The Effect of Omega-3 Polyunsaturated Fatty Acids on the Resolution of Inflammation in the Rodent Brain

Transcription

1 The Effect of Omega-3 Polyunsaturated Fatty Acids on the Resolution of Inflammation in the Rodent Brain By Marc-Olivier Trépanier A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Department of Nutritional Sciences University of Toronto Copyright by Marc-Olivier Trépanier 2016

2 The Effect of Omega-3 Polyunsaturated Fatty Acids on the Resolution of Inflammation in the Rodent Brain Abstract Marc-Olivier Trépanier Doctorate of Philosophy Department of Nutritional Sciences University of Toronto 2016 Resolution of inflammation in the periphery is believed to be mediated by omega- 3 polyunsaturated fatty acids (n-3 PUFA) derived specialized pro-resolving lipid mediators. However, the resolution of neuroinflammation, and the role of n-3 PUFA and their specialized pro-resolving lipid mediators in the resolution of neuroinflammation have yet to be studied. Moreover, while ischemia induces the production of various mediators in the brain, this effect has yet to be demonstrated for specialized pro-resolving lipid mediators. The first objective of this thesis was to develop a lipidomic approach to measure the rodent neurolipidome without the effect of ischemia using head-focused microwave fixation. Once a lipidomic approach was developed, we attempted to develop a selfresolving model of neuroinflammation and to determine the effect of increasing brain docosahexaenoic acid (DHA) on resolution of neuroinflammation. We demonstrated that microwave-fixation inhibits ischemia-induced production of bioactive mediators, including specialized pro-resolving lipid mediators, and changes in various intact lipid species. ii

3 We then developed a self-resolving model of neuroinflammation using intracerebroventricular injections of lipopolysaccharide (LPS). Following LPS injection, microglia activation peaked at 5 days and returned to baseline by 21 days. Using a microarray, we illustrated that various markers had varying time courses of inflammation. Interestingly, no neutrophil infiltration was detected. Since neutrophils carry the lipoxygenase enzyme, which produces specialized pro-resolving lipid mediators, we also did not detect specialized pro-resolving mediator production following LPS injection as measured by our new lipidomic approach combined with microwave fixation. In order to increase brain DHA, we compared a wildtype mouse fed a safflower diet deficient in n-3 PUFA to the fat-1 mouse and a wildtype mouse fed a fish oil diet high in n-3 PUFA. Increasing brain DHA resulted in modest increases in resolution of microglia activation and cyclooxygenase (COX)-2 mrna expression. However, many other inflammatory markers were unaffected by the increased brain DHA. In conclusion, we illustrated that microwave fixation inhibits the ischemiainduced changes on the rodent neurolipidome and that n-3 PUFA have small proresolving properties in a self-resolving model of neuroinflammation. These appear to be independent of specialized pro-resolving mediator production. iii

4 Acknowledgements First and foremost, I would like to thank my supervisor, Dr. Richard Bazinet, for mentoring me for the past 7 years. Your guidance throughout my academic career has been invaluable and I would not have achieved what I have achieved to this date without it. I do not know what my future career holds, but I am confident that because of you I am more than ready. Finally, I just wanted to say thank you for making us feel appreciated. All the dinners and the nights out have created memories I will not soon forget, and I always enjoy our conversation about food and drinks. I don t know where I will end up, but I truly hope we can keep in touch. I also need to thank my co-supervisor, Dr. Mojgan Masoodi. You have been a great help throughout my thesis. Moreover, you were a great host during my trip to Lausanne. I will never forget eating fondue in Gruyere. I also want to thank my advisory committee, Dr. Ali Salahpour and Dr. Romina Mizrahi. Your guidance throughout my thesis was much appreciated. Special thanks also need to go to Dr. W.M. Burnham, my first mentor. I would also like to thank my lab mates for being such great colleagues over the past 7 years. I would especially like to thank Katie. You were so generous with your time and working with you was a pleasure. Anthony, sharing an office with you and chatting about sports definitely made my day more entertaining. I d also like to thank Sarah for training me and helping me get my project off the ground. Vanessa, it was very enjoyable to train you and you have a bright future ahead of you. To the rest of you, Chuck, Lauren, Kayla, Lin, Shoug, Scott, Alex, and Adam, I want to thank iv

5 you for all the help you offered over the years and for making the work place so great. The Natural Science and Engineering Research Council should be thanked for providing me with a studentship over the course of my studies. The Canadian Institute of Health Research should be acknowledged for funding the project. Finally, I would like to thank the International Society for the Study of Fatty Acids and Lipids for awarding me with the International Research Exchange Scholarship to allow me to travel to Lausanne, Switzerland. Finally, I want to thank all my family and friends for being there for me over the years. To Louise and Pierre, thank you for always being there for me and being my home away from home. Sarah and Clayton, thank you for always making time for me either by coming to visit or hosting dinner. I can t wait to meet Raphael. To my parents, Mario and Francine, words can t describe the gratitude I have for you. This thesis would not have been possible if it weren t for you. I dedicate this thesis to you. And finally, Claudia, the love of my life, thank you for being in my life and for all the support you offer. I just can t wait to start this next chapter in my life with you. v

