Plasma non-esterified docosahexaenoic acid is the major pool supplying the brain

Despite being critical for normal brain function, the pools that supply docosahexaenoic acid (DHA) to the brain are not agreed upon. Using multiple kinetic models in free-living adult rats, we first demonstrate that DHA uptake from the plasma non-esterified fatty acid (NEFA) pool predicts brain uptake of DHA upon oral administration, which enters the plasma NEFA pool as well as multiple plasma esterified pools. The rate of DHA loss by the brain is similar to the uptake from the plasma NEFA pool. Furthermore, upon acute iv administration, although more radiolabeled lysophosphatidylcholine (LPC)-DHA enters the brain than NEFA-DHA, this is due to the longer plasma half-life and exposure to the brain. Direct comparison of the uptake rate of LPC-DHA and NEFA-DHA demonstrates that uptake of NEFA-DHA into the brain is 10-fold greater than LPC-DHA. In conclusion, plasma NEFA-DHA is the major plasma pool supplying the brain.

Brain DHA uptake. To examine the rate of uptake of NEFA-DHA into the brain, albumin-bound NEFA-14 C-DHA was continuously infused into the jugular vein of rats over a 5-minute period while blood samples were collected from the carotid artery (Fig. 1a). The 5-minute continuous infusion captured the appearance of NEFA-14 C-DHA in the plasma (Fig. 1b). Measures for the uptake coefficient (k * ) and rate of NEFA-DHA incorporation (J in ) were 0.20 ± 0.02 μ l/s/g and 80 ± 10 nmol/g/day, respectively (Eqs 4 and 5). To determine the relative contribution of esterified and NEFA-DHA to total brain DHA uptake, NEFA-14 C-DHA was gavaged (Fig. 1c) to rats resulting in the appearance of 14 C-DHA in several plasma esterified pools as well as the plasma NEFA pool ( Fig. 1d; Supplementary Fig. 2 for confirmation of tracer identity. Similar to others, we observed that DHA made up greater than 99% of the radioactivity 16 ). By continuous sampling from the carotid artery, it was determined that 14 C-DHA appeared largely in triacylglycerol (TAG; 84% of total plasma radiotracer) and relatively lower amounts were found esterified in phospholipids (total phospholipids excluding LPC; 4.7%), diacylglycerol (DAG; 2.9%), cholesteryl esters (CE; 0.4%) and lysophosphatidylcholine (LPC; 0.13%). NEFA-14 C-DHA was the second most abundant radiolabeled lipid pool in the plasma after NEFA-14 C-DHA gavage accounting for 7.6% of total plasma radiotracer area under the curve (Fig. 1d). Radiolabeled LPC-DHA did not appear in the plasma until 3 hours post-ingestion of NEFA-14 C-DHA which is consistent with it being secreted from the liver and not upon intestinal absorption (Fig. 1d). Using the measured k * for NEFA-DHA (Fig. 1f), it was determined that the uptake of esterified DHA (k * ES ) was − 0.0089 ± 0.002 μ l/s/g suggesting that esterified DHA including LPC-DHA is not, quantitatively, a major source for supplying the brain with DHA (Equations 1-3; Supplementary Table 1 for individual phospholipid species). Therefore, if it is assumed that only the radiolabeled DHA in the plasma NEFA pool post-gavage enters the brain, this leads to a k * of 0.098 ± 0.05 μ l/s/g and a net rate of incorporation (J in ) of 32 ± 9 nmol/g/day (Fig. 1f,h Eqs 4 and 5). Collectively, we find that the k * and J in upon a 5-min continuous infusion are slightly higher than would be predicted from the plasma NEFA pool upon oral gavage of the NEFA-14 C-DHA tracer (Fig. 1f,g). One reason for this might be error in estimating the AUC between the time points over a 4 hour study and due to acute catabolism of the tracer upon its entry into the brain during the 4 hours which has been estimated to be as much as 20% for PUFA 23 . While the mechanism for the decrease in PUFA radioactivity is not clear, it is possible that a proportion of the tracer enters an unstable pool within the brain.
