The low levels of eicosapentaenoic acid in rat brain phospholipids are maintained via multiple redundant mechanisms

Brain eicosapentaenoic acid (EPA) levels are 250- to 300-fold lower than docosahexaenoic acid (DHA), at least partly, because EPA is rapidly β-oxidized and lost from brain phospholipids. Therefore, we examined if β-oxidation was necessary for maintaining low EPA levels by inhibiting β-oxidation with methyl palmoxirate (MEP). Furthermore, because other metabolic differences between DHA and EPA may also contribute to their vastly different levels, this study aimed to quantify the incorporation and turnover of DHA and EPA into brain phospholipids. Fifteen-week-old rats were subjected to vehicle or MEP prior to a 5 min intravenous infusion of 14C-palmitate, 14C-DHA, or 14C-EPA. MEP reduced the radioactivity of brain aqueous fractions for 14C-palmitate-, 14C-EPA-, and 14C-DHA-infused rats by 74, 54, and 23%, respectively; while it increased the net rate of incorporation of plasma unesterified palmitate into choline glycerophospholipids and phosphatidylinositol and EPA into ethanolamine glycerophospholipids and phosphatidylserine. MEP also increased the synthesis of n-3 docosapentaenoic acid (n-3 DPA) from EPA. Moreover, the recycling of EPA into brain phospholipids was 154-fold lower than DHA. Therefore, the low levels of EPA in the brain are maintained by multiple redundant pathways including β-oxidation, decreased incorporation from plasma unesterified FA pool, elongation/desaturation to n-3 DPA, and lower recycling within brain phospholipids.

CPT-Ia, the other isoform of CPT-I that is localized to the mitochondria in the brain ( 36 ). Previously, Freed et al. ( 27 ) reported that MEP treatment reduced the ␤ -oxidation of 14 C-palmitate and 14 C-arachidonate. However, MEP treatment only increased the esterifi cation of 14 C-palmitate into brain total lipids and triacylglycerol. Overall, we found that MEP treatment led to signifi cant increases in the J in of EPA into ethanolamine glycerophospholipids (EtnGpl) and phosphatidylserine (PtdSer). Interestingly, there were also signifi cant increases in n-3 docosapentaenoic acid (n-3 DPA) (22:5n-3) within choline glycerophospholipids (ChoGpl) and EtnGpl. Therefore, ␤ -oxidation is involved, but not necessary for maintaining low EPA levels in brain phospholipids and elongation/desaturation of EPA into longer n-3 PUFA species may compensate for decreased ␤ -oxidation in order to maintain low levels of EPA in brain phospholipids.

MATERIALS AND METHODS
All procedures were performed in accordance with the policies set out by the Canadian Council on Animal Care and were approved by the Animal Ethics Committee at the University of Toronto. Male Sprague Dawley rats were purchased from Charles Rivers (Saint-Constant, QC, Canada) at 12 weeks of age and kept at the animal facility with an automated 12 h light-dark cycle and a constant temperature of 22°C for 3 weeks. The rats received ad libitum access to water and standard chow (Teklad 2018; Harlan, Madison, WI) which was composed of 54% of linoleate (18:2n-6), 5% of ␣ -linolenate (18:3n-3), and <0.3% of longer chained PUFAs (20:2n-6, 20:3n-3, 20:4n-6, EPA, 22:4n-6, 22:5n-6, 22:5n-3, and DHA), as measured by gas chromatography-fl ame ionization detection (GC-FID).

Surgery
Fifteen-week-old rats were anesthetized with isofl uorane inhalation (3% induction, 1-2% maintenance). Rats were given a lower back subcutaneous injection of 5 mg/kg of ketoprofen (MERIAL Canada, Inc., Baie d'Urfé, QC, Canada). Polyethylene catheters (PE 50, Intramedic ™ ; Becton Dickinson, Franklin Lakes, NJ) with silicone tubing [silicone tubing 0.020 inch inner diameter (I.D.) and 0.037 inch outer diameter(O.D .), VWR ® ; Mississauga, ON, Canada] fi lled with 0.9% saline were implanted into the right jugular vein. Surgery lasted for approximately 15 min. After surgery, all rats were singly housed to recover from anesthesia for at least 24 h with ad libitum access to food and water. Rats were not fasted for the radiotracer infusions. metabolism at a rate of 14% per day ( 6 ) as compared with DHA ( 15 ), ARA, ( 16 ) and palmitate ( 6 ), all at 2% per day. However, there are other potential differences in metabolism between DHA and EPA that could contribute to their large difference in brain phospholipid levels. Therefore, one of the objectives of this study was to investigate further differences in the metabolism of DHA and EPA upon entry into the brain.
In addition to quantifying loss kinetics of EPA from brain phospholipids, the kinetics of EPA uptake, incorporation, and turnover can also be quantified via in vivo intravenous radiotracer infusion in rodents as described by Rapoport and others (17)(18)(19). The method of Rapoport allows for the calculation of brain kinetic parameters upon intravenous infusion of a high specifi c activity radiotracer at steady-state into a plasma pool available to the brain. Upon oral administration of a radiotracer, it appears in multiple pools and is not always at steady-state, making the calculation of brain kinetic parameters diffi cult, if not impossible ( 20 ). In this report, we quantifi ed and compared four key kinetic parameters of palmitate, DHA, and EPA including k *, J in , J FA , and F FA . The incorporation coeffi cient ( k *) describes the proportional uptake of radiolabeled FA from the plasma unesterifi ed FA pool into stable brain phospholipids (17)(18)(19)(21)(22)(23)(24). The J in describes the net rate of plasma unesterifi ed FA incorporation into brain phospholipids, whereas the J FA describes the net rate of brain acyl-CoA incorporation into brain phospholipids ( 25 ). Lastly, the rate of turnover ( F FA ) describes the turnover of deesterifi ed FA from phospholipids that are reesterifi ed into brain phospholipid via Land's recycling ( 25 ).
Because the synthesis of PUFA within the brain is negligible relative to brain uptake ( 22,23 ), the mathematical model can predict the relative contribution of plasma pools from which FAs enter the brain via comparison of the J in to the net rate of loss from brain phospholipids, J out ( 26 ). If the plasma unesterifi ed PUFA pool is the major plasma contributor to brain phospholipid PUFA, then the J in should approximate the J out . However, if the J out exceeds the J in , then there may be additional plasma FA pools contributing to PUFA uptake into brain phospholipids. In contrast, if the J in exceeds the J out , then it would suggest that either the measured kinetic parameters of PUFA are overestimations or modifi cations to the current model may be necessary .
Because ␤ -oxidation appears to be a major contributor to the observed difference in brain PUFA concentrations ( 6 ), another objective of this study was to examine if mitochondrial FA ␤ -oxidation is necessary for maintaining low levels of EPA in brain phospholipids by irreversibly inhibiting carnitine palmitoyltransferase (CPT)-Ia, the rate-limiting enzyme in mitochondrial FA ␤ -oxidation that catalyzes the formation of acyl-carnitine from acyl-CoA, with methyl palmoxirate (MEP; methyl-2-tetradecylglycidate) (27)(28)(29). Although CPT-Ic is predominantly expressed in the brain ( 30 ), it is localized to the endoplasmic reticulum where it mediates food intake via endocannabinoids and ghrelin without modulating brain FA ␤ -oxidation (31)(32)(33)(34)(35). Therefore, the inhibitory effects of MEP would likely be through tandem mass spectrometry (LC-MS/MS) analysis as described below.