6 Table of Contents List of Figures... ix List of Tables... xi List of Abbreviations... xiii Chapter 1: Introduction Polyunsaturated fatty acids Sources of PUFA Synthesis Brain uptake N-3 PUFA and anti-inflammation Models of neuroinflammation The lipopolysaccharide (LPS) model of neuroinflammation The fat-1 mouse Objectives Chapter 2: N-3 Polyunsaturated Fatty Acids in Animal Models with Neuroinflammation: An Update Abstract Introduction Results n-3 PUFA and neuroinflammation in ischemia or ischemia/reperfusion n-3 PUFA and neuroinflammation in spinal cord injury n-3 PUFA and neuroinflammation in aging n-3 PUFA and neuroinflammation in Parkinson s disease n-3 PUFA and neuroinflammation with lipopolysaccharide n-3 PUFA and neuroinflammation in i.c.v. IL n-3 PUFA and neuroinflammation in traumatic brain injury n-3 PUFA and neuroinflammation in neuropathic pain n-3 PUFA and neuroinflammation in diabetes n-3 PUFA and neuroinflammation in other models Conclusion Acknowledgments Chapter 3: Objectives and Hypotheses Objectives Hypotheses Chapter 4: High-resolution lipidomics coupled with rapid fixation reveals novel ischemia-induced signaling in the rat neurolipidome Abstract Introduction Methods Subjects Treatment groups Microwave fixation Brain preparation Lipid extraction vi

7 Mass spectrometry analysis Data analysis Results Discussion Acknowledgements Author contributions Conflict of interest statement Chapter 5: N-3 polyunsaturated fatty acids mediate small changes in the resolution of neuroinflammation following intracerebroventricular lipopolysaccharide injection independent of pro-resolving lipid mediators Abstract Introduction Methods Diets Subjects Intracerebroventricular LPS injections Immunohistochemistry Genetic expression analysis Lipidomic analysis Bioactive mediator extraction Extraction of intact lipids from the brain Mass spectrometry analysis Total lipid extraction Fatty acid methyl ester analysis by gas-chromatography for Experiment Y-maze Statistics Results Experiment Microglial activation peaked by 5 days and resolved by 21 days, independent of neutrophil and macrophage infiltration Gene expression of various neuroinflammatory markers have different time courses of expression following LPS injection Neuroinflammation alters some intact lipid species, but does not affect the production of bioactive mediators Neuroinflammation does not affect cognitive abilities in the Y-maze Experiment The fat-1 gene and fish oil diet increases brain DHA Increased brain DHA increases microglial resolution Increased brain DHA decreases COX-2 expression but not the expression of other pro-inflammatory markers Discussion Chapter 6: Discussion Overall findings Limitations Future directions Significance Conclusions References vii

8 Appendix 1: Postmortem evidence of cerebral inflammation in schizophrenia: a systematic review viii

9 List of Figures Figure 1-1. The n-3 and n-6 synthetic pathways... 4 Figure 4-1. Flow of methods in Chapter Figure 4-2. Microwave fixation inhibits ischemia-induced production of bioactive lipid mediators Figure 4-3. Microwave fixation inhibits ischemia-induced changes of intact lipids.. 84 Figure 4-4. Correlation network between lipid mediators and intact lipids in the CO2 group Figure 5-1. Time course of Iba1 optical density in the hippocampus in the C57Bl/6 mouse following i.c.v. LPS Figure 5-2. Hippocampal microglial M1 markers response to i.c.v. LPS over time. 120 Figure 5-3. Hippocampal microglial M2 markers response to i.c.v. LPS over time. 121 Figure 5-4. Hippocampal mrna expression of infiltrating cell markers following i.c.v. LPS over time Figure 5-5. Hippocampal mrna expression of astrocytic markers following i.c.v. LPS over time Figure 5-6. Hippocampal cytokine mrna response to i.c.v. LPS over time Figure 5-7. Hippocampal NF- B pathways mrna markers response to i.c.v. LPS over time Figure 5-8. Hippocampal arachidonic cascade markers response to i.c.v. LPS over time Figure Spontaneous alternation performance in the Y-maze 7 days following i.c.v. LPS injection ix

10 Figure Increased brain DHA in the fat-1 mice and mice fed a fish oil diet at 12 weeks of age Figure Effect of increased brain DHA on the resolution of microglial activation Figure Effect of increased brain DHA on the time course of mrna expression of example pro-inflammatory markers x

11 List of Tables Table 2-1: Summary of studies investigating the effects of n-3 PUFA in ischemia and ischemia/reperfusion models Table 2-2: Summary of studies investigating the effects of n-3 PUFA in spinal cord injury models Table 2-3: Summary of studies investigating the effects of n-3 PUFA in aging and Alzheimer s disease models Table 2-4: Summary of studies investigating the effects of n-3 PUFA in Parkinson s disease models Table 2-5: Summary of studies investigating the effects of n-3 PUFA in lipopolysaccharide models Table 2-6: Summary of studies investigating the effects of n-3 PUFA in IL-1 models Table 2-7: Summary of studies investigating the effects of n-3 PUFA on traumatic brain injury models Table 2-8: Summary of studies investigating the effects of n-3 PUFA in neuropathic pain models Table 2-9: Summary of studies investigating the effects of n-3 PUFA in diabetes models Table 2-10: Summary of studies investigating the effects of n-3 PUFA on other neuroinflammatory models Table 4-1: Confusion matrix for PLS-DA calculated for lipid mediators xi

