Long-chain Acyl-CoA Synthetase 6 Preferentially Promotes DHA Metabolism*

Previously we demonstrated that supplementation with the polyunsaturated fatty acids (PUFA) arachidonic acid (AA) or docosahexaenoic acid (DHA) increased neurite outgrowth of PC12 cells during differentiation, and that overexpression of rat acyl-CoA synthetase long-chain family member 6 (Acsl6, formerly ACS2) further increased PUFA-enhanced neurite outgrowth. However, whether Acsl6 overexpression enhanced the amount of PUFA accumulated in the cells or altered the partitioning of any fatty acids into phospholipids (PLs) or triacylglycerides (TAGs) was unknown. Here we show that Acsl6 overexpression specifically promotes DHA internalization, activation to DHA-CoA, and accumulation in differentiating PC12 cells. In contrast, oleic acid (OA) and AA internalization and activation to OA-CoA and AA-CoA were increased only marginally by Acsl6 overexpression. Additionally, the level of total cellular PLs was increased in Acsl6 overexpressing cells when the medium was supplemented with AA and DHA, but not with OA. Acsl6 overexpression increased the incorporation of [14C]-labeled OA, AA, or DHA into PLs and TAGs. These results do not support a role for Acsl6 in the specific targeting of fatty acids into PLs or TAGs. Rather, our data support the hypothesis that Acsl6 functions primarily in DHA metabolism, and that its overexpression increases DHA and AA internalization primarily during the first 24 h of neuronal differentiation to stimulate PL synthesis and enhance neurite outgrowth.

20:4 n-6) and docosahexaenoic acid (DHA, 22:6 n-3) are the most abundant PUFAs, but cannot be synthesized de novo by mammals because the desaturase enzymes required to introduce double bonds at the n-3 and n-6 positions of FAs are not present (1). Therefore, AA and DHA, or their precursors, must be ingested from dietary sources and transported to the brain. During late gestation and early postnatal development neurodevelopment is especially rapid and AA and DHA are critical to ensure proper brain and retina growth. AA and DHA are normally transferred by the placenta to the fetus during the last trimester of pregnancy and in breast milk to infants (2). However, if the postnatal diet is not properly supplemented with these essential FAs, FA deficiency may manifest itself in the form of neurological and visual deficits. Premature infants are at even greater risk because they are deprived of the essential FA loading that normally occurs during the last trimester of gestational development (3).
Despite their obvious importance, the functions PUFAs perform during this critical period of development have yet to be defined. Several reports establish that supplementation of neuronal cells with AA or DHA can affect neurite outgrowth indicating that PUFAs may be critical for proper neuronal differentiation (4 -11). However, little is known about the mechanism or proteins that are responsible for PUFA internalization and their subsequent metabolism.
While it remains controversial how FA cross the plasma membrane, considerable evidence implicate diffusion through the PL bilayer as well as several integral and associated plasma membrane proteins as facilitators of free FA transport across the lipid bilayer (12)(13)(14). After internalization, FAs are activated primarily by acyl-CoA synthetase long-chain family members (Acsls, formerly ACSs), which utilize ATP to catalyze the formation of FA-CoA by esterifying the FA to coenzyme A (CoA). Once activated, the FA can function as a signaling molecule, be incorporated into phospholipids (PLs) or triacylglycerides (TAGs), or undergo ␤-oxidization in mitochondria for energy generation.
Because Acsl proteins catalyze an essential step in the metabolism of FA, it seemed likely that multiple isoforms exist with different preferences for FA species. This was first confirmed by the biochemical isolation of an arachidonoyl-specific acyl-CoA synthetase in human platelets that was different from the original acyl-CoA synthetase enzyme that had broad FA specificity (15). Subsequently, five mammalian Acsl proteins in all have been identified and biochemical studies using bacterially expressed Acsl proteins have shown that each protein activates different FAs with unique efficiencies. Acsl1 and Acsl5 activate most unsaturated FAs similarly (16,17), whereas Acsl3, Acsl4, and Acsl6 (formerly ACS2) preferentially activate PUFAs (17)(18)(19). In fact, Acsl6 activates AA and DHA 2-and 8-fold more efficiently, respectively, than does Acsl1. Compared with OA, the initial rate of AA and DHA uptake was increased more in PC12 cells overexpressing Acsl6 than in cells overexpressing Acsl1, providing in vivo evidence that Acsl6 preferentially metabolizes PUFAs (7).
