Metabolism of glycerophospholipids of myelin and microsomes in rat brain. Reutilization of precursors.

The metabolic turnover of phospholipids of rat brain myelin and microsomes was investigated after 17-day-old animals received intracranial injections of [Z-“HIglycerol and either [1,3-14C]glycerol, [1,2-14Clcholine, [l-‘%lacetate, [1,2%]ethanolamine, [methyl-14C1choline, o-[U-%lglucose, or [‘<‘P]orthophosphate. The turnover rate, with respect to the “H at the C-2 position of the glycerol moiety of individual lipids, was calculated for the first 15 days after injection (rapid phase) and for the time period between 15 and 80 days following injection (slow phase). The results for the half-life of phosphatidylcholine and phosphatidylethanolamine in microsomes were similar (3 to 4 days in the fast phase and 13 to 14 days in the slow phase, respectively). In myelin, the corresponding values were 6 to 10 days in the fast phase and 25 days in the slow phase. The isotope ratio (“H/Y or aH/:j”P) was determined for each lipid of microsomes and myelin at each time point studied. In each case, the ratios declined as a function of time, indicating a preferential recycling of the 14Cor ssPlabeled precursor relative to [2-“HIglycerol. Reference to the data for the turnover rate of [“HIglycerol in these phospholipids (see above), together with the isotope ratio data, made possible a calculation of the apparent half-life of the other radioactive moiety (incorporated as base, phosphate group, fatty acid, or vinyl ether) with many fewer data points than would be the case if experiments involving only a single radioactive label were carried out. When [‘%lcholine or [14Clethanolamine was a precursor, the half-lives of the base moieties of phosphatidylcholine and phosphatidylethanolamine were both 26 days in the microsomes and 39 and 33 days, respectively, for myelin. The half-life of ethanolamine in the plasmalogen was significantly longer, 40 and 58 days, in microsomes and myelin,

The metabolic turnover of phospholipids of rat brain myelin and microsomes was investigated after 17-day-old animals received intracranial injections of [Z-"HIglycerol and either [1,3-14C] The turnover rate, with respect to the "H at the C-2 position of the glycerol moiety of individual lipids, was calculated for the first 15 days after injection (rapid phase) and for the time period between 15 and 80 days following injection (slow phase). The results for the half-life of phosphatidylcholine and phosphatidylethanolamine in microsomes were similar (3 to 4 days in the fast phase and 13 to 14 days in the slow phase, respectively).
In myelin, the corresponding values were 6 to 10 days in the fast phase and 25 days in the slow phase.
The isotope ratio ("H/Y or aH/:j"P) was determined for each lipid of microsomes and myelin at each time point studied. In each case, the ratios declined as a function of time, indicating a preferential recycling of the 14C-or ssPlabeled precursor relative to  Reference to the data for the turnover rate of ["HIglycerol in these phospholipids (see above), together with the isotope ratio data, made possible a calculation of the apparent half-life of the other radioactive moiety (incorporated as base, phosphate group, fatty acid, or vinyl ether) with many fewer data points than would be the case if experiments involving only a single radioactive label were carried out.
When ['%lcholine or [14Clethanolamine was a precursor, the half-lives of the base moieties of phosphatidylcholine and phosphatidylethanolamine were both 26 days in the microsomes and 39 and 33 days, respectively, for myelin. The half-life of ethanolamine in the plasmalogen was significantly longer, 40 and 58 days, in microsomes and myelin, respectively. The half-lives for ["'PIphosphate in each of the individual lipid classes were slightly shorter or the same as that of the labeled base.
With respect to ['%lacetate as a precursor, the half-lives of phosphatidylcholine and phosphatidylethanolamine were 28 and 65 days, respectively.
The corresponding values for the myelin fraction were 54 and 125 days. Therefore, although the glycerol moieties of phosphatidylcholine and phosphatidylethanolamine are metabolized at about the same rate, acyl moieties are preferentially reutilized for synthesis for phosphatidylethanolamine.