12 Table 4-2: Class-based prediction statistics for PLS-DA calculated for lipid meditators Table 4-3: Confusion matrix for PLS-DA calculated for intact lipids Table 4-4: Class-based prediction statistics for PLS-DA calculated for intact lipids. 86 Table 4-5: Top 20 lipid mediators in PLS-DA discrimination of the four phenotypic groups Table 4-6: Top 20 intact lipids in PLS-DA discrimination of the four phenotypic groups Table 5-1: Percent of total fatty acids of the 3 experimental diets Table 5-2. Top 20 fold change in gene expression at each time point following LPS injection xii

13 List of Abbreviations A amyloid beta ANOVA analysis of variance ARA arachidonic acid BBB blood-brain barrier CD cluster of differentiation CCL chemokine (c-c motif) ligand CO2 5 minutes of CO2 asphyxiation (group) CO2 + MW 5 minutes of CO2 asphyxiation followed by microwave fixation (group) COX cyclooxygenase CXCL chemokine (c-x-c motif) ligand CX3CL chemokine (c-x3-c motif) ligand DHA docosahexaenoic acid EPA eicosapentaenoic acid F1SO fat-1 mice fed safflower oil (group) GFAP glial fibrillary acidic protein Iba1 ionized calcium-binding adaptor molecule 1 IFN interferon IL interleukin i.c.v. intracerebroventricular i.p. - intraperitoneal i.v. intravenous LCN - lipocalin LPS lipopolysaccharide or LPS injection 3 hr followed by microwave fixation (group) MHC major histocompability complex m/z mass/charge n omega NF- B nuclear factor kappa-light-chain-enhancer of activated B cells PET positron emission topography PG prostaglandin p.o. - per os PPAR peroxisome proliferator-activated receptor PUFA polyunsaturated fatty acid or acids qpcr quantitative polymerase chain reaction Ri resolution index SAA3 serum amyloid A3 s.c. subcutaneous SERPIN serpin peptidase inhibitor SEM standard error of the mean TNF tumor necrosis factor WT wildtype WTFO wildtype mice fed fish oil (group) WTSO wildtype mice fed safflower oil (group) xiii

14 Chapter 1: Introduction 1

15 1.1. Polyunsaturated fatty acids Fatty acids are defined as acyl chains containing a carboxylic group at the - carbon. There are several types of fatty acids, including saturated, monounsaturated, and polyunsaturated. Saturated fatty acids, which include stearic and palmitic acids, do not contain any double bonds. Monounsaturated acids, which include oleic acid, contain a single double bond. Polyunsaturated fatty acids (PUFA), such as docosahexaenoic acid (DHA) and arachidonic acid (ARA), have more than one double bond. The first double bond in relation to the methyl end of the fatty acids, known as the omega end, determines the family of polyunsaturated fatty acids. A double bond 3 carbons removed from the methyl carbon produces an omega-3 (n-3) fatty acid, such DHA, while a double bond 6 carbons away from the omega end generates an omega-6 (n-6) fatty acid, such as ARA Sources of PUFA Shorter chain PUFA cannot be made de novo and must be obtained through the diet. Although alpha-linolenic acid is most abundant in flaxseed, the major sources of n-3 PUFA in the western diet are soybean and canola oil 2. Longer chain n-3 PUFA are found in marine sources 2, 3. Salmon and herring are better sources of n-3 PUFA than other fish such as cod and catfish 3. Soybean oil is also a major source of n-6 PUFA in the western diet. Other rich sources of n-6 PUFA include corn and safflower oil 3. Since corn is a major source of feed in agriculture, most of the meat consumed in the western diet is high in n-6 PUFA, with 2

16 ratios of n-6 to n-3 PUFA that can reach as high as 25 to 1 4. The n-6/n-3 ratio of animal meat can return closer to 1 to 1 when animals are raised on grass or pasture 5, Synthesis As mentioned above, shorter chain fatty acids obtained from the diet can be elongated into longer chain PUFA. Alpha-linolenic acid, an 18 carbon omega-3 fatty acid is the precursor for the longer chain n-3 PUFA, while linoleic acid, an 18 carbon n-6 PUFA, is the precursor for longer n-6 PUFA. Once obtained from the diet, these precursor molecules can enter the elongation pathway (Figure 1-1). The fatty acids are desaturated by 5 and 6 desaturases and elongated by elongases. These enzymes involved in the elongation of longer chain PUFA are highly expressed in the liver 7, 8, representing the major site of DHA and ARA synthesis de novo. Other organs, such as the brain, have a lower expression of these enzymes and do not contribute significantly to the production of PUFA synthesis de novo 9. The synthesis rate of DHA from alpha-linolenic acid in the rat has been estimated to be approximately 1% 10. The rat is said to be a super converter, as human studies have reported much lower synthesis, as low as 0.01% However, due to methodological differences between animal and human studies, the conversion rate between the two species may be more similar than originally thought, with values in humans closer to 1% 14. 3