A series of biochemical and cell fractionation experiments established that different Acsl proteins have unique intracellular localizations, suggesting that each protein may partition FAs toward different metabolic fates. Acsl1 protein has been proposed to function in FA internalization and TAG synthesis because it localizes to the plasma membrane in 3T3-L1 cells (20,21) and to endoplasmic reticulum and mitochondrial associated membranes in primary rat hepatocytes (22). Consistent with these proposals, FA internalization in 3T3-L1 and PC12 cells (7,23) and TAG levels in PC12 cells (7,23) are increased in cells that overexpress Acsl1 protein. Acsl4 protein localizes to mitochondrial associated membranes and peroxisomes in primary rat hepatocytes suggesting it functions in TAG synthesis and FA oxidation (22). Acsl5 is the only Acsl localized to mitochondria in primary rat liver hepatocytes where it likely activates FAs for ␤-oxidation (22). The intracellular localizations of Acsl6 and Acsl3 have yet to be determined. However, Acsl6 overexpression in PC12 cells increases FA internalization, neurite outgrowth, and TAG accumulation, which are consistent with localization to the plasma membrane, endoplasmic reticulum, and mitochondrial associated membranes. Taken together, these data suggest that differential expression of Acsl proteins may contribute to PUFA enrichment in the PLs of neurons.
To test the hypothesis that Acsl6 enhances neurite outgrowth by preferential PUFA accumulation and partitioning into PLs, we overexpressed Acsl6 in differentiating PC12 cells and examined the metabolic fate of internalized FAs. We show that Acsl6 overexpression selectively enhances DHA internalization, activation to DHA-CoA and accumulation, but does not increase FA targeting to PLs relative to TAGs.

Generation of Stable Acsl6
Overexpressing Cells-To generate stable populations of control and Acsl6 overexpressing PC12 cells, retroviral supernatants containing control (pMSCVpuroIRES2-EGFP) or rat Acsl6 (pMSCVpuro-ratACS2-IRES2-EGFP) retrovirus were used to infect PC12 cells as described previously (7). Two days after infection, transformed cells were selected by incubation with 5 g/ml puromycin for 48 h. The level of GFP fluorescence was determined for 10,000 cells using a BD Biosciences FACScan to ensure that greater than 95% of the cells were transformed. The cells were then expanded for 2 weeks, the transformation rate confirmed again by FACS analysis, and aliquots of the cells were frozen until needed. The cells used for analysis were in culture for 2-8 weeks after infection.
Western Analysis of Acsl6 Protein-A peptide corresponding to the first 19 amino acids (MQTQEILRILRLPELSDLG) at the amino terminus of rat Acsl6 protein was synthesized and coupled to keyhole limpet hemocyanin. Antibodies to the peptide were raised in rabbits. Undifferentiated control and Acsl6 overexpressing cells were harvested, washed with 1ϫ phosphate-buffered saline, and resuspended in 1ϫ sample buffer (2% SDS, 150 mM Tris, pH 6.8, 10% glycerol, 700 mM 2-mercaptoethanol) at a concentration of 10 6 cells per ml. Cells were lysed and DNA sheared by titration through a Hamilton syringe and heated at 80°C for 10 min. Equal amounts of protein extract from control and Acsl6 overexpressing cells were separated on a 7% NuPAGE Tris acetate gel (Invitrogen), transferred to nitrocellulose, and probed with anti-Acsl6 antibody. The membrane was then probed with horseradish peroxidase-conjugated sheep anti-mouse secondary antibody and Acsl6 protein levels were visualized using an ECL Western blotting detection kit (Amersham Biosciences) followed by exposure to X-Omat Blue film (Kodak).
Quantitative PCR-Aliquots of undifferentiated control and Acsl6 overexpressing cells were collected at the same time that these cells were plated for differentiation experiments. Total RNA was extracted and the amount of Acsl1-6 mRNAs determined as described previously (7).
Acyl-CoA Synthetase Activity Assay-Undifferentiated control and Acsl6 overexpressing cells were treated for 5 min at 37°C with 0.25% trypsin and then washed three times with 1ϫ phosphate-buffered saline. Cells were pelleted, resuspended in STE (10 mM Tris base, 10 mM NaCl, 1 mM EDTA) at 10 7 cells/ml, and then sonicated for 2 min in a cup sonicator. The extract was then centrifuged for 10 min at 1500 ϫ g to remove insoluble material. The lysate was then diluted 1:4 (v/v) with STE. One hundred microliters of the extract was added to a reaction mixture containing 250 l of 2ϫ reaction mixture (0. Differentiation of Cells for Lipid Analysis-Control and Acsl6 overexpressing PC12 cells were maintained in 10% fetal bovine serum in RPMI 1640 medium on collagen-coated plates (VWR Biocoat) at 37°C in an incubator with 5% CO 2 . Cells were split bi-weekly at 5 ϫ 10 6 cells/100-mm plate. For each analysis, cells were treated for 8 min at 37°C with 0.25% trypsin, triturated through a 19-guage needle attached to a syringe, counted, and plated in growth media at the specified concentrations. Normal differentiation medium contains 4 mol % OA, which rises to 30.7% when supplemented with 5 M OA. Normal differentiation medium contains no AA or DHA, and the mole % rises to 19.5 and 22.0%, respectively, when supplemented with 5 M AA or DHA (Table I)  Detached and dead cells were removed by centrifugation at full speed in a microcentrifuge for 5 min and the amount of radioactivity in the media determined by adding the media to 4 ml of EcoScint H (National Diagnostics) followed by scintillation counting.

