Molecular species of mono-, di-, and triphosphoinositides of bovine brain

The mono-, di-, and triphosphoinositides of bovine brain were isolated by chromatography on columns of DEAE-cellulose, alumina, and silicic acid. The major molecular species in each phosphoinositide class were identified and quantitatively estimated by combined thin-layer and gasliquid chromatography of the component diglycerides, which were released by hydrolysis with a specific brain phosphodiesterase. The diglycerides were treated with pancreatic lipase, and the positional distribution of the fatty acids was determined. Over 27 molecular species were identified, and these accounted for about 95% of each phosphoinositide class, but the 1-stearate 2-arachidonate derivative contributed more than 40% of the total in each class. The other molecular species also were qualitatively and quantitatively similar in the three phosphoinositide classes. All the long-chain and polyunsaturated acids were confined to the 2-position and were preferentially paired with stearic acid in the 1-position. Oleic acid in the 2-position was about equally divided between species with palmitic and stearic acids in the 1-position. These results suggest that the mono-, di-, and triphosphoinositides of the bovine brain have similar compositions and that the various molecular species may he metabolically related. SUPPLEMENTARY


SUPPLEMENTARY KEY WORDS fatty acids
. positional distribution . diglycerides . diglyceride acetates R E C E N T studies have revealed that natural lecithins a n d cephalins are made u p of characteristic molecular species (l), which are metabolized a t distinctly different rates (2-4). Due to analytical difficulties it has not been possible to obtain evidence of the generality of this phenomenon in other glycerophosphatides, such as phosphatidylserine a n d phosphatidylinositols.
T h e treatment of phosphoinositides with a specific brain phosphodiesterase ( 5 ) of high activity makes it possible to isolate the diglyceride moieties, which can then be subjected to detailed chromatographic fractionation a n d quantitation. I n the present work we have identified a n d quantitatively estimated the molecular species of the mono-, di-, a n d triphosphoinositides of bovine brain.

MATERIALS AND METHODS
The methods a n d general experimental conditions were similar to those previously described (6, 7), except as noted below.

Preparation of Phosphoinositides
Phosphoinositides were prepared from ox brain obtained fresh from the slaughterhouse a n d transported to the laboratory on dry ice. Tri-a n d diphosphoinositides were isolated by a modified version of the procedure of Hendrickson a n d Rallou (8). T h e inositide-rich fraction of Hendrickson a n d Ballou was precipitated five times from chloroform by the addition of methanol; it was then dissolved in 500 ml of chloroform-methanol 1 : 1 (v/v).
T h e lipids were converted to the free acid form by vigorous shaking with 100 ml of 1 N HCl. T h e two phases were allowed to separate, the upper phase was discarded, a n d the lower chloroform phase was taken to dryness in Abbreviations: MPI, monophosphoinositides; DPI, diphosphoinositides ; TPI, triphosphoinositides ; GLC, gas-liquid chromatography; TLC, thin-layer chromatography; c 1 4 -c 2 2 , fatty acids with 14-22 carbon atoms; C d h , diglyceride acetates with a total number of fatty acid carbon atoms (including those of acetic acid) of 32-46. Fatty acids designated by number of carbon atoms: number of double bonds: number of carbons from the last double bond to end of chain. a rotary evaporator. The lipid residue was dissolved in a minimal volume (about 5 ml) of chloroform, and 7 volumes of methanol were added. The solution was left for 2 hr at -15"C, and the small precipitate that formed was centrifuged down and discarded. The methanolic solution was held in an ice bath, and 0.1 N NaOH i n methanol was carefully added to neutralize and precipitate the lipids as sodium salts. The precipitate was collected by centrifugation and redissolved in ehloroformmcthanol-water 20 : 9 : 1 (v/v/v) in the appropriate dilution for loading onto the DEAE-cellulose column. The linear gradient elution was begun with 1700 ml of chloroform-methanol-water 20 : 9 : 1 (v/v/v) in the mixing chamber and 1700 ml of 0.54 M ammonium acetate in chloroforn-methanol-water in the solvent reservoir. The less steep gradient improved the separation of diand triphosphoinositides. Hendrickson and Ballou (8) found that, i n addition to the major triphosphoinositide peak which was eluted with approximately 0.5 M ammonium acetate, a minor peak containing, presumably, a mixed calcium and magnesium chelate of triphosphoinositide was rapidly eluted in approximately 0.1 M salt. No such early component was observed in the modified procedure, and it is likely that the preliminary treatment with acid, in addition to saving considerable time, was more effective in removing divalent cations from the preparation. The characterization and chromatographic purity of the lipids were as previously described (5). Triphosphoinositide appeared as a single spot on TLC. The diphosphoinositide preparation contained less than 5Yc of triphosphoinositide.
For the isolation of monophosphoinositidc, 1 kg of ox brain was extracted with a total volume of 4 liters of chloroforn-methanol 1 : 1 (v/v), and the mixture was centrifuged. The lipid extract was taken to dryness under reduced pressure. The residue was dissolved in 500 ml of chloroforiii-inethanol 2: 1 (v/v), and the solution was shaken with 100 ml of 0.9% NaCl; the aqueous phase was discarded. The lipids were applied to a column of alumina (1 mg of phosphorus per g of alumina) in chloroform-methanol 1 : 1, and eluted as described by Dawson (9). The inositide-rich fraction, which was eluted with ethanol-chloroform-water 5 : 2: 2 (v/v/v), was taken to dryness, redissolved in chloroform-methanol 4 : l (v/v), and adsorbed on a column of silicic acid (1 0 ) . The monophosphoinositide was eluted in chloroform-methanol 3 : 2 (v/v), and samples were examined by T L C ( l l ) , and by paper chromatography and electrophoresis of the water-soluble phosphate esters obtained by mild alkaline hydrolysis (1 2). The preparation was found to be contaminated with a ninhydrin-positive material with properties consistent with those of lysophosphatidylethanolamine. This was completely removed from the monophosphoinositide by separation on a column of DEAE-cellulose (Whatman DE-32), acetate form, essentially as described by Rouser, Kritchevsky, Heller, and Lieber (1 3). The lysophosphatidylethanolamine was eluted with chloroform-methanol 7 : 3 (v/v), and the monophosphoinositide with chloroform-meth-anol4 : 1 (v/v) containing 10 ml of concentrated NH40H per liter. The monophosphoinositide contained phosphorus, inositol, and free fatty acids in the molar proportions 1 :0.97: 1.96 and traveled as a single spot on thin-layer chromatograms.