17 Figure 1-1. The n-3 and n-6 synthetic pathways n-3 pathway n-6 pathway Alpha-linolenic acid (18:3) Linoleic acid (18:2) Δ-6-Desaturase Octadecatrienoic acid (18:4) Gamma-linolenic acid (18:3) Elongase Eicosatetraenoic acid (20:4) Dihomo-gamma-linolenic acid (20:3) Δ-5-Desaturase Eicosapentaenoic acid (20:5) Arachidonic acid (20:4) Elongase n-3 Docosapentaenoic acid (22:5) Adrenic Acid (22:4) Elongase, Δ-6-Desaturase, β-oxidation Docosahexaenoic acid (22:6) n-6 Docosapentaenoic acid (22:5) Adapted from 15 4

18 1.4. Brain uptake Since PUFA synthesis does not occur in a significant quantity in the brain 16, 17, PUFA must be taken up from the periphery. Several mechanisms have been proposed for the uptake of PUFA by the brain 18, including passive diffusion of fatty acids 19 or lysophospholipids 20, or by lipoprotein transporters 21. Previous studies have demonstrated, however, that knocking out lipoprotein receptors does not affect PUFA levels, suggesting these receptors are not necessary for maintaining PUFA concentration 22, 23. Lysophosphatidylcholine has been proposed as the preferred source of PUFA in the rodent brain. The major facilitator superfamily domain-containing protein 2 (Mfsd2a), a lysophospholipid transport protein, knockout mouse model, has decreased brain DHA concentration suggesting lysophosphatidylcholine as the major source of brain DHA 24. Studies injecting either radiolabelled lysophosphatidylcholine DHA or unesterified DHA have reported more radioactivity entering the brain when DHA is delivered in the lysophosphatidylcholine form 25, 26. However, one study has demonstrated that the brain is exposed to lower radioactivity levels when injected with the unesterified form compared to the lysophosphatidylcholine form. This is explained by the shorter plasma half-life of the unesterified form compared to the lysophosphatidylcholine form. When correcting for radioactivity exposure, the brain uptake is higher for the unesterified form 26. Moreover, brain uptake rates of unesterified DHA closely match brain DHA consumption, suggesting that the unesterified pool is the major source of brain DHA 27. The process of uptake is thought to be passive and not transport facilitated 28. Despite the passive nature of the brain uptake, however, there are certain differences in 5

19 concentration between PUFA. ARA and DHA are highly concentrated in the brain (10,000 nmol/g of brain) 29, 30, while alpha-linolenic acid, EPA and linoleic acid are found in much lower concentrations, approximately 500-fold less concentrated 31. The low levels of alpha-linolenic acid 17 and EPA 32, 33 and linoleic acid 16 are due to beta-oxidation upon entry into the brain, and other redundant mechanisms to maintain the levels of these fatty acids 18, N-3 PUFA and anti-inflammation It has been suggested that n-3 PUFA have anti-inflammatory properties 15. In vivo, for instance, n-3 PUFA reduce inflammation in multiple models including stroke, spinal cord injury, Alzheimer s disease, Parkinson s disease, lipopolysaccharide (LPS), interleukin (IL)-1, and others. This is reviewed in Chapter 2. The anti-inflammatory properties of n-3 PUFA appear to be driven by multiple mechanisms 15. In vitro evidence suggests that one mechanism may relate to the action of n-3 PUFA on the peroxisome proliferator-activated receptor (PPAR)- The downstream action of n-3 PUFA on the PPAR- leads to a down regulation of the nuclear factor kappa-light-chain-enhancer of activated B cells (NF- B), resulting in a decrease in cytokine production 34, 35 and a reduction of adhesion molecules on monocytes 36. In microglial culture, n-3 PUFA reduce microglial hypertrophy 37, cytokine production 38-41, NF- B signalling 40, 41, and induce polarization to the anti-inflammatory M2 phenotype 38, 39 following an inflammatory insult. In addition, increasing doses of DHA and EPA 6

20 Figure 1-2. PUFA-derived mediator synthesis pathway Adapted from15, HETE 15-LO ARA EPA DHA PDX 12-LO 12-HETE Cyp450 5-LO PD1 15-LO 17S- H(p)DHA 12-LO 14- H(p)DHA 5-LO LXA4 15-LO 5-LO LTA4 Cyp450 EET 18-HpEPE 5-LO 5-LO 5-HpEPE 5-LO 17S-HDHA MaR1 MaR2 RVE1 RVE2 RVE3 RVD1 RVD6 LTB4 LTC4 COX-2 20-HETE LTA5 5-HEPE RVD2 RVD3 RVD4 RVD5 PGD2 PGH2 TXA2 PGES PGE2 PGF2 PGI2 8-keto- PGF1 LTB5 ARA, arachidonic acid; COX, cyclooxygenase; DHA, docosahexaenoic acid; EET, epoxyeicosatrienoic acid; EPA, eicosapentaenoic acid; HDHA, hydroxy DHA; HEPE, hydroxy eicosapentaenoic acid; HETE, hydroxyeicosatetraenoic acid; H(p)DHA, hydroperoxy DHA; H(p)EPE, hydroperoxy eicosapentaenoic acid; LO, lipoxygenase; LT, leukotriene; LX, lipoxin; MaR, maresin; PD, protectin D; PG, prostaglandin; RV, resolvin 7