Analysis of [ 14 C]FA Incorporation into Complex
Lipids-Cell pellets were resuspended in 0.8 ml of water and vortexed vigorously to disrupt the cell pellet. A 15-l aliquot of the cell suspension was removed and added to 45 l of counting solution (24), and the concentration of cells in the suspension determined using a hemacytometer. Two milliliters of methanol and 1 ml of chloroform with 50 g/ml butylhydroxytoluene were added to the remainder of the cell suspension, the sample was mixed vigorously and then sonicated for 30 min at room temperature in a bath sonicator. Non-solubilized material was removed by centrifugation at 3,000 ϫ g for 5 min in a table top centrifuge. The supernatant was transferred to a new tube, 1 ml of chloroform (with 50 g/ml butylhydroxytoluene) and 1 ml of water were added, the sample was vigorously mixed for 1 min, and the phases were clarified by centrifugation at 3,000 ϫ g for 5 min. The chloroform phase was collected, evaporated under a stream of nitrogen, and resuspended in chloroform at a concentration equivalent to 10 7 cells/ml. Ten microliters of the 14 C-labeled samples were added to 4 ml of EcoScint H and the amount of radioactivity was determined by scintillation counting. Twenty-five microliters of each 14 C-labeled sample was spotted on a silica G plate (Analtech) for separation by thin-layer chromatography (TLC). Complex lipid species were resolved by elution with petroleum ether/diethyl ether/acetic acid (90:10:1). PL species were resolved by elution with chloroform/triethylamine/ethanol/water (30:30:34:8). The plates were allowed to dry overnight and then exposed for 6 h to a phosphorimaging screen (FujiFilm). The plates were read with a FujiFilm BAS-2500 scanner and the amount of [ 14 C]FA incorporated into PL and TAG species were determined using Multiguage software (FujiFilm).
Determination of PL Levels-Twenty-five microliters of lipid extract from 3-day differentiated cells that were supplemented with 5 M unlabeled FA were evaporated to dryness in glass tubes under a stream of nitrogen. Fifty microliters of perchloric acid was added to each tube, then the tube was capped and placed in a 180°C heating block overnight to liberate phosphate from the PL. Ten microliters of each sample was added to wells of a 96-well plate. One-hundred microliters of assay solution (0.29% brilliant green, 1 M ammonium molybdate, 1.2 N hydrochloric acid, and 0.05% Tween 20) was added for 15 min at room temperature. The absorbance at 660 nm was measured using a Molecular Dynamics Versamax plate reader to determine the amount of free phosphate that is a measure of PL abundance. PL levels were determined by comparing the unknown samples to a standard of known amounts of PC (Sigma) that had been treated identically.
Gas Chromatography/Mass Spectrometry (GC/MS) of FAs-Ten microliters of 1 mg/ml 17:0 FA (internal standard) and 50 l of lipid extract from 3-day differentiated cells that were supplemented with 5 M unlabeled FA were evaporated to dryness in a glass tube under a stream of nitrogen. Once evaporated, 0.25 ml of toluene and 0.5 ml of 1% sulfuric acid in methanol were added, the tube were capped under nitrogen, and incubated at 50°C overnight in a heat block. The next day, 1.25 ml of 5% NaCl was added and the suspension was mixed vigorously. FA methyl esters were extracted twice with 1 ml of hexane followed by centrifugation at 3,000 ϫ g for 5 min in a table top centrifuge. The hexane phase was washed with 1 ml of 2% potassium bicarbonate and collected again before evaporation under a stream of nitrogen. The sample was resuspended in 25 l of methyl acetate with 50 g/ml butylhydroxytoluene. The fatty acid methyl esters were analyzed on an Agilent 6890 series gas chromatograph equipped with a 5873 mass selective detector. Identity of the fatty acids was made by comparison with the National Institute of Standards and Technology (NIST) data base and confirmed by comparison with standards purchased from commercial sources (Sigma and Avanti). Quantification was made by comparison with the added 17:0 internal standard.

