interior of the membrane by virtue of its inability to partition into the organic solvent phase and subsequently be carried out of the reactor with it.

It is recognized that the relatively thick microporous substrate region that underlies the relatively thin skin layer of integrally skinned asymmetric membranes will often contain so-called macrovoids or "fingers" that are characterized by dimensions an order of magnitude or more larger than the diameter of the more prevalent micropores comprising the bulk of the substrate. Such macrovoid-containing asymmetric membranes are also within the scope of the present invention. For instance, for the particular case where the membrane is hydrophilic in nature, it is contemplated that the small micropores will be filled with aqueous catalyst-containing solution — retained in the micropores by capillary action — while the much larger macrovoids extending through the substrate will be filled with organic solvent. Intrusion of organic solvent into the macrovoids can be made to occur by applying a small amount of positive pressure to the organic phase sufficient to overcome the relatively small intrusion pressure associated with the larger diameter macrovoids (see Equation I) . In this manner, the area of aqueous/organic interface can be made larger than the superficial geometric area of the outer envelope of the membrane. For purposes of this disclosure. such area of the membrane where the aqueous/organic interface is located will be referred to as one of the surfaces of the membrane. Hydrophobic asymmetric membranes containing macrovoids can also be employed for the containment of organic-soluble catalyst in the microporous region of the substrate, the macrovoid being filled with aqueous solution in this case;

Several variations on this general theme can be identified. For example, ^'the geometry of the asymmetric membrane is largely irrelevant to the present invention, whether the membrane be in the form of flat sheets, tubes, or hollow fibers (although the latter will usually be preferred from the points of vievrof maniϊfacturability and cost) . Moreover, both inside-skinned (as shown in Fig. 3) and outside-skinned hollow fibers, sheets or tubes may be employed, and the membrane material may be either hydrophilic (i.e., water-wet) or hydrophobic (i.e., preferentially wet by organic solvents) . Considering just these two dimensions of skin location (i.e., inside-vs. outside-skinned) and surface properties \i . &. , hydrophobic vs. hydrophilic) , it is apparent that four different configurations can be identified, each of which will have its own set of advantages, disadvantages, and potential applications:

o inside-skinned, hydrophilic, o inside-skinned, hydrophobic, o outside-skinned, hydrophilic, and o outside-skinned, hydrophobic.

For example, the outside-skinned and hydrophobic hollow-fiber of Fig. 4 may have utility in the conduct of phase-transfer catalyzed reactions, where the catalytic species is present predominantly in the organic phase. In this case, a acromolecular tail on the catalyst would assist in retaining it in the membrane matrix. The macromolecular tail may be comprised of, generally speaking, a macromolecule such as a polysaccharide, a protein, water soluble polymer (with hydrophilic membrane) or other polymer (i.e. polyethelene glycol). The macromolecular tails may be bonded or otherwise attached to the catalyst by standard methods,known in tl^e art such_:ιas .©gvalent_rattachment using cyanogen bromide or glutaraldehyde.

More specifically, the purpose of conjugation of. enzymes to peptides, polysaccharides .and polyfunctional, molecules is to increase the effective size of the enzyme and thus its retention by containment in an asymmetric membrane. Conjugation can be performed through a variety of techniques. One of the most used techniques relies on the use.of bifunctional reagents such_, as glutaraldehyde [Broun G.B., Methods in Enzymology, 4 263 (1796)]. Glutaraldehyde is believed to react with lysine residues in a protein. It can also be used to bind a protein to an inert matrix possessing free amino groups or to another protein. The conjugation chemistry using glutaraldehyde depends strongly on the nature of the protein being conjugated, the type of the inert matrix or protein (if is used) the concentrations of protein, glutaraldehyde, and inert matrix, and pH. Cantarella et al. [Biochem J., 179, 15, (1979)] have described a method for the preparation of soluble polymers consisting of acid phosphatase conjugated with human serum albumin. The report suggests that the molecular weight of this conjugate is in excess of 10 . Reichlin [Methods in Enzymology, 70 159 (1980)] has described methods to conjugate adenocorticotropic hormone (molecular weight 4,600) to bovine serum albumin (molecular weight 67,000). Water-soluble carbodiimides are another class of bifunctional reagent commonly used to make protein conjugates. Goodfriend et al. [Science, 144, 1344 (1964)] have employed l-ethyl-3-(3-dimethyl-aminopropyl) carbodiimide (ethyl CDI) to conjugate bradykinin to rabbit serum albumin. The resulting conjugate was found to have 12 moles of bradykinin per mole of albumin.

