Whole Body Synthesis of Docosahexaenoic Acid from Alpha-Linolenic Acid in. Rodents

Transcription

1 Whole Body Synthesis of Docosahexaenoic Acid from Alpha-Linolenic Acid in Rodents By Anthony Frank Domenichiello A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Nutritional Sciences University of Toronto Copyright by Anthony Frank Domenichiello 2015

2 ii Whole Body Synthesis of Docosahexaenoic Acid from Alpha-Linolenic Acid in Abstract Rodents Anthony Frank Domenichiello Doctor of Philosophy Department of Nutritional Sciences University of Toronto 2015 Docosahexaenoic acid (DHA) is an omega 3 polyunsaturated fatty acid (n-3 PUFA) that is thought to be important for neurological function partly through its role in the regulation of cell survival and neuroinflammation. The main dietary source of DHA is fish. Therefore, recommendations to increase DHA consumption may not be sustainable. Alternatively, DHA can be synthesized from precursor n-3 PUFA found in plants, primarily α-linolenic acid (ALA). DHA synthesis is generally accepted to be low, however, the requirement for DHA in the brain is not agreed upon and recent estimates for the brain DHA requirement range from mg/day (<0.3% of dietary ALA intake). Therefore, I hypothesize that DHA synthesis from ALA will be sufficient to maintain brain DHA. First, we determined if rats fed diets without DHA had deficits in brain DHA and found that neither concentration of DHA nor expression of genes that are regulated by DHA, in the brain, were different between rats fed a diet containing 2% of the fatty acids as ALA or DHA. We determined that this was likely due to the fact that rats consuming ALA synthesized between 3 and 100 fold more DHA than their brains were able to uptake and accrete. We next mimicked studies that measure DHA synthesis in humans, in

3 iii rats, and determined that the rat was an appropriate model for measuring DHA synthesis. The next two studies we performed determined how dietary fatty acids affect DHA synthesis rates. We found that low levels of dietary linoleic acid (LNA, the omega 6 PUFA equivalent to ALA) caused decreased DHA synthesis rates compared to higher levels of dietary LNA. We also found that increasing dietary ALA from 3% to 10% of the fatty acids did not lead to an increase in DHA synthesis, however, DHA synthesis rates were not maintained in rats fed diets containing ALA at 0.07% of the fatty acids. In conclusion, DHA synthesis from ALA can maintain brain DHA, and dietary LNA and ALA positively affect DHA synthesis rates.

4 iv Acknowledgements Many people have contributed this thesis and my graduate experience at the University of Toronto and I would like to extend my sincerest thanks to all of them especially those listed below: My parents who have sacrificed in ways I cannot comprehend to afford me every opportunity I could imagine and have always supported me and made my life as a graduate student as easy it could be. I nonni without whom our family would not be where it is today. My supervisor Dr. Richard Bazinet, who has led me through the world of graduate studies and provided me with an incredible example to aspire to become as a scientist. My labmates (who I am also proud to call my friends), you have all taught me so much both professionally and personally. You have taught me to see the world through a different perspective I hope you all know how grateful I am. My Buddies i.e. The Gentlemans Club (correct spelling) current, former and considered members (that covers all of you); you have all always been there for me and always supported me throughout these times and I am forever grateful and humbled by your friendship.

5 v The Battiston and Ceolin families who have welcomed me into their homes and become a second family to me. The cottage weekends will always be among my most treasured memories. Dr. Archer and Dr. Greenwood for their guidance as members of my thesis advisory committee. To all my other friends [especially The Crew (you know who you are)] for making my graduate school experience as enjoyable as it could have and always supporting me. I am grateful to all of you for your friendship.

6 vi Table of Contents List of Tables... ix List of Figures... x List of Equations... xi List of Abbreviations... xiii 1 Introduction... 2 Chapter 2 Literature Review Literature Review Current knowledge on DHA synthesis rates and brain DHA requirements Abstract Introduction Current intakes of n-3 PUFA and relation to brain function Current model for brain DHA uptake Pathway of DHA synthesis Estimates of DHA synthesis from ALA in humans Evidence that DHA synthesis affects blood DHA levels Estimates of DHA synthesis rates in rodents Concluding remarks Objectives and hypotheses Objectives Hypotheses Chapter 4 - Whole body synthesis rates of DHA from α-linolenic acid are greater than brain DHA accretion and uptake rates in adult rats Objectives 1 and 2: Comparing the rate of DHA synthesis from ALA to brain DHA uptake rates and comparing the effect of DHA and ALA on brain function Abstract Introduction Materials and Methods Animals Diets Balance study - Whole body fatty acid extraction Balance Study - Brain fatty acid extraction Balance study - Fecal PUFA analysis Transmethylation and gas chromatography-flame ionization detection Balance study - Brain RNA Extraction Balance study - Gene Expression Analysis Steady-state infusion study - Surgery and 2 H-ALA infusion Steady-state infusion study - Determination of Plasma Volume Plasma Lipid Extraction Steady-state infusion study Steady-state infusion study Thin Layer Chromatography Steady-state infusion study - Plasma lipid hydroxylation and esterification Gas Chromatography-Mass Spectrometry Brain DHA uptake study - 14 C-DHA infusion Brain DHA uptake study - Plasma radioactivity analysis... 58

7 vii Brain DHA uptake study - Liquid scintillation counting Brain DHA uptake study - Brain radioactivity analysis Gavage study - Surgery and blood sampling Liquid Chromatography(LC) Tandem Mass Spectrometry (MS/MS) Balance study - Equations Steady-state infusion study - Equations Brain DHA uptake study - Equation Statistics Results Body weight and food intake Balance study Fecal Excretion of PUFA Balance study Baseline PUFA concentrations Balance study Final PUFA concentrations Balance study PUFA accretion Balance study Gene Expression Balance study Plasma Volume Steady-state infusion study Plasma Concentrations Steady-state infusion study DHA synthesis Steady-state infusion study Brain DHA uptake rate Brain DHA uptake study Appearance of 2 H 5 -n-3 PUFA Gavage study Discussion Objective 3: To determine if high n-6 PUFA affect DHA synthesis Abstract Introduction Methods Animals Diets Surgery H 5 -ALA and U- 13 C 18 -LNA Infusion Plasma Volume Determination Plasma Lipid Extraction Separation of esterified and unesterified lipids by Thin layer Chromatography (TLC) Transmethylation and gas chromatography-flame ionization detection Fatty Acyl-CoA Extraction Quantification of fatty acid concentrations by Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) Kinetic Equations Statistics Results Plasma Volume Baseline Plasma Unesterified Lipid Concentrations Baseline Plasma Esterified Lipid Concentrations Plasma Unesterified 2 H 5 -ALA and 13 C 18 -LNA Concentrations n-3 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life n-6 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life Liver fatty acyl-coa concentrations Discussion Chapter 6 - A dose response study of dietary α-linolenic acid on the rate of synthesis of docosahexaenoic acid from α-linolenic acid in the free-living rat

8 viii 6 Objective 4: To determine how different levels of ALA affect the DHA synthesis rate Abstract Introduction Methods Animals Diets Surgery H 5 -ALA and 13 C 18 -LNA Infusion Plasma Lipid Extraction Separation of esterified and unesterified lipids by Thin layer Chromatography (TLC) Transmethylation and gas chromatography-flame ionization detection Preparation of plasma infusion samples Quantification of fatty acid concentrations by Liquid Chromatography-Tandem Mass Spectrometry (LC-MS/MS) Steady-State Infusion Kinetics Statistics Results Plasma Volume Determination Plasma n-3 and n-6 PUFA concentrations Plasma 2 H 5 n-3 PUFA and 13 C 18 n-6 PUFA n-3 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life n-6 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life Discussion General Discussion Discussion Overall findings Biological significance Clinical significance Limitations Future Studies Conclusions Appendix References

9 ix List of Tables Chapter 2 Table 1 - Summary of published studies that have used stable-isotope labeled ALA to measure DHA synthesis. Chapter 2 Table 2 - Summary of fractional conversion estimates in rats applying calculations used in human studies. Chapter 4 Table 1 - Percent composition of the three diets. Chapter 4 Table 2 Summary of n-3 PUFA balance. Chapter 4 Table 3 - Parameters for whole-body synthesis-secretion of DHA in rats consuming the control, ALA and DHA diet for 15 weeks. Chapter 5 Table 1 Percent composition of the low, medium and high LNA diets as measured by GC-FID. Chapter 5 Table 2 - Parameters for whole-body synthesis-secretion of DHA in rats consuming the control, ALA and DHA diet for 15 weeks. Chapter 5 Table 3 - Liver fatty acid CoA concentrations (nmol/g liver) for rats fed the low, medium or high LNA diet. Chapter 6 Table 1 - Fatty acid composition of the low, medium and high ALA diet as measured by GC-FID. Chapter 6 Table 2 - Synthesis parameters for ARA, EPA and DHA synthesis as determined from a 3 hour steady-state infusion of 2H5-ALA and 13C18-LNA. Chapter 7 Table 1 - Summary of results for DHA synthesis rates.

10 x List of Figures Chapter 2 Figure 1 - DHA is synthesized from ALA in the liver by a series of desaturations, elongations and a β-oxidation. Chapter 2 Figure 2 - When ALA is administered orally it is absorbed into the lymphatic system then deposited into systemic circulation. Chapter 2 Figure 3 - Quantitation of DHA synthesis rate using the steady-state infusion method. Chapter 4 Figure 1 - Whole body ALA and DHA concentrations and brain DHA concentrations in rats consuming the control, ALA or DHA diet. Chapter 4 Figure 2 DHA Kinetics Chapter 4 Figure 3 - Heat map depicting gene expression in cortex and striatum brain samples. Chapter 4 Figure 4 - Example of infusion curves for a rat from each diet group Chapter 4 Figure 5-2 H 5 -n-3 PUFA appearance in plasma of rats gavaged with 10 mg of 2 H 5 -ALA. Chapter 4 Figure 6 - Summary of the methods and results. Chapter 5 Figure 1 Plasma unesterified ALA and LNA concentrations (nmol/ml) in rats consuming the Low, Medium or High LNA diet. Chapter 5 Figure 2 - Plasma esterified concentrations for n-3 and n-6 PUFA. Chapter 5 Figure 3 Example infusion curves and chromatograms. Chapter 6 Figure 1 Plasma unesterified concentrations of ALA and LNA Chapter 6 Figure 2 Plasma esterified concentrations

11 xi List of Equations Chapter 4 Equation 1 Balance study equation Chapter 4 Equation 2 Slope of the appearance curve of labeled DHA Chapter 4 Equation 3 Maximal first derivative Chapter 4 Equation 4 DHA synthesis-secretion rate Chapter 4 Equation 5 Uptake coefficient for DHA into the brain Chapter 4 Equation 6 Brain DHA uptake rate Chapter 5 Equation 1 Slope of the DHA appearance curve Chapter 5 Equation 2 Slope of the ARA appearance curve Chapter 5 Equation 3 - Maximal first derivative Chapter 5 Equation 4 - DHA synthesis-secretion rate Chapter 5 Equation 5 Rate of turnover of plasma esterified DHA from synthesized DHA Chapter 5 Equation 6 Half-life of plasma esterified DHA Chapter 6 Equation 1 - Slope of the appearance curve for long chain n-3 PUFA Chapter 6 Equation 2 - Slope of the appearance curve for long chain n-6 PUFA Chapter 6 Equation 3 Maximal first derivative of the appearance curve for long chain n-3 PUFA Chapter 6 Equation 4 - Maximal first derivative of the appearance curve for long chain n-6 PUFA Chapter 6 Equation 5 Synthesis-secretion rate for long chain n-3 PUFA Chapter 6 Equation 6 Synthesis-secretion rate for long chain n-6 PUFA

12 xii Chapter 6 Equation 7 Rate of turnover of plasma esterified long chain PUFA Chapter 6 Equation 8 Half-life of plasma esterified long chain PUFA

13 xiii List of Abbreviations ALA ARA AUC BBB CoA COX-2 cpla2 CTX DHA Alpha-linolenic acid Arachidonic acid Area under the curve Blood-brain barrier Coenzyme A Cyclooxygenase-2 Calcium-dependent phospholipase A2 Cortex Docosahexaenoic acid DPA n-3 Docosapentaenoic acid n-3 DPA n-6 Docosapentaenoic acid n-6 EPA ER FAME GC-FID HIP IOM ipla 2 LC-MS/MS LCPUFA LNA Eicosapentaenoic acid Endoplasmic reticulum Fatty acid methyl ester Gas chromatography-flame ionization detection Hexane Isopropanol Institute of medicine Calcium-independent phospholipase A2 Liquid chromatography-tandem mass spectrometry Long chain polyunsaturated fatty acid Linoleic acid

14 xiv LPC Lysophosphatidylcholine n-3 PUFA Omega 3 polyunsaturated fatty acid n-6 PUFA Omega 6 polyunsaturated fatty acid PCR PE PFB Polymerase chain reaction Polyethylene Pentafluorobenzyl PGK1 Phosphoglycerate kinase 1 SNP spla 2 STR TLC TLDA TLE TNFα VLDL Single nucleotide polymorphism Secretory phospholipase A2 Striatum Thin layer chromatography TaqMan low density array Total lipid extract Tumor necrosis factor alpha Very low density lipoprotein

15 Chapter 1 Introduction 1

16 2 1 Introduction Dietary fats were first found to be essential in 1929 when Burr and Burr published a series of papers describing loss of hair, skin lesions and growth deficiencies in rats fed a fat-free diet that were reversed upon re-feeding with linoleic acid (LNA), an omega 6 polunsaturated fatty acid (n-6 PUFA) (1, 2). It was later discovered that both linoleic and alpha-linolenic acid (ALA, an n-3 PUFA) were not synthesized de novo and could both cure the symptoms of essential fatty acid deficiency (3, 4). Cases of LNA and ALA deficiency were later observed in humans (5, 6) and it was noted that ALA deficiency was associated with neurological abnormalities (6). The Institute of Medicine (IOM) concluded that the essentiality of ALA is likely due to it being a precursor to longer chain n-3 PUFA since ALA is not highly concentrated in neurological membranes (7). Docosahexaenoic acid (DHA), however, is the main n-3 PUFA in the brain making up approximately 10% of the fatty acids in the brain (8, 9). DHA is believed to be important for both cardiovascular and neurological function (10, 11) and there have recently been calls to establish a recommended intake for DHA (12). Establishing DHA recommendations is complicated as humans have the ability to synthesize DHA from ALA (13). While most studies measuring DHA synthesis conclude synthesis is low, these studies are plagued with methodological errors (14). Moreover, vegans and vegetarians have only 40% lower plasma DHA concentrations compared to fish eaters (15-17) and comparable neurological disease incidences (18-20). Another concern when making dietary recommendations is sustainability (21, 22), which

17 3 particularly impacts DHA, as its main dietary source is fish, a rapidly declining resource (23). The goal of this thesis is to evaluate whether DHA synthesis from ALA is sufficient to supply the brain by comparing DHA synthesis rates to brain DHA uptake rates. A novel in vivo kinetic model for measuring DHA synthesis rates will be validated against a gold standard in rats. Evidence will also be provided to verify that the rat is an appropriate model for measuring DHA synthesis rates. The validated, in vivo kinetic model will then be used to assess how dietary PUFA impact DHA synthesis rates. Chapter 2 of this thesis is a comprehensive literature review that will describe the history and current state of research focusing on the synthesis of DHA from ALA, as well as the current knowledge on the role of DHA in the brain. Chapter 2.1 is adapted from a published review that summarizes the current knowledge on DHA synthesis from ALA and assesses the sufficiency of this process from the point of view of a brain DHA requirement. The review in Chapter 2.1 also summarizes the current knowledge of DHA uptake into the brain and the role DHA plays in neurological diseases. Chapters 4-6 are the experimental chapters to this thesis. Chapter 4 is adapted from a published article containing the results of 4 studies aimed to address whether DHA synthesis from ALA can maintain brain DHA in the rat. It was found that brain DHA concentrations were maintained in rats consuming a diet containing 2% of the fatty acids as ALA. DHA synthesis rates were higher than brain DHA uptake rates (measured using 2 methods), likely explaining the ability of these rats to maintain brain DHA. We also provide evidence that this phenomenon does not appear to be specific to the rat.

18 4 Since the results in Chapter 4 indicate that DHA synthesis from ALA may be sufficient to maintain brain DHA, we next wanted to address how changes in diet will affect DHA synthesis rates. It has been hypothesized that dietary LNA competes with ALA for enzymes involved in long chain PUFA synthesis and, therefore, may lower DHA synthesis. Chapter 5 investigates how dietary LNA may affect DHA synthesis rates. DHA synthesis rates were measured in rats fed diets containing 3 different levels of LNA using an in vivo kinetic model. It was found that rats consuming higher amounts of LNA had higher DHA synthesis rates than rats consuming low amounts of LNA. In Chapter 6, we tested whether increased dietary ALA can increase the rate of DHA synthesis. Rats fed diets with higher amounts of ALA had higher DHA synthesis rates as measured by an in vivo kinetic model. Chapter 7 provides a summary of the results and overall conclusions of the thesis. The significance of the findings, both biological and clinical, are outlined. Limitations of the studies are thoroughly discussed with specific attention paid to the limitations of the in vivo kinetic model used to measure DHA synthesis rates.

19 Chapter 2 Literature Review 5

20 6 2 Literature Review 2.1 Current knowledge on DHA synthesis rates and brain DHA requirements Adapted from: Is Docosahexaenoic Acid synthesis from α-linolenic Acid sufficient to supply the adult brain? Anthony F. Domenichiello, Alex P. Kitson and Richard P. Bazinet. Prog Lipid Res. 59: Contribution: As first author I performed the majority of the background research for the paper. I designed the figures and worked with a graphic designer to make professional versions of them and researched for and made all the data tables. The majority of the writing was performed by me Abstract Docosahexaenoic acid (DHA) is important for brain function, and can be obtained directly from the diet or synthesized in the body from α-linolenic acid (ALA). Debate exists as to whether DHA synthesized from ALA can provide sufficient DHA for the adult brain, as measures of DHA synthesis from ingested ALA are typically <1% of the oral ALA dose. However, the primary fate of orally administered ALA is β-oxidation and long-term storage in adipose tissue, suggesting that DHA synthesis measures involving oral ALA tracer ingestion may underestimate total DHA synthesis. There is also evidence that DHA synthesized from ALA can meet brain DHA requirements, as animals fed ALA-only diets have brain DHA concentrations similar to DHA-fed animals, and the brain DHA requirement is estimated to be only mg/day in humans. This review summarizes evidence that DHA synthesis from ALA can provide sufficient DHA for the adult brain by examining work in humans and animals involving estimates of DHA synthesis and brain DHA requirements. Also, an update on methods to measure DHA synthesis in humans is presented highlighting a novel approach involving steady-state

21 7 infusion of stable isotope-labeled ALA that bypasses several limitations of oral tracer ingestion. It is shown that this method produces estimates of DHA synthesis that are at least 3-fold higher than brain uptake rates in rats.

22 Introduction Docosahexaenoic acid (DHA, 22:6n-3) is highly concentrated in the brain, and is important for brain function in part by regulation of cell survival and neuroinflammation (24-28). DHA cannot be synthesized de novo in mammals, and therefore, must be obtained in the diet primarily through fish, nutraceuticals and functional foods (29) or synthesized within the body from α-linolenic acid (ALA, 18:3n-3). While fish oil also contains the n-3 PUFA eicosapentaenoic acid (EPA), DHA is the main n-3 PUFA in the brain as it is concentrated at levels of nmol/g brain (about 10-15% of brain fatty acids or about 5 g in an adult brain (9, 30)), at least 50-fold more than EPA and 200-fold more than ALA (8, 9). In mammals, DHA synthesis rates from ALA are suggested to be low relative to dietary intake and tissue demand, however, debate exists as to whether the rate of DHA synthesis is sufficient to meet functional requirements for DHA. Estimates of DHA synthesis in humans are based on appearance of labeled DHA following oral ingestion of stable-isotope ALA, or changes in blood DHA following acute or chronic increases in ALA ingestion. Stable isotope methods have typically resulted in estimates of percent conversion of ALA to DHA being less than 1% of the ingested stable-isotope ALA, although estimates vary widely, ranging from 0-9.2% (Table 1). Also, there is typically no increase in plasma total lipid or phospholipid DHA when ALA intake is increased in humans (reviewed in (31, 32)), supporting the conclusion that DHA synthesis from ingested ALA is not an efficient process in humans.

23 9 Table 1 - Summary of published studies that have used stable-isotope labeled ALA to measure DHA synthesis. Reference Subjects (no. and sex) Dose (mg) Blood fraction Time (days) Conversion to DHA Method Emken et al (33) 7M 2.8 g TL % AUC Pawlosky et al (34) 4M, 4F 1 g TL % Modeling Burdge et al (35) 6M 700 mg TL 21 ND AUC Burdge et al (36) 6F 700 mg TL % AUC Burdge et al (37) 14M 700 mg TL % AUC Pawlosky et al (38) 5M, 5F 1 g TL % Modeling McCloy et al (39) 6F 47 mg TL % dose/l plasma AUC Hussein et al (40) 12M 400 mg TL 14 <0.01 % Modeling Goyens et al (41) 14M, 15F 190 mg PL % Modeling Gillingham et al (42) 14M, 25F 45 mg TL % of dose recovered Single blood sample

24 10 However, there is evidence that DHA synthesis from ALA can be sufficient to maintain brain function. For example, vegetarians and vegans, in which DHA derived from ALA is the sole source of DHA, have plasma DHA levels that are 0-40% lower than omnivores (15-17) and have neurological disease rates comparable to omnivores (18-20, 43) suggesting that ALA-derived DHA is sufficient to maintain brain function in these individuals. In addition, dietary ALA, with no DHA, is sufficient to completely restore brain DHA in rats (44) and non-human primates (45) following in utero DHA depletion, although retinal DHA was not completely restored in non-human primates. Taken together, evidence suggests that ALA-derived DHA is sufficient to maintain brain DHA levels and preserve function. In addition to a biological precedent for dietary ALA supplying adequate DHA for the brain, there is also environmental rationale to pursue this possibility. Concern has been raised regarding the environmental sustainability of current recommendations for DHA intake (21), as fish are the primary dietary source of DHA (46) and the world s fish stocks are declining (23). Although controversial (47) it has been estimated that 100% of the world s fish taxa will have collapsed by 2048 (23), indicating that strategies to reduce non-essential demands on fisheries be considered. Therefore, determining if DHA can be supplied by synthesis from ALA will be important to reduce the pressure on declining fish stocks. To accomplish this, it is essential that the extent to which ALA can be converted into DHA in humans is evaluated and compared to the requirement for DHA. This review critically examines the methodologies used to estimate DHA synthesis from ALA in humans and presents evidence suggesting that DHA synthesis capacity in humans may be greater than previously estimated. Studies measuring DHA

25 11 synthesis in adult humans will also be reviewed in the context of the brain. Additionally, a novel technique to measure DHA synthesis, that can be used in humans, the steady-state infusion method, is presented and evaluated as a means to determine, for the first time, a quantitative DHA synthesis-secretion rate in adult humans. In 2009, Barcelo-Coblijn and Murphy elegantly argued that ALA is a significant contributor to tissue DHA (48). Herein, we provide an update of the literature with a focus on brain DHA homeostasis Current intakes of n-3 PUFA and relation to brain function In North America, dietary intakes of DHA in adult (20-39 years of age) men and women are 70 and 60 mg/d respectively, while intakes of ALA in adult men and women are about 1700 and 1300 mg/d, respectively (49). Preformed DHA is found primarily in marine sources, while ALA is found in seeds and seed oils including flax, canola, and soy (49). The Institute of Medicine (IOM) recommends an adequate intake for ALA of % of total calories (7). There is only one documented case of specific n-3 PUFA deficiency observed in a patient undergoing total parenteral nutrition, with 0.6% of fatty acids as ALA in a dietary emulsion (equivalent to % of calories based on IOM acceptable macronutrient distribution ranges), that developed neuropathy and blurred vision that was reversed upon increasing ALA in the emulsion 10-fold (6). There is no specific recommendation for DHA; however, the Institute of Medicine (IOM) does state that 0-10% of the requirement for ALA can be made up from longer chain n-3 PUFA (7), corresponding to approximately % of calories, or mg/d based on a 2000 kcal diet. Recommendations for daily intake of EPA and DHA for primary prevention of coronary heart disease range from 200 to 700 mg/d (reviewed in (12)), but we are not

26 12 aware of any specific recommendations regarding DHA intakes pertaining to the adult brain. DHA is highly concentrated in the brain and retina, and reductions in brain and retina DHA in rodents and non-human primates are associated with cognitive impairments such as severe learning deficits and anxiety, as well as visual impairments such as lower electroretinogram amplitude and longer electroretinogram recovery time (reviewed in (50)). Supplemental DHA is associated with improved visual acuity in preterm infants (51), and infant formula containing DHA and arachidonic acid (ARA, the main n-6 PUFA in the brain) improves cognitive development up to one year post-partum (52). However, the effect of these PUFA treatments later in childhood is not clear (53, 54). Lower post-mortem brain DHA is present in major depressive disorder relative to controls (55-57), however, supplemental EPA, but not DHA, appears to be effective in the management of depressive symptoms (58, 59). Although somewhat controversial (60), brain DHA may also be lower in Alzheimer s disease (61-64) as compared to normal aging in which brain DHA is relatively stable (65) and prospective studies demonstrate a protective effect of fish intake on Alzheimer s disease incidence (Reviewed in (12) and (66)). However, it should be mentioned that clinical trials investigating the use of fish oil supplements to prevent/reverse cognitive decline associated with Alzheimer s disease have produced largely neutral findings in their preregistered endpoints (67-70). The effect of DHA on brain function has previously been reviewed in detail (11, 66, 71-74) Current model for brain DHA uptake PUFA such as DHA are present in the circulatory system in either the unesterified form, bound to albumin, or in the esterified form as cholesteryl esters, phospholipids and