Since the work of Smith (see Ref. 1 for a review), it has been generally accepted that the lipids in myelin are not metabolically stable. Values reported for the half-life of phospholipids from rat brain subcellular fractions vary considerably (e.g. the reported half-life of phosphatidylcholine ') in myelin varies from less than 10 days (2) to 167 days (3). Sources of experimental variation that might explain these differences within a given species include injection routes, age of animals, and the use of different labeled precursors, which may be reutilized to different extents. The suggestion that the use of different labeled precursors can lead to differences in observed half-life for a given membrane phospholipid is supported by previous results in diverse membrane systems (4, 5), including brain (6-10).

Metabolism
of Glycerophospholipids in Rat Brain by a rapid interchange of glycerol 3-phosphate with dihydroxyacetone phosphate, with accompanying loss of "H (via FHlNADH+) to water Cl), a reaction catalyzed by glycerophosphate dehydrogenase (EC 1.1.1.8). The "H may also be lost to the respiratory chain by a mitochondrial glycerophosphate dehydrogenase (EC 1.1.99.5) and transferred to water without the mediation of a reduced pyridine nucleotide.
The experimental approach consisted of giving injections to young rats of [2-"HIglycerol and a W-or :j2P-labeled precursor of another phospholipid constituent (e.g. ['YIJ]acetate to label acyl or alkenyl moieties, [32Plphosphate to label the phosphoryl moiety, or ['Clcholine to label the base of phosphatidylcholine). The change in "H/W (or "H/"'P) isotope ratio with time was used to determine the extent of reutilization of the W-or "'P-labeled moiety for de nouo synthesis of phospholipids relative to the turnover of [2-"HIglycerol.
These experiments were carried out simultaneously with lipids of the microsomal and myelin fractions to clarify the possible role of subcellular compartmentation with regards to reutilization of phospholipid precursors. The data obtained in this study also made possible certain conclusions about the relative roles of the glycerol 3-phosphate and dihydroxyacetone phosphate pathways in the synthesis of diacyl phospholipids and alkenyl-acyl phospholipids. . This procedure separates the C-l carbon of glycerol (precipitate) from the C-2 and C-3 carbons (soluble).
The :%H in the precipitate and the soluble fraction were determined as described below, and the proportion of the label in the soluble fraction was calculated.
This corresponded to radioactivity in the C-2 position (see below). Although no "H radioactivity was expected at the C-3 position, a control was used to confirm this. Individual lipids, isolated from the membrane fractions at various times after injection with [2-3Hlglycerol, were subjected to phospholipase D treatment (19). The phosphatidic acid product was degraded as described by Agranoff and Hajra (18). This procedure, which distinguishes between 3H label on each of the glycerol carbons, showed no label at the C-3 position.
Measurement of Radioactivity -For determination of radioactivity in individual lipids following thin layer chromatography, the silica gel containing the phospholipid was scraped into a scintillation vial, the silica gel was deactivated by addition of 0.4 ml of water, and 10 ml of Triton X-lOO/toluene (1:4 by volume) scintillation solvent (10) was added. Samples from alkaline methanolysis or from the glycerol degradation procedure were also counted in this solvent (samples in organic solvents were first evaporated to dryness the Pearson product moment correlation coefficient (r) was calculated (21) and the probability (p) of r equal to zero was determined.
For determination of the half-life of a specific lipid, the points within the desired time interval were fit by linear regression using the log of the counts and time after injection.
The range was determined by taking 1 S.D. of the slope and adding it to and subtracting it from the calculated slope. These slopes were divided into log 2 to give the extreme values for half-lives. Because this is a logarithmic function, the range, in days, is not symmetrical on both sides of the calculated half-life. The smooth curve fit to the experimental points of Fig. 3 is a linear regression to the exponential function: Y = P,e"%' + P,,e"" + P, where Y equals log dpm and X equals time after injection in days.
This function gave a better fit than either a hyperbolic or a simple exponential equation.

Experimental Plan
Litters of 8 to 10 animals were given injections of ["HIglycerol and one of the other isotopes, and individual animals were killed at various times over the subsequent 80day period. Myelin and microsomes were isolated, and lipids were extracted. The total lipid extract obtained from each subcellular fraction was divided and subjected to thin layer chromatography in duplicate. Individual lipid classes from one chromatography plate were scraped directly into scintillation vials, radioactivity was quantitated, and the absolute 3H radioactivity as well as the "H/14C or "H/"*P ratio was tabulated.
