Supranormal Electroretinogram in F at-1 Mice with Retinas Enriched in Docosahexaenoic Acid and n-3 Very Long Chain Fatty Acids (C24–C36)

Abstract

purpose. Fat-1 mice can convert n-6 to n-3 fatty acids endogenously, resulting in the accumulation of n-3 fatty acids in major tissues. This was a study of how this conversion affects the major fatty acid found in retina, n-3 docosahexaenoic acid (DHA), the very long chain fatty acids (VLCFA, C24–C36), and retinal function.

methods. Both wild-type (WT) and fat-1 mice were fed a modified AIN-93G diet containing 10% safflower oil, high in 18:2n-6. Fatty acid composition of individual phospholipids was analyzed in total lipid extracts from whole eyes excluding the lens. Retinal function and levels of proteins involved in cellular stress were assessed with full field electroretinogram (ERG) recordings and immunohistochemistry, respectively.

results. Compared with WT mice, DHA levels in fat-1 mice increased two to five times in all phospholipid classes, whereas n-6 fatty acid levels decreased. Levels of C32 and C34 n-3 pentaenoic and hexaenoic VLCFA in phosphatidylcholine increased whereas n-6 VLCFAs were depleted. Scotopic and photopic ERGs showed unusually high amplitudes for both a- and b-waves and lower thresholds in fat-1 mice. Glial fibrillary acidic protein (GFAP) and carboxyethylpyrrole (CEP, protein adducts produced from DHA oxidation) were respectively increased in Müller cells and photoreceptors of fat-1 mice.

conclusions. Highly enriched DHA and n-3 VLCFA in the retina lead to supernormal scotopic and photopic ERGs and increases in Müller cell reactivity and oxidative stress in photoreceptors. The regulation of n-3 fatty acids levels and of the n-6/n-3 fatty acid ratio are essential in preserving retinal integrity.

The molecular architecture of the photon-capturing site in the retina (the outer segment; OS) is unique, with its high phospholipid content. ^{1 2} These membrane phospholipids are heavily acylated with polyunsaturated fatty acids (PUFAs), which comprise more than 50% of all fatty acids in the OS membrane. ³ The most abundant PUFA is docosahexaenoic acid (DHA; C22:6n-3). The OS also contains significant amounts of polyenoic very long chain fatty acids (VLCFAs) with carbon chain lengths encompassing C24 to C36. ^{4 5} Increases in DHA and n-3 VLCFA and decreases in arachidonic acid (AA, C20:4n-6) and n-6 VLCFA in OS are characteristic developmental changes in the OS. ⁶ Balance in the levels of these fatty acids is thought to play a role in retinal function. 

DHA was shown to prevent photoreceptor cell apoptosis and to promote photoreceptor cell differentiation by acting as a trophic factor in vitro. ^{7 8 9} Furthermore, DHA is a precursor of a potent neurotrophic factor, neuroprotectin D1 (NPD1), which protects photoreceptors against injury-induced oxidative stress and also enhances retinal pigment epithelial (RPE) cell survival. ^{10 11} DHA-derived NPD1, resolvin D1 (RvD1), and eicosapentaenoic acid (EPA, C20:5n-3)-derived resolvinE1 (RvE1) also potentially protect against retinal angiogenesis. ¹² Deficiency in DHA, due to diets with an unbalanced n-6 to n-3 ratio, leads to impaired rod cell renewal and abnormal electroretinogram (ERG) responses in animals and infants. ^{13 14 15} We have shown that small amounts of dietary DHA alter the retina lipid composition and are accompanied by changes in rhodopsin content, rhodopsin phosphorylation, and rhodopsin kinetics in rats. ^{4 6} These studies strongly indicate the importance of n-3 fatty acid balance in normal cellular and physiological function of the retina. However, it is unclear how much and what balance of dietary n-3 fatty acids are required as an intake and as the final retinal concentration for optimal retina function. 

Transgenic fat-1 mice developed by Kang et al. ¹⁶ convert n-6 PUFA to n-3 PUFA, thereby accumulating n-3 PUFA levels in many organs and tissues. ^{16 17} These mice express the fat-1 (f atty acid metabolism 1) gene from Caenorhabditis elegans, ¹⁸ which encodes for an n-3 fatty acid desaturase, absent in mammals, that can synthesize n-3 PUFA from n-6 PUFA. The promoter driving the expression of fat-1 is chicken β-actin, which was chosen because it yields high-level and ubiquitous expression of the transgene in mice. We used this model to examine the impact of partial replacement of n-6 by n-3 fatty acids on the retina. Our study complements previous studies that used dietary manipulation to produce n-3 fatty acid–deficient animal models. We characterized the fatty acid profiles in individual phospholipids of fat-1 transgenic mice retina and the impact of these profiles on retina function by recording the ERG. 