RESULTS
Previously we reported that supplementation of differentiating PC12 cells with PUFAs enhances neurite outgrowth and that overexpression of Acsl6 further increases PUFA-mediated neurite outgrowth (7). For the metabolic tests to determine whether Acsl6 overexpression selectively enhanced PUFA accumulation or targeted PUFAs into PL relative to TAG, large numbers of control and Acsl6 overexpressing cells were required. Therefore, Eco1 PC12 cells (PC12 cells overexpressing the mouse ecotropic retrovirus receptor) were infected with control or Acsl6 overexpressing retroviruses that also encoded GFP and a puromycin resistance gene. After 48 h, the cells were selected for 2 days with 5 g/ml puromycin to obtain highly enriched populations of transformed cells. As identified by FACS analysis of GFP expression, over 95% of the cells contained an integrated retrovirus and synthesized additional Acsl6 protein (Fig. 1, A-C). To obtain sufficient numbers of cells to perform kinetic and biochemical analysis of the metabolic fate of exogenous OA, AA, and DHA during PC12 cell differentiation, these populations of transformed cells were expanded for 2-8 weeks before differentiation.
Quantitative PCR was used to measure the levels of Acsl mRNAs and determine whether long-term overexpression of Acsl6 affected the expression of other Acsl mRNAs (Fig. 1D). Compared with the level of Acsl6 mRNA in control cells, the level of Acsl6 mRNA in Acsl6 overexpressing cells was increased ϳ100-fold. The levels of the other Acsl mRNAs were not significantly altered in Acsl6 overexpressing cells, demonstrating that the other Acsl mRNA are not down-regulated to compensate for increased Acsl6 mRNA expression.
The level of Acsl6 protein in control and Acsl6 overexpressing cells was measured to determine whether the 100-fold increase in Acsl6 mRNA translated to a similar increase in Acsl6 protein. In Acsl6 overexpressing cells, the level of Acsl6 protein increased by ϳ15-fold compared with control cells (Fig. 1C), confirming that the increase in the level of Acsl6 mRNA led to a highly significant increase in Acsl6 protein accumulation.
To determine whether Acsl6 overexpression selectively increased total OA-CoA, AA-CoA, or DHA-CoA synthetase activity, the conversion of each [ 14 C]FA to [ 14 C]FA-CoA was measured using total cellular extracts from undifferentiated control and Acsl6 overexpressing cells (Fig. 2). In control cells, the total OA-CoA synthetase activity was only 1/7 th and 1/3 rd the level of total AA-CoA and total DHA-CoA synthetase activities, respectively (Fig. 2). Acsl6 overexpression increased DHA-CoA synthetase activity 5-fold compared with control cells. In contrast, OA-CoA and AA-CoA synthetase activities were increased only 1.7-and 0.4-fold in Acsl6 overexpressing cells compared with control cells. These findings provide direct evidence that Acsl6 preferentially converts DHA to DHA-CoA.
Acsl6 Overexpression Selectively Increases DHA Metabolism-To determine whether Acsl6 overexpression specifically increased AA or DHA metabolism over the course of differentiation, control and Acsl6 overexpressing cells were supplemented with 5 M [ 14 (Fig. 3, A and C, respectively, compare solid to dotted curves). After 72 h of differentiation, each [ 14 C]FA was depleted similarly from the media by control and Acsl6 overexpressing cells, except for [ 14 C]DHA, which was depleted 26% more from the media by Acsl6 overexpressing cells than by control cells.
To confirm that the selective [ 14 C]DHA removal from the media by Acsl6 overexpressing cells presented in  1 and 4 to 3). Similarly, after 72 h of incorporation the level of [ 14 C]DHA was increased in Acsl6 overexpressing cells relative to control cells by a modest 28% (Fig. 4B, lanes 5 and 6), but the levels of [ 14 C]OA and [ 14 C]AA in Acsl6 overexpressing cells (lanes 1-4) were similar to those of control cells. Increased DHA extraction from the media and increased accumulation of [ 14 C]DHA in Acsl6 overexpressing cells are consistent with the notion that Acsl6 preferentially functions in DHA metabolism for lipid synthesis, and not ␤-oxidation.