21 increase phagocytosis of amyloid- (A 39. DHA is also suggested to promote antiinflammatory properties through its binding to G-coupled receptor 120, which has been demonstrated to reduce cytokine production 43. Alternatively, it has been proposed that n-3 PUFA may exert their antiinflammatory properties through their conversion to specialized pro-resolving lipid mediators by lipoxygenases, including protectin D1, resolvins and maresins 42, 44. The synthesis pathways of these specialized pro-resolving lipid mediators are illustrated in Figure 1-2. In microglial culture, similar to DHA, specialized pro-resolving lipid mediators reduce cytokine production 45, increase A phagocytosis 46, reduce microglial marker expression 46, increase anti-inflammatory M2 phenotype 47 and increase neuronal survival 46. These specialized pro-resolving lipid mediators appear to work through different receptors than G-coupled receptor 120, with different specialized pro-resolving lipid mediators activating different receptors 48. These receptors include G-coupled receptor 18 (resolvin D2) 49, chemerin receptor 23 (resolvin E1) 50, G-coupled receptor 32 (resolvin D1, resolvin D3, resolvin D5) 51, 52, lipoxin A4 receptor (resolvin D1) 53, and leukotriene B4 receptor (resolvin E1) 50. N-3 PUFA are also thought to have indirect anti-inflammatory properties through their reduction of n-6 PUFA in the phospholipid membrane. N-6 PUFA, and more specifically ARA, are located in the sn-2 position of the phospholipid membrane and can be released by cytosolic calcium-dependent phospholipase A2. Once released by cytosolic calcium-dependent phospholipase A2, ARA can be metabolized by cyclooxygenase (COX) and lipoxygenase to form pro-inflammatory mediators such as prostaglandins 8

22 (PG), thromboxanes, hydroxyeicosatetraenoic acids and leukotrienes. The synthesis of these mediators is also represented in Figure 1-2. N-3 PUFA also occupy the sn-2 position in the phospholipid membrane. Therefore, by increasing n-3 PUFA concentration in the phospholipid membrane, n-6 PUFA are displaced which reduces the substrate availability for pro-inflammatory mediator production 15, 54. Moreover, both n-3 and n-6 PUFA use the same enzyme for mediator production. By increasing n-3 PUFA concentration, these enzymes become saturated which limits the production of proinflammatory mediators by producing more anti-inflammatory mediators Models of neuroinflammation Neuroinflammation is present in many common animal models, including transgenic models, traumatic brain injury models and models involving the injection of neurotoxic agents. A number of models of central nervous system disorders, for instance, have a neuroinflammatory component. For example, Alzheimer s disease is associated with neuroinflammation, and many of the Alzheimer s disease transgenic mouse models, including the 3xTg and APPsw mice, have a neuroinflammatory phenotype 55, 56. This includes microglial activation 57, 58 and the production of cytokines 59, 60. It is thought, however, that the neuroinflammation found in these models is secondary to the betaamyloid production 56. Similar observations are also seen in Parkinson s disease transgenic mice 61, 62. Neuroinflammation is also observed in traumatic injury models. In stroke models, the ischemia causes activation of microglia in the penumbra shortly after the induction of 9

23 ischemia-induced cell death 63. This induces secondary cell death through the release of pro-inflammatory cytokines, such as IL-1 and tumor necrosis factor (TNF) Similar increases in microglial activation and cytokine production are observed in traumatic brain injury 67, 68 and spinal cord injury 69 models. The inflammatory component of these models appears to be secondary to the injury itself, and to be induced by endogenously produced danger-associated molecular patterns secreted by damaged neurons 70. Breakdown of the blood-brain barrier (BBB) also occurs following traumatic injury, which allows for the movement of inflammatory blood-born immune cells from the periphery into the brain 70, 71. This is important to consider when assessing neuroinflammation following trauma. Neurotoxins have also been linked to neuroinflammation. For example, injection of 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine results in the degeneration of dopaminergic neurons and is associated with increased microglial activation 72. Neurodegeneration in this model occurs independent of microglial activation 73, and neuroinflammation appears to be secondary to cell death 74. Other similar models include 6-hydroxydopamine and rotenone The lipopolysaccharide (LPS) model of neuroinflammation The LPS model is one of the most common models of neuroinflammation. LPS is a component the outer wall of Gram-negative bacteria. It causes inflammation through the activation of the Toll-like receptor 4 76, 77. These receptors are highly expressed on microglia, while not present on neurons, astrocytes or oligodendrocytes Activation of 10

