Chronic dietary n-6 PUFA deprivation leads to conservation of arachidonic acid and more rapid loss of DHA in rat brain phospholipids.

To determine how the level of dietary n-6 PUFA affects the rate of loss of arachidonic acid (ARA) and DHA in brain phospholipids, male rats were fed either a deprived or adequate n-6 PUFA diet for 15 weeks postweaning, and then subjected to an intracerebroventricular infusion of 3H-ARA or 3H-DHA. Brains were collected at fixed times over 128 days to determine half-lives and the rates of loss from brain phospholipids (Jout). Compared with the adequate n-6 PUFA rats, the deprived n-6-PUFA rats had a 15% lower concentration of ARA and an 18% higher concentration of DHA in their brain total phospholipids. Loss half-lives of ARA in brain total phospholipids and fractions (except phosphatidylserine) were longer in the deprived n-6 PUFA rats, whereas the Jout was decreased. In the deprived versus adequate n-6 PUFA rats, the Jout of DHA was higher. In conclusion, chronic n-6 PUFA deprivation decreases the rate of loss of ARA and increases the rate of loss of DHA in brain phospholipids. Thus, a low n-6 PUFA diet can be used to target brain ARA and DHA metabolism.

with policy statements of the Canadian Council on Animal Care. One hundred and twelve male Fischer (CDF) rats were purchased from Charles River Laboratories (Saint-Constant, QC, Canada) and arrived at the Division of Comparative Medicine animal facility at 21 days of age. Rats were housed in a 22°C environment with a 12 h light-dark cycle, and they received ad libitum access to water and food throughout the study. Upon arrival, they were randomized to receive either the n-6 PUFA-deprived (n = 56) or n-6 PUFA-adequate diet (n = 56) (supplementary Fig. 1). Measurements of body weight and food intake were carried out on a weekly basis for 15 weeks. Following 15 weeks of feeding, 8 rats from each dietary group were euthanized by high-energy, head-focused microwave irradiation (13.5 kW for 1.6 s; Cober Electronics Inc., Norwalk, CT) as previously described (18)(19)(20)(21). Their brains were removed, dissected sagittally, and stored at Ϫ 80°C for measurements of eicosanoid levels, docosanoid levels, and baseline brain phospholipid fatty acid concentrations.