Taken together, these results are consistent with previously reported short-term uptake studies (7) and biochemical studies overexpressing cells were harvested, total RNA was extracted and analyzed by quantitative PCR to determine the level of each Acsl mRNA. All mRNA levels are relative to the amount of Acsl1 mRNA in control cells, which was defined as 1. As previously described (7), to correct for differences in amplification rates among the Acsl genes because of primer sequence variation, the amplification rate for each primer pair was determined for genomic DNA where each gene is present in exactly two copies. Cyclophilin B mRNA levels were used to normalize expression between samples. (17) that suggest Acsl6 preferentially promotes AA and DHA metabolism. The highly significant increase in [ 14 C]DHA removal from the media by Acsl6 overexpressing cells during the entire 3-day period of differentiation strongly support the hypothesis that DHA-enhanced neurite outgrowth is a consequence of increased DHA availability for PL synthesis. In contrast, the modest increase in [ 14 C]AA extraction from the media only during the first 24 -48 h of differentiation suggest that AA-enhanced neurite length in Acsl6 overexpressing cells must be a consequence of other mechanisms that could include enhanced AA incorporation into PLs, preferential delivery of AA containing PLs to the plasma membrane, or increased AA signaling that leads to increased endogenous FA synthesis.
Acsl6 Overexpression Does Not Preferentially Increase FA Targeting to PLs-The existence of five Acsl proteins with unique subcellular localization patterns is consistent with the hypothesis that different Acsl proteins target FAs to different complex lipid pools. To test whether the PUFA-enhanced neurite outgrowth that occurs with Acsl6 overexpression is a consequence of Acsl6 preferentially targeting DHA or AA to PLs, the lipid extracts of control and Acsl6 overexpressing cells  (Fig. 5B, lane 6 relative to lane 5) was increased by ϳ250% compared with control cells. In other words, of the additional [ 14 C]DHA incorporated by Acsl6 overexpressing cells relative to control cell, only 66% was incorporated into PLs and 33% was incorporated into TAGs. This was in contrast to control cells, where 90% of the [ 14 C]DHA was incorporated into PLs and 10% of the [ 14 C]DHA was incorporated into TAGs. These results are inconsistent with Acsl6 overexpression increasing the targeting of FAs for PL synthesis rather than TAG synthesis.

Distribution of FAs Among PL Classes Is Not Altered by Acsl6
Overexpression-Earlier studies showed that in PC12 cells AA accumulates primarily in phosphatidylethanolamine (PE) and phosphatidylinositol (PI), DHA accumulates primarily in PE, and OA accumulates primarily in phosphatidylcholine (PC) (25,26). We hypothesized that DHA-enhanced neurite outgrowth in Acsl6 overexpressing cells was a consequence of Acsl6 preferentially targeting of AA and DHA to their preferred PL species. To test this hypothesis, the relative distributions of PI, and PE species in Acsl6 overexpressing cells was not significantly different from control cells (Fig. 6, A and B). These findings strongly suggest that Acsl6 does not target FAs to specific PL species.
Especially after 72 h of incubation, there were several significant differences in the distributions of [ 14  C]DHA incorporated into PLs by Acsl6 overexpressing cells were not substantially increased after a 3-day differentiation, it was unclear whether PUFA-enhanced neurite outgrowth was a consequence of increased levels of total cellular PLs. Therefore, the levels of total cellular PL of control and Acsl6 overexpressing cells differentiated for 3 days in media supplemented with 5 M OA, AA, or DHA were measured (Fig.  7). Consistent with increased PL synthesis, the level of total cellular PL in Acsl6 overexpressing cells supplemented with AA and DHA were significantly increased by 11  Importantly, the levels of total PLs in control and Acsl6 overexpressing cells supplemented with AA (lanes 3 and 4) or DHA (lanes 5 and 6) were increased by 13-38% compared with equivalent cells supplemented with OA (lanes 1 and 2). These small but significant changes in total PL levels that occur with PUFA supplementation and Acsl6 overexpression support the hypothesis that increased levels of intracellular AA and DHA stimulate PL synthesis, and that the resulting increase in total cellular PLs stimulates neurite outgrowth.
PUFA Supplementation Increases Levels of Both Saturated and Polyunsaturated FAs-We have shown that Acsl6 overexpression primarily increased DHA and to a lesser extent AA internalization (Figs. 3 and 4A), but did not preferentially target [ 14 C]FAs into PLs (Figs. 5 and 6). To determine whether Acsl6 overexpression increased the total amount of cellular AA and DHA and other FA species, control and Acsl6 overexpress- ing cells were differentiated for 3 days with 0.25% horse serum supplemented with 5 M OA, AA, or DHA. After extraction, the total lipid extracts were analyzed by GC/MS to measure the levels of several saturated, monounsaturated, and PUFA species. Compared with control cells supplemented with DHA, the level of DHA in DHA-supplemented Acsl6 overexpressing cells was increased by a highly significant 23% (Fig. 8C, 22:6(n-3)). This finding supports the hypothesis that the majority of internalized DHA was incorporated into PLs and TAGs and was not metabolized to other FA species. Additionally, compared with control cells, the level of stearic acid (18:0) in Acsl6 overexpressing cells supplemented with DHA was increased by a significant 38%, and the PUFA eicosapentanoic acid (20:5(n-3)) was increased by a highly significant 77% (Fig. 8C, ii).