Materials and Methods

Animals and Diets

Fat-1 mice on a mixed C57Bl/6xC3H background were originally obtained from author JXK. Fat-1 mice on a mixed C57Bl/6xC3H background are maintained through brother and sister matings. For experiments, fat-1 mice were mated with C57Bl/6 female mice (Charles River Laboratories, Saint-Constant, QC, Canada). Wild-type (WT) and fat-1 mice were fed a modified AIN-93G diet containing 10% safflower oil high in the n-6 PUFA, linoleic acid (LA, 18:2n-6; product no. D04092701; Research Diets, New Brunswick, NJ). All mice were housed in a temperature- and humidity-controlled animal facility with a 12-hour light/dark cycle. At 12 weeks of age, male WT mice and fat-1 mice were culled by CO₂ asphyxiation. The eyes were excised, snap frozen in liquid nitrogen and kept at −80°C in DWLM’s laboratory (University of Toronto) until shipping on dry ice to MS’s laboratory (University of Manitoba). Age-matched WT and fat-1 mice, raised in DWLM’s laboratory, were shipped to the University of Manitoba for ERG recordings. All mice were continuously fed the diet until culling. The respective Animal Ethics Committees of the University of Manitoba and of the University of Toronto approved all aspects of the study. Animal care procedures were based on guidelines described in the Canadian Council for Animal Care and adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Lipid Analysis

Six to eight eyes were pooled for lipid analysis. Because of the frozen condition, total lipids were extracted from total eyes without the lens by the method described by Folch et al. ¹⁹ Individual phospholipids were separated on hexane-washed silica-gel H-plates (20 × 20 cm), which were developed in a solvent system composed of chloroform/methanol/2-propanol: 0.25% (wt/vol) KCl/thiethylamine (30:9:25:6:18, by volume) as described by Touchstone et al. ²⁰ The plates were air dried at room temperature for 5 minutes, then sprayed with 0.1% (wt/vol) 8-anilino-1-naphthalene-sulfonic acid (ANSA), to enable visualization under UV light. Each phospholipid band was scraped and kept at −80°C until methylation. 

Samples were prepared by adding 2 mL hexane and 1.5 mL 14% boron trifluoride in methanol to methylate the fatty acids (FAME). ²¹ Separation of FAME was performed on a capillary column (35 m × 25 mm inner diameter with 0.25-μm thickness; BPX70; SGE, Austin, TX) and a chromatograph system (GC17A; Phenomenex Shimadzu, Torrance, CA). Hydrogen was used as the carrier gas with nitrogen as the make-up gas, under a flow rate of 2.5 mL/min. The temperature program profile adhered to a previously published protocol. ⁴ Injection temperature was 275°C, and detector temperature was 320°C. Peaks were identified by using standard FAME 461 (Nucheck Prep, Inc., Minneapolis, MN) for PUFA. The standard for VLCFA was prepared in MS’s laboratory using pig retinas. 

Electroretinography

For the ERG measurement, wild-type mice (C57Bl/6) fed commercial chow (no. 5001 LabDiet, St. Louis, MO) were added as an additional comparison group (WT chow). ERG responses were recorded bilaterally. One eye per animal was studied, and the eye yielding the highest mixed scotopic a-wave amplitude was selected for each animal, from the following three groups: WT chow-fed mice (n = 4 eyes); experimental n-6 diet-fed WT mice (WT n-6, n = 6 eyes); and experimental n-6 diet-fed fat-1 mice (fat-1, n = 7 eyes). After a minimum of 1-hour dark adaptation, the animals were prepared under dim red light for the recordings. In mice under anesthesia with a mixture of ketamine (62.5 mg/kg IP) and xylazine (12.5 mg/kg IP), body temperature was maintained at 38°C with a homeothermic blanket system, the pupils were fully dilated with 1% tropicamide, and a drop of methylcellulose was placed on the cornea to prevent corneal dehydration and make electrical contact with gold loop electrodes without exerting any pressure on the cornea. Platinum needles (25-gauge; Grass Telefactor, West Warwick, RI) implanted subcutaneously behind each eye served as reference electrodes. The ground electrode consisted of a platinum needle inserted subcutaneously behind the neck. Amplification (at 1–1000-Hz band-pass, without notch filtering), stimulus presentation, and data acquisition were provided by an integrated system (UTAS-4000; LKC Technologies, Gaithersburg, MD). The mixed scotopic ERG was recorded as described elsewhere. ²² In brief, stimuli consisted of single white (6500 K, xenon bulb) flashes (10 μs duration), repeated three to five times to verify the responsiveness reliability. For intensity responses, stimuli were presented at 16 increasing intensities varying from –3.7 to 2.9 log cd · s/m² in luminance. To allow for maximum rod recovery between consecutive flashes, interstimulus intervals were increased (as the stimulus intensities were progressively increased) from 10 seconds at lowest stimulus intensity up to 2 minutes at highest stimulus intensity. The amplitude of the b-wave was measured from the a-wave negative peak relative to the b-wave positive apex, and not up to the peak of oscillatory potentials, which can exceed the b-wave apex. After scotopic recordings, animals were light adapted for 10 minutes (30 cd/m² background) before the photopic intensity responses were recorded with stimulus intensities ranging from −1.6 to 2.9 log cd · s/m². Finally, the flicker ERG was recorded with a flash intensity of 1.37 log cd · s/m² presented at frequencies ranging from 5 to 40 Hz at each 5-Hz incremental step. Criterion amplitudes were set at 20 μV for a- and b-waves and at 5 μV for flicker amplitudes. Therefore, the minimal light intensity required for a component to exceed the criterion amplitude was considered as the threshold. 

