NO172064B - PROCEDURE FOR ENZYMATIC SYNTHESIS OF PEPTIDES AND PROCEDURE FOR ENZYMATIC SEPARATION OF RACEMIC CARBOXYLIC ACID COMPOUNDS - Google Patents

Description

The present invention relates to a method for the enzymatic synthesis of peptides, and the distinctive features of this method are the steps of: an N-acyl-p-substituted-L-aspartic acid with an α-carboxylate group is condensed with a phenylalanine-lower alkyl ester with an alpha- ammonium group, in an aqueous reaction mixture including a condensation enzyme and thereby forms N-acyl-L-aspartyl-((3-substituted)-L-phenylalanine (lower alkyl ester) in the form of an uncharged peptide,

this uncharged peptide is transported from the aqueous reaction mixture through an antifreeze membrane into a product mixture by osmosis, the membrane preferably comprising a water-immiscible organic liquid immobilized in the pores of a microporous membrane, and

the uncharged peptide in the product mixture is converted into a charged compound form that cannot diffuse back through the membrane to the initial reaction mixture,

with the conversion of the uncharged peptide preferably occurring by enzymatic cleavage.

The invention further relates to the separation of racemic carboxylic acid compounds, and the peculiarity of this method is that it comprises the steps of: hydrolyzing a racemic carboxylic acid ester compound in an aqueous reaction mixture which includes a hydrolyzing enzyme to form a charged enantiomeric compound and an uncharged enantiomeric ester compound in the aqueous reaction mixture, and

transporting the uncharged enantiomeric ester compound from the aqueous reaction mixture through an ion-repelling membrane into a product mixture.

These and other features of the invention appear in the patent claims.

The use of proteolytic enzymes as condensation catalysts for the stereospecific coupling of two L-amino acids to give L,L-peptides is known since they

earlier days for protein chemistry. As early as 1909

Mohr and Strohschein described formation of the water-insoluble

equal dipeptide Bz-Leu-Leu-NH^ by reacting Bz-Leu-OH and H-Leu-Nr^ in the presence of the proteolytic enzyme

papain. E. Mohr and F. Strohschein, Ber 42, 2521 (1909).

The Mohr and Strohschein-type reaction is possible only between diamino acids forming peptide bonds that are exposed

for cleavage with papain or another enzyme used. The condensation reaction's equilibrium between the amino acid reaction components and the peptide product is strongly shifted towards the reacting amino acids. Still can

the condensation reaction is driven to completion by means of mass action if - e.g. the dipeptide product is poorly soluble and precipitates out of the reaction phase.

Because of the commercial importance of certain peptides

and the fact that enzymes are known to catalyze peptide-

formation under mild conditions, much research has been carried out on the enzymatic synthesis of peptides and particularly simple dipeptides. K. Oyama and K. Kihara, Kagaku Sosetsu 35, 195 (1982); K. Oyama and K. Kihara, ChemTech.

14, 100 (1984).

The process for the enzymatic synthesis of the peptide derivative aspartame, described in US Patent No. 4,165,311, hereinafter referred to as the 311 process, involves the thermolysin-catalyzed condensation of N-carbobenzoxy-L-aspartic acid with D,L-phenylalanine -methyl ester and

precipitation of an intermediate complex, D-phenylalanine methyl ester salt of N-carbobenzoxy-aspartame, to drive the reaction to the peptide product side. Further treatment of this intermediate complex allows recovery of D-phenylalanine methyl ester which can be recycled after racemisation, and of the N-carbobenzoxy-aspartame derivative which can be converted to aspartame by removal of the N-carbobenzoxy protecting group. The 311 process must be performed on a batch basis which is cumbersome and complicates the recovery of enzyme. See also K. Oyama, S. Irino, T. Harada, and N. Hagi, Ann. NEW. Acad. Pollock. 434, 95 (1985). Willi Kullmann, Enzymatic Peptide Synthesis (1987).

The N-carbobenzoxy protecting group plays an essential role in the 311 process: by fulfilling the structural requirement set by the active point in thermolysin, and by contributing to the insolubility of the intermediate complex so that the yield of the reaction increases. Removal of the N-carbobenzoxy protecting group from the aspartame derivative must be carried out under mild conditions, e.g.

catalytic hydrogenation, to prevent cleavage of the methyl ester function. Catalytic hydrogenation involves the disadvantage of handling hydrogen gas on a large scale.

Alternative attempts to drive enzymatic condensation reactions to completion are also described in the chemical literature. E.g. the use of organic solvents as reaction media has been found to be effective in increasing peptide product yields although the concomitant decrease in enzyme stability has prevented its use on a large scale, K. Oyama, S. Nishimura, Y. Nonaka, K. Hihara and T .Hashimoto, J. Org. Chem. 46, 5241

(1981); H. Ooshima, H. Mori and Y. Harano, Biotechnology Letters 7, 789 (1985) and K. Nakanishi, T. Kamikubo and

R. Matsuno, Biotechnology 3, 459 (1985).

On the basis of the above-mentioned difficulties in carrying out previously known methods for the enzymatic synthesis of peptides, in particular the requirements for the precipitation of an intermediate product complex and the handling of dangerous reagents, it has been desirable to provide an improved method which avoids these difficulties and which certainly provides effective yields without rapid deactivation of the enzyme catalysts.

The present invention fulfills these purposes.

The foregoing and other objects and advantages are achieved by the invention as can be seen from the subsequent detailed description seen in connection with the attached drawings in which: Figure 1 is a schematic illustration of the enzymatic synthesis of aspartame in accordance with the present invention. Figure 2 is a schematic view of an apparatus for carrying out the method according to the present invention. Figure 3 is a graphical representation illustrating the amount of product (aspartame derivative) found over time in examples 1 and 2. Figure 4 is a graphical representation illustrating the amount of product (aspartame derivative) formed over time in example 4. Figure 5 is a graphical representation illustrating the amount of product (aspartame derivative) formed over time in example 5. Figure 6 is a graphical representation illustrating the amount of product (aspartame derivative) formed over time in example 6. Figure 7 is a graphical representation which illustrates the amount of product (aspartame derivative) formed over time i

example 7.

