ABSTRACT: We have studied the effects of dietary FA on theaccumulation and secretion of [3H]glycerolipids by salmon he-patocytes in culture. Atlantic salmon were fed diets supplementedwith either 100% soybean oil (SO) or 100% fish oil (FO), andgrew from an initial weight of 113 ± 5 g to a final weight of 338± 19 g. Hepatocytes were isolated from both dietary groups andincubated with [3H]glycerol in an FA-free medium; a mediumsupplemented with 0.75 mM of one of three FA—18:1n-9, 20:5n-3,or 22:6n-3—or a medium supplemented with 0.75 mM of thesulfur-substituted FA analog tetradecylthioacetic acid (TTA),which cannot undergo β-oxidation. Incubations were allowed toproceed for 1, 2, 6, or 24 h. The rate of the secretion of radioac-tive glycerolipids with no FA added was 36% lower from hepato-cytes isolated from fish fed the FO diet than it was from hepato-cytes isolated from fish fed the SO diet. Hepatocytes incubatedwith 18:1n-9 secreted more [3H]TAG than when incubated withno FA, whereas hepatocytes incubated with 20:5n-3 or TTA se-creted less labeled TAG than when incubated with no FA. Thisobservation was independent of the feeding group. Hepatocytesincubated with 22:6n-3 secreted the highest amounts of total[3H]glycerolipids compared with the other treatments, owing toincreased secretion of phospholipids and mono- and diacylglyc-erols (MDG). In contrast, the same amounts of [3H]TAG were se-creted from these cells as from cells incubated in an FA-freemedium. The lipid-lowering effect of FO is thus independent of22:6n-3, showing that 20:5n-3 is the FA that is responsible for thelipid-lowering effect. The ratio of TAG to MDG in lipids secretedfrom hepatocytes to which 20:5n-3 or TTA had been added waslower than that in lipids secreted from hepatocytes incubatedwith 18:1n-9 or 22:6n-3, suggesting that the last step in TAG syn-thesis was inhibited. Morphometric measurements revealed thathepatocytes incubated with 20:5n-3 accumulated significantlymore cellular lipid than cells treated with 18:1n-9, 22:6n-3, TTA,or no treatment. The area occupied by mitochondria was alsogreater in these cells. The present study shows that dietary FO re-duces TAG secretion from salmon hepatocytes and that 20:5n-3mediates this effect.

Paper no. L9540 in Lipids 40, 477–486 (May 2005).

Fish oil (FO), which is particularly rich in the PUFA 20:5n-3and 22:6n-3, has traditionally been used as the predominant

lipid component of salmon feed. However, the uncertain sup-ply and variable price of FO have led to an interest in using themore readily available plant oils rich in 18-carbon n-6 [soybeanoil (SO) and sunflower oil] and n-9 (rapeseed oil) FA in salmondiets. Little information is available, however, concerning howthese plant oils affect lipid deposition in salmon hepatocytesand how they affect the secretion of TAG-rich VLDL fromthese cells. There is considerable evidence that consumption ofFO lowers plasma TAG concentrations in humans (1,2) andprotects against coronary heart diseases (1,3,4). It is particu-larly important to study these processes in salmonids, sincethey are more predisposed to coronary arteriosclerosis thanother fish species (5). Several mechanisms are probably in-volved in the inhibiting effect of FO on the secretion of TAG-rich VLDL particles in mammals, and the detailed mechanismshave not been completely elucidated. The effect has been at-tributed to the content of n-3 FA in the feed (6,7). Studies usingprimary hepatocytes from rats have shown that n-3 PUFA in-hibit the secretion of TAG-rich VLDL (8–11). The PUFA in-hibit diacylglycerol acyltransferase (DGAT, EC 2.3.1.20)(12–15), which is the rate-limiting enzyme in TAG synthesis,and they inhibit the assembly process of the VLDL particles(9,10,16). Furthermore, several studies in mammals haveshown that the lipid-lowering effect of FO is mediated by20:5n-3 and not by 22:6n-3 (13–15,17,18).

We have investigated the effects of several FA on lipid ac-cumulation and on the secretion of glycerolipids in Atlanticsalmon. We isolated hepatocytes from salmon fed diets supple-mented with either FO or SO and then added the two mainPUFA in FO (20:5n-3 and 22:6n-3) separately to the hepato-cyte cultures. We incubated other cultures with 18:1n-9, whichis a common FA in vegetable oils and which stimulates lipidsecretion by cultivated rat hepatocytes (9,13,19). Finally, wecompared the effects of these FA with the effects of a sulfur-substituted FA analog, tetradecylthioacetic acid (CH3–(CH2)13–S–CH2–COOH) (TTA), on glycerolipid secretion.TTA is unable to undergo β-oxidation (for review see Ref. 20),and it inhibits TAG secretion in mammals (14,21).

MATERIALS AND METHODS

Materials. TTA was produced at Dr. R. Berge’s laboratory(Haukeland Hospital, University of Bergen, Norway). The ra-dioactive isotope [1,2,3-3H]glycerol (30–60 Ci/mmol) was ob-

Copyright © 2005 by AOCS Press 477 Lipids, Vol. 40, no. 5 (2005)

*To whom correspondence should be addressed at AKVAFORSK, Instituteof Aquaculture Research, P.O. Box 5010, NO-1432 Ås, Norway.E-mail: anne.vegusdal@akvaforsk.nlh.noAbbreviations: DGAT, diacylglycerol acyltransferase (EC 2.3.1.20); FO, fishoil; MDG, mono- and diacylglycerol; PL, phospholipid; SO, soybean oil;TTA, tetradecylthioacetic acid.

Effect of 18:1n-9, 20:5n-3, and 22:6n-3 on Lipid Accumulationand Secretion by Atlantic Salmon Hepatocytes

A. Vegusdala,*, T. Gjøenb, R.K. Bergec, M.S. Thomassena, and B. Ruytera

aAKVAFORSK, Institute of Aquaculture Research, NO-1432 Ås, Norway, bDepartment of Microbiology, Instituteof Pharmacy, University of Oslo, NO-0316 Oslo, Norway, and cDepartment of Clinical Biology, Division of Biochemistry,

Haukeland University Hospital, University of Bergen, NO-5021, Bergen, Norway

tained from American Radiolabeled Chemicals, Inc. (St. Louis,MO). Nonlabeled glycerol, Leibowitz-15 medium, FBS, an-tibiotics, HEPES, L-glutamine, collagenase, trypsin, FA, essen-tial FA-free BSA, collagenase type 1, and 2′,7′-dichlorofluo-rescein were all supplied by Sigma-Aldrich (St. Louis, MO).Tissue culture plastic ware was obtained from NalgeNunc In-ternational (Naperville, IL). Acetic acid, chloroform, petroleumether, diethyl ether, and methanol were purchased from Merck(Darmstadt, Germany). Benzene was obtained from RathburnChemicals, Ltd. (Walkerburn, Scotland). Methanolic HCl and2,2-dimethoxypropane were purchased from Supelco Inc.(Bellefonte, PA). Glass-baked silica gel K6 plates were ob-tained from Whatman International, Ltd. (Maidstone, Eng-land). Glutaraldehyde, epon resin, copper grids, and lead cit-rate were supplied by Electron Microscopy Sciences (FortWashington, PA).