27 13 triacylglycerides. To enter the brain, DHA must cross the blood brain barrier (BBB), a process that can be mediated either by receptor-facilitated transport or passive diffusion. The endothelium of brain capillaries contain lipoprotein receptors (75), however, lipoprotein receptor knock-out mice do not have lower brain DHA levels (76, 77). It has also been suggested that the major plasma pool supplying the brain is the unesterified DHA pool (78). Additionally, in rodents, unesterified DHA crosses the BBB rapidly in a non-competitive manner suggesting that the mechanism by which DHA crosses the BBB is via passive diffusion (79, 80). Based on the model that plasma unesterified DHA is the major DHA pool that enters the brain, brain DHA uptake rates in the rat can be measured by infusing radiolabeled unesterified-dha and measuring how much gets incorporated into the brain, after correcting for plasma radioactivity (i.e. plasma exposure to radioactivity) and the pool size (81). More recently, this concept was applied to humans using positron emission tomography to image the incorporation of [1-11 C]-DHA into the brain and quantify a rate of DHA uptake into the brain (82). The rate of DHA uptake into the brain is assumed to be replacing DHA that is consumed in the brain, and therefore, can be used as an estimate for the brain DHA requirement. It has been reported that the brain DHA uptake rate in humans is between 2.4 and 3.8 mg/day (82, 83). Based on current estimates of ALA consumption in adult males of 1700 mg/day, the percent conversion of ALA to DHA would need to be % to match the brain DHA requirement (84). Therefore, it is possible that even a small amount of DHA synthesis may be sufficient to meet adult brain DHA uptake demands. We have found that in rats DHA synthesis rates are at least 3-fold higher than brain DHA uptake rates indicating that rats may synthesize enough DHA to support the brain (85). However, the conclusion that

28 14 ALA is sufficient to support the brain will depend on what proportion of synthesized DHA is available to the brain (i.e. the brain-body partition coefficient for DHA). It is also important to recognize that if another plasma DHA pool contributes to brain DHA, current estimates of brain DHA uptake will be underestimates. It is possible that the plasma lysophosphatidylcholine (LPC) pool is a major contributor to brain DHA, especially in the rodent pup where i.v. injections of radiolabeled-lpc-dha resulted in 12-fold higher brain radioactivity compared to pups injected with radiolabeledunesterified-dha (86, 87). Additionally, the orphan receptor Mfsd2a has recently been shown to transport LPC-DHA in vitro, and ablation of Mfsd2a resulted in decreased uptake of LPC-DHA and lowered brain DHA composition compared to wild-type controls (88). Serum LPC-DHA levels, as measured by liquid-chromatography tandem mass spectrometry, range from μM (89-92), while measures of serum unesterified- DHA range from 1-4µM to (93-95) as measured by TLC-GC-FID. However, measurement of LPC- and unesterified DHA in single studies in rodents shows that unesterified DHA ranges from 10-fold higher than LPC-DHA to approximately equal (92, 96). The apparent discrepancy between the contribution of non-esterified and LPC DHA to brain phospholipid DHA may be explained by different half-lives and different experimental procedures between laboratories. A more comprehensive comparison of non-esterified DHA and LPC-DHA concentrations using the same technique within studies is required. Regardless, current estimates of brain DHA uptake may be underestimates if LPC-DHA proves to be a major contributor to adult brain DHA uptake Pathway of DHA synthesis Figure 1 depicts the synthesis pathway of DHA from ALA. The desaturase and elongase enzymes that are used to synthesize longer chain PUFA (for example DHA) are

29 15 most highly expressed in the liver as compared to heart or brain (97-99), corresponding to more than 30-fold higher rates of DHA synthesis in this organ (100). ALA is desaturated by the rate-limiting Δ6-desaturase enzyme in the endoplasmic reticulum (ER) to form 18:4n-3 (101, 102), followed by elongation to 20:4n-3 and desaturation by Δ5-desaturase to form 20:5n-3 (EPA). EPA can be elongated further to 22:5n-3 (docosapentaenoic acid DPAn-3), and 24:5n-3. 24:5n-3 is desaturated by Δ6-desaturase forming 24:6n-3, which is transferred from the ER to the peroxisome where it is β-oxidized to form 22:6n-3 (DHA) (13, 32, 101, 103). DHA is then transferred back to the ER where it can undergo esterification, lipoprotein packaging and excretion to the blood.

30 Figure 1: DHA is synthesized from ALA in the liver by a series of desaturations, elongations and a β-oxidation. Enzymes involved in the synthesis of DHA from ALA are also used by n-6 PUFA and n-9 fatty acids (not shown) leading to competition between n-3 and n-6 PUFA for these enzymes. This competition is most apparent for the Δ6 desaturase, where 4 PUFA (2 n-3 PUFA and 2 n-6 PUFA) compete for a single enzyme. The desaturations and elongations occur in the endoplasmic reticulum and the β- oxidation occurs in the peroxisome, to where 24-carbon PUFA are transferred. The final products (DHA and 22:5n-6) are then transferred back to the endoplasmic reticulum where they along with other PUFA can be esterified to very-low density lipoproteins (VLDL) and secreted into the blood. 16

31 17 The pathway is active towards both n-3 and n-6 PUFA as well as n-9 fatty acids, resulting in potential competition for enzyme activity between the families of fatty acids. This is particularly important for the rate-limiting enzyme, Δ6-desaturase, which is active towards both ALA and linoleic acid (LNA), as well as 24-carbon n-3 and n-6 PUFA (104, 105). Dietary PUFA down regulate the expression and activity of the enzymes involved in DHA synthesis in the liver (97, 98, 100, 106), thus decreasing the hepatic DHA synthesis rate (107). The brain is capable of synthesizing DHA (100, 108), however, brain DHA synthesis is approximately 100-fold lower than brain DHA uptake and consumption rates, indicating that brain DHA synthesis does not contribute significantly to brain DHA homeostasis (109). Interestingly, dietary n-3 PUFA deprivation does not affect the expression of the desaturases or elongases or the DHA synthesis rate in the brain, in contrast to increased synthesis found in the liver (100). DHA synthesis-secretion in the liver is at least 3-10 fold greater than brain DHA consumption rates (85, 110), which, combined with the finding of up-regulated hepatic DHA synthesis during n-3 deprivation, suggests that hepatic DHA synthesis is capable of maintaining brain DHA homeostasis. Recently, alternative mechanisms for DHA synthesis have been proposed (103, ). An experiment performed in baboons determined that the Δ6-desaturase enzyme also has Δ8-desaturase activity (111). Based on this finding the authors proposed an alternative pathway for DHA synthesis from ALA that functions in parallel with the classical pathway and involves an initial elongation of ALA to 20:3n-3 followed by Δ8- desaturation to make 20:4n-3, which is then desaturated and elongated to become DHA (111). Another recent study questioned the Δ6-desaturation as the sole rate-limiting step

32 18 in the synthesis pathway. The authors found that the elongation of DPA n-3 to 24:5n-3 may be another crucial control point in DHA synthesis (112). This reaction is catalyzed by the enzyme elovl2, and lack of expression of this enzyme in heart is believed to be the reason why heart tissue has very low DHA synthesis rates (114). These novel insights into DHA synthesis merit further investigation to determine how much they contribute to DHA synthesis in vivo Estimates of DHA synthesis from ALA in humans Evidence from ALA Feeding The simplest means of estimating DHA synthesis in humans is measuring changes in DHA status in response to acute or chronic increases in dietary ALA consumption, and these studies have been previously reviewed in detail (31, 32, 115). In general, these studies increase subjects ALA consumption and measure DHA in the blood. While most studies report that plasma and erythrocyte EPA increase with ALA feeding, most do not detect an increase in plasma or erythrocyte DHA ( ). Reviews of these studies have pointed out two important points pertaining to the lack of plasma DHA increases after ALA feeding. Firstly, in humans with low DHA diets (vegans and vegetarians), ALA feeding increases plasma DHA (115). Additionally, plasma DHA tends to increase to a greater extent when ALA consumption is increased in combination with decreased LNA consumption (31, 32). It should be recognized that these studies only measure DHA in blood lipids (plasma, erythrocytes, or leukocytes) as opposed to tissues. While plasma DHA may be a reliable marker for dietary DHA intake, the applicability of this pool to the brain is not agreed upon. This is because most of these studies measure percent composition of DHA in the esterified blood lipid pools, which are not thought to be available to the brain (81).

33 19 A recent rodent study performed in our laboratory highlights this point (85). We fed rats a diet that was either low in n-3 PUFA (0.25% fatty acids as ALA) or contained either ALA or DHA. After 15 weeks on these diets, levels of DHA in the body and plasma were significantly higher in rats fed DHA compared to rats fed the ALA and control diet (2.4 and 11-fold higher, respectively, for the body and 2 and 5-fold higher, respectively, for plasma). However, brain DHA levels were not different between ALA- and DHA-fed rats, similar to previous studies in rats (44) and non-human primates (45), suggesting that changes in blood DHA concentration do not necessarily reflect the magnitude of changes in brain DHA, with some exceptions (136, 137). Interestingly, graded ALA deprivation from 4.6% (considered adequate to maintain brain function and DHA concentrations) to 0.2% (considered inadequate based on decreased DHA concentration and metabolism and deleterious metabolic adaptations) of fatty acids in a diet lacking DHA results in decreased brain DHA only when the ALA content of the diet is decreased to 0.8% or lower (138). This indicates that extreme cases of ALA deprivation are required to affect brain DHA concentrations. Accordingly, the only recorded case of n-3 PUFA deficiency in humans resulted from total parenteral feeding of an emulsion containing only 0.6% of fatty acids as ALA (6). This supports the hypothesis that extremely low ALA intakes are required to significantly affect brain DHA levels and function, assuming however, that the neurological impairments observed with ALA deficiency are caused by decreases in brain DHA. It is possible that though plasma esterified DHA is unchanged with chronic increases in ALA feeding, dietary ALA may be sufficient to maintain brain DHA concentrations, possibly via the plasma unesterified fatty acid pool. The plasma

34 20 unesterified fatty acid pool is fold smaller than the esterified pools (107, 139, 140) and is maintained largely via the adipose (fasting state) and hydrolysis from plasma lipoproteins (post-prandial) (141). Also, the DHA concentration of the plasma unesterified fatty acid pool decreases only when extreme n-3 PUFA deprivation occurs (138). Moreover, few studies have examined the effect of increasing dietary DHA intake on unesterified DHA concentrations in humans, with some studies reporting an increase and others reporting no increase (95, 139, ). Adipose, the tissue that maintains plasma unesterified fatty acid concentrations, has been estimated to contain 1-4 and g of DHA in the infant (146, 147) and adult (148), respectively. Using the previously measured brain DHA uptake rate of 3.8 mg/day in adult humans, it can be calculated that adult human adipose contains enough DHA to supply the brain for years. It is important to note that the estimate for how long adipose DHA can supply the brain is an overestimate because DHA released from the adipose is used by other tissues as well as the brain. Therefore, to determine the actual amount of time that adipose DHA can supply the brain, the proportion of DHA that is released from the adipose and taken up into the brain (brain-body partition coefficient) must be determined Evidence from stable isotope administration DHA synthesis from ALA in humans has been examined by administering an oral dose of stable isotope-labeled ALA and measuring the appearance of labeled DHA in blood lipids over time. Through repeated blood sampling, concentration-time plots of the appearance of labeled n-3 PUFA are obtained, and the area under the curve (AUC) for DHA is compared to either the AUC for all labeled PUFA (33-38) or expressed relative to the administered dose to calculate the fractional conversion of ALA to DHA (39, 42). The fractional conversion of DHA from ALA is, therefore, a measure of the percentage

35 21 of labeled n-3 PUFA that appears in the plasma as DHA or the percentage of a single oral dose of ALA administered at one time that appears in the plasma as DHA. Estimates of fractional conversion of an oral dose of ALA to DHA using this technique have ranged from below the detection limit in one study to 9.8%, however, the majority of studies using this technique report fractional DHA conversion of <1% (Table 1). Alternatively, the relative conversion of each intermediate within the pathway can be estimated by using compartmental modeling. This approach is based on the assumption that the relative concentration of pathway intermediates in plasma is representative of the relative concentrations in liver, the primary site of DHA synthesis. The amount of orally administered ALA utilized for DHA synthesis using this technique has been estimated to be between % (34, 38, 40, 41). Taken together, these measures have led to a general consensus that DHA synthesis in humans is insufficient to meet DHA demands; however, care must be taken in interpreting these estimates of DHA synthesis in humans, especially in reference to the brain. In general, there are considerations regarding the oral administration of an ALA tracer to estimate DHA synthesis, as this type of experiment represents DHA synthesis from postprandial ALA only, rather than total DHA synthesis from ALA. For example, the extent to which orally administered ALA is available for DHA synthesis is not known. Fatty acids absorbed in the intestine are packaged into lipoproteins, and the majority are transported through lymphatic circulation and secreted into the blood through the thoracic duct (Figure 2). Fatty acids are taken up by tissues following hydrolysis of lipids by endothelial lipase and lipoprotein lipase or by endocytosis of the lipoprotein. Approximately 72% and 64% of orally administered 13 C-ALA is β-oxidized

36 hours after dosing in humans (39) and after 24 hours in rats (149), respectively. This value for β-oxidation of ALA in humans is similar to that of oleic and elaidic acids, but slightly higher than LA, at least between 9 and 24 hours post dose (39). Studies in rats demonstrate that the adipose AUC makes up 75% of the whole-body AUC for orally gavaged 2 H-ALA after 600 hours with progressive enrichment of adipose tissue with ALA (150). Balance studies performed in rodents have also found that the majority of dietary ALA that is not β-oxidized accumulates in the adipose tissue (151). Moreover, in adult females after one week it has been estimated that upon an oral dose of labeled ALA, up to 57% of the tracer is in the adipose (39). The fate of ALA that is deposited into adipose tissue is not clear, however, the adipose fatty acid half-life is approximately 1 year (152, 153) indicating that long-term storage would make a large proportion of oral ALA tracer unavailable for DHA synthesis measures. Taken together, this indicates that the major fate of orally administered ALA tracer, that is not β-oxidized, is adipose sequestration with a long half-life. In fact, enrichment of plasma with gavaged ALA peaks at only 5% of the whole-body tracer content and progressively declines over time (150). Moreover, in rats, less than 5% of 2 H-EPA, DPAn-3 and DHA derived from gavaged 2 H-ALA is found in plasma with the majority found in nervous system, liver, and adipose with a progressive enrichment in nervous tissue (150). Thus, the amount of tracer that is found in plasma represents a very small proportion of the total tracer that is provided orally, and is likely an underestimate of the total whole-body DHA synthesized and accreted (150).

37 Figure 2: When ALA is administered orally it is absorbed into the lymphatic system then deposited into systemic circulation. This is problematic for human tracer studies that administer ALA orally and measure the appearance of labeled n-3 PUFA products in the plasma, as a large portion of the tracer will get taken up into the tissues and adipose and not reach the liver for the duration of the study. 23

38 24 This suggests that DHA synthesis measures from ingested ALA tracer likely represent only DHA synthesized from postprandial ALA, but do not necessarily reflect the total pool of ALA that is available for DHA synthesis. As fractional conversion of DHA from ingested ALA represents only the proportion of the dose that is found in the blood compartment, which is a very small portion of the DHA synthesized from ALA, these estimates of fractional conversion are likely underestimates of actual DHA synthesis in humans (150, 154). Estimates of DHA synthesis from ALA using this method range from <0.01-1% of oral dose of ALA (39, 40, 42). Fractional DHA synthesis has also been estimated by comparing the plasma AUC for labeled n-3 PUFA to estimate the percentage of plasma ALA that is converted into DHA. In these studies fractional conversion is measured by determining what percentage of the total labeled n-3 PUFA that appeared in the plasma was labeled DHA. By adjusting for the appearance of labeled fatty acids, this method is less likely to underestimate fractional DHA synthesis rates by accounting for loss of label associated with adipose sequestration. The fractional conversion of 13 C- or 2 H- ALA to DHA in young men using this technique has been measured as 3.8% after 48 hours (33) and below the detection limit after 504 hours in one study (35), and 9.2% after 504 hours in young women (36). However, the extent to which the fractional conversion quantifies actual DHA synthesis is not clear, as it is only a relative measure (155). In addition, the AUC comparisons do not take into account differences in the plasma half-lives of the different n-3 PUFA. It has been estimated that the half-life for plasma esterified ALA is 1 hour, while that of DHA is 20 hours (34). This difference in plasma half-life would result in equal amounts of DHA and ALA eliciting a much greater AUC for DHA than that of ALA. Therefore,

39 25 these methods are also susceptible to factors that affect plasma half-life of DHA, such as diet (38, 85). Compartmentalized modeling procedures are another method to determine DHA synthesis from orally administered ALA and also provide measures of rate of flow of labeled fatty acids between compartments, half-lives, loss rates, as well as conversion rates from one fatty acid to another. Compartmentalized modelling describes the flow of materials, in this case n-3 PUFA, from one compartment to another. For modelling n-3 PUFA metabolism, stable isotope-labeled ALA is provided orally and the appearance of ALA and its longer-chain derivatives, including DHA, is measured over time. Each fatty acid between ALA and DHA is considered a compartment in the model, and when the data is corrected for unlabeled n-3 PUFA concentrations the transfer of label from one compartment to another describes the rate constants for conversions between fatty acids within the DHA synthesis pathway. A major advantage of this type of modelling is that it can potentially yield conversion rates in µg/h rather than relative data such as percent conversion. However, numerous assumptions are required for this type of modeling that likely affect conclusions rendered from the data. For example, the kinetics that are modelled in this analysis are based on oral consumption of an ALA tracer, and as such may not represent the kinetics of all sources of ALA that compose steady-state serum ALA concentration, such as ALA secreted from adipose or liver stores. Also, the majority of the tracer is lost to uptake by adipose or other tissues and/or β-oxidation based on very low appearance of the ALA tracer in plasma after ingestion (34). This type of modeling is an approximation of hepatic conversion of ALA into longer-chain n-3 PUFA based on appearance of label in plasma (34), however, important differences in plasma and hepatic

40 26 n-3 PUFA composition (eg. ratio of DHA to ALA is 2-fold higher in liver than in plasma total lipids (156, 157)) suggest this approximation is limited. The rate constants that are calculated, therefore, represent the cumulative process involved in conversion of one plasma tracer to another, including uptake by the liver, conversion, and secretion back into the plasma (34). This will also lead to an underestimation of DHA synthesis as it has been reported that after the consumption of a labeled ALA tracer, approximately 15% of DHA is synthesized fully in the liver before appearing in the plasma based on comparison of compartmental DHA metabolism (41). Interestingly, in one study the compartmental model predicted that the amount of dietary DHA required to maintain serum DHA concentration was 2.2-fold higher than what was directly measured by food duplicate, and the authors concluded that maintenance of DHA status requires greater DHA output from body store utilization or ALA synthesis than was measured in this study (34). Estimates of fractional DHA synthesis from ALA using this method range from % (34, 38, 40, 41) Considerations for oral stable isotope infusions A factor that contributes to the significant variation in estimates of DHA synthesis in humans, and therefore, adds significant uncertainty to conclusions regarding DHA synthesis, is heterogeneity between studies in background fatty acid intake. Dietary fatty acid composition has significant effects on the DHA synthesis rate (33, 38, 100, 107). Specifically, DHA is known to down regulate enzymes involved in its own synthesis (98, 100, 158, 159). In addition, n-6 PUFA may compete with n-3 PUFA for the enzymes involved in DHA synthesis (33, ). For example, higher fractional conversion of ALA into DHA has been shown in response to increased ALA:LNA ratio in the diet using compartmentalized modeling in humans (132).

41 27 Although methods utilizing oral administration of isotope-labeled ALA to estimate DHA synthesis in humans may not directly measure a DHA synthesis rate, these measures do have utility in comparing DHA synthesis between individuals or groups in the same study (42, 155, 163). In general, conclusions can be drawn about the relative differences in DHA synthesis between groups, such as the finding that women utilize a greater proportion of n-3 DPA for DHA synthesis as compared with men (164). However, based on factors discussed previously, absolute DHA synthesis rates cannot be quantified with this method. Ingested fatty acid tracers also appear to poorly model the pharmacokinetics of in situ PUFA metabolism, in addition to having only a fraction of the tracer appear in the blood. For example, compartmental analysis revealed that stable isotope-labeled EPA is 40% less effectively utilized for DHA synthesis when ingested as compared with EPA that has been synthesized from ALA (165). This may also be true for ALA, in that ingested labeled ALA may poorly represent unlabeled ALA derived from body stores, although this has not been examined. The use of stable isotope tracers to measure DHA synthesis has another general consideration, as one must by definition change the substrate concentration in the form of an administered tracer. This may increase flux through a pathway, result in substrate inhibition, or result in additional effects that might otherwise not occur under normal circumstances. Therefore, one must use the smallest amount of tracer that allow for reliable quantitation of the measure of interest so as not to influence the physiological process being measured. There is also some concern regarding deuterium exchange while using deuterium-labeled stable isotopes, in which deuterium atoms are exchanged with

42 28 unlabeled hydrogen atoms. Though this exchange rate has not been quantified in DHA synthesis studies, hydrogen exchange between water and fatty acids has been found to be negligible under typical experimental conditions (166), suggesting that deuterium exchange is a quantitatively minor process. Also, deuterium exchange would most likely affect tracer/tracee ratio of both products and substrates in DHA synthesis (assuming all fatty acids have equal deuterium exchange rates). Therefore, studies calculating DHA synthesis as percent dose of oral dose are susceptible to underestimation if using 2 H- ALA, while studies calculating percent conversion based on comparisons between AUCs of 2 H-ALA and 2 H-DHA would likely be unaffected Evidence that DHA synthesis affects blood DHA levels In addition, there is evidence that significant changes in DHA status can occur independent of changes in n-3 PUFA intake, likely through increased synthesis of DHA from ALA. For example, women have higher DHA in plasma phospholipids and erythrocytes compared with men (167), which is associated with much higher rates of DHA synthesis in women (35, 36, 164). The higher DHA synthesis in women corresponds to higher hepatic expression of the Δ5- and Δ6-desaturase enzymes in female compared with male rodents (156, 168). Female rats also have higher expression of fatty acid binding protein in hepatocytes (169), suggesting that binding and trafficking of ALA towards DHA synthesis may be higher in females as compared with males, and it is also possible that the half-life of DHA in the plasma is longer in women. Another example of DHA status being affected independent of changes in n-3 PUFA intake is altered fatty acid profiles associated with single nucleotide polymorphisms (SNP) in the human Fatty Acid Desaturease 2 gene (FADS2), the gene that encodes for the Δ6-desaturase enzyme.

43 29 The majority of these polymorphisms affect EPA concentrations, but not DHA concentrations, in phospholipids of plasma (170), serum (171), and erythrocytes (172); while analysis of a particular haplotype (with 28 SNP) has shown increased levels of DHA in plasma total lipids in the Northern Swedish Population Health Study (173). Also, a Δ6-desaturase SNP associated with increased Δ6-desaturase product:precursor ratios is associated with increased DHA percent composition in maternal erythrocytes during pregnancy (174) and colostrum (175) and a SNP with lower Δ6-desaturase activity is associated with lower levels of DHA in erythrocytes in pregnancy and breast milk (176). A recent study using orally administered ALA tracer found that some minor allele variants were associated with lower labeled EPA enrichment in the plasma as well as lower concentrations of ARA and EPA (42). These studies provide some evidence that DHA levels can be altered with no change in n-3 PUFA intake, with evidence that these changes are due, at least in part, from differences in DHA synthesis Estimates of DHA synthesis rates in rodents Examination of DHA synthesis in rodent models allows for more invasive analytical methods which can assist in validation of less invasive methods that can be applied to human subjects. Estimates of DHA synthesis from ALA based on isotope administration can be compared with whole-body DHA synthesis-accretion rates in ALAfed animals to validate the isotope method. For example, rates of DHA synthesis in rats achieved using the balance method (described below) and steady-state 2 H-ALA infusion (which can be applied to humans and is also described below) provide estimates of 4.4 µmol/d and 1.5 µmol/d, respectively. In this way, the balance method validates the steady-state infusion method and suggests that the infusion method can provide an accurate measure of DHA synthesis in humans.

44 30 There is concern that the rat is a more rapid converter of ALA to DHA as compared with humans (177), resulting in concern regarding the generalizability of DHA synthesis measures in rats to humans. This notion stems largely from the comparison of microsomal desaturase enzyme activities measured in rats and humans (177, 178). However, no study has directly compared human and rat desaturase activities or DHA synthesis rates. Moreover, the methods used to estimate DHA synthesis rates from ALA in the rat differ from those in the human, and the method used to measure DHA synthesis rates in the human have not been validated in the rat. To examine this, our laboratory orally administered 2 H 5 -ALA to rats and sampled blood over a 6-hour experiment to measure 2 H 5 -DHA and apply calculations used previously in studies providing a single oral bolus of labeled ALA in humans (85). Depending on the calculation used, the percentage of ALA dose converted to DHA ranged from 0.12% to 0.64%,which are not higher than previous estimates of DHA synthesis in humans using the same calculations (Table 2) (33, 39, 42), suggesting that DHA synthesis estimated by oral dose methodology is similar between rats and humans and that the rat may be a suitable model for validation of human DHA synthesis methods.