Figure 8 is a graphic representation illustrating the amount of product (aspartame derivative) formed over time in example 8. Figure 9 is a schematic representation of an apparatus for practicing the present invention and illustrates the containers on the product side as described in example 9. Figure 10 is a graphic representation illustrating the amount of product (aspartame derivative) formed over time in Example 9 with and without the use of an ion exchange resin.

The invention disclosed herein provides a procedure for driving the enzymatic synthesis of peptides in an aqueous reaction mixture to completion by

separating the uncharged peptide intermediate, or derivative thereof, from the reaction mixture by means of a membrane which selectively transports the uncharged peptide out of the reaction mixture into a product mixture. Because the membrane is essentially impermeable to reaction components (charged molecules), removal of the peptide intermediate from the reaction mixture causes a decrease in the peptide concentration at equilibrium which shifts the reaction towards completion by mass action.

The membranes most suitable for practicing the present invention are "Immobilized Liquid Membranes" (ILM) comprising a non-polar liquid embedded

in a microporous support material, preferably a polymeric sheet which is substantially impermeable to the enzymes, reaction components and products. Hydrophobic polymers such as polypropylene are preferred carrier materials. ILM modules can be manufactured during application

of polypropylene hollow fibres. "Celgard" is an example of commercially available polypropylene hollow fibers. Embedding compounds known in the field and polyvinyl chloride pipes or hoses can optionally be used when manufacturing an ILM module. A further polymer for making the microporous support material is "TEFLON" as an example

fluorinated hydrocarbon polymers.

The micropores pass through the carrier material and can be sized so that an immobilized "liquid" will be retained therein by capillarity forces and will not escape from there when the material is exposed, for example, to pressure differentials across the membrane or other common reaction conditions. Based on the above limitations, it is advantageous to maximize the contact area between the immobilized liquid and the reaction mixture to maximize the rate of transfer (flux) of the uncharged peptide product through the membrane. It will be appreciated that the preferred pore size will vary depending on the properties of the particular immobilized liquid, reaction components used, products being prepared and similar factors, and further that the optimal pore size can be determined empirically by experts in the field. A useful discussion of pore size selection can be found in US Patent No. 4,174,374. The use and manufacture of immobilized liquid membranes is described in the following reference is, S.L.. Matson, J.A. Quinn, Coupling Separation and Enrichment to Chemical Conversion in a Membrane Reactor, a paper presented at the AICHE Annual Meeting, New Orleans, Louisiana (November 8-12, 1981) and S.L. Matson and J.A. Quinn, Membrane Reactors in Bioprocessing, Ann. NEW. Acad. Pollock. 469, 152 (1986).

The immobilized liquid retained in the microporous support by capillarity forces should be water-immiscible and a good solvent for the uncharged peptide product to be transported through the membrane (diffuse)

at a reasonable rate, i.e. good transport properties/high flux, while charged or ionized molecules on both the reaction and product side of the membrane are mostly not transported through the membrane in any direction, i.e. good selectivity/ion repulsion.

Selection of the best combination of carrier material and immobilized liquid for use in an enzymatic synthesis of a peptide in accordance with the present invention will depend in part on the nature of the particular reaction components used, desired products and solvents in the system.

The generally preferred immobilized liquids for practicing the present invention include water-immiscible organic solvents, such as e.g. branched and unbranched alcohols with 6 to 12 carbon atoms, e.g. n-decanol, n-dodecanol, n-hexadecanol and mixtures thereof. Also preferred are mixtures of water-immiscible organic solvents including mixtures thereof. Such solvents include, but are not limited to, N,N-diethyldodecanamide, dodecane, and 2-undecanone.

A further type of membrane suitable in the practice of the present invention comprises hydrophobic solid films made from organic polymers such as e.g. polyvinyl chloride or the like. The production of these polymer membranes is well described in the literature, e.g.

O.J. Sweeting, publisher, Science and Technology of

Polymer Films, Interscience, New York (1968), while extensive use of these membranes in the separation of gases and liquids is discussed in S.T. Hwang and K. Kammermeyer, Membranes in Separations, Techniques of Chemistry, Vol. VII, (A. Weissberger, Publisher) John Wiley & Sons, Inc., New York (1975) .

A preferred embodiment of the invention utilizes a membrane reactor/separator system which provides an aqueous reaction mixture or phase circulating in contact with one side of an ILM membrane and an aqueous product phase or mixture circulating countercurrently on the opposite side of the membrane. S.L. Matson and J.A. Quinn, Ann. NEW. Acad. Pollock. 469, 152 (1986). The pH and temperature of the reaction and product phases are maintained at a value that keeps the reaction components in a form that reduces their transport through the membrane at pH between 4.0 and 9.0 to a minimum. Transport of the uncharged peptide intermediate from the reaction phase to the product phase is driven by the concentration gradient across the membrane established by increasing neutral or uncharged peptide concentration in the reaction phase. The transport activity or the flux through the membrane can be significantly improved by the simultaneous, irreversible transformation of the transported peptide in the product phase into a species that cannot diffuse back. E.g. can the latter conversion result in the formation of a polar peptide which cannot diffuse back and thus supplement the driving force to achieve completion of the coupling reaction. An example of a membrane reactor/separator that can be adapted for the practice of the present invention can be found in US Patent No. 4,187,086.

Available alternative membrane reactor/separator configurations that can be adapted to the practice of the present invention include the hollow fiber modules described in UK Patent Application No. 2,047,564 A, and conventional plate and frame type filter devices which are well known in the art.

In addition to selective transport of the uncharged peptide, the membrane provides a barrier between the reaction phase and the product phase that prevents unwanted mixing of and reactions between the components of each phase.

In a preferred membrane reactor/separator in accordance with the present invention, the chemical equilibrium between the reaction components is actually "shifted" through the membrane by carrying out an irreversible conversion of the transported uncharged peptide, into a membrane-impermeable species, on the product side of the membrane. This type of membrane reactor/separator uses a coupled two-enzyme (E 1 and E 2) process of the general type:

The charged reaction components A and B are amino acids and/or small peptides which are condensed by the peptide forming enzyme E"<*>" to form the uncharged intermediate peptide C which is selectively transported through the membrane to the product side. It is clear that the reactive functional groups in the reaction components which do not participate in the desired reaction can be protected or blocked if necessary to prevent unwanted side reactions and/or changes in the products. On the product side of the membrane side, the uncharged peptide C is converted into the charged peptide D which cannot diffuse back through the membrane so that the chemical equilibrium in the reaction mixture is shifted in the direction of the production of more C.