Supplementation of Acsl6 overexpressing cells with AA selectively and significantly increased the levels of palmitic (16:0) and stearic acid (18:0) compared with control cells (Fig. 8B). Consistent with no change in [ 14 C]AA accumulation after 72 h of incubation (Fig. 4B, lanes 4 -6), the level of AA was unchanged in Acsl6 overexpressing cells supplemented with AA compared with control cells (Fig. 8B, 20:4(n-6)). However, it is interesting that the level of DHA was significantly increased by 63% (Fig. 8B, i) in Acsl6 overexpressing cells supplemented with AA compared with control cells. This finding is consistent with the hypothesis that Acsl6 preferentially promotes DHA metabolism. Supplementation of Acsl6 overexpressing cells with OA resulted in no significant change in the level of any FA compared with control cells (Fig. 8A), consistent with no change in total PL levels (Fig. 7).
Independent of Acsl6 overexpression, supplementation with AA instead of OA increased the amount of AA by 3.5-fold (compare Fig. 8, A to B, 20:4(n-6)) and supplementation with DHA in place of OA increased the amount of DHA by 9 -13-fold (compare Fig. 8, A-C, 22:6(n-3)). Additionally, AA or DHA supplementation in place of OA primarily increased the level of stearic acid (18:0) by 40 -70% (compare Fig. 8, B and C to A, 18:0) and not surprisingly reduced the level of oleic acid by 30 -40% (compare Fig. 8, B and C to A, 18:1). This is consistent with published reports that PUFAs reduce the activity of stearoyl-CoA desaturase, which converts stearic acid (18:0) to oleic acid (18:1) (27). Taken together, these findings are consistent with the hypothesis that increased levels of PUFAs lead to increases in PL levels and neurite length. It seems likely that PUFA-enhanced neurite outgrowth in Acsl6 overexpressing cells is simply an exacerbation of this process because of increased AA and DHA internalization during the first 24 h of differentiation.  7. Total levels of cellular phospholipids are increased with AA and DHA supplementation and further increased by Acsl6 overexpression. Control or Acsl6 overexpressing cells were differentiated for 3 days in 0.25% horse serum/ RPMI 1640 supplemented with 5 M OA, AA, or DHA. Lipids were extracted and phospholipid levels were determined for 5 ϫ 10 4 cells using a molybdate/malachite green assay to determine phosphate levels as described under "Materials and Methods." PL levels were determined by comparing experimental samples to a standard series of known amounts of PL. The graph represents the average with S.E. from five independent experiments. Note that these samples were differentiated in parallel with the cells that were supplemented with [ 14 C]FAs in Figs. 4-6. *, p value Ͻ 0.05 and considered significantly different from control cells by the paired Student's t test. **, p value Ͻ 0.05 and considered significantly different from OA-supplemented cells by the paired Student's t test.

DISCUSSION
This article makes several key points concerning the relationship of PUFA internalization and metabolism to PL synthesis and neurite outgrowth. Upon differentiation of PC12 cells, and independent of Acsl6 overexpression, internalized PUFAs were preferentially metabolized into PLs, relative to TAG, and their distribution among the major PL species differed significantly from internalized OA. After 3 days of differentiation, the respective levels of AA or DHA, as well as the level of total cellular PLs, were significantly increased in cells that were supplemented with AA or DHA relative to cells that were supplemented with OA. Additionally, compared with control cells, the levels of PLs, SFAs, and PUFAs were increased in Acsl6 overexpressing cells that were supplemented with AA or DHA. The most striking finding was that, relative to OA and AA metabolism, DHA internalization, activation to DHA-CoA, and incorporation into lipids was significantly increased in Acsl6 overexpressing cells compared with control cells. Finally, Acsl6 overexpression did not increase the partitioning of internalized FAs exclusively into PLs or TAGs nor did it alter the distribution of FAs among the major PL species. Thus Acsl6 preferentially promotes DHA uptake and metabolism, presumably by preferentially activating DHA to DHA-CoA. AA uptake was only slightly enhanced by Acsl6 overexpression, yet supplementation with AA also increased total PL accumulation, presumably by stimulating FA and/or PL synthesis.