n-6 PUFA-adequate and n-6 PUFA-deprived diets
The n-6 PUFA-adequate and n-6 PUFA-deprived rodent diets were based on the AIN-93G formulation with a 10% fat composition ( 22,23 ) as used by others (15)(16)(17). The diets were purchased from Dyets Inc. (Bethlehem, PA), under the following product names: Revised Modifi ed n-6 PUFA Adequate Diet (Dyet# 180780) and Custom Modifi ed n-6 PUFA Defi cient Diet (Dyet #180784). In order to be consistent with the literature (15)(16)(17), we chose to use the terms "adequate" and "deprived" to describe the 24% and 2% of total fatty acids LA levels in the diets. However, we recognize that LA requirements are controversial and point the reader to the methods of Igarashi et al. ( 17 ) as well as other papers for more details ( 24,25 ). The compositions of both diets are shown in ( Table 1 ). As in previous studies (15)(16)(17), the n-6 PUFA-adequate diet contained saffl ower oil (32.3 g/kg), hydrogenated soybean oil (5 g/kg), and coconut oil (55 g/kg). The n-6 PUFA-deprived diet did not contain saffl ower oil, a significant source of LA. Instead, it contained hydrogenated coconut oil (87.3 g/kg) and olive oil (5 g/kg). Both diets had equal amounts of fl axseed oil (7.7 g/kg).
Total lipids were extracted from ‫ف‬ 0.5 g of each diet (n = 3) and analyzed by gas chromatography with fl ame-ionization detection (GC-FID) as described below. Resultant fatty acid concentrations are shown in ( Table 2 ). LA accounted for ‫ف‬ 24% of total where there is a decrease in the activity of the DHAmetabolizing enzymes, calcium-independent phospholipase A 2 (iPLA 2 ) and cyclooxygenase (COX) 1, and an increase in the activity of the ARA-selective enzymes calcium-dependent cytosolic phospholipase A 2 (cPLA 2 ) and secretory phospholipase A 2 as well as COX-2 ( 12,13 ). Lower concentrations of DHA in the brains of n-3 PUFA-deprived rats are likely due to a reduction in the incorporation of DHA into brain phospholipids, as well as a reduction in DHA recycling ( 14 ).
Compared with dietary n-3 PUFA manipulation, few studies have examined dietary n-6 PUFAs and the brain. Reducing the percentage of total fatty acid from 28% LA to 2% LA decreases brain ARA by 28% and increases DHA by 11% ( 15 ). Dietary n-6 PUFA deprivation also reduces expression of ARA-selective enzymes (cPLA 2 and COX-2) and increases expression of DHA-selective enzymes [iPLA 2 and 15-lipoxygenase (LOX)] ( 16 ). There was a corresponding decrease in the protein levels of activator protein (AP)-2 ␣ and nuclear factor B p65 (transcription factors for cPLA 2 and COX-2), as well as an increase in the levels of sterol-regulatory element binding protein 1 (an iPLA 2 transcription factor) ( 16 ). Furthermore, the DHA uptake rate into brain phospholipids is increased by 45% in rats fed an n-6 PUFA-deprived diet, which could account for the increase in phospholipid DHA ( 17 ). In the same study, the turnover of DHA in brain total phospholipids was increased 30-84% in the choline glycerophospholipid (ChoGpl), phosphatidylinositol (PtdIns), and phosphatidylserine (PtdSer) fractions, but not the ethanolamine glycerophospholipid (EtnGpl) fraction.
As of yet, no study has investigated the effect of n-6 PUFA deprivation on the rate of loss of ARA and DHA in brain phospholipids. The objective of this study was to determine the rate of loss of both ARA and DHA from brain phospholipids in rats that are fed either the adequate or deprived n-6 PUFA diet as used in previous studies (15)(16)(17). We hypothesized that upon chronic low n-6 PUFA consumption, ARA would be conserved while DHA would be lost at a more rapid rate. To test this hypothesis, rats consumed either an adequate or a deprived n-6 PUFA diet for 15 weeks postweaning, after which they were infused with either 3 H-ARA or 3 H-DHA via an intracerebroventricular infusion. At set time points post intracerebroventricular infusion, rats were euthanized and the radioactivity of their brains was measured to plot a curve depicting the loss of radioactive ARA or DHA over time. After a logarithmic transformation of the curves, a linear regression analysis was done, and the regression slopes were used to calculate the ARA and DHA half-lives in brain phospholipids ( t 1/2 ), as well as the rate of loss of ARA and DHA on a molar basis ( J out ). ARA was found to be conserved in the n-6 PUFA-deprived rats, while DHA appeared to be lost more rapidly from the brain phospholipids.

Animals
All procedures were approved by the Animal Ethics Committee at the University of Toronto (Protocol # 20010100), in accordance The only difference between the two diets is the amount of hydrogenated coconut oil, olive oil, saffl ower oil, and hydrogenated soybean oil. The saffl ower oil and soybean oil in the n-6 PUFA-adequate diet were replaced by hydrogenated coconut oil and olive oil. lateral ventricle of the brain (+1.5 mm lateral/medial, Ϫ 1 mm anterior/posterior, and Ϫ 4 mm dorsal/ventral to bregma). After the infusion, the needle was slowly removed, and the hole in the skull was sealed with cranioplastic cement (Stoelting). The incision was closed with self-dissolving sutures. For recovery, rats were placed under a heating lamp for 20-30 min before being returned to their respective cages. Rats received a second dose of ketoprofen 1 day postsurgery. For the rest of the study, rats continued to consume their respective diets.

Collection of radioactive brains
Rats were euthanized by microwave irradiation (13 kW for 1.6 s) at the following time points: 4, 16, 32, 64, and 128 days post intracerebroventricular infusion (supplementary Fig. 1). Four n-6 PUFA-deprived rats and 4 n-6 PUFA-adequate rats were euthanized at each time point from both the 3 H-ARA-infused and 3 H-DHA-infused groups. Brains were removed and stored at Ϫ 80°C.