This concept is illustrated in fig. 1 for the special case of the enzymatic condensation of D,L-phenylalanine methyl ester with (N- and -protected) N-formyl-g-benzyl-L-aspartate, in the presence of thermolysin at about pH 5.5.

In the reaction scheme illustrated in fig. 1 is the reaction component A<+> D, L-phenylalanine methyl ester and B_ is N-formyl-3-benzyl-L-aspartate. The reaction components are condensed during the reaction time by the membrane with the help of the enzyme E"<*>" thermolysin which forms the uncharged peptide C. The pH is chosen so that the reaction components are maintained in their charged state and thus reduces their diffusion through the membrane together with the uncharged peptide C to a minimum .

Although the chemical equilibrium for the condensation reaction largely favors reaction component A<+ >and B species, diffusion of the uncharged peptide product C through the membrane to the product side requires constant production of C to maintain the chemical equilibrium on the reaction side of the membrane.

On the product side of the membrane, an esterase enzyme E 2 quickly and irreversibly converts the uncharged peptide C that has diffused through the membrane into charged peptide D that cannot diffuse back to the reaction side. The chemical equilibrium on the reaction side thus becomes effective

"pulled" through the membrane and towards the production of uncharged product C. Peptide D is then converted to aspartame by acid hydrolysis, which removes formyl and benzyl protecting groups, followed by C-terminal end-esterification with methanol.

The above method can be carried out in a counter-flow membrane reactor/separator 10, as shown in fig. 2, by working under controlled conditions with regard to pH and temperature, so that a reaction mixture comprising pea N-acyl-g-substituted-L-aspartic acid, e.g. N-formyl-3-benzyl-L-aspartate, with a free α-carboxylate group

(electronegative species) which is allowed to condense with a reaction component, e.g. D,L-phenylalanine methyl ester,

with a protonated free α-amino group (electropositive species) under the catalytic action of protease

active in the pH range approximately 4.0 to 9.0, to give a fully protected L-aspartyl-L-phenylalanine dipeptide bearing no ionized groups (electroneutral species). On the reaction side of the membrane, the reaction mixture is circulated from the reactor tank 12 by means of the pump 14, through the inlet line 16 through the separator 18 to the outlet line 20 which returns to the reactor tank 12. On the product side of the membrane, a product mixture or uptake mixture that includes fully protected uncharged peptide, e.g. N-formyl-ø-benzyl-L-aspartyl-L-phenylalanine methyl ester which is transported through the membrane, circulated from the product reactor tank 22, by means of the pump 24, through the product supply line 26, through the product side of the separator 18,

to the product outlet line 30. This product mixture includes another enzyme E 2 in the form of an esterase, or other suitable reagent, which can cleave at least one of the protected groups carried by the uncharged peptide so that an electro-charged species is developed which cannot escape the uptake current by diffusing back through the membrane. If an esterase is used, a preferred esterase will have a proteolytic activity and a preferred pH range of from about 6.0 to 9.0. Aminoacylace I, α-chymotrypsin and subtilisin A are examples of esterases which are considered useful in the present invention.

The charged product can be periodically emptied out and/or continuously removed from the product phase by conventional means such as e.g. ion exchange resins, and the remaining effluent can be recycled through the system. The resulting product bound to the ion exchange resin can be desorbed and recovered using conventional procedures.

The selection of suitable deprotection reagents is determined by the chemical nature of the protecting groups applied to the reaction components, such as N-protected-L-aspartic acid, and as indicated above, the choice of protecting groups is in turn dictated by structural constraints imposed at the active site of the condensing enzyme.

In general, the enzymes usable in the practice of the present invention are proteolytic enzymes, sometimes referred to as proteases, which can be divided into two groups.

The most commonly used are endopeptidases, which only cleave internal bonds, and exopeptidases, which only cleave terminal end bonds. Useful enzymes include serine proteinases (EC 3.4.21) such as chymotrypsin, trypsin, subtilisin BNP' and Achromobacter protease, further thiol proteinases (EC 3.4.22), such as e.g. papain, carboxyl proteinases (EC 3.4.23) such as pepsin, and metallo-proteinases (EC 3.4.24) such as thermolysin, prolysin, Tacynasen N (St. caespitosus) and Dispase. Binding of the enzymes to insoluble supports, following procedures well known to those skilled in the art, may be incorporated in the practice of the invention, and although binding of the enzymes is not an essential step, it may be desirable for certain applications. Among the many proteases potentially useful in the practice of the invention, thermolysin (EC 3.4.24) is the most preferred condensing enzyme because of its remarkable thermostability, easy availability, low cost, and wide usable pH ranges between about 5.0 and 9.0. Other preferred proteases include pepsin and penicillo-pepsin (T- Hofmann and R. Shaw, Biochim. Biophys. Acta.

92 543 (1964)) and thermostable protease from penicillium duponti (S. Emi, D. V. Myers and G. A. Iacobucci, Bio-chemistry 15, 842 (1976)). These would be expected to

act at pH approximately 5.5, exhibit good stability at such pH values and do not require the presence of Zn<++> and Ca<++> ions to maintain their activity.

The practical establishment of enantiomeric selectivity, which in the case described above is the production of only the L,L isomer of the peptide C, is directly related to the chosen enzyme, optimal mode of action of the membrane, chemical nature of the carrier material and pH in the aqueous reaction phase.