Primary Antibodies for Immunohistochemistry

Two types of primary antibodies were used mouse monoclonal anti-glial fibrillary acidic protein (GFAP; Covance, Princeton, NJ; SMI-22R, clone SMI-22; 1:1000 dilution) and rabbit anti-carboxyethylpyrrole (CEP) antibody to CEP adducts from DHA (1:100 dilution; generous gift from Roger G. Salomon, Case Western Reserve University). The monoclonal anti-GFAP antibody cocktail (raised against GFAP derived from the Bigner-Eng clones MAb1B4, MAb2E1, and MAb4A11) labels a major band at ∼50-kDa on Western blot analysis of mouse retina. ²³ This antibody labels retinal glia as previously reported in the mouse. ^{23 24} The polyclonal anti-CEP antibody was raised in rabbits against mouse CEP-KLH (keyhole limpet hemocyanin) and the IgG fraction of anti-CEP antibodies was purified using immobilized protein G. ²⁵ Western blot analysis of anti-CEP antibodies (1 μg/mL) on dissected human Bruch’s membrane and RPE choroid tissues show multiple bands corresponding to the various CEP protein adducts. ²⁶ Studies involving CEP immunostaining with the same antibody (provided by Roger Salomon) in the mouse retina ²⁵ have reported intense labeling in photoreceptor OS and RPE. Less intense staining was evident in the inner plexiform layer (IPL), and little if any staining was seen in the outer limiting membrane (OLM), outer nuclear layer (ONL), outer plexiform layer (OPL), inner nuclear layer (INL), or ganglion cell layer (GCL). 

Histologic Procedure on Retinal Cross-Sections

Immediately after pentobarbital-induced euthanasia, the eyes of fat-1, WT n-6, and WT chow mice were excised and fixed overnight with a solution of 4% paraformaldehyde (pH 7.4). After removal of the cornea and lens, the eyes were cryoprotected through immersion in a series of 10%, 20%, and 30% sucrose, then cryosectioned at a 20-μm thickness. After extensive washing in TBS (0.1 M; pH 7.5; three times for 10 minutes each), the sections were blocked for 2 hours in a medium containing PBS 0.1 M + 0.3% Triton X-100 + 10% goat (or horse) serum, and reacted overnight with primary antibodies diluted in a 1:10 solution of the previous blocking medium. On the following day, the sections were washed with TBS, blocked again for 1 hour, exposed for 2 hours to appropriate secondary antibodies (diluted to 1:500 in a 1:10 solution of the blocking medium), washed extensively in TBS, covered with a DAPI (4′,6-diamidino-2-phenylindole-dihydrochloride-)-containing antifade solution (ProLong gold antifade reagent, P36939; Molecular Probes, Eugene, OR), and coverslipped. Each primary antibody was applied alone on a minimum of 12 retinal sections (placed on a minimum of three separate slides). Secondary antibodies were goat anti-rabbit-Alexa 488, goat anti-mouse-Alexa 594, and donkey anti-goat-Alexa 594 (Molecular Probes). Control labeling without primary antibody remained negative. All reactions were run at room temperature. 

For Nissl staining, the retina cryosections were stained in 0.1% cresyl violet solution for 3 to 10 minutes at room temperature, and after excess stain was removed, the sections were dehydrated in solutions containing increasing concentrations of ethanol for 5 minutes. The sections were dehydrated in 95% ethyl alcohol for 5 minutes and then dehydrated in 100% ethanol two times for 5 minutes each, cleared in xylene two times for 5 minutes each and mounted on glass slides. 

Imaging and Data Analysis

Representative samples were imaged for illustrations with a laser confocal microscope (LSM 510 Axiovert 100M; Carl Zeiss Meditec, Inc., Dublin, CA), and processed with image-analysis software (Photoshop 6.0 software; Adobe, San Jose, CA) to adjust contrast levels if required. 

Statistical Analysis

The effects of genotype on phospholipid fatty acid composition and ERG results were analyzed by one-way analysis of variance (ANOVA; SAS 9.1; SAS, Cary, NC). Significant effects of treatment for ERG components were defined by using the Mann-Whitney U test for comparing maximum values. Fatty acid data are expressed as the mean ± SD. ERG results are expressed as the mean ± SEM. Statistical significance was set at P < 0.05. 