Lipid classes isolated from the second plate were collected and subjected to chemical degradation to determine label distribution within the lipid class. Despite the fact that a different 14C or :j2P compound was injected in each experimental protocol, each litter represented a replicate experiment with regards to [3Hlglycerol. Since some of the experiments were repeated, approximately 70 experimental points were available to determine half-lives of the glycerol moiety of each lipid in the two subcellular fractions. Results regarding phosphatidylcholine, phosphatidylethanolamine, and phosphatidylethanolamine plasmalogen are given in detail.
Results, but not individual data points, are described for phosphatidylcholine plasmalogen. Although statistical analysis of the phosphatidylcholine plasmalogen data indicates a high degree of significance, it is possible that the results were systematically biased during the collection of samples from the chromatography plates by inclusion of small amounts of phosphatidylcholine.
Because of the much greater amount of phosphatidylcholine relative to phosphatidylcholine plasmalogen (22) and the concomitant greater levels of radioactivity, a trailing of 2% of the phosphatidylcholine into the phosphatidylcholine plasmalogen region can introduce a possible systematic error of up to 50% for phosphatidylcholine plasmalogen. Thus, the results, although briefly reported, may simply reflect contamination of the phosphatidylcholine plasmalogen by phosphatidylcholine.
This was potentially more of a problem in the present work (since we were interested in quantitative recovery and therefore scraped a large area of the silica gel) than in a previous study (10) (in which we were interested only in isotope ratios and could selectively scrape only the phosphatidylcholine plasmalogen area not likely to be contaminated by phosphatidylcholine).
The possibility that a small fraction of one of the ester-linked fatty acids of phosphatidylcholine is hydrolyzed, leading to contamination of the phosphatidylcholine plasmalogen, also has not been eliminated.
Data for phosphatidylserine were also subject to sampling errors because, under our chromatography conditions, this lipid trailed severely. The data for this lipid in myelin was so scattered that it was not significant and is not reported. The statistically significant data for phosphatidylserine in microsomes are briefly indicated.

Metabolism of the Glycerol Moiety
Relative Distribution of 3H between Glycerol and Acyl Moieties - Fig. 1 shows the percentage of lipid 'H in glycerol for phosphatidylcholine, phosphatidylethanolamine, and phosphatidylethanolamine plasmalogen in microsomes and myelin. The results, with respect to initial incorporation of 3H into 100 ,  I  I  I  ,  I  1  ,  I,  I  I  ,  1  Individual animals were killed at the indicated times, the myelin and microsomal fractions were isolated, total lipids were extracted, and a portion (50%) was fractionated by thin layer chromatography. Individual lipid classes were eluted and subjected to alkaline methanolysis. The proportion of 3H in the aqueous phase (presumably in the glycerol moiety of glycerophosphoryl base) and organic phase (acyl methyl esters) was quantitated.
Lines were fit to the data points by a linear regression program (p i 0.001 in each case), with the exception of the myelin phosphatidylethanolamine plasmalogen (PE PI) data, which was fit by visual inspection.
In this and subsequent figures, the data for phosphatidylethanolamine plasmalogen does not include the radioactivity in the alkenyl moiety, which is cleaved during the thin layer chromatography procedure.
In these experiments, the animals received either a ['Qlipid precursor or [32Plorthophosphate in addition to the [2-3Hlglycerol.
In this (and subsequent figures), the YH radioactivity levels studied can be back-calculated from data in Fig. 3. Even at the later time points, radioactivity levels were well above those necessary for statistically significant results. The average values for 3H in glycerol at the last time point for phosphatidylcholine (PC); phosphatidylethanolamine GPE), and phosphatidylethanolamine plasmalogen of microsomes were 1657, 534, and 228 dpm, respectively.
For myelin, the equivalent values were 4488, 965, and 1085. These numbers represent radioactivity on the thin layer plate from one-half of the total lipid extracted from the subcellular fractions. Material remaining after a portion was taken for this label distribution study was used for quantitation of "H radioactivity ( Fig.  3) and for determination of isotope ratios (following figures).