A preferred membrane for practicing the specific method described above is a microporous polypropylene support material which includes a mixture of n-hexadecanol and n-dodecane immobilized therein. This membrane is available under the trade name

"Type 1 Hollow Fiber Selective Dialysis Mebrane" and is preferred with the reactions in Examples 1 to 4. Another preferred membrane uses "Celgard"

Type 2400 polypropylene hollow fiber which is a

microporous carrier material which includes a mixture of 30% v/v N,N-diethyl dodecanamide in dodecane as a water-insoluble organic liquid which is immobilized by capillarity in the pores of a microporous sheet of "Celgard". This membrane is designated as "Type 2 Hollow Fiber Selective Dialysis

Membrane" and is preferred in the reactions in Examples 5-9. "Celgard" type 2400 polypropylene hollow fibers with a pore size of 0.025 to 0.050 } in and a wall thickness of

25 µm was used in Type 2 Hollow Fiber

Selective Dialysis Membrane". When man

operated at a pH of approximately 5.5, these membranes exhibited a high selectivity, e.g. when carrying out the process in fig. 1, with selectivities in the range of approximately 500:1 (w/w) in favor of the uncharged peptide species have been measured. That is, 500 mg of the uncharged peptide (C) is transported through the membrane for every mg of charged reaction component (A<+> or B~) that is transported.

As mentioned above, for the application of the present invention, it will be necessary or desirable to block or protect various functional groups on the reaction components and products in order to prevent unwanted side reactions that could prevent the production of the desired product and/or reduce its yield and suppress electro-charges in intermediate peptide. Table I sets forth a series of selected combinations of protecting groups and deprotection conditions useful in the practice of the present invention.

As a result of the enantio-selectivity of selected condensation enzymes and the functional discrimination exercised by the membrane, in practicing the invention using N-formyl-L-aspartyl-g-benzyl ester and D,L-phenylalanine- methyl ester, a 99.8% enantiomeric resolution of the racemic phenylalanine methyl ester reaction component is obtained, the L-enantiomer occurring as N-formyl-L-aspartyl(g-benzyl)-L-phenylalanine methyl ester, an aspartame derivative, together with the unreacted D-enantiomer is back in the reaction phase.

The D-phenylanaline methyl ester that is back in the reaction phase can be recovered from there, racemized back to the D,L step, and recycled in the feed. This is

an advantage common to many processes using racemic amino acid reaction components, e.g. 311 process described above.

The economic advantages of the invention are derived at least in part from the use of a racemate feed-reaction mixture rather than a more expensive predissolved L-enantiomer. This advantage is made possible by in situ optical separation of the D,L-phenylalanine methyl ester due to (a) the enantioselectivity of the condensing enzyme and (b) the high selectivity of the membrane in favor of the uncharged species. Preferred methods for the inexpensive synthesis of racemic phenylalanine are those based on the utilization of benzaldehyde via 5-benzalhydantoin sometimes referred to as the Grace process, or the catalytic carbonylation of 'benzyl chloride to phenylpyruvic acid, a procedure developed by the Sagami Chemical Research Center, Tokyo, Japan (sometimes referred to as the Toyo Soda process developed by Toyo Soda manufacturing Co., Ltd., Yamaguchi, Japan).

Example 1.

An aspartame derivative was prepared in accordance with the present invention as follows: To a solution containing 5.02 g (20 mmol) of N-formyl-L-aspartyl-B-benzyl ester, 4.31 g (20 mmol) of L-phenylalanine methyl ester in 100 ml of water, adjusted to pH 5.5, was added 500 mg of thermolysin enzyme (Daiwa Chem. Co., Osaka, Japan) representing a total of 8 x 10 5 proteolytic units.

The resulting clear reaction mixture was incubated for 15 hours at 40°C where the presence of insoluble dipeptide-N-formyl-3-benzyl-L-aspartyl-L-phenylalanine methyl ester was revealed. The resulting mixture was then placed in a 200 ml container, connected to the reaction side, in this case the tube side of an experimental hollow fiber separator of a Bend Research, Inc. "Type 1 Hollow Fiber Selective Dialysis Membrane" giving approximately 9 dm 2 of area. The product side of the membrane (sheath side of the separator) was connected to a source (product) mixture (total volume 200 ml) containing 500 mg of the enzyme Acylase I (EC 3.5.1.14) from Aspergillus Sp.

(Sigma A 2156) at pH 7.5. This enzyme, commonly described as an aminoacylase, was found to act as a C-terminal endesterase, on both N-acetyl- and N-formyl-B-benzyl-aspartame.

Reaction component and product mixtures were circulated at room temperature through the hollow fiber separator in co-current at a rate of 600 ml/min., using peristaltic pumps. The configuration of this apparatus is similar to that illustrated in fig. 2. As both reactions (condensation and ester hydrolysis) are protogenic, the pH of both the reactant and product mixtures falls as the process progresses. Constant pH, both in the reaction component and product mixtures, was maintained by using pH-stats.

The formation of N-formyl-3-benzyl-L-aspartyl-L-phenylalanine was monitored by HPLC. Chromatographic analysis was carried out on a Tracor model 995 instrument together with an LDC spectromonitor II detector set at 254 nm for detection of the amino acids, fully protected product dipeptide and dipeptide. The column used was a NOVA-PAK C^g Radial-Pak cartridge, 8 mm x 10 cm, which was housed in a Millipore Waters RCM-100 Radial Compression Module.

The mobile phase used for the detection of the fully protected dipeptide was a volume/volume mixture of 45% methanol (HPLC purity) 5% tetrahydrofuran

(HPLC purity) and 50% of a 1% KH 2 PO 4 buffer solution. For the detection of the product dipeptide, the mobile phase consisted of a volume/volume mixture of 40% methanol and 60%

of a 1% KH 2 PO 3 buffer solution (1 ml of triethylamine per liter of solvent was added to minimize lag and the pH was adjusted down to 4.3 using 80% phosphoric acid). The flow rate was maintained at 1 ml/min.

HPLC data regarding formation of N-formyl-benzyl-L-aspartyl-L-phenylalanine are summarized in Table II below and are expressed as total amount (mg) of product dipeptide accumulated in the product solution as a function of time.