To examine whether these rates of brain DHA uptake were similar to rates of DHA loss, the rate of loss of DHA from brain phospholipids was determined by administering NEFA-14 C-DHA Figure 1. NEFA-DHA uptake by the brain. (a) Schematic diagram of kinetic modeling following continuous intravenous (iv) infusion. (b) Plasma lipid radioactivity upon continuous iv infusion of 86 μ Ci NEFA-14 C-DHA over 5 minutes (n = 6). Radioactivity from other plasma lipid pools was below detection, 23 pCi. (c) Schematic diagram of kinetic modeling following NEFA-14 C-DHA gavage. (d) Plasma lipid radioactivity upon NEFA-14 C-DHA oral gavage over 4 hours (n = 4) and the percent composition of total radiolabeled lipid pools in the plasma post-gavage of NEFA-14 C-DHA. (e) Brain radioactivity of total phospholipids from rats gavaged or continuously iv infused with NEFA-14 C-DHA. (f) The uptake incorporation coefficient of NEFA-14 C-DHA (k * ) into brain total phospholipids from rat gavage or continuous iv infusion. (g) The plasma concentration of NEFA-DHA prior to NEFA-14 C-DHA gavage or continuous iv infusion. (h) The net incorporation rate of DHA (J in ) to brain total phospholipids from rats gavaged or iv infused with NEFA-14 C-DHA. (i) Schematic diagram of kinetic modeling following intracerebroventricular (icv) infusion. (j) Brain loss curve of esterified 14 C-DHA from brain total phospholipids including lysophosphatidylcholine over 128 days (n = 3-4 independent samples per time point). (k) The loss of DHA from brain total phospholipids (rate of loss, J out ) from rats infused icv with NEFA-14 C-DHA which was not significantly different from the J in of rats continuously iv infused with NEFA-14 C-DHA. (l) Calculated predictive k * from the NEFA-14 C-DHA icv infusion which was not significantly different from k * from NEFA-14 C-DHA iv infusion. All data are expressed as mean ± SEM; whereas percent plasma radioactivity data are expressed as mean. * P < 0.05 and ** P < 0.01 indicate significant difference from gavage using Students' t-test. AUC, area under the curve; CE, cholesteryl ester; DAG, diacylglycerol; LPC, lysophosphatidylcholine; NEFA, non-esterified fatty acid; PL, phospholipids; TAG, triacylglycerol.
intracerebroventricularly and calculating the slope of DHA loss from total phospholipids (Fig. 1i, Supplementary Fig. 3 for individual phospholipid species). This measured loss half-life (t 1/2 ) of 64 ± 7 days (Equation 6) translated to rate of loss from brain phospholipids (J out ) of 65 ± 7 nmol/g/day (Fig. 1k,  Equation 7) and a predictive k * of 0.17 ± 0.02 μ l/s/g (Fig. 1l, Equation 8). The t 1/2 and J out are in broad agreement with previous reports by DeMar et al. and Lin et al., with minor discrepancies possibly related to differences to animal ages, diets and strain 16,26 . Therefore, it is important that comparisons of fatty acid kinetic parameters from various models be from studies conducted under similar conditions. The k * and J in from the continuous iv infusion are not significantly different from the predictive k * and J out again suggesting that plasma NEFA-DHA can replace brain DHA loss to maintain homeostatic levels of brain DHA. The results predict that plasma NEFA-DHA is the major contributor to brain DHA in vivo.