Results

Fatty Acid Composition in Phospholipids

The level of DHA in fat-1 mice was significantly (P < 0.05) increased in all phospholipids when compared to WT n-6 mice: phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylserine (PS), and phosphatidylinositol (PI), 2.6, 2.5, 2.3, and 4.7 times (P < 0.05), respectively (Table 1 ; Fig. 1 ). In contrast, levels of the major n-6 PUFA, AA, was decreased by 10% to 42% compared to levels in WT n-6 mice (Table 1 ; Fig. 1 ). Specific increases of C22:5n-3, C22:6n-3, C24:5n-3, and C24:6n-3 (PC and PS only) were identified in all phospholipids while most showed decreases in C20:4n-6, C22:4n-6, C22:5n-6, C24:4n-6 and C24:5n-6. The resulting n-6 to n-3 PUFA ratio was less than 1.0, with the exception of PI: 0.9, 0.6, 0.5, and 2.0 for PC, PE, PS, and PI, respectively, which was five to six times less than in WT n-6 mice. There were no overall differences in unsaturated index (UI) in all phospholipids, since the increase of n-3 fatty acids in fat-1 mice was reciprocally replaced by n-6 fatty acids in WT, suggesting no changes in membrane fluidity. The total of saturated and monounsaturated fatty acid levels in all individual phospholipid fractions remained unchanged in fat-1 compared with WT n-6 mice. 

VLCFA containing carbon chain lengths of up to C24, -32, -34, and -36 were found in the PC fraction. The n-3 series of penta- and hexaenoic C32 and C34 VLCFA levels increased significantly, whereas n-6 tetra- and pentaenoic VLCFA levels decreased (Table 1) . This finding suggests that normal elongation and desaturation takes place in fat-1 mice. The intermediate metabolites, C26–C30 VLCFA were not detected, possibly because of insufficient sample quantity. 

Electroretinograms

ERG responses were recorded to compare retinal function between fat-1, WT n-6, and WT chow. Shown in Figures 2a 2b 2care representative ERG traces recorded in fat-1 and WT n-6 fed mice. Mixed scotopic response (Fig. 2a) , photopic response (Fig. 2b) , and photopic flicker (Fig. 2c)in the fat-1 mice had unusually higher amplitudes in comparison to those in the WT n-6 mice. In addition, WT chow mice had even lower ERG response amplitudes (Figs. 2d 2e 2f 2g) . All three types of ERG intensity response curves and the flicker frequency response curves showed clear differences between the three experimental groups. Thresholds were lower for both mixed scotopic (a- and b-waves) and photopic (b-waves) responses in fat-1 mice compared with WT n-6 mice and WT chow mice. Fat-1 mice had higher amplitudes for scotopic a- and b- wave, photopic b-wave, and flicker amplitude, and fusion, followed by WT n-6 mice and WT chow mice. The amplitudes of pure scotopic rod b-waves (elicited by −2.0 log cd · s/m², the intensity just below the one eliciting a-waves) were elevated (P < 0.05) in fat-1 mice compared with WT n-6 mice (133% ± 30%) and WT chow mice (175% ± 42%). The amplitudes of saturated mixed scotopic a-waves were increased (P < 0.05) in fat-1 mice compared with WT n-6 mice (145% ± 28%) and WT chow mice (224% ± 45%; Fig. 2h ). Similar amplitude elevations (P < 0.05) were seen for the saturated mixed scotopic b-wave: 136% ± 28% and 209% ± 42% in fat-1 mice compared with WT n-6 mice and WT chow mice, respectively (Fig. 2h) . The total amplitude of all oscillatory potentials (isolated with 60–300 Hz band-pass), for the saturated mixed scotopic responses, was higher in fat-1 mice than in WT chow mice (188% ± 13%; P < 0.05) but not higher than in WT n-6 mice (111% ± 11%; Fig. 2h ). Saturated photopic b-waves were increased by 160% ± 36% and 245% ± 79% in fat-1 mice compared with WT n-6 mice (P < 0.05) and WT chow mice (P < 0.05), respectively (Fig. 2h) . The total amplitude of the oscillatory potentials, for the saturated photopic responses, was higher (P < 0.05) in fat-1 mice than in both WT n-6 mice (159% ± 20%) and WT chow mice (243% ± 58%). The flicker amplitude at 20 Hz was higher (P < 0.05) in fat-1 mice than in WT n-6 mice (133% ± 17%) and WT chow mice (316% ± 68%). The flicker fusion was also higher in fat-1 mice than in WT chow mice (P < 0.05); however, it was not different from WT n-6 mice. 

Finally, implicit times were shorter for both fat-1 and WT n-6 mice when compared with WT chow mice (P < 0.05). The respective implicit times, recorded at saturation levels, for fat-1, WT n-6, and WT chow mice were: 6.0 ± 0.2, 6.0 ± 0.6, and 8.6 ± 0.6 ms for the mixed scotopic a-wave; 32 ± 5, 31 ± 5, and 48 ± 6 ms for the mixed scotopic b-wave; and 34 ± 2, 36 ± 5, and 43 ± 5 ms for the photopic b-waves. 