Acyl-CoA analysis
Long chain acyl-CoA was extracted from the brains of vehicleand MEP-treated rats as well as the brains of radiotracer-infused rats using a modifi ed affi nity chromatography method ( 40 ). Brains were homogenized by a probe sonicator in isopropanol:25 mM KH 2 PO 4 :acetonitrile (1:1:2 by volume) with 10 nmol of added internal standard, heptadecanoyl-CoA (17:0-CoA). Subsequently, protein was precipitated from the homogenate by the addition of saturated (NH 4 ) 2 SO 4 . After centrifugation for 5 min at 3,000 rpm (2,000 g ), supernatant containing acyl-CoA was extracted and diluted with a 1.25-fold volume of 25 mM KH 2 PO 4 . The diluted supernatant was then repeatedly passed through an oligonucleotide purifi cation cartridge (ABI Masterpiece™, OPC ® ; Applied Biosystems, Foster City, CA) for three times at a rate of 5 ml/min. Afterwards, the cartridge was washed with 25 mM KH 2 PO 4 and the bound acyl-CoA was eluted with isopropanol/1 mM glacial acetic acid (75:25 by volume). Acyl-CoA samples were reconstituted in elution buffer (100 l) and stored at Ϫ 80°C until HPLC analysis as described below.

HPLC-ultraviolet photodiode array detection and LSC
FA methyl esters and radioactive acyl-CoA analysis were previously described by Chen, Liu, and Bazinet ( 6 ).

LC-MS/MS
Acyl-CoA concentrations were detected using an Agilent 1200 Binary LC pump (Agilent Technologies, Wilmington, DE) equipped with a Zorbax SB-Phenyl column (3 × 50 mm, 3.5 m spherical size; Chromatographic Specialties, Brockville, ON, Canada). The initial conditions of elution were set at 600 l/min gradient system consisting of (A) 70% 10 mM ammonium acetate in water and (B) 30% 10 mM ammonium acetate:acetonitrile (10:90 by volume). The gradient started with 70% (A) and 30% (B) and maintained for 1 min, decreased to 30% (A) and 70% (B) over a 5 min period where it was maintained for 6 min before returning to 70% (A) and 30% (B) over a 6.2 min period and maintained for 10 min to complete the total run of 28.2 min. Mass spectrometry analyses were carried out on API4000 QTRAP (AB SCIEX, Concord, ON, Canada) quadruple-linear ion trap (QqLIT) mass spectrometers. The QTRAP analyses were conducted in positive ion mode under multiple reaction monitoring conditions. The turbospray temperature was set to 600°C, the curtain gas fl ow to 30 psi, and the ion spray voltage to 5,500 V. The collision energy, declustering potential, and collision cell exit potential were optimized and were set to 45 Fig. I). Concentrations were corrected for percent recovery of heptadecanoyl-CoA and expressed as nmol/g brain.
Brain unesterifi ed EPA was detected using an Agilent HPLC 1200 equipped with a Zorbax SB-Phenyl column. HPLC solvent contained 4 l/l propionic acid. The initial conditions of elution were set at 400 l/min gradient system consisting of (A) water and (B) acetonitrile. The gradient started with 80% (A) and 20% (B) and maintained for 2 min, decreased to 75% (A) and 25% (B) for 0.5 min, then further decreased to 50% (A) and 50% (B)