Fish and experimental design. The feeding study was per-formed as described by Grisdale-Helland et al. (22). Briefly,Atlantic salmon in tanks containing seawater at 12°C were fedfor approximately 950 day-degrees with one of two test diets.There were three tanks per diet. The diets, containing 30% fat,were based on fish meal and differed only in the type of sup-plementary oil that they contained [see Grisdale-Helland et al.(22) for a detailed description of the diets]. The fish were fedad libitum using automatic band-feeders every 10 min, 24 h/d.Supplementary lipid in one diet, called “FO,” came solely fromfish oil (capelin oil), whereas the supplementary lipid in theother diet, called “SO,” came solely from crude soybean oil.The salmon grew from an initial weight of 113 ± 5 g to a finalweight of 338 ± 19 g.

Isolation of hepatocytes. Three fish from each tank were usedfor isolation of hepatocytes at the end of the growth period. Thefish were anesthetized with metacain. The abdominal cavity wasexposed and the vena porta cannulated. The liver was perfusedfollowing a two-step collagenase procedure developed by Seglen(23) and modified by Dannevig and Berg (24), to isolate hepato-cytes. The hepatocytes were easily isolated after the collagenasedigestion by gentle shaking of the digested liver in Leibowitz-15medium. The suspension of parenchymal cells was filteredthrough a 100 µm nylon filter. Hepatocytes were isolated fromthis suspension by three sedimentations each of 2 min at 50 × g.The hepatocytes were resuspended in Leibowitz-15 culturemedium containing 2% FBS, 2 mM L-glutamine, and 0.1 mgmL−1 gentamycin. Cell viability was assessed by staining withTrypan blue (0.4%). The protein content of the cell suspensionwas determined using the method described by Lowry et al. (25).Approximately 5 × 106 hepatocytes (approximately 8 mg of pro-tein) were plated onto 75-mL Nunc flasks and left to attachovernight at 12°C.

Incubation with FA and [3H]glycerol. The hepatocytes werethoroughly washed with Leibowitz-15 medium without serumsupplementation, before the incubation with FA and [3H]glyc-erol. Each cell flask with hepatocytes was incubated with 0.5Ci/mmol [3H]glycerol (final glycerol concentration 25 µM), 10mM lactate, and either no FA (Ctr), 18:1n-9, 20:5n-3, 22:6n-3,or TTA in a total volume of 5 mL Leibowitz-15 medium. A

final FA concentration of 0.75 mM was used, a concentrationoften used in similar studies with mammalian cells (13,19). TheFA were added to the medium in the form of their potassiumsalts bound to BSA (the molar ratio between an FA and BSAwas 2.5). The cells were incubated for 1, 2, 6, or 24 h at 12°C,and then the medium from each culture flask was transferred tovials and centrifuged for 5 min at 50 × g. The supernatant (cul-ture medium) from a flask was either immediately frozen at−70°C and stored until the analysis of radiolabeled lipids or di-rectly used for the isolation of VLDL as described next. Hepa-tocytes were washed twice with 0.1 M PBS (pH 7.4), followedby preparation for electron microscopy.

Isolation and analysis of VLDL. To determine the most rep-resentative method for investigating the secretion of glyc-erolipids from cultivated hepatocytes, we compared the lipidanalysis of secreted VLDL fractions with the analysis of totalesterified lipids secreted to the culture media. Hepatocytes fromSO-fed fish that had been incubated for 1, 2, 6, or 24 h in amedium without FA supplementation, but with [3H]glycerol,were used for the comparison of the two methods. The culturemedia were divided in two halves at each time point; one forVLDL separation and one for determination of total secretedesterified lipids. A discontinous NaCl/KBr gradient wasformed by filling Bell-top Quick-Seal tubes with two layerswith different densities, using a peristaltic pump. A layer ofaqueous NaCl solution (10 mL, 0.9%) was pumped into thebottom of the tube, and then culture medium (5 mL), the den-sity of which had been adjusted to 1.30 g mL−1 with KBr, waspumped underneath. The tubes were sealed with a cordless tubestopper and then ultracentrifuged (4 h, 4°C, average accelera-tion 167,153 × g) in a vertical rotor, using slow accelerationand deceleration below 160 rpm. The procedure was a modi-fied version of the procedure developed by Chung et al. (26).Rotor shut-down added approximately 20 min to the total run-time. The top of the tube was gently sliced off using a scalpel,and the gradient was unloaded into 30 fractions (each of vol-ume 0.5 mL) from the bottom of the tube using a peristalticpump. The fractions covered the density range from 1.01 to1.28 g mL−1. The density of a fraction was determined by mea-suring its refractive index (n) in a refractometer, and calculat-ing the density according to:

d = 3.298n – 3.396G [1]

where d is the density of the fraction and G is the density of theNaCl/KBr gradient at which this fraction was collected.

VLDL was recovered from the surface of the gradient withdensity lower than 1.015 g mL−1. The VLDL fraction wasfrozen at −70°C and stored until the analysis of radiolabeledlipid classes and protein composition. Protein samples (0.5 µgprotein) from the fraction associated with VLDL were addedto a buffer containing SDS and boiled for 5 min. They werethen loaded onto a 7% polyacrylamide gel for further analysesby SDS-PAGE as described by Burnette (27).

Lipid extraction and analysis of lipid classes and FA com-position. Total lipids were extracted from culture media and

478 A. VEGUSDAL ET AL.

Lipids, Vol. 40, no. 5 (2005)

from VLDL by the method described by Folch et al. (28). Thechloroform phase was dried under N2, and the residual lipid ex-tract was redissolved in hexane. Phospholipids (PL), mono-and diacylglycerols (MDG), and TAG were separated by TLCusing a mixture of petroleum ether, diethyl ether, and aceticacid (113:20:2, by vol) as the mobile phase. The lipids were vi-sualized by spraying the TLC plates with 0.2% (wt/vol) 2′,7′-dichlorofluorescein in methanol, and they were identified bycomparison with known standards under UV light. The spotscorresponding to PL, MDG, and TAG were scraped off intovials containing liquid scintillation fluid (InstaGel II Plus;Packard Instrument, Downers Grove, IL), and the radioactivitywas measured in a liquid scintillation counter (TRI-CARB1900 TR; Packard Instrument). There are certain limitations tothe method used, and what is actually measured is the recoveryof tritium (derived from tritiated glycerol) in the different lipidclass fractions. The total FA profiles in the diets were deter-mined by GC as described by Grisdale-Helland et al. (22). Thetotal endogenous FA composition of hepatocytes was deter-mined essentially as described by Ruyter et al. (29).