45 31 Table 2: Summary of fractional conversion estimates in rats applying calculations used in human studies (adapted from (85)) Method Description Human Rat Conversion to DHA Conversion to DHA McCloy et al. (39) % oral dose of ALA appearing as AUC DHA 0.99% 0.31% Emken et al (33) % of total n-3 AUC as DHA AUC 3.8% 0.64% Gillingham et al. (42) % recovery of oral ALA tracer in blood sample at conclusion of study 0.19% 0.12%

46 Measurements of dha synthesis from balance studies The balance method, developed by Cunnane et al. for use in examination of essential fatty acid accretion and metabolism, requires feeding rats a diet with ALA as the only n-3 PUFA then measuring the accretion of DHA in the rat whole body (146, 151, 179). Previously published balance studies have reported that the DHA synthesis rate in rats to be between 4.4 and 15 μmol/day (85, 151, 180). The balance method does provide an estimate of the net DHA synthesis and accretion; however, it cannot account for DHA that has been synthesized and then metabolically consumed. Even so, this potential limitation to the balance study would, at most, result in an underestimate in the actual DHA synthesis rate. Interestingly, balance studies also provide more evidence that the major fate of orally ingested ALA and DHA is apparent β-oxidation. By feeding rats only either ALA or DHA, it has been shown that β-oxidation of these fatty acids is between approximately 60% after 8 weeks of feeding (180) and 90% after 15 weeks (85), with no differences between ALA and DHA loss rates Measurements from steady-state ALA infusion studies The steady-state infusion model, developed by Rapoport et al. (181) and recently applied with modifications by our laboratory (85), involves infusing isotope-labeled unesterified ALA such that plasma concentration of the tracer achieves steady-state and measuring the appearance of labeled DHA in the plasma (181). The major strength of this method is that it provides a whole-body DHA synthesis rate (μmol or mg synthesized per unit time) as opposed to many methods using ingestion of an oral bolus which provide relative estimates of DHA synthesis from ingested ALA only (the exception being some applications of compartmental modelling (34), but these are still based on ingested ALA

47 33 tracers). Another advantage of the steady-state infusion method is that it measures DHA derived from serum ALA, rather than only dietary ALA, and as such likely represents the entirety of ALA that is available to the liver for DHA synthesis. By more closely representing the substrate pool available for DHA synthesis (serum ALA from all sources, not just post-prandial, gut-derived ALA), the steady-state infusion method may be a more representative estimate of DHA synthesis relative to oral dosing methodology. The steady-state infusion model involves infusing unesterified albumin-bound labeled ALA intravenously at a constant rate to achieve a steady state concentration in the blood, and through repeated blood sampling the appearance of labeled DHA in esterified lipids is measured. The appearance of labeled DHA using this method can be fit to a sigmoidal curve (Figure 3). Generally, the first derivative at any point of a curve represents the rate-of-change of the measured variable at that point, and as such the derivative of the sigmoidal DHA appearance curve represents the rate-of-change of labeled DHA (i.e. the rate of change of the concentration of labeled DHA in plasma) at that point. By determining the maximal first derivative of the sigmoidal labeled-dha curve, the maximum rate-of-change of labeled DHA is obtained. This maximal rate-ofchange is taken as the labeled DHA synthesis rate from labeled ALA, and by correcting this rate by the unlabeled ALA concentration the whole-body DHA synthesis rate is obtained (see description of calculations, figure 3).

48 Figure 3: Quantitation of DHA synthesis rate using the steady-state infusion method. Upon infusion of labeled ALA at steady-state, appearance of labeled DHA in the plasma is sigmoidal (top panel). The first derivative of this curve can be determined and is equal to the rate of appearance of labeled DHA in the plasma at every time point during the infusion (bottom panel). The maximal first derivative (circled in the top and bottom panel) is taken to be the maximal rate of DHA appearance in the plasma (S max ). S max is assumed to be the point where tissue uptake of labeled DHA is negligible meaning that the appearance of labeled DHA at this point is due solely to DHA synthesis from ALA. By correcting S max for the tracee to tracer ratio the DHA synthesis rate can be determined. 34

49 35 There are several important limitations of this method, including that the serum concentration of labeled DHA represents the equilibrium between synthesis-secretion and tissue uptake. While the contribution of tissue uptake to labeled DHA concentration is lowest relative to synthesis-secretion at the maximal first derivative, it is necessarily defined as zero in the calculations. The half-life of serum DHA is approximately 20 hours (34) suggesting that tissue uptake DHA would not be a major contributor to changes in labeled DHA over the time-course of most experiments (infusions are typically 3 hours, with the maximal derivative obtained within 2 hours). Despite this, the assumption of zero tissue uptake at the maximal first derivative means that the steady-state infusion yields only a lower bound estimate on whole-body DHA synthesis measures. There is also potential for dilution of the ALA tracer in the liver acyl-coa pool (the pool that is primarily utilized for DHA synthesis), such that measured tracer-tracee ratios in the plasma may not represent the ratio in the liver, the primary site of DHA synthesissecretion. This dilution has been estimated in the rat where after a 5 min infusion of radiolabeled ALA the tracer:tracee ratio was 60-80% lower in liver acyl-coa compared with plasma unesterified ALA (100, 107). While the dilution factor is unlikely to be measurable in the human, it should be noted that this limitation will also result in an underestimation of the actual DHA synthesis rate (for more details refer to (85)). Additionally, it should also be pointed out that due to the small pool size of unesterified ALA in the plasma, minimal amounts of labeled unesterified ALA should be infused during these studies to avoid altering the physiological pathway that is being studied (as previously discussed). Finally, synthesis is measured only over a short period of time (i.e. 2-3 hours) in the rat and may miss any diurnal variation that occurs in the synthesis rate.

50 36 The achievement of a steady-state of ALA tracer concentration is essential for calculating a DHA synthesis rate as the rate of DHA synthesis is not influenced by changes in plasma ALA tracer. This is in contrast to the single oral dosing method used in humans, in which labeled ALA concentrations in the plasma do not reach steady state, therefore, limiting the possibility of estimating DHA synthesis rates as discussed previously (163). The steady-state infusion method bypasses ingestion, digestion, and absorption of the labeled ALA, and therefore, measures the whole-body DHA synthesis rate from circulating unesterified ALA, whereas estimates of DHA synthesis from orally provided ALA model only post-prandial ALA. The method provides a measure of wholebody DHA synthesis, and would account for synthesized DHA that occurs when other secretory tissues, such as adipose or gut, take up the labeled ALA and secrete labeled DHA back into circulation. The steady-state infusion method has been used to estimate DHA synthesis rates in rats, and found DHA synthesis rate to be approximately 1.5 μmol/day (85), which is lower, but in broad agreement with the rates determined by the balance method in rats fed the same diet (4 μmol/day) (85) and growing rats fed a similar diet (11-14 μmol/day (151, 180)). Using the steady-state method, the DHA synthesis rate is at least 3-fold higher than the daily brain DHA uptake rate in rats (85), suggesting that DHA synthesis may be sufficient to provide the brain with DHA. Importantly, the steadystate infusion method can be performed in humans, and is currently being applied as part of a larger clinical study (ClinicalTrials.gov Identifier: NCT ) Concluding remarks There is considerable debate as to whether the human capacity to synthesize DHA from ALA is sufficient to meet brain DHA requirements. This debate has been further

51 37 complicated by lack of agreement regarding the brain DHA requirement, and methodological inconsistencies in attempts to quantify the rate of DHA synthesis from ALA. The IOM did not assign a dietary reference intake for DHA, and other recommendations for DHA and EPA intake pertain to cardiovascular disease prevention (7, 12, 182, 183) and not specifically to support the brain, in part reflecting uncertainty in the role of dietary DHA in maintaining brain DHA. Fortunately, an estimate of human brain DHA uptake is now available ( mg/day (82, 83)), and novel approaches to measure whole-body DHA synthesis from serum ALA using steady state isotope infusion will allow for quantitative comparison of DHA synthesis rate to brain DHA uptake rate, as done previously in rats (85). This approach will supplement previous measurements of DHA synthesis from ingested stable isotope ALA, which provide estimates of DHA synthesis from postprandial ALA, and produce a more complete understanding of DHA homeostasis in humans. Despite limitations in comparing rates of DHA synthesis and brain uptake rates in humans to date, there is considerable evidence from animals showing that brain DHA levels are similar when fed ALA as the only n-3 PUFA as opposed to DHA or ALA+DHA, as reviewed extensively (84), although there are some exceptions (137) possibly related to dose-, duration-, and species-specific effects. The brain has mechanisms whereby it can conserve DHA that may explain similar brain DHA between DHA- and ALA-fed rats (184). For example, the expression of DHA-catabolizing enzymes, such as group VIB calcium-independent phospholipase A 2, can be reduced, resulting in decreased catabolism of DHA and a longer brain DHA half-life (138, 184, 185). The effect of altering brain DHA turnover on brain function is not clear. Also, n-3

52 38 PUFA deficiency increases aspects of ARA turnover and decreases DHA turnover (138, ), suggesting that the brain may metabolize ARA to spare DHA. The effect of increased utilization of ARA relative to DHA by the brain, on brain function or in disease is not currently known. However, vegans and vegetarians have similar prevalence of neurological diseases as compared with omnivores suggesting that any altered brain DHA metabolism in these individuals does not manifest neurologically (18-20, 43, ). Studies that have used ingested stable-isotope ALA to measure DHA synthesis in humans have for the most part reported that DHA synthesis from ALA is thought to be an inefficient process (generally <1% conversion). The calculations used in these studies are inconsistent (155), and we have shown that they yield different values for percent conversion depending on the calculation used (85). In addition, these methods may only provide relative as opposed to absolute quantifications of DHA synthesis rate (115, 155) and only represent the DHA synthesized from postprandial ALA. However, if the low brain DHA uptake rate is an accurate measure of the brain DHA requirement then a low fractional conversion may still be sufficient to supply DHA to the brain. It must be stressed that the focus of this review is the capacity of DHA synthesis from ALA to supply the brain in healthy adults. Situations that may affect DHA synthesis rates and/or brain DHA uptake rates (such as diet, development, genetics, brain injury, disease or aging) must be examined to determine if ALA-derived DHA can meet brain DHA requirements in these cases. For example, during infancy the brain accretes a large amount of DHA as it grows and post-mortem studies have found that breast-fed infants have significantly higher brain DHA concentrations than infants fed formula that contains ALA but not DHA (192). Therefore, this may be an instance where DHA synthesis from

53 39 ALA is not sufficient to supply the brain, and preformed DHA is required. However, methods are now available that can be applied to both rodents and humans to measure brain DHA uptake and DHA synthesis rates, allowing for estimation of sufficiency of DHA synthesis and recommendation for DHA intake. In 2009, Rapoport and colleagues developed an in vivo steady-state, stable-isotope infusion method to measure the DHA synthesis rate from serum ALA in rats (181). By infusing the tracer intravenously this method avoids some of the considerations of oral tracer administration. The steady-state infusion method allows for the direct quantification of the DHA synthesis rate, rather than a relative conversion measure. Importantly, the synthesis rates measured using this method (85) are in line with rates that were measured using balance studies (151, 181). Application of this method to humans would represent the first quantification of the DHA synthesis rate from blood ALA in humans, which could be compared to the brain DHA uptake rate. It is of importance to know how much DHA can be synthesized by humans, in order to properly set guidelines for ALA and DHA consumption.

54 40 3 Objectives and hypotheses 3.1 Objectives The overall objective of this thesis is to determine if DHA synthesis from ALA can maintain brain DHA. This will be assessed by addressing a series of sub objectives: 1. To determine if rats not consuming DHA have altered DHA concentration and gene expression in brain. 2. To compare brain DHA uptake and accretion rates to whole body DHA synthesis rates. 3. To determine if high dietary LNA affects the whole body DHA synthesis rate. 4. To determine how different levels of dietary ALA affect the whole body DHA synthesis rate. 3.2 Hypotheses Overall Hypothesis DHA synthesis from ALA will be sufficient to maintain brain DHA. Specific Hypotheses 1. Brain DHA concentrations will be lower in rats not consuming DHA, however, gene expression will not be different between rats consuming DHA and rats not consuming DHA. 2. Brain DHA uptake and accretion rates will be lower than whole body DHA synthesis rates. 3. High dietary LNA levels will decrease the DHA synthesis rate.

55 4. High levels of dietary ALA will increase the DHA synthesis rate 41

56 42 Chapter 4 - Whole body synthesis rates of DHA from α- linolenic acid are greater than brain DHA accretion and uptake rates in adult rats. Domenichiello AF, Chen CT, Trepanier MO, Stavro PM, Bazinet RP. J Lipid Res Jan;55(1): Contribution: As first author I performed all of the experimental procedures, all of the data analysis and all of the manuscript drafting.

57 43 4 Objectives 1 and 2: Comparing the rate of DHA synthesis from ALA to brain DHA uptake rates and comparing the effect of DHA and ALA on brain function 4.1 Abstract Docosahexaenoic acid (DHA) is important for brain function, however, the exact amount needed for the brain is not agreed upon. While it is believed that the synthesis rate of DHA from α-linolenic acid (ALA) is low, how this synthesis rate compares with the amount of DHA required to maintain brain DHA levels is unknown. The objective of this work was to assess whether DHA synthesis from ALA is sufficient for the brain. To test this, rats consumed a diet low in n-3 PUFA, or a diet containing ALA or DHA for 15 weeks. Over the 15 weeks, whole body and brain DHA accretion were measured, while at the end of the study, whole body DHA synthesis rates, brain gene expression and DHA uptake rates were measured. Despite large differences in body DHA accretion there was no difference in brain DHA accretion between rats fed ALA and DHA. In rats fed ALA, DHA synthesis and accretion was 100 fold higher than brain DHA accretion of rats fed DHA. Also, ALA fed rats synthesized approximately 3 fold more DHA than the DHA uptake rate into the brain. This work indicates that DHA synthesis from ALA may be sufficient to supply the brain.

58 Introduction α-linolenic acid (ALA) is the most accessible and sustainable source of omega-3 polyunsaturated fatty acids (n-3 PUFA) in the global diet (46). ALA is also a precursor to docosahexaenoic acid (DHA), an n-3 PUFA that is particularly enriched within the brain (9). While it is generally accepted that DHA is important for normal brain function, the amount of DHA required by the brain is not agreed upon (84, ). n-3 PUFA cannot be synthesized by mammals de novo, therefore, DHA must be consumed from dietary sources or be synthesized from shorter chain n-3 PUFA (i.e ALA). To date, reports suggest that the synthesis rate of DHA from ALA is low and perhaps even below detection (33-40, 197, 198). However, plasma concentrations of DHA in vegans are only 0-40% lower than fish eaters despite having no dietary DHA (15, 17). Furthermore, vegan and vegetarian populations do not have an increased risk of neurological disorders (19, 43). The lack of concordance between the low DHA synthesis rates and the relatively normal plasma DHA concentrations in vegans may be due to the methods used to measure DHA synthesis (155). In humans, an ALA tracer is administered orally and the appearance of labeled DHA in the plasma is measured for up to 2 weeks (33-40, 197, 198). From the area under the curve of labeled DHA appearance, the fractional conversion of DHA is calculated (33, 35-37, 39, 197). Alternatively, plasma labeled DHA concentrations over time can be used in modeling programs to determine fractional conversion (34, 38, 40). In both cases, the calculations may preclude a quantitative measurement of the DHA synthesis rate and allow only a comparison between study groups (155). Another concern is the finding that up to 57% of the tracer remains in the adipose after oral consumption of labeled ALA (39). Since the human adipose half-life

59 45 can be longer than one year, it is possible that the tracer is unavailable for DHA synthesis during the study period (152). Recently, Rapoport and colleagues developed a new method in rats to estimate the DHA synthesis rate from ALA (199). This method requires a steady-state infusion of labeled ALA and uses non-linear regression to determine the DHA synthesis rate. Importantly, this method estimates the DHA synthesis rate in rats to be 9.8 μmol/day (199) matching the 11 μmol/day synthesis and accretion rate of longer chain n-3 PUFA, which we estimated from a published balance study performed in growing rats (151). The steady-state infusion method may have advantages because: 1) infusing a tracer to achieve steady-state in the plasma eliminates issues around adipose tissue storage of the tracer and 2) it allows for a quantitative determination of the DHA synthesis rate. The goal of this study was to determine if DHA synthesis from ALA can maintain brain DHA in rats. We addressed this objective by measuring (i) brain and whole body DHA accretion, (ii) DHA synthesis rates from ALA, and (iii) brain DHA uptake rates in rats fed 3 different diets a control diet (low n-3 PUFA), an ALA diet (2% ALA), or a DHA diet (2% DHA). We found that rats fed ALA and DHA accreted similar amounts of brain DHA, which together with our kinetic findings suggest DHA synthesis from ALA is likely sufficient to maintain brain DHA levels. We also mimicked, in rats, the methods used in humans to determine DHA synthesis from ALA by subjecting rats to a gavage with labelled ALA. We found that the rates from this experiment, in rats, were comparable to the results of previously published human studies. Collectively, these results indicate that the rat is an appropriate model for measuring brain DHA synthesis and that brain DHA can be supplied from dietary ALA.

60 Materials and Methods Animals All procedures were performed in accordance with the policies set out by the Canadian Council on Animal Care and were approved by the Animal Ethics Committee at the University of Toronto. Three Long Evans dams each with 18-day-old, male, Long Evans pups were ordered from Charles River Laboratories (St. Constant, QC, Canada). Pups were non-littermates and each dam was housed with pups. Upon arrival dams were allocated to one of the three diets (described below). When the pups were 21 days old, 2 pups from each dam were euthanized by high-energy, head-focused microwave fixation (4.5 kw for 0.9 s). Brains were removed immediately and stored at -80 C until lipid analysis. Bodies were also stored at -80 C immediately after euthanization until analysis. The brains and bodies of these pups were used to determine the PUFA content of the rats at weaning (i.e. baseline PUFA content). The remaining pups were weaned, singly housed and continued on the diet of their respective dam for 15 weeks and were allocated to either the balance study (n=11), the steady-state infusion study (n=4) or the brain DHA uptake study (n=3). For the balance study, the rats were weighed, food intake was determined, and uneaten food was replaced with fresh food on a weekly basis. Rats were euthanized by CO 2 asphyxiation. Brains were removed and dissected sagitally. Carcasses and one half brain per rat were stored immediately in -80 C (to be later analyzed for lipid content). The other half brain was flash frozen with 2-methylbutane on dry ice, and then stored in -80 C until analyzed for mrna expression. The PUFA content in the brains and bodies of these rats after the 15-week feeding period was compared to the baseline PUFA content (described above) to determine PUFA accretion (Eq 1, described below). For the steady-state infusion study

61 47 and the brain DHA uptake study, a jugular vein catheter was surgically implanted into the rats. After allowing 24 hours recovery, rats were infused with 2 H 14 -ALA or 14 C-DHA, respectively (described below). The 2 H 14 -ALA infusion was used to determine DHA synthesis rates following a previously published method (199) and the 14 C-DHA infusion was utilized to measure the brain DHA uptake rate by following previously published methods (81). The study design is illustrated in supplementary figure 1. For the gavage study (n=4), a cartotid artery catheter was surgically implanted in rats that had consumed a diet with ALA but no DHA for 9 weeks post-weaning. After 24 hours recovery the rats were gavaged with 10 mg of 2 H 5 -ALA (described below) Diets Diets were modified from the custom low n-3 AIN-93G purified rodent diet (Dyets inc. Bethlehem, PA) (184). The diet contained 10% lipids (by weight). The fat content of the diet was 32.8% (by weight) safflower oil, 65.2% hydrogenated coconut oil and 2% added oil. The added oils were either DHA ethyl ester (Equateq, Callanish, Scotland), ALA ethyl ester (Equateq), or oleate ethyl ester (Nucheck Prep, Elysian, MN). Each oil was determined to be >98% pure by GC-FID and each diet contained 1 added oil to make the DHA, ALA and control diets, respectively. The custom low n-3 AIN-93G diet is designed to be deficient in n-3 PUFA and as a result the only n-3 PUFA in the diets was that added as ethyl ester oils (DHA or ALA) and a residual amount of ALA that made up 0.25% of the fatty acids. Oleate ethyl ester was added to the control diet to keep total fat content of the diets consistent, and to ensure a constant n-6 PUFA level across all three diets. The fatty acid composition of each diet as measured by gas chromatography flame ionization detection (GC-FID) is shown in Table 1.

62 Table 1. Percent composition of the three diets Fatty Acid Control Diet ALA Diet DHA Diet Gavage Study Diet 10: ND 12: ND 14: : :1n : :1n :1n :2n :3n DHA (22:6 n-3) ND ND 2.00 ND Data shown are means (n=3) expressed as percent of total fatty acids. ND fatty acid concentration below limit of detection 48

63 49 The composition of the diet consumed by rats for the gavage study is shown in Table 1. This diet contained 52 and 5% of the fatty acids as linoleic acid (LNA) and ALA, respectively, with all other PUFA <0.5% Balance study - Whole body fatty acid extraction Carcasses were thawed overnight in a fridge at 3 C, cut into pieces and passed through a #12 hand grinder. After the whole body had passed through the grinder once, the homogenate was mixed together by hand and passed at least one more time through the grinder until the mixture was sufficiently homogeneous. Due to the small size of the pups all of the 21-day-old pup bodies were passed through the grinder together. A portion of the whole body homogenate was weighed and further homogenized (done in duplicate for each 18-week-old rat and the baseline homogenate) using a Polytron Benchtop Homogenizer (Brinkman Instruments, Toronto, ON, Canada) in a mixture of chloroform:methanol:0.88%kcl (2:1:0.8 by volume) (200). The mixture was centrifuged at 500 g for 10 min, and the chloroform layer was extracted. New chloroform was added to the remaining aqueous phase, the mixture was once again centrifuged and the chloroform layer was extracted and added to the previously extracted chloroform layer. This total lipid extract (TLE) was evaporated under N 2, reconstituted in a known volume of chloroform and stored under N 2 in -80 C until further analysis Balance Study - Brain fatty acid extraction One hemisphere was thawed briefly and dissected over ice to separate the brainstem, cerebellum, cortex, hippocampus, striatum and rest of brain. Total lipids were extracted from the whole brain of 21-day-old rats and each region of the dissected 15-

64 50 week half brains by following the method of Folch et al, 1957 (200). Briefly, whole brains, or brain regions were weighed in glass mortars, homogenized with glass pestles, in 0.88% KCl and transferred into a clean test tube. The homogenizer was then washed with methanol, which was then transferred to the test tube containing the homogenized brain and KCl. The homogenizer was washed a final time with chloroform and the chloroform was transferred to the test tube containing the KCl and methanol mixture (2:1:0.8 chloroform:methanol:0.88% KCl by volume). The homogenate mixtures were left at 3 C overnight and were centrifuged at 500 g for 10 min the following morning. The chloroform layer was extracted and new chloroform was added to the aqueous phase. This mixture was centrifuged at 500 g for 10 min and the chloroform layer was extracted and added to the previously extracted chloroform phase Balance study - Fecal PUFA analysis At 8 weeks post-weaning all the feces produced by a rat in 24 hours were collected. Approximately 1 g of feces from 6 rats was then analyzed for PUFA content. Fatty acids were extracted with chloroform:methanol:0.88% KCl (2:1:0.8 by volume) as described above and quantified by GC-FID to determine an estimate of n-3 and n-6 PUFA fecal excretion. Mean PUFA excretion (percentage of intake) was determined and applied to all animals to determine PUFA excretion Transmethylation and gas chromatography-flame ionization detection A known amount of heptadecanoic acid (17:0 - Nu-Check Prep. Elysian, MN) was added to the TLE. Samples were transmethylated using 14% boron trifluoride in methanol at 100 C for 1 hr. Fatty acid methyl esters (FAME) were extracted with hexane

65 51 and quantified using GC-FID. FAME were analyzed using a Varian-430 gas chromatograph (Varian, Lake Forest, CA, USA) equipped with a Varian FactorFour capillary column (VF-23ms; 30 m x 0.25 mm i.d. x 0.25 lm film thickness) and a FID. Samples were injected in splitless mode. The injector and detector ports were set at 250 C. FAME were eluted using a temperature program set initially at 50 C for 2 min, increasing at 20 C/min, and held at 170 C for 1 min, then at 3 C/min and held at 212 C for 5 min to complete the run at 32 min. The carrier gas was helium, set to a constant flow rate of 0.7 ml/min. Peaks were identified by retention times of authentic FAME standards (Nu-Chek Prep, Inc., Elysian, MN, USA). The concentration of each fatty acid was calculated by comparison with the internal standard (17:0) (201). The concentrations were expressed on a per gram basis then multiplied by total weight of the tissue to determine the total amount of each fatty acid in the tissue Balance study - Brain RNA Extraction RNA was extracted from the flash frozen, half brains of the 15-week postweaning rats. The brain was briefly thawed and dissected to isolate the brainstem, cerebellum, cortex, hippocampus, striatum and rest of brain. RNA was extracted from the brain regions using Trizol Reagent (Ambion by Life Technologies, Burlington, ON, Canada) according to the manufacturer s protocol. The tissues were placed in a volume of Trizol Reagent that was 10 times greater than the volume of the tissue and homogenized using a Kontes tissue grinder with plastic pestles (Daigger, Vernon Hills, IL, USA) or for larger regions, a Tissue Ruptor (Qiagen, Germantown, MD, USA). Chloroform was added to the Trizol Reagent at a ratio of 1:5 (Chloroform:Trizol Reagent), the solution was mixed and incubated at room temperature. The solution was then centrifuged at 12