Hydrolysis of Phosphoinositides
Triphosphoinositide was hydrolyzed to a diglyceride and a water-soluble inositol triphosphate as previously described (5) by a specific phosphodiesterase present in high concentration in aqueous extracts from acetone powders of ox brain. The conditions were modified for diphosphoinositide degradation as follows : duplicate samples of diphosphoinositide (26.3 mg) were incubated in 42 ml of 0.05 M Tris-HC1 buffer (pH 7.2) containing brain extract (9.5 mg of protein) and 0.025% cetyltrimethylammonium bromide at 37OC for 60 min. Since the activity of the ox brain extracts was somewhat variable with respect to monophosphoinositide, this lipid was hydrolyzed by the enzyme obtained from the supernatant fraction of a rat brain homogenate (14). Duplicate samples of monophosphoinositide (17.5 mg) were incubated in 30 nil of 0.05 M acetate buffer (pH 5.4) containing brain extract (6.2 mg of protein), 0.001 M CaC12, and 0.01 M cysteine at 37OC for 90 min. In each case, after incubation, the diglycerides were removed by extracting the incubation mixture four times with 40 ml of diethyl ether. Samples of the aqueous phase were analyzed for water-soluble phosphates (1 5) to determine the percentage hydrolysis of the substrates. It was calculated that each lipid was hydrolyzed to the extent of 60-7070.

Preparation and Resolution of Diglyceride Acetates
The diglycerides were purified by T L C and acetylated as previously described (6). The acetates were resolved on the basis of total number and distribution of double bonds per molecule by argentation T L C (6, 7). The diglyceride acetates corresponding to the various groups of unsaturation were separately recovered by suspending the gel scrapings in diethyl ether-methanol-acetic acid 60:40:1 (v/v/v) (0.5 g/5 ml), diluting the mixture to 25% water, and extracting the three-phase suspension with an equal volume of petroleum ether (three times). The ratios of the diglycerides among and within the various bands were determined by GLC (6, 7) using tridecanoin (50-300 pg) as an internal standard.

Analysis of Fatty Acids
The fatty acids were quantitatively estimated by GLC following methylation or transmethylation (6, 7). The identities of the unsaturated acids were confirmed by argentation T L C and hydrogenation. The positional distribution of the fatty acids in the mono-, di-, and triphosphoinositides was determined by analysis of the acids which were released when the corresponding diglycerides (7) were treated with pancreatic lipase. The acid distribution in the various diglyceride acetates recovered from the silver nitrate fractionation was determined in a similar manner. The positional placement of the fatty acids in the tri-and diphosphoinositides found by pancreatic lipase agreed well with that obtained previously (5) for the intact phosphoinositides in experiments with phospholipase A2 of bee venom. In all cases, the reaction products as well as any unreacted starting materials were purified and recovered by TLC, prior to GLC analysis of the fatty acids.