The value of the uncharged dipeptide concentration at equilibrium corresponding to its saturation point in water at pH 5.5 was found to be approximately 0.05% at 25°C. The amount of uncharged

2

dipeptide that is transported per dm membrane per hour was found to be approximately 18 mg, indicating maintenance of equilibrium required dissolution of insoluble dipeptide and/or dipeptide resynthesis. Almost complete dissolution of insoluble uncharged dipeptide phase in the reaction mixture was observed after about 5 hours, and then the reaction was stopped. A deposition of data in table II is shown in fig. 3. The linear function indicates that transfer of peptide through the membrane takes place in a stable state. The observed formation rate of product dipeptide of approximately 18 mg/dm 2/hour (Table II) was confirmed corresponding to the flux of uncharged dipeptide through the membrane (17 mg/dm 2/hour) measured at 0.05% in water as expected on theoretical basis.

At this point, the described system would be expected to continue in steady state if continuous addition were made to the reaction mixture of the two amino acid reaction components, at a rate of approximately 120 mg/hr each, to keep the system saturated with uncharged dipeptide.

To fully realize the selectivity of the membrane, the insertion of another membrane in series with the first membrane before the contact with the second enzyme may be necessary because the selectivity through a single membrane is lower due to the high amino acid concentration in the reaction mixture.

The product solution (200 mL) was recovered, adjusted to pH 2.5 and cooled at 4°C overnight. The combined precipitate was isolated and recrystallized from MeOH:K 2 O to give 307 mg of N-formyl-3-benzyl-L-aspartyl-L-phenyl-npO

alanine [a]D = -5.6° (C=1.2; EtOH), identical by (IR, 13C-NMR) to an authentic sample prepared by batch hydrolysis of N-formyl-3-benzyl-aspartame with Aspergillus esterase , [a]D 0 c = -5.3 (C=1.3; EtOH).

Example 2

An experiment similar to the experiment in example 1 was carried out with the exception of using 8.63 g of D,L-phenylalanine methyl ester instead of the L-enantiomer. The results are summarized in Table III. Isolation of the uncharged peptide from 200 ml of product solution gave 310 mg of product [a]p = -6.4 (C=1.4, EtOH), identical in all respects (IR, 13C-NMR) to an authentic sample of N -formyl-o-benzyl-L-aspartyl-L-phenylalanine.

A plot of data summarized in Table III (Fig. 3) again showed the presence of a steady state process when the reactor was operated with D,L-phenylalanine methyl ester. The stereospecificity of thermolysin is demonstrated by the exclusive formation of the same L,L-dipeptide as described in example 1. The D-phenylalanine methyl ester retained in the tube phase (reaction mixture) did not inhibit the overall kinetic conditions of the peptide formation.

Example 3

A mixture of 1.0 g of N-formyl-3-benzyl-L-aspartyl-L-phenylalanine, prepared in accordance with example 1.4, 0 ml of water, 4.0 ml of tetrahydrofuran and 1.0 ml of conc. hydrochloric acid (12N) was heated under reflux for 9 hours. The mixture was then cooled and the pH adjusted to 4.0 with 50% NaOH solution. Tetrahydrofuran was then removed by evaporation at less than 35°C and 20 mm Hg. Crystallization was completed at <35°C and 20 mm Hg. Crystallization was completed by storage at 5° for 1 hour, the sample was then filtered, washed with 1 mL ice water and dried under vacuum to give 367 mg of white solid. This material was identical to an authentic sample of aspartyl-phenylalanine by HPLC and IR comparison. [α]2^°= + 12° (c = 0.5; 0.1 N HCl in 50% MeOH).

Aspartyl-phenylalanine was converted to aspartame by treatment with methanol and hydrochloric acid, as described in G.L. Bachman and B.D. Vineyard, US Pat

No. 4,173,562, Example No. 1.

Example 4

An experiment similar to the experiment in example 1 was carried out, with the exception of the use of 5.65 g (20.1 mmol) of N-carbobenzoxy-L-aspartic acid-g-methyl ester and 4.38 g (20.3 mmol ) L-phenylalanine methyl ester as reaction components. The amino acids were dissolved in 100 ml of water, the pH of the solution adjusted to 5.5, and 500 mg of thermolysin Daiwa (8 x 10^ proteolytic units) was added. The solution was pre-incubated for 15 hours at 40°C where a larger amount of N-CBZ-(3-methyl-ester)-L-asp-L-phenylalanine-methyl-ester was precipitated. The suspension was connected to the tube side of a "type 1 Hollow Fiber Selective Dialysis Membrane" (Bend Research Inc.) containing approximately 10 dm 2 of membrane surface area, and the machine was operated at room temperature for 5 hours against a mantle side phase of 200 ml of water containing 500 mg of acylase I (Sigma) at pH 7.5. Accumulation of peptide product in the coat phase was monitored by HPLC

and the results are reproduced in table IV and fig. 4.

After running for 5 hours, the reaction was stopped, the jacket-side phase (200 ml) was isolated, adjusted to pH 2.5 and stored overnight at 4°C. The precipitated product was collected and recrystallized from CH^OHtP^O to give 405 mg (86% recovery) N-CBZ-(g-methyl ester)-L-asp-L-phenylalanine (N-CBZ-iso-APM) [a]^<5>= +6.0° (C = 1.1, EtOH), identical ( 13C-NMR) to an authentic sample of N-CBZ-iso-APM, [a]p = +5, 5° (C = 1.1, EtOH), prepare by chemical coupling and partial esterolysis with acylase I.

As observed earlier in the experiments in examples 1 and 2, the deposit in fig. 4 that accumulation of N-CBZ-iso-APM takes place at a stable rate of approx

200 mg/hour.

The conversion of N-CBZ-iso-APM to AMP can be carried out under the conditions described in Example 3.

Example 5

The aspartame derivative, N-formyl-(3-methyl)-asp-phe was prepared in accordance with the present invention as follows: To a solution containing 6.00 g (35 mmol) of N-formul-(g-methyl)- L-aspartic acid, 10.00 g (48 mmol) L-phenylalanine methyl ester in 100 ml of water, adjusted to pH 7.0, was added 770 mg of thermolysin enzyme (Daiwa

Chemical Co., Osaka, Japan) representing total

1.2 x 10 proteolytic units. The resulting clear solution was incubated for one hour at 40°C when HPLC analysis indicated the presence of 433.8 mg of N-formyl-((3-methyl)-asp-phe-OMe. The solution was cooled to 25°C, pH set to 5.0, and the solution placed in a 200 ml container connected to the tube side by hollow fiber separators ("Type 2

Hollow Fiber Selective Dialysis Membrane”, Bend Research

2

Inc.) which provided approximately 4.5 dm of an ILM

prepared from 30% v/v N,N-diethyl dodecanamide in dodecane. The jacket side of the separator was connected to the product container containing 500 mg of the enzyme ancyle I (EC 3.5.1.14) from Aspergillus Sp. (Aminoacylase AMANO, Nagoya, Japan), at pH 7.5.