Extraction and isolation of brain phospholipids
Brains were homogenized, and their total lipids were extracted by the chloroform-methanol-0.88% KCl (2:1:0.75) Folch, Lees, and Stanley method ( 27 ). TLC was used to isolate the total phospholipids, as well as the phospholipid classes from the total lipid extract as previously described ( 11,26 ). TLC plates were washed in chlorofrom-methanol (2:1 by volume) and were activated for 1 h at 100°C. TLC G-plates (EMD Chemical, Gibbstown, NJ) in a heptane-diethyl ether-glacial acetic acid solution system (60:40:2 ml by volume) were used for neutral lipid separation. TLC Hplates (Analtech, Newark, DE) in a chloroform-methanol-2propanol-0.25% KCl-triethylamine (30:9:25:6:18 by volume) system were used to separate the phospholipid classes. The plates were sprayed with 0.1% 8-anilo-1-naphthalene sulfonic acid for UV visualization of the fatty acid bands. The bands containing total phospholipids, and the phospholipids fractions (ChoGpl, EtnGpl, PtdSer, and PtdIns) were collected into tubes. For GC-FID analysis, a known amount of heptadecanoic acid (17:0) standard was added. In preparation for both GC-FID and HPLC analysis, the fatty acids were converted into fatty acid methyl esters (FAMEs) by treatment with 14% boron trifl uoride-methanol at 100°C for 1 h.

Quantitation of baseline brain phospholipid fatty acids by GC-FID
FAMEs were analyzed by a Varian-430 gas chromatograph (Varian, Lake Forest, CA) with an FID and a Varian FactorFour capillary column (VF-23ms; 30 m × 0.25 mm inner diameter × 0.25 m fi lm thickness). The FAMEs were dissolved in hexane and injected in splitless mode. The injector and detector ports were set at 250°C. The FAMEs were eluted with increasing temperatures. The temperature program started at 50°C for 2 min, increased 20°C/min, held at 170°C for 1 min, increased 3°C/min, and fi nally, held at 212°C for 5 min. The helium carrier gas had a fl ow rate of 0.7 ml/min. Output peaks were identifi ed using known retention times of authentic FAME standards (Nu-Chek Prep Inc., Elysian, MN). Fatty acid concentrations were calculated by comparison of the GC fatty acid peak areas to the internal 17:0 standard peak area ( 11,26,28 ).

Quantitation of baseline brain eicosanoids and docosanoids
As described previously ( 29 ), composite standards of lipid metabolites (natural or deuterated; Cayman Chemicals Co., Ann Arbor, MI) were diluted in ethanol from stock solutions to perform an eight-point calibration curve (0.05 to 5 ng). Internal standard fatty acids in the n-6 PUFA-adequate diet, and only 2% of total fatty acids in the n-6 PUFA-deprived diet. Approximately 4% of total fatty acids were ALA in both diets. There were negligible amounts of both ARA and DHA (<0.05%). 5,6,8,9,11,12,14,  To create each infusate, the tracer was dissolved in a 5 mM HEPES buffer solution (pH 7.4) with a ratio of tracer molecules to fatty acid-free BSA that was larger than 3:2. The mixture was sonicated and stored at Ϫ 80°C ( 11 ). Tracer purity was confi rmed by HPLC and liquid scintillation counting (LSC).

Calculations and statistics
The data were expressed as means ± SEM. Differences in body weights and food intake between the two dietary groups were assessed using repeated-measures ANOVA (SigmaPlot 12.5; SigmaPlot Software, San Jose, CA). Curves depicting the loss of radioactivity over time post intracerebroventricular infusion were logarithmically transformed and fi t with linear regression. The slopes of these linear regressions were tested using an ANOVA to determine whether they were signifi cantly different from zero as well as if they differed between the adequate and deprived n-6 PUFA groups (GraphPad Prism 5; GraphPad Software, La Jolla, CA). Statistical signifi cance was taken at P < 0.05. Loss half-lives of 3 H-ARA and 3 H-DHA were calculated from the slopes of the linear regressions using equation 1, and the rate of loss ( J out ) in nmol/g brain/day was calculated using equation 2 ( 10, 11, 32 ): where C FA is the baseline brain phospholipid concentration of the fatty acid of interest (ARA or DHA). For the purpose of calculating J out , the half-life that was calculated from the slope of the linear regression was treated as a constant and applied to all measurements of baseline brain phospholipid concentrations (n = 8 baseline measures per dietary group). This produced a distribution of J out values, for which we calculated the mean and SEM. Differences in J out between the deprived and adequate n-6 PUFA rats were assessed using the Student's t -test.