24 the Toll-like receptor 4 directly activates the NF- B pathway, resulting in increased production of pro-inflammatory cytokines 81, 82. There are many variations of the LPS model involving different routes of administration, doses, and durations of infusion. Systemically, LPS has been administered through a variety of routes, including s.c., i.p., and i.v., and at doses ranging from to 200 mg/kg 83. Overall, LPS induces rapid microglial activation within 6 hours of injection, which can last up to 3 days 83, 84. By one week, microglial activation has usually returned to baseline 84, 85. However, some studies have reported microglial activation lasting months 86 following LPS injection, or even for up to one year 87. Following microglial activation, cytokine production and astrocyte activation are observed 88, 89. Chronic administration of LPS can result in less severe symptoms compared to acute administration 88. Systemic delivery of LPS results in peripheral inflammatory effects through the activation of the Toll-like receptor 4 on macrophages. This can result in increased circulating cytokines, which in turn can signal to the brain and increase BBB permeability 90. In order to focus on neuroinflammation, without the effects from the periphery, the present project focuses on neuroinflammation following the intracerebroventricular (i.c.v.) injection of LPS. This model has been shown to increase cytokine production, such as TNF- and IL-1, and microglial activation within hours to 1 day of injection Microglial activation persists up to at least 3 days 94, and has been shown to still be increased at 4 weeks in rats receiving 25 g LPS 96. A study using positron emission topography (PET) imaging has reported that microglial activation peaks at 3 days post injection 97. The inflammatory insult induced by i.c.v. LPS is 11

25 exacerbated in older (26 month old) rats 98. Memory appears to be affected in this model. Repeated administration of 50 g of LPS in Wistar rats results in decreased spatial memory in the Morris water maze 99. Wistar rats receiving 10 g of i.c.v. LPS show decreased spontaneous alternating behaviour in the Y-maze 100. Depressive-like behaviour in the forced swim test is also present 24 hr following the injection of 100 ng i.c.v. LPS 95, The fat-1 mouse The fat-1 mouse is a model developed by Dr. Jiang Kang at Harvard University 102. The fat-1 gene from c. elegans codes for a n-3 desaturase that converts n-6 to n-3 PUFA by adding a double bond at the third carbon from the methyl end of the fatty acid. The introduction of the fat-1 gene into a mouse allows for the endogenous conversion of n-3 PUFA from n-6 PUFA. When fed an n-3 PUFA deficient diet, fat-1 mice have higher cortical DHA and lower cortical ARA as compared to wildtype (WT) littermates 103, 104. In fact, fat-1 mice fed a diet deficient in n-3 PUFA have brain DHA concentrations similar to mice on an adequate diet 104. The amount of DHA that can accumulate in the brain appears to have a plateau, the fat-1 mice and WT littermates consuming fish oil have similar brain DHA concentrations 104. Dietary interventions, such as fish oil, have the potential for confounders, such as having increased vitamin D or the subtraction of n-6 PUFA for the introduction of n-3 PUFA. The fat-1 model allows us to isolate the potential effect of increased brain DHA on neuroinflammation. 12

26 The fat-1 mouse has reduced systemic inflammation in various disease models. The fat-1 mouse, for instance, displays a lower inflammatory phenotype, with decreased pro-inflammatory signals such TNF- and IL-6, in models of colitis 105, osteoporosis 106, ethanol-induced liver steatosis 107 and in the apoe knockout 108. Serum IL-6 is also reduced in the fat-1 mouse in a cerulein-induced pancreatis model 109. Similarly, expression of the NF- B pathway is decreased in the fat-1 mouse in the streptozotocin induced-diabetes model 110, in colon tumorigenesis 107 and in the fat-1/apoe knockout mouse fed a western diet 108. The fat-1 gene also appears to enhance the calorie-induced reduction of serum pro-inflammatory cytokines such as IL-1 and TNF Aside from increases in n-3 PUFA, bioactive mediators are also elevated in the fat-1 mouse in inflammatory models. Lipidomic analysis of fat-1 mouse plasma shows a specific anti-inflammatory signature, with increases in 17-hydroxy DHA, a precursor to resolvin D1 and resolvin D2, and decreases in pro-inflammatory mediators such as hydroxyeicosatetraenoic acid 112. Increases in 17-hydroxy DHA have also been found in the fat-1 mouse in a model liver carcinogenesis. This was correlated with decreases in serum TNF- and COX In an obesity-linked inflammation model, the fat-1 mouse has increased protectin D1 in muscle and adipose tissue, which is correlated with increased macrophage clearance and decreased cytokine production 114. In the brain, it has been reported that COX-2 is lower in fat-1 mice than in WT littermates, while no differences in cytosolic phospholipase A2 are found between the two groups 103. A microarray analysis has shown a decrease in calcium independent phospholipase A2 mrna expression in the fat-1 brain as compared to the brain of a WT mouse on an n-3 PUFA deficient diet 115. These two studies point to a potential lower 13

27 inflammatory environment in the fat-1 brain as compared to the WT littermate brain. Several studies have reported lower neuroinflammation in various disease models in the fat-1 mouse (reviewed in Chapter 2) Objectives The goal of this thesis was to define a self-resolving model of neuroinflammation following the i.c.v. injection of LPS. This model was to be defined based on protein expression, gene expression and lipidomic profile. Once the model was defined, this thesis was designed to evaluate whether n-3 PUFA could modulate the resolution of neuroinflammation. Chapter 2 is a review of the literature on the effect of n-3 PUFA on neuroinflammation. This chapter is adapted from a published article that summarizes all known studies which have measured the effect of n-3 PUFA on neuroinflammatory markers in a variety of models, including stroke, spinal cord injury, Alzheimer s disease, LPS injection, traumatic brain injury and others. This paper is published in the European Journal of Pharmacology. Chapter 3 reviews the objectives and hypotheses of this thesis. Chapter 4 and 5 are the experimental chapters of the thesis. Chapter 4 is a method paper that attempts to describe a novel method for measuring the brain neurolipidome. As PUFA-derived mediators are thought to regulate resolution of inflammation, and these mediators had yet to be measured in the brain without the artifact of ischemia, a novel method was developed combining high-energy head-focused microwave fixation with high-resolution lipidomics. 14