The two phases were circulated at 25°C in countercurrent with beat

50 ml/min. (tube phase) and 500 ml/min. (jacket phase) by means of two peristaltic pumps using the configuration illustrated in fig. 2. Constant pH in both phases was ensured by the use of pH-stats.

Formation of N-formyl-(-methyl)-asp-phe was monitored

by HPLC using a Tracor Model 995 instrument coupled with an LDC spectromonitor II

detector set at 254 nm. The column used was a NOVA-PAK C18 radial-Pak cartridge, 8 mm x 100 mm, enclosed in a Millipore Waters RCM-100 Radial Compression Module.

The mobile phases used for the analysis were:

(a) for N-formyl-(8-methyl)-asp-phe-OMe: 40% v/v methanol in 0.1% KH 2 PO 4 buffer pH 4.6; (b) for N-formyl-(g-methyl)-asp-phe: 20% v/v

methanol in 0.1% KH2P0^ pH 4.6. The flow rates used were 1 ml/min. for both analyses.

Data regarding formation of N-formyl-(8-methyl)-asp-phe product dipeptide) are reproduced in the following V and are expressed as total amount (mg) accumulated in 200 ml of coat phase as a function of time.

A provision of data in table V is shown in fig. 5.

At the end of the experiment, HPLC analysis indicated the presence of 283 mg of N-formyl-(g-methyl)-asp-phe-OMe in the tube phase. These values allowed calculation of the amount of peptide synthesized during the operation of the reactor.

3 36 P = peptide transferred in the coat phase = 400 " = 417.4 mg Pq = initial peptide in the tube phase = 433.8 mg

P = peptide back in the tubular phase at the end of the experiment

273 mg.

V of 53.5 mg/hour 100 ml coincides with synthesis view * 3 J

rate (500 mg/hour L) measured for the rate of synthesis in equilibrium studies carried out with N-formyl-(-methyl)-L-aspartic acid and L-phenylalanine methyl ester in the presence of thermolysin.

The product solution (200 mL) was isolated, adjusted to pH 2 with 1 N HCl and extracted twice with 200 mL EtOAc. The combined extracts left a white residue on evaporation which, after recrystallization from EtOAc^hexane, gave 100 mg of N-formyl-(8-methyl)-L-asp-L-phe, [a]^<5> = + 0.70 ° (c, 0.29, MeOH), identical (IR, <13>C-NMR) to an authentic sample prepared by batchwise hydrolysis of synthetic N-formyl-(8-methyl)-L-asp-L-phe- OMe with Aspergillus esterase,

[α]p<5> = +0.80° (c, 0.29, MeOH).

Example 6

The experiment described in example 5 was scaled up in a type 2 Hollow Fiber Dialysis Membrane (Bend Research Ind.) containing 9 dm <2> liquid membrane (30% volume/volume N,N-diethyl-dodecantamide in dodecane). The tube phase contained 40 g L-phe-OMe, 24 g N-formyl-(8-methyl)-L-asp and 3.08 g thermolysin Daiwa (a total of 5 x 10 proteolytic units) in 400 ml water, adjusted to pH 7 ,0. After an incubation period of 1 hour at 40°C, an amount of 1.068 g (2.7 g/L) of N-formyl-(8-methyl)-L-asp-L-phe-OMe was found to be present. The solution was cooled at 25°C, adjusted to pH 5.0 with 1N HCl and connected to the tube side of the hollow fiber separator. The jacket phase consisted of 400 ml of water, pH 7.5, containing 2 g of aminoacylace I (Amano). The two phases were circulated in countercurrent at 25°C for 5 hours, as described in example 5. The results are summarized in table VI and fig. 6.

At the end of the experiment, the amount of N-formyl-(8-methyl)-asp-phe-OMe back in the tube phase was 586 mg.

The highest transport value observed (404 mg/hour) during the first 30 min. is the result of the high initial peptide concentration produced during the pre-incubation period. Deviation from equilibrium caused by the transport of peptide led to the synthesis of more peptide so that a stable state was established after the first hour of the experiment, at an expected level of 200 mg/hour (fig. 6).

Example 7

An experiment similar to Example 5 was carried out, with the exception of the use of 10.00 g of D,L-phenylalanine methyl ester in place of the L-enantiomer. The results,

presented in table VII and fig. 7, clearly shows that the observed rate of formation of N-formyl-(8-methyl)-L-asp-L-phe was half that seen with L-phe-OMe (Example 5, Table V) as expected from the enantioselectivity of thermolysin and the previous results in examples 1 and 2.

Example 8

Membrane-assisted enzymatic resolution of D,L-phenylalanine methyl ester was used in accordance with the present invention in the following way: To a solution of 1.0 g (5.6 mmol) D,L-phenylalanine methyl ester in 100 ml of water, pH 7.5, was added 500 mg of amino-acylase I (Amano Pharmaceutical Co., Nagoya, Japan). The mixture was allowed to react at 25°C for 30 min. and at the end of this period the presence of 266 mg of L-phe (1.6 mmol) and 712 mg of D,L-Phe-PMe (4 mmol) was observed by HPLC analysis. This solution was introduced into the tube side (reaction side) of a "Celgard" hollow fiber-supported ILM separator ("Type 2 Hollow Fiber Selective Dialysis Membrane", Bend Research Inc.) containing 4.5 dm 2 of a 30% v/v N ,N-diethyl dodecanamide in dodecane liquid film in the membrane. The product side of the membrane (sheath side of the separator) was filled with 200 ml of water adjusted to pH 2.0 with dilute HCl. The two phases were circulated through the separator in countercurrent at a rate of 200 ml/min., using peristaltic pumps (Fig. 2). The separation took place by the continuous addition of D,L-Phe-0Me to the tube phase, carried out at a rate of approximately 1 g D,L-Phe-0Me per hour. A total of 30 mL of a 7% solution of D,L-Phe-0Me (2.1 g, 11.7 mmol) in water at pH 7.5 was added over a period of 2 hours. The pH in both phases was kept constant using pH-stats.