Free-living intravenous tracer infusion
Twenty-four hours postsurgery, a second catheter (iv catheter 24 gauge/0.75 inch, Angiocath ™ ; Becton Dickinson) was implanted into the tail vein where rats received either vehicle or 10 mg/kg of MEP. Ten minutes post injection of vehicle or MEP, the tail vein catheter was connected to a syringe containing radiolabeled 14 C-palmitate (n = 8), 14 C-DHA (n = 6), or 14  C-EPA. Thus a total of 1.39 mol of palmitate, 1.42 mol of DHA, or 1.44 mol of EPA were infused over 5 min. During the 5 min infusion, blood samples were collected from the jugular vein at approximately 0, 0.25, 0.5, 0.75, 1.5, 3, 4, and 5 min while the unanesthetized rat was mobile in the infusion box with food and water. In a pilot study, we found that there were no signifi cant differences in the plasma unesterifi ed FA concentrations between carotid artery (palmitate, 163 ± 20 nmol/ml; EPA, 0.29 ± 0.017 nmol/ml, DHA, 0.63 ± 0.059 nmol/ml) and jugular vein (palmitate, 163 ± 22 nmol/ml; EPA, 0.31 ± 0.025 nmol/ml; DHA, 0.70 ± 0.058 nmol/ml). After 5 min, the rats were rapidly euthanized by head-focused high-energy microwave irradiation (13.5 kW for 1.6 s; Cober Electronics Inc., Stratford, CT). The radiotracer infusions continued until rats were euthanized. The brain was excised and cut sagittally. Both hemispheres were stored at Ϫ 80°C for radioactive and biochemical analyses. Plasma was isolated from whole blood via microcentrifugation at 6,200 rpm (2,000 g ) for 5 min and stored at Ϫ 80°C.

Lipid analysis
Total lipids from one brain hemisphere and from plasma were extracted by the method of Folch, Lees, and Sloane Stanley ( 39 ). Isolation of various neutral lipid and phospholipid classes from the total lipid extract was previously described by Chen, Liu, and Bazinet ( 6 ).

Unesterifi ed FA analysis
At 15 weeks of age, 16 rats were randomized to receive vehicle or 10 mg/kg MEP via tail vein catheter as described above. Fifteen minutes post injection, rats were euthanized by headfocused high energy microwave irradiation (13.5 kW for 1.6 s) and brains were excised and sagittally cut for unesterifi ed FAs/lipid mediators and acyl-CoA analyses as described below. Brain hemispheres were homogenized in ethanol to yield a concentration of 100 mg tissue/ml. The unesterifi ed FAs from 100 mg of brain tissue were isolated. The unesterifi ed FA bands were collected and extracted twice from silica by hexane:isopropanol (3:2 by volume) with 5.5% water. Extracted unesterifi ed FAs were dried down with nitrogen gas and added with 100 l of freshly made pentafl uorobenzylbromide (PFB) cocktail consisting of acetonitrile:N, N-diisopropylethylamine:PFB (1 ml:100 l:10 l by volume). The mixtures were shaken for 15 min and dried down with nitrogen gas. FA-PFB esters (FAPEs) were reconstituted in 100 l hexane for GC-MS analysis as described below.
For measurements of unesterifi ed EPA, brain samples were homogenized in ethanol on ice. An aliquot of 100 mg tissue was spiked with ARA-d8 (20 ng; Cayman Chemical, Ann Arbor, MI) and dried under nitrogen gas in reduced light conditions. Residues were dissolved in ethanol, acidifi ed to pH 4 with 1 N HCl, and extracted three times with ethyl acetate. After washing to neutrality with water, the ethyl acetate fraction was dried under nitrogen and transferred to siliconized minivials for liquid chromatography- is the conversion-incorporation coeffi cient and * brain, (n-3 DPA or DHA) i c is the brain radioactivity of n-3 DPA or DHA as determined by HPLC and LSC.
Because the incorporation coeffi cient applies to both radiolabeled and nonradiolabeled FAs, we can determine the rate of incorporation of nonradiolabeled plasma unesterifi ed FAs into stable brain lipid pools as unmetabolized FAs, in, (palmitate, DHA, or EPA) i J , or as elongation/desaturation products in, (EPA n-3 DPA or DHA) i J (nmol/g brain/day).
where plasma(palmitate, DHA ,or EPA) c is the plasma unesterifi ed FA concentration.
In addition to the rate of incorporation from plasma to stable brain lipid pools, we can also calculate the net rate of incorporation, FA, (palmitate, DHA, or EPA) i J (nmol/g brain/day) ( 18,41 ), from the brain acyl-CoA pool to stable brain lipid pools via correction for the steady-state ratio of specifi c activity of the acyl-CoA pool over the specifi c activity of the radiotracer in plasma which is defi ned as the dilution factor, λ palmitate, DHA, or EPA, λ * brain(palmitate, DHA, or EPA-CoA) brain(palmitate, DHA, or EPA-CoA) palmitate, DHA, or EPA * plasma(palmitate, DHA, or EPA) plasma(palmitate, DHA, or EPA) where the numerator and denominator are the steady-state specifi c activities of brain acyl-CoA and plasma unesterifi ed FAs, respectively. Because the infusion is 5 min, contributions of FA from de novo synthesis and esterifi ed plasma FA are negligible ( 18,20,23,41 ); thus only contributions from plasma unesterifi ed FAs and acyl-CoA pools were considered in the calculation of λ palmitate, DHA, or EPA .
FA, (palmitate, DHA ,or EPA) i J of nonradiolabeled FAs and EPA elongation/desaturation products from acyl-CoA pools into stable brain lipid pools i are calculated as followed: (%/day), and halflife, t 1/2 (day) ( 18,41 ), within stable brain lipid pools i as unmetabolized FAs or EPA elongation/desaturation products are quantifi ed as, FA, (palmitate, DHA, or EPA) FA, (palmitate, DHA, or EPA) brain, (palmitate, DHA, or EPA) FA, (EPA n-3 DPA or DHA) FA, (EPA n-3 DPA or DHA) brain, (n-3 DPA or DHA) where brain, (palmitate, DHA, or EPA) i c is the brain FA concentrations of stable brain lipid pools i . for 5 min, then to 45% (A) and 55% (B) for 6.2 min, and 100% (B) for 11 min. Mass spectrometry analyses were carried out on API5500 triple quadruple mass spectrometer (AB SCIEX, Concord, ON, Canada). The QTRAP analyses were conducted in electrospray ionization negative ion mode. The turbospray temperature was set to 500°C and the ion spray voltage to 4,500 V. The collision energy, declustering potential, and collision cell exit potential were optimized and were set to 15, 50, and 11, respectively. Concentration was quantified by comparing the deuteriumto-protium ratio of brain unesterifi ed EPA with standard lines generated from authentic standards. Authentic standards in appropriate dilutions (0.002-2 ng) were prepared and analyzed simultaneously with brain samples. The lower limit of quantifi cation (LLQ) was 0.002 ng (6.3 fmol of EPA).

