WO1989001524A1

WO1989001524A1 - Method of preparing phosphatidylglycerol

Info

Publication number: WO1989001524A1
Application number: PCT/US1988/002784
Authority: WO
Inventors: Carl Redemann
Original assignee: Liposome Technology, Inc.
Priority date: 1987-08-14
Filing date: 1988-08-15
Publication date: 1989-02-23
Also published as: IL87449A0

Abstract

A method of producing phosphatidylglycerol (PG) by enzymatic conversion of phosphatidylcholine (PC) in the presence of glycerol. The reaction is carried out in a single-phase reaction system containing phosphatidylcholine (PC), phospholipase D enzyme, glycerol, and a soluble calcium salt in an aqueous solvent containing between about 5 %-25 % of a water-miscible alkyl diether, such as 1,1-dimethoxymethane, 1,2-dimethoxyethane, 1,2-dimethoxypropane, and 1,1-diethoxymethane. The reaction goes through a PG maximum which determines the optimum amount of enzyme added to the mixture. After termination of the reaction, the insoluble calcium salts in the mixture, including the calcium salt of PG, are removed and extracted to obtain purified PG, and/or the reaction mixture, which contains phosphatidic acid (PA), is treated further with glycidol, in a suitable alkylgbher, to covert PA to PG. Also disclosed 656a method for converting PA to PG.

Description

METHOD OF PREPARING PHOSPHATIDYLGLYCEROL

1. Field of the Invention The present invention relates to an improved method for converting phosphatidylcholine (PC) enzymatically to phosphatidylglycerol (PG) by phospholipase D, and more generally, to a single-phase re¬ action system suitable for phospholipase reactions with phospholipids. The invention also relates to an improved method of converting phosphatidic acid (PA) to PG and to a combined reaction method in which PC is first converted to a mixture of PG and PA, and PA is then further converted to PG.

2. References

ALLGYER, T.T., and WELLS, M.A. , "Phospholipase D from

Savoy Cabbage: Purification and Preliminary Kinetic Characterization", Biochemistry, (1979), 1_8 5348- 5353.

ATWAL, A.S., ESKIN, N.A., and HENDERSON, H.M., "Isolation and Characterization of Phospholipase D from Fababeans", Lipids, (1979), l±, 913-917.

DAVIDSON, F.M., and LONG, C, "The Structure of the

Naturally Occurring Phosphoglycerides . 4. Action of Cabbage-Leaf Phospholipase D on Ovolecithin and Related Substances", Biochem. J. (1958) 6 S_, 458-466.

DAWSON, R. .C., "The Formation of phosphatidylglycerol and other Phospholipids by the Transferase Activity of Phospholipase D", Biochem. J., 102, 205-210. HELLER, M. , ..and ARAD, R. , "Properties of the Phospholi¬ pase D ^"*from Peanut Seeds", Biochem. Biophys. Acta, (1970) HO 276-286.

KATES, M. , "Lecithinase Systems ^'in Sugar Beets, Spinach, Cabbage and Carrot", Can. J. Biochem. and Physiol. , (1954) 22 , 571-583.

LEE, S.Y., HIBI, N., YAMANE, T., and SHIMIZU, S.,

"Phosphatidylglycerol Synthesis by Phospholipase D in a Microporous Membrane Bioreactor", J^ Ferment. Tech. , (1985), 6_3, 37-44.

PARSONS, J.G., et al. , J. Lipid Res. ■ (1967) 8 , 696-698.

REDEMANN, C.T., "An Assay for the Transferase Activity of Phospholipase D Using Tritiated Glycerol", (1987) LTI Report TD-LTI-21-03-87.

YANG, S.F., FREER, S., and BENSON, A.A. , "Transphospha- tidylation by Phospholipase D", J^ Biol. Chem. , (1967) , 242., 477-484.

3. Background of the Invention

Phospholipases carry out a variety of specific ester-bond reactions on phospholipids. The specific sites of ester-bond hydrolysis by phospholipase A., A-, C, and D are indicated in the figure below.

Many of the reactions catalyzed by phosp^' olipases have commercial importance in the prepara¬ tion of selected phospholipids or diacylglycerol compounds. For example, phospholipase A., and A~ have been used to prepare phospholipids with specific acyl chain groups. Phospholipase D (PLD) is useful for converting phosphatidy choline (PC) to phosphatidic acid (PA) and, in the presence of a suitable head-group substrate, such as glycerol, PLD can catalyze the transfer of one head group (the OR, head group in the figure above) for another—for example glycerol for choline—to convert PC to phosphatidylglycerol (PG) (Yang) . The latter reaction is an important commercial method for preparing PG.

Since phospholipids themselves are quite in- soluble in conventional aqueous media, such as a low-salt buffered media, it is usually necessary to modify the re¬ action medium to enhance the solubility of the lipid. This can be done, according to one prior art method, by adding a detergent such as sodium dodecyl sulf te (Atwal), Triton-X or deoxycholate to the reaction, in an amount sufficient to suspend the phospholipid in a predominantly micellar form. This approach is not practical for com¬ mercial production of phospholipids or phospholipid derivatives, however, because of the difficulty in remov- ing the detergent from the end product.

Alternatively, the phospholipid substrate may be dissolved in a lipophilic organic solvent which is added to an aqueous solution of the enzyme, to form a two-phase reaction mixture. The two-phase system is shaken during reaction, to promote contact between lipid and enzyme.

Many researchers have used ethyl ether as the second-phase solvent (e.g., Davidson, 1958), and others have reported that di-isopropyl ether or n-butyl ether are effective (Lee) . After a suitable reaction period, the organic phase material is separated from the aqueous phase, and the desired PG product is isolated from other components in the organic phase extract, typically by chromatographic methods .. The system has a number of limitations which tend to increase the cost of the product PG. First, the method requires solvent separation, followed by a -4-

chromatographic lipid-isolation procedure, making product isolation somewhat laborious and expensive. Secondly, the organic-phase solvent, the presence of oil-emulsion inter¬ faces, and the need to agitate the mixture during re¬ action, all have the potential for inactivating the PLD enzyme relatively rapidly. Finally, the ether solvents which have shown best enzyme conversion efficiency in the reaction are volatile and highly flammable, making the method difficult to scale up for commercial production.

4. Summary of the Invention

It is therefore one object of the present inven¬ tion to provide, in a reaction system for enzymatically treating a phospholipid with a phospholipase, an improved method which substantially overcomes or reduces problems associated with the prior art.

A more specific object of the invention is to provide such a method for the preparation of PG by enzymatic conversion of PC by phospholipase D in the pres- ence of glycerol in a single-phase reaction mixture.

A related object of the invention is to provide such a method which allows for efficient, non- chrσmatographic isolation of the PG product.

Still another object of the invention is to provide a method of converting PA to PG, particularly PA produced as a by-product of the PC to PG reaction of the invention.

In the method for producing PG by enzymatic conversion of PC, a mixture of PC, PLD, glycerol, and a soluble calcium salt are prepared in an aqueous solvent containing between about 5%-25% of a water-miscible alkyl diether. After reacting the mixtures under conditions favoring the conversion of PC to PG, the product PG is removed, preferably as an insoluble calcium salt. A preferred reaction mixture contains between 10 and 100 mM 'calcium salt, between 2-10% by volume glycerol, and between 5-20% by volume diether co-solvent, at a pH between about 6-9. The diether is preferably one whose boiling point is less than that of water, allowing solvent re¬ covery by distillation. Exemplary diether co-solvents are 1,1-dimethoxymethane, 1,2-dimethoxyethane, 1,2- dimethoxypropane, 1, 1-diethoxymethane, and 1-ethoxy-l- methoxy ethane. Another alkyl diether whose boiling point is below 100 C is 2-2-dimethoxypropane. The amount of active enzyme present in the mixture is preferably controlled such that at the termination of the reaction, (a) no more than about 10% of the original amount of PC remains at the

of the reaction, and (b) the level of PG present in the reaction mixture is no less than about 50% of the maximum PG level which occurs during the course of the reaction. This maximum level is achieved when the amount of PG produced by conversion of PC to PG (catalyzed by the transferase activity of the PLD enzyme) less the amount of PG lost by the conversion of PG to PA (catalyzed by the hydrolase activity of the enzyme) is a maximum.

The PLD enzyme may be prepared conveniently by heat-treating a cabbage juice filtrate. Preincubating the enzyme with glycerol increases the initial rate of PG formation, thus increasing the ratio of PG to PA formed in the reaction.