The course of the dissolution was followed by HPLC, on samples taken from both phases every 30 min. HPLC instrumentation and procedures are as described in Example 5. The mobile phase used in this case was 20% vol/vol methanol in 0.1% KK^ PO^ buffer pH 4.6, with a flow rate of 1 ml /my. The results, presented in table VII and set out in fig. 8, showed accumulation of L-Phe in the tube phase and of D-Phe-OMe in the sheath phase. At the end of the experiment, both phases were isolated and worked up as follows: (a) Tube phase: The contents (130 ml) were adjusted to pH 8.5 with 1N NaOH and then extracted with 2 x 50 ml EtOAc. The aqueous phase was then adjusted to pH 4.0 with 1N HCl and the solution passed through a 2.5 x 20 cm "Dowex" 50 (NH 2 ) column. After washing with 200 ml of water, the product was eluted with 200 ml of 10% NH 3 OH. The eluate was concentrated to 50 ml under vacuum and the solution freeze-dried. Yield: 249 mg,

white solid.

L-phenylalanine, [a] = -29.2° (c, 2, HO), lit.

(Aldrich) [αJD 25 = D-35.0 (c, 2, H 2 O) optical purity 92%.

(b) Shell phase: (200 ml) was adjusted to pH 8.5

with dilute NaOH and extracted with 2 x 50 mL EtOAc.

The organic extract was dried over anhydrous. Na 2 SO 4 , evaporated to dryness, dissolved in 50 mL water acidified to pH 3.0 with 1N HCl and then freeze-dried to give 989 mg (4.6 mmol) D-Phe-OMe HCl, white solid<off,> [ a]<25>° = _2i.o° (c, 2; EtOH), lit.

(Aldrich) [a] 5D °= -32.4° (c, 2; EtOH), optical purity 83%.

Based on the permeability value of o 32 mg/cm 2 ■min. found for Phe-OMe (water, pH 8, 25°C) on this membrane, the expected flux for a membrane area of 4.5 dm was 2880 mg/h. Table VIII shows that the amount of Phe-OMe transferred in the mantle phase at the end of the first hour was 860 mg, which suggests that the transport was carried out under membrane boundary conditions.

Batch resolution of D,L-Phe-OMe by enantioselective hydrolysis of the methyl ester function catalyzed by subtilisin A (a serine-type alkaline protease) has recently been demonstrated (Shui-Tein Chen, Kung-Tsung Wang and Chi-Huey Wong, J . Chem. Soc. Chem. Commun. 1986 1514).

A hydrolysis is necessary for membrane dissolution of racemic carboxylic acids. A hydrolase which is an esterase such as e.g. aminoacylase I, α-chymotrypsin and subtilisin A can be used to dissolve a D,L-amino acid compound such as D,L-phenylalanine methyl ester. Amino-acylase I is generally preferred for the demethylation of peptides over other esterolytic enzymes such as subtilisin A and α-chymotrypsin with strong proteolytic effects.

If the above-described solution of the D,L-phenylalanine methyl ester is used for the production of a peptide as described with the present invention, the L-phenylalanine which is produced is methylated using standard methylating agents known in the field.

This example can be adapted for the resolution of other racemic carboxylic acids. E.g. a racemic carboxylic acid ester compound in an aqueous reaction mixture that includes a hydrolyzing enzyme can be hydrolyzed to form a charged enantiomeric compound and an uncharged enantiomeric ester compound in the aqueous reaction mixture. The uncharged enantiomeric ester compound is then transported from the aqueous reaction mixture across an ion-repelling membrane in accordance with the present invention including Type 1 or Type 2 "Hollow Fiber Selective Dialysis Membranes" from Bend Research, Inc. In one type of example like for example. e.g. 8, the racemic carboxylic acid ester compound is a D,L-amino acid ester compound, the charged enantiomeric compound is an L-amino acid compound and the uncharged enantiomeric ester compound is a deamino acid ester compound. It can be seen that selection of enzymes and reaction conditions are within the understanding and knowledge of persons with expertise in the field corresponding to the present invention.

Continuous or batch-wise processing agents are provided in this example by the fact that the desired enantiomer of the reaction component can be prepared and added to the reaction mixture.

Example 9

Synthesis of N-formyl-(g-methyl)-L-asp-L-phe according to the present invention was carried out using immobilized amino-acylase I and ion exchange resins to remove permeable products as follows: To a solution of 10.0 g (48 mmol) of L-phenylalanine methyl ester and 6.0 g (35 mmol) of N-formyl-(B-methyl)-L-aspartic acid in 100 L of deionized water, adjusted to pH 7, were added 770 mg of thermolysin (Daiwa Chemical Comapny, Osaka, Japan) which represented a total of 1.2 x 10 proteolytic units. The resulting solution was incubated for 1 hour at 40°C and HPLC analysis then indicated the presence of 383 mg of N-formyl-(g-methyl)-asp-phe-OMe. The solution was cooled to 25°C, pPT set to 5.0, and the solution was placed in a 200 ml container 40 connected to the tube side 46 of a hollow fiber separator 44 ("Type 2 Hollow Fiber Selective Dialysis Membrane", Bend Research, Inc .)

2

containing 900 cm of a hydrophobic liquid membrane prepared from 30% N,N-diethyl dodecanamide in dodecane. The casing side of the separator 48 was arranged as a closed circuit made of a series of connected containers illustrated in fig. 9. The solution returned to separator 44 was adjusted to pH 4.0 to protonate L-phe-OMe which permeated along with the dipeptide N-formyl-(B-methyl)-asp-phe-OMe. Circulation through a column of "Dowex" 50 (Na<+>) 50 removed positively charged L-phe-OMe leaving the uncharged dipeptide in solution. The effluent was adjusted to pH 7.0 and exposed to aminoacylase I, immobilized over DEAE-Sephadex 52 (T. Tosa, T. Mori and I Chibata, Agr. Biol. Chem. 33, 1053 (1969)). The resulting dipeptide N-formyl-(g-methyl)-asp-phe was negatively charged at this pH and was then captured by the Dowex 1 (C1-) resin 54. The effluent was returned to the membrane separator before adjustment to pH 4.0 and closed thus the circuit.