Calculations
The structure of the original phosphoinositides was determined by proportional summation and normalization of all the analytical data. The accuracy of the final result was estimated by matching the mole percentage composition, and positional distribution of the fatty acids in the original phosphatides, against those determined for the derived acetates. In several instances the experimental data were compared with random distribution data which were derived by calculation (6). The three types of phosphoinositides were compared with respect to the positional distribution and molecular association of the fatty acids as described previously for phosphatidylcholine and phosphatidylethanolamine of egg yolk (7).

RESULTS
The yields of phosphoinositides obtained from 1 kg of ox brain were 550, 400, and 500 pmoles, respectively, for the tri-, di-, and monophosphoinositides. Wells and Dittmer (1 6 ) found concentrations of 500-800 pmoles/kg for triphosphoinositide and 650-1 000 pinoleslkg for diphosphoinositides, depending on the extraction procedures used. Dawson and Eichberg (1 7) have reported values of 61 5 pmoles/kg for triphosphoinositide and 45 pmoles/kg for diphosphoinositide. These discrepancies emphasize the problem of postmortem autolysis as well as the consequences of the procedures employed to extract fresh brain. No comparable values appear to be available for the monophosphoinositide content of ox brain. Our yield is likely to be low since the isolation procedure involved three separate column steps, and narrow cuts of the column eluates were taken to reduce contamination with other fractions. Because of the large amounts of brain required to prepare the lipids in substrate quantities and the relative inefficiency of some of the methods of enrichment, the monophosphoinositides and polyphosphoinositides came from different brain samples.
In view of the rapid autolysis and interconversion of the phosphoinositides that may occur during extraction and the different methods employed for their isolation, we cannot be absolutely certain that the final preparation represented all the native molecular species. Table 1 gives the over-all composition and the positional distribution of the fatty acids of the three phosphoinositide classes. All are characterized by a high content of stearic and arachidonic acids, which is similar to the findings of others (5,8,18) for the di-and triphosphoinositides. The arachidonate is specifically found in position 2, while the stearate is attached to position 1 of the glyceride residue as also noted previously (5). Oleic acid is the other major unsaturated acid in these phosphatides; it is located in both positions 1 and 2. All phosphoinositides also contained significant amounts of eicosatrienoic acid, which was made up of nearly equal proportions of the 5, 8, 11 (n-9) and the 8, 11, 14 ( n -6 ) isomers. Both acids occurred exclusively in the 2-position. The identification of the two trienoic acids in the phosphoinositides of the bovine brain coincides with the findings in the human brain (19). There are minor differences in the relative amounts of the trienoic fatty acids in the three phosphoinositides, but the importance of these discrepancies remains to be established in view of the differences in the origin of the samples. Table 2 gives the fatty acid composition of the silver nitrate subfractions of the diglycerides derived from the mono-, di-, and triphosphoinositides. The estimated molar proportions of the acids are seen to approximate closely the ratios required for the pairing of all components according to the degree of unsaturation of the subfraction. The diglyceride moieties analyzed were free of alkenyl ethers as indicated by the absence of dimethylacetal peaks from the gas chromatograms of the fatty acids. This is supported by the nearly theoretical proportions of phosphorus, inositol, and fatty acids in these phosphatides. Appreciable quantities of plasmalogens are found in ox brain, but these appear to be largely confined to the ethanolamine phosphoglycerides (20). Table 3 gives the molecular weight distribution of the total mono-, di-, and triphosphoinositides of the bovine brain as estimated by GLC of derived diglyceride acetates. In all phosphatide classes the major constituents   * Diglyceride acetates resolved by TLC on silver nitrate-treated plates. Each fatty acid reported as a percentage of saturation class. t Estimates of reconstituted composition obtained by algebraic summation and normalization of data from $ Random distribution determined as previously described (6).

Molecular Weight Distribution
argentation TLC.
are those with one Cl8 and one CZO (C,,), with two C18 ( c 3 8 ) , and with one (218 and one CZZ (c,~) fatty acid per molecule. The similarities in the proportional contributions of the phosphoinositides of the three types reflect similarities in both the fatty acid composition and the pairing. The finding of large amounts of the species containing CIS and CZO acids was expected in view of the high proportion of stearic and arachidonic acids in these phosphatides. There exists, however, considerable selectivity in the over-all association of the fatty acids in the phosphatide molecules as shown by the difference between the experimental and the randoln values derived by calculation from the fatty acid composition. Table 3 also includes the reconstitution values obtained for the diglyceride acetates by algebraic summation and normalization of the data from argentation TLC.
The close agreement between these and the original total indicates that the execution of the detailed analytical procedure was not accompanied by significant losses of the polyunsaturated acids due to autoxidation. Table 4 gives the molecular weight distribution of the major diglyceride acetates recovered from the silver nitrate fraction for each class of phosphoinositides, along with the random values calculated on the basis of the fatty acid composition of the appropriate fractions. It is seen that only a few major diglyceride acetates or phosphatidylinositols occur in any one group of molecules having a given degree of unsaturation. This is true for all three phosphoinositide classes, and is i n contrast with the random values, which usually cover a wider range of molecular weights ( Table 3 ) .
The proportions of the different unsaturation groups in the total phosphatide mixture, however, show some variation. Thus, while the monoenes make up about 10% of the mono-and diphosphoinositides, the triphosphoinositides contain only 6% of this fraction. In contrast, both tri-and diphosphoinositides contain more than twice as much trienes as do the unonophosphoinositides. The triphosphoinositides contain about three times as much pentaenes as the monophosphoinositides and twice as much as the diphosphoinositides. Although the saturated fatty acids accounted for 41-46y0 of the total, no fully saturated species were recognized in any of the phosphoinositide classes.  that as much as 20Q/, of saturated phosphatides could have occurred by a random association of the fatty acids in these glycerol esters. The random estimates for the other saturation classes were similarly quite different from the experimental values which illustrate the high degree of specificity noted for most fatty acid pairings.