The relative partitioning of LPC-DHA into the brain compared with NEFA-DHA. Earlier work demonstrated that upon iv infusion of a bolus of radiolabeled NEFA-DHA or LPC-DHA, that LPC-DHA preferentially targeted the brain 19,20 . Recently, with the discovery of a putative LPC-DHA transporter, and using similar experiments, it was concluded that LPC-DHA represents the major mechanism by which DHA enters the brain 22 . However, the results presented above (Fig. 1), are not compatible with the hypothesis that LPC-DHA is the major plasma pool supplying the brain. Thus, we set out to reproduce the finding of higher brain radioactivity following iv LPC-14 C-DHA infusion. This time, however, the amount of DHA exposure to the brain was measured, which was not considered in the previous report. Similar to other reports 19,22 , a bolus iv infusion of radiolabeled NEFA-DHA or LPC-DHA was used and brains were collected 2 hours post-infusion (Fig. 2a). Consistent with the short half-life of plasma NEFA of approximately 30 seconds, NEFA-14 C-DHA decreased 207-fold from the peak initial level within 5 minutes (Fig. 2b). In contrast, radiolabeled LPC-DHA was still elevated at 5 minutes and had only decreased 7-fold from the peak. Furthermore, at 30 minutes, the reduction in plasma LPC-14 C-DHA levels was merely 88-fold as compared to a 339-fold reduction in plasma NEFA-14 C-DHA levels, suggesting a much longer half-life for LPC-DHA (Fig. 2c). This translated to higher total relative LPC-14 C-DHA exposure (61.8% of total plasma radiotracer) in rats receiving LPC-14 C-DHA as compared to NEFA- 14 C-DHA exposure (44.9% of total plasma radiotracer) in rats receiving NEFA-14 C-DHA ( Fig. 2d-f). Collectively, the AUC for NEFA-DHA in the iv NEFA-14 C-DHA infused rats was 282 ± 32 nCi*min/ml, while the AUC for plasma LPC-DHA in the iv LPC-14 C-DHA infused rats was 1478 ± 80 nCi*min/ml (Fig. 2g). Similar to other reports, we found that radiolabeled brain DHA was 3-fold higher in iv LPC- 14 C-DHA infused animals compared those receiving NEFA-14 C-DHA (Fig. 2h). However, because the radiolabel appeared in multiple plasma pools over the 2 hours, it is unclear which plasma pool would have supplied the brain with DHA. Nevertheless, we corrected brain radioactivity for either plasma LPC-14 C-DHA or NEFA-14 C-DHA exposure, which was 5.2 fold higher for LPC-DHA than NEFA-DHA (Fig. 2g). An uptake coefficient (k * ) can be calculated, which was significantly higher for NEFA-DHA (0.21 ± 0.05 μ l/s/g) than LPC-DHA (0.12 ± 0.009 μ l/s/g) (Fig. 2i). Finally, to determine the rate of brain DHA uptake, the concentration of plasma NEFA-DHA and LPC-DHA was measured using LC/MS/MS and found that the concentration of LPC-DHA was 64% lower than NEFA-DHA (Fig. 2j). Applying the plasma concentration to the k * for LPC-DHA and NEFA-DHA, the net rate of entry (J in ) into the brain from the plasma NEFA pool (46 ± 10 nmol/g/day) is 5-fold higher than the rate of entry from the plasma LPC-DHA pool (9 ± 0.6 nmol/g/day) (Fig. 2k). Because a small fraction of LPC-DHA is in the form of lipoproteins, this method slightly overestimates albumin-bound plasma LPC-DHA. However, similar to the gavage experiment, if other pools are entering the brain, these rates could be overestimates. Since the k * from the current bolus iv NEFA-DHA infusion study is similar to our results with continuous iv NEFA infusion, it suggests that, at least for this pool, it is not an overestimate.