The tube 46 (100 mL) and jacket 48 (500 mL) phases were circulated at 25°C countercurrently through the membrane separator at rates of 50 mL/min. (tube phase) and 120 ml/min.

(mantle phase).

Periodic sampling of the jacket phase was carried out on the effluent from the second enzyme (E2), (Fig. 9, sample port 56, before entering the "Dowex" 1 (Cl-) column 54) and the samples were monitored using HPLC following the procedures described in Example 5. As expected, at this point in the circuit (Fig. 9) no L-phe that could result from the enzymatic hydrolysis of

L-phe-OMe of E2 (see Ex. 8) detected. Only a circulating steady-state level of N-formyl-(8-methyl)-asp-phe (mean concentration 54 mg/L) was observed and reflected the continuous transfer of the dipeptide N-formyl-(6-methyl-APM) across the membrane and its subsequent hydrolysis of E2. The effective capture of the charged bipeptide by "Dowex" 1 resin is indicated by the low concentration of the dipeptide observed at point A throughout the experiment. A similar parallel experiment carried out in the absence of "Dowex" 1 resulted in peptide in the envelope phase, as would be expected from the results discussed above. Again, no L-phe was found in the circulating envelope phase. A comparison of both experiments is seen in Fig. 10 and Table IX.

In addition to separating phenylalanine lower alkyl ester

which also penetrates through the ion-repellent membrane into the product mixture using ion exchange resins as described in this example, aspartic acid which also penetrates through the ion-repellent membrane into the product mixture can be separated using such resins. Species or product that cannot diffuse back through the membrane from the product mixture can be removed using such ion exchange resins. Other separation methods known in the field including, but not limited to electrophoresis, electrodialysis and membrane separations which are equivalent to ion exchange resin separations can also be used, in the present invention.

Immobilization of the condensing enzyme allows the enzymatic reaction in the tube phase to be carried out at an initial pH in the reaction mixture preferred for optimal efficiency of the enzymatic reaction in consideration of the reaction components, products and enzyme including the desired equilibrium of the enzymatic reaction. Optionally, the initial pH in the reaction mixture in the tube phase can be re-adjusted to a pH in the second reaction mixture before this is brought into contact with the membrane so that the pH in the second reaction mixture will maximize the membrane efficiency by transporting the uncharged product from the tube phase through the membrane into the jacket phase. Fig . 9 does not show setting the pH in the initial reaction mixture in the pipe phase to a pH in the second reaction phase.

Similarly, the esterase in the coat phase can be immobilized and

The pH of the product mixture in the jacket phase can be adjusted and re-adjusted as needed to achieve the most efficient process execution. Example 8 and the preceding example provide additional means for efficient continuous or batchwise process execution

using the present invention. With continuous processing, the desired enantiomer of reaction components and any compounds that penetrate together with these can be returned to the tube phase or the reaction mixture.

Claims (8)

1. Process for the enzymatic synthesis of peptides, characterized by the steps of: an N-acyl-B-substituted-L-aspartic acid with a cc-carboxylate group is condensed with a phenylalanine lower alkyl ester with an alpha-ammonium group, in an aqueous reaction mixture including a condensation enzyme and thereby forms N-acyl-L-aspartyl-(.B-substituted)-L-phenylalanine (lower alkyl ester) in the form of an uncharged peptide, this uncharged peptide is transported from the aqueous reaction mixture through an ion-repelling membrane into a product mixture by osmosis , as the membrane preferably comprises a water-immiscible organic liquid immobilized in the pores of a microporous membrane, and the uncharged peptide in the product mixture is converted into a charged compound form that cannot diffuse back through the membrane into the initial reaction mixture, as the conversion of the uncharged peptide preferably takes place by enzymatic cleavage . 2. Method as set forth in claim 1, characterized in that the condensation enzyme is a peptide-forming protease which is active in the pH range approximately 4.0 to 9.0. 3. Process as stated in claim 1, characterized in that the uncharged peptide is converted into a charged species by cleavage with an esterase enzyme active in a pH range from approximately 6.0 to 9.0, and selected from the group consisting of aminoacylase I , α-chymotrypsin and subtilisin A. 4. Method as stated in claim 1, characterized in that the membrane comprises a water-immiscible organic liquid which is immobilized by capillary action in the pores of a microporous sheet, the organic liquid being a solvent for the uncharged peptide, the microporous sheet being made of a material selected from a group consisting of polytetrafluoroethylene and polypropylene and that the solvent comprises at least one compound selected from the group consisting of n-l-decanol, n-hexadecanol, n-dodecanol, N,N-diethyldodecanamide, dodecane, 2-undecane and mixtures thereof. 5. Method as stated in claim 1, characterized in that D,L-phenylalanine methyl ester is condensed with N-acyl-B-substituted-L-aspartic acid with an α-carboxylate group. 6. Method as stated in claim 5, characterized in that the N-acyl group is selected from the group consisting of ØCH2OCO-, -CH=0 and ØCH2CO-, and the B-substituent is selected from the group consisting of ØCH2O-, ter-BuO -, CH3O-, NH2-. 7. Method as stated in claim 1, characterized by the further step of separating N-acyl-^-substituted-L-aspartic acid which also penetrates through the ion-repelling membrane into the product mixture and/or phenylalanine lower alkyl ester which also penetrates the ion-repellent membrane into the product mixture. 8. Process for enzymatic separation of racemic carboxylic acid compounds, characterized in that it comprises the steps of: hydrolyzing a racemic carboxylic acid ester compound in an aqueous reaction mixture that includes a hydrolyzing enzyme to form a charged enantiomeric compound and an uncharged enantiomeric ester compound in the aqueous reaction mixture, and transporting the uncharged enantiomeric ester compound from the aqueous reaction mixture through an ion-repelling membrane into a product mixture.