Body weights and food intake
Body weights increased over time as expected (supplementary Fig. 2). Although there were some signifi cant differences between body weights of adequate and n-6 PUFA-deprived rats at certain points in time ( P < 0.05), the magnitude of these differences were small as shown previously by others ( 15 ). After 15 weeks of feeding, in the 3 H-ARA infusion group, the n-6 PUFA-adequate and n-6 PUFA-deprived rats had a mean weight of 363 ± 2.9 g and 356 ± 2.2 g, respectively. In the 3 H-DHA infusion group, the n-6 PUFA-adequate rats had a mean weight of 347 ± 3.7 g and the n-6 PUFA-deprived rats had a mean weight of 344 ± 3.1 g.
Food intake, like body weight, was largely the same between both dietary groups (supplementary Fig. 3). The few signifi cant differences at specifi c time points were small, and the pattern of food intake differences did not match the pattern of body weight differences. The mean weights of the adequate and deprived n-6 PUFA rat brains after 15 weeks of feeding were 1.7 ± 0.006 g and 1.7 ± 0.02 g, respectively ( P > 0.05).

Radiotracer identifi cation
Brain samples (4 days postinfusion) were analyzed by HPLC and LSC to confi rm radiotracer identity. ARA elutes at 35 min ( 11 ). All the radioactivity in the 3 H-ARA-infused rat mixtures in ethanol were added to both the composite standards and the samples prior to extraction. Extraction and sample preparations were performed in siliconized glassware. To minimize autooxidation, fatty acids were extracted on ice, in a reduced light condition, using solvents that contained 0.1% butylated hydroxyltoluene. The frozen brain halves were homogenized in methanol. One nanogram of internal standard mixture was added to a 250 mg aliquot of each homogenized brain. External ARA, EPA (20:5n-3), and DHA standards were prepared in a similar way. The samples were mixed for 1 min, incubated on ice for 30 min, and centrifuged at 1,000 g for 10 min. The supernatants were collected. The pellet was resuspended in ethanol for 1 min and centrifuged again for a second extraction. The resultant ethanolic supernatants were combined with the methanolic supernatants, previously extracted. After evaporation with nitrogen gas, the supernatants were suspended in 10% ethanol, acidifi ed to pH 3 with 1 N HCl, and triply extracted with ethyl acetate. The ethyl acetate layer was washed to neutrality with water and dried under nitrogen gas. The residues from the brain and external standard samples were reconstituted in acetonitrile-water (1:1 by volume) and transferred into the inserts of amber vials for immediate LC/MS/MS analysis. LC/ MS/MS was performed using a 1290 UHPLC System (Agilent Technologies, Santa Clara, CA) and a QTRAP5500 Mass Spectrometer (ABSciex, Framingham, MA). The chromatography was done at a 600 l/min fl ow rate on a Zorbax SB-Phenyl column (Agilent Technologies; 3.0 × 50 mm, 3.5 m). The gradient started at 80% water and, over 9 min, ramped up to 100% acetonitrile. The mass spectrometer was operated in negative electrospray ionization mode with a source temperature setting of 600°C and a voltage setting of 4,500 V. The precursors to product ion mass transitions were obtained through scheduled multiple reaction monitoring. Quantitative analysis was performed by Analyst 1.5.2 Software (AB-Sciex). The area ratios of the integrated peaks (natural to deuterated standard) were plotted against the standard curves for quantifi cation. The limit of quantifi cation was 0.025 ng per sample, and values between 0.005 ng and 0.025 ng were considered semiquantitative.