Müller Cell Reactivity

The levels and distribution of GFAP, a marker of Müller cell reactivity, were examined immunohistochemically. In WT chow mice, GFAP-labeled Müller cells were detected only in the extreme periphery of the retina (marginal zone), with processes encompassing the OPL, to the GCL (Fig. 3) . In opposition to the WT chow mice, strongly GFAP-labeled Müller cell processes were seen to extend radially between the OPL to the GCL throughout the whole retina, with equal extent and strength of labeling in the center versus periphery in fat-1 mice. In addition, in fat-1 mice, tangential processes were seen to extend along the OPL, both in the center and periphery of the retina; such processes were seen only in the periphery of the WT chow mice, and these were less dense and more scattered. Finally, the inner limiting membrane showed constitutive labeling in both groups; these cells and processes correspond to astrocytes. 

CEP-Modified Proteins

The distribution of CEP (protein adducts produced by DHA oxidation) was assessed immunohistochemically. Both WT chow and fat-1 mice retina exhibited CEP immunoreactivity in the photoreceptor inner and OS. In WT chow mice, CEP was evenly distributed along tightly packed IS and OS (Fig. 4a) . In fat-1 mice (Fig. 4b) , CEP-labeling revealed disorganized, and swollen segments, with a gap in staining at the level of the cilium (the structure linking the IS and OS; see arrow). The end of the inner segment was swollen into an ampulla shape just before the cilium. The OS were often tortuous and appeared broken into fragments. 

Gross Anatomy of the Retina

Despite clear phenotypes related to GFAP and CEP expression, the overall retina of fat-1 mice did not display any gross morphologic defects in comparison to WT n-6 and WT chow retinas, when examined using Nissl staining (Figs. 4c 4d) . The stratification had the same appearance. The number of rows in each nuclear layer (ONL, INL, and GCL) was not different between WT chow, WT n-6 and fat-1 mice retinas. The thickness of all layers was similar in all the three groups studied. To make sure that the sections quantified were cut at the same angles (to avoid diagonally cut sections), ascertained that the ONL-INL thickness ratios were the same in sections from all groups. 

Discussion

In the present study, we examined the effect of enriched n-3 fatty acids on retina function using fat-1 transgenic mice fed n-6 fatty acid. Fat-1 mice accumulated high levels of n-3 PUFA, especially DHA, in the retina by converting n-6 into n-3 PUFA. This increase resulted in a significantly low ratio of n-6/n-3 PUFA—in the range of 0.5 to 0.9 in major phospholipids—but no overall change in membrane fluidity, as the UI was similar between two animal groups (Table 2) . This result may be due to a reciprocal replacement of these two types of fatty acids. The low ratio of 0.5 to 0.6 in diet-induced or in fat-1 mice has been shown to have a protective effect against pathologic angiogenesis after oxygen-induced retinopathy in 17-day-old mice. ¹² This indicates that the balance of n-6/n-3 PUFA is an important factor in ocular neovascularization leading to blindness. In this regard, fat-1 mice are a useful model for exploring the impact of DHA and n-6/n-3 PUFA ratio on retina functional integrity without having to supplement dietary n-3 fatty acids. 

Fat-1 mice also had significantly increased n-3 VLCFA retina levels, especially C32 and C34 VLCFA in PC, which is highly concentrated in OS. ^{4 5} Deficiency in these VLCFA may lead to macular degeneration or macular dystrophy such as Stargardt-like (STGD3) disease leading to blindness. Mutations in the gene encoding for an enzyme involved in elongation of VLCFA, called ELOVL4, are responsible for this autosomal dominant disease. ^{27 28} The recently developed STGD3-knockin mice carrying ELOVL4 mutation were shown to have a selective deficiency in C32–C36 VLCFA in their retina, ²⁹ providing evidence for the involvement of VLCFA in macular pathophysiology. Mice expressing the STGD3 human mutated ELOVL4 gene (encoding a truncated protein) showed declines in maximum amplitudes of rod and cone b-wave. ³⁰ Unlike these mice models, fat-1 mice accumulate very high n-3 VLCFA. Macdonald et al., ³¹ studied the potential benefit of DHA supplementation (20 mg/kg body weight) in a patient with STGD3. The treatment led to improved visual acuity as well as foveal and parafoveal function measured with the multifocal electroretinogram (mfERG). These results were correlated with increased plasma DHA levels. These preliminary results indicate that DHA may have a beneficial effect in STGD3 even though the mutated gene encodes for an enzyme involved in the elongation of fatty acids longer than DHA itself. The potential therapeutic effect of DHA might involve neuroprotection rather than n-3 VLCFA replenishment. ^{10 11}  