Peptides can also be conjugated to polysaccharides. Molteni {Methods in Enzymology, 112, 2^*85 (1985) describes a method for activating dextrans with CNBr and subsequently coupling the activated dextrans,with primary »amino groups of drugs.,. Axen efc al_* [Biopolypers, 9, 410 (1970) describes another method to produce soluble enzyme conjugates with polysaccharides. This method describes the use of conventional CNBr activation and chemistry for coupling the activated enzyme to a solid support, followed by digestion of the solid polysaccharide support with the enzyme dextranese. Examples 10 and 11 which follow, demonstrate these polysaccharide conjugation techniques being applied to the apparatus invention described herein. The present invention can be further categorized according to the nature of the catalysts and reactions involved. For example, the present invention is useful both for localizing enzymes that are dissolved in the aqueous phase, as well as those that operate at aqueous/organic- phase boundaries, such as certain of the lipases. Moreover, despite the focus of the above discussion, it is important to note that the utility of the invention is not limited to "bioconversions" such as those catalyzed by enzymes and viable or non-living 'whole cells. Various catalytic reactions of synthetic organic chemistry involve multiple phases (e.g., phase-transfer catalyzed reactions), and the present invention is equally useful in these cases. Finally, both soluble (typically, macromolecular) and particulate catalysts can be localized according to the method of the present invention.

The membrane structure of the present invention is operated in a diffusive mode, i.e., with diffusive transport of reactants into and products out of the catalytic region of the membrane; convective flow through the membrane is to be avoided. Preferably, reactants diffuse in on one side of the structure, and products diffuse out on the other, so that a separation and/or purification is accomplished simultaneously with the catalytic conversion.

The present invention is particularly useful:in the-conduct-Λf catalyzed

or "extractive" membrane reactors, where two process streams — one aqueous and one organic —^■■^■ are located on opposite surfaces of the catalyst-containing membrane and serve the purpose of supplying reactant or removing product. For example, in cases where a reactant is soluble in organic solvents but not in water and where the reaction product is water soluble, the'■-reactant may be red to the reactor via a stream of organic solution directed past/one surface ofr^the membrane of the present invention, while the water-soluble product may be withdrawn from the opposite surface of the membrane via a second aqueous process stream. In other cases, a water-soluble reactant may be supplied via an aqueous stream directed past one surface of the membrane of the present invention while the product is made to partition and is thereby removed into a stream of organic solvent flowing past the other surface of the membrane.

The manner of loading the membrane with catalyst is illustrated here for the case of an enzymatic reaction conducted in the hydrophilic and inside-skinned hollow-fiber membrane of Fig. 3. Initially, aqueous enzyme solution is charged to the shell (or outer) side of the hollow-fiber module and passed though the fiber wall in an ultrafiltration process under a modest pressure difference (i.e., a pressure insufficient to cause disruption or loss of integrity of the skin under "back-flush" conditions) .

During this step, enzyme is accumulated in the porous substrate region of the fiber. Next, excess aqueous enzyme solution is displaced from the shell side of the fiber bundle by flushing it with an immiscible fluid such as air or the organic process solvent. If air or another gas is used in this step, the shell is filled with the organic solvent in a subsequent step. The module is then operated with the organic solvent on. the shell side and an aqueous solution in the lumen of the fiber_^with a slight.excess pressure on the shell side. This pressure difference is too smalil- to1cause intrusioftn of the org -π.ani*c p**hi Calse into. - the „ substrate region of the fiber on the qne hand and is in the wrong direction to cause ultrafiltration of aqueous solution on the other hand.

When the enzyme becomes deactivated and must be recharged, a positive pressure is applied to the aqueous solution on the interior or lumen side of the fibers, thereby causing ultrafiltration (i.e., convective flow) through the membrane and displacement both of organic solvent from the shell side of the module as well as deactivated enzyme from the fiber walls. In order to reload the membrane with catalyst, the steps of the preceding paragraph are repeated.