28 Chapter 5 develops a self-resolving model of neuroinflammation following i.c.v. LPS injection. It is based on inflammatory markers and lipidomic profile as measured by our new method to eliminate the ischemia-induce artifact on pro-resolving mediator production. Once resolution of neuroinflammation is established, Chapter 5 sets out to determine whether or not increasing brain DHA, utilizing either dietary approaches or the fat-1 transgenic model, has an effect on the resolution of neuroinflammation. Chapter 6 summarizes the findings of this thesis, along with the significance and implications for the field. The chapter concludes that ischemia induces a unique lipidomic signature which is inhibited by microwave fixation, and that increasing brain DHA only has subtle effects on the resolution of neuroinflammation. Limitations of the thesis are also discussed in this chapter. Appendix 1 is a systematic review of the literature on neuroinflammation in postmortem schizophrenia brains. Neuroinflammation has been found in most neurological disorders, and evidence is starting to suggest that neuroinflammation is associated with psychiatric disorders as well. In particular, in vivo imaging and clinical trials suggest a possible link between neuroinflammation and schizophrenia. Some postmortem studies have also demonstrated elevated neuroinflammation in schizophrenia. However, to date, a review of all of the postmortem data had not yet been published. This paper is published in Molecular Psychiatry. 15

29 Chapter 2: N-3 Polyunsaturated Fatty Acids in Animal Models with Neuroinflammation: An Update Adapted from: Marc-Olivier Trépanier, Kathryn E Hopperton, Sarah K. Orr and Richard P Bazinet. Eur J Pharmacol May 30 [DOI: /j.ejphar ] Contribution: Expanding on our previous article, I found all new published articles published since 2012, along with expanding the scope of research included. I extracted all information and wrote the first draft of the paper. 16

30 2.1. Abstract Neuroinflammation is a characteristic of a multitude of neurological and psychiatric disorders. Modulating inflammatory pathways offers a potential therapeutic target in these disorders. Omega-3 polyunsaturated fatty acids have anti-inflammatory and pro-resolving properties in the periphery, however, their effect on neuroinflammation has been less studied. This review summarizes 61 animal studies that have tested the effects of omega-3 polyunsaturated fatty acids on neuroinflammatory outcomes in vivo in various models including models of stroke, spinal cord injury, aging, Alzheimer s disease, Parkinson s disease, lipopolysaccharide injection, IL-1β injection, diabetes, neuropathic pain, traumatic brain injury, depression, surgically induced cognitive decline, whole body irradiation, amyotrophic lateral sclerosis, and lupus. The evidence presented in this review suggests that while there are anti-neuroinflammatory properties of omega-3 polyunsaturated fatty acids, it is not clear by which mechanism omega-3 polyunsaturated fatty acids exert their effects. Future research should aim to isolate the effects of omega-3 polyunsaturated fatty acids on neuroinflammatory signaling in vivo and to elucidate the mechanisms underlying these effects. 17

31 2.2. Introduction Inflammation is a characteristic of many neurological and psychiatric illnesses, including Alzheimer s disease, multiple sclerosis, depression, schizophrenia and Parkinson s disease 116, 117. While some inflammation is integral to pathogen and debris clearance, as well as to wound healing, excessive, dysregulated inflammation can exacerbate tissue injury 116, 118, 119. Indeed, inflammation has been suggested as a mechanism by which Alzheimer s and Parkinson s disease pathologies potentiate neuronal death 116, 120. The brain is an immunologically unique environment, and, as such, knowledge about inflammation and its resolution in the periphery may not apply directly to the brain 121. The brain is separated from the periphery by the blood-brain-barrier (BBB), and houses its own population of immune effector cells: astrocytes and microglia. Microglia are the macrophages of the brain, and, under normal conditions, exist in the M0 resting phenotype, surveying the neurological environment for insult or injury 122. Microglia can be activated from their resting M0 state to a M1 pro-inflammatory state by cytokines such as tumor necrosis factor-alpha (TNF- ) and interferon gamma (IFN-, produced either by the microglia themselves or by astrocytes, the major glial cells of the brain, in response to insult recognition 123, 124. Once activated, M1 microglia are characterized by the production of pro-inflammatory cytokines and chemokines, such as interleukin (IL)-6, IL-1β, IL-12, IFN-, IL-1α, and chemokine (c-x-c motif) ligand (CXCL) 11. Moreover, M1 microglia have increased activity of cyclooxygenase (COX)-2 and production of proinflammatory lipid mediators such as prostaglandin (PG) E2. They also exhibit increased production of reactive oxygen and nitrogen species via activity of inducible nitric oxide 18