Preparation of hepatocytes for electron microscopy. Afterbeing washed in 0.1 M PBS, cells were scraped from the flaskinto 2 mL of 0.1 M PBS (pH 7.4), gently centrifuged, and thenresuspended and fixed in 2% glutaraldehyde in 0.1 M cacodylatebuffer (pH 7.4) at 4°C for 24 h. The cells were rinsed in the samebuffer and post-fixed for 60 min in 2% OsO4 containing 1.5%potassium ferrocyanide, followed by en bloc staining with 1.5%uranyl acetate. Cells were dehydrated in a series of ethanol solu-tions (70, 90, 96, and 100%) and propylene oxide, and then em-bedded in epon resin, which was polymerized at 60°C for 12 h.Ultrathin sections (approximately 50 nm) were cut on a ReichertUltracut E ultramicrotome using a diamond knife. The sectionswere placed onto formvar/carbon-coated 75-mesh copper grids,post-stained for 2 min with 0.2% lead citrate solution in 0.1 MNaOH, and examined in a Philips CM 100 transmission electronmicroscope at an accelerating voltage of 80 kV.

Morphometry. Hepatocytes isolated from three fish in eachfeeding group were incubated for 24 h in medium containingeither no FA (Ctr), 18:1n-9, 20:5n-3, 22:6n-3, or TTA. Lipiddroplets and mitochondria in these samples were then analyzedby morphometry. Micrographs for morphometry were takenfrom 20 cells per treatment at a magnification of 2950×, at ran-dom positions in the sample. A square lattice grid was super-imposed onto the micrographs, and the number of intersectionsfalling on lipid droplets or on mitochondria was counted. Wedetermined the total area of the mitochondria and of the lipiddroplets in the cells relative to the total area of the cells (ex-cluding the nuclei), as described by Hexeberg et al. (30). Inbrief, the percent area, A, of the cells constituting lipid dropletsor mitochondria was calculated from:

A = (Plip /Pcell) × 100 [2]

where Plip is the number of intersections falling on the lipiddroplets and Pcell is the number of intersections falling on thecells (excluding the nuclei).

Statistical analysis. ANOVA was carried out using a modelthat included effects of diet, incubation time, and exogenousFA, and all interactions between these factors, to analyze theeffect of endogenous FA composition on the secretion of totalradiolabeled lipids. A reduced model in which only main ef-fects were included was used where the interactions includedin the full model gave nonsignificant results. Treatment differ-ences were ranked using Duncan’s multiple range test for themain effects. Pearson’s correlation coefficient was calculatedand used to analyze the correlation between the mitochondrialarea and the lipid area.

One-way ANOVA (general linear model) was used to testthe significance of the analysis of the level of [3H]glycerolipidssecreted at different time points and at different additions ofexogenous FA. Significant differences between means were an-alyzed using Duncan’s test. Differences were considered to besignificant at a 5% level. SAS computer software was used forall the statistical analyses (31).

RESULTS

FA composition. The SO diet contained more than eight timesmore n-6 FA (45.4%) than the FO diet (5.3%). The FO diet, onthe other hand, contained almost twice as much n-3 FA (21.3%)as the SO diet (10.9%). The percentage of total monounsatu-rated FA was twice as high in the FO diet (49.3%) than in theSO diet (25.3%). The FA composition of hepatocytes isolatedfrom fish fed either the SO or the FO diets was substantially af-fected by the dietary FA composition (Table 1). The percent-age of n-3 FA in the hepatocyte lipids was 1.6 times greater incells from fish fed the FO diet (39.3%) than in cells from fishfed the SO diet (24.5%). The percentage of n-6 FA was approx-imately seven times greater in cells from fish fed the SO diet(31%) than in cells from fish fed the FO diet (4.5%). There wasan enrichment of 22:6 n-3 in hepatocytes isolated from bothSO-fed (17%) and FO-fed fish (27%) compared with the corre-sponding percentages in the two diets (3 and 7%, respectively).

There were no major differences in the percentages of totalsaturated FA between cells in the two dietary groups.

[3H]Glycerolipids in the total extracted lipid fractions fromculture media and from isolated VLDL particles. There was alower recovery of secreted radiolabeled glycerolipids in theVLDL fraction than in the total culture media (Table 2). Lipidclass analysis showed that whereas TAG dominated at all timepoints in the VLDL fractions, it constituted only 13% of theglycerolipid fraction in the total culture media at the early timepoints (1 and 2 h) and increased thereafter to 87% at the latertime points (Table 2). The discrepancy between the results ob-tained by these two analytical methods can most probably beexplained by the fact that exogenously added radiolabeledglycerol exhibits a natural time lag in its appearance in secretedglycerolipids. In addition, the procedure used for VLDL isola-tion (density gradient centrifugation) also may influence the re-sults. In this method, the most TAG-rich VLDL particles willbe more buoyant than the VLDL particles that are less TAG-rich. Only the TAG-rich VLDL particles will therefore be col-

LIPID ACCUMULATION AND SECRETION BY ATLANTIC SALMON HEPATOCYTES 479

Lipids, Vol. 40, no. 5 (2005)

lected on the top of the gradient. Based on these results, the totalrecovery of glycerolipids in the culture media, and not the recov-ery of glycerolipids in isolated VLDL fraction, was used for thefurther analysis of glycerolipid secretion from hepatocytes.

Effect of endogenous FA composition on the secretion oftotal [3H]glycerolipids. The rate of secretion of total [3H]glyc-erolipids (the sum of PL, MDG, and TAG) was 36% lower inhepatocytes from fish fed the FO diet than in hepatocytes fromfish fed the SO diet (P = 0.003) (Fig. 1).