In another embodiment, a composite membrane structure is employed in place of the asymmetric, integrally skinned membranes contemplated above. As shown in Fig. 5, this composite consists of a thin, permselective surface layer of one material supported on a highly porous and much thicker nonselective substrate membrane, generally fashioned from a different material. Techniques for the fabrication of multilayer composite, laminated and coated membrane structures are well known in the art and are the subject of published review articles. Matson, S. L. , Lopez, J. and J. A. Quinn, Chem. Enq. Scl., 38, 503 (1983); Lonsdale, H. K., J. Memb. Sci. , 10, 81 (1982). For example, composite membrane structures suitable for the localization of various biocatalysts might be manufactured based on the use of thin surface coatings of regenerated cellulose dialysis-type membrane supported on microporous membranes, particularly microporous hollow fibers. Alternatively, a suitably microporous layer might be deposited upon or within a regenerated cellulose hollow fiber. Hydrophilic polyacrylonitrile-based copolymer membranes also appear to be well suited to construction of such types of composite membrane structures. ;

The following examples are provided in order to further illustrate the invention disclosed_^herein. %-τ_*.-

ENZYME CONTAINMENT EXAMPLE EXAMPLE 1

An enzyme solution was prepared by dissolving 50 grams of Candida lipase (Mol. Wt. 100,000; Sigma Chemical 5 Co. Cat # L 1754) in 1.25 liters of water and then filtering this solution to remove the insoluble material. This inter-facially acting enzyme is known to hydrolyze a large number of organic esters, among them, phenoxyacetate methyl ester and amyl acetate.

1_Q The enzyme was loaded into a i 2 c*ustom-made solvent-resistant membrane module fabricated with aniso- tropic polyacrylonitrile (PAN) hollow fibers taken from a PAN-200 hemofilter (ASAHI Medical Co.). The morphology of this membrane is such that it can be described as an

-_|5 asymmetric hydrophilic inside-skinned hollow fiber characterized by 90% rejection of proteins with a molecular weight higher than 50,000. The enzyme solution was recirculated from the shell side to the lumen side and back to the solution reservoir in an ultrafiltration mode. 0 Throughout the loading process the pressure difference between the shell and lumen compartments was kept to 8 psi by adjusting the ultrafiltration rate (generally between 200 to 20 ml/min) . The procedure was completed in one hour. The initial and final enzyme solution activities are shown 5 below in Table 1.

TABLE I 0

Specific activity* Total activity μmole/min-ml μmole/min

Initial solution 7.84 9800

Final solution 0.05 63 5

* determined by measuring the rate of addition of 25 mM NaOH required to maintain the pH at 7.8 in a solution of 20.ml of 0.2 M NaCl + 0.5 ml of phenoxyacetate methyl ester (Aldrich Co.) + 2.5 ml of enzyme solution.

After loading the enzyme to the reactor, recirculation of 1140 ml of phenoxyacetate methyl ester on the shell side was started. The recirculation rate was 150 ml/min and the average pressure on the shell compartment was kept at 6.5 psi by adjusting a^'throttling valve at the shell side exit. On the lumen side 2 liters of 0.1 M NaHCO. were ³ recirculated at a rate of 300 ml/mm. The pH of the aqueous reservoir was kept at 7.8 by addition of 50% NaOH. The reactor was run continuously for five days with daily replacement of the buffer and reaction products solution in the aqueous-phase reservoir. Throughout the experiment, enzyme assays of the buffer reservoir showed no detectable enzymatic activity in the aqueous phase.

The reaction progress and rate were monitored by following the caustic consumption and observing the organic phase reservoir level. At the beginning of the experiment the rate of ester hydrolysis was 3000 μmoles/min and at the end it was 1500 μmoles/min. The phenoxyacetic acid product in the aqueous reservoirs was subsequently recovered by acidifying to a pH of 1.0 with concentrated HC1 and filtering the precipitated solids. After drying, the solids were assayed titrimetrically and found to be 96.3 % acid. A sample of the dried solids was dissolved in a mixture of chloroform and water with a subsequent drying/evaporation of the chloroform phase. The remaining solid from this purification step was titrimetrically assayed to be 99.5% pure with a melting point of 98-103^βC (melting point of phenoxy-acetic acid is 98-100^βC) . The total amount of phenoxyacetic acid recovered was 0.953 kilogram. EXAMPLE 2

Using the same membrane module described in Example 1 with the enzyme still in the membrane, the hydrolysis of amyl acetate was conducted. ^"Recircύlalzion of

— -* • -. . : ^"- ^*'*" fjϊ- ^{*** *}T^*»->- ^" ' '. ' ^{' '}■ '

5400 ml of amyl acetate on the shell side was started and as before, the recirculation rate was §50 ml/min and the average pressure on the shell compartment was kept at 6.5 psi by adjusting a throttling valve at the shell side exit.