In another aspect of the invention, the method of the invention can be used more generally for treating a phospholipid enzymatically with phospholipase A.. , A-, C, or D. The method includes preparing a mixture of the phospholipid and phospholipase in an aqueous solvent containing between about 5% - 25% of a water-miscible alkyl diether of the type mentioned above, and reacting the mixture under conditions which allow enzymatic conver¬ sion of the^:" phospholipid by the phospholipase.

In still another aspect, the invention includes a novel method for converting PA to PG, by reaction of PA with glycidol in an alkyl ether solvent. Preferably, PA is reacted with glycidol in a low-boiling alkyl ether solvent at a temperature between about 20-40 C, and the resulting PG isolated. Preferably, PA is reacted with glycidol in a low-boiling alkyl ether solvent at a temperature between about 20-40 C, and the resulting PG isolated. The method may be coupled with the PC to PG reaction above, to convert PA produced in the reaction to PG, thus increasing PG yields and eliminating PA as a contaminant. These and other objects and features of the present invention will become more fully apparent when the following detailed description of the invention is read in conjunction with the accompanying drawings.

Brief Description of the Drawings

Figure 1 illustrates the enzymatic conversion by PLD of (A) PC to PA, (B), PG to PA, in the presence of glycerol, and (C), PC to PG or any related ester of PA; Figure 2 illustrates the effect of glycerol concentration on PG yield over a 5-hour reaction period, measured in 2% (circles) or 4% (triangles) glycerol;

Figure 3 shows the effect of glycerol concentra¬ tion on percent PG yield in 20% glyme; Figure 4 shows plots of PLD inactivation in 10% glyme, as measured by maximum PG yield assayed by a tritiated PG assay (circle'..), or TLC densitometry (triangles) ;

Figure 5 plots the kinetics of PG formation in 7.5% (triangles) and 12.5% (circles) glyme; Figure 6 shows plots of the kinetics of PG formation a^ concentrations of glyme between 10 and 20%;

Figure 7 shows the kinetic of PG formation and conversion to PA at increasing ^'concentrations of PLD; Figure 8 shows curves of PG production in a re¬ action in which the PLD enzyme has been preincubated with glycerol (triangles) or has not been preincubated (circles), showing the reduction in lag phase of PG production when the enzyme is preincubated with glycerol; and

Figure 9 is a flow chart of a reaction process based on the method of the invention.

Detailed Description of the Invention

Forming PG from PC

A. Reaction Components

In one aspect, the invention includes a reaction method for enzymatic conversion of PC to PG in a single- phase reaction mixture. With reference to Figure 1, the PLD enzyme used in the reaction acts on the phosphate ester bond linking the glycerol phosphate moiety and the phospholipid head group. The enzyme reaction may be one of two types. The first is a simple hydrolytic cleavage to convert either PC (Figure 1A) or PG (Figure IB) to PA, involving a hydrolase activity of the enzyme. The second reaction type is a concerted transferase reaction in which the choline head group of PC is replaced, in the presence of glycerol, with a glycerol head group, to form PG

(Figure 1C). It is this latter transferase reaction which is desired in the PC-to-PG conversion method. The hydrolytic reactions which convert PC or PG to PA both contribute to reduced PG yield in the method, although the PA can then be converted to PG according to the PA-to-PG conversion method of the invention.

The reaction mixture used in the method includes PC, glycerol, a source of PLD enzyme, and a soluble calcium salt which is necessary for the desired transferase activity of the enzyme. According to an important feature of the invention, the reaction is car¬ ried out in a single-phase aqueous solvent system contain¬ ing between 5-50 percent, and preferably 5-20 percent by volume of a water-miscible alkyl diether.

The PC is selected to correspond, in its fatty acyl composition, to the fatty acyl composition of the desired PG product. PCs having a variety of acyl composi¬ tions, including selected acyl chains with partial or complete saturation, and different fatty acyl chain lengths, are available from commercial sources, or can be synthesized by well-known methods. The amount of PC added to the final reaction mixture is preferably between about 0.5 to 2.0 percent by weight, and is added in the diether solvent, as will be described.

Studies on the effect of glycerol concentration on PG production are discussed in Example 3C , and the results of the studies are summarized in Figures 2 and 3. The first of these figures shows quantity of PG formed, expressed as a percent of original PC, at 2% and 4% glycerol weight concentrations, over a 5-hour reaction period. Two features of the curves are of interest. First, both reactions pass through a maximum PG level, achieved when the amount of PG produced by conversion of PC to PG (catalyzed by the transferase activity of the enzyme) less the amount of PG lost by the conversion of PG to PA (catalyzed by the hydrolase activity of the enzyme) is a maximum. As will be seen below, this feature is important in determining enzyme levels which optimize the amount of PG produced in the reaction. The second feature is the clearly superior reaction performance at the higher glycerol concentration, both in terms of greater maximum PG levels and longer period of net PG formation. The percent PG formed at several glycerol concentrations between 2.5 and 15 percent is given in Figure 3. As seen, glycerol concentrations between 2.5 and 10 percent all support efficient reaction levels.

The concentration of soluble calcium ions in the reaction mixture is preferably between about 10 to 100 mM. The calcium ions are preferably supplied in a salt form which acts to buffer the mixture. Published reports on the enzymatic conversions of PC with PLD generally suggest that increasingly high ionic concentrations are detrimental to enzyme action. In the usual and preferred reaction mixture, the calcium is supplied as calcium acetate, and the material is adjusted to the desired buffered pH (below) with acetic acid.

PLD preparations from Savoy cabbage (Davidson, 1958; Dawson; Algyer; Yang), from peanuts (Heller), from sugar beets (Kates), and from fababeans (Atwal) have been described. PLD enzyme obtained from peanuts, cabbage, and different strains of Streptomyces are commercially avail¬ able, e.g., from Sigma Chemicals (St. Louis, MO). Although these sources of PLD are suitable, they are relatively expensive for scale-up operations. Similarly, methods reported in the literature (e.g., Yang) for obtaining purified PLD generally involve the use of organic solvents, such as acetone, making large- preparation isolation of the enzyme expensive. Also the necessity of flammable solvents is a serious drawback in a large-scale operation.

Because of these limitations, alternative enzyme preparative methods which (a), do not involve organic solvents or chromatographic separation procedures, and (b) yield an enzyme prep which is not contaminated with phospholipa.se A, or A_ were developed in conjunction with the invention. One preferred enzyme prep is formed from Savoy cabbage, according to the general method detailed in Example 1. Briefly, cabbage is homogenized ice-cold in a suitable buffer, centrifuged to remove particulate matter, and the clarified supernatant is treated by quickly heat¬ ing to 50-55 C for 5 minutes. After cooling, the material is again centrifuged to remove heat-denatured material. The resulting supernatant shows little lyosphospholipase activity, and shows good PLD activity in the reaction of the invention. The amount of enzyme which is added to the reaction mixture to optimize PG production will be discussed in Section II below.

According to an important feature of the inven- tion, it has been discovered that the PLD enzyme is active and relatively stable in a single-phase aqueous medium containing between 5-50 percent by volume of a water- miscible alkyl diether. Exemplary diether co-solvents include 1,1-dimethoxymethane, 1,2-dimethoxyethane (glyme), 1,2-dimethoxyproρane, 1,1-diethoxymethane, and 1-ethoxy-l- methoxymethane. Another alkyl diether whose boiling point is below 100 C is 2-2-dimethoxypropane. ^■ All of the above co-solvents have boiling points below that of water, and thus can be recovered readily from a spent reaction mixture by distillation. Where solvent recovery is not crucial, higher boiling point diethers, such as 1,2- diethoxyethane, (126 C bp) , l-ethoxy-2-methoxyethane (102 C bp) and 1,3-diethoxypropane (106° C bp) can also be used. The stability of PLD in aqueous solvent mixtures containing alkyl diether co-solvents has been examined, such as reported in Example 2. The studies indicate that increasing amounts of co-solvent, in the 5-50 volume percent range, produce increasing enzyme inactivation, but that at all co-solvent concentrations, the enzyme shows appreciable enzyme activity at least up to about 5 hours incubation ^:at room temperature. The inactivation of PLD in 10% glyme (1,2-dimethoxyethane) is illustrated in Figure 4. The enzyme was preincubated for increasing periods up to five hours in the glyme mixture before addi¬ tion of PC, for enzymatic conversion to PG, in the pres¬ ence of radiolabeled PG. The amount of PG formed was as¬ sayed both by a ^ritiated PG assay, in which total re¬ action phospholipids are counted by scintillation for in- corporation of labeled glycerol or by a TLC densitometer assay in which the total lipids are fractionated by thin layer chromatography (TLC), and the lipids stained for densitometric determination. Both methods gave substantially identical results, as seen. The study, which is detailed in Example 2, indicates that the enzyme loses about half its transferase activity over a 5 hour preincubation period. There is evidence from other stud¬ ies conducted in support of the invention that the hydrolyze activity of the enzyme (which converts PC and PG to PA) is similarly inactivated along with the transferase activity.