Effects of time and exogenously added FA on the secretion of[3H]glycerolipids. Figure 2 shows the secretion of total [3H]glyc-erolipids (the sum of PL, MDG, and TAG) after 1-, 2-, 6-, and24-h incubations with either no FA (Ctr), 18:1n-9, 20:5n-3,22:6n-3, or TTA. Hepatocytes from SO-fed fish secreted moreesterified lipids than cells from FO-fed fish at the early time-points (1 and 2 h) irrespective of FA supplementation. Incuba-tion of the hepatocytes with a relatively high concentration (0.75

480 A. VEGUSDAL ET AL.

Lipids, Vol. 40, no. 5 (2005)

TABLE 1FA Composition of the Total Lipid Fraction of the Diets and Hepatocytes Isolated from Either Soybean Oil(SO)- or Fish Oil (FO)-Fed Fisha

SO diet FO diet SO cells FO cells

14:0 0.9 5.7 1.9 ± 0.22a 2.5 ± 0.01a

16:0 12.7 14.0 16.2 ± 1.18a 15.5 ± 0.22a

18:0 3.7 1.6 7.9 ± 0.31b 4.9 ± 0.02a

16:1n-7 1.3 7.5 1.1 ± 0.19a 3.8 ± 0.21b

18:1n-7 1.6 2.8 1.5 ± 0.18a 3.6 ± 0.13b

18:1n-9 20.3 10.3 15.9 ± 2.18a 13.9 ± 0.68a

20:1 (sum isomers) 1.1 12.8 3.2 ± 0.37a 5.0 ± 0.11b

22:1 (sum isomers) 1.0 15.9 0.4 ± 0.06a 0.9 ± 0.12b

18:2n-6 45.4 4.4 22.4 ± 0.99b 2.3 ± 0.18a

20:2n-6 ND ND 2.4 ± 0.71b 0.5 ± 0.00a

18:3n-3 5.3 0.8 ND ND20:3n-6 ND ND 3.2 ± 0.54b 0.2 ± 0.20a

18:4n-3 0.5 3.2 0.2 ± 0.23a 0.5 ± 0.11b

20:4n-3 ND ND 2.9 ± 1.65b 0.8 ± 0.15a

20:4n-6 ND ND 2.9 ± 0.15b 1.4 ± 0.22a

20:5n-3 2.1 9.0 3.3 ± 0.32a 8.5 ± 0.40b

22:5n-3 0.2 0.6 1.5 ± 0.22a 3.0 ± 0.25b

22:6n-3 2.8 7.0 16.6 ± 1.11a 26.5 ± 0.33b

ΣOthers 1.1 4.4 0.2 ± 4.76a 6.2 ± 0.60b

ΣSaturated 17.3 21.3 26.0 ± 0.22b 22.9 ± 0.08b

ΣMonoene 25.3 49.3 22.1 ± 0.37b 27.2 ± 0.21b

Σn-3 10.9 21.3 24.5 ± 5.21b 39.3 ± 0.31b

Σn-6 45.4 5.3 31.0 ± 4.76b 4.5 ± 0.60b

aThe quantity of each FA is given as a percentage of the total FA. Data are means ± SEM (n = 3). ND, not detectable. Val-ues with different roman superscripts are significantly different (P ≤ 0.05; n = 3).

TABLE 2 Percentage Distributions of [3H]Glycerol in TAG, MDG, and PL of Isolated VLDL Particles and in Mediuma

Incubation VLDL fractions Recovery VLDL Total media

time (h) TAG MDG PL (%) TAG MDG PL

1 81.9 ± 4.68a 3.4 ± 1.54a 14.7 ± 3.19b 21.0 13.3 ± 6.05a 9.8 ± 3.78a 76.9 ± 10.32b

2 88.3 ± 0.71a,b 3.5 ± 0.83a 8.2 ± 0.16a,b 32.2 30.5 ± 5.75b 7.9 ± 1.59a 61.6 ± 5.10b

6 93.8 ± 1.86b 2.4 ± 0.09a 3.8 ± 1.93a 68.1 82.5 ± 3.33c 4.9 ± 0.53a 12.6 ± 2.79a

24 94.4 ± 0.1b 1.6 ± 0.68a 4.0 ± 0.68a 91.4 87.4 ± 1.13c 4.2 ± 0.09a 8.3 ± 1.05a

aData are shown as means ± SEM (n = 3). Values marked with different roman letters indicate significant differences (P ≤0.05) between lipid classes. VLDL recovery was calculated as the amounts of [3H]glycerolipids recovered in the isolatedVLDL fraction as a percentage of the amounts of total [3H]glycerolipids recovered in the culture medium. Isolated VLDLand total media were from hepatocyte cultures isolated from SO-fed fish. MDG, mono- + diacylglycerols; PL, phospho-lipids; for other abbreviations see Table 1.

FIG. 1. Secretion rate of [3H]glycerolipids by hepatocytes from fish fedsoybean oil (SO) or fish oil (FO) [the sum of TAG, mono- and diacyl-glyerols (MDG) and phospholipids (PL)]. The secretion rate is calculatedas nmol [3H]glycerolipids secreted per min per mg protein. Data aremeans ± SEM (n = 3). The difference is significant at P = 0.003. Eachdata point (n) is the average of 20 individual measurements.

LIPID ACCUMULATION AND SECRETION BY ATLANTIC SALMON HEPATOCYTES 481

Lipids, Vol. 40, no. 5 (2005)

FIG. 2. Effect of exogenously added FA on the secretion of [3H]glycerolipids (the sum of TAG, MDG and PL) from fish fed SO or FO after incuba-tions with an FA-free medium (Ctr) or media supplemented with 0.75 mM of either 18:1n-9, 20:5n-3, 22:6n-3, or tetradecylthioacetic acid (TTA)for 1, 2, 6, or 24 h. Data are means ± SEM (n = 3). For other abbreviations see Figure 1.

482 A. VEGUSDAL ET AL.

Lipids, Vol. 40, no. 5 (2005)

FIG. 3. Percentage distribution between the different [3H]glycerolipids (TAG, MDG, and PL) after incubations with an FA-free medium (Ctr) ormedia supplemented with 0.75 mM of either 18:1n-9, 20:5n-3, 22:6n-3, or TTA for 1, 2, 6, or 24 h. Each data point shown is a mean value of thetwo dietary groups ± SEM (n = 6). For abbreviations see Figures 1 and 2.

mM) of exogenously added FA for a longer period of time, how-ever, reduced the effect of the feeding group that had been ob-served at 1 and 2 h. Approximately 6–7 times more [3H]glyc-erolipids were secreted at 24 h than at 1 h from hepatocytes in-cubated with 22:6n-3 or 18:1n-9, irrespective of dietary group.In comparison, there were only approximately 2–3 times more[3H]glycerolipids secreted at 24 h than at 1 h from control cellsor cells incubated with 20:5n-3 or TTA. This shows that hepato-cytes to which 18:1n-9 or 22:6n-3 had been added secreted[3H]glycerolipids more than twice as rapidly as control cells orcells to which 20:5n-3 or TTA had been added.

Figure 3 shows the effect of exogenously added FA on thesecretion of different [3H]glycerolipid classes (TAG, MDG,and PL). No significant differences in percent distribution ofglycerolipid classes were found between the two feedinggroups at any time point. The results are therefore presented asmean values of the two feeding groups.