To circumvent the repackaging of infused radiotracer into multiple plasma pools and to control for exposure, the uptake of NEFA-14 C-DHA and LPC-14 C-DHA was quantified using the continuous 5-minute infusion model (Fig. 3a). Consistent with the shorter half-life of NEFA-DHA, the plasma radioactivity appeared to plateau around 1 min (Fig. 3b), while plasma radioactivity increased during the study for the LPC-DHA (Fig. 3c). At the 5-minute time point, 97% of the total lipid radioactivity in the 14 C-DHA-infused rats was in the plasma NEFA pool and 97% of the total lipid radioactivity in LPC-14 C-DHA-infused rats was in the plasma phospholipid pool. As predicted from the previous experiments, brain radioactivity was significantly higher upon 5-minute LPC-14 C-DHA infusion (Fig. 3d). After correction for plasma exposure, however, the incorporation coefficient (k * ) was significantly higher in the group receiving NEFA-14 C-DHA (0.18 ± 0.04 μ l/s/g) than LPC-14 C-DHA (0.06 ± 0.009 μ l/s/g) (Fig. 3e). The elevated k * in the 14 C-DHA-infused rats was magnified upon correction for the plasma pool sizes (Fig. 3f) as plasma NEFA-DHA pool (J in = 46 ± 11 nmol/g/day) contributes 10-fold more DHA than plasma LPC-DHA (J in = 4.7 ± 1.2 nmol/g/day) (Fig. 3g). Importantly plasma concentrations of NEFA-DHA and LPC-DHA is this and the previous experiments are similar to the normal physiological range reported in the literature 27,28 . The replacement half-life of DHA in brain total phospholipids would be 65 ± 12 days from the plasma NEFA-DHA pool and 676 ± 122 days from the plasma LPC-DHA pool (calculated based on the concentrations from Supplementary Fig. 1). Furthermore, if use the experimentally derived k* (Fig. 3e) for LPC-DHA to predict the amount of brain radioactive DHA upon the oral gavage of NEFA-14 C-DHA (Fig. 1e), this leads to a 148 fold underestimate of the measured radioactivity. The incorporation coefficients and net rate of entry for NEFA-DHA were consistent across all studies for the plasma NEFA-DHA, despite differences in time and dose. This is not surprising as uptake of plasma NEFA-DHA is thought to be via passive diffusion and the models used in this study are time-independent 29-33 .

Discussion
Using several kinetic models, these studies demonstrated that plasma NEFA-DHA is the major plasma pool supplying the brain with DHA in free-living adult rats. The finding that plasma NEFA-DHA is the major plasma pool supplying the brain does not negate earlier important findings of preferential LPC-DHA incorporation, as compared to NEFA-DHA, into the brain 19,20,22 . Similar to the reports of The net incorporation rate of DHA (J in ) into brain total phospholipids from rats iv infused with bolus NEFA-14 C-DHA or LPC-14 C-DHA. Percent plasma radioactivity data are expressed as mean. Remaining data are expressed as mean ± SEM. * P < 0.05, ** P < 0.01 and *** P < 0.001 indicate significant difference from NEFA-14 C-DHA infused rats. + P < 0.01 indicates significant difference between plasma LPC-DHA and NEFA-DHA concentration. # P < 0.001 indicates significant difference between radiolabeled LPC pool from LPC-14 C-DHA infused rats and radiolabeled NEFA-DHA pool from NEFA-14 C-DHA infused rats. Students' t-test was used to determine significant difference. See Fig. 2 for abbreviations. others 20,22,34 , our results suggest that upon a single intravenous bolus dose or infusion of LPC-DHA, more LPC-derived DHA will target the brain as compared to NEFA-DHA. This has therapeutic importance for studies aiming to deliver DHA to the brain and is consistent with preclinical findings regarding the efficacy of LPC-DHA analogues targeting the brain in models of stroke 35 . However, despite this "preferential LPC-DHA uptake", LPC-DHA is not the major mechanism by which DHA enters the brain under normal conditions. LPC-DHA appears to circulate with a longer plasma half-life than NEFA-DHA. When controlling for plasma exposure and/or the steady-state non-labeled "cold" pool size, plasma NEFA-DHA is the major pool supplying the brain with DHA in vivo. It is important to note, that while the NEFA and LPC-DHA tracers disappear from the plasma with different half-lives, the cold concentrations remain at steady-state due to replacement from other pools. Based on results from experiments similar to ours (Fig. 2) and because mice lacking the candidate transporter for LPC-DHA Mfsd2a had deficits in brain DHA levels (~8% in phospholipids) compared to wild-type controls (~20%), Nguyen and colleagues proposed that the major mechanism by which DHA enters the brain is from plasma-derived LPC 22 . The large decrease in brain phospholipid DHA in Mfsd2a knockout mice may appear to be at odds with our finding that LPC-DHA is a relatively minor source of brain DHA, but there are many possible explanations for this apparent discrepancy, including: (a) the low DHA levels in the Mfsd2a knockout are due to the large loss of DHA-rich neuronal cells in the knockout (over 50% loss of Purkinje cells), possibly The net incorporation rate of DHA (J in ) into brain total phospholipids from rats continuously iv infused with NEFA-14 C-DHA or LPC-14 C-DHA. All data are expressed as mean ± SEM. * P < 0.05, ** P < 0.01 and *** P < 0.001 indicate significant difference from NEFA-14 C-DHA infused rats using Students' t-test. + P < 0.01 indicates significant difference between plasma LPC-DHA and NEFA-DHA concentration. See Fig. 2  also explaining why DHA supplementation does not restore brain DHA levels or (b) that the Mfsd2a knockout was life-long and much brain DHA accretion occurs during development which may require LPC-DHA. Furthermore it is worth noting that despite life-long deficiency of Mfsd2a, knockout mice still have 8% DHA which is similar to what is reported during omega-3 deficiency in the mouse, albeit methodological differences may account for this result 36,37 .