In order to minimize enzyme inactivation in the co-solvent, it is important to add a protective agent which reduces peroxide formation and damage, since diether solvents tend to promote peroxide reactions. Reducing agents, such as thiosulfate or dithiothreitol, are effec¬ tive for this purpose. Typically, 0.01% by weight thiosulfate was added to the reaction mixture. Although enzyme stability was not significantly enhanced by thorough purging of the mixture with nitrogen, it is advisable to run the reaction under an inert atmosphere ^'to further reduce po ential peroxide formation.

The volume percent of diether co-solvent is generally that which optimizes the yield of PG in the re- action, under the reaction conditions employed. At below optimal co-solvent concentrations, although enzyme in¬ activation ^:is minimized, the low solubility of PC in the reaction medium may significantly reduce the amount of PC which can react with the enzyme to form PG. At above optimal co-solvent concentrations, enzyme inactivation may seriously reduce PG yield. Studies conducted in support of the invention, and reported in Example 3, indicate that diether concentrations between about 10-20 volume percent, and particularly, about 12.5% are optimal. The kinetics of conversion of PC to PG in 7.5 and 12.5 percent glyme are seen in Figure 5. The higher glyme concentration clearly favors higher PG yields. The reaction kinetics shown also indicates the effect of adding a second enzyme dose to the reaction mixture, in this case, 20 hours after initiation of the reaction. At the higher glyme concentration, the second enzyme dose carried the yield of PG up to about 60% after 40 hours.

Figure 6 shows the results of another study in which PG formation at different glyme concentrations between 10-20% was examined. There is little difference in the kinetics of product formation among the different reaction mixtures, although the 10% glyme mixture shows a slight lag in PG formation, and less rapid decline of maximum PG levels, presumably because of slower PG-to-PA conversion.

B. Reaction Method

This section describes the method for production of PG from PC, including those factors which are important • to optimizing PG yields, and methods for obtaining PG from the reaction fixture in relatively pure form.

As indicated in Section I, the conversion of PC to PG in the reaction of the invention proceeds through a maximum PG concentration, at which the amount of PG produced by the transferase activity of the enzyme less the amount ^"of PA produced by enzymatic hydrolysis of PG to PA is a maximum. Following this maximum, the concentra¬ tion of PG declines, as the hydrolytic conversion of PG to PA exceeds conversion of PC to PG. If optimal PG were the only consideration in optimizing the reaction, the amount of enzyme present in the reaction mixture would be adjusted to produce a maximum PG level, at which point the reaction could be terminated. However, it is also advantageous to minimize the amount of PC remaining when the reaction is terminated, since PC tends to contaminate PG in the final product isolation steps. In particular, it is desired to have no more than about 10% of the original PC present at the termination of reaction. As a rule, the desired low PC concentration is reached well after the point PG maximum, so that the re¬ action must be run for some period on the "down" side of the PG curve. It is therefore important that the reaction conditions be adjusted to minimize loss of PG during this latter phase of the reaction period, i.e., that the PG curve remain as flat as possible during the latter re¬ action phase. More specifically, the amount of active enzyme present in the mixture should be controlled such that at the termination of the reaction (a) no more than about 10% of the original amount of PC remains, and (b) the level of PG is no less than about 50% of the maximum level achieved during the coarse of the reaction.

Optimal enzyme levels can be attained in two ways. First, the amount of enzyme added can be such that the enzyme is largely inactivated by the time the PG maximum is reached, allowing relatively slow conversion of PC to PG and PG to PA in the latter reaction phase. The importance of relatively low enzyme activity after the PG maximum is reached is illustrated by the study reported in Example 3E. Here the kinetics of PG production at increasing concentrations of enzyme, expressed in terms of volume percent of enzyme solution, were analyzed, with the results shown in Figure 7. As seen, the lowest enzyme concentration—corresponding to about 4% by volume of cab- bage juice—gave very low loss of PG beyond the PG maximum, indicating substantial loss of enzyme activity at this stage of the reaction. By contrast, the 2X and 4X enzyme concentrations gave rapid loss of PG following the PG maximum. It can be appreciated that an enzyme concentration similar to or slightly greater than the IX mixture is optimal. Since the specific activity of enzyme may vary from one preparation to another, depending on preparation and storage variables, it is advisable to establish empirically the optimal enzyme concentration with each batch

Secondly, the PLD enzyme can be added in one or more additional increments during the course of the re¬ action, at a sufficiently low enzyme concentration that the conversion of PG to PA is never significantly greater than PG conversion to PC. In practice, the two above ap¬ proaches may be combined as follows: The initial enzyme amount is such that under the reaction conditions employed, the enzyme is largely inactivated before the PC has been reduced to the desired low level. Based on the amount of PC remaining at this first low-activity stage, a proportionately smaller amount of fresh enzyme is added to further reduce PC levels, and this addition process may be repeated with successively smaller amounts of enzyme until the desired end point is reached. This feature is il- lustrated in the reaction method described in Example 5. Figure 5 illustrates the favorable reaction kinetics which can be achieved with a second enzyme addition midway through the reaction.

Another factor which is important in reducing the relative final amounts of PG and PA is the relative activity of the PLD enzyme for the PC and PG substrates . In rate studies conducted in support of the present inven¬ tion, it has been repeatedly observed that the formation of PG from PC lags the conversion of PC and PG to PA by 3- 4 hours. One possible explanation for this lag is that the transferase activity of the enzyme requires a combined enzyme interaction with both glycerol and PC, and that the glycerol binding to the enzyme occurs at a rate which is slow in comparison with enzyme binding to PC (or PG) . If so, preincubation of the enzyme with glycerol for a 3-4 hour period, before addition of PC in co-solvent, may prevent the lag in PG production.

The more immediate PG production achieved with preincubation is shown in Figure 8, which plots the kinet- ics of PG production with (triangle) or without (circles) preincubation of the enzyme in 4% glycerol before addition of PC in glyme. As seen, preincubation largely eliminates the lag in PG conversion curve seen without preincubation, and thus enhances the production of PG over PA. The pH of the reaction mixture is adjusted to a selected pH compatible with high transferase activity, and preferably between pH 6-9. Reports from the literature indicate that PLD is active over a wide pH range, and that maximum activity occurs at a pH of about 6. There is also reason to believe that optimal transferase activity may occur at pH 8-9. The formation of PG in reaction mixtures initially adjusted to either pH 6.0 or 8.5 are reported in Example 3B. Although the higher pH mixture gave higher levels of PG, the pH of this reaction mixture also dropped during the reaction to a final pH of about 6. Readjust¬ ment of the pH values and addition of fresh enzyme resulted in higher PG production in the pH 6 mixture. The results indicate that (a) the reaction can be carried out over a range of pH between 6 and 9, and (b) at a pH greater than about 6, it is necessary to monitor pH during the reaction, to maintain the desired pH. As indicated above, a reaction pH of about 6 can be achieved conveniently in an acetate buff «er which also provides a source of calcium ions. On the other hand, PG and PA phospholipids formed in the reaction are converted completely to calcium salts by raising the pH of the reaction mixture to between 8 and 9, as will be seen below. Thus a potential advantage of running the reaction at this higher pH is more complete removal of PG as it is formed, with the potential for reduced conversion of PG to PA.

The reaction is carried out at a temperature preferably between about 20-30 C. Example 3A describes a study which compares PG production at 21 C and 30°C. As reported, the lower temperature gave higher PG production in a four-hour incubation period. The reaction is run until the desired final level of reactant PC is reached, e.g., 5-10% of the original PC.