On the lumen side 1 liter of 0.05 M NaHCO. was recirculated at a rate of 300 ml/min. The pH of the aqueous reservoir was kept at 7.8 by addition of 5.57 M NaOH. The rate of amyl acetate hydrolysis was determined to be 250 μmoles per minute. Once the rate of amyl acetate hydrolysis was measured the reaction was stopped and the system rinsed 5 with water both on the lumen and the shell side. The reactor was then backflushed for §5 hours with filtered (0.2 μm filter) tap water entering on the lumen side and exiting on the shell side at a flow rate of 50 ml/min. Two liters of 8 M urea were backflushed in the same manner as the tap water and then both, the shell and lumen compartments, were rinsed with 4 liters of distilled water.

The rate of amyl acetate hydrolysis in the reactor was measured in exactly the same manner described above with the exception that 25 mM NaOH was used. The reaction rate was 6.8 μmoles per minute which corresponds to 3% of the initial activity.

EXAMPLE 3

At the conclusion of the membrane regeneration procedure described in Example 2, the module was charged with 20 grams of Candida lipase (the same type and in the same concentration described in Example 1) . Amyl acetate was used as a substrate and the reactor operated in exactly the same manner as described in Example 2. The rate of reaction was measured at 70 μmoles/min.

^■-^t : EXAMPLE 4^"* '

An enzyme^' solution was prepared by dissolving ^10.8 ml of a pig liver esteraεe preparation (Mol. Wt.

150,000, 11 mg/ml, Sigma Chemiύal Co.," Cat # E 3128) in 300 ml of 0.2 M phosphate buffer pH 8.0. This enzyme, falling under the category of an esteraεe, will hydrolyze ethyl butyrate^'dissolved in water, i.e., the reaction is a 0 homogeneous one and does not require an organic/ aqueous interface to be present.

The enzyme was loaded into the same membrane module described in Example 1 by recirculating the enzyme solution from the shell side to the lumen side and back to 5 the solution reservoir in an ultrafiltration mode. Throughout the loading process the pressure difference between the shell and lumen compartments was kept to 9.5 psi by adjusting the ultrafiltration rate (generally between 200 to 20 ml/min) . The'procedure was completed in one hour. 0 The initial and final enzyme solution activities are shown below in Table 2.

TABLE 2 5

Specific activity* Total activity μmole/min-ml μmole/min

Initial solution 49.2 14800

Final solution 0.51 153 0

* determined by measuring the rate of addition of 20 mM NaOH required to maintain the pH at 8.0 in a solution of 20 ml of 0.1 M phosphate buffer pH 8.0 + 0.2 grams of ethyl .butyrate (Aldrich Co.) + 1.0 ml of enzyme solution.

5 After loading the enzyme to the reactor, recirϊulation of 500 ml of ethyl butyrate on the shell side was started. The recirculation rate was 500 ml/min and the average pressure on the shell compartment was kept at 6.5 psi by adjusting a throttling valve at the shell side exit. on the lumen side, 1 liter of 0.2 M phosphate buffer pH 8.0 was recirculated at a rate of 500 ml/min. The pH of the aqueous reservoir was kept at 8.0 by addition _£>£ 6.0 M NaOH. The rate of ethyl butyrate hydrolysis was determined to be 9600 μmoles per minute. Once the rate of ethyl butyrate hydrolysis was measured, the reaction was stopped and the system rinsed-with water both on the lumen and the shell side.

The enzyme was removed from the reactor through the following procedure: - backflushed with 6 liters of distilled water entering on the lumen side and exiting on the shell side at a flow rate of 50 ml/min

- rinse on both the shell and lumen side with a) 4 liters of 1.0 M NaCl - b) 500 ml of 12% (NH₄)₂S0₄. c) 500 ml of 8 M urea d) 2 liters of 1.0 M NaCl.

The rate of ethyl butyrate hydrolysis in the reactor was measured in exactly the same manner described above with the exception that 25 M NaOH was used. The reaction rate was 30 μmoles per minute corresponding to 0.3% of the initial activity in the module.