66 g for 15 min at 4 C. The aqueous phase was transferred into a new tube, mixed with isopropanol (1:1 isopropanol:original volume of Trizol Reagent) and incubated for 10 min. The samples were then centrifuged at g for 10 min at 4 C to precipitate the RNA pellet. Following the precipitation of the RNA pellet, isopropanol was removed and the pellet was washed with 75% ethanol (1:2, ethanol:original volume of Trizol Reagent). The samples were then centrifuged at g for 5 min at 4 C and the wash was discarded. The pellet was allowed to air dry, and dissolved in RNase-free water. A Nanodrop 1000 (NanoDrop Technologies Inc. Wilmington, DE, USA) was used to determine the concentration and purity of each sample by measuring absorbance at 260 and 280 nm. Each sample was aliquoted to make 1 μg RNA per 10 μl of sample and stored in -80 C Balance study - Gene Expression Analysis A High-Capacity cdna Reverse Transcription Kit (Applied Biosystems, Burlington, ON, Canada) was used to reverse transcribe 1 μg of RNA, according to the manufacturer s instructions. Newly synthesized cdna was stored in -20 C until the following day when they were loaded onto TaqMan Low Density Array (TLDA) Plates. Quantitative real-time PCR was performed using TLDA (Applied Biosystems) on the 7900HT Real-Time PCR Systems (Applied Biosystems). Fifty ng of cdna was diluted with water to make a volume of 50 μl and mixed with 50 μl of TaqMan Universal PCR Master Mix (Applied Biosystems). The mixture was then loaded onto a customized, preconfigured 384-well TLDA plate according to the manufacturer s protocol. TaqMan gene expression assays were used to assess 21 genes that were previously reported to be differentially expressed with n-3 deprivation. Among these were genes involved in the

67 53 arachidonic acid (ARA) cascade: group 6 ipla 2 (Rn _m1) (185), group 2a spla 2 (Rn _m1) (185), group 4 cpla 2 (Rn _m1) (185) and COX-2 (Rn _m1) (185); genes involved in neuroplasticity: brain derived neurotrophic factor (Rn _s1) (186), transthyretin (Rn _m1) (202), T-cell intracellular antigen 1 (Rn _m1) (202), α-synuclein (Rn _m1) (203); genes involved in the dopaminergic system: dopamine receptor D2 (Rn _m1) (204), vessicular monoamine transporter 2 (Rn _m1) (204), tyrosine hydroxylase (Rn_ _m1) (204); genes involved in learning and memory: retinoic acid receptor α (Rn _m1) (205), retinoid X receptor α (Rn _m1) (205), retinoid X receptor β (Rn _m1) (205), peroxisome proliferator-activated receptor Υ (Rn _m1) (205); genes involved in neurodegeneration and neuroinflammation: uncoupling protein 2 (Rn _m1) (202), TNF α receptor member 1a (Rn m1) (203), heme oxygenase 1 (Rn _m1) (206), 15-lipoxygenase (Rn _m1) (207); as well as epidermal growth factor receptor (Rn _m1) (202), prostaglandin E synthase 3 (Rn _m1) (202). Endogenous controls measured included 18S RNA, phosphoglycerate kinase 1 (PGK1; Rn _g1) and β-actin (Rn _m1). All genes were normalized to PGK1. Similar results were found when normalizing to β-actin or 18S rrna Steady-state infusion study - Surgery and 2 H-ALA infusion At 15 weeks post-weaning rats were subjected to surgery to implant a catheter into their jugular vein. The animals were anesthetized using isoflurane inhalation (5% induction, 1-3% maintenance). Before the incision was made, hair was shaved from the incision site and the site was sterilized with iodine and ethanol. A transverse incision was

68 54 made anterior to the upper thorax. The jugular vein was located by blunt dissection. The vein was isolated and tied off. The vein was then nicked and a catheter (PE 50, Intramedic, Sparks, MD, USA) with a 3.5 cm silastic tubing end (VWR, Mississauga, ON, Canada) was inserted into the vein. The catheter was secured using 3.0 silk suture and a 16 gauge angiocath (Becton Dickinson, Mississauga, ON, Canada) was used to guide the catheter subcutaneously to a site outside the body near the scapula. An incision was made at this site and the catheter was tucked beneath the skin at the incision site. The incision site was stapled to protect the catheter from the rats. The incision site on the chest was closed with 4.0 silk sutures and the rats were allowed to recover from the anesthetic under a heat lamp. Approximately 24 hours after the surgery, the tail vein of the rats was cannulated with a 24 gauge angiocath (Becton Dickinson). While the rats were restrained, the staple closing the incision site on the scapula was removed and the jugular vein catheter was connected to a longer polyethylene catheter. The rats were then placed in an infusion box that contained food, a chew toy and bedding from the rat s cage. The tail vein catheter was then connected to an infusion line. Rats were able to move freely within the infusion box throughout the infusion. Modified from the method of Rapoport et al, 2010; 4.5 μmol/100 g body weight of 2 H-ALA ( 2 H 14 -ALA, purity > 95% confirmed by GC-FID and GC-MS; Cayman Chemical, Ann Arbor, MI, USA) was infused into the tail vein for 3 hours (181). To prepare the infusate, a known amount of 2 H-ALA was dissolved in 5 mm HEPES buffer (ph 7.4) containing 50 mg/ml fatty acid-free bovine serum albumin. The infusate was mixed by sonication at 37 C. An infusion pump (Harvard Apparatus PHD 2000; Holliston, MA, USA) was used to infuse 3.78 ml of tracer solution at a constant rate of

69 ml/min for 3 hours. Immediately before the infusion and every 30 min during the infusion 0.2 ml of blood was drawn from the jugular vein. The jugular vein catheter was flushed with heparinized saline (5% by volume) after every blood draw to prevent coagulation in the line. After 180 min of the infusion, 1 ml of blood was drawn from the jugular vein and the animals were euthanized with a lethal injection of T-61 into the tail vein. All blood samples were centrifuged for 10 min (PC-100 microcentrifuge, Diamed, ON, Canada) and the plasma was collected and stored at -80 C Steady-state infusion study - Determination of Plasma Volume Plasma volume was determined using the method of Schreihofer et al 2005 (199, 208). Briefly, a known amount of Evans Blue dye was injected into the tail veins of the rats. After 15 min, 1 ml of blood was drawn from the jugular vein, twice. The plasma was collected as described above, and 0.1 ml of plasma was diluted into 1 ml of saline. Absorbance at 604 nm was measured with a Nanodrop 1000 and by comparing the absorbance to a standard curve, the concentration of the dye was determined Plasma Lipid Extraction Steady-state infusion study Fifty μl of plasma was added to a test tube which contained known amounts of unesterified 17:0 standard and di-17:0 phosphatidylcholine standard. Lipids were extracted from the plasma using the method of Folch et al 1957 (200), as described above Steady-state infusion study Thin Layer Chromatography Thin layer chromatography (TLC) was used to separate esterified and unesterified lipids. The TLE was dried under N 2 and reconstituted in 250 μl chloroform. TLC plates

70 56 (TLC Silica gel 60, EMD) that were washed in chloroform and methanol (2:1) were activated by heat at 100 C for 1 hr. TLE was loaded onto the TLC plates and the plates were run in heptane-diethyl ether-glacial acetic acid (60:40:2 v/v/v) alongside authentic standards (Nu-Check Prep, MN). The plates were sprayed with 0.1% (w/v) 8-anilino-1- naphthalenesulfonic acid. Total phospholipid, unesterified fatty acid, triglyceride and cholesteryl ester bands were identified under UV light by comparison to standards. Esterified lipid (phospholipid, triglyceride and cholesteryl ester) bands were collected and transferred to a glass test tube. The unesterified fatty acid band was scraped off and transferred to a separate glass test tube Steady-state infusion study - Plasma lipid hydroxylation and esterification Plasma samples taken at time 0 min of the infusion were transmethylated as described above and run on GC-FID. Unesterified fatty acids were extracted from the silica using chloroform:methanol:0.88% KCl (2:1:0.8) as described above. Esterified lipids were hydrolyzed with 1 ml of 10% methanoic KOH at 70 C for 1 hr (199). After the reaction 1 ml of concentrated HCl was added to the sample followed by 1 ml of dh 2 O. The fatty acids were extracted twice with 3 ml of hexane. The unesterified lipids (free fatty acids and hydrolyzed esterified fatty acids) were converted to fatty acidpentafluorobenzyl (PFB) esters by following the method of Strife et al 1984 (209). One hundred μl of a mixture of pentafluorobenzylbromide-diisopropylamine-acetonitrile (10:100:1000 by volume) was added to the samples, which were then shaken for 15 min. The PFB mixture was then evaporated under N 2 gas and the fatty acid-pfb esters were reconstituted in 50 μl of hexane and ran on the GC-Mass Spectromemeter.

71 Gas Chromatography-Mass Spectrometry Fatty acid-pfb esters were analyzed using an Agilent 6890 series gas chromatograph (Agilent Technologies, Wilmington, DE, USA) equipped with a DB- FFAP capillary column (30 m X 0.25 mm ID, 0.25 mm film thickness; J and W Scientific, Folsom, CA, USA) according to the method of Pawlosky et al 1992 (210). Fatty acid-pfb esters dissolved in hexane were injected with a splitless injection technique. The GC oven temperature was programmed from 80 C to 185 C at 20 C/min then to 240 C at 10 C/min and held for 30 min. Injector and transfer line were maintained at 250 C and 280 C, respectively. The NCI source temperature was 150 C. Methane was the ionization gas. Selected ion mode was used to analyze fatty acids using the M-PFB ion for PFB derivatives. The m/z values for 2 H-ALA and 2 H-DHA were 291 and 337, respectively. Concentrations were determined by using calibration equations that relate the fatty acid peak area:standard peak area ratio to a fatty acid concentration Brain DHA uptake study - 14 C-DHA infusion At 15 weeks post-weaning a surgery was performed to implant a catheter into the jugular vein (described above). Approximately 24 hours after the surgery the tail vein of the rats was cannulated and 14 C-DHA (76 μci per rat) was infused into the tail vein at a rate of 0.223(1+e -19.2t ) ml/min (t is infusion time in minute) for 5 min. At approximately 0, 15, 30, 45, 90, 180, 240 and 300s during the infusion 200 μl of blood was drawn from the jugular vein and the plasma was extracted and stored in -80 C as described above. After 5 min of 14 C-DHA infusion rats were euthanized by high-energy, head-focused microwave fixation (13.5 kw for 1.6 s). The brains were then removed, dissected sagitally and stored in -80 C until analysis.

72 Brain DHA uptake study - Plasma radioactivity analysis Lipids were extracted from 50 μl of plasma using the method of Folch et al 1957 (200), as described above. A known portion of the plasma TLE was then added to a scintillation vial with 5 ml of scintillation cocktail (GE Healthcare Life Sciences, Baie d Urfe, QC, Canada). Radioactivity was measured using liquid scintillation counting to determine the plasma area under the curve (AUC) for radioactivity. For plasma samples drawn at time 0 s of the infusion the unesterified lipids were isolated by TLC and quantified using GC-FID (described above) Brain DHA uptake study - Liquid scintillation counting Radioactivity was quantified by a Packard TRI-CARB2900TR liquid scintillation analyzer (Packard, Meriden, CT, USA) with a detector efficiency of 48.8%. Radioactivity was expressed in disintegrations per minute; then converted to nci Brain DHA uptake study - Brain radioactivity analysis Brain hemispheres were homogenized and lipids extracted as described above. A known portion of the TLE was then added to a scintillation tube with 5 ml of scintillation cocktail. Radioactivity was quantified using liquid scintillation counting Gavage study - Surgery and blood sampling At 10 weeks post-weaning rats were subjected to a surgery to implant a catheter into their carotid artery. The animals were anesthetized using isoflurane inhalation (5% induction, 1-3% maintenance). Before the incision was made, hair was shaved from the incision site and the site was sterilized with iodine and ethanol. A transverse incision was made anterior to the upper thorax. The carotid artery was located by blunt dissection. The

73 59 artery was isolated and tied off. The artery was then nicked and a catheter (PE 50, Intramedic, USA) was inserted into the vessel. The catheter was secured using 3.0 silk suture and a 16 gauge angiocath (Becton Dickinson) was used to guide the catheter subcutaneously to a site outside the body near the scapula. The incision site on the chest of rats was closed with 4.0 silk sutures and the rat was then injected with 1ml of saline solution, subcutaneously and allowed to recover from the anesthetic under a heat lamp. The day after the surgery 200 μl of blood was drawn from the carotid artery catheter, the rat was then gavaged with 10 mg of 2 H 5 -ALA that was dissolved in 1 ml of olive oil (Sigma-Aldrich Corporation, St. Louis, MS, USA). To prepare the gavage solution 2 H 5 - ALA ethyl ester (Cambridge Isotope Laboratories Inc. Andover, MA, USA), which was generously donated by Dr. Stephen Cunnane, was hydrolyzed with 10% methanolic KOH at 70 C for 1 hour. Unesterified fatty acids were formed by adding 1 ml of HCl followed by 1 ml of dh 2 O. Lipids were extracted twice with 3 ml of hexane. The unesterified 2 H 5 - ALA was then dried down and mixed with olive oil to make a solution that was 10 mg 2 H 5 -ALA per ml of olive oil. Blood samples (200 μl) were drawn at 30, 60, 90, 120, 180, 240, 300, and 360 min after the gavage. Plasma was separated and stored in -80 C as described above. Lipids were extracted from 50 μl of plasma by the Folch method, and the TLE were hydrolyzed to create unesterified fatty acids (described above). Hydrolyzed TLE were stored in -80 C until analyzed with liquid choromatography tandem mass spectrometry (LC-MS/MS) Liquid Chromatography(LC) Tandem Mass Spectrometry (MS/MS) Half of the TLE was evaporated under N 2 gas and reconstituted in 100 μl of water/acetonitrile (80:20 v/v). Fatty acids were detected using an Agilent HPLC 1290

74 60 (Agilent Technologies, Santa Clara, CA, U.S.A) equipped with an Agilent Zorbax SB- Phenyl column (3 x 50 mm, 3.5μm; Agilent, Santa Clara, CA, U.S.A). The initial HPLC conditions of elution were set at 500 μl/min gradient system consisting of (A) 50% water and (B) 50% acetonitrile. The gradient started with 50% (A) and 50% (B) and maintained for 1.5 minutes, increased to 100% (B) from 1.5 to 6 minutes and maintained at 100% B for 4 minutes to complete the total run of 10 minutes. Mass spectrometry analyses were carried out on QTRAP 5500 triple quadruple mass spectrometer (AB SCIEX, Framingham, MA, U.S.A) in electrospray ionization, negative ion mode. The source temperature was 600 C and the ion spray voltage was ev. The optimized parameters were as follows: de-clustering potential -40, entrance potential -10, collision energy -20, and collision cell exit potential -11. Mass transitions for 2 H 5 -ALA, 2 H 5 - eicosapentaenoic acid (EPA), 2 H 5 -docosapentaenoic acid n-3 (DPAn-3) and 2 H 5 -DHA were: to 59.0, to 262.2, to m/z and to 288.2, respectively. Concentration was quantified by comparing the peak area ratios (peak of interest:internal standard) and correcting for a response factor that was determined for each fatty acid of interest. Response factors were determined by analyzing a standard mixture of 100ng/mL each of DHA, EPA, ALA, DPA, and 2 H 8 -AA (arachidonic acid) by LC/MS/MS and comparing peak areas for each of the 4 fatty acids in relation to the peak area for 2 H 8 -AA to generate response factors. The response factors were 10, 0.75, 0.25, 0.75 for ALA, EPA, DPAn-3 and DHA, respectively Balance study - Equations In the balance study, PUFA content in the whole body of animals at time 0 (21 days of age) and after 15 weeks (126 days of age) consuming the diets were determined.

75 61 Since PUFA cannot be synthesized de novo it is possible to determine the accretion of a specific PUFA by subtracting the baseline amount from the amount of that PUFA in rats at 15 weeks post-weaning. Using equation 1 it is then possible to determine the metabolic consumption of the dietary PUFA (151). MetabolicConsumption consumption i excretion i (accretion i accretion x ) (Eq. 1) Where i refers to a PUFA consumed in the diet of the animals and x refers to a longer chain PUFA synthesized from i Steady-state infusion study - Equations To determine the DHA synthesis rate (steady-state infusion study), appearance of 2 H-DHA in the plasma-esterified pool was measured and fit to a Boltzmann sigmoidal curve ([ 2 H-DHA]*plasma volume vs. time) using non-linear curve (199) (Graphpad Prism Version 4.0, La Jolla, CA, USA). At any point on this curve the, slope ( ) will be determined by the ability of the body to synthesize 2 H-DHA from 2 H-ALA and the ability of the periphery to uptake 2 H-DHA (Eq. 2). S k 1,DHA [ 2 H ALA] unesterified k 1,DHA [ 2 H DHA] esterified (Eq. 2) Where k 1,DHA is the steady-state synthesis-secretion coefficient for DHA, [ 2 H- ALA] unesterified is the plasma concentration of the infusate, k -1,DHA is the disappearance coefficient for DHA and [ 2 H-DHA] esterified is the concentration of DHA in the plasma that has been synthesized from the infusate, packaged into a lipoprotein and secreted into the plasma. The maximum first derivative (S max ) of this curve is assumed to be the time point when the uptake of esterified DHA from the periphery is negligible, i.e. 0 (Eq. 3). S max k 1,DHA [ 2 H ALA] unesterified 0 (Eq. 3)

76 62 Therefore, the derivative at this point is equal to the rate of 2 H-DHA synthesis. By correcting the S max by the tracee:tracer ratio, the rats actual DHA synthesis is determined, J syn,dha (nmol/min) (199). J syn,dha S max [ALA] unesterified [ 2 H ALA] unesterified k 1,DHA [ALA] unesterified (Eq. 4) Brain DHA uptake study - Equation To determine the brain DHA uptake rate, the incorporation coefficient for DHA from the unesterified plasma pool into the brain total lipid pool is (81): k * c * br(t) T c * pl dt 0 (Eq. 5) Where c * br (T) is the total brain radioactivity at the end of the infusion and c * pl dt is the plasma radioactivity AUC. Multiplying the incorporation coefficient by the concentration of plasma unesterified DHA (c pl ) allows for the determination of the brain DHA uptake rate: J in k * c pl (Eq. 6) T Statistics Mean lipid concentrations, and accretions as well as mean RQ for genes and brain DHA uptake rates were compared by one-way ANOVA using Tukey s test for multiple comparison (Graphpad Prism version 4.0, La Jolla, CA, USA). If variances were determined to be unequal, by Bartlett s test for equality of variances, then the Kruskal-

77 63 Wallis test was used to compare the means followed by Dunn s test for multiple comparisons. Mean plasma lipid concentrations were compared using Kruskal-Wallis test followed by Dunn s test for multiple comparisons. DHA synthesis rates between rats consuming the ALA and control diet was compared using Student s t-test due to the fact that DHA synthesis rates were not quantifiable in rats consuming the DHA diet. 4.4 Results Body weight and food intake Balance study There were no statistical differences in body weight and food intake between rats consuming different diets throughout the 15-week feeding period. At 15 weeks, mean weights were g, g, and g in rats fed the control, ALA, and DHA diet, respectively. Average weekly food intake throughout the study was g, g, and g for the control, ALA, and DHA diet, respectively Fecal Excretion of PUFA Balance study On average, 0.27 ± 0.002, and 0.5 ± 0.05 % of dietary ALA, DHA and LNA respectively, were excreted in the feces Baseline PUFA concentrations Balance study Total n-6 and n-3 PUFA in 21-day-old rats was 1505 ± 35 μmol and 148 ± 4 μmol, respectively (Supplementary Table 1 and Table 2). The majority of PUFA were in the carcasses of the rats. LNA was the main n-6 PUFA at baseline 1279 ± 30 μmol (data

78 64 not shown). DHA and ALA were the most abundant n-3 PUFA at baseline (55 ± 1 and 41 ± 0.9 μmol, respectively, data not shown) Final PUFA concentrations Balance study Concentrations of major n-3 PUFA in the brain and whole body are shown in Figure 1. Whole body ALA concentrations (Figure 1a) were 30 fold greater in rats consuming the ALA diet compared to rats consuming the DHA and control diet (ALA diet>dha diet=control diet, p<0.05). Whole body DHA concentrations (Figure 1b) were highest in rats consuming the DHA diet, followed by the rats consuming the ALA diet and lowest in rats consuming the control diet (p<0.05). Regional brain DHA concentrations did not differ significantly between rats consuming the DHA and ALA diet (Figure 1 c-h). However, DHA concentrations were significantly lower in rats consuming the control diet compared to rats consuming the ALA and DHA diets in all brain regions (p<0.05). Total (body + brain) n-6 PUFA concentrations did not differ between rats consuming different diets (Supplementary Table 1). However, docosapentaenoic acid n-6 (DPAn-6) in both the brain and body were highest in the rats consuming the control diet, followed by the rats consuming the ALA diet and lowest in rats consuming the DHA diet (Supplementary Tables 1 and 2, p<0.05). Whole body ARA concentrations only differed between rats consuming the control and DHA diet (2280 ± 98 μmol vs ± 44 μmol respectively, p<0.05). Brain ARA concentrations only differed in the brainstem of rats fed the control and DHA diet, while brainstem ARA concentrations for rats fed the ALA diet were similar to those of rats fed both diets (Supplementary Tables 3-8, P<0.05).

79 Figure 1 Whole body ALA and DHA concentrations and brain DHA concentrations in rats consuming the control, ALA or DHA diet. (a) Whole body ALA concentration (μmol/g) is highest in rats consuming ALA and is not different in animals consuming DHA or the control diet. (b) Whole body DHA concentration (μmol/g) is highest in rats consuming DHA diet>ala diet>control diet (c) Brain DHA concentrations (nmol/g) are not different in rats consuming the ALA and DHA diet but are significantly lower in rats consuming the control diet in the cortex, (d) cerebellum, (e) striatum, (f) hippocampus, (g) brainstem and (h) rest of brain. All data are mean ± SEM. Different letters signify the means are significantly different (p<0.05) measured by One-Way ANOVA followed by Tukey s multiple comparison test or Kruskal-Wallis test followed by Dunn s multiple comparison test (if variances were significantly different). n=11 65

80 PUFA accretion Balance study Table 2 and supplementary table 1 summarize the n-3 and n-6 PUFA balance. Total and body DHA accretion (Figure 2a and b) was highest in rats consuming the DHA diet, followed by rats consuming the ALA diet followed by rats consuming the control diet (p<0.05). Brain DHA accretion (Figure 2c), however, did not differ in rats consuming the ALA and DHA diets but was significantly lower in rats consuming the control diet (P<0.05). Rats consuming the ALA diet synthesized and accreted 4.2 ± 0.4 μmol/day of total DHA (Figure 2d). The DHA uptake and accretion rate into the brain, for rats consuming the DHA diet was 42 ± 7 nmol/day (figure 2e).

81 67 Table 2. Summary of n-3 PUFA Balance Dietary Group: Control ALA DHA Intake of n-3 PUFA (μmol) 2560 ± 69* ± 661* ± 627** Fecal Excretion (μmol) 7 ± ± 2 10 ± 0.4** Body Content of n-3 PUFA (μmol) Day ± ± ± 3 Day ± 23 a 2352 ± 166 b 1607 ± 106 b Brain Content of n-3 PUFA (μmol) Day 0 10 ± ± ± 0.6 Day ± 1 a 14 ± 2 b 15 ± 3 b Total Accretion (μmol) ALA 47 ± 6 a 1500 ± 100 b 51 ± 7 a 20:3n-3 84 ± 7 a 110 ± 11 ab 164 ± 21 b EPA 45 ± 3 a 70 ± 10 b 71 ± 8 b DPAn-3 5 ± 2 a 90 ± 7 b 66 ± 7 b DHA 60 ± 6 a 440 ± 39 b 1119 ± 62 c Total n-3 PUFA 240 ± 17 a 2209 ± 107 b 1472 ± 99 b Metabolic consumption of dietary PUFA (μmol) 2317 ± 61 a ± 617 b ± 548 b ** Data are means SEM, different letters signify means are significantly different (p<0.05) measured by One-Way ANOVA followed by Tukey s test for multiple comparisons. n=2 for day 0 measurements and n=11 for day 105 measurements. *Consumption of ALA **Refers to DHA

82 Figure 2 DHA kinetics. (a) Total DHA accretion (body + brain DHA accretion) and (b) body DHA accretion (μmol) was highest in rats consuming DHA>ALA>control diet. (c) Brain DHA accretion (μmol) was significantly lower in rats consuming the control diet compared to the rats consuming the ALA and DHA diet. There was no difference in brain DHA accretion between rats consuming the ALA and DHA diets. (d) Total DHA accretion rate (μmol/day) in rats consuming the 3 diets. For rats consuming the ALA diet this is the synthesis-accretion rate (4.2 μmol/day). (e) Brain DHA accretion rate (μmol/day) of rats consuming the 3 diets. Rats consuming the DHA diet accreted μmol DHA/day in their brains. (f) Brain DHA uptake rate (nmol/day). All data are mean ± SEM. Different letters signify the means are significantly different (p<0.05) measured by One-Way ANOVA followed by Tukey s multiple comparison test or Kruskal-Wallis test followed by Dunn s multiple comparison test (if variances were significantly different). n=11 for a-e and n=3 for ALA and control group and 2 for DHA group. 68

83 Gene Expression Balance study Relative gene expression of 21 genes was measured in all 6 brain regions. A heat map representing gene expression in the cortex and striatum illustrates regional differences in gene expression (Figure 3). While there were strong effects of brain region, diet did not have an effect on gene expression for most of the measured genes. There were only 11 differences in gene expression out of 378 comparisons. A full list of relative gene expression of all the genes measured is available in supplementary tables Due to the high number of comparisons, it is not possible to rule out these differences as a chance finding Plasma Volume Steady-state infusion study Mean plasma volume (V plasma ) was measured to be ml/kg which agrees with a previous report (211) Plasma Concentrations Steady-state infusion study Unesterified ALA in the plasma was higher in rats consuming the ALA diet compared to rats consuming the control diet (Supplementary Table 15, p<0.05) but did not differ from rats consuming the DHA diet (4 0.9 nmol/ml, p>0.05). Rats consuming the DHA diet had significantly higher plasma unesterified DHA levels than rats consuming the control diet (65 21 vs nmol/ml p<0.05). Rats consuming the ALA diet did not have statistically different concentrations of plasma unesterified DHA compared to rats consuming the DHA or control diet (4 0.9 nmol/ml, p>0.05). Plasma esterified ALA and DHA followed the same pattern as plasma unesterified ALA and DHA (Supplementary Table 16).