At the termination of the reaction, the PG is extracted from the mixture in a procedure which preferably involves (a) forming the insoluble calcium salts of PG and PA (which also contain PC), (b) removing the salts from the reaction mixture, and (c) selectively extracting PG from the removed salts. At a reaction pH of about 6 , a portion of the PG and PA which form during the reaction are converted to and precipitate as calcium salts . To insure complete conver¬ sion to insoluble calcium salts, the reaction mixture is brought to pH 8-9 with addition of calcium hydroxide. To this mixture is added a filter aid, such as Celite 545® which facilitates column solvent extraction of PG. The particulate material, including tϊ.-_ calcium phospholipid salts and filter aid, is removed by filtration or centrifugation, washed, and placed in a column for solvent extraction of PG, as exemplified in the method detailed in Example 5A.^;'

Solvent extraction of PG from the insoluble PG/ PA material can be performed wi^'th a suitable solvent system, such as the t-butanol:water system described in Example 5B, or the chloroform:methanol:water system described in Example 5C. Following solvent extraction, the material may be treated with an ion exchange resin, to form a more soluble lipid salt, such as the ammonium salt, substantially as described in Example 5B. Typically, the product at this point is between about 70-95% pure depend¬ ing on the solvent system used for lipid extraction. If necessary, the preparation can be further purified by silica gel chro atography, using conventional solvent separation methods, such as that described in Example 5A. Product yields are typically in the range of about 25%- 35%, based on the original quantity of PC used, at PG purity is between about 90%-95%.

Alternatively, and according to another feature of the invention, the mixture of PG and PA produced in the reaction are further reacted with glycidol to convert the PA to PG. This PA-to-PG conversion reaction will be detailed in Section II below.

It can be appreciated from the above that the features of the invention which are applicable to the conversion of PC to PG by PLD are more generally ap¬ plicable to phospholipid reactions with phospholipases A. , A~, C and D. These features include the use of water- miscible diether co-solvents to solubilize the reactant phospholipid, with solvent compatibility for the phospholipase enzyme, and selection of reaction conditions and enzyme quantities, in relation to the rate of enzyme inactivation, which maximize product yields. The inven¬ tion thus includes, more generally a method of treating a phospholipid enzymatically with phospholipase A_]_, 2 , C, or D. The method involves preparing a mixture of the phospholiprci and phospholipase in an aqueous solvent containing between about 5% - 25% of a water-miscible alkyl diether, and reacting the^' mixture under conditions which allow enzymatic conversion of the phospholipid by the phospholipase. Preferred diether co-solvents are those mentioned above, including 1,1-dimethoxy ethane, 1,2-dimethoxyethane, 1,2-dimethoxypropane, and 1,1- diethoxymethane. The method is illustrated in Example 6, which describes the conversion of PC to PA by phospholipase D in aqueous, single-phase reaction mixtures formed with 1,2-diethoxymethane, 1,2-dimethoxypropane, or 1,2-diethoxyethane.

C. Reaction System

Figure 9 is a flow diagram of a reaction system designed for production of PG according to the method of the invention. The system employs the optimized reaction conditions discussed above and is designed particularly for scale up to commercial scale production.

The reaction is carried out in a fermentation tank which initially receives the mixture of glycerol in calcium acetate, pH 6 and the source of PLD for preincubation of the enzyme, as indicated. After preincubation, a solution of PC in diether solvent is added to a final desired diether and PC concentration. The PC-to-PG conversion reaction is carried out until the desired end point, preferably determined by final Pc concentration. Reaction at 21 C for 24-40 hours is typical. At the end of the reaction period, the mixture is tested for total PC remaining, and if necessary, ad¬ ditional preincubated enzyme is added to reduce final PC concentration to less than about 10%.

The reaction mixture is then brought to pH 8-9 with calcium hydroxide and, after addition of filter aid, is filtered to removed the filter aid and insoluble salts of PG and PA. The insoluble material is packed into a column and extracted with a suitable solvent for extrac¬ tion of the calcium salt of PG.^" Treatment of the 5 extracted material with an ion exchange resin yields the ammonium salt of PG which can be further purified, and/or treated for solvent.

Considering now material recovery operations, the filtrate from the co-solvent reaction mixture is

10 treated by distillation to recover the diether co-solvent which has a boiling point preferably below about 90 C. Simple distillation of the filtrate through a 10-plate Oldershaw column has been used to recover glyme substantially quantitatively from a 12.5% glyme reaction

15. mixture. The recovered product has been reused in PLD fermentations without any observable untoward effects.

Glycerol can be recovered by conventional re¬ covery techniques used in the soap-making industry. PC can be recovered by conventional lipid extraction

20 techniques, after filtration to remove insoluble reaction products, and distillation to remove diether co-solvent. Alternatively, the lipid can be removed prior to distilla¬ tion by lipid adsorption by hydrophobic resin particles, such as phenolic resin beads. Lipid removal from the

25 washed beads is by conventional lipid extraction.

PA, the major by-product of the PLD fermenta¬ tion, can be recovered after PG extraction with further extraction in a more lipophilic solvent, such as chloroform:methanol (2:1). The extracted PA can be puri-

30 fied or further treated with phospholipases to convert to other phospholipids. As indicated above, PA in the mixture of PG and PA formed in the reaction may alternatively be reacted with glycidol, to convert the PA to PG. This reaction, which is detailed in Section II,

35 provides increased yields of PG and eliminates the need to purify PG from PA in the present reaction products.

T-butanol, if used for selected solvent, is eas¬ ily recovered by distillation, since it boils at 83 C. Similarly, chloroform and methanol are readily recovered by distillation. Ion exchange resins used for calcium ion replacement in the PG salts can be reused repeatedly.

From the foregoing, it can be appreciated how various objects and features of the invention are met. According to one advantage of the invention, the enzymatic conversion of PC to PG is performed in a single=ρhase re¬ action mixture which allows efficient conversion of PC to PG without the requirement for energy input, to maintain a two-phase emulsion, and avoids variations in reaction kinetics and enzyme inactivatisjji relating to variations in the emulsion and the agitation needed to maintain the emulsion. Further, flammable ether solvents which have been used heretofore in two-phase reactions are avoided.

A preferred source of PLD enzyme in the reaction is readily obtained from cabbage homogenate without need for costly solvent extraction or chromatographic procedures. The enzymatic conversion of PC to PG can also be significantly enhanced in the reaction by preincubation of the enzyme material with glycerol. This feature, combined with solvent inactivation of the enzyme at a selected rate, can be exploited to maximize the PG to PA ratio of reaction products.

The reaction method produces relatively high product yields, and is compatible with simple product re- covery methods based on filtration and solvent extraction, to yield-a high purity product. Virtually all of the materials used in the reaction and product extraction steps, except the enzyme itself, are readily recoverable for economy in a large-scale operation. The following examples illustrate methods for converting ^"PC to PG and PA, and for obtaining substantially purified PG product. The examples are intended to illustrate, but not limit, the invention.

Materials 1,2-dimethoxyethane, 1,2-dimethoxypropane, and 1,2,diethoxyethane were obtained from Aldrich Chemicals (Milwaukee, WI) . Egg PC was obtained from Asahi Chemical Ind. (Asahi, Tokyo, Japan) . Egg PG and PA were obtained from Avanti Polar Lipids (Birmingham, AL) . Thin layer chromatography plates (silica gel) were supplied by J.T. Baker (Phillipsburg, NJ) , and Chelex-100 ion exchange resin, from Bio-Rad Labs (Richmond, CA) .

Example 1 Preparation of Phospholipase D Phospholipase D (PLD) was prepared from Savoy cabbage by the following procedure. A fresh head of cab- bage was passed through a meat grinder, and the freshly ground tissue was minced with about 1/4 of its volume of crushed ice, and homogenized in a blender until liquefied. The homogenate was pressed through a sheet of muslin to remove coarser particulate matter, and then centrifuged at about 500 x g for 20 minutes. The clarified supernatant was decanted and stored frozen until needed. The yield ranged from 500 ml to 1 liter, depending on the size of the head of cabbage used.

Shortly before use, the frozen supernatant material was thawed, adjusted to pH 6.5 with the dropwise addition of 0.1 N NaOH, and quickly heated to 50°-55°C for 5 minutes. The material was then cooled in an ice bath to about 30 C within less than 2 Minutes from the end of the heating step. The cooled preparation was again centrifuged at about 1,000 x g for 10 minutes to remove coagulated proteins. The clarified supernatant was used as the source of PLD.

Example 2 Co-solvent Inactivation of Phospholipase D

Several alkyl ether co-solvents, including those listed in Table 1 have been assessed for suitability in the present invention. Three co-solvents having boiling points below 100 C were examined in the present study: 1,2-dimethoxyethane, 1,1-dimethoxymethane, and 1,1- diethoxymethane.

The conversion of PC to PG in the reactions discussed below was assayed by measuring levels of radio- activity in lipids incubated in the presence of tritiated glycerol and/or by thin layer chromatography (TLC) scan¬ ning of stained TLC spots. In the first method, the re¬ action sample was acidified with 6.IN trichloroacetic acid (TCA) to pH 2, and extracted with an equal volume of chloroform. The chloroform phase was separated and washed 2 times with water to extract free glycerol. The chloro¬ form solution was than counted by conventional scintilla¬ tion counting.