Confi rmation of radiotracer identity by HPLC
Total phospholipids were extracted from brain homogenate and methylated as described above. Samples were reconstituted in acetonitrile. As described in previous studies ( 11,26,28,30 ), FAMEs were separated by HPLC (Waters 2690, Boston, MA) with a Luna C18 reverse column (4.6 × 250 mm, 100 Ǻ; Phenomenex, Torrance, CA) and an in-line UV photodiode array detector (Waters 996) set at a 242 nm wavelength. The system was fi rst stabilized at a 1 ml/min fl ow rate with a gradient system consisting of i ) 100% water and ii ) 100% acetonitrile. The gradient was then set to 85% (ii) for 30 min, and then increased to 100% (ii) over 10 min. It was held there for 20 min before returning back to 85% (ii) over a 5 min period. Fractions were collected at 1 min intervals for 55 min, and each of the 55 fractions was measured for radioactivity by LSC. Similar to what has previously been reported, ARA and DHA had elution times of 35 and 31 min, respectively ( 11,31 ).

Quantifi cation of radioactivity by LSC
LSC was used to measure the radioactivity of the total phospholipids, fractions, as well as 4-day brain phospholipid HPLC fractions. Samples were put into scintillation vials, and 5 ml of scintillation cocktail (GE Healthcare, Life Sciences, Baie d'Urfe, QC, Canada) was added. Radioactivity was quantifi ed using a Packard TRI-CARB2900TR liquid scintillation analyzer (Packer, Meriden, CT) with a detector effi ciency of 61.07% for tritium. The measurements were given in disintegrations per minute and were converted to nCi/brain ( 11,26 ).

Baseline brain phospholipid fatty acid concentrations
The changes in total phospholipid concentrations were refl ected by changes in the phospholipid fractions, although some fractions changed more than others for certain fatty acids. ARA was ‫ف‬ 20% lower ( P < 0.05) in the ChoGpl, EtnGpl, and PtdSer fractions but not the PtdIns fraction of the n-6 PUFA-deprived rats ( Table 4 ). DPAn-6 brain phospholipids eluted at 35 min and was identifi ed as ARA in both the adequate and deprived n-6 PUFA rats ( Fig.  1A ). DHA elutes at 31 min. All of the radioactivity in the 3 H-DHA-infused rat brain phospholipids eluted at 31 min and was identifi ed as DHA in both dietary groups ( 33 ) ( Fig. 1B ).