EXAMPLE 5

An enzyme solution of alpha-Chymotrypsin (Mol. Wt. 23,000, Sigma Chemical Co. Cat # C 4129) was prepared by dissolving 0.5 grams of the enzyme in I liter of 0.1 M K.HPO./l.O M NaCl pH 7.8. This solution was recirculated on the shell side of a 1 m² ASAHI PAN-150 hemofilter (ASAHI Medical Co.) at a flowrate of 50 ml/min for 1 hour. Because there^"was no flow of enzyme solution from the shell to the lumen side, the enzyme was loaded into the membrane solely in a diffusive mode. After draining the shell side of the enzyme solution, recirculation of ϊ liter of silicone oil (Petrarch Systems Inc.) was started on the shell compartment while maintaining a pressure of 9 psi. A 0.2 M solution of N-Benzoyl-L-Ty*rosine Ethyl Ester (BTEE, Sigma Chemical Co.) in 0.1 M K-HPO./l M NaCl pH 7.8 was then passed through the

*_» -i lumen side of the module at a flow rate of 1 liter/min. The activity of the module was calculated by measuring the amount of BT acid that was present in the aqueous effluent from the reactor. The module activity was determined to be 80 μmoles/min. The membrane module was then drained of solvent and buffer and backflushed with 10 liters of 0.1 M phosphate buffer. The membrane module activity was measured in the same manner described above with the exception that the BTEE solution was pumped into the reactor at a rate of 53 ml/min. The activity of the membrane module was 4 μmole/min corresponding to 5% of the initial activity.

EXAMPLE 6

An enzyme solution was prepared by dissolving 100 mg of the same alpha-Chymotrypsin used in Example 4 in 500 l of 0.1 M phosphate buffer pH 7.0. The enzyme was loaded on a 1 m_{2 ASAH1 PAN}_₁₅₀ hemofilter identical to the one used in Example 5 by recirculating the enzyme solution from the shell side to the lumen side and back to the solution reservoir in an ultra-filtration mode. The procedure was completed in 2.5 hours.

After loading the enzyme to the reactor, recirculation of 1 liter of 10 mM BTEE in amyl acetate was started on the shell side. The recirculation rate was 10 ml/min and the average pressure on the shell compartment was kept at 6.5 psi by adjusting a throttling valve at the shell side exit. On the lumen side, 200 ml of 2 mM phosphate buffer pH 7.0 was recirculated at a rate of 250 ml/min. The pH of the aqueous reservoir was maintained at"7.0 by addition of 1 M NaOH. The initial rate of BTEE hydrolysis was determined to be 45 μmoles/min.

EXAMPLE 7

In order to increase the retention^feof alplia- Chymotrypsin by the hollow fiber membrane used in Examples 05-6, the molecular weight of the enzyme was increased by crosslinking with Bovine Serum Albumin (Sigma Chemical Co., Cat # A 4503) using glutaraldehyde and following conventional protocols for such chemistry^'.

540 mg bovine serum albumin and 3.5 ml of 2.5 %

*5 glutaraldehyde were added to 31.5g of 10 mM citrate/20 mM CaC12/pH 5.00 buffer and 560 mg of alpha-chymotrypsin. The solution was kept at 4^βC for 3 hours, after which 1.75 g of glycine was added and the solution allowed to stand at 4°C for an additional hour. Two batches of enzyme were prepared 0 in this manner. Gel permeation chromatography revealed that over 80 % of the protein conjugate had a molecular weight in excess of 100,000. The two enzyme batches were diluted to a final volume of 1 liter with 0.1 M K-HPO /1M NaCl/pH 7.8 buffer. The measured activity of this solution was 101 5 moles/min-ml per standard BTEE assay. The enzyme was loaded on a 1-m AHAHI PAN-150 hemofilter identical to the one used in Example 4 by ultrafiltering the solution once from the shell side to the lumen side at a flow rate of 20 ml/min. After loading the enzyme to the reactor, ⁰ recirculation of 500 ml of 40 mM N-benzoyl-L-tyrosine ethyl ester (BTEE) in octanol (Aldrich Co.) was started on the shell side. The recirculation rate was 500 ml/min and the average pressure on the shell compartment was kept at 6.5 psig by adjusting a throttling valve at the shell side exit. ⁵ On the lumen side, 1 liter of 0.1 M K-HPOl MNaCl buffer pH 7.0 was recirculated at a rate of 500 ml/min. The pH of the aqueous reservoir was maintained at 7.0 by addition of 1 M NaOH. The initial rate of BTEE hydrolysis as catalyzed by the membrane-contained enzyme conjugate was determined to be 42,000 μmoles/hr. 5 EXAMPLE 8