Retinal Function

The present study shows that highly enriched DHA and VLCFA in phospholipids of retina led to markedly high scotopic and photopic ERG amplitudes. This phenomenon is linked with the specific elevation in n-3 PUFA and not n-6 PUFA levels since the overall membrane fluidity, as evidenced by the UI shown in Table 2 , was not affected when similar amounts of dietary n-6 fatty acids are fed to fat-1 and WT mice. However, the amount of total fat in the diet may also affect retina function, since WT chow mice (5% fat, wt/wt) showed the lowest amplitudes in comparison to fat-1 mice and WT n-6 mice (10% fat, wt/wt). The results in the present study also imply that the rods, cones, and inner neural cells are not lost in fat-1 mice. This observation is supported by the many studies showing that loss of rods and cones leads to decreased a- and b-wave amplitudes. ^{29 32} The a- and b-wave implicit times were reduced in fat-1 mice compared with the two other groups. This correlates with the fact that n-3 PUFA-deficient animals have prolonged a- and b-wave implicit times. ³³ Sprague-Dawley rats fed a high-fat diet (20%, wt/wt) containing 5.0% dietary n-3 fatty acids ³⁴ showed compatible levels of DHA in individual phospholipids of retina as found in fat-1 mice, but no functional measures were performed. Overall, this present study strongly supports a clear relationship between n-3 PUFA and retinal function. However, a longitudinal study of fat-1 mice will be essential to determine whether the unusually high amplitude ERG response in fat-1 mice could be a sign of retina disease, as described by Heckenlively et al. ³⁵  

Since the ERG is a voltage measure, it is proportional to the current and resistance along the pathway formed between the active electrode on the cornea and the subdermal reference electrode placed in the temporal aspect of the eye. Therefore, it cannot be excluded that changes in fatty acid composition of the retina in fat-1 mice could lead to an increase in the resistance along this electrical pathway, thereby resulting in an increase in ERG amplitudes. In vitro measures of retina resistance must be considered in future experiments. Changes in pupil size and eye sizes could also affect ERG amplitudes. However, both of these variables were identical between groups. As for eye size, no differences were seen between fat-1, WT chow, and WT n-6 mice. Other variables known to influence ERG amplitudes, such as body temperature, corneal or lens opacity, and pressure from the electrode on the cornea, were all carefully controlled. There was no opacity of the cornea or lens, and the recording electrode never made physical contact with the cornea. Although changes in resistance can account for changes in ERG amplitudes, they cannot be the cause of the changes in implicit times observed in fat-1 mice. A plausible explanation for the reduced implicit time of dark-adapted a- and b-waves in fat-1 (compared to WT mice) is an increase in OS membrane fluidity, as would be expected due to higher levels of VLCFA. 

There are multiple additional factors that must be considered when trying to understand the physiological basis underlying the ERG changes we reported in the fat-1 mice retinas. Changes in the number of photoreceptor number could alone explain changes in ERG amplitudes. Our results show that there are no changes in the number of photoreceptors as inferred by the same ONL thickness as well as the same photoreceptor nuclei row number between fat-1, WT chow, and WT n-6. We did not use cone specific staining in the context of this present study. The observation that pure rod b-wave amplitudes were equally increased as pure cone photopic b-waves points to a change that affects both rod and cone photoreceptors. We cannot exclude, however, that changes in PUFA levels might influence with the developmental processes involved in the maturation of the retina and possibly other organs. To address this possibility, detailed studies of retina anatomy and function will have to be undertaken at specific time points during the embryonic and postnatal development of the retina in fat-1 compared to WT mice. Levels of molecules such as retinoids, which rely on liver rather than the retina for their synthesis, have not been examined in the retina of fat-1 mice. Changes in the levels of retinoids would likely interfere with retina development and function. Therefore, multiple additional studies will have to be undertaken to fully understand the physiological basis underlying the changes reported here in fat-1 mice retinas. 

GFAP and CEP

The impact of highly enriched n-3 PUFA in the retina on physical and oxidative cell stress was examined in this present study by measuring GFAP and CEP, respectively. Müller cells play an important role in maintaining retina tissue integrity, in modulating neuronal activity, in the regeneration of photopigments, and in the regulation of neurotransmitter metabolism. ³⁶ Upregulation of intermediate filament proteins, such as GFAP, is a characteristic feature of nerve tissue injury or stress. ^{37 38} We observed higher GFAP expression in the filaments of both central and peripheral Müller cells of fat-1 mice in comparison to WT chow mice. This suggests that an increase in n-3 PUFA retina levels might cause physical stress in this neuronal environment. Furthermore, CEP, an adduct that forms from the oxidation of DHA, which is abundant in the outer retina, ^{26 37} was accumulated in the inner and OS of fat-1 mice. This oxidation-generated CEP has been linked with the oxidative damage that occurs in age-related macular degeneration. ^{25 39} Whether increased CEP in photoreceptors influences the physical integrity of Müller cells or vice versa is not known. Considering the increased expression of GFAP and CEP in fat-1 mice, highly enriched DHA may jeopardize cell integrity and increase oxidative stress. Bazan ¹⁰ and Mukherjee et al., ¹¹ found that another byproduct of DHA, NDP1, protects against cell-injury–induced oxidative stress and enhances RPE cell survival. Considering these data, there appears to be two opposing physiological events, dependent on the nature and levels of DHA byproducts. Enticing the production of NPD1 rather than CEP in retinas with high DHA levels, such as by increasing the levels of anti-oxidants, could be a useful endeavor. Such a combined approach is currently being studied in humans on a large scale (see ARED2: http://www.areds2.org/). 