GC-FID
Brain FA concentrations from total and phospholipid classes were quantifi ed as described by Chen, Liu, and Bazinet ( 6 ).

GC-MS
FAPEs were identifi ed with an Agilent 7890A gas chromatograph (Agilent Technologies) equipped with a SP-2380 (Supelco) fused silica column (Agilent Technologies; 30 m × 0.25 mm I.D. × 0.2 m fi lm thickness) and an Agilent 5975C quadruple mass spectrometry detector (Agilent Technologies). The sample was injected in split mode (10:1). The injection port temperature was set to 240°C and the ionization mode was set to negative chemical ionization using methane. FAPEs were eluted using a temperature program initially set at 150°C for 1 min, increase at 12°C/min to 270°C, and then at 40°C/min to 275°C for 3 min. The carrier helium gas was set to a constant fl ow of 1 ml/min. The LLQ for n-3 docosapentaenoate was 1 ng/20 mg brain. The LLQ for palmitate, palmitoleate, ␣ -linolenate, and EPA was 5 ng/20 mg brain. The LLQ for ARA and linoleate was 10 ng/20 mg brain. The LLQ for oleate, stearate, and DHA was 20 ng/20 mg brain.

Kinetics
Total and phospholipid class radioactivity were adjusted by the percentage of radiolabeled palmitate, DHA, and EPA as measured by HPLC and LSC for kinetic modeling. The model for in vivo kinetics of brain FAs in rats has been previously described ( 17,18,21,23,25,37,41 ).
The unidirectional incorporation coeffi cient, * i(palmitate, DHA, or EPA) k (ml plasma/day/g brain), which represents the incorporation of plasma radiotracers into stable brain lipid pools i , was calculated as: where * brain, (palmitate, DHA, or EPA) i c is the radioactivity in i (nCi/g brain) from palmitate, DHA, or EPA at time T , and * plasma(palmitate, DHA, or EPA) c is the plasma radioactivity (nCi/ml plasma) of 14 C-palmitate-, 14 C-DHA-, or 14 C-EPA-infused rats. Because elongation/desaturation products of 14 C-EPA, n-3 DPA, and DHA were detected, we can determine their incorporation into stable lipid pools i with following adjustment to the equation: exceptions of a 9% reduction in linoleate from EtnGpl and an 8% reduction in ARA from PtdSer upon MEP treatment ( Table 2 ). After adjustments for pool size, palmitate was primarily esterifi ed to ChoGpl and as the major component in ChoGpl, it accounted for 48% of total ChoGpl FAs ( Table 2 ). As for DHA, it was primarily esterifi ed to EtnGpl and PtdSer which accounted for 17% of total FAs in both EtnGpl and PtdSer ( Table 2 ). As for EPA, albeit, the esterifi cation of EPA to phosphatidylinositol (PtdIns) was the lowest among four major phospholipid classes, when pool size is considered, 0.5% of total PtdIns FAs were EPA as compared with 0.1, 0.2, and 0.4% of total ChoGpl, EtnGpl, and PtdSer FAs, respectively.

Identifi cation of radioactivity in total and phospholipid fractions
After a 5 min infusion of radiolabeled 14 C-palmitate, the only radioactive peak in plasma total lipids and brain total phospholipids of the vehicle-and MEP-treated rats was palmitate ( Fig. 1B, C ). Similarly, upon 14 C-DHA infusion, the only radioactive peak in plasma total lipids and brain total phospholipids of vehicle-and MEP-treated rats was DHA ( Fig. 1B, C ). Upon 14 C-EPA infusion, while plasma total lipids of the vehicle-and MEP-treated rats only contained radioactive EPA peak ( Fig. 1B ), brain total phospholipids from the vehicle-and MEP-treated rats all contained radiolabeled EPA, DHA, n-3 DPA, and palmitate ( Fig. 1C ). However, we observed higher n-3 DPA and lower DHA in the MEP-treated rats. The major radioactive peak in brain total phospholipids of the vehicle-and MEP-treated rats was EPA, accounting for 62 and 56% of total radiolabeled FAs, respectively; whereas the minor peak was palmitate, accounting for 4 and 2% of brain total radiolabeled FAs in the vehicle-and MEP-treated rats, respectively. In the vehicletreated rats, 24 and 10% of brain total radiolabeled FAs corresponded to n-3 DPA and DHA, respectively, elongation and desaturation products of EPA; while in the MEP-treated rats, the composition of radiolabeled n-3 DPA and DHA was 38 and 3% of brain total radiolabeled FAs, respectively.
Because we detected elongation and desaturation products in the 14 C-EPA-infused rats, the percent composition of radiolabeled FAs in each major phospholipid class was also measured ( Fig. 1D ). The radiolabeled FA composition of ChoGpl from the vehicle-treated rats was 65% EPA, 11% DHA, 18% n-3 DPA, and 6% palmitate; whereas the composition from the MEP-treated rats was 66% EPA, 4% DHA, 26% n-3 DPA, and 3% palmitate. In contrast to ChoGpl, only three of the four radiolabeled FAs were detected in EtnGpl. From EtnGpl of the vehicle-treated rats, the radiolabeled FA composition was 24% EPA, 29% DHA, and 47% n-3 DPA; while in the MEP treated rats, the composition was 25% EPA, 14% DHA, and 61% n-3 DPA. In PtdIns and PtdSer fractions, there were only two detectable radiolabeled peaks, EPA and n-3 DPA. In PtdIns of the vehicle-and MEP-treated rats, EPA accounted for the majority of brain radiolabeled FAs at 66 and 65%, respectively. However, in PtdSer, there was more radiolabeled EPA (54% of total radiolabeled FAs) in the MEP-treated rats as opposed to the vehicle-treated rats