Brain phospholipid ARA and DHA rate of loss
The loss of 3 H-ARA and 3 H-DHA from brain phospholipids was plotted from 4 to 128 days postinfusion and then logarithmically transformed for linear regression analysis ( Figs. 3 , 4 ). All slopes were negative and signifi cantly different from zero ( P < 0.0001) ( Table 5 ). Linear regression diets used in this study were similar to those used in previous studies examining body and organ weights, enzyme expression, and brain DHA uptake and turnover (15)(16)(17). In brain total phospholipids, there was a 15% reduction in ARA concentration and an 18% increase in DHA concentration with the n-6 PUFA-deprived diet. These changes are comparable to the 28% reduction in ARA and the 11% increase in DHA found previously in brain total lipids ( 15 ). For the fi rst time, we show that lowering the amount of n-6 PUFA in the diet leads to longer ARA loss half-lives in brain total phospholipids, ChoGpl, EtnGpl, and PtdIns pools. After factoring in the concentration of ARA in the baseline, uninfused rats, the n-6 PUFA-deprived rats had a slower net rate of loss (a lower J out ) of ARA in the total phospholipids, ChoGpl, EtnGpl, and PtdSer pools. This may have been caused by a decrease in cPLA 2 and COX-2 activity, which has been reported in rats upon 15 weeks consumption of a low n-6 PUFA diet ( 16 ). It would have also been interesting to calculate the half-lives of ARA and DHA within brain neutral lipids as fatty acid turnover within these lipids is PLA 2 independent, and future studies should consider this. Decreased cPLA 2 activity likely refl ects a decreased ARA turnover and less opportunity for ARA to be lost through eicosanoid production or ␤oxidation. The decrease in eicosanoid production shown here may not be enough to account for the decreased loss of ARA, unless the eicosanoids have very short half-lives. Thus, there is also likely a decreased amount of ␤ -oxidation or other catabolic processes, which is an area requiring further research.
In contrast to our observations with ARA, there was a more rapid net loss (a higher J out ) of DHA from the brain total phospholipids and the ChoGpl pool in the n-6 PUFAdeprived rats, but not from the other phospholipid fractions. This difference in the rate of loss of DHA was a function of the higher brain DHA concentrations in the deprived versus adequate n-6 PUFA rats because the fractional losses and the loss half-lives for DHA were not signifi cantly different between the two dietary groups in any of the phospholipid pools. In the n-6 PUFA-deprived rats, the J out of DHA for total phospholipids was smaller than the J in (net rate of incorporation) of DHA found previously, whereas the J in approximately matched the J out in the n-6 PUFA-adequate rats ( 17 ). A more rapid daily uptake rate compared with loss rate of DHA may account for the higher concentration of DHA in the n-6 PUFA-deprived rats than in the n-6 PUFA-adequate rats upon 15 weeks of feeding. However, caution should be taken when comparing and combining results for kinetic analyses from two different studies, and future experiments should be completed under similar conditions. One important distinction between our intracerebroventricular method to calculate phospholipid half-lives and the use of a pulse intravenous infusion can be seen in glycerophospholipid species. While the half-life or the net rate of entry ( J in ), as calculated upon an intravenous pulse infusion, approximates the half-life or J out for brain total phospholipids, this does not hold true for measured glycerophospholipid species. One reason for the discrepancy in glycerophospholipid of brain/day in the n-6 PUFA-deprived group ( P < 0.05). This corresponds to a daily fractional loss of 2.2% versus 1.6% in the adequate versus deprived groups. The J out (nmol/g of brain/day) in the phospholipid fractions ranged from 6.9 ± 0.06 (PtdSer) to 43 ± 0.6 (ChoGpl) in the n-6 PUFA-adequate group, and from 4.0 ± 0.1 (PtdSer) to 24 ± 0.3 (ChoGpl) in the n-6 PUFA-deprived group. Overall, ARA was lost from brain phospholipids at a slower rate in the n-6 PUFA-deprived group.
There were no differences in the slopes and thus, no difference in the loss half-lives for 3 H-DHA between the adequate and deprived n-6 PUFA rats. However, due to the higher baseline DHA concentrations in the deprived rats, the n-6 PUFA-deprived group appeared to have a higher J out for DHA than the adequate n-6 PUFA group. This difference was only signifi cant in the total phospholipid pool and the ChoGpl fraction. The J out for total phospholipids in the adequate and n-6 PUFA-deprived groups was 129.1 ± 11.4 nmol/g of brain/day and 145.7 ± 15.5 nmol/g of brain/ day, respectively ( P < 0.05). The J out (nmol/g of brain/day) for the phospholipid fractions ranged from 2.3 ± 0.3 (Pt-dIns) to 113.5 ± 7.8 (EtnGpl) in the n-6 PUFA-adequate group, and from 2.4 ± 0.3 (PtdIns) to 121 ± 10.2 (EtnGpl) in the n-6 PUFA-deprived group. Overall, DHA seemed to be lost, when measured in terms of J out , from brain phospholipids at a more rapid rate in the n-6 PUFA-deprived rats.