The majority of the [3H]glycerolipids secreted from all cellincubations after 24 h were in the form of TAG, except fromhepatocytes incubated with 20:5n-3. The percentage of MDGsecreted from the cells incubated with 20:5n-3 increased dur-ing the time period, giving a ratio of TAG to MDG of approxi-mately 1 after 24 h of incubation. In comparison, this ratio was23.5 in control cells after 24 h of incubation. The ratios of TAGto MDG after 24 h were approximately 10, 3, and 2.5, for cellsincubated with 18:1n-9, 22:6n-3, and TTA, respectively. Thesecretion patterns for the different lipid classes from cells incu-bated with TTA were similar to those for cells incubated with20:5n-3.

TAG secretion from 20:5n-3-treated cells was approxi-mately 70% lower after 24 h of incubation than it was for con-trol cells (Fig. 4). The FA 22:6n-3, in contrast to 20:5n-3, had

no major effect on the amount of TAG secreted. The only ex-ogenously added FA that stimulated TAG secretion comparedwith control cells during the experiment was 18:1n-9. Theamount of TAG secreted from cells incubated with TTA for 24h was approximately 40% lower than that of control cells.

Morphometric quantification of lipid and mitochondrialarea. The mitochondrial area of the hepatocytes was not af-fected by the composition of FA in the feed. There was aslightly higher intracellular lipid area in cells from the FO-fedfish than in cells from the SO-fed fish. Owing to a relativelyhigh SD between the analyzed samples, the differences were,however, not significant. This allows us to present the morpho-metric results as mean values from all observed cells, irrespec-tive of feeding group. The FA added to the cells in culture, onthe other hand, affected the lipid area and the mitochondrialarea more profoundly.

Figure 5 shows the percent area of lipid droplets and mito-chondria after incubation with either no FA (Ctr), 18:1n-9,20:5n-3, 22:6n-3, or TTA for 24 h, and Figure 6 shows repre-sentative micrographs of hepatocytes after each incubation.Lipid accumulations were observed as light and/or blackdroplets in the cytoplasm of hepatocytes. Cells incubated with20:5n-3 had the largest lipid area (12.0%) (Figs. 5, 6C),whereas cells incubated with 18:1n-9 or 22:6n-3 had a signifi-cantly smaller lipid area: 4.8 and 6.7%, respectively (Figs. 5,6B, 6D). Cells incubated with TTA had the lowest lipid area,2%, a level similar to that of control cells (Figs. 5, 6A, 6E). Thelipid area and the mitochondrial area were highly correlated (P= 0.002). The mitochondrial area was highest in cells incubatedwith 20:5n-3 (13.5%) and was significantly higher than in cellsincubated with TTA (10.7%) and in control cells (9.7%), whichhad the lowest mitochondrial area.

LIPID ACCUMULATION AND SECRETION BY ATLANTIC SALMON HEPATOCYTES 483

Lipids, Vol. 40, no. 5 (2005)

FIG. 4. Relative TAG secretion after incubations with an FA-freemedium (Ctr) or media supplemented with 0.75 mM of either 18:1n-9,20:5n-3, 22:6n-3, or TTA for 24 h. The control TAG levels are set to100% in each incubation for both dietary groups, and the mean valuesof the two dietary groups are presented. The mean values of the TAGlevels of the different FA incubations are presented relative to the con-trol TAG levels. Data are means ± SEM (n = 6). Different letters indicatesignificant (P < 0.05) differences. For abbreviations see Figures 1 and 2.

FIG. 5. Morphometrical estimates of lipid area and mitochondrial areaof hepatocytes incubated for 24 h in an FA-free medium (Ctr) or mediasupplemented with 0.75 mM of either 18:1n-9, 20:5n-3, 22:6n-3, orTTA. Data are the relative areas of lipid droplets or mitochondria of totalcell area (exclusive of the nuclei) and are shown as the mean values ofthe two dietary groups ± SEM (n = 6). Different letters indicate signifi-cant (P < 0.05) differences between the different FA-incubations. Forabbreviation see Figure 2.

DISCUSSION

The endogenous FA composition of hepatocytes was influ-enced by the identity of the oil used as lipid source in the diet.This agrees with several studies using different types of veg-etable oil in salmonid diets. Generally, the replacement of FOby SO results in higher levels of n-6 FA and lower levels of n-3FA in the tissues of Atlantic salmon and rainbow trout(29,32–35). The marked differences in FA composition due tothe diets had a pronounced effect on the secretion of glyc-erolipids. Lipid secretion from hepatocytes from fish fed theFO diet was 36% lower than the secretion from hepatocytesfrom fish fed the SO diet. A number of in vivo studies in mam-mals during the past few decades have shown that FO diets in-hibit lipid secretion by hepatocytes (2,16,36). Our results fromfish hepatocytes agree with the findings from mammals andshow that an endogenous FA composition with a high n-6/n-3ratio promotes lipid secretion to a level that is higher than thatwith a lower n-6/n-3 ratio. Other components in diets however,

may also influence the TAG-rich lipoprotein secretion from he-patocytes. SO may contain significant quantities of lecithin(37), whereas FO may contain relatively high levels of choles-terol (38), both of which are known to modify lipoprotein me-tabolism in mammals (39–41). We do not know whether di-etary lecithin and cholesterol affect these processes in fish, butthis topic will be important for further studies.

The mechanisms by which FO in the diet reduces TAG se-cretion in fish are not known, nor have they been completelyelucidated in mammals. However, several mechanisms andregulatory sites may be involved. The present study demon-strates that 20:5n-3, and not 22:6n-3, inhibits TAG secretion insalmon hepatocytes, in accordance with the results of manystudies in rat hepatocytes (13–15,17,19,42,43). Studies inmammals have shown that dietary FO interferes with the as-sembly of the initial VLDL precursor particles in hepatocytesand inhibits VLDL secretion in this way (9,10,16,44,45).Whether VLDL secretion in fish is regulated in the same man-ner is not known.

Several studies in mammals have shown that 20:5n-3 re-duces the availability of intracellular TAG for incorporationinto VLDL by inhibiting the activity of DGAT (13,14), that is,the enzyme catalyzing the conversion of 1,2DAG into TAG(13–15,19). In contrast, 22:6n-3 does not affect DGAT activityin rat hepatocytes (14). Although we did not measure the activ-ity of DGAT, the reduced lipid secretion from hepatocytes in-cubated with 20:5 n-3 has led us to suggest that DGAT alsomay be a regulated step in fish. The inhibition of TAG synthe-sis should occur before or at the DGAT step, since the ratio ofTAG to MDG secreted from these cells is lower than that se-creted from control cells. The seemingly contradictory resultsof a reduced DGAT activity and the observed increase in intra-cellular lipid level in 20:5n-3 treated cells may be explained byan inhibiting effect of 20:5n-3 on the VLDL assembly step,leading to a net increase in cytosolic lipid droplets. Many fac-tors seem to affect the different steps of the lipid secretionprocess and its regulation, and we cannot on the basis of ourresults draw conclusions about which regulatory mechanismsare most important for the lowering effect of 20:5n-3 on TAGsecretion. However, lipid class composition of cellular lipidsmay give valuable additional information about the DGAT ac-tivity and TAG synthesis.