In the alternate method, the chloroform extract from above was dried and redissolved in chloroform/ methanol/water/28% ammonia (130:70:8:0.5) and chromatographed by TLC. PC migrated with a R_f value of 0.13-0.16 and PG, with an R_f of 0.74-0.80. The TLC plate was stained with I- vapor for two hours, and then spots quantitated by densitometry scanning, according to conventional methods. Typically two values were measured for each time point.

For each co-solvent reaction mixture the enzyme was incubated at 21 C in 0.02 M calcium acetate, pH 6.0, containing 4% tritiated glycerol and 10% co-solvent for varying periods of time before addition of the PC substrate. Enzyme activity was assayed by percent conver¬ sion of PC to PG after 24 hours. The rate curve enzyme inactivation in 10% 1,2-dimethoxyethane (glyme) is shown in Figure 3, where the graph symbols represent percent maximum yield of PG, as determined by tritiated PG counts (solid circles) and thin layer chromatography (TLC) densitometric scan (solid triangles).

As seen, about 50% of the PLD transferase activ- ity remained after 5 hours at 21 C. Inactivation of transferase activity was slightly greater in the other two co-solvents. Thus 1,2-dimethoxyethane was selected as a preferred co-solvent of the PC-to-PG conversion reaction.

Example 3

Optimizing Reaction Conditions The enzymatic conversion of PC to PG, according to the method of the invention, involves at least seven variables capable of optimization. These are: temperature, calcium ion concentration, pH, glycerol concentration, co-solvent concentration, enzyme concentra¬ tion, and time and sequence of addition of reaction components. Studies examining several of these variables are presented in this example and the following example. A. Temperature

Two duplicate reactions were set up. Each contained 0.02 M calcium acetate, pH 6.0, 4% glycerol, 12.5% glyme, and 1% PC, in addition to 4% (by volume) heat-treated enzyme solution. One preparation was incubated under nitrogen at 21 C for 4 hours, and the other at 30 C for 4 hours. At the end of the incubation period, sample of the reaction were removed and analyzed by TLC. The lipids in the 21 C reaction contained 34.8% PG, and those in the 30°C reaction, 24.4% PG. It is thus concluded that the lower temperature is more favorable for PC-to-PG conversion.

B. Reaction pH Reaction at pH 6 and 8.5 were examined.

Duplicate fermentations each contained 0.02 calcium acetate, 4% glycerol, 12.5% glyme, 1% PC, and 0.10% sodium thiosulfate, and 4% cabbage juice. One of the samples was adjusted to pH 6.0 with acetic acid, and the other, to pH 8.5 with sodium hydroxide. The samples were incubated at 21 C under a nitrogen atmosphere for 16 hours, then as¬ sayed by TLC. During the reaction period, the pH of the higher-pH mixture fell to about 6. Final lipid content was about 53% PG. The pH 6.0 mixture retained its pH, and its lipids contained about 39% PG.

Readjustment of pH values, followed by ad¬ ditional equal portions of enzyme and a second 16-hσur incubation period led to a lipid content of 45% PG at pH 6.0, and 42% at pH 8.5.

C. Glycerol Concentration

In a first study, reaction mixtures containing 0.02 M calcium acetate, pH 6.0, 0.5% PC,. 10% cabbage juice, 10% glyme and either 5% or 10% glycerol were incubated at 30 C. After three days incubation, 76% of the PC had been consumed at 5% glycerol, but only about 50% PC at 1^:6% glycerol.

In a second experiment, similar reaction mixtures, but containing only 2.5% cabbage juice and either 2% or 4% glycerol were incubated at the same temperature, and aliquσts examined for PG at 1, 3, and 5 hours incubation. The results are seen in Figure 2. As seen, a substantially higher PG maximum, at 3 hours incubation, is reached with 4% glycerol. A third experiment examined glycerol formation in 0.02 M calcium acetate, pH 6.0 with 1% PC in 20% glyme, and at glycerol concentrations of 2.5%, 5%, 10%, and 15%. PG was measured after 24 hours incubation at 23 C, with the results seen in Figure 4. The data indicate high conversion between 2.5%-10% glycerol, with a maximum at about 5%.

D. Concentration of Co-Solvent

To establish a lower limit of effective co- solvent concentration, reaction mixtures containing 0.02 M calcium acetate, pH 6.0, 0.5% PC, 4% glycerol, 5% cabbage juice in either 5% or 15% glyme were prepared, and were examined for PG production by TLC after 20 hours incuba¬ tion at 30 C. The mixture with 5% glyme contained 81.8% PC, but only 4.4% PG and 5.7% PA. By contrast, the 15% glyme mixture contained 6.2% PC, 27.9% PG, and 63.1% PA. The relatively poor conversion of PC at the lower glyme concentration presumably reflects, at least in part, relatively PC solubility in the reaction mixture. At the upper limit, PC-to-PG conversion under substantially similar reaction conditions, but in 50% glyme, was examined. An identical reaction mixture, but containing only 10% glyme was run in parallel. From TLC analysis, two features of the reaction were observed. First, at the high co-solvent concentration, the calcium salt of PC tends to precipitate before significant PC conversion 'occurs. Secondly, despite PC precipitation, the reaction mixture contained about 21% PG and 19% PA and 53% PC after 4 hours incubation at 21°C. A similar study, comparing PC-to-PG conversion in 7.5% and 12.5% glyme, is shown in Figure 5. As indicated in the figure, a second portion of the cabbage enzyme was added to the reaction mixtures at 20 hours. As seen, 12.5% glyme yielded nearly 60% conversion to PG, after 40 hrs incubation at 30°C.

In another study, reaction mixtures containing 0.04 M calcium acetate, pH 6.0, 5% glycerol, and cabbage juice was preincubated for 4 hours at 21 C prior to addi¬ tion of glyme, at concentrations between 10% and 20%, and containing PC and sodium thiosulfate. Incubation at 21 C for periods up to 2Q hours gave the percent PG formation shown in Figure 6. All glyme concentrations between 10%- 20% gave effective PC-to-PG conversion.

E. Quantity of Enzyme

To demonstrate that the rate of conversion of PC to PG is dependent on enzyme concentration, reactions containing 10% glyme, 0.5% egg PC, 4% glycerol, 0.02M calcium acetate, pH 6.0, and 0.1% thiosulfate were incubated with IX (squares in Figure 7), 2X (triangles), or 4X (circles) volumes of cabbage juice, where IX cor¬ responds roughly to 2% of the reaction mixture by volume. The mixtures were incubated at 21 C for the increasing time periods shown in Figure 7. As seen, all three percent PG passed through similar maxima, and doubling and quadrupling the enzyme concentration resulted in maxima at one-half and one-fourth the reaction times, respectively. The percent PG measure by scintillation counting corr responded closely to PG levels measured by TLC. Example 4 Preincubation with Glycerol A reaction sample containing 0.04 M calcium acetate, pH 6.0, and cabbage juice was preincubated in the presence of 5% glycerol for 4 hours at 21 C prior to addi¬ tion of 0.1 M sodium thiosulfate, 1% PC, and 12.5% glyme. The PC/glyme mixture was added directly to a second identical sample which had not been preincubated with glycerol. The two reaction mixtures were incubated at 21°C for the time indicated in Figure 8, and aliquots were assayed for percent PG, as above. As seen in the figure, the lag time seen in the reaction without enzyme preincubation (circles) was eliminated by enzyme preincubation for 4 hours (triangles).

Example 5 PG Preparation

A. Enzymatic Reaction Into a one liter flask was added 50 g glycerol.

Into a second flask 5.6 g calcium acetate hydrate was dis¬ solved in 800 ml distilled water, and enough acetic acid was added to lower the pH to 6.0. The resulting calcium acetate buffer was added to the glycerol with stirring until well mixed. To this solution was then added enough cabbage juice (Example 1) to convert 10 g of PC to PG and PA in 4-8 hours. Typically the final volume percent of cabbage juice in the final reaction mixture was between 4- 6%. The reaction mixture was incubated under a nitrogen atmosphere for 4 hours at room temperature.

A lipid solution was prepared'by dissolving 10 g egg PC in 175 ml glyme (spectroscopic grade) . "After all of the PC had dissolved, the 500 mg of sodium thiosulfate was added with stirring until dissolved. The lipid solu- tion was then added to the preincubated enzyme mixture, and the milky dispersion is incubated under nitrogen overnight (^"about 8 hours), then analyzed for residual PC. If more than about 5% of the original PC remained, ad¬ ditional cabbage juice was added. The amount of enzyme to be added was estimated by multiplying the amount of enzyme (volume of cabbage juice) originally added by percent of PC remaining after 24 hours. Thus, if 10% PC remained after 24 hours, and the original volume f enzyme was 40 ml, the volume addition of enzyme after 24 hours was 4 ml. The reaction mixture was incubated for another 24 hours at 21°C.