The proteolytic enzyme derived from Aspergillus oryzae and obtained commercially from Amano Enzymes Inc. (tradename SEAPROSE) has a molecular weight of 18-30,000 0 Daltons. This enzyme is not retained as effectively as desired by the polyacrylonitrile membrane in hemofilters manufactured by ASAHI Inc., which are rated at a molecular weight cutoff of 50,000 Daltons. This enzyme has been demonstrated to hydrolyze the sulfomethyl ester of R(-)

15 ibuprofen. In order to effectively entrap the enzyme as a conjugate in this type of membrane, the following experiment was carried out. The enzyme conjugate was prepared by making 4 liters of a solution containing 2 mg/ml of Seaprose enzyme, 1 mg/ml of polyethyleneimine (MW 70,000; Aceto 0 Corp., Flushing NY), and 0.005 M CaCl₂. The pH of the solution was adjusted to pH 7.00 with concentrated HC1. The crosslinking was initiated by adding 16 g of a 25 weight % glutaraldehyde solution. After 40 minutes the pH of the solution was readjusted to pH 7.00 by adding 10% NH40H, 5 followed by addition of 35 g of alanine. The enzyme conjugate was loaded onto a PAN-200 ASAHI hemofilter (ASAHI Medical Co.). This membrane is an asymmetric, hydrophilic, inside-skinned hollow fiber characterized by 90 % rejection of proteins with a molecular weight higher than 50,000. The

³⁰ enzyme solution was ultrafiltered from the shell side to the lumen side and the filtrate was collected. When all of the enzyme solution was ultrafiltered, the shell was pressurized with compressed air at 5 psig. No esterase activity was detected in the filtrate. A liter of a solution made from

35 50 iK racemic ibuprofen sulfomethyl ester (IbuSME) in 0.1 M phosphate buffer pH 7.00 was recirculated through the lumen to .test for esterase activity in the membrane. Activity was measured by monitoring the amount of NaOH needed to maintain the pH of the ester solution at 7.00. The membrane activity was 25,500 moles of ester hydrolyzed per hour, demonstrating 5 retention of the enzyme conjugate during the loading step; The module described above was subsequently incorporated in a membrane reactor process used for the resolution of ibuprofen. One liter of an ester solution consisting of 0.5 moles/L of racemic ibuprofen -sulfomethyl

10 ester and 0.1 moles/L of KH2P04, pH 7.00, was recirculated at a rate of 400 ml/min on the lumen side of the enzyme loaded membrane module. On the shell side,of this module., were recirculated 450 ml of cyclohexane at a rate of 400 ml/min. The shell exit was connected to the shell inlet of

15 separate membrane device used as a back-extractor for removal of the R-ibuprofen reaction product from the cyclohexane phase. The shell-side organic stream exiting from the back extractor module was returned to the organic reservoir. On the lumen side of the back extractor module,

20500 ml of 0.1 N NaHC03, pH 9.5 were recirculated at a rate of 400 ml/min. , with the pH maintained at this value by addition of base. (During the operation of the bioreactor, the ibuprofen acid that is produced by the esterase action of the entrapped enzyme is extracted into the organic phase

25 and subsequently back extracted into the carbonate buffer.) After 46 hours of operation, no more base was consumed in the back-extractor buffer in order to maintain the pH of 9.5, thus demonstrating completion of the hydrolysis reaction as catalyzed by the membrane-contained enzyme

³0 conjugate.