The finding that plasma NEFA is the major plasma pool supplying the brain with DHA suggests that the previous studies that have estimated brain DHA uptake to be between 2.4 and 3.8 mg/brain/day, by imaging brain DHA uptake upon infusion of labeled NEFA-DHA in humans, are only slight underestimates of the total brain DHA uptake rate 7,38,39 . While it is known that the plasma NEFA pool is supplied by the adipose and circulating lipoproteins 40 , studies on the source of DHA in the plasma NEFA pool are, to the best of our knowledge, lacking. The DHA content in rat white adipose tissue was recently reported to be 13 mg 41 . Given our calculated rate of uptake of about 20 μ g/day, this would suggest that adipose could supply for the brain for 1.8 years. However, this is an overestimate as DHA released from adipose does not solely supply to the brain and released DHA would be taken up by other tissues. Even though the brain/body partition coefficient for DHA has not been reported for rats, the partition coefficient has been estimated to be 5% for palmitate in rats and 0.5% for DHA in a monkey 23,39 . It is important to note that these might also be underestimates as fatty acids that are taken up by other tissues may be released and taken up by the brain at a later time. Also, a proportion, if not all, of the released DHA from adipose is being replaced from the diet 4 . Using similar calculations, we have recently estimated that adipose in adult humans contains enough DHA (20-50 g) to supply the brain for 14-36 years 42 .
In summary, our results demonstrate that the NEFA pool of DHA is the major source of DHA supplying the rat brain in vivo and confirm that a bolus of plasma LPC-DHA will target the brain more than a bolus of NEFA-DHA. Collectively, our results resolve a more than 20 year-old controversy and will inform studies aiming to deliver or image/quantify DHA entry into the brain which is being examined in stroke 43,44 , traumatic brain injury 8,45 and other neurological diseases 12,38,46,47 . Future studies should examine the effects of diet, genetics, brain injury, disease, development and aging to test if the mechanisms presented here persist. Radiotracer perfusate. Radiolabeled NEFA-14 C-DHA ([1-14 C]-DHA, specific activity: 58 mCi/ mmol, Moravek Biochemical Inc., Brea, CA, USA) were solubilized in olive oil for gavage, and 5 mM HEPES buffer containing 50 mg/ml fatty acid-free bovine serum albumin for intravenous (iv) infusion (continuous and bolus) or 5 mM HEPES buffer containing 100 mg/ml fatty acid-free bovine serum albumin for intracerebroventricular (icv) infusion. The NEFA-14 C-DHA perfusates were prepared to a concentration of 100 μ Ci/ml, 78 μ Ci/ml or 2 μ Ci/μ l for the gavage, iv or icv study, respectively. The purity of 14 C-radiotracer was > 99.9% as confirmed by high performance liquid chromatography (HPLC) and liquid scintillation counting (LSC) (Supp Fig. 1). Radiolabeled LPC- 14  Mississauga, ON, Canada) filled with 0.9% saline were implanted into the carotid artery. Surgery lasted for approximately 15 minutes. After surgery, all rats were singly housed to recover from anesthesia for 24 hours with ad libitum access to food and water. Rats were not fasted for the radiotracer oral gavage. Twenty-four hours post-surgery, rats were gavaged with 500 μ Ci of NEFA-14 C-DHA (8.62 μ mol DHA) in olive oil. Blood samples were obtained prior to gavage and at 15, 30, 45, 60, 90, 120, 180, and 240 minute minute post-gavage. After 4 hours, the rats were rapidly euthanized by head-focused, high-energy microwave irradiation (MI; 13.5 kW for 1.6 seconds; Cober Electronics Inc., Norwalk, CT, USA). Brains were excised and stored at − 80 °C for radioactive and biochemical analyses. Plasma was isolated from whole blood via microcentrifugation at 6200 rpm (2000 g) for 10 minutes and stored at − 80 °C.