DISCUSSION
This study investigated the effect of an adequate (24% LA) versus deprived (2% LA) n-6 PUFA diet on the rate of loss of ARA and DHA from rat brain phospholipids. The Brain fatty acid concentrations in phospholipid fractions of baseline uninfused rats after 15 weeks of feeding (n = 8 per dietary group). Data are means ± SEM. a Indicates signifi cant difference from n-6 PUFA adequate group ( P < 0.05). must be taken with this in mind. Nonetheless, the fi nding is consistent with the signifi cantly higher DHA levels in the brain phospholipids of the n-6 PUFA-deprived rats. It should also be noted that intracerebroventricular administration of fatty acids can lead to neuroinfl ammation and damage to the blood-brain barrier, which could alter our results. To minimize the effect of intracerebroventricular administration, similar to our previous experiments ( 11,26 ), we used a 33-gauge needle and allowed the rats to recover for 4 days before beginning our analyses.
One of the fates of unesterifi ed ARA and DHA is metabolism into eicosanoids and docosanoids. We reported decreases in the levels of certain ARA-derived eicosanoids in rats fed the deprived versus adequate n-6 PUFA diet. The largest reduction was seen in the enzymatically derived PGF 2 ␣ . As seen in previous studies, an n-6 PUFAdeprived diet causes a reduction in the activity of cPLA 2 and COX-2, which are both enzymes required for the synthesis of PGF 2 ␣ ( 16 ). However, this effect was selective for PGF 2 ␣ because there were no signifi cant reductions in the levels of the other nonautooxidative eicosanoids that were detected (PGE 2 and PGI 2 ). This suggests that the reductions in enzyme activity can selectively reduce certain eicosanoids but not others. The mechanism behind this is unknown. Furthermore, it could be that the differences in the concentration of other eicosanoids may only appear when the brain is responding to stress. For example, in a neuroinfl ammatory state, there is an increased production of PGE 2 ( 5 ). Perhaps, in such a state, the n-6 PUFA-adequate rats would have a larger increase in eicosanoid production than the n-6 PUFA-deprived rats. The remaining eicosanoids that changed are thought to be autooxidative, meaning that they are capable of being species half-lives is likely due to remodeling/exchange of radiolabeled fatty acids between phospholipid species that occurs over time, which does not occur upon acute pulse labeling ( 34 ). However, it is also possible that fatty acids other than the plasma unesterifi ed pool enter the brain, which would be captured in our study.
Similar to previous studies, one limitation of this study was that measuring the radioactivity of the rat brains required euthanization of the rats and removal of their brains. Thus, different rats had to be used at each time point postinfusion, and the actual loss of radioactivity in the each individual rat brain could not be tracked over 128 days. This posed a problem when trying to calculate the SEM of J out because the J out calculation was the quotient of two measurements: the baseline ARA or DHA concentration and the loss half-life of ARA or DHA. Only the baseline ARA or DHA concentration measurements had discernable sample sizes. Thus, consistent with past literature, the calculated loss half-lives were treated as constants and were applied to all baseline concentration values in order to create a distribution of J out values from which the J out SEM could be calculated ( 10,11 ). It is noteworthy, however, that incorporation of the slope SEM would result in more conservative analyses of the difference between J out values of the adequate and deprived n-6 PUFA rats. For ARA, this assumption does not change the overall conclusion, as the differences in ARA J out between the dietary groups are signifi cant up to a very high SEM. For DHA, however, an incorporation of larger slope error would likely result in no signifi cant differences between the DHA J out values of the n-6 PUFA-adequate and n-6 PUFA-deprived rats in any of the phospholipids. Thus, the conclusions made about DHA metabolism the n-6 PUFA-deprived rats. Brain phospholipid ARA concentrations were 6,077 ± 103 nmol/g of brain and 5,191 ± 158 nmol/g of brain, respectively. This raises an interesting question: why does a reduction to 5,191 nmol/g of brain lead to an ‫ف‬ 0.0006 nmol/g of brain reduction in 5-HETE when the amount of ARA available is still >2 million times the amount needed to supply the 0.0019 nmol of 5-HETE in the n-6 PUFA-adequate rats? Perhaps this highlights how important phospholipid ARA is in its other roles aside from eicosanoid production or that ARA is regulating the enzymatic synthesis of the eicosanoids. It may also be nonenzymatically derived ( 29 ). This suggests that these eicosanoid reductions could be due to the lower ARA concentration in the brain phospholipids of n-6 PUFAdeprived rats. One thing to note is that in the brain, eicosanoids are present in the fentomole to picomole per gram range, whereas concentrations of ARA in brain phospholipids are just over a micromole per gram. The most concentrated eicosanoid detected was 5-HETE, which had a concentration of 0.0019 ± 0.00001 nmol/g of brain in the n-6 PUFAadequate rats and 0.0013 ± 0.000007 nmol/g of brain in  In rats, the level of LA at which tissue ARA concentrations plateau is 1,200 mg of LA per 100 g of diet, and this intake is considered the recommended minimum LA intake ( 25 ). This level is comparable to the recommended level of LA for humans: 1,000-1,500 mg of LA per 100 g of food, which equates to ‫ف‬ 2-3% of energy (42)(43)(44). However, more recent studies suggest that these are overestimations of the actual LA requirement, and there is a concern that the amount of LA in the average human diet is too high ( 45 ). For rats, the level at which tissue ARA levels plateau cannot necessarily be substituted for the LA requirement level, as there is no evidence that ARA at this peak level is critical for health. For instance, the n-6 PUFAdeprived diet used in our study and in previous studies had a LA level that was 10% of the suggested requirement, and rats did not show signifi cant signs of LA defi ciency ( 15 ). One of the problems with earlier experiments that formed the basis of LA requirements is that many of them did not ensure an adequate level of ALA (45)(46)(47)(48)(49). The effects of an inadequate level of ALA were attributed to having an inadequate level of LA, and thus may have led to an overestimation of LA requirement. The n-6 PUFAdeprived diet in our study contained an adequate amount of ALA, allowing for the effects here to be attributed to changes in dietary n-6 PUFAs, without the infl uence of dietary n-3 PUFAs.