Metabolism of Glycerophospholipids in Rat Brain
the glycerol portion of the phospholipid relative to acyl moieties (80 to 95%) and with respect to decrease with time of percentage of lipid "H in glycerol, do not vary markedly for any of the lipids, except for the initially more severe drop in the case of myelin phosphatidylethanolamine plasmalogen. For phosphatidylethanolamine plasmalogen, the amount of label in the long chain hydrocarbons would be approximately doubled if the cleaved alkenyl residue was accounted for. Results were also obtained for microsomal phosphatidylserine (y intercept = 89, slope = -0.141, microsomal phosphatidylcholine plasmalogen (y intercept = 93, slope = -0.28), and myelin phosphatidylcholine plasmalogen (y intercept = 94, slope = -0.30).
Percentage of rSHIGlycerol at the C-Z Position-Phosphatidylcholine and phosphatidylethanolamine in myelin and microsomes initially have all of the [3Hlglycerol in the C-2 position ( Fig. 2). Even by 80 days following injection, most of the label is retained at the C-2 position, and in the extreme case, that of phosphatidylethanolamine in microsomes, less than 20% of the label is in the C-l position at 80 days following injection. The phosphatidylethanolamine plasmalogen data show a contrast between the two subcellular fractions. In microsomes, phosphatidylethanolamine plasmalogen behaves in a manner analogous to the other lipids, whereas in the myelin lipid a large percentage of the label is found at the C-l of the glycerol moiety of phospholipids following intracranial injection of [2-3Hlglycerol into l7-day-old rats. Animals were given injections, and individual lipid classes were obtained from microsomes and myelin at various times after injection (legend to Fig. 1). Glycerol was isolated, the proportion of 3H in this moiety present at the C-2 position was established following chemical degradation (see "Experimental Procedures"), and the data points were fit by a linear regression program. For microsomal phosphatidylethanolamine (PE) and phosphatidylethanolamine plasmalogen (PE P1) and myelin phosphatidylethanolamine plasmalogen, p < 0.05. For microsomal phosphatidylcholine (PC) and myelin phosphatidylcholine and phosphatidylethanolamine, p is large as would be expected, since it is a measure of the significance that the points fall on a line with a slope equal to zero, i.e. the estimated slope of these lines is close to zero. The actual range of total radioactivity (C-l and C-2) measured for each lipid class was similar to that described in Fig. 1.
Turnover of [2-'%IIGlycerol -"H radioactivity in individual lipids was calculated, on the basis of the fraction of the original homogenate analyzed, to give the value expected if the entire preparation of microsomes or myelin had been used for analysis. Since the actual amount of ["HIglycerol varied from experiment to experiment, this value was normalized to that expected if 100 &i of [3H]glycerol had been injected. These values were further corrected to include only the "H in the C-2 position of glycerol using the data in Figs. 1 and 2. The resultant data were plotted as log disintegrations per min versus time after injection (Fig. 3). There is clearly more than one component of decay; straight lines were arbitrarily fitted separately for the first 15 days and for the later time period (dashed lines in Fig. 31, and the half-lives were calculated from these lines. In each case, the half-life for the first 15 days following injection was significantly shorter than the half-life for the subsequent 2-month period. Half-lives for individual lipids in each membrane fraction and the statistical significance of the linear approximations (dashed line) are indicated in Fig. 3. Although the data points are not shown, the half-life of phosphatidylserine in microsomes in the first 14 days was 4.5 days (range, 3.5 to 6.0 days), and for the subsequent 2-month period it was 18 days (range, 15 to 23 days). Microsomal phosphatidylcholine plasmalogen turnover for the first 14 days was 5.0 days (range, 4.0 to 7.0 days), and for the next 2 months it was 11 days (range, 10 to 12 days). Myelin phosphatidylcholine plasmalogen turnover for the first 15 days was 8.5 days (range, 6.5 to 121, and for the subsequent 2-month period it was 31 days (range, 27 to 38 days).
A smooth curve (solid lines on Fig. 3) was also fit to the experimental points. No biological interpretation of the complex equation (see "Experimental Procedures") that fits the curve is feasible. However, readings from this curve were used to normalize the isotope ratio data (see below).
For establishment of the proportion of lipid 14C that could be attributed to glycerol, an aliquot of the lipids was subjected to alkaline methanolysis, and the percentage of lipid 14C in glycerol was plotted as a function of time (Fig. 4).