EXAMPLE 9

In this example, Aspergillus oryzae proteases (Prozymes 6 and 10™) obtained from Amano Enzymes Inc. were crosslinked for use in the enzyme membrane reactor of this invention using dimethyl adipimidate obtained from the Aldrich Chemical Company. One sample each of solutions of Prozyme 6. and Prozyme 10* were prepared by dissolving 5 mgs of each enzyme in 2 mis of 500mM sodium phosphate buffer at pH 8.0. Solutions of_dimethyl adipimidate.were "t-h ., prepared as follows [D.R, Doddε, J.B. Jones, "Enzymes in Organic Synthesis 17.", Can. J. Chem., 57,2533 (1977)]. 8 mmols of the adipimidate were dissolved, in approximately 2 mis of water and the pH adjusted to 8.0 by the addition of concentrated sodium hydroxide solution. The final volume was adjusted to 4 is, giving a final concentration of 2 M adipimidate. 10011 aliquots of the dimethyl adipimidate solution were added to one of the Prozyme 6 solutions and one of the Prozyme 10"' solutions every 15 minutes for 1 hour (four additions) , a fresh adipimidate solution being prepared for each addition.

A sample o'f each enzyme mixture was then run on an isoelectric focusing gel,, using ihe Pharmacia Phast* electrophoresis system. The gel system used does not perturb the native structure of the protein sample applied. Samples of the enzymes treated with dimethyl adipimidate showed radical changes in their behavior upon isoelectric focusing. In the case of Prozyme 6™, clear bands were not detected in the gel lane, suggesting that the protein species in the reaction mixture were unable to even enter the gel due to their increased molecular weight. The Prozyme 10* gel lane did show bands, but the nebulous and smeared appearance of the bands indicated that a significant increase in molecular weight had been achieved through the dimethyl adipimidate crosslinking.

In companion experiments performed as controls, solutions of each of these two enzymes were also treated in an exactly analogous manner with ethyl acetimidate, also obtained from Aldrich Chemical Co. Unlike the situation with dimethyl adipimidate, this ethyl acetimidate reagent is incapable of crosslinking these enzymes. As expected, these enzyme preparations appeared identical to untreated enzymes when run on isoelectric focusing gels.

EXAMPLE 10

Dextran dialdehyde was prepared by periodic acid oxidation [following a method described in Organic Reactions, Vol 2., p 363]. The molecular weight of the Dextran was 5000, and it was obtained from Sigma Chemical co. Dextran dialdehyde was collected and air dried. Two grams (2.0 g) of the prepared dextran dialdehyde were then dissolved in 100 ml of 0.05 M phosphate buffer, pH 6.5 by warming at 50^βC for 18 hours. Prozyme 10™ (1.0 g) was dissolved in the dextran dialdehyde solution at room temperature. After the addition of enzyme, 0.4 of sodium cyanoborohydride was added and the solution was stirred for 6 hours. The activity of the solution was measured and found to be 141 U/ml (1 U is the amount of enzyme required to hydrolysis l μmole of substrate per hour) . In the absence of dextran crosslinking the activity is 185 U/ml, thus demonstrating maintenance of enzyme activity during this chemical procedure.

The crosslinked enzyme was then ultrafiltered through a PAN membrane characterized by 95 % rejection of molecules with a molecular weight higher than 50,000 Daltons [the native enzyme has a molecular weight of 18-23,000 Daltons] . The filtrate was assayed for activity. The percent of rejected enzyme was 70 %. In the absence of crosslinking, the amount of rejected enzyme is below 50%, thus demonstrating increased molecular weight and retention.

EXAMPLE 11

Calcium alginate beads were prepared by the dropwise addition of a 2 % (w/w) sodium alginate solution in water into 350 ml of 0.1 M CaC12. The-beads were subsequently activated with cyanogen bromide by adding 50 ml of a 40 mg/ml CNBr solution in water. The pH was maintained at 11 during the activation by the addition of concentrated NaOH. After consumption of NaOH stopped, the beads were washed with water and subsequently filtered and isolated. Cytochrome C (Sigma Chemical Co., 50 mg) was dissolved in water (50 ml) . The alginate beads were added to the bright red Cytochrome C solution. After 1 hour the solution was crystal clear and the beads had a dark red color indicating that all the protein had been coupled to the beads. The protein beads were then washed in phosphate buffer to disrupt the alginate gel. After 18 hours, the phosphate solution had a distinct red color. These observations indicate that the protein was bound to the - alginate polymer resulting in an alginate/Cytochrome C conjugate and subsequently was made soluble by the addition of phosphate ion. This example illustrates yet another chemistry for the preparation of high-molecular-weight protein applicable to enzymes to be contained within the asymmetric membrane structures of the present invention.