Furthermore, several studies in mammals have shown that20:5n-3 reduces TAG secretion by stimulating mitochondrialβ-oxidation, thus reducing the availability of FA for TAG syn-thesis (15,18,42), whereas 22:6n-3 is a poor substrate for mito-chondrial β-oxidation in rats (14,15). Frøyland et al. (18)showed that the plasma TAG level decreased in rats fed 20:5n-3, but not in rats fed 22:6n-3. The decrease was caused by anincrease in the number of mitochondria and an increase in thelevel of mitochondrial β-oxidation. We have not measured theactivities of enzymes involved in β-oxidation. However, wehave shown that incubation with 20:5n-3 leads to an increasedmitochondrial area and that it reduces the secretion of glyc-erolipids. These results suggest that 20:5n-3 also stimulates thecapacity of mitochondria to β-oxidize FA in salmon.

484 A. VEGUSDAL ET AL.

Lipids, Vol. 40, no. 5 (2005)

FIG. 6. Electron micrographs of hepatocytes incubated for 24 h in (A) anFA-free medium, or a medium supplemented with 0.75 mM of either (B)18:1n-9, (C) 20:5n-3, (D) 22:6n-3, or (E) TTA. The micrographs shownare representative cells for each of the five treatments. Lipid accumula-tions (arrows) were observed as light and/or black droplets in the cyto-plasm. The highest accumulation of lipid droplets was observed in hepa-tocytes incubated with 20:5n-3, whereas hepatocytes incubated with18:1n-9 or 22:6n-3 showed a significantly lower lipid droplet area. No oronly a few lipid droplets were observed in hepatocytes incubated withTTA or an FA-free medium. Bars = 5 µm. For abbreviation see Figure 2.

The microscopy study revealed only normal-sized mito-chondria in all hepatocyte samples, without any observationsof megamitochondria. Other investigators have found moredramatic effects on mitochondrial number and size after feed-ing trials in rats. It has been shown that EPA may initiate a sig-nal for megamitochondria formation and subsequent prolifera-tion (30,46). We suggest that a 24-h incubation is too short atime to achieve megamitochondria formation and high mito-chondrial proliferative activity in our cell culture experiment.The observed increase in mitochondrial area in EPA-treatedcells can therefore be a result of a moderate increase in thenumber of mitochondria.

The relative area of lipid droplets was highly correlated (P= 0.002) with the relative area of mitochondria, and both weresignificantly higher in cells incubated with 20:5n-3 than theywere in control cells. It may be that the amount of accumulatedlipid per se induces mitochondrial proliferation directly by anunknown mechanism, and that the potency of 20:5n-3 as a mi-tochondrial proliferator in salmon hepatocytes is related to theincreased intracellular lipid level estimated in the cells throughlipid droplet areas.

Although 22:6n-3 had no effect on TAG secretion, this FA re-sulted in the highest secretion of total glycerolipids. This wasmainly due to a higher incorporation of 22:6n-3 into secreted PL.Previous results from salmon hepatocytes showed that 22:6n-3was preferentially incorporated and conserved in PL (47,48).

Cells incubated with TTA secreted less TAG than controlcells, in accordance with studies in mammals showing thatTTA inhibits TAG secretion (14,21). This FA analog directs FAaway from esterification and toward FA oxidation in rats(14,17,21) through regulations resulting in increased activitiesof enzymes involved in mitochondrial and peroxisomal β-oxi-dation (18,30), and by inhibiting DGAT (14).

We have shown that 18:1n-9 stimulates TAG secretion, whichis in accordance with results from rat hepatocytes (13,19,43).However, the observed stimulating effect of SO and 18:1n-9 onlipid secretion, and the reported negative effects of some veg-etable oils on fish heart histology (49,50), should be considered.These effects may give rise to problems that are similar to prob-lems related to coronary heart diseases in humans. Such prob-lems have been linked to a diet low in n-3 FA (1–4).

In conclusion, we have shown that lipids are secreted fromcultivated salmon hepatocytes at a rate that is significantlylower in fish fed a diet containing FO as the only lipid sourcethan in fish given a diet that has been supplemented with SO.Further, we have shown that 20:5n-3, and not 22:6n-3, some-how mediates this effect, whereas 18:1n-9 stimulates TAG se-cretion to levels that are higher than those in control cells. Ourresults indicate that in dietary FO, 20:5n-3 is the FA that re-duces TAG secretion in fish and may be responsible for the hy-potriglyceridemic effect of FO observed in fish and humans.

ACKNOWLEDGMENTS

The authors are grateful to Inger Ø. Kristiansen for excellent techni-cal assistance and to Gerd Berge for help with the statistical analy-sis. The work was supported by the Norwegian Research Council.

REFERENCES

1. Harris, W.S., Connor, W.E., and McMurphy, M.P. (1983) TheComparative Reductions of the Plasma Lipids and Lipoproteins byDietary Polyunsaturated Fats: Salmon Oil versus Vegetable Oils,Metabolism 32, 179–184.

2. Nestel, P.J. (1990) Effects of n-3 Fatty Acids on Lipid Metabolism,Annu. Rev. Nutr. 10, 149–167.

3. Bang, H.O., Dyerberg, J., and Nielsen, A.B. (1971) Plasma Lipidand Lipoprotein Pattern in Greenlandic West-Coast Eskimos,Lancet 1, 1143–1145.

4. Phillipson, B.E., Rothrock, D.W., Connor, W.E., Harris, W.S., andIllingworth, D.R. (1985) Reduction of Plasma Lipids, Lipoproteinsand Apoproteins by Dietary Fish Oils in Patients with Hypertriglyc-eridemia, N. Engl. J. Med. 312, 1210–1216.

5. Farrell, A.P. (2002) Coronary Arteriosclerosis in Salmon: GrowingOld or Growing Fast? Comp. Biochem. Physiol. A Mol. Integr.Physiol. 132, 723– 735.

6. Childs, M.T., King, I.B., and Knopp, R.H. (1990) Divergent Lipopro-tein Responses to Fish Oils with Various Ratios of EicosapentaenoicAcid and Docosahexaenoic Acid, Am. J. Clin. Nutr. 52, 632–639.

7. Rambjor, G.S., Walen, A.I., Windsor, S.L., and Harris, W.S. (1996)Eicosapentaenoic Acid Is Primarily Responsible for the Hypotriglyc-eridemic Effect of Fish Oil in Humans, Lipids 31, S45–S49.

8. Otto, D.A., Tsai, C.E., Baltzell, J.K., and Wooten, J.T. (1991) Ap-parent Inhibition of Hepatic Triacylglycerol Secretion, Independentof Synthesis, in High-Fat Fish Oil-Fed Rats: Role for Insulin,Biochim. Biophys. Acta 1082, 37–48.