The present study provides evidence that n-3 DHA and VLCFA levels influence retina function as assessed with the ERG and stress markers in the retina. A unique advantage of the fat-1 mouse is that it allows study of the impact of n-3 and n-6 fatty acid balance in the retina on retina physiology without dietary n-3 fat supplementation. In addition, the fat-1 mouse model has invaluable pertinence in the development of therapies to tackle retinal diseases. For instance, studies with the fat-1 mice could assist in determining whether altered PUFA levels are associated with a protective effect in various models of photoreceptor degeneration such as phototoxicity and chemotoxicity. Finally, crossing the fat-1 mouse with specific mouse models would allow exploration of the effect of altered PUFA levels (achieved through genetic manipulations) on murine analogues of defined human diseases. 

 Supported by the Natural Sciences and Engineering Research Council of Canada.

 Submitted for publication July 13, 2008; revised January 8, 2009; accepted July 6, 2009.

 Disclosure: M. Suh, None; Y. Sauvé, None; K.J. Merrells, None; J.X. Kang, None; D.W. L. Ma, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Table 1.

 

Changes in Fatty Acid Composition of Phosphatidylcholine in the Retina of Fat-1 Compared with WT Mice

Fatty Acid (%, wt/wt Total Fatty Acids)	WT n-6 (n = 4)	Fat-1 (n = 5)
C14:0	0.91 ± 0.18	0.83 ± 0.28
C15:0	0.14 ± .0.04	0.13 ± 0.01
C16:0	30.88 ± 0.99	32.14 ± 0.94
C16:1n7+n5	1.74 ± 0.16	1.58 ± 0.31
C17:0	0.12 ± 0.08	0.36 ± 0.38
C18:0	14.94 ± 0.11	15.24 ± 0.45
C18:1n9+n7	18.09 ± 1.11	18.93 ± 1.07
C18:2n6	5.94 ± 2.25	5.98 ± 0.48
C18:3n3	0.00 ± 0.00	0.00 ± 0.00
C20:0	0.00 ± 0.00	0.00 ± 0.00
C20:1	0.33 ± 0.03	0.35 ± 0.06
C20:2n9/20:2n6	0.46 ± 0.31	0.55 ± 0.04
C20:3n6	0.55 ± 0.05	0.51 ± 0.03
C20:4n6	8.60 ± 0.35	5.00 ± 0.49*
C20:5n3	0.16 ± 0.11	0.27 ± 0.08
C22:4n6	1.61 ± 0.37	0.48 ± 0.05*
C22:5n6	6.69 ± 0.47	0.43 ± 0.25*
C22:5n3	0.09 ± 0.06	0.85 ± 0.06*
C22:6n3	5.15 ± 0.86	13.61 ± 2.24*
n-6 VLCFA
C24:4n6	0.11 ± 0.07	0.01 ± 0.02*
C24:5n6	0.10 ± 0.07	0.01 ± 0.02*
C32:4n6	0.35 ± 0.07	0.01 ± 0.01*
C32:5n6	0.44 ± 0.08	0.00 ± 0.00*
C34:4n6	0.65 ± 0.06	0.06 ± 0.01*
C34:5n6	0.36 ± 0.24	0.00 ± 0.00*
C36:4n6	0.12 ± 0.12	0.00 ± 0.00*
C36:5n6	0.14 ± 0.13	0.00 ± 0.00*
n-3 VLCFA
C24:5n3	0.00 ± 0.00	0.05 ± 0.03*
C24:6n3	0.00 ± 0.00	0.17 ± 0.05*
C32:5n3	0.05 ± 0.04	0.09 ± 0.02
C32:6n3	0.13 ± 0.02	0.53 ± 0.17*
C34:5n3	0.13 ± 0.12	0.24 ± 0.07
C34:6n3	0.28 ± 0.07	0.66 ± 0.21*
C36:5n3	0.00 ± 0.00	0.00 ± 0.00
C36:6n3	0.08 ± 0.06	0.11 ± 0.03

 Values are the mean (n = 4–5) ± SD. For each, 6–8 eyes were pooled. Both animal groups were fed an n-6 fatty acid diet. VLCFA, very long chain fatty acids (C24–C26).

*  Significant difference between WT and fat-1 mice (P < 0.05).

Figure 1.

 

Fatty acid composition of phosphatidylethanolamine, phosphatidylserine, and phosphatidylinositol in fat-1 mice retina. Data are the mean (n = 3–5) ± SD. For each n, 6 to 8 retinas were pooled. *Significant difference between WT n-6 mice and fat-1 mice (P < 0.05).

Figure 2.

 

ERG recordings in WT chow mice, WT n-6 mice, and fat-1 mice. Representative ERG traces for the various tests in WT n-6 mice (left column) and fat-1 mice (right column) are presented: (a) scotopic intensity response; (b) photopic intensity response; and (c) photopic flicker frequency series. Responses to increasing intensities are presented from bottom to top, and responses to increasing flicker frequencies are presented from top to bottom. Average values for each step are presented for the three experimental groups: (d) mixed scotopic a-wave; (e) mixed scotopic b-wave; (f) photopic b-wave; and (g) flicker frequency series. Maximum values are presented in (h). *Significant difference between two experimental groups, P < 0.05. ERG, electroretinogram.