Following the enzymatic reaction, PG and PA were converted to the corresponding insoluble calcium salts by slow addition of calcium hydroxide with stirring, until a pH of between 8 and 9 was reached. To this mixture was added 15 grams of Celite 545® and the mixture was filtered with suction. The filter cake was washed with 10 to 20 ml distilled water, then with 25 ml acetone. The solid was dried in a vacuum desiccator to obtain about 24 grams of colorless, friable powder consisting primarily of Celite and the calcium salts of PG and PA. The powder was poured into a 22 mm chromatography column plugged loo'sely with glass wool. Either of the two methods described in parts B and C below were used to obtain PG.

B. PG Extraction with t-butanol:water

A total volume of 300 ml of solvent mixture containing tert-butanol:water (5:2) was poured over the column and allowed to elute. Total elution time was about 48 hours. TLC analysis of the phospholipids in the efflu¬ ent revealed a PG content of about 76% and PA content of about 8%.

The effluent was next passed through a 22 x 250 mm column of Chelex 110®, ammonium form, and the column was washed with 300 ml of the t-butanol:water mixture to bring all of the PG (ammonium ion form) through the column. The Chelex 100 column eluate was vacuum evaporated to a constant weight of about 5 g of a color¬ less wax. TLC analysis of this^' product showed about 70% PG as the ammonium salt, making the overall yield of PG about 36%.

To increase the purity of the PG, the above wax¬ like product was dissolved in 35 ml methylene chloride and poured into a 22 mm chromatographic adsorption column packed with 50 g silica gel. The solution was permitted to percolate through the column until the meniscus is at the top of the silica gel. Development was with 300 ml methylene chloride containing 10% ammoniacal methanol (methanol containing 1% concentrated ammonium chloride), followed by 300 ml of methylene chloride containing 50% ammoniacal methanol. Upon evaporation, the 50% ammoniacal mixture yielded 3.6 grams of a waxy product containing 92% PG. The overall yield was 33%.

C. PG Extraction with chloroform: ethanol:water

A total volume of about 400 ml of solvent mixture containing chloroform:methanol:water (2:3:1) was poured over the column from part A and allowed to elute. The first 50 ml of effluent were saved for recycling PC. The next 300 ml of effluent were collected for PG extrac¬ tion. This material was passed through a Chelex 100 column as above, and the column was washed with 300 ml of the chloroform:methanol:water mixture. The eluates from the Chelex column were combined and to this was added an equal volume of chloroform, resulting in a phase separa¬ tion. After shaking, the chloroform phase was evaporated to dryness nder vacuum, to yield 2.6 gram of a colorless wax-like material. TLC analysis of the dried product showed about 95% PG, as the ammonium salt, bringing the yield to about 25%. Example 6 Conversion of PC to PA Egg PC (0.5 g) was dissolved in 10 ml of 1,2- diethoxymethane. The lipid soϊution was added to 40 ml of 0.02 M calcium acetate adjusted to pH 6.0 with dilute acetic acid. To the resulting reaction mixture is added 10 ml of the phospholipase D solution (Materials) . The mixture was incubated at 23 degrees C for 4 hours without stirring, progressively accumulating a calcium/PA precipitate.

The precipitate was collected by filtration.

TLC analysis of the filtrate, in a chlorofσrm/methanol/ water system, showed about 35% conversion of PC to PA.

Similar reactions were carried out with 1,2- dimethoxypropane and 1,2,-diethoxyethane. The percent conversion of PC to PA in each reaction is shown in Table 1 below. Efficient enzymatic conversion occurred in each of the solvent systems.

Table 1

Co-Solvent % PA Formed 1, 1-Diethoxymethane 22%

1,2-Dimethoxypropane 20%

1,2-Diethoxyethane 68%

II . Forming PF from PA

A. Definitions The term "suitable alkyl ether", as used with respect to the PA-to-PG conversion reaction, refers to an alkyl e*her having a boiling point at atmospheric pressure less than about 150°C, preferably less than about 100°C. Suitable alkyl ethers may be straight, branched, or cyclic in structure. Suitable alkyl ethers for the practice of the invention must be capable of dissolving PA and glycidol, must be liquid over a temperature range of about 15-40°C, and should not react chemically with the re- actants. Preferably, the ether will also be inexpensive. Exemplary low-boiling alkyl ethers include, without limitation, 1,4-dioxane, tetrahydrofuran (THF) , dimethoxymethane (DMM), diethoxymethane (DEM), 1,2- diethoxyethane (DEE), and 1,2-dimethoxyethane (DME) . DEM is presently preferred. Glycidol refers to l,2-epoxypropan-3-ol.

"Phosphatidylcholine, " "phosphatidylglycerol, " and "phosphatidic acid", as used with respect to both PC- to-PG and PA-to-PG conversion reactions, refer to phospho¬ lipids irrespective of their source of origin. It is understood that the fatty acyl chains of the phospholipids may be the same or different, may be saturated or unsaturated, and may be natural or synthetic. Further, phospholipids within the scope of this invention may comprise mixtures of molecules having different fatty acid chains.

In order to effect nucleophilic opening of the glycidol epoxide ring, the pH of the reaction mixture must be low enough to protonate the epoxide oxygen. If the reaction mixture is prepared with PA in the free acid form, the pH will be sufficiently low without further adjustment. If the PA is added to the reaction mixture as a salt (e.g., the sodium or ammonium salt), then a suf¬ ficient amount of a strong protic, non-nucleophilic acid must be added. The term "non-nucleophilic protic acid" refers to an acid strong enough to protonate the glycidol epoxide oxygen, and which is substantially less nucleophilic than PA. The acid must be less nucleophilic than PA or it will compete with PA for epoxide opening and addition. Exemplary non-nucleophilic protic acids include, without limitation, trichloroacetic acid and iodoacetic ^"acid.

It is also preferred that the reaction mixture be substantially anhydrous, to minimize competing side reactions. "Substantially anhydrous", as used herein, refers to reaction mixtures containing less than about 10% water, preferably less than 5%, and most preferably less than about 1% water.

B. General Method

In general, a suitable solvent for the reaction is first selected, based upon its ability to dissolve PA and glycidol. Although it appears that the reaction rate of the invention process is not dependent upon reactant concentration, economic savings are still effected by minimizing the amount of solvent required to perform the reaction. Similarly, it is less energy-intensive to remove a solvent having a low boiling point, thus making low-boiling solvents more economic than high-boiling solvents. However, the solvent must remain liquid over the temperature range for performing the reaction. Thus, suitable solvents for the practice of the invention must be liquid over the range of about 10-60°C, and in the preferred embodiment, over a range of about 20-40°C. Preferred solvents in the practice of the invention are 1,4-dioxane (Example 7), tetrahydrofuran (THF) (Example 8), 1,2-diethoxymethane (DEM) (Example 11), 1,2-diethoxy- ethane (DEE) (Example 10), and 1,2-dimethoxyethane (DME) (Example 9), especially DEM. Phosphatidic acids (PA) are commercially avail¬ able from a number of suppliers (e.g., Sigma Chemical Co., St. Louis, MO), or may be prepared by hydrolysis of phosphatidylcholine, for example using phospholipase D, as described above. The particular PA used (i.e., regarding the fatty acid substitution) will be selected on the basis of the composition desired in the product. It is believed that the me^'chanism of reaction requires that the glycidol epoxide oxygen be protonated before the ring is opened. PA is a sufficiently strong acid that the pH of a reaction mixture prepared with PA in the free acid form will be sufficiently low to protonate glycidol. However, PA is frequently supplied in the form of a salt, particularly the sodium or ammonium salt. Accordingly, where a PA salt is employed, it will also be necessary to add a sufficient amount of acid strong enough to assure that the glycidol is protonated under the reaction conditions. The acid should also be less nucleophilic than the PA, to avoid competing for glycidol epoxide opening. Suitable acids include, for example, trichloroacetic acid and -iodoacetic acid.

Glycidol is also available from commercial sources, or may be prepared by the epoxidation of allyl alcohol using conventional methods. In the instant re¬ action, it is preferred to use an excess of glycidol over PA due to the competing side reactions (mainly glycidol polymerization) . The ratio of glycidol to PA should be between about 1:1 and about 10:1, preferably between 2:1 and 7:1, and most preferably about 5:1.