Statistics
Concentrations and rates are expressed as mean ± SD. HPLC profi les were analyzed as pooled samples and do not have SD. Statistical comparisons of kinetic parameters between 14 C-DHA and 14 C-EPA infusions were performed, a priori, using two-tailed t -test. Differences between 14 C-DHA and 14 C-EPA infusions upon vehicle administration were statistically signifi cant at # P < 0.05, ## P < 0.01, and ### P < 0.001. Comparisons were not performed with 14 C-palmitate which served as positive control to confi rm the activity of MEP ( 27 ). Statistical comparisons of FA and acyl-CoA concentrations, radioactivity, and kinetic parameters between vehicle-and MEP-treated rats were performed using two-tailed t -test. All data had passed the normality and equal variance test (SigmaStats 3.5). Differences between vehicle-and MEPtreated rats were statistically signifi cant at * P < 0.05, ** P < 0.01, and *** P < 0.001.
MEP had no effect on the concentration of palmitoyl-CoA but reduced the concentration of stearoyl-CoA ( Table 1 ). MEP increased linoleoyl-CoA and ␣ -linolenoyl-CoA and decreased EPA-CoA. Interestingly, while MEP had no effect on unesterifi ed DHA, it signifi cantly reduced DHA-CoA in the brain.
Acute administration of MEP was insuffi cient to significantly alter total phospholipid FA concentrations ( Table 1 ). There were no signifi cant differences between the brain FA concentrations of all radiotracer-infused rats in each treatment; therefore, data were pooled for comparison between vehicle and MEP treatments. In accordance with previous reports, palmitate, stearate, and oleate were the major constituents of brain phospholipids consisting of 25, 22, and 21% of the total FAs, respectively. Similarly, the major PUFA species in brain phospholipids were ARA and DHA, which accounted for 8% and 9% of total FAs, respectively. Lastly, the level of EPA in brain total phospholipids was relatively low, as compared with DHA, at 0.2% of total FAs which corresponded to 225 ± 12 nmol/g brain in vehicle-treated rats and 202 ± 8.6 nmol/g brain in MEP-treated rats.
After fractionation into the four major phospholipid classes, there was no signifi cant effect of MEP on the FA compositions of individual phospholipid classes with the total phospholipids (vehicle, 17 ± 0.9 nCi/g brain; MEP, 22 ± 0.6 nCi/g brain; P < 0.05) (data not shown). However, upon adjusting the radioactivity for percent composition of 14 C-EPA, we observed no signifi cant effect of MEP treatment on esterifi cation of 14 C-EPA into brain total phospholipids ( P = 0.06), but esterifi cation into PtdSer was increased upon MEP treatment ( Fig. 2 ). It is possible with a larger sample size that we would have observed a signifi cant effect of MEP on 14 C-EPA incorporation into total phospholipids and this result should be interpreted with caution. The increase in total radioactivity upon MEP treatment was largely due to a signifi cant increase in the esterifi cation of radiolabeled n-3 DPA into brain total phospholipids (vehicle, 4.1 ± 0.2 nCi/g brain; MEP, 8.3 ± 0.2 nCi/g brain; P < 0.001) (data not shown). There was also a signifi cant 49% increase in radioactivity of PtdSer with MEP treatment; while other phospholipid classes were unaffected by MEP ( Fig. 2 ).
Furthermore, to account for the amount of infused radiotracer in the plasma available to the brain, incorporation coeffi cients [ k i * (Eq. 1)] were determined. There was a signifi cant increase in k i * of 14 C-palmitate into ChoGpl and PtdIns upon MEP treatment ( Table 3 ). Upon 14 C-DHA infusion, there was no effect of MEP on k i * ( Table 3 ). In 14 C-EPAinfused rats, there was no effect of MEP on k i * of brain total phospholipids. However, there were signifi cant increases in k i * for 14 C-EPA into EtnGpl in addition to PtdSer ( Table 3 ).
(44% of radiolabeled FAs). Lastly, there was no esterifi cation of 14 C-EPA into ceramide phosphocholine after 5 min of infusion (data not shown).