The isotope ratio data were corrected to include only the "H at the C-2 position of glycerol (using Figs. 1 and 2) and only the 14C in glycerol (using Fig. 4), and the "H/14C ratio for each lipid was plotted as a function of time (Fig. 5). The initial :'H/'V ratio was less than 3 for myelin phosphatidylethanolamine plasmalogen, relative to the 3H/14C ratio of over 30 for myelin phosphatidylethanolamine and phosphatidylcholine. This marked difference is most easily accounted for by the assumption that dihydroxyacetone phosphate is an intermediate in the synthesis of the plasmalogen. This effect is not as marked in microsomal phosphatidylethanolamine plasmalogen, although the differences in its isotope ratio as compared to the microsomal diacyl lipids are still highly significant. The contrast between the long term stability of the isotope ratio for Procedures"). Data for time points 15 days or more after injection were treated in a similar fashion. Note that the r values refer to the fit of the linear approximation rather than to the much better fit of the solid line curue (an arbitrary fit by a computer program, see "Experimental Procedures").
The curve was used as a convenient means for normalizing data obtained as an isotope ratio (see text nolamine determination was corrected for the large error due to injection variability by use of the "H data from many animals). The utility of the normalization procedure is made clear by the considerable scatter of the uncorrected data points (Fig. 7). The half-life for time intervals equal to or greater than 15 days following injection was determined from the normalized data points. Data points obtained from animals sacrificed at times less than 15 days after injection were fit by visual inspection. Because of potential problems of slow equilibration with the radioactive precursor pool, half-lives were not calculated for this time period (e.g. it is clear that the phosphatidylethanolamine plasmalogen in myelin does not equilibrate with the precursor pool of radioactive ethanolamine until 2 weeks following injection (Fig. 7)).
Choline Turnover -Experiments analogous to those described above for ethanolamine were carried out by giving animals injections of [2-3H]glycerol and [Wlcholine, isolating microsomes and myelin, and separating phosphatidylcholine from other lipids by thin layer chromatography.
A typical autoradiograph of a sample (Fig. 8) demonstrated that injected ['4Clcholine was quite specific as a precursor for the cholinecontaining lipids (see also Ref. 24). An aliquot of each phosphatidylcholine sample was subjected to alkaline methanolysis; more than 96% of the total radioactivity was in the aqueous-soluble fraction, indicating that very little of the intracranially injected choline was metabolized to other products. The remainder of each sample was used to quantitate the 3H/14C ratio of isolated phosphatidylcholine, and this value and 17 &i of [1,3-'4Clglycerol into 17-day-old rats. Animals were killed at the indicated times following injection with both radioactive precursors, and lipids were extracted from microsomal and myelin fractions (Fig. 4). "H and 14C radioactivity in individual lipid classes obtained from the microsomal and myelin fractions was determined. The 3H radioactivity was corrected to include only that present at the C-2 position of glycerol (Figs. 1 and 21, and the 14C radioactivity was corrected to include only that present in the glycerol moiety (Fig. 4) was corrected to include only the "H in the C-2 position of glycerol. The [2-"HIglycerol label is lost from phosphatidylcholine more rapidly than [Wlcholine (Fig. 9). As a control, a reverse label experiment was conducted with  (Fig. 9) were divided into the absolute [sH]glycerol radioactivity (Fig. 3) at the same time point as described above. The log of these resultant values was plotted as a function of time after injection (Fig. 10) Radioactivity in the 2 hydrophobic residues accounted for 91 and 97% of the total 14C of microsomes and myelin, respectively.
In each case, the isotope ratio decreased as a function of time. After correction of the 'JH/'"C ratios to include the :'H only in the C-2 position of glycerol, the 14C radioactivity of each phospholipid in both subcellular fractions was calculated by dividing the "H/W ratio into the absolute l"SH1glycerol radioactivity (Fig. 31, IA  I  I  I  I  I  I  I  1 The &to points (0) were obtained by dividing the SH/'4C ratio (corrected as described in Fig. 6) into the absolute [3Hlglycerol radioactivity given for that subcellular fraction at the same time (see text and Fig.  3). The data points corresponding to times equal to or greater than 15 days after injection were fit by a linear regression program, and halflives and ranges were calculated.