84 70 Figure 3 Heat map depicting gene expression in cortex and striatum brain samples. Figure illustrates that there are no differences within brain regions for any of the measured genes (average linkage and Pearson s distance metric). The clustering of all the cortex (CTX) samples and all the striatum (STR) samples together indicate that samples of the same brain region are highly correlated. There is also a clear difference in dopamine receptor D2 indicating brain regions were properly dissected as there was less expression of this receptor in the cortex relative to the striatum. Expression profiles are illustrated as ΔCt values with red indicating higher expression and green indicating lower expression compared to a striatal sample from a rat fed the ALA diet. Dendrograms indicate the correlation between groups of samples and genes. Samples are in columns and transcripts in rows. Gene names are available in the methods section of this paper. n=10

85 DHA synthesis Steady-state infusion study Curves in Figure 4a show V plasma x [ 2 H-DHA] esterified plotted vs. time for 3 rats (1 per dietary group). Due to the fact that [ 2 H-DHA] esterified did not increase throughout the infusion, in rats consuming the DHA diet, and thus, the data did not fit to a sigmoidal curve, a DHA synthesis rate was not calculated in these animals. Mean infusion parameters are presented in Table 3. Mean [ 2 H-ALA] unesterified was nmol/ml and nmol/ml in the ALA diet and control diet groups, respectively. Mean S max for rats consuming the ALA diet was nmol/min and nmol/min in rats consuming the control diet (p=0.12). Corresponding daily DHA synthesis rates were nmol/day and nmol/day in rats consuming the ALA and control diet, respectively (p<0.05, table 3).

86 Figure 4 Example of infusion curves for a rat from each diet group. (a) Plasma volume (ml) x concentration of esterified 2 H-DHA (nmol) plotted against time (min) and fit to a sigmoidal curve. This figure shows the resulting infusion curve for one rat from each dietary group that was infused with unesterified 2 H-ALA for 3 hours. (b) First derivatives of the curves from figure 4a. The maximum first derivative is used to determine the DHA synthesis rate. For the rat consuming DHA there was no increase in esterified 2 H-DHA throughout the infusion resulting in an artificially high maximum first derivative therefore the DHA synthesis rate in this rat was deemed unquantifiable. 72

87 73 Table 3. Parameters for whole-body synthesis-secretion of DHA in rats consuming the control, ALA and DHA diet for 15 weeks Daily Secretion Rate Diet S max, i (nmol/min) k 1,DHA (ml/min) J syn (nmol/min) (nmol/day) F DHA (/min) t 1/2 (days) Control ± ± ± ± 19 2E-05 ± 5E ± 29 ALA ± ± 0.04* ± 0.274* 1452 ± 395* 2E-04 ± 4E-05* 3 ± 1* DHA ND ND ND ND ND ND Data are mean SEM (n=4, independent samples per group), S max, i = maximum first derivative, k 1, DHA = synthesis-secretion coefficient for DHA synthesis from ALA, j syn = Synthesis rate of DHA from ALA, F DHA = Turnover rate of esterified plasma DHA, t 1/2 = half-life of esterified plasma DHA * p<0.05 vs. control diet

88 Brain DHA uptake rate Brain DHA uptake study There was no significant difference in k * between rats fed the different diets ( x 10-4, x 10-4, and x 10-4 ml/s/g; control, ALA, DHA diet respectively). Brain DHA uptake rate was significantly higher in the DHA fed rats vs. the ALA and control fed rats ( > = nmol/day respectively, p<0.05). There were no statistical differences in brain DHA uptake rates between the ALA and control fed rats (Figure 2f) Appearance of 2 H 5 -n-3 PUFA Gavage study Figure 5 illustrates the appearance of 2 H 5 labeled n-3 PUFA in the plasma of rats that were gavaged with 10 mg of 2 H 5 -ALA. ALA was the first labeled n-3 PUFA to appear in the plasma and was consistently at the highest concentration. Labeled longer chain n-3 PUFA EPA, DPA n-3 and DHA appeared in the plasma at later time points and concentrations of these PUFA were lower compared to ALA.

89 75 Figure 5 2 H 5 -n-3pufa appearance in plasma of rats gavaged with 10 mg of 2 H 5 - ALA. Data are expressed as mean percentage of dose recovered SEM at time points after the gavage. Different colored lines represent different 2 H 5 -n-3 PUFA. Data from this graph was used to model the methods used in humans to calculate DHA synthesis. n=4.

90 Discussion We showed that brain DHA levels in the adult rat can be maintained by dietary ALA just as well as by dietary DHA. This was supported by the finding that dietary ALA and DHA resulted in the same level and accretion of brain DHA after 15 weeks. From the 15 week balance study, the accretion of brain DHA in the ALA and DHA fed rats did not significantly differ, and equaled ± and ± mmol per day, respectively (Figure 2e). When we compare these daily brain accretion rates to the whole body DHA synthesis rate in the ALA fed rats in the balance study, we see that DHA synthesis rates exceed brain uptake rates by 100 fold (Figure 2d). To illustrate this, a summary of study designs and results is shown in Figure 6. Using the steady-state infusion method we estimated synthesis rates of DHA from ALA to be 1.5 and μmol/day in animals consuming the ALA and control diets, respectively. These rates are lower, but in line with previously published estimates using the steady-state infusion technique (199, 212, 213). Differences could be due to the fact that we performed the steady-state infusion in rats fed a diet containing 2% of the total fatty acids as ALA, whereas Gao et al. 2009, fed their animals a diet containing approximately 5% of the total fatty acids as ALA. The differences in synthesis rates could also be due to the differences in age and strain of the rats. In particular, different rat strains are known to have different desaturase enzyme activity (214). Another important difference between our kinetic studies compared to the studies done by others is that our infusions were performed in completely free-living rats, 24 hours after recovery from anesthesia (199, 215). Additionally, using a free-living infusion model, we determined the rate of DHA uptake

91 77 from the unesterified plasma pool into the brain. Brain DHA uptake rates were between 189 and 618 nmol/day, which is similar to rates that have been previously published by others (215, 216).

92 Figure 6 Summary of the methods and results. Three studies were performed in rats fed the control, ALA or DHA diet from weaning for 15 weeks. After 15 weeks the brain and body were collected to perform the balance study. For the steady-state infusion study a jugular vein and tail vein catheter were implanted into the rats and the rats were infused with 2 H-ALA to measure whole body DHA synthesis from ALA in rats fed the 3 diets. For the brain DHA uptake study, tail vein and jugular vein catheters were implanted into the rats and 14 C-DHA was infused to determine the brain DHA uptake rate in rats fed the 3 diets. 78

93 79 Despite sizeable differences in n-3 PUFA concentrations and accretions in the bodies, we were unable to detect differences in brain DHA concentrations and accretions between rats consuming diets with DHA or ALA as the only n-3 PUFA source. In our balance study we calculated a DHA synthesis rate of 4.2 μmol/day in animals fed the ALA diet. We also calculated, in rats consuming the DHA diet, a brain uptake and accretion rate for DHA of 42 nmol/day. Previous work in our lab found that mice fed DHA at a concentration 2% of the fatty acids attained maximal DHA concentrations in the brain (217). Since rats fed the ALA diet were able to synthesize 100 fold more DHA than the amount of DHA accreted in the brains of rats consuming the DHA diet; and there were no significant differences in brain DHA concentrations or accretions in rats fed these two diets, it is likely that rats fed the ALA diet were able to synthesize sufficient DHA to maintain brain DHA levels. The finding that brain DHA concentrations are not different between rats fed the ALA and DHA diet is in contrast to previously published work (136). Abedin et al in 1999 reported that DHA in the brain phosphatidylethanolamine fraction was higher in Guinea pigs fed a diet containing DHA as compared to a diet containing ALA. Our study differed from this work, however, in that we measured brain total lipids, not brain phospholipid fractions, and we fed our rats pure fatty acids ethyl esters whereas Abedin et al used oil sources to formulate their diets. As such, the high DHA diet also contained ALA in the study by Abedin et al. Furthermore, the results of our kinetic studies support the idea that rats consuming the ALA diet were able to synthesize enough DHA to supply the brain. Using the steadystate infusion method we calculated that rats consuming the ALA diet synthesized 1.5 μmol/day of DHA which is about 3 fold higher than the amount of DHA these rats uptake

94 80 into the brain (468 nmol/day) as calculated by a 5 min infusion of radiolabeled DHA. Also, DHA synthesis rates were higher in rats fed a diet containing ALA compared to rats fed a diet with no added n-3 PUFA. This suggests that the ALA substrate was the ratelimiting factor for DHA synthesis from ALA. Contrary to others, we were unable to measure a DHA synthesis rate in rats fed a diet containing DHA (199). Our work indicates that 2% dietary DHA reduced DHA synthesis by greater than 30-fold (199). It has been previously reported that n-3 PUFA feeding downregulates the expression of the hepatic desaturase enzymes required for DHA synthesis from ALA (100). Despite the similar brain DHA concentrations in rats fed the ALA and DHA diet, rats consuming DHA had an almost 2-fold greater uptake rate of DHA into the brain. This means that the brains of rats fed the DHA diet took up and metabolically consumed more DHA. Despite increased metabolism of DHA in the brain, there were no differences in gene expression between rats on either diet. It is conceivable that exposing these rats to a stressor (such as brain trauma, neuroinflammation, etc.) would result in differential gene expression in the brains of these rats. Therefore, future experiments should measure the effect of diet on brain gene expression in rats that have been exposed to stress. The steady-state infusion method developed by Rapoport et al in 2009 is an in vivo kinetic approach to measure DHA synthesis rates at a given time whereas the balance method measures an average synthesis rate over the balance period (218). Using the steady-state infusion method we calculated that rats consuming the ALA diet were synthesizing 1.5 μmol of DHA per day, at 15 weeks post-weaning; whereas using the balance method we calculated that over the 15 weeks these rats synthesized, on average, 4.2 μmol of DHA per day. One explanation as to why the balance method gives a higher

95 81 DHA synthesis rate than the steady-state infusion method is because the balance study measures an average DHA synthesis rate over the 15 weeks. Therefore included in the average rate is the growth and development period of the rat, when DHA synthesis rates are likely to be highest. In contrast, the steady-state infusion method measured DHA synthesis rates at the end of the 15-week feeding period. This rate is only a measure of the synthesis rate in these rats at this time, when the rats are older and when the DHA synthesis rate is also likely low. It has been shown previously that the DHA synthesis rate decreases with age in rats (213). Using the balance method to calculate DHA synthesis rates is advantageous because it measures DHA in all body pools to determine how much DHA was synthesized and accreted. This method is limited, however, because it cannot determine how much DHA was synthesized and then metabolically consumed. Therefore this method likely underestimates the actual DHA synthesis rate. While the steady-state infusion method only measures the DHA in plasma pool, this is unlikely to be a limitation because the method utilizes kinetic modeling to calculate the actual DHA synthesis rate. The steady-state infusion method is advantageous because it measures a DHA synthesis rate in conditions where the substrate may be rate limiting which can be different from in vitro kinetic models where the substrate is assumed to not be rate limiting. More importantly, the steady-state infusion method may be advantageous to the oral administration method that is performed in humans to measure DHA synthesis. When ALA is consumed orally, the majority will be β-oxidized or stored in the adipose. Our study as well as other balance studies found that the majority of ALA intake is metabolically consumed and not accumulated in the tissues (151, 180). In our study

96 82 approximately 90% of dietary ALA was metabolically consumed. While we did not measure adipose composition, other balance studies have found the majority of PUFA that are not metabolically consumed are stored in the adipose (151). These findings are in agreement with oral tracer administration studies performed in humans (39). This is problematic because tracer that is stored in the adipose would not be available for DHA synthesis in the liver over the duration of the study. The steady-state infusion method eliminates this problem because the tracer is infused in a way that it achieves a steadystate level in the plasma despite uptake into the adipose. It is generally accepted that the rat can synthesize more DHA than the human, however, the methods used to measure DHA synthesis in the human have not been validated in the rat. In humans, DHA synthesis is measured by administering an oral bolus of labeled ALA and measuring the appearance of labeled DHA in plasma. These calculations can be applied to the data from our gavage study to mimic the human method for determining DHA synthesis. When we applied 3 of the calculations previously performed in humans (33, 39, 42) to our data in rats, we obtained 3 different values for DHA synthesis. When using the calculation of McCloy et al. 2004, mean AUC for 2 H 5 -DHA in our rats was 0.31 % of dose which was in line with the value previously published in humans (0.99 % of dose, corrected for plasma volume assuming plasma volume equals 4.5% of body weight) (39). When we applied the calculation of Emken et al. 1994, to our data in rats, the rat had lower average percent conversion to DHA compared to previously published values in humans (0.64% vs. 3.8%, respectively) (33). Finally, when we used the calculation of Gillingham et al. 2013, we found that the value of apparent conversion to DHA in the rat was comparable to that found in humans (0.12

97 83 % of dose recovered as DHA in rat vs % of dose recovered as DHA in human) (42). Therefore, the results from the oral gavage study are inconsistent with the belief that the rat is more efficient at synthesizing DHA than the human. Also, the fact that applying different kinetic calculations to the same data gave markedly different results for DHA synthesis indicates that the calculations used to measure DHA synthesis are inconsistent and should be used largely to compare relative DHA synthesis between experimental groups within a study (155). While pilot data indicated that labeled DHA peaked within 6 hours of an oral gavage in a rat, if this study was extended beyond 6 hours the DHA synthesis rates measured could be higher. However, we do not expect that extending this study beyond 6 hours would increase synthesis rates enough to change our conclusions. This study had several limitations. Firstly, with respect to the gene expression data, we analyzed different sub-regions of the brain than others. For example, many studies analyzing expression of genes involved in the ARA cascade focused on the frontal cortex whereas our study only investigated gene expression in the cortex, which could partly explain why we were unable to reproduce previously published findings (185, 186). This study was also limited by the 15-week feeding period. As all experiments were conducted on the rats at 15 weeks post-weaning, we cannot draw conclusions about outcomes during the growth and development phase of the rat s life, when brain accretion peaks. It is possible that differences in brain DHA concentrations were more pronounced during adolescence and started to equilibrate in adulthood. In fact, it has been reported that infants who were breastfed had higher brain DHA concentrations those fed formula, therefore, the results from our study do not apply to infants (192). Also, heparin was used as an anti-coagulant for our infusion studies. Heparin is known to activate lipoprotein

98 84 lipase and may have contributed to the large variability in plasma unesterified fatty acid concentrations (219, 220). However, the effect of heparin is likely an overestimation of the rate DHA uptake into the brain. Another limitation to this study was the use of pure fatty acid ethyl esters. Pure oil sources were used so that the effect of n-3 PUFA could be compared directly (ALA vs. DHA). However, the use of pure oil sources limited the applicability of the study to free living situations because n-3 PUFA are consumed from food or oil sources, not as pure fatty acids, and most diets that contain DHA also contain ALA. This study showed that despite large differences in fatty acid accumulation in the body, rats fed a diet containing DHA or ALA making up 2% of the fatty acids did not have differences in brain DHA accumulation. Using in vivo kinetic approaches we were able to determine that animals consuming the ALA diet synthesized DHA at rates that exceed the rate of DHA uptake from the plasma into the brain. Importantly, rats consuming the ALA diet had a lower uptake rate of DHA into the brain than rats consuming DHA. As the uptake rate of DHA into the brain has been shown to match rates of brain DHA metabolism (184), it is likely that decreased brain DHA metabolism, in combination with an increased rate of DHA synthesis from ALA is the reason that brain DHA accretion in rats fed the ALA diet did not differ from the rats fed the DHA diet. The overall results from this study indicate that DHA synthesis from ALA in the rat may be sufficient to maintain brain DHA concentrations in the absence of dietary DHA consumption. Importantly, the steady-state infusion method can be used in humans to calculate an actual DHA synthesis rate that can be compared to brain DHA uptake rates measured in humans with positron emission tomography scanning (82).

99 85 Chapter 5 A dose response study on the effect of linoleic acid on the rate of whole body synthesis of docosahexaenoic acid from α-linolenic acid in free-living rats. Anthony F. Domenichiello, Alex Kitson, Chuck T. Chen, Marc-Olivier Trepanier, P. Mark Stavro, Richard P Bazinet Contribution: As first author I performed all of the experimental procedures, all of the data analysis and all of the manuscript drafting.

100 86 5 Objective 3: To determine if high n-6 PUFA affect DHA synthesis 5.1 Abstract Docosahexaenoic acid (DHA) is thought to be important for the brain function. The main dietary source of DHA is fish, however, DHA can also be synthesized from precursor n-3 PUFA, the most abundantly consumed being α-linolenic acid (ALA). The enzymes required to synthesize DHA from ALA are also used to synthesize long chain omega-6 polyunsaturated fatty acids (n-6 PUFA) from linoleic acid (LNA). The large increase in LNA consumption that has occurred over the last century has led to concern that LNA and other n-6 PUFA outcompete n-3 PUFA for enzymes involved in DHA synthesis, and therefore, decrease overall DHA synthesis. To assess this, rats were fed diets containing LNA at 53 (high LNA diet), 11 (medium LNA diet) or 1.5% (low LNA diet) of the fatty acids with ALA being constant across all diets (about 4% of fatty acids). Rats were maintained on these diets from weaning for 8 weeks, at which point they were subjected to a steady-state infusion of labeled ALA and LNA to measure DHA and ARA synthesis rates. DHA and ARA synthesis rates were highest in rats fed the medium and high LNA diets, while the plasma half-life of DHA was longer in rats fed the low LNA diet. Therefore, increasing dietary LNA, in rats, did not impair DHA synthesis, however, low dietary LNA led to a decrease in DHA synthesis with plasma concentrations of DHA being maintained by a longer DHA half-life.

101 Introduction Docosahexaenoic acid (DHA) is the main omega 3 polyunsaturated fatty acid (n-3 PUFA) in the brain, comprising approximately 10% of brain fatty acids (9). DHA is thought to be important for brain function, and relatively lower brain DHA levels have been reported in post-mortem brain samples of patients with various neurological disorders (55-57, 61-63, ). DHA cannot be synthesized de novo it must either be consumed directly from diet or synthesized from precursor n-3 PUFA. The main dietary source of DHA is fatty fish, which makes increasing DHA consumption especially problematic in light of the worlds collapsing fish stocks (23). Therefore, knowing the extent to which DHA can be synthesized from precursor n-3 PUFA is critical for formulating dietary guidelines. Previous work performed in adult rats, from our laboratory found that a diet containing α-linolenic acid (ALA), the main precursor to DHA in human diets, was sufficient to maintain brain DHA concentrations (85). Moreover, rats consuming this ALA diet synthesized DHA at a rate that was at least 3- fold higher than their brain DHA uptake and accretion rates; indicating that DHA synthesis from ALA was sufficient to maintain brain DHA in these rats (85). DHA is synthesized from ALA in the liver by a series of desaturations, elongations and a β-oxidation (13, 104). Dietary DHA has been shown to downregulate expression of enzymes involved in its synthesis (100). Also, the desaturase and elongase enzymes are reactive to both n-3 and omega 6 (n-6) PUFA, which may result in competition between the 2 classes of PUFA for these enzymes (13, 97, 98, 104). Linoleic acid (LNA) is the main n-6 PUFA in the diet and is considered the n-6 PUFA equivalent

102 88 to ALA. LNA uses the same desaturase and elongase enzymes as ALA to be converted to arachidonic acid (ARA), the main n-6 PUFA in the brain, and docosapentaenoic acid (DPA n-6). Dietary intakes of LNA in the United States have increased dramatically over the past century, due to a 1000-fold increase in soybean oil consumption (46). Such an increase in LNA consumption has led to concern that LNA is outcompeting ALA for desaturation causing a decrease in DHA synthesis. It was recently shown that lowering dietary LNA in human clinical trials resulted in increases in plasma phospholipid and cholesteryl ester DHA concentrations (93, 224). While this would indicate that DHA synthesis was higher in subjects with lower LNA diets, it is also possible that higher phospholipid and cholesteryl ester DHA in the plasma reflect increased secretion from the liver due to less competition with n-6 PUFA for esterification into lipoproteins or decreased plasma half-life. Also, once ALA enters the liver it must be converted to ALA- CoA in order to be synthesized to DHA, therefore, in order for competition between ALA and LNA to have an effect on DHA synthesis the ALA-CoA to LNA-CoA ratio, in the liver, must be altered. Whether or not dietary LNA can alter the ALA to LNA ratio in the hepatic CoA pool is unknown. It is plausible that the CoA pool, being about 500-fold lower than the esterified lipid pools (225), is tightly regulated and maintained, even with extreme deprivation, by deacylation of the esterified lipids. If diet does not alter the CoA pool, it is possible that despite the high LNA to ALA ratios seen in plasma with high LNA diets, there will be no effect of diet on DHA synthesis. Th3 goal of this study was to feed rats diets with different amounts of LNA and observe the effect of these diets on DHA and ARA synthesis measured in vivo using a modified version of the steady-state infusion method designed by Rapoport et al (181).

103 Methods Animals All procedures were performed in accordance with the policies set out by the Canadian Council on Animal Care and were approved by the Animal Ethics Committee at the University of Toronto. Three long evans dams were ordered from Charles River Laboratories (Saint Constant, QC, Canada) each with, 18-day-old, male, long evans pups. Dams and pups were maintained on standard chow, which contained no long chain n-3 PUFA (Chapter 4 Table 1), for 3 days. When pups were 21 days old they were weaned and allocated to one of three test diets containing high, medium or low LNA. After the rats consumed these diets for 8 weeks, jugular vein and carotid artery cannulae were surgically implanted in these rats. Three days after surgery, rats were infused via the jugular vein with 2 H 5 -ALA and U- 13 C-LNA for 3 hours to measure the synthesis rates of longer chain PUFA (Supplementary Figure 2) Diets The diets were based off the AIN93 custom low n-6 diets (Dyets Inc, Bethlehem, PA, USA). These diets were modified such that 50% of the fatty acids was added ethyl ester oils. The ethyl ester oils were used to control the LNA content of these diets without altering the fatty acid content of the whole diet, therefore, these diets differed only in their oleate and LNA content. The high LNA diet contained 50% of the fatty acids as LNA ethyl ester (Nu-Chek Prep, Elysian, MN, USA). The medium LNA diet contained 40% of the fatty acids as oleate ethyl ester (Nu-Chek Prep) and 10% of the fatty acids as

104 90 LNA ethyl ester. The low LNA diet contained about 50% of the fatty acids as oleate ethyl ester with no added LNA ethyl ester. Full diet fatty acid composition is found in Table 1.

105 91 Table 1: Percent composition of the low, medium and high LNA diets as measured by GC-FID Low LNA Medium LNA High LNA 10: : : : : :1n :2n :3n Data shown are means (n=3) expressed as percent of total fatty acids.

106 Surgery After the rats consumed their respective diets for 8 weeks they were subjected to surgery to implant jugular vein and carotid artery catheters. The animals were anesthetized using isoflurane inhalation (5% induction, 1-3% maintenance). Before the incision was made, hair was shaved from the incision site and the site was sterilized with iodine and ethanol. A transverse incision was made anterior to the upper thorax. The jugular vein was located by blunt dissection. The vein was isolated and tied off. The vein was then nicked and a catheter (PE 50, Intramedic, Sparks, MD, USA) with a 3.5 cm silastic tubing end (VWR, Mississauga, ON, Canada) was inserted into the vein. The catheter was secured using 3.0 silk suture. Next, the carotid artery was located and isolated by blunt dissection. The vessel was tied off and nicked, and a catheter was inserted 3.5 cm into the artery. The catheter was tied to the artery using silk suture. A 16 gauge angiocath (Becton Dickinson, Mississauga, ON, Canada) was used to guide the catheters subcutaneously to a site outside the body near the scapula. An incision was made at this site and the catheters were fed through a skin button (Braintree Scientific, Braintree, MA USA), which was then sutured to the subcutaneous tissue at incision site. The incision site was then sutured shut and the catheters were taped to the skin button for protection. The rats were allowed to recover from the anesthetic under a heat lamp. One day after surgery, blood was drawn from the carotid artery cannula (baseline blood sample) and both the jugular vein and carotid artery catheters were locked with heparinized saline (5% heparin by volume, Sandoz Canada Inc. Canada). The blood was centrifuged and plasma was isolated and stored at -80 C until analysis.