Online Methods
Scientific RepoRts | 5:15791 | DOi: 10.1038/srep15791 Intravenous (iv) continuous tracer infusion. week-old rats were anesthetised with isofluorane inhalation (3% induction, 1-2% maintenance). Rats were implanted with polyethylene catheters into the right carotid artery and right jugular vein that were buried in a skin pouch above the scapula and randomized to receive either an iv dose of 86 μ Ci NEFA-14 C-DHA, 10 μ Ci NEFA-14 C-DHA or 10 μ Ci LPC-14 C-DHA which translated to 1.48 μ mol, 0.17 μ mol or 0.18 μ mol of DHA over 5 minutes as previously described 25 . week-old rats were implanted with catheters in the right carotid artery and right jugular vein which were externalized and protected by skin button (Braintree Scientific, Braintree, MA, USA) sutured to the subcutaneous tissue at the scapula, and hand-infused with 13 μ Ci of either NEFA-14 C-DHA or LPC-14 C-DHA which translated to 0.22 μ mol or 0.24 μ mol of DHA, respectively. Following hand-infusion of radiotracer, the catheter was immediately flushed with saline. Blood samples were collected at 0, 0.25, 0.5, 1, 5, 15, 30, 45, 60, 90, 120 minutes post-infusion. After two hours, the rats were rapidly euthanized by MI. Brains were excised and stored at − 80 °C for radioactive and biochemical analyses. Plasma was isolated from whole blood via microcentrifugation and stored at − 80 °C. Intracerebroventricular (icv) tracer infusion. At 15 weeks of age, eight rats were euthanized by MI for baseline brain fatty acid concentration measurements and seventeen 15-week-old rats were administered a 5 μ l icv injection containing 10 μ Ci of 14 C-DHA as previously described by Chen et al. 48 .

Intravenous (iv) bolus tracer infusion.
Lipid analysis. Total lipids from whole brain (icv), one brain hemisphere (gavage and iv) and plasma (gavage and iv) were extracted and various neutral lipid and phospholipid classes from total lipid extract were isolated as previously described by Chen et al. 48  Gas chromatography (GC)-flame ionization detection (FID). Plasma NEFA concentration and brain fatty acid concentration at baseline were quantified as described by Chen et al. 48 .
High performance liquid chromatography (HPLC)-UV photodiode array detection and liquid scintillation counting (LSC). Fatty acid methyl esters (FAME) analysis were previously described by Chen et al. 48 .