PUFA Metabolism, PL Synthesis, and Neurite Outgrowth-Several reports established that supplementation of neuronal cells with AA or DHA can enhance neurite outgrowth (4 -11), but the mechanisms by which these FAs exert their modifying effects remain unclear. Previously, we showed that supplementation with PUFAs, not OA, increased neurite outgrowth. Overexpression of Acsl6, but not overexpression of Acsl1, stimulated neurite outgrowth further, but again only when the medium was supplemented with AA or DHA, not OA. Consistent with the hypothesis that selective PUFA internalization contributes to PUFA-enhanced neurite outgrowth, upon PC12 differentiation, the initial rate of PUFA internalization was increased more than the initial rate of OA internalization (7). Additionally, compared with overexpression of Acsl1, overexpression of Acsl6 caused the initial rate of PUFA internalization to be increased relatively more than that of OA (7).
Based on these findings, it seemed likely that the PUFAenhanced neurite outgrowth might simply be a consequence of PUFAs being internalized more efficiently than OA, with Acsl6 overexpression further increasing PUFA internalization. Higher levels of PUFAs would provide more FA substrate for PL synthesis that would be inserted into the plasma membrane to increase neurite length. Consistent with this hypothesis, we found that after 72 h of incubation, the levels of total cellular PLs were significantly increased in cells supplemented with AA and DHA, compared with cells supplemented with OA. Acsl6 overexpression increased the level of PLs further when the medium was supplemented with PUFAs, not OA. However, after 72 h of incubation, nearly equal amounts of OA, AA, and DHA were taken up from the medium and incorporated into lipids of control cells, except for Acsl6 overexpressing cells where DHA internalization and incorporation were increased further. This finding suggests that PUFA-enhanced neurite outgrowth is not simply because of an increase in AA and DHA accumulation, relative to OA accumulation.
Rather, these results suggest several possible alternative mechanisms as to how PUFAs stimulate neurite outgrowth. These include preferential incorporation of PUFAs into PLs; PUFAs functioning as signaling molecules to stimulate an increase in the level of other FAs (i.e. saturated FAs) by either increasing their uptake into the cells or their synthesis within the cell; or preferential targeting of PLs containing PUFAs to the plasma membrane relative to PLs not containing PUFAs.
Consistent with preferential partitioning of PUFAs into PLs, we found that compared with [  C]palmitic acid, were incorporated into brain PLs even though equal amounts of label were found in the lipid fraction (28,29). However, our finding that after 72 h, the levels of [ 14 C]PLs were similar regardless of the [ 14 C]FA used suggest that preferential internalization and incorporation of PUFAs into PLs does not provide sufficient amounts of FAs to increase the level of PLs over the entire differentiation period.
DAG is the common precursor for de novo PL and TAG synthesis and its level increases significantly upon PC12 differentiation (30). During the synthesis of PLs, saturated FAs (i.e. palmitic acid (16:0) or stearic acid (18:0)) are typically acylated into the sn-1 position of glycerol 3-phosphate and PUFAs (i.e. 20:4 or 22:6) are acylated into the sn-2 position to generate phosphatidic acid, which is the precursor for PI. However, most of the phosphatidic acid is dephosphorylated to generate DAG, which is further metabolized to PC, PE, or TAG. There is some evidence that DAG molecules that contain PUFAs in the sn-2 position are preferentially converted to PLs. For example, DAG species that contain DHA in the sn-2 position are the preferred substrate of ethanolaminephosphotransferase, which converts DAG to PE (31), and DAG molecules that contain AA in the sn-2 position are phosphorylated by DAG kinase ⑀ (32, 33). Consistent with this hypothesis, the distributions of OA, AA, and DHA among PC, PS, PI, and PE are significantly different. The bulk of DHA is incorporated in PE and to a lesser extent in PS. Likewise, the majority of AA is incorporated into PI and PE, whereas the bulk of OA is incorporated into PC (26,34). Consistent with this result, we found that after 72 h of incubation, a higher percentage of [ 14 C]DHA and [ 14 C]AA were incorporated into PS, PI, and PE, whereas the bulk of [ 14 C]OA was incorporated into PC.
Not surprisingly, when we examined the FA composition of the cells after 72 h of differentiation, the greatest difference between the PUFA-supplemented cells and the OA-supplemented cells was a massive increase in the amount of each respective PUFA in the cells. Notably, however, there also were increases in the levels of several other FA species that were highly significant, most prominently in the levels of the saturated FAs stearic acid (18:0) and palmitic acid (16:0). This finding is consistent with studies that have shown that high concentrations of PUFAs repress stearoyl-CoA-⌬9-desaturase expression and activity, which in turn reduces the conversion of stearic acid (18:0) to OA (18:1) (27). This may be important to ensure that saturated FAs, whether newly synthesized or taken up from the external environment, are available for insertion into the sn-1 position of PLs as they are synthesized. Maintaining the proper PUFA/non-PUFA ratio in the PLs of the plasma membrane is critical for normal membrane fluidity and possibly for optimal neurite outgrowth.