107 H 5 -ALA and U- 13 C 18 -LNA Infusion U- 13 C 18 -LNA ethyl ester (Cambridge Isotope Laboratories, Andover, MA, USA) was converted to free fatty acid by hydrolysis with 10% KOH/Methanol at 70 C for 1 hour with hexane used to extract the free acid. Modified from Domenichiello et al (85) infusate was made by dissolving 4.5 μmol/100g body weight of U- 13 C 18 -LNA (free acid) and 2 H 5 -ALA free acid (Cambridge Isotope Laboratories) into 5 mm HEPES buffer (ph 7.4) containing 50 mg/ml fatty acid-free bovine serum albumin. The infusate was mixed by sonication at 37 C. An infusion pump (Harvard Apparatus PHD 2000; Holliston, MA, USA) was used to infuse 3.78 ml of tracer solution at a constant rate of ml/min for 3 hours into the jugular vein. Immediately before (time = 0 min), and every 30 min during the infusion 0.2 ml of blood was drawn from the carotid artery and the catheter was flushed with heparinized saline (5% by volume) between blood draws. After 180 min of the infusion, 1 ml of blood was drawn from the carotid artery and the animals were euthanized with a lethal injection of T-61 into the jugular vein and livers were collected and flash frozen in liquid N 2. All blood samples were centrifuged for 10 min (PC-100 microcentrifuge, Diamed, ON, Canada) and the plasma was collected and stored at -80 C as previously described (85) Plasma Volume Determination Plasma volume was determined using previously described methods with modifications (85). A known amount of Evans Blue dye was injected into the jugular veins of the rats (n=6). After 15 min, 2 blood samples of 1 ml were drawn from the carotid artery. Plasma was collected as described above, and 0.1 ml of plasma was diluted into 1 ml of saline. Absorbance at 604 nm was measured with an imark microplate

108 94 absorbance reader (bio-rad laboratories, Hercules, CA, USA) and by comparing the absorbance to a standard curve, the concentration of the dye was determined Plasma Lipid Extraction Lipids were extracted from 50 μl of plasma using the method of Folch et al (200). For baseline plasma samples a known amount of heptadecanoic acid (17:0 - Nu- Check Prep. Elysian, MN) and di-17:0 phosphatidylcholine (Nu-Check Prep.) internal standard were added. Next, chloroform, methanol, and 0.88% KCl (4:2:1.75 v/v/v) were added to plasma. The mixture was centrifuged and the chloroform phase was extracted twice. Total lipid extract (TLE) was stored under nitrogen gas at -80 C Separation of esterified and unesterified lipids by Thin layer Chromatography (TLC) TLE from plasma samples were loaded onto TLC plates and run in heptane/diethylether/glacial acetic acid (60:40:2 v/v/v) to separate esterified and unesterified lipids, as described previously (85). Baseline samples were then prepared for transmethylation (described below) by adding hexane to the esterified and unesterified fractions and storing under nitrogen at -80 C. Infusion samples (time min) were prepared for LC-MS/MS (described below) by adding a known amount of internal standard ( 2 H 8 -ARA, Cayman Chemical, Ann Arbor, MI, USA) to the esterified and unesterified fractions then adding 4 ml Hexane/Isopropanol (HIP, 3:2 v/v) with 5.5% dh 2 O, and storing under nitrogen at 5 C.

109 Transmethylation and gas chromatography-flame ionization detection Baseline samples were transmethylated using 14% boron trifluoride in methanol and fatty acid methyl esters (FAME) were extracted with hexane and quantified using GC-FID as previously described (85). FAME were analyzed using a Varian-430 gas chromatograph (Varian, Lake Forest, CA, USA) equipped with a Supelco capillary column (SP-2560; 100 m x 0.25 mm i.d. x 0.20 μm film thickness) and a FID. Samples were injected in splitless mode. The injector and detector ports were set at 250 C. FAME were eluted using a temperature program set initially at 60 C for 2 min, increasing at 10 C/min to 170 C, and held at 170 C for 4 min, then increasing at 6.5 C/min to 175 C, 2.6 C/min to 185 C and 1.3 C/min to 190 C, increasing at 8 C and held at 240 C for 11 min to complete the run at min. The carrier gas was helium, set to a constant flow rate of 3 ml/min. Peaks were identified by retention times of authentic FAME standards (Nu-Chek Prep, Inc., Elysian, MN, USA). The concentration of each fatty acid was calculated by comparison with the internal standard (17:0) (201). The concentrations were expressed as nmol/ml of plasma Fatty Acyl-CoA Extraction Acyl-CoA were extracted from approximately 0.7 g of liver following the method of Chen et al (8). Briefly, livers were homogenized, on ice, by sonication in 25mM KH 2 PO 4 and isopropanol. Saturated (NH 4 ) 2 SO 4 was used to precipitate protein from the homogenate and acyl-coa was extracted and diluted. Diluted Acyl-CoA extract was purified using an oligodendrocyte purification cartridge (ABI Masterpiece, OPC ; Applied Biosystems, Foster City, CA) and acyl-coa concentrations were measured using LC-MS/MS as described below.

110 Quantification of fatty acid concentrations by Liquid Chromatography- Tandem Mass Spectrometry (LC-MS/MS) After overnight storage, lipids were extracted from silica by centrifuging the samples at 500 g for 10 min, and transferring the HIP phase to a new test tube (2 times). Samples were evaporated under nitrogen gas and unesterified samples were reconstituted in 2 ml of hexane for analysis by LC-MS/MS. Esterified samples were hydrolyzed with 1 ml of 10% KOH in methanol, as previously described (85). The hydrolyzed esterified fatty acid samples were extracted twice with hexane, evaporated under nitrogen and reconstituted in 2 ml of hexane for analysis by LC-MS/MS. Half of each plasma and liver sample were evaporated under N 2 gas and reconstituted in 100 μl of water/acetonitrile (80:20 v/v). Fatty acids were detected using an Agilent HPLC 1290 (Agilent Technologies, Santa Clara, CA, U.S.A) equipped with an Agilent Zorbax SB-Phenyl column (3 x 50 mm, 3.5μm; Agilent, Santa Clara, CA, U.S.A). The initial HPLC conditions of elution were set at 500 μl/min gradient system consisting of (A) 50% water and (B) 50% acetonitrile. The gradient started with 50% (A) and 50% (B) and maintained for 1.5 minutes, increased to 100% (B) from 1.5 to 6 minutes and maintained at 100% B for 4 minutes to complete the total run of 10 minutes. Mass spectrometry analyses were carried out on QTRAP 5500 triple quadruple mass spectrometer (AB SCIEX, Framingham, MA, U.S.A) in electrospray ionization, negative ion mode. The source temperature was 600 C and the ion spray voltage was ev. The optimized parameters were as follows: de-clustering potential -40, entrance potential -10, collision energy -20, and collision cell exit potential -11. Mass transitions for 2 H 5 - ALA, 2 H 5 -eicosapentaenoic acid (EPA), 2 H 5 -docosapentaenoic acid n-3 (DPAn-3) and

111 97 2 H 5 -DHA were: to 59.0, to 262.2, to m/z and to 288.2, respectively. Concentration was quantified by comparing the peak area ratios (peak of interest:internal standard) and correcting for a response factor that was determined for each fatty acid of interest. Response factors were determined by analyzing a standard mixture of 100ng/mL each of DHA, EPA, ALA, DPA, and 2 H 8 -AA (arachidonic acid) by LC/MS/MS and comparing peak areas for each of the 4 fatty acids in relation to the peak area for 2 H 8 -AA to generate response factors. The response factors were 10, 0.75, 0.25, 0.75 for ALA, EPA, DPAn-3 and DHA, respectively Kinetic Equations To determine the DHA and ARA synthesis rates, appearance of 2 H-DHA (or 13 C 18 -ARA) in the plasma-esterified pool was measured and fit to a Boltzmann sigmoidal curve ([labeled-long chain PUFA]*plasma volume vs. time) using non-linear curve (Graphpad Prism Version 4.0, La Jolla, CA, USA) (85). At any point on this curve the slope (S) will be determined by the ability of the body to synthesize labeled-long chain PUFA from their respective labeled-precursor and the ability of the periphery to uptake the labeled-long chain PUFA (Eq. 1). S k 1,DHA [ 2 H ALA] unesterified k 1,DHA [ 2 H DHA] esterified (Eq. 1) S k 1,ARA [U 13 C 18 LNA] unesterified k 1,ARA [ 13 C 18 ARA] esterified (Eq. 2) Where k 1,DHA and k 1,ARA are the steady-state synthesis-secretion coefficients for DHA and ARA respectively, [ 2 H-ALA] unesterified and [U- 13 C 18 -LNA] unesterified are the plasma concentrations of the infusates, k -1,DHA and k -1,ARA are the disappearance coefficients for DHA and ARA respectively and [ 2 H-DHA] esterified and [ 13 C 18 -ARA] esterified are the

112 98 concentrations of DHA and ARA in the plasma that has been synthesized from the infusate, packaged into a lipoprotein and secreted into the plasma. The maximum first derivative (S max ) of the curves described by Eq.1 and 2 is assumed to be the time point when the uptake of esterified DHA from the periphery is negligible, i.e. 0 (for example Eq. 3). S max k 1,DHA [ 2 H ALA] unesterified 0 (Eq. 3) Therefore, the derivative at this point is equal to the rate of 2 H-DHA synthesis. By correcting the S max by the tracee:tracer ratio, the rats actual DHA synthesis is determined, J syn,dha (nmol/min) (85). J syn,dha S max [ALA] unesterified [ 2 H ALA] unesterified k 1,DHA [ALA] unesterified (Eq. 4) Since the study diets did not contain long chain PUFA (EPA, ARA or DHA) and esterified plasma lipids were constant throughout the study, the turnover rate (F i ) and half-life (t 1/2 ) of esterified lipids in the plasma can be determined by Eq 5 and 6 (shown below) using DHA as an example: F DHA and J syn,dha [DHA] esterified V plasma (Eq. 5) t 1/2,DHA F DHA (Eq. 6) Statistics All data are presented as mean SEM and were compared by one-way ANOVA using Tukey s test for multiple comparison if ANOVA p-value 0.05 (GraphPad Prism version 4.0, La Jolla, CA, USA). When data was not normal or sample sizes were too low

113 99 to test normality of the data, determined by Kolmogrov-Smirnov normality test, then data was log transformed and then compared by one-way ANOVA using Tukey s test for multuple comparisons. A p-value 0.05 with Tukey s multiple comparison test was considered significant. 5.4 Results Plasma Volume Plasma volume was ml/g of body weight, similar to previous estimates of long evans rat plasma volume (85) Baseline Plasma Unesterified Lipid Concentrations Plasma unesterified ALA concentrations (Figure 1a) were not significantly different between rats fed the different diets ( , , nmol/ml for the low, medium and high LNA diet, respectively). Plasma unesterified LNA concentrations (Figure 1b) were highest in rats fed the high LNA diet but did not differ between rats fed the low and medium LNA diet ( , , nmol/ml p<0.05, low, medium, and high LNA diet, respectively) Baseline Plasma Esterified Lipid Concentrations Plasma esterified ALA concentrations (Figure 2a) did not differ in rats fed different diets, however, plasma esterified EPA (Figure 2b) was significantly higher in rats fed the low LNA diet compared to medium LNA diet and rats fed the medium diet had higher plasma esterified EPA concentrations than rats fed the high LNA diets ( > > nmol/ml, p<0.05). Plasma esterified DHA (Figure 2c) concentrations were not significantly different between rats fed either diet ( ,

114 nmol/ml, nmol/ml low, medium and high LNA diet respectively). Plasma esterified LNA and ARA concentrations (Figure 2d and e) were highest in rats fed the high LNA diet. Plasma esterified ARA did not differ between rats fed the medium and low LNA diets ( > = nmol/ml, p<0.05), however, plasma esterified LNA was significantly higher in rats fed the medium diet compared to rats fed the low LNA diet ( > > nmol/ml, p<0.05).

115 101 Figure 1: Plasma unesterified (a) ALA and (b) LNA concentrations (nmol/ml) in rats consuming the Low, Medium or High LNA diet. Plasma unesterified ALA concentrations did not differ between rats fed the 3 diets, while plasma unesterified LNA concentrations were highest in rats fed the high LNA diet, but did not differ between rats fed the low or medium LNA diet (High > medium = low LNA diet p<0.05). All data are mean ± SEM. Different letters signify the means are significantly different (p<0.05) measured by One-Way ANOVA followed by Tukey s multiple comparison test, data that was not normally distributed was log transformed. n=7,7,4 (low, medium, high LNA diet respectively). a) b)

116 102 Figure 2: Plasma esterified concentrations for n-3 and n-6 PUFA. (a) Plasma esterified ALA concentrations did not differ between rats fed the 3 diets. (b) Plasma esterified EPA concentrations were significantly higher in rats fed the low LNA diet than rats fed the medium and high LNA diet (p<0.05). (c) Plasma esterified DHA concentrations were not significantly different in rats fed the three diets (p>0.05). Rats fed the high LNA diet had the highest concentrations of plasma esterified (d) LNA followed by rats consuming the medium and low LNA diet respectively (high>medium>low p<0.05). Plasma esterified (e) ARA concentrations were lowest in rats consuming the low LNA diet but there were no differences in plasma esterified ARA between rats fed the high and medium LNA diets. All data are mean ± SEM. Different letters signify the means are significantly different (p<0.05) measured by One-Way ANOVA followed by Tukey s multiple comparison test, data that was not normally distributed was log transformed. n=7,7,4 (low, medium, high LNA diet respectively). a) b) c) d) e)

117 Plasma Unesterified 2 H 5 -ALA and 13 C 18 -LNA Concentrations Mean concentrations of plasma unesterified 2 H 5 -ALA throughout the infusion were , and nmol/ml (low, medium and high LNA diet, respectively) with the rats consuming the high LNA diet having a significantly higher 2 H 5 -ALA concentration than those consuming the low LNA diet (p<0.05). Plasma unesterified 13 C 18 -LNA concentrations were , , nmol/ml (low, medium and high LNA diet, respectively) n-3 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life Sample infusion curves for one representative animal with respective chromatograms for DHA, EPA and ARA are shown in figure 3 and table 2 shows the synthesis parameters for EPA and DHA. For DHA, rats consuming the high LNA diet had a significantly higher synthesis-secretion rate than rats consuming the low LNA diet (66.7 ± 20.3 < ± 91.3 nmol/day, p<0.05), while the synthesis-secretion rate in rats fed the medium LNA diet did not differ from rats consuming either diet (320.5 ± nmol/day). K 1,DHA, the synthesis-secretion coefficient for DHA, was higher in rats consuming the medium LNA diet compared to rats consuming the low LNA diet, but not different from rats consuming the high LNA diet ( , , ml/min, low, medium and high LNA diet, respectively). DHA turnover was significantly slower in rats consuming the low LNA diet compared to rats consuming the medium and high LNA diet (0.03 ± 0.02, 0.17 ± 0.09, 0.17 ± 0.06 /day, low, medium, high LNA diets, respectively, p>0.05) and half-life of DHA was significantly longer in rats consuming the low LNA diet compared to rats consuming the medium and high LNA diet (65.0 ± 15.5, 12.4 ± 4.0, 6.96 ± 3.30 days, low, medium and high LNA diet, respectively).

118 104 Figure 3: Example infusion curves and chromatograms (inserts) for a rat. (a) Plasma volume (ml) x concentration of esterified 2 H- EPA (nmol) plotted against time (min) and fit to a sigmoidal curve. (b) Plasma volume (ml) x concentration of esterified 2 H-DHA (nmol) plotted against time (min) and fit to a sigmoidal curve. (c) Plasma volume (ml) x concentration of esterified 13 C-ARA (nmol) plotted against time (min) and fit to a sigmoidal curve. This figure shows the resulting infusion curve for one rat from that was infused with unesterified [ 2 H 5 ]-ALA and [U- 13 C]-LNA for 3 hours. Inserts show chromatograms for the labeled fatty acid measured for each point on the infusion curve. The chromatograms are overlaid in the insert to illustrate the increase in the amount of the labeled fatty acids throughout the infusion. Arrows indicate the time-point on the infusion curve that corresponds to each peak. a) b) c)

119 105 Table 2: Kinetic parameters for synthesis-secretion of DHA, EPA and ARA following 3 hour steady-state infusion of [ 2 H 5 ]-ALA and [U- 13 C]-LNA in rats fed a low, medium or high LNA diet. Fatty Acid, i Diet S max,i (nmol/min) K 1,i (ml/min) Daily Synthesis Rate (nmol/day) F i (/day) t 1/2, i (day) Low (n = 7) a a a a a DHA Medium (n = 7) b b ab b b High (n = 4) b ab b b b Low (n = 7) a a EPA Medium (n = 7) b b High (n = 4) b b Low (n = 6) a a a a a ARA Medium (n = 5) b b b ab ab High (n = 3) ab ab b b b Data are mean SEM. S max, i = maximum first derivative, K 1,i = synthesis-secretion coefficient for PUFA i, F i = Turnover rate of esterified plasma PUFA i, t 1/2,i = half-life of esterified plasma PUFA i. Differences between diets were measured by oneway ANOVA followed by Tukey s multiple comparison test. Data that was not normally distributed was log-transformed and then analyzed by one-way ANOVA followed by Tukey s multiple comparison test. Different letters signify that means are significantly different (p<0.05).

120 106 The daily synthesis rate for EPA was not significantly different between rats fed the different diets (472.6 ± 186.3, ± 423.2, ± nmol/day for rats fed the low, medium and high LNA diet respectively, p>0.05). Turnover of EPA was significantly slower in rats consuming the low LNA diet compared to rats consuming the other two diets (0.13 ± 0.05, 1.2 ± 0.4, 6.2 ± 2.7 /day, low, medium, high LNA diet, respectively) and the half-life of EPA was longer for rats consuming the low LNA diet compared to those consuming the low LNA diet (19.8 ± 8.6, 4.3 ± 2.4, 0.2 ± 0.09 days, low, medium, high LNA diet, respectively) n-6 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life 13 C 18 -ARA was the only n-6 PUFA that was synthesized from 13 C 18 -LNA and detected in the plasma by LC-MS/MS. Therefore, ARA was the only n-6 PUFA for which synthesis parameters could be calculated (Table 2). 13 C 18 -ARA was not detected in 3 rats (1 per dietary group). Daily ARA synthesis rates were calculated to be higher in rats fed the high and medium LNA diet compared to those fed the low LNA diet ( ± , and ± 20.3 nmol/day, low, medium and high LNA diet, respectively, p<0.05). Daily ARA synthesis rates were not different in rats fed the medium LNA diet compared to either the high LNA diet. Rats fed the high LNA diet had significantly faster ARA turnover ( > /day, p<0.05) and shorter ARA half-life ( < days, p<0.05) compared to rats fed the low LNA diet. ARA turnover and half-life in rats fed the medium LNA diet did not differ from rats fed the other 2 diets ( /day and days for turnover and half-life, respectively).

121 Liver fatty acyl-coa concentrations Acyl-CoA concentrations were measured for unlabeled ALA-CoA, EPA-CoA, DHA-CoA, LNA-CoA and ARA-CoA (Table 3). ALA-CoA concentrations in the livers were not significantly different between dietary groups ( , , nmol/g liver, low, medium, high LNA diets, respectively). EPA-CoA concentrations were higher in rats fed the low LNA diet compared to the high LNA diet but rats fed the medium LNA diet had EPA-CoA concentrations that were not different from the other dietary groups ( , , nmol/g liver low, medium, high LNA diet, respectively). DHA-CoA concentrations ( , , nmol/g liver low, medium, high LNA diet, respectively) and ARA-CoA concentrations ( , , nmol/g liver, low, medium, high LNA diet, respectively) did not differ between dietary groups. LNA-CoA concentrations were highest in rats fed the high LNA diet compared to those fed the low and medium LNA diets ( , , nmol/g liver, low, medium, high LNA diet, respectively).

122 108 Table 3: Liver fatty acid CoA concentrations (nmol/g liver) for rats fed the low, medium or high LNA Diet. Diet [ALA-CoA] (nmol/g) [EPA-CoA] (nmol/g) [DHA-CoA] (nmol/g) [LNA-CoA] (nmol/g) [ARA-CoA] (nmol/g) Low (n=5) 3.9 ± ± 49.0 a 49.3 ± ± 4.1 a 92.3 ± 36.7 Medium (n=6) 4.4 ± ± 21.7 ab 29.2 ± ± 5.9 a 94.9 ± 15.4 High (n=3) 4.1 ± ± 1.5 b 9.0 ± ± 35.9 b 95.1 ± 17.7 Data are mean SEM. Differences between diets were measured by one-way ANOVA followed by Tukey s multiple comparison test or Kruskal-Wallis test followed by Dunn s multiple comparison test (if data was not normally distributed). Different letters signify that means are significantly different (p<0.05).

123 Discussion We found that rats fed a diet low in LNA had lower plasma concentrations of LNA and ARA and higher plasma concentrations of EPA compared to rats fed diets with higher levels of LNA. This is consistent with findings of others comparing rats fed diets with high and low amounts of n-6 PUFA (226, 227). Previous work has also shown that feeding diets with low levels of n-6 PUFA will result in higher concentrations of DHA in the plasma (226) and rat brain (227) suggesting that consuming high levels of LNA impair DHA brain accretion, potentially leading to impaired brain function (207, 226). We failed to detect a difference in plasma DHA concentrations, however, there was a trend towards increased plasma DHA in rats consuming the low LNA diet (p=0.07, t-test low vs. medium LNA diet). n-6 PUFA compete with n-3 PUFA for enzymes involved in the synthesis of long chain PUFA (104), therefore, a logical explanation for the decrease in plasma long chain n-3 PUFA concentrations in rats fed high levels of LNA (observed in this report as well as by others (226)) is decreased DHA synthesis from ALA. Alternatively, a decrease in plasma DHA can also be explained by an increased turnover of DHA in the plasma, without an increase in the DHA synthesis rate. We performed a steady-state infusion of stable-isotope labeled ALA and LNA to measure the EPA, DHA and ARA synthesis rates in rats fed diets containing different levels of LNA. We found that rats fed a diet with low dietary LNA had significantly lower EPA and DHA synthesis-secretion rates as well as lower K 1,DHA compared to rats fed higher levels of LNA. Using the calculated synthesissecretion rates and the plasma concentrations of the different PUFA, the plasma half-life of longer chain PUFA from their dietary precursor (LNA or ALA) can be measured. We

124 110 calculated that rats fed a low level of LNA in the diet had longer half-lives for EPA and DHA in the plasma than rats fed a diet containing high levels of LNA. Therefore, the higher concentration of EPA (and trend towards higher DHA) observed in the plasma of rats fed a low LNA containing diet are due to lower turnover and longer half-lives of these PUFA in the plasma and not due to higher synthesis rates of these PUFA. However, it is important to note that the measure of PUFA turnover and half-life is not a direct measurement, but rather, is an inferred estimate of the how quickly a PUFA is replaced by PUFA that is newly synthesized from precursors (ALA and LNA). Since only the precursor PUFA are consumed in the diet, it is assumed that plasma concentrations of longer chain PUFA are maintained by synthesis from ALA or LNA. It is possible that PUFA that are synthesized from ALA are stored in the adipose (or other tissues) and secreted into the plasma at a later time making the measurement of PUFA turnover used in this study an overestimate. Therefore, future studies should be conducted utilizing a more direct method to measure the half-lives of PUFA in the plasma in rats fed diets with different levels of LNA. Other possible explanations for the lack of effect of dietary LNA on DHA and EPA synthesis rates are tracer dilution in the liver and inability of rats fed the low LNA diet to secrete n-3 PUFA from the liver. Though our experimental procedure did not allow for a proper measurement of the labeled acyl-coa pool we were able to measure non-labeled n-3 and n-6 PUFA concentrations in the acyl-coa pool. LNA-CoA was the only fatty acyl-coa that differed between the 3 dietary groups so it is unlikely that there were significant differences in tracer dilution between the 3 dietary groups in this study. Still, future studies that utilize the steady-state infusion method for measuring DHA

125 111 synthesis rates should euthanize rats and collect livers in such a manner as to preserve the labeled fatty acyl-coa concentrations in order to properly measure tracer dilution in the liver. Moreover, it is possible that rats fed the low LNA diet have high n-3 PUFA synthesis but impaired secretion of these products from the liver. We assessed this hypothesis by measuring the concentrations of labeled n-3 PUFA and n-6 PUFA in the livers of these rats. We also measured ARA and DPAn-6 synthesis rates using the steady-state infusion method (table 2). We were unable to detect labeled DPAn-6 in the plasma, and therefore, could not determine a synthesis rate. Rats fed the low LNA diet had lower ARA synthesis-secretion rates than rats fed the diets containing higher levels of LNA, as would be expected by the higher plasma ARA concentrations in the plasma in these rats. We found that were no differences in K 1,ARA in rats fed the high and low LNA diet, however, K 1,ARA was higher in rats fed the medium ARA diet compared to the low, but not the high LNA diet. Therefore, the similar plasma ARA concentrations found in rats consuming the medium and high LNA diet were maintained by two different mechanisms, increased substrate (LNA) availability in the high LNA group and increased synthesis capacity (K 1,ARA ) in the medium LNA group. Moreover, the low LNA diet group did not seem to adapt to a lower substrate availability as K 1,ARA was lowest in rats consuming this diet (compared the rats consuming the medium LNA diet). Throughout the 20 th century consumption of LNA has increased almost 3-fold in US diets leading to an increase in the dietary LNA to ALA ratio (46). An increased ratio of LNA to ALA can possibly lead to decreased levels of DHA being synthesized because of increased competition for the desaturase and elongase enzymes involved in DHA

126 112 synthesis. In our study, rats were fed diets with LNA levels ranging from 2-50% of the fatty acids and found higher dietary LNA levels were associated with higher EPA and DHA synthesis rates. Therefore, this work indicates that high levels of LNA in current US diets may not be impairing DHA synthesis. However, this hypothesis should be tested in humans, as it is possible that DHA synthesis rates in humans are affected differently than rats to high levels of dietary LNA. In conclusion, we found that rats fed diets containing high levels of LNA had increased plasma ARA and decreased plasma EPA concentrations compared to rats fed diets with low levels of LNA. Plasma ARA concentrations were elevated in rats fed high levels of LNA, likely as a result of higher ARA synthesis rates. However, the augmented plasma n-3 PUFA concentrations appear to be due to slower n-3 PUFA turnover in the plasma rather than increased synthesis from ALA.

127 113 Chapter 6 - A dose response study of dietary α-linolenic acid on the rate of synthesis of docosahexaenoic acid from α-linolenic acid in the free-living rat. Anthony F. Domenichiello, Alex P. Kitson, Adam H. Metherel, Chuck T. Chen, Kathryn Hopperton, P. Mark Stavro, Richard P Bazinet Contribution: As first author I performed all of the experimental procedures, all of the data analysis and all of the manuscript drafting.