DISCUSSION
On the basis of the analytical data presented above we reconstituted the original composition of the three classes of phosphoinositides. Table 5 lists the major individual nionophosphoinositides as well as their mono-and diphospho derivatives. The molecular species have been specified as previously explained (6). The species with reversed fatty acid placement occur in much smaller amounts, but can be estimated from the fatty acid distributions reported in Table 1. The estimates were derived by reconstitution of the over-all molecular weight distribution (Table 3), the molecular weight distribution of the subfractions (Table 4), and the fatty acid compo-sition of the subfractions (Table 2). O n the basis of the data given in Tables 2-4 and the knowledge of the positional placement of the fatty acids, it can be seen that the monounsaturated inositides in each phosphatide class are largely made up of the palmitoyl and stearoyl oleates, which occur in nearly equal proportions, although the combined total in the triphosphoinositides is only about one-half of the total in the mono-and diphosphoinositides. In view of the relatively small total amount of palmitic compared with stearic acid in these phosphatides, the finding of equal proportions of palmitic and stearic acid species in the monoenes represents a preferential enrichment of palmitic acid in this fraction.
The major dienes in all three phosphatide classes were made up of the dioleoyl and the 1-stearoyl 2-eicosadienoyl species, which occurred in nearly equal proportions in any one phosphatide, although the total differed in each of the phosphatides. Only the monophosphoinositides contained a significant proportion of 1 -stearoyl 2-linoleoyl species. The small amount of the l-palmitoyl 2-linoleoyl species in this inositide was in keeping with the proportion of palmitic acid in the total saturated acids. half the amount of this triene found in the di-and triphosphoinosi tides. The main tetraene in each phosphatide was the 1stearoyl 2-arachidonoyl species. In addition, the monophosphoinositide also contained significant amounts of the 1 -palmitoy1 2-arachidonoyl species as well as thc 1-oleoyl 2-docosatrienoyl species. The diphosphoinositides had particularly high levels of the latter species.
The pentaenes were the most complex of all the fractions examined. The data in Table 5 give only a partial indication of the complexity, as many species had to be deleted because their proportion of the total was so minor. Nevertheless, in all three phosphatides, the 1oleoyl 2-arachidonoyl species could be recognized as the major component; it was most abundant in the triphosphoinosi tides. These data show that the phosphoinositides possess glyceride structures comparable with those identified for other natural glycerophosphatides (1, 4, 6, 7). Furthermore, the three inositides are made up of molecular    Table 5. verted to the higher phosphates to the same extent. Alternatively, the difference in the quantitative coniposition could have resulted from a rapid postmortem degradation of the polyphosphoinositides. In the latter case, however, a differential hydrolysis would have to be implicated to account for the discrepancies.
In view of this uncertainty, we compared the molecular species in the three inositides on the basis of the nature of the major fatty acids esterified and the degree of preference shown (7). It was hoped that such a comparison might reveal similarities among the phosphatides that are no longer perceptible due to secondary changes in the composition of the inolecular species. The degree of preference was calculated by setting up a characteristic ratio of proportions as shown in Table 6, which also gives the results of some of the associations thus tested. It is seen that in all three phosphoinositides, for each molecular species there is a strikingly similar pairing of the major unsaturated fatty acid with a particular saturated acid. All the phosphatides specifically combine palmitic and oleic acids in the monoenes although there are significant amounts of stearic and oleic acid pairs. In contrast, stearic acid is predominant over palmitic acid in the formation of the dienes, trienes, and tetraenes of the corresponding eicosaenoic acids.
The present findings are consistent with the concept that polyphosphoinositides originate from the corresponding monophosphoinositides, but that rates of phosphorylation of different species may be different. These conclusions require further confirmation by metabolic experiments.