Earlier time points were fit by visual inspection.
values was plotted as a function of time (Fig. 12) to include only the 3H radioactivity at the C-2 position of glycerol ( Figs. 1 and 2) and only the 14C in glycerol (Fig. 4). The curues were tit by visual inspection.
The l*C radioactivity values for phosphatidylcholine at the last time point were 1891 and 5337 dpm for microsomes and myelin, respectively. Following sacrifice at various times after injection, individual phospholipids from myelin and microsomes were isolated, and the 3H/14C ratios were plotted as a function of time (Fig. 13). There was a slight trend for i*C to be reutilized more efficiently than 3H, but this was of marginal significance over a 2-month period.
Synthesis of fatty acid by the cytoplasmic pathway would be expected to result in a loss of methyl hydrogen and equilibration of a second hydrogen with the reduced pyridine nucleotide pool. Therefore, almost two-thirds of the 3H from [methyl-3H1acetate would be lost in each round of synthesis. The observation that the 3H/14C ratio of membrane phospholipids from animals given injections of [3H]-and [14C]acetate remained almost constant (Fig. 13) indicates that fatty acid and aldehyde moieties released during catabolism were not degraded to acetate before reutilization, since these steps would result in an extensive loss of 3H. In addition, this experiment demonstrates the suitability of the [3Hlacetate as a precursor for labeling of phospholipids. were killed at various times after injection, and microsomal and myelin lipids were isolated. An aliquot was subjected to alkaline methanolysis to determine the percentage of lipid l*C in the acyl moiety (Fig. 14). Due to the relatively low levels of radioactivity and the limited material available, distribution of 14C could accurately be determined only in microsomal and myelin phosphatidylcholine and myelin phosphatidylethanolamine plasmalogen. The high proportion of label initially in the glycerol backbone is as expected from production of three carbon products by the glycolytic pathway. With time the more efficiently reutilized hydrophobic moieties account for an increasing percentage of the total radioactivity.
The remaining lipid was used for determination of the 3H/ 14C ratio, which was corrected to include only the 3H at the C-2 position of glycerol and only the 14C in the acyl moiety. The microsomal phosphatidylethanolamine and phosphatidylethanolamine plasmalogen and myelin phosphatidylethanolamine ratios were corrected assuming the same distribution of Y! radioactivity as that observed for phosphatidylcholine ( Fig.  141 for  The "H/i% ratios (corrected to include only 3H at the C-2 position of glycerol and i4C only in the acyl groups) decreased as a function of time after injection. These data were used to calculate the half-lives and range for the turnover of acyl groups (Fig. 151,  rats. Animals were killed at various times following injection, myelin and microsomal fractions were isolated, total lipids were extracted, and aliquots were subjected to thin layer chromatography for separation of individual lipid classes. Phospholipid samples were eluted and subjected to alkaline methanolysis. The myelin (---) and microsomal (-) 3H/14C ratios of material in the organic phase were determined.
The aldehyde obtained from the cleavage of phosphatidylethanolamine plasmalogen (PE Pl Aid) was collected directly from the thin layer chromatography plate (Fig. 8)  The double label protocol provides direct evidence for preferential recycling of phosphate, base, and hydrophobic moieties relative to glycerol, without the experimental error involved in the calculation of half-lives from two different precursors. Furthermore, the ratio data allow for a calculation of the absolute turnover rate of the I%-or 32P-labeled precursor by reference to the turnover of [2-3H1glycerol. The results obtained in this manner represent a level of significance that could be achieved in single isotope experiments only by the use of a much greater number of animals. The intracranial injection procedure was a major source of variability in the present study. However, a more uniform systematic route, such as intraperitoneal injection, was not feasible because [3Hlglycerol injected by this route is incorporated into brain lipid 2 orders of magnitude less efficiently than intracranially injected [SHlglycerol (11). In interpreting the results it should also be kept in mind that myelin is a relatively homogeneous membrane produced by oligodendroglial cells, whereas the microsomal fraction is heterogeneous, with respect to both the type of membrane and the cells of origin.