128 114 6 Objective 4: To determine how different levels of ALA affect the DHA synthesis rate 6.1 Abstract Docosahexaenoic acid (DHA) is the main omega 3 polyunsaturated fatty acid (n-3 PUFA) in the brain and is thought to be important for brain function. Though the main source of DHA is fish, DHA can also be synthesized from α-linolenic acid (ALA), which is mainly found in plants. While it is generally believed that DHA synthesis is low, there is some evidence that suggests DHA synthesis from ALA is sufficient to maintain brain DHA. We aimed to determine how different levels of dietary ALA would affect the whole body DHA synthesis rate by feeding rats diets containing ALA at 0.07, 3 and 10% of the fatty acids (low, medium and high ALA diet respectively) for 8 weeks LNA was constant across all diets and amounted to approximately 26% of the fatty acids). The steady-state infusion method was then performed to determine the in vivo synthesis rate of DHA from ALA. We found that there was no difference in DHA synthesis between rats fed the medium and high ALA diet. However, rats fed the low ALA diet had a lower DHA synthesis rate compared to the other two diets. Therefore, DHA synthesis was maintained with dietary ALA concentrations from 3-10% of the fatty acids, however, when dietary ALA is too low DHA synthesis cannot be maintained.

129 Introduction The omega-3 polyunsaturated fatty acid (n-3 PUFA) docosahexaenoic acid (DHA) is highly concentrated in the brain (9, 57, 63, 228) where it regulates numerous functions including cell survival and neuroinflammation (24, 26, 27, 73). The brain takes up DHA directly from the plasma, which is maintained directly via the diet or from DHA that has been synthesized in the liver (211). The main precursor to DHA in western diets is α-linolenic acid (ALA) (7, 17, 229, 230). In general, DHA synthesis rates are believed to be low and insufficient to maintain optimal brain DHA, especially during development (31, 35, 115, 231). However, a recent study performed in rodents, using a novel steadystate infusion to measure DHA synthesis, concluded that DHA synthesis from ALA may be sufficient to supply the brain (85, 232). How dietary components affect DHA synthesis is not completely understood. The enzymes used to synthesize DHA from ALA are thought to also be used by the omega-6 (n-6) PUFA linoleic acid (LNA), to synthesize arachidonic acid (ARA) and docosapentaenoic acid n-6 (DPAn-6), therefore, it is possible that competition exists between n-6 and n-3 PUFA for these enzymes. The most competition exists for the Δ6 desaturase enzyme, which is believed to be rate limiting for DHA synthesis, as it is reactive to both 18- and 24-carbon fatty acids of the n-3, n-6 and n-9 family (104, 105). Additionally, it has been shown that dietary DHA can decrease the expression of enzymes involved in its own synthesis (158) and decrease the rate of its synthesis (85). Similar evidence has been found for ALA, with studies showing rats consuming dietary ALA have decreased expression of enzymes involved in the DHA synthesis pathway (100) and decreased DHA synthesis capacity compared to rats fed a diet deficient in n-3

130 116 PUFA (107). Despite dietary ALA decreasing DHA synthesis capacity, net DHA synthesis rates were higher in rats fed ALA compared to those fed a diet deficient in n-3 PUFA due to lack of substrate in the deficient animals (85, 107). In this study, we performed a dietary dose response to determine how different levels of dietary ALA affect the synthesis-secretion rates of ALA and LNA to their long chain PUFA (LCPUFA) products in rats. Using a novel steady-state infusion technique the synthesis-secretion rates of DHA, eicosapentaenoic acid (EPA, an n-3 PUFA also found in fish) and ARA were determined in rats fed diets with different levels of dietary ALA. The steady-state infusion works by infusing labeled ALA and LNA in a manner such that they achieve a constant level in the plasma (85, 181, 232). The appearance of labeled LCPUFA in the plasma is graphed and a sigmoidal curve is fit to the data, which is then integrated to determine the synthesis rate of the LCPUFA. We found that DHA synthesis rates were maintained in rats fed diets with ALA concentrations ranging from 3-10% of the fatty acids but DHA synthesis rates were decreased in rats fed ALA at a concentration of 0.07% of the fatty acids. 6.3 Methods Animals All procedures were performed in accordance with the policies set out by the Canadian Council on Animal Care and were approved by the Animal Ethics Committee at the University of Toronto. Long evans dams (Charles River Laboratories) arrived at our facility with 18-day-old, male, long evans pups. Dams and pups were maintained on standard chow, which contained no long chain n-3 or n-6 PUFA (Chapter 4 Table 1). Supplementary figure 3 illustrates the design of this study. Pups were weaned at 21 days

131 117 of age and allocated to a low, medium or high ALA diet (containing 0.07, 3 or 9.9% of fatty acids as ALA). Rats were maintained on these diets for 8 weeks upon which time they underwent a surgery to implant catheters in the carotid artery and jugular vein. After 3 days recovery from surgery, rats were subjected to an infusion of 2 H 5 -ALA and 13 C 18 - LNA (described below) Diets The diets were based off of the AIN-93G custom low n-3 diets (Dyets Inc, Bethlehem, PA, USA). These diets contained 90% of the fatty acids (by weight) as hydrogenated coconut oil (57.2% by weight) and safflower oil (32.8% by weight) and 10% of the fatty acids (by weight) consisted of added ethyl ester oils which allowed for manipulation of the ALA content of the diet without changing the whole fatty acid composition of the diet. The only fatty acids that differed between the 3 diets used in this study were ALA and oleate (18:1 n-9). The 3 diets for this study were the low, medium and high ALA diets. The low ALA diet contained ethyl ester oleate (Nu-Chek Prep, Elysian, MN, USA) added at a concentration of 10% of the fatty acids (by weight). The medium ALA diet contained ethyl ester oleate added at 7% of the fatty acids and ethyl ester ALA (generously donated by BASF Pharma Callanish Ltd. Isle of Lewis, UK) added at 3% of the fatty acids (by weight). The high ALA diet contained ethyl ester ALA added at 10% of the fatty acids (by weight) and no ethyl ester oleate. The full dietary fatty acid composition as measured by GC-FID is reported in Table 1.

132 118 Table 1. Fatty acid composition of the low, medium and high ALA diet as measured by GC-FID LOW ALA DIET (n = 3) MEDIUM ALA DIET (n = 3) HIGH ALA DIET (n = 3) 10: : : : :1n : :1n :1n :2-9t,12c :2 n-6 (LNA) : :1n :3n-3 (ALA) : Data are means SEM

133 Surgery After the rats consumed their respective diets for 8 weeks they were subjected to surgery to implant catheters in the jugular vein and carotid artery. Animals were anesthetized using isoflurane inhalation (5% induction, 1-3% maintenance). Before the incision was made, hair was shaved from the incision site and the site was sterilized with iodine and ethanol. A transverse incision was made anterior to the upper thorax. The jugular vein was located by blunt dissection. The vein was isolated and tied off. The vein was then nicked and a catheter (PE 50, Intramedic, Sparks, MD, USA) with a 3.5 cm silastic tubing end (VWR, Mississauga, ON, Canada) was inserted into the vein. The catheter was secured using 3.0 silk suture. Next, the carotid artery was located and isolated by blunt dissection. The vessel was tied off and nicked, and a catheter with a small (less than 0.5 cm) silastic tip was inserted 3.5 cm into the artery. The catheter was tied to the artery using silk suture. Using a 16-gauge angiocath (Becton Dickinson, Mississauga, ON, Canada) the catheters were guided, subcutaneously, to a site outside the body slightly posterior to the scapula and along the midline. An incision was made at this site and the catheters were fed through a skin button (Braintree Scientific, Braintree, MA USA), which was then sutured to the subcutaneous tissue at incision site. The incision site was then sutured shut and the catheters were tapped to the skin button for protection. The rats were allowed to recover from the anesthetic under a heat lamp. One day after surgery blood was drawn from the carotid artery catheter (baseline blood sample) and both the jugular vein and carotid artery catheters were locked with heparinized saline (5% heparin by volume, Sandoz Canada Inc. Canada). The blood was centrifuged and plasma was isolated and stored at -80 C until analysis.

134 H 5 -ALA and 13 C 18 -LNA Infusion U- 13 C 18 -LNA ethyl ester (Cambridge Isotope Laboratories, Andover, MA, USA) was converted to free fatty acid by hydrolysis with 10% KOH/Methanol at 70 C for 1 hour with hexane used to extract the free acid. Modified from Domenichiello et al (85) infusate was made by dissolving 4.5 μmol/100g body weight of U- 13 C 18 -LNA (free acid) and 2 H 5 -ALA free acid (Cambridge Isotope Laboratories) into 5 mm HEPES buffer (ph 7.4) containing 100 mg/ml fatty acid-free bovine serum albumin. The infusate was mixed by sonication at 37 C. Three days after the surgery catheters were connected to a dual channel stainless steel swivel that was connected to a single axis counter balance arm (Instech Laboratories, Plymouth, PA, USA) and placed in a cage. A stainless steel tether (Instech Laboratories) was secured to the swivel and to the rats skin button. This protected the catheters from being tangled or chewed by the rat allowing for the infusion to be performed in a completely free-living environment. Throughout the infusion the rats had access to food and water. An infusion pump (Harvard Apparatus PHD 2000; Holliston, MA, USA) was used to infuse 3.78 ml of tracer solution at a constant rate of ml/min for 3 hours into the jugular vein via the swivel. Immediately before (time = 0 min), and every 30 min during the infusion 0.2 ml of blood was drawn from the carotid artery and the catheter was flushed with heparinized saline (5% by volume) between blood draws. After 180 min of the infusion, 1 ml of blood was drawn from the carotid artery and the animals were euthanized with a lethal injection of T-61 into the carotid artery, while tracer was still being infused into the jugular vein. Livers were collected immediately and flash frozen in

135 121 liquid N 2. Animals were euthanized while still being infused in order to ensure an accurate measurement of labeled fatty acyl-coenzyme A (AcylCoA) products in the liver. All blood samples were centrifuged for 10 min (PC-100 microcentrifuge, Diamed, ON, Canada) and the plasma was collected and stored at -80 C as previously described (85) Plasma Lipid Extraction Lipids were extracted from 50 μl of plasma using the method of Folch et al (200). For baseline plasma samples a known amount of heptadecanoic acid (17:0 - Nu- Check Prep. Elysian, MN) and di-17:0 phosphatidylcholine (Nu-Check Prep.) internal standard were added. For infusion samples a known amount of 2 H 8 -ARA (Cayman Chemical, Ann Arbor, MI, USA) was added as internal standard. Next, chloroform, methanol, and 0.88% KCl (4:2:1.75 v/v/v) were added to plasma. The mixture was centrifuged and the chloroform phase was extracted twice. Total lipid extract (TLE) was stored under nitrogen gas at -80 C Separation of esterified and unesterified lipids by Thin layer Chromatography (TLC) TLE from baseline plasma samples were loaded onto TLC plates and run in heptane/diethylether/glacial acetic acid (60:40:2 v/v/v) to separate esterified and unesterified lipids, as described previously (85). Samples were then prepared for transmethylation (described below) by adding hexane to the esterified and unesterified fractions and storing under nitrogen at -80 C.

136 Transmethylation and gas chromatography-flame ionization detection Baseline samples were transmtehylated using 14% boron trifluoride in methanol and fatty acid methyl esters (FAME) were extracted with hexane and quantified using GC-FID as previously described (85). FAME were analyzed using a Varian-430 gas chromatograph (Varian, Lake Forest, CA, USA) equipped with a Supelco capillary column (SP-2560; 100 m x 0.25 mm i.d. x 0.20 μm film thickness) and a FID. Samples were injected in splitless mode. The injector and detector ports were set at 250 C. FAME were eluted using a temperature program set initially at 60 C for 2 min, increasing at 10 C/min to 170 C, and held at 170 C for 4 min, then increasing at 6.5 C/min to 175 C, 2.6 C/min to 185 C and 1.3 C/min to 190 C, increasing at 8 C and held at 240 C for 11 min to complete the run at min. The carrier gas was helium, set to a constant flow rate of 3 ml/min. Peaks were identified by retention times of authentic FAME standards (Nu-Chek Prep, Inc., Elysian, MN, USA). The concentration of each fatty acid was calculated by comparison with the internal standard (17:0) (201). The concentrations were expressed as nmol/ml of plasma Preparation of plasma infusion samples TLE from plasma samples taken during the infusion were dried down and reconstituted in 1ml of hexane. The TLE was then split into two aliquots of 500 μl each. One aliquot of the TLE was analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS) to determine the concentration of the infusate (unesterified 2 H 5 -ALA and 13 C 18 -LNA) in the plasma throughout the infusion. The other aliquot of TLE was used to determine the concentration of labeled, longer chain n-3 and n-6 PUFA that was synthesized from 2 H 5 -ALA and 13 C 18 -LNA throughout the infusion. As most of

137 123 these products were found in the esterified fraction of the plasma (unpublished observations) this aliquot was hydrolyzed using 10% KOH/Methanol at 70 C for 1 hour. After 1 hour this reaction was neutralized by adding HCl and H 2 O to the samples and the lipids were extracted with hexane twice and analyzed by LC-MS/MS (85) Quantification of fatty acid concentrations by Liquid Chromatography- Tandem Mass Spectrometry (LC-MS/MS) Labeled PUFA in plasma samples were measured by LC-MS/MS. Half of each sample was evaporated under N 2 gas and reconstituted in 100 μl of water/acetonitrile (80:20 v/v). Fatty acids were detected using an Agilent HPLC 1290 (Agilent Technologies, Santa Clara, CA, U.S.A) equipped with an Agilent Zorbax SB-Phenyl column (3 x 50 mm, 3.5μm; Agilent, Santa Clara, CA, U.S.A). The initial HPLC conditions of elution were set at 500 μl/min gradient system consisting of (A) 50% water and (B) 50% acetonitrile. The gradient started with 50% (A) and 50% (B) and maintained for 1.5 minutes, increased to 100% (B) from 1.5 to 6 minutes and maintained at 100% B for 4 minutes to complete the total run of 10 minutes. Mass spectrometry analyses were carried out on QTRAP 5500 triple quadruple mass spectrometer (AB SCIEX, Framingham, MA, U.S.A) in electrospray ionization, negative ion mode. The source temperature was 600 C and the ion spray voltage was ev. The optimized parameters were as follows: de-clustering potential -40, entrance potential -10, collision energy -20, and collision cell exit potential -11. Mass transitions for 2 H 5 -ALA, 2 H 5 - eicosapentaenoic acid (EPA), 2 H 5 -docosapentaenoic acid n-3 (DPAn-3) and 2 H 5 -DHA were: to 59.0, to 262.2, to m/z and to 288.2, respectively. Concentration was quantified by comparing the peak area ratios (peak of interest:internal

138 124 standard) and correcting for a response factor that was determined for each fatty acid of interest. Response factors were determined by analyzing a standard mixture of 100ng/mL each of DHA, EPA, ALA, DPA, and 2 H 8 -AA (arachidonic acid) by LC/MS/MS and comparing peak areas for each of the 4 fatty acids in relation to the peak area for 2 H 8 -AA to generate response factors. The response factors were 10, 0.75, 0.25, 0.75 for ALA, EPA, DPAn-3 and DHA, respectively Steady-State Infusion Kinetics The steady-state infusion method used to measure LCPUFA synthesis was designed by Rapoport et al (181) and requires that labeled ALA and LNA, in the unesterified form, be infused at a steady rate such that the concentration of these tracers achieves steady-state in the plasma. Throughout the infusion labeled LCPUFA, that are synthesized from the tracers being infused, will appear in the plasma esterified lipid pool in a sigmoidal fashion (85). Therefore, when the appearance of labeled n-6 or n-3 LCPUFA are plotted on a graph of Plasma Volume (V plasma ) x [labled-n-6 or n-3 LCPUFA] vs. time the data can be fit to a Boltzman sigmoidal curve using non-linear regression (Graphpad Prism Version 4.0, La Jolla, CA, USA) (85). The first derivative (i.e. slope, S) of any point on this curve is equal to the rate of appearance of a given LCPUFA at that particular time during the infusion and is determined by Eq. 1 and 2 for n-3 and n-6 PUFA, respectively. S k 1,n 3PUFA [ 2 H 5 ALA] unesterified k 1,n 3PUFA [ 2 H 5 n 3LCPUFA] esterified S k 1,n 6PUFA [ 13 C 18 LNA] unesterified k 1,n 6PUFA [ 13 C 18 n 6LCPUFA] esterified (Eq1) (Eq2) Where k 1,n-3LCPUFA is the synthesis-secretion coefficient for a long chain n-3 LCPUFA that is synthesized from ALA, k 1,n-6LCPUFA is the synthesis-secretion coefficient for a long

139 125 chain n-6 LCPUFA that is synthesized from LNA and k -1,n-3LCPUFA and k -1,n-6LCPUFA are the disappearance coefficients for the long chain n-3 and n-6 LCPUFA, respectively. The maximum first derivative (S max ), is assumed to be the point that the disappearance coefficient is equal to 0, or negligible. According to this assumption, S max is determined solely by the synthesis-secretion rate for a given PUFA as described by Eq3 and 4. S max k 1,n 3PUFA [ 2 H 5 ALA] unesterified 0 S max k 1,n 6PUFA [ 13 C 18 LNA] unesterified 0 (Eq3) (Eq4) Therefore, if S max is known the synthesis-secretion coefficient can be determined. The actual synthesis rate of a LCPUFA (J LCPUFA ) is the product of the synthesis-secretion coefficient for that LCPUFA multiplied by the plasma unesterified concentration of its precursor and can be described by Eq5 and 6. J n 3PUFA k 1,n 3LCPUFA [ALA] unesterified S max [ALA] unesterified [ 2 H 5 ALA] unesterified (Eq5) J n 6PUFA k 1,n 6LCPUFA [LNA] unesterified S max[lna] unesterified [ 13 C 18 LNA] unesterified (Eq6) The diets consumed for this study contained no LCPUFA and it can be assumed that the plasma esterified lipids were constant during the infusion period. Therefore, the turnover rate (F LCPUFA ) and half-life (t 1/2,LCPUFA ) of esterified LCPUFA in the plasma can be determined by Eq7 and 8, respectively.

140 126 J LCPUFA F LCPUFA V plasma [LCPUFA] esterified (Eq7) t 1/2,LCPUFA F LCPUFA (Eq8) Statistics All data are presented as mean SEM and were compared by one-way ANOVA using Tukey s test for multiple comparison if ANOVA p-value 0.05 (GraphPad Prism version 4.0, La Jolla, CA, USA). When data was not normally distrubuted as determined by Bartlett s test for equal variances, then data was log transformed and then compared by one-way ANOVA using Tukey s test for multuple comparisons. A p-value 0.05 with Tukey s multiple comparison test was considered significant. 6.4 Results Plasma Volume Determination Plasma volume (V plasma ) was previously determined in our laboratory to equal ml/g body weight, in long evans rats at 8 weeks post-weaning (Chapter 4). Using this estimate, plasma volume equaled , , ml (low, medium, high ALA diet respectively). Plasma volume was significantly higher in rats fed the medium diet compared to those fed the low ALA diet (p<0.05) Plasma n-3 and n-6 PUFA concentrations Plasma unesterified ALA concentrations (Figure 1a) were highest in rats fed the high ALA diet ( , and nmol/ml for rats fed the low, medium

141 127 and high ALA diet, respectively) while plasma unesterified LNA concentrations (Figure 1b) did not differ ( , , nmol/ml; low, medium and high, respectively). Rats fed the high and medium ALA diet had higher concentrations of plasma esterified ALA (7 1.7, 57 7, nmol/ml for the low, medium and high ALA diet, respectively), EPA ( , , nmol/ml; low, medium and high, respectively) and DHA (28 2, , ; low, medium, high, respectively) compared to rats fed the low ALA diet (Figure 2a-c). No differences were found for plasma esterified LNA ( , , nmol/ml; low, medium, high, respectively) and ARA ( , , nmol/ml, respectively, Figure 2d and e).

142 128 Figure 1: Plasma unesterified concentrations of ALA (a) were higher in rats the high ALA diet compared to those fed the medium and low ALA diet. While plasma unesterified LNA (b) was not different in rats fed either diet. Data are expressed in nmol/ml as mean SEM. Different letters signify that means are significantly different as measured by one-way ANOVA using Tukey s test for multiple comparisons (p<0.05). Data that was not normally distributed, as measured by Bartlett s test for unequal variances, was log transformed. n= 8,8,6 (low, medium and high ALA diet, respectively). a) b)

143 129 Figure 2: Plasma esterified concentrations of ALA (a) EPA (b) and DHA (c) were higher in rats the high ALA diet compared to those fed the medium and low ALA diet. While plasma esterified LNA (d) and ARA (e) was not different in rats fed either diet. Data are expressed in nmol/ml as mean SEM. Different letters signify that means are significantly different as measured by one-way ANOVA using Tukey s test for multiple comparisons (p<0.05). Data that was not normally distributed, as measured by Bartlett s test for unequal variances, was log transformed. n= 8,8,6 (low, medium and high ALA diet, respectively). a) b) c) d) e)

144 Plasma 2 H 5 n-3 PUFA and 13 C 18 n-6 PUFA The average concentration of the n-3 PUFA tracer ( 2 H 5 -ALA) in the unesterified plasma lipid fraction, throughout the infusion, did not differ between rats fed the 3 diets and equaled , , nmol/ml (low, medium, high ALA diet respectively). The average concentration of the of the n-6 PUFA tracer ( 13 C 18 -LNA) in the unesterified plasma lipid fraction, throughout the infusion, was higher in the high ALA diet group compared to the medium ALA diet group ( > nmol/ml for rats fed the high and medium ALA diet respectively, p<0.05). The average plasma unesterified 13 C 18 -LNA concentrations, throughout the infusion, in rats fed the low ALA diet did not differ from rats fed the other diets ( nmol/ml) n-3 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life Synthesis parameters for DHA, EPA and ARA synthesis are found in table 2. For both EPA and DHA there were no significant differences with respect to S max between rats fed any diet ( , , and nmol/min for EPA and , and nmol/min for DHA, low medium and high ALA diet respectively). k 1,EPA was not significantly different across the diets ( , and ml/min; low, medium and high, respectively), however, the EPA synthesis rate was highest in rats fed the high ALA diet compared to those fed the low and medium ALA diet ( = < nmol/day; low = medium < high diet, respectively). F EPA was significantly higher in rats fed the low and high ALA diets compared to those fed the medium diet ( , and /day low, medium, high diets, respectively) and t 1/2,EPA was

145 131 significantly lower in rats fed the low and high ALA diets compared to those fed the medium ALA diet ( , and days; low, medium, high, respectively).

146 132 Table 2. Synthesis parameters for ARA, EPA and DHA synthesis as determined from a 3 hour steady-state infusion of 2 H 5 -ALA and 13 C 18 -LNA. Fatty Acid (n) DHA Diet S max (nmol/min) k 1,n (ml/min) J n (nmol/min) Daily Synthesis- Secretion Rate (nmol) F n (/day) t 1/2, n (day) Low ± ± a ± a 6.2 ± 1.6 a 0.29 ± ± 21.5 Medium ± ± b ± b 46.8 ± 12.5 b 0.22 ± ± 31.7 High ± ± a ± b 64.5 ± 18.0 b 0.29 ± ± 25.3 EPA Low 0.29 ± ± ± a 85.7 ± 22.6 a 51.1 ± 35.8 a 0.15 ± 0.07 a Medium 0.40 ± ± ± ab ± 89.2 ab 2.6 ± 1.0 b 1.2 ± 0.43 b High 0.37 ± ± ± 0.24 b ± b 5.3 ± 3.2 ab 1.1 ± 0.68 ab Low ± ± ± ± ± ± 10.4 ARA Medium ± ± ± ± ± ± 9.8 High ± ± ± ± ± ± 21.6 Data are expressed as means SEM. S max = maximum first derivative, k 1,n = synthesis-secretion coefficient, J n = synthesis rate, F n = turnover, t 1/2,n = half-life. Sample size = 8,8,6 for the low, medium and high ALA diets, respectively. Different letters within a fatty acid indicate the means are significantly different as measured by one-way ANOVA using Tukey s test for multiple comparisons (p<0.05). Data that was not normally distributed, as measured by Bartlett s test for unequal variances, was log transformed.