Figure 3.

 

Representative confocal images of GFAP immunoreactivity in retinal sections of WT chow and fat-1 mice: (a) WT mouse central retina; (b) WT mouse peripheral retina; (c) fat-1 mouse central retina; (d) fat-1 mouse peripheral retina. In WT chow mouse, Müller cells expressed GFAP only in the extreme periphery of the retina. In fat-1 mouse, there was an increased expression of GFAP in Müller cells from the central and peripheral retina. Note that astrocytes (localized in the inner limiting membrane) constitutively expressed GFAP. Scale bar, 100 μm. GFAP, glial fibrillary acidic protein; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer; ILM, inner limiting membrane.

Figure 4.

 

Representative photomicrographs of CEP immunoreactivity and Nissl staining. (a, b) Examples of CEP expression are shown in the central retina of a WT chow mouse and a fat-1 mouse. Staining in the WT chow mouse retina shows densely packed fine inner and OS (a), whereas in the fat-1 mouse retina (b), stained segments are swollen with a gap at the level of the cilium (arrow). Many OS are tortuous and fragmented. (c, d) Despite these distortions, Nissl staining in fat-1 mice showed no gross morphologic changes when compared with WT chow retina (d); however, the lack of inner and outer segment organization in fat-1 compared with WT mice was also apparent on Nissl-stained retinas. Scale bar: (a, b) 40 μm; (c) 100 μm. CEP, carboxyethylpyrrole; OS, outer segment; IS, inner segment; ONL, outer nuclear layer; OPL, outer plexiform layer; INL, inner nuclear layer; IPL, inner plexiform layer; GCL, ganglion cell layer.

Table 2.

 

Changes in Fatty Acid Composition of Individual Phospholipids in the Retina of Fat-1 Compared to WT Mice

Fatty Acid (%, wt/wt of Total Fatty Acids)	WT n-6	Fat-1
Phosphatidylcholine
ΣSat	46.98 ± 0.89	48.70 ± 1.18
ΣMono	20.16 ± 1.37	20.87 ± 1.51
Σn-6 PUFA	23.85 ± 2.20	12.95 ± 0.60*
Σn-3 PUFA	5.41 ± 0.96	14.73 ± 2.23*
Σn-6 VLCFA	2.26 ± 0.19	0.09 ± 0.04*
Σn-3 VLCFA	0.67 ± 0.11	1.88 ± 0.55*
n-6/n-3 ratio	4.41 ± 0.98	0.81 ± 0.16*
UI	155.0 ± 6.10	158.04 ± 15.29
Phosphatidylethanolamine
ΣSat	30.86 ± 0.91	30.02 ± 2.00
ΣMono	14.64 ± 0.70	13.69 ± 2.06
Σn-6 PUFA	40.99 ± 1.19	19.35 ± 1.51*
Σn-3 PUFA	12.17 ± 0.87	32.45 ± 3.23*
n-6/n-3 ratio	3.38 ± 0.34	0.60 ± 0.07*
UI	263.94 ± 0.41	279.08 ± 21.02
Phosphatidylserine
ΣSat	36.74 ± 1.49	39.90 ± 2.52
ΣMono	16.77 ± 1.64	17.99 ± 3.02
Σn-6 PUFA	32.72 ± 0.42	13.71 ± 0.69*
Σn-3 PUFA	10.35 ± 2.61	26.14 ± 4.70*
n-6/n-3 PUFA	3.31 ± 0.93	0.54 ± 0.11*
UI	229.91 ± 21.79	230.49 ± 30.39
Phosphatidylinositol
ΣSat	47.13 ± 1.16	47.39 ± 2.85
ΣMono	11.04 ± 4.79	10.84 ± 2.36
Σn-6 PUFA	35.68 ± 6.45	26.21 ± 4.22*
Σn-3 PUFA	3.68 ± 1.25	13.22 ± 1.13*
n-6/n-3 ratio	10.74 ± 4.61	1.98 ± 0.26*
UI	173.42 ± 21.90	188.07 ± 26.21

 Data are the mean (n = 3–5) ± SD. For each, 6–8 eyes were pooled. Both animal groups were fed an n-6 fatty acid diet. ΣSat, sum of saturated fatty acids; ΣMono, sum of monounsaturated fatty acids; Σn-6 PUFA, sum of n-6 polyunsaturated fatty acids; Σn-3 PUFA, sum of n-3 polyunsaturated fatty acids; Σn-6 VLCFA, sum of n-6 very long chain fatty acids; Σn-3 VLCFA, sum of n-3 very long chain fatty acids; n-6/n-3 ratio, the ratio of n-6 to n-3 fatty acids; UI, unsaturation index.

*  Significant difference between WT and fat-1 mice (P < 0.05).

The authors thank Peter Jones, Director, at the Richardson Centre for Functional Foods and Nutraceuticals at the University of Manitoba, for purchasing the electroretinograph; Chantal Murray and Dennis Labossiere for assistance in statistical analysis and fatty acid analysis; and Sharee Kuny and Silvina Mema for assistance with the histology.