[2-zHlGlycerol As Label for Brain Phospholipids Following the injection of [2-3H]glycerol, most of the radioactivity in the glycerol phospholipids was initially present in the C-2 position (Figs. 1 and 2). However, the proportion of 3H  Fig. 14. The data plotted were obtained by dividing the 3H/'"C ratio (corrected to include only 3H in the C-2 position of glycerol and to include only 14C in the acyl position) into the absolute [3Hlglycerol radioactivity (see text). The distribution of 14C between acyl groups and the glycerophosphoryl base of microsomal phosphatidylethanolamine (PE) and phosphatidylethanolamine plasmalogen (PE PI) was assumed to be similar to that of microsomal phosphatidylcholine (PC), and the distribution of 14C between the glycerophosphoryl base of myelin phosphatidylethanolamine was assumed to be similar to the results obtained from myelin phosphatidylcholine (Fig. 14). For diacyl lipids, a half-life was calculated from the data points corresponding to times equal to or greater than 15 days after injection.
For myelin phosphatidylethanolamine plasmalogen, a half-life could not be calculated. Data corresponding to times before 15 days following injection were fit by visual inspection.
randomized to the acyl moieties or to the C-l carbon of glycerol increased as a function of time after injection. The extent to which this took place was unexpected, emphasizing the need for chemical localization of the label in long term experiments. The most probable explanation for the data in Fig. 1 involves reutilization of the 3H from C-2, possibly by a pathway where label lost from glycerol 3-phosphate to ["HINADH is conserved for synthetic purposes, via ["HINADPH, which might be formed by a transhydrogenase.
Alternatively, contamination of the [2-3H]glycerol by a fraction of a per cent of a substrate that labels fatty acids effciently might account for the small amount of label initially present in the hydrophobic moieties. Due to the relative metabolic stability of the acyl and alkenyl side chains (see below), they would rapidly account for an increasing percentage of the total lipid label.
With respect to Fig. 2, the results are best explained by an argument presented by Agranoff and Hajra (18), that label lost from the C-2 position of glycerol 3-phosphate is partially recovered in the C-l position of glycerol 3-phosphate via ["HINADH (and glyceraldehyde 3-phosphate) by reversal of the glycolysis pathway.
Another possible drawback to the use of 12-"HIglycerol for studies of phospholipid synthesis concerns the isotope effect that has been shown with the mitochondrial glycerophosphate dehydrogenase (EC 1.1.99.5); [2-"HIglycerol is not a good substrate and is enriched in the tissue relative to [14Clglycerol (26). That this enrichment also occurs in brain is suggested by the high :'H/'% ratio in phosphatidylcholine and phosphatidylethanolamine relative to the injected isotope ratio (Fig.  5) (26).

Metabolism of Base Moieties
The isotope ratio data for ethanolamine ( Fig. 6) and choline ( Fig. 9)  Although other reports have been published in which labeled base and phosphate precursors have been used to follow the turnover of microsomal and myelin lipids (7, 27, 331, these results are not readily comparable with ours due to experimental differences (e.g. differences in techniques or time period of the study). Another critical variable is animal age, since Horrocks (36) has shown, at least for the short turnover phase, that the half-life of ethanolamine lipids in mouse brain subcellular fractions increases with increasing age at time of isotope administration.

Turnover of Acyl and Alkenyl Moieties
-The results relating to the reutilization of the hydrophobic moieties are not exactly analogous to the reutilization data reported for ethanolamine and choline. In the case of the bases, ethanolamine and choline are presumably reutilized intact without any metabolic alteration.
The acyl moieties, however, can undergo chain elongation and desaturation before reincorporation into phospholipids.
This has been demonstrated and used as evidence for recycling of acyl moieties of phospholipids (8, 37). The reutilization of essential fatty acids for synthesis of brain phospholipids has also been investigated (38). Our ["HIacetate/ ['Clacetate experiment (Fig. 13) supplements this work by demonstrating that reutilization of the acyl precursor does not involve recycling through the acetate pool. In addition, this experiment demonstrates the suitability of the [sH]acetate as a precursor for labeling of phospholipids.