Radioactivity in brain aqueous and lipid fractions
MEP signifi cantly decreased radioactivity in the brain aqueous fraction (marker of ␤ -oxidation) for all radiotracers ( Fig. 2 ). Upon vehicle injections, the radioactivity of aqueous fractions between radiotracer-infused rats were similar ( 14 C-palmitate, 17 nCi/g brain; 14 C-DHA, 17 nCi/g brain; 14 C-EPA, 21 nCi/g brain) ( Fig. 3 ). However, post MEP treatment, radioactivity in the brain aqueous fraction of the 14 C-palmitate-infused rats was signifi cantly reduced by 74%, whereas in the 14 C-EPA-infused rats and the 14 C-DHA-infused rats the radioactivities of the brain aqueous fractions were reduced by 54 and 23%, respectively.
In 14 C-palmitate-infused rats, MEP had no signifi cant effect on esterifi cation into brain total phospholipids ( Fig. 2 ). However, there was a signifi cant 65% increase in radioactivity of PtdIns in MEP-treated rats; while no signifi cant differences were observed in ChoGpl, EtnGpl, and PtdSer between the brains of the vehicle-and MEP-treated rats ( Fig. 2 ). In 14 C-DHA-infused rats, MEP had no signifi cant effect on brain total phospholipids or any individual phospholipid classes ( Fig. 2 ). Finally, upon 14 C-EPA infusion, MEP signifi cantly increased total radioactivity in brain Unesterifi ed FAs (n = 8 per treatment), acyl-CoA (n = 8 per treatment), and total phospholipids (total PL) (n = 10-11 per treatment). Data are mean ± SD and are expressed in nmol/g brain. Unesterifi ed FA concentrations were quantifi ed by GC-MS with exception of EPA + which was determined by LC-MS/MS. Total phospholipid FA concentrations were quantifi ed by GC-FID. Acyl-CoA concentrations were quantifi ed by LC-MS/ MS. P values indicated signifi cantly different from vehicle-treated rats; * P < 0.05, ** P < 0.01, *** P < 0.001. Brain 22:5n-3 (n-3 DPA)-CoA was not determined (ND). Data are mean ± SD and are expressed in nmol/g brain (n = 10-11 per treatment). FA concentrations were quantifi ed by GC-FID. P values indicated signifi cantly different from vehicle-treated rats; * P < 0.05, ** P < 0.01 . total phospholipids and ChoGpl of the MEP-treated rats ( Table 3 ).
MEP did not signifi cantly affect the J FA (Eq. 4) of palmitoyl-CoA into brain phospholipids ( Table 3 ). In 14 C-DHAinfused rats, MEP signifi cantly reduced the J FA of DHA-CoA in brain total phospholipids, ChoGpl and EtnGpl ( Table 3 ). Similarly, in 14 C-EPA-infused rats, there were signifi cant reductions in the J FA of EPA-CoA in brain total phospholipids and ChoGpl with MEP treatment ( Table 3 ). Additionally, MEP did not affect the J FA of EPA-synthesized n-3 DPA-CoA into brain phospholipids ( Table 3 ). However, in accordance with MEP's effect on the J FA of DHA-CoA, MEP also signifi cantly reduced the J FA of EPA-synthesized Net rates of incorporation of palmitate, DHA, and EPA from plasma unesterifi ed FA and brain acyl-CoA pools into brain phospholipids The J in (Eq. 2) of palmitate was signifi cantly higher in ChoGpl and PtdIns upon MEP treatment ( Table 3 ). There were no signifi cant effects of MEP on the J in of DHA into brain phospholipids ( Table 3 ). Upon MEP treatment, the J in for EPA was signifi cantly increased in EtnGpl and Ptd-Ser ( Table 3 ). Additionally, there were signifi cant increases in the J in of EPA-synthesized n-3 DPA into brain total phospholipids, ChoGpl and EtnGpl of the MEP-treated rats ( Table 3 ). However, this was accompanied by signifi cant decreases in the J in of EPA-synthesized DHA into brain Fig. 2. Radioactivity of the aqueous (AQ) and organic fractions including total phospholipids (total PL) and four major phospholipid fractions (n = 3-4) upon HPLC adjustment in (A) 14 C-palmitate-infused, (B) 14 C-DHA-infused and (C) 14 C-EPA-infused rats. Data are mean ± SD and are expressed in nCi/g brain. Fractions were isolated by TLC and radioactivity counted by LSC. P values indicated signifi cantly different from vehicle-treated rats; * P < 0.05, ** P < 0.01, *** P < 0.001. Fig. 3. Kinetic summary of palmitate, DHA, and EPA in brain total phospholipids for vehicle (black) and MEP-treated (gray) rats. Kinetic rates ( J in and J FA ) are nmol/g brain/day and radioactivity is nCi/g brain. P values indicated signifi cantly different from vehicle-treated rats; * P < 0.05, *** P < 0.001, ! P = 0.06. P values indicating signifi cant differences between 14 C-DHA-infused and 14 C-EPA-infused rats are ## P < 0.01 .  n-6 PUFA in rat brain phospholipids ( 27 ). While there was no effect of MEP on the esterifi cation of 14 ( 27 ). Furthermore, there were signifi cant increases in unesterifi ed FAs known to be extensively ␤ -oxidized upon entry to the brain including palmitate ( 27,42 ), linoleate ( 22 ), ␣ -linolenate ( 23 ), and EPA ( 5,6 ); whereas the concentration of relatively metabolically stable unesterifi ed PUFAs, including ARA ( 16 ) and DHA ( 15 ), were unaffected. Previously, using GC-MS, we did not detect unesterifi ed EPA in the brain, but our detection limit was 60 pmol ( 6 ). In the current study, again we did not detect EPA by GC-MS (data not shown), but upon more sensitive LC-MS/MS we estimated, for the fi rst time, the brain unesterifi ed EPA pool to be 19 pmol/g brain.
In accordance with previous studies ( 5,6 ), ␤ -oxidation of palmitate, EPA, and DHA was confi rmed by the synthesis of radiolabeled cholesterol at 1.4, 1.1, and 0.8 nCi/g brain, respectively (data not shown). Moreover, we observed that within 5 min, EPA was metabolized into longer chain PUFAs such as n-3 DPA and DHA via elongation and desaturation, as well as ␤ -oxidized and resynthesized into palmitate via FA synthase. When ␤ -oxidation was inhibited by MEP, there was an increase in EPA elongation to n-3 DPA, but not DHA. This suggests that without ␤ -oxidation to remove the infl ux of EPA, the brain compensates by elongating some EPA to n-3 DPA, which emerging evidence suggests may be bioactive in the brain ( Fig. 3 ) ( 43,44 ). Subsequently, the signifi cant increase in total radioactivity of brain total phospholipids with MEP treatment was partly the result of increased esterifi cation of radiolabeled n-3 DPA. However, the brain synthesis of DHA from EPA appears inadequate to maintain the turnover of DHA in brain phospholipids because it would require 768 days to replace phospholipid DHA with DHA-CoA synthesized from EPA; whereas it only required about 1 day to replace phospholipid DHA with intact DHA-CoA.
In addition to the metabolism of EPA via elongation and desaturation, we also observed a 12% reduction in (vehicle, 0.12 ± 0.014; MEP, 0.24 ± 0.022; P < 0.01) and a 16% reduction in EPA-CoA upon MEP treatment suggesting that inhibition of ␤ -oxidation further reduced the recycling of EPA into brain total phospholipids ( Fig. 3 ). Interestingly, upon MEP treatment, DHA-CoA was also reduced by 26%; whereas linoleoyl-CoA and ␣ -linolenoyl-CoA concentrations increased by 25 and 49%, respectively. Although unclear, these changes in acyl-CoA concentrations may be explained by the selectivity of long chain fatty acyl-CoA synthetases where the inhibition of ␤ -oxidation leads to an infl ux of unmetabolized linoleate and ␣ -linolenate which competes for long chain fatty acyl-CoA synthetases thereby reducing synthesis of EPA-CoA and DHA-CoA ( 45,46 ). DHA-CoA into brain total phospholipids, ChoGpl and EtnGpl ( Table 3 ).
In comparing k i * between DHA and EPA, there was no signifi cant difference between the k i * ( Table 3 ). However, the J in of DHA into brain total phospholipids was 3-fold higher than EPA upon vehicle-injection ( Table 3 ). In addition, the J FA of DHA-CoA into total phospholipids was 156-fold higher as compared with EPA-CoA upon vehicle injection ( Table 3 ).