147 133 The daily synthesis rate for DHA was lower in the low ALA diet group compared to rats fed the medium and high DHA diet ( < = nmol/day, low, medium and high, respectively). k 1,DHA was significantly higher in rats fed the medium ALA diet as compared to those fed the high and low ALA diet ( > = ml/min; medium, high and low ALA diet, respectively). F DHA ( , , /day; low, medium and high diet, respectively) and t 1/2,DHA ( , , days; low, medium and high, respectively) did not differ in rats consuming either diet n-6 PUFA synthesis-secretion coefficients, rates, turnover rates and half-life None of the synthesis parameters measured for ARA were significantly different between rats fed the 3 diets (table 2). S max equaled , and for rats fed the low, medium and high ALA diet. K 1,ARA equaled , and ml/min (low, medium and high ALA diet, respectively). Daily ARA synthesis rates were equal to , and nmol/day for rats fed the low, medium and high ALA diet, respectively. Furthermore, measures of ARA turnover (F ARA and t 1/2,ARA ) did not differ between rats consuming the 3 diets. F ARA equaled , and /day and t 1/2,ARA equaled , and days (low, medium and high ALA diet, respectively). 6.5 Discussion We found that after 8 weeks of dietary feeding rats that consumed a low ALA diet (<0.1% of the fatty acids by weight as ALA) had significantly lower plasma esterified

148 134 ALA concentrations compared to those consuming higher amounts of ALA. Additionally, plasma esterified EPA and DHA were lower in rats fed the low ALA diet compared to rats fed the higher ALA diets, while plasma esterified ARA was unchanged. The observed responses of plasma esterified lipids to these dietary conditions is in concordance with previous reports (107, 138). Contrary to the changes in esterified ALA we found that rats that were fed a high ALA diet had significantly higher plasma unesterified ALA concentrations than rats fed diets containing lower amounts of ALA which also agrees with previously published observations (85, 107, 138). Contrary to previous reports, we did not find a significant difference in unesterified ALA between rats fed the medium and low ALA diets. However, it is possible that we were underpowered to detect such a difference, as there was a trend towards a higher plasma unesterified ALA concentration in rats fed the medium ALA diet compared to those fed the low ALA diet (>2-fold difference, p=0.12). We directly measured DHA, EPA and ARA synthesis in rats in response to dietary ALA and found that DHA synthesis was highest in rats fed the high and medium ALA diets compared to rats fed the low ALA diet. This indicates that the higher plasma DHA concentrations observed in rats fed the high and medium ALA diets were due, at least in part, to increased DHA synthesis from ALA. Though there was no significant difference in the daily synthesis rate between rats fed the high and medium ALA diet, k 1,DHA was significantly higher in rats fed the medium ALA diet, meaning that rats fed the medium ALA diet had higher capacity to synthesize DHA compared to the high ALA diet. The lower k 1,DHA, therefore, contributed to the similar DHA synthesis rates between rats fed the medium and high ALA diet, despite the fact that rats fed the high ALA diet

149 135 had significantly higher concentrations of plasma unesterified ALA (the substrate for DHA synthesis). This finding indicates that the body adapts to low substrate availability for DHA synthesis by increasing the capacity to synthesize DHA, therefore, maintaining a stable DHA synthesis rate across a wide range of ALA intakes. Alternatively, the higher k 1,DHA that was found in rats fed the medium ALA diet may be due to a downregulation of enzymes involved in DHA synthesis in rats fed the high DHA diet, as it has been previously reported that dietary ALA decreases the expression of the enzymes involved in DHA synthesis in the liver (100). EPA synthesis rates were highest in rats fed the high ALA diet compared to those fed the medium and low ALA diet. EPA turnover was most rapid and its half-life was shortest in rats fed the low ALA diet indicating rapid uptake of EPA from the plasma. Future studies should aim to confirm this finding using a method that can directly measure PUFA turnover within the plasma as the steady-state infusion method can only estimate PUFA turnover, when that PUFA is absent from the diet, based on the synthesis rate of that PUFA. ARA synthesis rates were unaffected by diet though there is a trend towards lower ARA synthesis in rats fed the high ALA diet these differences did not reach statistical significance (t-test between the high and low ALA diet groups yielded a p-value=0.1). It should be noted that the estimates for the synthesis parameters were highly variable and limited our statistical power. Future studies should consider adjustments to the experimental protocol that will decreases variability in the measure and strengthen statistical power. Furthermore, none of the synthesis-secretion parameters for ARA were statistically different between rats fed the 3 diets. Since there were no differences in

150 136 plasma concentrations of ARA it would be unexpected to have found differences in ARA synthesis-secretion rates between rats fed the different diets. It is known that both n-6 and n-3 PUFA are substrates for the desaturase and elongase enzymes involved in the synthesis of long chain n-6 and n-3 PUFA (104, 105), however, the results of this study do not indicate that competition between these two families of PUFA affected synthesis/secretion rates. An unexpected observation was that k 1,DHA was similar between rats fed the high and low ALA diet while k 1DHA was increased in rats fed the medium ALA diet. It has previously been reported that the expression of the desaturase and elongase enzymes involved in DHA synthesis is upregulated in liver and DHA synthesis capacity is increased in rats consuming a diet deprived of n-3 PUFA (100, 107). While Igarashi et al. found that rats fed a low ALA diet had increased DHA synthesis capacity compared to rats fed adequate n-3 PUFA, we found that rats fed the low ALA diet, which contained no added n-3 PUFA, had a lower synthesis-secretion capacity compared to rats fed a diet with ALA added at a concentration of 3% of the fatty acids. This may be explained by experimental differences, diets in this study were formulated by adding pure ALA ethyl esters whereas Igarashi et al. formulated their diets with vegetable oils. Alternatively, while Igarashi et al. measured DHA synthesis in the liver we determined synthesissecretion capacity by measuring labeled n-3 PUFA that were secreted into the blood. Therefore, it is possible that the lower k 1,DHA observed in rats fed the low ALA diet compared to those fed the medium ALA diet is due to a decreased capacity for the liver to secrete lipoproteins containing DHA when rats are fed a diet that is low in n-3 PUFA. Previously, we have reported that in rats fed a diet containing 2% of the fatty acids as

151 137 ALA, DHA synthesis-secretion capacity was increased compared to rats fed a control diet containing 0.2% of the fatty acids as ALA (85). We have previously performed the steady-state infusion method to estimate DHA synthesis rates in rats fed diets containing 2% and 0.2% of the fatty acids as ALA (85). The synthesis rates calculated in this report were fold lower than our previous study. This difference in synthesis rates may be explained by several methodological differences. Previously, we determined DHA synthesis rates in rats at 15 weeks postweaning whereas synthesis-secretion rates were measured in this report at 8 weeks postweaning. Another difference was the use of heparin as an anti-coagulant. Heparin is a known activator of lipoprotein lipase (219, 220), which may lead to an overestimation of the baseline plasma unesterified ALA concentration. The current study used a nonheparinized plasma sample to determine the baseline plasma ALA concentration whereas we previously used a plasma sample from rats that had been treated with heparin. Overestimation of the baseline plasma unesterified ALA concentration would lead to an overestimation of the DHA synthesis rate. The results of this study indicate that DHA synthesis is relatively constant across a wide range of ALA intakes, however, there appears to be a level of ALA intake below which DHA synthesis is decreased significantly. One possible explanation for the lower DHA synthesis rates found in rats consuming the low ALA diet is the lack of substrate as plasma unesterified ALA concentrations were significantly lower in rats fed the low ALA diet compared to rats fed the high ALA diet. However, rats fed the medium ALA diet also had significantly lower plasma unesterified ALA concentrations compared to those fed the high ALA diet. The synthesis-secretion capacity of DHA (k 1,DHA ) in rats fed the

152 138 medium ALA diet was increased leading to no difference in DHA synthesis rates compared to rats fed the high ALA diet. The lack of an observed increase in k 1,DHA, as we reported previously (85), in rats fed the low ALA diet could be due to either a decreased capacity to synthesize DHA, or a decreased capacity for organs that synthesize DHA to secrete DHA into the plasma. Current intakes of ALA in North America range from % of calories (7) with the majority of ALA being consumed in vegetable oils (canola in Canada and soybean in the United States) (46, 233). Recently, industry is moving towards producing largely high-oleic acid varieties of vegetable oils, which can reduce the ALA content up to 4-fold (234, 235). In light of the results in this study, lowering dietary ALA may negatively affect rates of DHA synthesis. This study found that rats fed the low ALA diet, which contained 0.02% of calories as ALA had a significantly lower DHA synthesis rate than rats fed the medium and high ALA diets (which provided 0.7 and 2.3% of calories as ALA, respectively). Therefore, based on the results of this study lowering population ALA intakes may put members of the population at risk of having decreased DHA synthesis. In conclusion, DHA synthesis remained constant in rats fed diets consisting of 3-10% of the fatty acids as ALA. However, there is a level of ALA intake below 3% of the fatty acids at which DHA synthesis cannot be maintained, either due to decreased substrate availability or decreased synthesis-secretion capacity.

153 7 General Discussion 139

154 Discussion Overall findings The research presented in this thesis supports the overall hypothesis that DHA synthesis from ALA is sufficient to maintain brain DHA. DHA synthesis from ALA is at least 3-fold greater than brain DHA uptake, which likely explains the lack of differences in DHA concentrations and expression of genes regulated by DHA in the brain in rats fed diets with ALA but not DHA. Furthermore, dietary LNA does not appear to compete with ALA for enzymes involved in DHA synthesis and DHA synthesis is maintained across a wide range of ALA intake levels. There is a level of ALA intake below which DHA synthesis cannot be maintained which can explain the decreased brain DHA concentrations in rats fed low levels of ALA (control diet chapter 4 and low ALA diet chapter 6). A summary of the affect of dietary fatty acids on DHA synthesis rates (as measured by the steady-state infusion method) is presented in table 1.

155 141 Table 1: Summary of results for DHA synthesis rates. Chapter 4 Chater 5 Chapter 6 Control Diet ALA Diet DHA Diet Low LNA Diet Medium LNA Diet High LNA Diet Low ALA Diet Medium ALA Diet High ALA Diet dietary ALA dietary LNA dietary ALA:LNA ratio [ALA] unesterified [DHA] esterified k 1,DHA - ND - - DHA synthesis rate - ND - - T 1/2,DHA - ND - - Arrows indicate that means are increased, decreased or not significantly different when compared to the control or low diet group within a study. A double arrow indicates that the mean is significantly higher in the high diet group than the medium diet group. ND means that values were not determinable. p<0.05 [ALA] unesterified = plasma unesterified ALA concentrations [DHA] esterified = plasma esterified DHA concentrations k 1,DHA = synthesis-secretion capacity for DHA T 1/2,DHA = plasma half-life DHA

156 Biological significance DHA makes up 10% of brain lipids and is thought to be important for brain function at least in part by regulating cell survival and neuroinflammation (24, 25, 28). We have shown that brain DHA concentrations are maintained when rats are fed a diet containing 2% of the fatty acids as ALA and expression of genes, in the brain, that are regulated by DHA were maintained with a diet containing as low as 0.2% of the fatty acids as ALA. DHA synthesis rates were higher than brain DHA uptake rates when rats were fed a 2% ALA diet, therefore, DHA synthesis rates in these rats was likely sufficient to maintain brain DHA concentrations. There is a belief among experts that n-6 and n-3 PUFA compete for desaturase and elongase enzymes involved in DHA synthesis. Evidence for this belief comes from higher plasma and tissue levels of DHA in rats fed LNA deprived diets compared to LNA adequate diets (227, 236). Results from this thesis show that this does not occur across a wide range of dietary LNA concentrations. In fact, it was found that rats fed a low amount of LNA, in fact, have impaired DHA synthesis, slower turnovers and longer plasma half-lives of DHA, which may explain the higher plasma concentrations of DHA when rats are fed a LNA deprived diet. Though not measured in this thesis higher DHA synthesis rates found in rats consuming higher amounts of LNA may be due to increased expression of desaturase and elongase enzymes involved in DHA synthesis that has been previously reported to occur in piglets fed high amounts of LNA (237). In chapter 6 of this thesis we evaluated the effect of increased dietary ALA on DHA synthesis. While it is logical that increasing the substrate for DHA would increase its synthesis, it is also possible that increasing ALA will increase all n-3 PUFA

157 143 intermediates in the FADS pathway, and therefore, increase competition for the rate limiting Δ6 desaturase enzyme and decrease DHA synthesis. We found no difference in DHA synthesis rates between rats fed the diets containing 3% (medium ALA diet) and 10% (high ALA diet) of the fatty acids as ALA, however, the synthesis-secretion capacity was highest in rats fed the 3% ALA diet. This supports the hypothesis of increased competition in rats fed the high ALA diet; however, it is also possible that the increased synthesis-secretion capacity found in rats consuming the medium ALA diet was due to increased expression of the enzymes involved in DHA synthesis, as n-3 PUFA have been previously shown to decrease desaturase and elongase enzyme expression (100). Some have proposed an alternate mechanism for LC PUFA synthesis which involves 2 sets of enzymes, n-3 PUFA specific and n-6 PUFA specific (238). The results from chapter 5 seem to support this hypothesis, however, the results from chapter 6 do not seem to support this hypothesis as there was a trend towards higher ARA synthesis in rats fed diets with lower levels of ALA. There is also some debate pertaining to how the pathway for LC PUFA synthesis operates. For example, it is generally accepted that the Δ6 desaturase enzyme is the rate limiting step for DHA synthesis, yet recently, it has been proposed that an elongase enzyme may be rate limiting in this pathway (112). Importantly, this in vivo kinetic model can be used to test the assumptions of candidate pathways in vivo either by feeding different n-3 PUFA and observing the effect on DHA synthesis rates, or by infusing labeled versions of other n-3 PUFA that are intermediates in the FADS pathway and measuring DHA synthesis rates from these intermediates.

158 Clinical significance The experiments presented in this thesis used a novel in vivo kinetic model to calculate DHA synthesis rates in rats. Rates calculated using this method were found to be in line with DHA synthesis rates measured by the balance method (a gold standard), thus, validating the in vivo kinetic model. Unlike the balance study the in vivo kinetic model requires an intravenous infusion and periodic blood sampling, which can be utilized to measure, for the first time, the DHA synthesis rates in humans. Moreover, PET imaging has already been used to estimate brain DHA uptake rates (estimated to be mg/day (82, 83)), therefore, once a DHA synthesis rate is quantified in humans it can be compared to human brain DHA uptake rates to estimate the sufficiency of DHA synthesis in maintaining brain DHA (as we have done in chapter 4 of this thesis). Chapter 5 of this thesis addressed a major concern that the large increase in LNA consumption, in western diets over the past century has led to lower n-3 PUFA levels in the body due to increased competition for desaturase and elongase enzymes (46). We have shown that diets with large amounts of LNA do not lead to lower DHA synthesis rates. Furthermore, we have shown in chapter 6 that a large increase in ALA consumption does not lead to increased DHA synthesis. Though rats consuming very low levels of ALA had significantly lower DHA synthesis than those consuming higher levels of ALA, there was no increase in DHA synthesis, or plasma DHA concentrations when rats consumed ALA at a concentration of 10% of the fatty acids compared to 3% of the fatty acids. Therefore, the results from chapters 5 and 6 of this thesis should serve as a guide for selecting dietary patterns of subjects in clinical trials where DHA synthesis rates will be measured.

159 145 The findings in chapter 6 also explain results from many human studies where plasma DHA concentrations are not increased by increases in plasma ALA (129, 131, 134, 135). While some experts in the field use this as evidence for a lack of DHA synthesis in humans (31, 115), our data suggest that DHA synthesis may be maximal and increasing dietary ALA does not lead to increased plasma DHA because DHA is already being synthesized at a maximal rate. It should be noted that based on our previous results (Chapter 4) we would expect that rats consuming the medium and high ALA diet attained maximal brain DHA concentrations, which may explain why DHA synthesis is maximal with a relatively low level of dietary ALA intake. If DHA synthesis is also maximal in humans with regular ALA intake, this may indicate that DHA synthesis rates are sufficient to meet the requirement for DHA. However, until the requirement for DHA is determined the sufficiency of DHA synthesis rates remain speculative. 7.2 Limitations There were several limitations to the studies described in chapters 4-6 of this thesis. Firstly, the fat content of our diets consisted of a large amount of fatty acid ethyl esters, which was necessary to keep the fatty acid content of the diets as consistent as possible. While the use of pure fatty acid ethyl esters allowed us to isolate effects to a single fatty acid of interest, these results may not be applicable to free living situations where fatty acids are being consumed in food and oil sources containing a mixture of fatty acids. Another limitation to this work is a lack of consistency in the experimental protocol across studies. While rats in chapter 4 were fed diets for 15 weeks post-weaning before being subjected to the steady-state infusion, rats in chapters 5 and 6 were fed diets for only 8 weeks post-weaning. Moreover, the steady-state infusion was originally performed

160 146 via the tail vein in chapter 4, however, by chapter 6 the method improved to a jugular vein infusion that could be performed directly in the rats cage. While, the method has improved to reduce stress on the rats and maximize success rates the lack of consistency limits the comparisons that can be made between studies. The DHA synthesis rates estimated using the steady-state infusion method are highly variable. As a result of this variability some differences in synthesis rates between dietary groups may not have been detected. For example, there is an approximately 40% increase in the DHA synthesis rate in rats fed the high ALA diet compared to rats fed the medium ALA diet (Chapter 6) which, was not statistically significant. When determining an appropriate sample size for these experiments we used previously published studies to estimate the variance and effect sizes. These previously published papers have since been retracted and a power analysis of our data indicate that we were underpowered (less than 80% power) to determine differences in DHA synthesis rates between certain groups (in particular, Chapter 5 low LNA diet vs. medium LNA diet and Chapter 6 medium ALA diet vs. high ALA diet). Future experiments should be performed in fasting conditions, which might limit variability in baseline plasma unesterified ALA (and LNA) concentrations. Future experiments should also consider using only two dietary groups (i.e. low vs. high) as sample sizes required to achieve appropriate power to detect smaller differences may be implausible. The different sample size between groups within experiments is also a limitation to this work. Initially an equal number of rats are distributed to each diet. However, throughout the course of the study rats die due to various causes. Particularly, the fact that anti-coagulants are not used throughout the surgeries leads to an increased chance of

161 147 stroke before animals are infused. Rats often die due to complications with the surgery and infusion, which was a major cause of the low sample sizes in Chapters 4 and 5. The free-living infusion model was improved in Chapter 6 to prevent animal deaths during the infusion procedure. Moreover, animal deaths during the study are a phenomenon that occurs randomly and should not bias the study results. As a large portion of this thesis relies on the steady-state infusion model for measuring DHA synthesis it is important to discuss the limitations to this method. Some major limitations such as labeled DHA tissue uptake, tracer dilution in the liver and assuming that the DHA synthesis rate is constant throughout the day have been discussed in detail in chapter It is important to understand that these limitations would likely result in underestimations to the DHA synthesis rate. Furthermore, since measuring the dilution of tracer in the liver is likely not feasible in humans, if the dilution is large or changes due to dietary conditions then the steady-state infusion method may not be able to be used in humans to determine the actual rate of DHA synthesis. Additionally, tracer dilution in the liver will affect the accuracy of the DHA synthesis measurement. However, accuracy cannot be determined as the steady-state infusion model is a novel model to measure DHA synthesis. While in Chapter 4 the rates determined using the steady-state infusion method were compared to the balance study DHA synthesis rates, the rates measured by these two methods are not totally comparable. Another limitation to the steady-state infusion method is the reliance on hepatic secretion of labeled products into the plasma. Since labeled PUFA products need to be synthesized and secreted into the plasma to be measured we essentially measure a DHA synthesis-secretion rate. Therefore, if an experimental condition impairs secretion of fatty

162 148 acids from the liver (or any other tissues where they are being synthesized) then the measured DHA synthesis-secretion rate will be an underestimate of the actual DHA synthesis rate. Moreover, if diet affects the rate of DHA secretion then comparisons of rates between diets will reflect differences in DHA secretion more so than differences in DHA synthesis and comparison of DHA synthesis rates will not be possible. 7.3 Future Studies We estimated how DHA synthesis rates were affected by different levels of ALA and LNA were consumed in the diet. It would next be useful to determine how dietary DHA affects DHA synthesis rates. DHA has been shown to downregulate enzymes involved in its own synthesis (158), therefore, it would be important to determine how this affects the DHA synthesis rate. Also, studies should be performed in rats to determine the effect of the limitations described in chapter 7.2. Specifically, the dilution of tracer in the liver acyl-coa pool should be quantified and the effect of diet on this dilution should be investigated. Furthermore, it should be determined if diet can affect the secretion of DHA from the liver. This is of particular interest for rats consuming a low LNA diet to confirm if the results of the experiment presented in chapter 5 are due to DHA synthesis or impaired DHA secretion. Additionally, more studies need to be done to elucidate details of the FADS pathway. Many findings presented in this thesis were contrary to what would be expected based on the accepted pathway for DHA synthesis from ALA. Some debate exists around the generally accepted FADS pathway as it has recently been shown that the Δ6 desaturase enzyme also has the Δ8 and Δ4 desaturase activity, which provides evidence for an alternative pathway for DHA synthesis (111, 239). Furthermore, the accepted

163 149 belief that the Δ6 desaturase enzyme is the rate limiting enzyme in DHA synthesis has also been questioned (112). It should be investigated if these results differed from what we would hypothesize because the accepted DHA synthesis pathway has not been characterized correctly, or if the details on how the enzymes involved in the pathway function are not correctly understood. Importantly, the steady-state infusion method can be used to test these hypotheses in vivo in humans by feeding intermediates in the FADS pathway, or by infusing stable isotope labeled intermediates in the FADS pathway and measuring the DHA synthesis rates. Another study of importance would be to measure the DHA synthesis rate and brain DHA uptake rate in single human subjects. This can be achieved by utilizing the steadystate infusion method (descrbied in this thesis) to measure DHA synthesis rates and PET imaging to measure brain DHA (82, 83). Such a study would provide an estimate of how much of the brain DHA requirement can be accounted for by DHA synthesis from ALA. Also, more studies need to be done investigating the role of DHA in the brain in order to properly determine a brain DHA requirement. Currently, the best estimate available for a brain DHA requirement is the brain DHA uptake rate, measured by PET imaging to mg/day (82, 83), which is not a high (about 0.3% of North American ALA intakes and 5% of North American DHA intakes). If this is an accurate estimate of the brain DHA requirement then it is possible that even if DHA synthesis rates are low they may be sufficient to support the brain. Finally, as discussed in chapter of this thesis conditions that affect DHA synthesis or brain DHA uptake rates should be investigated to determine if DHA synthesis is sufficient to maintain brain DHA in these instances. In particular, infants

164 150 accrete large amounts of DHA, and therefore, likely have a higher brain DHA requirement. DHA synthesis rates in these instances should be measured and compared to brain DHA uptake rates to determine the sufficiency of DHA synthesis to maintain brain DHA. 7.4 Conclusions The research presented in this thesis support the hypothesis that DHA synthesis from ALA can maintain brain DHA because: 1. DHA concentrations and expression of genes that are regulated by DHA in the brain were maintained when rats consume ALA and not DHA. 2. DHA synthesis rates were higher than brain DHA uptake and accretion rates. Dietary LNA and ALA affected the DHA synthesis rates in the following manner: 3. DHA synthesis rates were not lower when rats are fed diets with high amounts of LNA. 4. DHA synthesis rates were maintained across a wide range of dietary ALA concentrations but at low levels of ALA, DHA synthesis rates were not maintained.

165 151 8 Appendix Supplementary Figure 1: Study 1 design A) Balance Study B) Steady-state infusion study and DHA uptake study Non-littermate pups arrive with lactating dams (1 dam per diet). Dams are allocated to one of the 3 diets. At weaning 2 pups per litter are collected to determine baseline brain and body fatty acid composition (Balance Study only). Pups are maintained on the diet of their respective dam for 15 weeks post weaning. After 15 weeks on the diets, rats are euthanized (balance study) or subjected to a surgery and infusion (steady-state infusion study and DHA uptake study).

166 152 Supplementary Figure 2: Study 2 design Lactating dams with pups are fed standard chow. At weaning pups are allocated to one of the three diets and fed this diet for 8 weeks. After 8 weeks of feeding rats are subjected to a surgery and, 3 days later, an infusion to determine DHA synthesis rates.

167 153 Supplementary Figure 3: Study 3 design Lactating dams with pups are fed standard chow. At weaning pups are allocated to one of the three diets and fed this diet for 8 weeks. After 8 weeks of feeding rats are subjected to a surgery and, 3 days later, an infusion to determine DHA synthesis rates.

168 154 Supplementary Table 1: Summary of n-6 PUFA balance Dietary Group: Control (n=10) ALA (n=10) DHA (n=10) Intake of LNA (μmol) ± ± ± 6855 Fecal Excretion of LNA (μmol) 950 ± ± ± 34 Body Content of n-6 PUFA (μmol) Day ± ± ± 35 Day ± ± ± 2116 Brain Content of n-6 PUFA (μmol) Day 0 14 ± ± ± 0.6 Day ± ± ± 0.9 Total Accretion (μmol) LNA ± ± ± :3n-6 57 ± 5 52 ± 5 41 ± 5 20:2n ± ± ± 17 ARA 2132 ± 97 a 1806 ± 93 ab 1539 ± 85 b 22:4n ± 13 a 138 ± 12 a 97 ± 12 b DPAn ± 16 a 168 ± 9 b 67 ± 13 c Total ± ± ± 2111 Metabolic consumption of dietary PUFA (μmol) ± ± ± 5519 Data are means SEM, different letters signify means are significantly different (p<0.05) measured by One-Way ANOVA followed by Tukey s test for multiple comparisons

169 155 Supplementary Table 2: Whole body fatty acids for rats fed the control, ALA or DHA diet for 15 weeks Fatty Acid Control ALA DHA 12: ± ± ± : ± ± ± :1n ± ± ± : ± ± ± :1n ± ± ± : ± ± ± :1n ± ± ± :1n ± ± ± :2n ± ± ± :3n-6 73 ± 5 67 ± 5 56 ± 5 18:3n-3 89 ± 6 a 1557 ± 99 b 93 ± 7 a 20:1n ± ± ± 17 20:2 206 ± ± ± 20 20:3n ± 7 a 137 ± 11 ab 192 ± 21 b ARA (20:4n-6) 2280 ± 98 a 1954 ± 93 ab 1688 ± 44 b EPA (20:5n-3) 55 ± 3 a 76 ± 9 b 80 ± 8 b 22:4n ± 16 a 155 ± 12 ab 115 ± 6 b 22:5n ± 21 a 124 ± 6 b 52 ± 3 c 22:5n-3 19 ± 2 a 101 ± 7 b 80 ± 7 b DHA (22:6n-3) 106 ± 6 a 481 ± 40 b 1160 ± 63 c Data are expressed in μmol, as mean +/- SEM. Different letters signify the means are significantly different (p<0.05) measured by One-way ANOVA followed by Tukey s test for multiple comparisons or Kruskal-Wallis test followed by Dunn s multiple comparison test (if variances were significantly different).

170 156

171 157

172 158

173 159

174 160

175 161

176 162

177 163

178 164

179 165

180 166

181 167