The reaction is performed by combining the re- actants in the selected solvent, and adjusting the temperature of the reaction mixture to a suitable temperature for a period of time sufficient to convert a substantial amount of the PA to PG. The glycidol may, if desired, be added in several portions at varying times. By adding a portion of the glycidol later during the re¬ action, one can further reduce the incidence of side re¬ actions which consume glycidol. "The reaction temperature must be high enough to ensure a reasonable rate of re¬ action, yet low enough to minimize competing side re- actions. Hence, the reaction temperature will be between about 10°C and about 60°C, preferably between 20-40°C, and most preferably at about 30 C. The yield of PG varies with time, typically reaching a maximum at about 60 hours. The reaction period is selected to optimize conversion of PA to PG, while minimizing the side reactions which dominate with increasing time. Thus, reaction times vary from about 10 hours to about 2 weeks, preferably 1-6 days, and most preferably about 3-4 days.

The course of the reaction may be monitored by standard methods, e.g., by thin layer chromatography

(TLC). A suitable TLC system is, for example, silica gel plate using a solvent mixture of chloroform, methanol, water, and aqueous ammonia in a ratio of 130:70:8:0.5 (Parsons) . When the reaction is judged complete, the solvent is removed from the reaction mixture under reduced pressure, and the residue taken up in methylene chloride and treated with anhydrous ammonia. The solution is then applied to a chromatographic absorption column, e.g., using silica gel, and the product eluted using methylene chloride containing an increasing amount of concentrated ammonium hydroxide in methanol. PG elutes when the solvent reaches about 60% methylene chloride/40% MeOH-

NH₄OH.

C. Combined PG Generation

This section describes a combined reaction in which PC is first converted to a mixture of PG and PA, according to the reaction method described in Section I, and the PA in the reaction mixture is further converted to PG, according to procedures described in Section IIB above.

As discussed above, the PC to PG reaction produces a mixture of PG and PA, and may also contain unreacted PC at the optional PG product level. The PA and PG are recovered in enriched form (selective removal of PC) by addi^ction of calcium hydroxide to complete the formation of the calcium salts of the two lipids, as discussed above. The mixture p^'referably contains a filter aid, such as Celite 545^w, and the precipitate is removed by filtration or centrifugation and washed, as described in Section I.

The calcium salts of PG and PA are preferably converted to acid form and dissolved in a suitable alkyl ether solvent by one of the two following methods. In the first method, the precipitated salts are suspended in a two-phase solvent mixture containing a lipid solvent, such as chloroform or methylene chloride, and an aqueous solu¬ tion of a non-nucleophilic protic acid, such as trichloroacetic acid (TCA) or iodoacetic acid. The aque¬ ous solution, which is preferably a 0.1 to 1 N concentra¬ tion of acid, is preferably present in slight molar excess of the total calcium present in the lipid precipitate, as determined by weighing or the like. One preferred two- phase solvent mixture is chloroform:l N TCA (5:1).

The lipids are shaken in the solvent mixture, and the two-phase system is then allowed to separate. After removal of the upper aqueous phase, the organic phase is washed two times with water. The resulting lipid solution, which contains PG and PA in protonated form, is then dried, e.g., by low-temperature rotovap. The dried lipids are suspended in a suitable alkyl ether solvent and reacted with glycidol, as above, to convert the PA in the mixture to PG. The PG may be purified, if desired, by conventional chromatographic methods, such as described in Section I.

In the second method for recovering PG and PA in protonated form, the precipitated calcium salts of the two lipids are extracted with a solvent system such as t- butanol:water (5:2) or chloroforrmmethanol:water (2:3:1), as described in Section I. The dissolved lipid solution is then pas'sed through a suitable ion exchange resin, in protonated form, such as acid-treated Chelex 100. The ion exchange resin is washed with s^'everal volumes of the extraction solvent, and the total eluate is combined, washed two times with water, then dried as above to yield a lipid residue enriched for PG and PA. The lipids are dissolved in a suitable alkyl ether solvent for reaction with glycidol, as above, to convert the PA in the mixture to PG.

Examples The following examples illustrate methods for converting PA to PG, and for obtaining substantially pure PG. The examples are intended to illustrate, but not limit, the invention.

Example 7 Preparation of PG in Dioxane Phosphatidylglycerol was prepared from phosphatidic acid and glycidol in p-dioxane as follows:

PA (2.85 g, 4.39 mmol) was dissolved in p- dioxane (8.6 mL) . Glycidol (345 uL, 5.20 mmol) was added, and the reaction mixture allowed to stand at 21 C for 24 hours. A second aliquot of glycidol (345 uL) was added, and allowed to react for an additional 72 hours.

The solvent was then removed under vacuum to provide the product as an amber colored wax (3.41 g) . The product was dissolved in methylene chloride (20 mL), and anhydrous ammonia gas bubbled in until the solution was saturated. The resulting ammonium salt solution was then applied to a 21 mm x 200 mm chromatographic adsorption column of silica gel (Merck Kieselgel 60). The column was eluted with CH₂C1₂ (100 L) , followed by 100 mL of 95% CH2CI2 with 5% methanol containing 1% saturated aqueous NH₄0H ( "MeOH/NH₄OH") , followed by 90% CH₂Cl₂-10% MeOH/ NH₄OH (100 ^'-mL ) , followed by 80% CH₂Cl₂-20% MeOH/NH₄OH (100 mL), followed by 60% CH₂Cl₂-40% MeOH/NH₄OH (200 mL) .

The eluate was collected in 50 mL fractions, and each fraction assayed by TLC on silica gel coated plates, using a mixture of chloroform, methanol, water, and aque¬ ous ammonia in a ratio of 130:70:8:0.5 as the developing solvent. Plates were dried and exposed to iodine vapor overnight to visualize the phospholipids. PG appeared in the 60% CH₂C1₂ fractions.

The PG-containing fractions were then evaporated to dryness under vacuum to provide 1.35 g of PG as a nearly colorless wax. TLC analysis indicated a purity of >97%, thus providing a total yield of 41%.

Example 8 Preparation of PG in THF PG was reacted with glycidol as in Example 17 above, but substituting 8.6 mL of tetrahydrofuran (THF) for p-dioxane.

The reaction yielded PG (1.24 g) as a nearly colorless wax. TLC analysis indicated a purity of about 97%, for a total yield of about 38%.

Example 9

Preparation of PG in 1,2-Dimethoxyethane

PA (3.3 g, 5.1 mmol) was dissolved in 1,2- dimethoxyethane (10 mL) . Glycidol (400 uL, 6.0 mmol) was added, and the reaction mixture allowed to stand at 21 C for five days. A second aliquot of glycidol (400 uL) was added, and allowed to react for an additional 24 hours. The solvent was then removed under vacuum to provide the product as an amber colored wax (3.41 g) . The product was dissolved in methylene chloride (20 mL) , and anhydrous ammonia gas bubbled in until the solution was saturated. The resulting ammonium salt solution was then applied to ^:a 21 mm x 200 mm chromatographic adsorption column of silica gel (Merck Kieselgel 60) . The column was eluted with CH₂C1₂ (100 mL) , followed by 100 mL of 95% CH₂C1₂ with 5% MeOH/NH₄OH, followed by 90% CH₂Cl₂-10%

MeOH/NH₄OH (100 mL) , followed by 80% CH₂C1₂~20% MeOH/NH₄OH (100 mL), followed by 60% CH₂Cl₂-40% MeOH/NH₄OH (200 L) , followed by 100% MeOH/NH₄OH (100 mL) .

The eluate was collected in 50 L fractions, and each fraction assayed by TLC on silica gel coated plates, using a mixture of chloroform, methanol, water, and aque¬ ous ammonia in a ratio of 130:70:8:0.5 as the developing solvent. Plates were dried and exposed to iodine vapor overnight to visualize the phospholipids. PG appeared in the 60% CH₂C1₂ fractions.

The PG-containing fractions were then evaporated to dryness under vacuum to provide 1.59 g of PG as a nearly colorless wax. TLC analysis indicated a purity of 84%, thus providing a total yield of 35%.

Example 10 Preparation of PG in 1,2-Diethoxyethane

PG was prepared from PA and glycidol following the procedure of Example 7, but substituting 8.6 L of 1,2-diethoxyethane for p-dioxane.

The reaction yielded 1.28 g of PG as a nearly colorless wax. TLC analysis indicated a purity of about 99%, thus providing a total yield of about 40%.