Matrix-assisted laser desorption/ionization (MALDI)-mass spectral imaging. The brain tissue block was mounted onto the specimen disc of a cryostat (Leica Microsystems) using optimal cutting temperature (OCT) compound (Sakura Finetek, Torrance, CA). The tissue sections were sliced at a thickness of 12 μ m at − 15 °C and mounted onto indium tin oxide (ITO)-coated glass slides. A thin matrix layer was applied to the tissue sections using an automated MALDI plate matrix deposition system (TM-Sprayer ™ , Leap Technologies, Carrboro, NC). A total of 5 ml of DHB solution (15 mg/ml in 50% Scientific RepoRts | 5:15791 | DOi: 10.1038/srep15791 acetonitrile/0.1% trifluoroacetic acid) or 9-aminoacridine (9AA, 15 mg/mL in methanol) was sprayed per slide during 4 passes at 140 °C (DHB) or at 80 °C (9AA) with a velocity of 400 mm/min and a line spacing of 3 mm. A time-of-flight tandem mass spectrometer (AB Sciex TOF/TOFTM 5800 System, AB SCIEX, Ontario, Canada) was used to acquire the images. MALDI-mass spectra were obtained using a Nd:YAG laser (349 nm) at 3 ns pulse width and 400 Hz firing rate. To install the ITO-coated glass slides in the ionization chamber, we used a special holder (AB Sciex) having concavities. The data were acquired in the positive-ion or negative-ion reflector mode using an external calibration method. The external calibration lipids (Avanti Polar Lipids) were deposited on the ITO-coated slides to minimize mass shift. In the imaging experiment, a total of 200 laser shots per point were irradiated (1s/point) and the interval between data points were 80 μ m. The mass spectrometric data were processed using a specialized script of Analyst software (AB Sciex) at a mass resolution of 0.1 amu and images were visualized using TissueView software (AB Sciex). After MALDI imaging was completed, the glass coverslip containing the placenta section was carefully removed from the MALDI plate, dipped in methanol to remove the matrix and fixed prior to DAPI staining.
Slices were prepared with Vectashield (Vector Laboratories) mounting medium containing DAPI and visualized at 368nm on a widefield epi-fluorescent microscope at 4× magnification. Multiple images were stitched together to make full image using Nikon Elements Software (NIS-Elements Basic Research 3.10).

Kinetics.
To estimate the relative contribution of plasma NEFA (UE) and esterified (ES) DHA to brain phospholipids, the following equation were derived: where C * brain is the radioactivity in the brain (nCi/g brain) from DHA at time, T, k * UE/ES is unidirectional incorporation coefficient (μ l/s/g brain) which represents the incorporation of plasma 14 C-DHA into stable brain lipid pools, and C * plasma (UE/ES) is the radioactivity in the plasma (nCi/ml plasma) from DHA at time, T, of NEFA-14 C-DHA gavaged rats. Integration of Equation 1 from the start of radiotracer infusion to time of death at T will yield: However, to calculate k * ES , k * UE was separately measured from intravenous infusion of NEFA-14 C-DHA where an acute five-minute infusion measures the uptake of albumin-bound NEFA-14 C-DHA without 14 C-DHA incorporation into phospholipids, triacylglycerol, cholesteryl esters, and diacylglycerol via hepatic lipoprotein packaging. k * UE is often referred to as k. * Since the incorporation coefficient applies to both radiolabeled and non-radiolabeled fatty acids, we can further calculate the rate of incorporation of non-radiolabeled plasma DHA into stable brain lipid pools, J in .
where C plasma is the plasma NEFA-DHA concentration. In addition to the calculation of relative contribution of NEFA-DHA to brain phospholipid DHA uptake, the comparison of J in from iv infusion and rate of loss, J out , from icv infusion can infer to the amount of plasma lipid pool(s) contributing to the maintenance of brain phospholipid DHA levels.
Loss half-lives, t 1/2 (day), of 14 C-DHA in brain total and phospholipid classes were calculated from the slope of regression lines (day −1 ) which were obtained by plotting log-transformed 14  The t 1/2 of DHA was used to calculate the J out (nmol/g brain/day) from brain phospholipids by the following equation: = .
( ) / J C t 0 693 7 out brain 1 2 where C brain is the baseline DHA concentration in a stable phospholipid pool (nmol/g brain). If the NEFA-DHA pool is the major contributor to total brain DHA uptake, then J in would equal J out . By rearranging Equation 5, predictive k * UE can be solved: Finally, to calculate a fractional loss (%/day) of DHA from stable phospholipid pool. analyzed as pooled samples and do not have SEM. Statistical comparisons of gavage k * UE /J in and iv k * UE /J in as well as icv predictive k * UE and gavage/iv k * UE were performed, a priori, using two tailed Student's t-test. Statistical significant differences were denoted as * p < 0.05, ** p < 0.01 and *** p < 0.001.