Rate of turnover of palmitate, DHA, and EPA in brain phospholipids
While MEP did not affect the rate of turnover, F FA (Eq. 5), of palmitate and EPA in brain total phospholipids, MEP signifi cantly reduced the F FA of DHA in brain total phospholipids by 24% per day ( Table 3 ). When individual phospholipid classes were analyzed, MEP did not significantly alter the F FA of palmitate in any phospholipid classes ( Table 3 ). However, MEP signifi cantly reduced the F FA of EPA and DHA in ChoGpl ( Table 3 ). Similar to MEP's effect on the F FA for DHA, MEP also signifi cantly reduced the F FA of EPA-synthesized DHA to brain total phospholipids, ChoGpl and EtnGpl; whereas there was no effect on EPA-synthesized n-3 DPA ( Table 3 ). The F FA of DHA into brain total phospholipids was 4-fold higher as compared with EPA upon vehicle injection ( Table 3 ).
In regard to half-life, t 1/2 (Eq. 6), MEP did not signifi cantly affect the half-life of palmitate, DHA, or EPA in brain total phospholipids ( Table 3 ). Similarly, there was no effect of MEP on the t 1/2 of palmitate, DHA, and EPA in phospholipid fractions except for EPA in ChoGpl where MEP treatment signifi cantly increased the t 1/2 of EPA by 1.7-fold ( Table 3 ). In addition, MEP did not signifi cantly affect the t 1/2 of EPAsynthesized n-3 DPA except in PtdSer where there was a signifi cant increase in the t 1/2 by 1.6-fold ( Table 3 ). Lastly, MEP signifi cantly increased the t 1/2 of EPA-synthesized DHA in brain total phospholipids, ChoGpl, and EtnGpl by 4.2-, 4.3-, and 2.5-fold, respectively ( Table 3 ). The t 1/2 of DHA was 4-fold lower than EPA upon vehicle injection ( Table 3 ).