Variations of the above described methods which involve minor changes are clearly contemplated to be within the scope of the present invention. In addition, minor variations in the design, or materials of the apparatus of the present invention are also contemplated to be within the scope of the present invention. These modifications and variations of the above invention may be made without departing from it spirit and scope, as will become apparent to those skilled in the art. The specific embodiments described herein are offered by way of example only, and the invention is limited only by the terms of the appended claims.

Claims

What is claimed is:

1. An apparatus for containing a catalyst within an asymmetric membrane between two immiscible liquids said asymmetric membrane, catalyst and immiscible liquids contained within a module comprising: . %

a. an -asymmetric membrane having first and second surfaces and a porous substrate region, said region being located between said first and second asymmetric membrane surfaces;

b. said asymmetric membrane first surface having a skin having pores that are large enough to allow permeation into said porous substrate region by reactants or products but small enough to substantially prevent catalyst leakage, which first surface is in contact with a first liquid that wets said asymmetric membrane; and

c. said asymmetric membrane second surface having pores that are large enough to allow permeation into said porous substrate region by reactant, products and catalyst, which second surface is in contact with a second liquid that is substantially immiscible with the first liquid and in which the catalyst is not appreciably soluble;

d. a catalyst contained within said porous substrate region of said asymmetric membrane;

e. a first liquid; and

f. a second liquid that is substantially immiscible with said first liquid and in which the catalyst is not appreciably soluble said second liquid under greater pressure than said first liquid so as to maintain the first liquid-second liquid interface at said second 5 surface of said asymmetric membrane;

2. The apparatus of claim 1, wherein the catalyst is attached to a macromolecule.

0 3. The apparatus of claim 1, wherein the catalyst is attached to a polysaccharide.

4. The apparatus of claim 1, wherein the catalyst is attached to a protein. 5

5. The apparatus of claim 1, wherein the catalyst is attached to a polymer.

6. The apparatus of claim 1, wherein the 0 catalyst is attached to a water soluble polymer.

7. The apparatus of claim 2, 3, 4, 5 or 6 wherein the catalyst is attached covalently.

5 8. The apparatus of claim 2, 3, 4, 5 or 6 wherein the catalyst is a protease derived from Aspergillus oryzae.

9. The apparatus of claim 2, 3, 4, 5 or 6 ³⁰ wherein the catalyst is alpha-chymotrypsin.

10. The apparatus of claim 7 wherein the catalyst is covalently attached with glutaraldehyde.

3 "^5 11. A method for containing a catalyst within an asymmetric membrane between two immiscible liquid streams comprising:

(a) providing an asymmetric membrane having* (i) a first surface having a skin having pores that are large enough to allow permeation by reactants or products but small enough to substantially prevent catalyst leakage; and (ii) ^'a second surface Graving pores that are large enough to allow permeation by reactants, products and catalyst;

(b) charging a catalyst into said asymmetric membrane by ag ing said catalyst to a first liquid stream and contacting the second surface of said asymmetric membrane with said •- first liquid stream containing said catalyst, said first liquid stream wetting said asymmetric membrane and thereby providing said catalyst and said first liquid stream into said asymmetric membrane; (c) providing said first liquid stream without said catalyst to the first surface of said asymmetric membrane;

(d) replacing said first liquid stream containing said catalyst used in charging said asymmetric membrane at the second surface with a second liquid stream in which said catalyst is not appreciably soluble, which second liquid stream is substantially immiscible with said first liquid stream present at said first surface of said asymmetric membrane; and

(e) providing said second liquid stream under a greater pressure than said first liquid stream so as to maintain the interface between said first and second liquid streams at said second surface; thereby containing said catalyst within said asymmetric membrane between said first and second liquid streams which are immiscible.

12. The method of claim 11, wherein the catalyst is attached to a macromolecule.

13. The method of claim 11, wherein the catalyst is attached to a polysaccharide.

14. The method of claim 11, wherein the catalyst is attached to a protein.

15. The method of claim 11, wherein the catalyst is attached to a polymer.

16. The method of claim 11, wherein the catalyst is attached to a water soluble polymer.

17. The method of claim 12, 13, 14, 15 or 16 wherein the catalyst is attached covalently.

18. The method of claim 11, 12, 13, 14, 15 or 16 wherein the catalyst is a protease derived from Aspergillus oryzae.

19. The method of claim 11, 12, 13, 14, 15 or 16 wherein the catalyst is alpha-chymotrypsin.

20. The method of claim 17 wherein the catalyst is covalently attached with glutaraldehyde.