9. Lang, C.A., and Davis, R.A. (1990) Fish Oil Fatty Acids ImpairVLDL Assembly and/or Secretion by Cultured Rat Hepatocytes, J.Lipid Res. 31, 2079–2086.

10. Brown, A.M., Castle, J., Hebbachi, A.M., and Gibbons, G.F. (1999)Administration of n-3 Fatty Acids in the Diets of Rats or Directlyto Hepatocyte Cultures Results in Different Effects on Hepatocel-lular ApoB Metabolism and Secretion, Arterioscler. Thromb. Vasc.Biol. 19, 106–114.

11. Zheng, X., Avella, M., and Botham, K.M. (2001) Comparison ofthe Effects of Dietary n-3 and n-6 Polyunsaturated Fatty Acids onVery Low-Density Lipoprotein Secretion When Delivered to He-patocytes in Chylomicron Remnants, Biochem. J. 357, 481–487.

12. Haagsman, H.P, de Haas, C.G., Geelen, M.J., and van Golde, L.M.(1982) Regulation of Triacylglycerol Synthesis in the Liver. Modu-lation of Diacylglycerol Acyltransferase Activity in vitro, J. Biol.Chem. 257, 10593–10598.

13. Rustan, A.C., Nossen, J.Ø., Christiansen, E.N., and Drevon, C.A.(1988) Eicosapentaenoic Acid Reduces Hepatic Synthesis and Se-cretion of Triacylglycerol by Decreasing the Activity of Acyl-Coenzyme A:1,2-Diacylglycerol Acyltransferase, J. Lipid Res. 29,1417–1426.

14. Berge, R.K., Madsen, L., Vaagenes, H., Tronstad, K.J., Göttlicher,M., and Rustan, A.C. (1999) In Contrast with DocosahexaenoicAcid, Eicosapentaenoic Acid and Hypolipidaemic Derivatives De-crease Hepatic Synthesis and Secretion of Triacylglycerol by De-creased Diacylglycerol Acyltransferase Activity and Stimulation ofFatty Acid Oxidation, Biochem. J. 343, 191–197.

15. Madsen, L., Rustan, A.C., Vaagenes, H., Berge, K., Dyrøy, E., andBerge, R.K. (1999) Eicosapentaenoic and Docosahexaenoic AcidAffect Mitochondrial and Peroxisomal Fatty Acid Oxidation in Re-lation to Substrate Preference, Lipids 34, 951–963.

16. Kendrick, J.S., and Higgins, J.A. (1999) Dietary Fish Oils InhibitEarly Events in the Assembly of Very Low Density Lipoproteinsand Target ApoB for Degradation Within the Rough EndoplasmicReticulum of Hamster Hepatocytes, J. Lipid Res. 40, 504–514.

17. Frøyland, L., Vaagenes, H., Asiedu, D.K., Garras, A., Lie, Ø., andBerge, R.K. (1996) Chronic Administration of EicosapentaenoicAcid and Docosahexaenoic Acid as Ethyl Esters Reduced PlasmaCholesterol and Changed the Fatty Acid Composition in Rat Bloodand Organs, Lipids 31, 169–178.

LIPID ACCUMULATION AND SECRETION BY ATLANTIC SALMON HEPATOCYTES 485

Lipids, Vol. 40, no. 5 (2005)

18. Frøyland, L., Madsen, L., Vaagenes, H., Totland, H., Auwerx, J.,Kryvi, H., Staels, B., and Berge, R.K. (1997) Mitochondrion Is thePrincipal Target for Nutritional and Pharmacological Control ofTriglyceride Metabolism, J. Lipid Res. 38, 1851–1858.

19. Nossen, J.Ø., Rustan, A.C., Gloppestad, S.H., Målbakken, S., andDrevon, C.A. (1986) Eicosapentaenoic Acid Inhibits Synthesis andSecretion of Triacylglycerols by Cultured Rat Hepatocytes,Biochim. Biophys. Acta 879, 56–65.

20. Berge, R.K., Skorve, J., Tronstad, K.J., Berge, K., Gudbrandsen,O.A., and Grav, H. (2002) Metabolic Effects of Thia Fatty Acids,Curr. Opin. Lipidol. 13, 295–304.

21. Skrede, S., Bremer, J., Berge, R.K., and Rustan, A.C. (1994) Stim-ulation of Fatty Acid Oxidation by a 3-Thia Fatty Acid ReducesTriacylglycerol Secretion in Cultured Rat Hepatocytes, J. Lipid Res.35, 1395–1404.

22. Grisdale-Helland, B., Ruyter, B., Rosenlund, G., Obach, A., Hel-land, S.J., Sandberg, M.G., Standal, H., and Røsjø, C. (2002) Influ-ence of High Contents of Dietary Soybean Oil on Growth, FeedUtilization, Tissue Fatty Acid Composition, Heart Histology andStandard Oxygen Consumption of Atlantic Salmon (Salmo salar)Raised at Two Temperatures, Aquaculture 207, 311–329.

23. Seglen, P.O. (1976) Preparation of Isolated Rat Liver Cells, Meth-ods Cell Biol. 1, 29–59.

24. Dannevig, B.H., and Berg, T. (1985) Endocytosis of Galactose-Ter-minated Glycoproteins by Isolated Liver Cells of Rainbow Trout,Comp. Biochem. Physiol. 82B, 683–688.

25. Lowry, O.H., Rosebrough, N.J., Farr, A.L., and Randall, R.J.(1951) Protein Measurement with the Folin Phenol Reagent, J. Biol.Chem. 193, 265–275.

26. Chung, B.H., Wilkinson, T., Geer, J.C., and Segrest, J.P. (1980)Preparative and Quantitative Isolation of Plasma Lipoproteins:Rapid, Single Discontinuous Density Gradient Ultracentrifugationin a Vertical Rotor, J. Lipid Res. 21, 284–291.

27. Burnette, W.N. (1981) ‘Western Blotting’: Electrophoretic Transferof Proteins from Sodium Dodecylsulfate–Polyacrylamide Gels toUnmodified Nitrocellulose and Radiographic Detection with Anti-body and Radioiodinated Protein A, Anal. Biochem. 112, 195–203.

28. Folch, J., Lees, M., and Sloane-Stanley, G.M. (1957) A SimpleMethod for Isolation and Purification of Total Lipids from AnimalTissues, J. Biol. Chem. 226, 497–509.

29. Ruyter, B., Røsjø, C., Grisdale-Helland, B., Rosenlund, G., Obach,A., and Thomassen, M.S. (2003) Influence of Temperature andHigh Dietary Linoleic Acid Content on Esterification, Elongation,and Desaturation of PUFA in Atlantic Salmon Hepatocytes, Lipids38, 833–840.