DISCUSSION
When ␤ -oxidation was inhibited by MEP, we observed signifi cant reductions in radioactivity of the aqueous fractions for 14 C-palmitate-infused rats by 74%, 14 C-EPAinfused rats by 54%, and 14 C-DHA-infused rats by 23% ( Figs. 2 and 3). The relatively small reduction in the brain aqueous fraction of 14 C-DHA-infused rats suggests that ␤ -oxidation products are a small percentage of the radioactivity in the aqueous fraction, and the majority of radioactivity in the aqueous fraction may be attributed to other water-soluble DHA metabolites or glycolipids. We also observed signifi cant increases in the esterifi cation of 14 C-palmtiate into PtdIns and of 14 C-EPA into PtdSer; whereas there was no significant change in esterifi cation of 14 C-DHA into brain phospholipids ( Fig. 2 ). The lack of increase in esterifi cation of DHA into brain phospholipids upon MEP treatment is similar to a previous report with ARA, which is the major the use of heparin. In this study, as opposed to previous kinetic reports, we did not administer heparin which activates endothelial and hepatic lipoprotein lipase which promotes lipolysis of triacylglycerol into unesterifi ed FAs ( 57,58 ). This highlights the importance of comparing kinetics within the same model strains with the same dietary regimen and under the same experimental conditions. In comparison to the previous reported J out for DHA (58-257 nmol/g/day) ( 15 ), the J out exceeded the J in suggesting that multiple plasma pools may be required to maintain DHA levels in brain phospholipids. However, as mentioned previously, differences in experimental conditions may account for the discrepancy; hence, comparison of the J in and the J out under similar experimental conditions is warranted. In the case of EPA, the calculated J in of EPA (5.7 nmol/g/day) accounted for 36% of the J out (16 nmol/g/ day) ( 6 ). This implies that the maintenance of EPA levels in brain phospholipids may require other plasma pools in addition to the unesterifi ed FA pool.
In our previous investigations, we consistently found higher esterifi cation of 14 C-EPA into PtdIns as compared with other phospholipid classes and DHA ( 5,6 ). This is of interest because PtdIns is a key modulator in several signaling cascades ( 59 ) and is a candidate therapeutic target for drugs used to treat bipolar disorder, a disorder that where EPA may be effi cacious (60)(61)(62)(63)(64). Therefore, an interesting aspect of this study was to examine if EPA in PtdIns would increase upon MEP treatment. Upon inhibition of ␤ -oxidation by MEP, we observed no signifi cant changes in the J in for EPA into PtdIns. Furthermore, in contrast to our in situ investigation ( 5 ), we found that the k i * of 14 C-EPA into PtdIns was signifi cantly lower when compared with 14 C-DHA. The majority of 14 C-EPA esterification was into ChoGpl and EtnGpl as opposed to PtdIns. There are two possible explanations for these discrepancies: 1 ) the brain concentration of EPA in PtdIns may be tightly regulated acutely in vivo, or 2 ) increased esterifi cation of 14 C-EPA into PtdIns may require phospholipid remodeling that does not occur upon acute intravenous infusion in vivo ( 6 ). Therefore, a study that traces the time course of the metabolism and remodeling of EPA containing phospholipids is warranted.
In conclusion, the 250-to 300-fold difference in DHA and EPA brain phospholipid levels may be due to multiple redundant mechanisms including ␤ -oxidation, decreased incorporation from the plasma unesterified FA pool, elongation/desaturation to n-3 DPA, and lower recycling within brain phospholipids ( Fig. 3 ). While ␤ -oxidation may play a role in removing EPA from the brain, this process is not necessary to maintain low levels of EPA because inhibition of ␤ -oxidation can be compensated by increasing EPA elongation/desaturation and reducing EPA recycling.
Computer programmable pump software was designed by Brian Scott. Mass spectrometry analyses of unesterifi ed FAs and acyl-CoA concentrations from the brains treated with vehicle or MEP were performed at the Analytical Facility for Bioactive Molecules (AFBM) by Michael Leadley, Ashley St.
Previously, we had demonstrated that there were differences between DHA and EPA metabolism in the brain, including ␤ -oxidation ( 5 ) and loss kinetics ( 6 ), which may partially explain the 250-to 300-fold difference in their brain phospholipid levels. In this study, we further explored additional differences in the metabolism of DHA and EPA to explain large differences in their levels. First, there were no signifi cant differences in the k i * between DHA and EPA, which recapitulated our previous in situ fi nding ( 5 ). Although there was no difference in the k i * of DHA and EPA, the net rate of incorporation ( J in ) of plasma unesterifi ed DHA into brain phospholipids was 3-fold higher than EPA. Specifi cally, we found that the most striking difference was in the net rate of incorporation ( J FA ) of brain DHA-CoA and EPA-CoA into brain phospholipids. The J FA of DHA-CoA into brain phospholipids was 154-fold higher than EPA-CoA ( Fig. 3 ). This implies that the major difference in brain DHA and EPA concentration is not due to uptake from the plasma, but rather from recycling within the brain acyl-CoA pool. This large difference in esterifi cation from the acyl-CoA pool may be attributed to a 36-fold difference in brain acyl-CoA concentrations and a 50-fold difference in (DHA, 0.0024 ± 0.0003 vs. EPA, 0.12 ± 0.01; P < 0.01). This translated to 75% recycling of DHA per day (t 1/2 of 22 h) and 21% recycling of EPA per day (t 1/2 of 3.3 days) in brain phospholipids. The lack of EPA recycling in brain phospholipids may explain the rapid loss of EPA (loss t 1/2 : 5 days or 14% per day) ( 6 ) from brain phospholipids as compared with DHA (loss t 1/2 : 33 days or 2% per day) ( 15 ). Although not measured in this study, it would be interesting to investigate if EPA released from brain phospholipids is converted to bioactive lipid mediators including E-series resolvins ( 47,48 ).
In comparison to previous reports calculating the J in of palmitate (724-822 nmol/g/day) ( 25,49,50 ), our calculated J in of palmitate (2,004 nmol/g/day) for adult rats was 2.4-to 2.8-fold higher. However, this difference appears to be driven by the plasma unesterifi ed palmitate concentration as the k i * of palmitate in our study was comparable to previous reports ( 19,20,25,27 ). As compared with our previously reported J out for palmitate (469 nmol/g/day) ( 6 ), the J in exceeded the J out , but the J in may be an overestimate. Albeit, we only detected radiolabeled palmitate in the brains of 14 C-palmitate-infused rats; we did not identify the position of radiolabeled carbon. Therefore, it is not possible to differentiate between intact infused 14 C-palmitate and resynthesized 14 C-palmitate from recycling of 14 C in de novo FA synthesis (51)(52)(53)(54)(55)(56). Similarly, the J out may be underestimated because of extensive palmitate ␤ -oxidation and active resynthesis. Future studies that identify the position of radiolabeled carbons could improve the quantifi cation of J in and J out for palmitate. The J in of DHA (17 nmol/g/day) was 89-91% lower than the reported J in of 150-190 nmol/g/day ( 21,38 ). The difference was again driven by a discrepancy in plasma unesterifi ed DHA concentration. There are several possible explanations for this discrepancy including: 1 ) different strains of rats; 2 ) our rat chow did not contain EPA and DHA, which may decrease circulating unesterifi ed EPA and DHA; and/or 3 )