The decrease in the "H/Y ratio in each phospholipid as a function of time (not illustrated, since it can be calculated from data in Fig. 3 and in Fig. 12, for acetate, and Fig. 15, for glucose) indicates that radioactivity from acyl groups is reutilized more efficiently than is radioactivity from phosphate or base moieties (Table I). The same presumably holds true of the alkenyl groups of phosphatidylethanolamine plasmalogen since, as indicated in the results, their reutilization appears to be the same as that of the acyl moiety. Although the turnover time for the glycerol moiety of phosphatidylcholine and phosphatidylethanolamine in myelin is the same, radioactivity in acyl groups is much more stable in phosphatidylethanolamine than in phosphatidylcholine ( Table I). The same observation holds with respect to the microsomal lipids. This indicates that labeled acyl groups were even more efficiently reutilized for the resynthesis of phosphatidylethanolamine than for the resynthesis of phosphatidylcholine.
Our reported half-lives for the hydrophobic moiety of phosphatidylcholine in myelin with acetate or glucose as a precursor, 54 or 56 days, are almost the same as the 2-month half-life reported by Smith and Eng (39), who used acetate as a precursor, and are similar to the 42-day value later reported by Smith (1) using glucose as a precursor. In both of these studies, it was reported that ethanolamine phospholipids (a mixture of the diacyl and the plasmalogen compounds) were more stable than phosphatidylcholine.
We are now in a position to confirm and interpret these earlier data of Smith and Eng in greater detail and to emphasize that phosphatidylcholine and phosphatidylethanolamine in myelin really have the same turnover rate of 25 days with regard to the [2-3H1glycerol, but that the acyl and alkenyl moieties are reutilized more efficiently for the synthesis of phosphatidylethanolamine and phosphatidylethanolamine plasmalogen than for phosphatidylcholine synthesis. This difference between the metabolism of ethanolamine-and choline-containing lipids (demonstrated by the use of either ['Qacetate or ['Qglucose as precursor) is reminiscent of metabolic differences between phosphatidylethanolamine and phosphatidylcholine with regards to short term isotope experiments studying the assembly of myelin (13). As discussed in that reference, the metabolic differences between ethanolamine-and choline-containing lipids may be related to preferential localization of ethanolamine lipids in the inner half of the bilayer and choline lipids on the outer surface, in analogy to the architecture of the red cell membrane (40).

Involvement
of Dihydroxyacetone Phosphate in the Synthesis of Phosphatidylethanolamine Plasmalogen -Following simultaneous injection of [2-"HIglycerol and [1,3-'%]glycerol, myelin phosphatidylethanolamine plasmalogen had a lo-fold lower initial "H/l% ratio than did phosphatidylethanolamine or phosphatidylcholine in the same subcellular fraction (Fig.  5). This observation is consistent with the hypothesis that myelin phosphatidylethanolamine plasmalogen is primarily synthesized by a route in which dihydroxyacetone phosphate is an obligatory intermediate (41,42) with consequent loss of the "H at the C-2 position. The significance of this pathway in brain has been demonstrated (43,44). This result is in contrast to the preferential involvement of the glycerol 3-phosphate pathway (45, 46) for synthesis of phosphatidylcholine and phosphatidylethanolamine.
The preferential loss of the C-2 tritium for synthesis of phosphatidylethanolamine plasmalogen of myelin relative to synthesis of phosphatidylethanolamine of myelin can also be ascertained from examination of the "H/l% ratios for the various substrate pairs (Fig. 6 for ["Clethanolamine; see "Results" for data obtained with [%lacetate or ['4Clglucose). The "H/J*P ratio data for the individual myelin lipids supports the hypothesis of involvement of the dihydroxyacetone phosphate pathway for synthesis of phosphatidylethanolamine plasmalogen of myelin.
The results with regard to phosphatidylethanolamine plasmalogen of microsomes were not as clear-cut as for the myelin data. There was a preferential loss of the C-2 tritium during synthesis of phosphatidylethanolamine of myelin relative to that observed for synthesis of phosphatidylethanolamine plasmalogen in microsomes. However, the magnitude of this effect was not always as prominent in microsomes as in myelin (Figs. 5 and 6). This might be due to the more rapid turnover of the microsomal lipids, resulting in a greater loss of C-2 tritium from glycerol before the first time point was taken. Another possible complication is the heterogeneous cellular origin of the microsomal fraction.