Acsl6 Overexpression Does Not Preferentially Target FAs to PLs-Several reports suggest that some Acsl or FATP proteins can influence the partitioning of FAs into either TAG or PLs (20,22,35). Therefore we had hypothesized that the increase in total cellular PL levels that occurs with Acsl6 overexpression may in part be because of an increase in the partitioning of PUFAs into PLs relative to TAG. However, after 15 min of incubation, the relative incorporation of [ 14  C]DHA were incorporated similarly into PC and PI, regardless of whether Acsl6 was overexpressed by the cells. Taken together, these data suggest that Acsl6 does not preferentially or exclusively partition FAs into TAGs, PLs, or specific PL species. However, it should be noted that the subcellular distribution of Acsl6 in Acsl6 overexpressing cells may differ from its normal distribution, thus altering its normal function in FA partitioning.
Acsl6 and DHA Metabolism-As stated earlier, we previously reported that, compared with overexpression of Acsl1, overexpression of Acsl6 preferentially increased the initial rates of AA and DHA internalization relative to the rate of OA internalization (7). Other evidence is consistent with the notion that Acsl6 contributes to DHA enrichment in neurons. In PC12 (7) and reaggregated rat brain cell cultures (36), the level of Acsl6 mRNA is increased upon differentiation, but the level of Acsl1 remains unchanged. Additionally, biochemical studies using bacterially expressed proteins established that Acsl6 activates AA 2-fold and DHA 8-fold more efficiently than Acsl1 does (17). Here we found that, relative to control cells, [ 14 C]DHA internalization, activation to DHA-CoA, and metabolism into complex lipids was significantly and selectively en-hanced in Acsl6 overexpressing cells especially during the first 24 h of differentiation (Figs. 3 and 4). This is the first report to provide in vivo evidence for preferential DHA metabolism by any Acsl protein.
The precise mechanism of how FAs cross the plasma membrane is controversial, but it most likely involves integral and plasma membrane-associated proteins that facilitate FA internalization (13,14,23,37). In the first step, unesterified FAs partition into the outer leaflet of the plasma membrane because of their hydrophobicity or by their interaction with putative FA transporters such as FATP. Once in the plasma membrane, "flip-flop" of the FA from the outer leaflet to the inner leaflet occurs spontaneously or may be enhanced by FA transporters. FAs are then extracted from the inner leaflet of the plasma membrane by membrane-associated FATPs or Acsls or by cytosolic FABPs. However, if the FA is not extracted, the process can be reversed and the FA may repartition back to the extracellular environment. Unesterified FAs extracted from the plasma membrane by FABPs must be transported to subcellular locations where the FAs are esterified to CoA by Acsls or possibly FATPs because there is now evidence that some FATPs possess acyl-CoA synthetase activity (38,39). Equally likely, Acsls or possibly FATPs may couple the extraction of the FA from the membrane with its activation to FA-CoA (21,40). Once activated, FAs cannot repartition back into the plasma membrane because of their decrease in hydrophobicity. FA activation also decreases the concentration of unesterified FA in the cell, thus maintaining a concentration gradient that is favorable for the entry of more unesterified FAs into the cell (13,23).
There are several possible steps in FA internalization and metabolism where Acsl6 may function to promote DHA accumulation. An increase in the initial rate of FA uptake (7,23) and an increase in the accumulation of FAs in Acsl6 overexpressing over a 15-min time frame (Fig. 4A) are consistent with a fraction of Acsl6 localizing to the plasma membrane where it may activate FAs, especially DHA, as it extracts them from the plasma membrane or accepts them from another protein such as a FATP. Either way, esterification of the FA to CoA will maintain the outside Ͼ inside FA concentration gradient in the cell, thus promoting the influx of more FA into the cell. Alternatively, FABPs may extract FA from the membrane and transport the FA to subcellular compartments where the Acsl proteins are located. The FA may be transferred from the FABP to the Acsl, where it would be activated, thus "freeing up" the FABP to bind an additional FA. Regardless of the specific intracellular localization, increased Acsl6 expression increases the rate of FA activation and metabolism, thereby promoting DHA internalization.
It is clear that AA and DHA are critical for proper brain and retina growth. Our study suggests that PUFAs may function to stimulate neuronal differentiation by being preferentially metabolized into PLs and stimulating PL synthesis. Our finding that Acsl6 preferentially promotes DHA metabolism suggests that it and perhaps other proteins may function to internalize DHA efficiently.