Example 11

Preparation of PG in Diethoxymethane (A) PA (1.48 g, 2.28 mmol) was dissolved in diethoxymethane (3.7 mL) , and glycidol (0.50 mL, 7.54 mmol) and Fuller's Earth (0.5 g) added. The mixtures was heated at 30 C for 96 hours. The solvent was then removed under vacuum, and the resulting waxy residue dissolved in CH₂C1₂ (15 mL), and treated with NH, gas until saturated.

The resulting ammonium salt solution was then applied to a 21 mm x 200 mm chromatographic adsorption column of silica gel (Merck Kieselgel 60, 70-230 mesh). The column was eluted with CH^Cl^ (100 mL) , followed by MeOH/NH₄OH (100 mL) , followed by 90% CH₂C1₂-10% MeOH/NH₄OH (100 mL), 80% CH₂Cl₂-20% MeOH/NH₄OH (100 mL) , and 60% CH₂Cl₂-40% MeOH/NH₄OH (200 mL) . The eluate was collected in 50 L fractions, and each fraction assayed by TLC on silica gel coated plates, using a mixture of chloroform, methanol, water, and aque¬ ous ammonia in a ratio of 130:70:8:0.5 as the developing solvent. Plates were dried and exposed to iodine vapor overnight to visualize the phospholipids. PG appeared in the 60% CH₂C1₂ fractions.

The P^'G-containing fractions were then evaporated to dryness under vacuum to provide 1.06 g of PG as a pale amber wax. TLC analysis indicated a purity of 91%, thus providing a total yield of 58%, corrected for purity. (B) PA (1.53 g, 2.36 mmol) was dissolved in diethoxymethane (10 mL) , and glycidol (0.39 mL, -5.89 mmol) added. The solution was allowed to stand at 21°C for 3 days. Then, an additional 0.39 mL of glycidol was added, and the solution allowed to stand for an additional day. The solvent was evaporated in vacuo to provide 3.67 g of light amber wax, which was then dissolved in CH₂C1₂ (20 mL) and saturated with ammonia gas.

The resulting ammonium salt solution was then applied to a 21 mm x 200 mm chromatographic adsorption column of silica gel (Merck Kieselgel 60, 70-230 mesh). The column was eluted with CH₂C1₂ (100 mL) , followed by MeOH/NH₄OH (100 mL) , followed by 90% CH₂C1₂-10% MeOH/NH₄OH (100 mL), 80% CH₂Cl₂-20% MeOH/NH₄OH (100 mL), and 60% CH₂Cl₂-40% MeOH/NH OH (200 mL). The eluate was collected in 50 mL fractions, and each fraction assayed by--TLC on silica gel coated plates, using a mixture of chloroform, methanol, water, and aque¬ ous ammonia in a ratio of 130:70:8:0.5 as the developing solvent. Plates were dried and exposed to iodine vapor overnight to visualize the phospholipids. PG exhibits an R- of between 0.7 and 0.8 in this system. PG appeared in the 60% CH₂C1₂ fractions.

The PG-containing fractions were then evaporated to dryness under vacuum to provide 1.24 g of PG as a pale amber wax. TLC analysis indicated a purity of 98.%, thus providing a total yield of 71% corrected for purity.

Although the invention has been described with respect"-^o particular materials and methods, it will be apparent to one skilled in the art that various changes and modifications may be made without departing from the invention.

Claims

IT IS CLAIMED :

1. A method of producing phosphatidylglycerol comprising preparing a mixture of phosphatidylcholine, phospholipase D enzyme, glycerol, and a soluble calcium salt in an aqueous single-phase solvent containing between about 5%-25% of a water-miscible alkyl diether, reacting the mixture for a period sufficient to convert a substantial portion of the phosphatidylcholine to phosphatidylglycerol, and extracting the phosphatidylglycerol from the mixture.

2. The method of claim 1, wherein the diether is s'elected from the group consisting of 1,1- dimethoxy ethane, 1,2-dimethoxyethane, 1,2- dimethoxypropane, 1,1-diethoxymethane, and 1-ethoxy-l- methoxymethane.

3. The method of claim 1, wherein the concentration of glycerol is between about 2%-10% by volume.

4. The method of claim 1, wherein the concentration of diether co-solvent is between about 10%- 20% by volume.

5. The method of claim 1, wherein the pH of the mixture is between about 6.0 and 9.0.

6. The method of claim 1, wherein the concentration of calcium ion is between about 10 to 100 mM.

7. The method of claim 1, wherein said reacting produces a maximum level of PG in the mixture, achieved when the amount of phosphatidylglycerol produced by conversion of phosphatidylcholine to phosphatidylglycerol catalyzed by the transferase activity of the enzyme, less the amount of phosphatidylglycerol lost by the conversion of phosphatidylglycerol to phosphatidic acid catalyzed by the hydrolase activity is a maximum, and the amount of active enzyme present in the mixture is controlled such that at the termination of the reaction, (a) no more than about 10% of the original amount of phosphatidylcholine remains at the termination of the reaction, and (b) the level of phosphatidylglycerol present in the reaction mixture is no less than about 50% of such maximum level.

8. The method of claim 7, which further includes adding additional phospholipase D to the mixture during said reacting, at an enzyme level sufficient to reduce the total phosphatidylcholine level to less than 10% of its original level.

9. The method of claim 1, which further includes preincubating the enzyme with glycerol prior to addition of phosphatidylcholine to said mixture.

10. The method of claim 1, wherein said enzyme is prepared by homogenizing cabbage to form a homogenate, forming a filtrate by removing particulate matter from the homogenate, heat-treating the filtrate, and removing particulate matter from the heat-treated filtrate.

11. The method of claim 1, wherein said extracting ^"includes adding calcium hydroxide to the re¬ action mixture to a final pH of between about 8-9 to convert phosphatidylglycerol and phosphatidic acid substantially completely to insoluble calcium salts, removing such insoluble salts from the reaction mixture, and extracting phosphatidylglycerol selectively from the removed insoluble material.

12. The method of claim 11, which further includes treating the extracted phosphatidylglycerol with an ion exchange resin to convert the calcium phosphatidylglycerol salt to an ammonium phosphatidylglycerol salt.

13. The method of claim 1, wherein said extracting yields a mixture of phosphatidylglycerol and phosphatidic acid, which further includes converting the phosphatidic acid in the mixture to phosphatidylglycerol by the steps of: forming a second reaction mixture containing the phosphatidylglycerol and phosphatidic acid and glycidol in a suitable alkyl ether; adjusting the temperature of the second reaction mixture to about 10-60°C, for about 10 hours to about 2 weeks; and isolating the phosphatidylglycerol produced thereby.

14. The method of claim 13, wherein said extracting includes adding calcium hydroxide to the first- mentioned reaction mixture to a final pH of between about 8-9 to convert_, phosphatidylglycerol and phosphatidic acid substantially completely to insoluble calcium salts, removing such insoluble calcium salts from the first- mentioned reaction mixture, and converting the calcium salts to protonated forms of phosphatidic acid and phosphatidylglycerol.

15. A method of treating a phospholipid enzymatically with phospholipase A., , A₂, C, or D, said method comprising preparing a mixture of the phospholipid and phospholipase in an aqueous single-phase solvent contain- ing between about 5% - 25% of a water-miscible alkyl diether, and reacting the mixture under conditions which al¬ low enzymatic conversion of the phospholipid by the phospholipase.

16. The method of claim 15, wherein the diether is selected from the group consisting of 1,1- di ethoxymethane, 1,2-dimethoxyethane, 1,2- dimethσxypropane, 1,1-diethoxymethane, and 1-ethoxy-l- methoxymethane.

17. A process for preparing phosphatidyl¬ glycerol from phosphatidic acid, comprising: dissolving or suspending phosphatidic acid and glycidol in a suitable alkyl ether to form a reaction mixture; adjusting the temperature of the reaction mixture to about 10-60 C, for about 10 hours to about 2 weeks; and isolating the phosphatidylglycerol produced thereby.

18. he_t process of claim 17, wherein said phosphatidic acid is supplied in the form of a salt, and said reaction mixture is adjusted to a pH low enough to substantially protonate said glycidol using a non- nucleophilic protic acid.

19. The process of claim 17, wherein said suit- able alkyl ether is 1,4-dioxane, tetrahydrofuran, diethoxymethane, dimethoxymethane, 1,2-diethoxyethane, or 1,2-dimethoxyethane, and the ratio of phosphatidic acid to glycidol is about 1:1 to about 1:10.

20. A composition useful for preparing phosphatidylglycerol, which composition comprises: phosphatidic acid; glycidol in a ratio of about 10:1 to about 1:1 glycidol:phosphatidic acid; and a suitable alkyl"ether, in an amount sufficient to dissolve said phosphatidic acid and said glycidol.