CONCLUSION
In summary, rats fed an n-6 PUFA-deprived diet (2% LA) or an n-6 PUFA-adequate diet (24% LA) for 15 weeks were infused with 3 H-ARA or 3 H-DHA, which allowed for the determination of the rate of loss of ARA and DHA that there is a set rate at which ARA autooxidation occurs, which cannot be altered to compensate for alterations in brain phospholipid ARA concentration.
There was a signifi cantly higher level of all detectable EPA-derived eicosanoids in the n-6 PUFA-deprived rats, except for 12-HEPE. These EPA-derived eicosanoids were all autooxidation products, and the increase we observed may be due to the 10-fold increase in brain total phospholipid EPA concentration in the n-6 PUFA-deprived rats. In our study, and as reported by others, the brain maintains very low levels of EPA ( 26,(36)(37)(38). Even supplementation of EPA in the diet does not create such large increases in EPA. For example, in fi sh a 4-fold increase in the EPA content of the diet only results in a 70% increase in brain EPA concentration ( 39 ). Recent meta-analyses suggest that EPA may be protective in major depression ( 40,41 ). Thus, being able to raise EPA levels 10-fold by lowering n-6 PUFA may be therapeutic in certain mood disorders, including major depression, and should be further investigated.
There were no changes in the levels of docosanoids, though the only four docosanoids detected were also considered autooxidative ( 29 ). It is interesting here how even though the concentration of brain phospholipid DHA increased by ‫ف‬ 1,000 nmol/g of brain in the n-6 PUFAdeprived rats, there was no change in the level of these autooxidative docosanoids, unlike what was seen with the eicosanoids. Perhaps the extra DHA is largely being shunted toward increasing the levels of enzymatically derived docosanoids, which, similar to others, we were unable to detect in microwave-fi xed rat brains ( 29 ). Previous studies using similar diets have reported that an n-6 PUFAdeprived diet increases the activity of iPLA 2 and expression of 15-LOX, which are enzymes that metabolize DHA into resolvins, protectins, and maresins ( 16 ). Loss half-lives were calculated from the slopes of the linear regressions using the equation t 1/2 = log 10 2/(slope of the regression line). Baseline fatty acid concentrations were used to calculate the rates of loss ( J out ) from the different fatty acid pools using the equation J out = 0.693 C FA / t 1/2 , where C FA is the baseline brain total phospholipids and fractions fatty acid concentration after 15 weeks of feeding. Data are mean ± SEM. Differences in slope and J out were assessed with an ANOVA and Student's t -test, respectively. a P < 0.05 versus adequate n-6 PUFA.
from their brain phospholipids. ARA had a longer half-life and was lost at a slower rate in the total phospholipids, ChoGpl, EtnGpl, and PtdSer pools of the n-6 PUFAdeprived rats, illustrating the brain's ability to conserve ARA in response to lower dietary LA. In contrast, DHA was lost more rapidly in the total phospholipids and ChoGpl pools of the n-6 PUFA-deprived rats, but not the other phospholipids fractions. These effects are approximately opposite of what was observed with the low versus high n-3 PUFA diets in previous studies where low n-3 PUFA leads to conservation of DHA but not ARA ( 10,11 ). Recently, the Nurses' Health Study reported that an elevated intake of LA was associated with an increased risk of developing major depression ( 50 ), and preclinical studies suggest that drugs used to manage bipolar disorder decrease brain arachidonic acid metabolism ( 51,52 ). While much interest in dietary LA levels have focused around coronary heart disease risk, understanding how dietary n-6 PUFAs regulate the metabolism of ARA and DHA in the brain may also be important for determining dietary requirements.