30. Hexeberg, S., Frøyland, L., Asiedu, D.K., Demoz, A., and Berge,R.K. (1995) Tetradecylthioacetic Acid Reduces the Amount ofLipid Droplets, Induces Megamitochondria Formation and In-creases the Fatty Acid Oxidation in Rat Heart, Mol. Cell. Cardiol.27, 1851–1857.

31. SAS (1990) SAS/STAT User’s Guide, SAS Institute, Cary, NC,1686 pp.

32. Hardy, R.W., Scott, T.M., and Harrell, L.W. (1987) Replacementof Herring Oil with Menhaden Oil, Soybean Oil, or Tallow in theDiets of Atlantic Salmon Raised in Marine Net-Pens, Aquaculture65, 267–277.

33. Thomassen, M.S., and Røsjø, C. (1989) Different Fats in Feed forSalmon: Influence on Sensory Parameters, Growth Rate and FattyAcids in Muscle and Heart, Aquaculture 79, 129–135.

34. Greene, D.H.S., and Selivonchick, D.P. (1990) Effects of DietaryVegetable, Animal and Marine Lipids on Muscle Lipid and Hema-tology of Rainbow Trout (Oncorhynchus mykiss), Aquaculture 89,165–182.

35. Finstad, B., and Thomassen, M.S. (1991) Does Dietary Lipid Com-position Affect the Osmoregulatory Ability of Rainbow Trout (On-corhynchus mykiss) at High and Low Temperatures? Comp.Biochem. Physiol. 99A, 463–471.

36. Willumsen, N., Vaagenes, H., Lie, Ø., Rustan, A.C., and Berge,

R.K. (1996) Eicosapentaenoic Acid, but Not DocosahexaenoicAcid, Increases Mitochondrial Fatty Acid Oxidation and Upregu-lates 2,4-Dienol-CoA Reductase Gene Expression in Rats, Lipids31, 579–592.

37. Chazan, J.B. (1996) Pharmacological and Dietetic Properties of Oilsand Fats, in Oils and Fats Manual, (Karleskind, A., ed.), Vol. 1, pp.675–702, Intercept Limited, Andover, Hants., United Kingdom.

38. Sébédio, J.L. (1996) Marine Oils, in Oils and Fats Manual, (Karle-skind, A., ed.), Vol. 1, pp. 266–276, Intercept Limited, Andover,Hants., United Kingdom.

39. Mastellone, I., Polichetti, E., Gres, S., de la Maisonneuve, C.,Domingo, N., Marin, V., Lorec, A.M., Farnarier, C., Portugal, H.,Kaplanski, G., and Chanussot, F. (2000) Dietary Soybean Phos-phatidylcholines Lower Lipidemia: Mechanisms at the Levels ofIntestine, Endothelial Cell, and Hepato-biliary Axis, J. Nutr.Biochem. 11, 461–466.

40. LeBlanc, M.J., Brunet, S., Bouchard, G., Lamireau, T., Yousef,I.M., Gavino, V., Levy, E., and Tuchweber, B. (2003) Effects ofDietary Soybean Lecithin on Plasma Lipid Transport and HepaticCholesterol Metabolism in Rats, J. Nutr. Biochem. 14, 40–48.

41. Madani, S., Frenoux, J.M., Prost, J., and Belleville J. (2004)Changes in Serum Lipoprotein Lipids and Their Fatty Acid Com-positions and Lipid Peroxidation in Growing Rats Fed SoybeanProtein Versus Casein With or Without Cholesterol, Nutrition 20,554–563.

42. Aarsland, A., Lundquist, M., Borretsen, B., and Berge, R.K. (1990)On the Effect of Peroxisomal β-Oxidation and Carnitine Palmitoyl-transferase Activity by Eicosapentaenoic Acid in Liver and Heartfrom Rats, Lipids 25, 546–548.

43. Martin, L.J., Reaidi, G.B., Gavino, G.R., and Gavino, V.C. (1991)Effect of 4,7,10,13,16,19-Docosahexaenoic Acid on TriglycerideAccumulation and Secretion in Rat Hepatocytes in Culture, Lipids26, 374–380.

44. Wang, H.X., Chen, X., and Fisher, E.A. (1993) n-3 Fatty AcidsStimulate Intracellular Degradation of Apoprotein B in Rat Hepa-tocytes, J. Clin. Invest. 91, 1380–1389.

45. Botham, K.M., Zheng, X., Napolitano, M., Avella, M., Cavallari,C., Rivabene, R., and Bravo, E. (2003) The Effects of Dietary n-3Polyunsaturated Fatty Acids Delivered in Chylomicron Remnantson the Transcription of Genes Regulating Synthesis and Secretionof Very Low-Density Lipoprotein by the Liver: Modulation by Cel-lular Oxidative State, Exp. Biol. Med. 228, 143–151.

46. Totland, G.K., Madsen, L., Klementsen, B., Vaagenes, H., Kryvi,H., Frøyland, L., Hexeberg, S., and Berge, R. (2000) Proliferationof Mitochondria and Gene Expression of Carnitine Palmitoyltrans-ferase and Fatty Acyl-CoA Oxidase in Rat Skeletal Muscle, Heartand Liver by Hypolipidemic Fatty Acids, Biol. Cell 92, 317–329.

47. Ruyter, B., and Thomassen, M.S. (1999) Metabolism of n-3 and n-6 Fatty Acids in Atlantic Salmon Liver: Stimulation by EssentialFatty Acid Deficiency, Lipids 34, 1167–1176.

48. Ruyter, B., Røsjø, C., Einen, O., and Thomassen, M.S. (2000) Es-sential Fatty Acids in Atlantic Salmon: Time Course of Changes inFatty Acid Composition of Liver, Blood and Carcass Induced by aDiet Deficient in n-3 and n-6 Fatty Acids, Aquacult. Nutr. 6,109–117.

49. Bell, J.G., McVicar, A.H., Park, M.T., and Sargent, J.R. (1991)High Dietary Linoleic Acid Affects the Fatty Acid Compositionsof Individual Phospholipids from Tissues of Atlantic Salmon(Salmo salar): Association with Stress Susceptibility and CardiacLesion, J. Nutr. 121, 1163–1172.

50. Bell, J.G., Dick, J.R., McVicar, A.A., Sargent, J.R., and Thomp-son, K.D. (1993) Dietary Sunflower, Linseed and Fish Oils AffectPhospholipid Fatty Acid Composition, Development of Cardiac Le-sions, Phospholipase Activity and Eicosanoid Production in At-lantic Salmon (Salmo salar), Prostaglandins Leukotrienes Essent.Fatty Acids 49, 665–673.

[Received July 7, 2004; accepted March 24, 2005]

486 A. VEGUSDAL ET AL.

Lipids, Vol. 40, no. 5 (2005)