32 synthase and NADPH oxidase 119, 122. Pro-inflammatory cytokines also activate astrocytes, which contribute to cytokine, reactive oxygen species and nitric oxide production 125. Exaggerated innate immune responses, or the failure to clear insults, can lead to excessive production of cytokines and reactive oxygen species by astrocytes and microglia, which triggers neuronal death by apoptosis or necrosis, feeding forward to 116, 122, further activate microglia by releasing ATP and calcium into the extracellular space 125. Upon neutralization of the initial insult and/or in response to cytokine IL-4 and chemokine (c-c motif) ligand (CCL) 2, M1 microglia switch to a M2 anti-inflammatory phenotype, promoting phagocytosis, wound healing and a return to homeostasis 123, 124. Despite the presence of the BBB, neuroinflammation can also be influenced by peripheral factors. Pro-inflammatory cytokines such as IL-1α, IL-1β, IL-6 and TNF-α have all been shown to cross the BBB, seemingly regulated by specific transporters 126. Permeability of the BBB to these factors increases under some neurological conditions, allowing peripheral macrophages, neutrophils and T cells to enter the brain 121, 122, 126, 127. Clearly, neuroinflammation is a distinct and complex process that results from interplay between a variety of cell types and mediators. As neuroinflammation has been implicated in the pathogenesis of various neurological disorders, there has been interest in the role of anti-inflammatory drugs for prevention and treatment. In human observational studies, the use of aspirin and other non-steroidal anti-inflammatory drugs is associated with a decreased risk of Alzheimer s disease, with longer-term users exhibiting the greatest risk reduction 128. Ibuprofen use is also associated with a decreased risk of Parkinson s disease, although aspirin and other non-steroidal anti-inflammatory drugs do not seem to exhibit the same protective 19

33 effects 129. Randomized clinical trials, on the other hand, generally do not support the positive effects of non-steroidal anti-inflammatory drugs in these neurological diseases. For instance, the only randomized control trial testing non-steroidal anti-inflammatory drugs in primary prevention of Alzheimer s disease found that neither celecoxib nor naproxen reduced the risk of Alzheimer s Disease onset, although this trial was stopped with an average of 15 months follow-up, well short of the target 7 years, due to concerns over increased cardiovascular risk with celecoxib treatment 130. Randomized clinical trials of anti-inflammatory drugs in patients with Alzheimer s disease or mild cognitive impairment have also generally failed to show any benefits 131, and, in some cases, have reported serious adverse events, with one trial finding that rofecoxib (selective COX-2 inhibitor) increased the risk of patients with mild cognitive impairment progressing to Alzheimer s disease 132. The evidence suggests that, although neuroinflammation is implicated in neurological disease, blocking inflammation may not be therapeutic. In animal models, blocking inflammation via reduced activity of microglia exacerbates acute neural injury to ischemia 133, and acute administration of exogenous activated microglia immediately following ischemia-reperfusion improves recovery 134. In mice, deletion or disruption of the COX-2 gene exacerbates the neuroinflammatory response to lipopolysaccharide (LPS) and fails to provide any benefit in models of Parkinson s disease and traumatic brain injury, while pharmaceutical COX-2 inhibitors have mixed effects in neuroinflammatory disease models 135. In a transgenic model of Alzheimer s disease, a mildly pyrogenic agonist of Toll-like receptor 4, a receptor on the surface of microglia, improved amyloid-β (A ) clearance and cognitive measures, while the much more potent 20

34 Toll-like receptor 4 agonist, LPS, did not 136. Thus, interventions that can modulate, as opposed to block, neuroinflammation may be a useful therapeutic approach. The resolution of inflammation was historically thought to be a passive process resulting from the dissipation of pro-inflammatory mediators 137. A novel class of molecules produced from the omega (n)-3 polyunsaturated fatty acids (PUFA) docosahexaenoic acid and eicosapentaenoic acid (DHA and EPA, respectively), collectively referred to as specialized pro-resolving lipid mediators, stimulate resolution, actively returning tissue to homeostasis following inflammation 42, 137. Specialized proresolving lipid mediators, comprised of the resolvin, protectin and maresin families, offer a potential mechanism for the protective effects of n-3 PUFA on neurological diseases that have been observed in animal models and human observational studies 117, 138, 139. The two main PUFA species in the brain are DHA, an n-3 PUFA, and arachidonic acid (ARA), an n-6 PUFA. DHA and ARA can be consumed pre-formed from the diet, or can be synthesized from dietary precursors, α-linolenic (n-3) or linoleic (n-6) fatty acids, primarily in the liver. While the brain expresses enzymes that can synthesize DHA and ARA from their dietary precursors, these synthesis rates appear to be much lower than the rate of brain PUFA uptake from the plasma, suggesting the brain is largely dependent on 16, 140, preformed DHA and ARA synthesized in the liver, or supplied directly from the diet 141. Brain lipid metabolism is a complex and evolving field (for review see 142 ). Briefly, upon entry into the brain, DHA and ARA are mostly esterified to phospholipids at the stereospecifically numbered-2 position. DHA and ARA are both released from the phospholipid membrane by phospholipase A2, with ARA preferentially cleaved by 21

35 calcium dependent cytosolic phospholipase A2, and DHA by calcium independent phospholipase A While over 90% of DHA and ARA released from the phospholipid membrane are rapidly re-esterified to phospholipids, a process known as the Lands cycle, a proportion of these unesterified fatty acids can be used as substrates for the synthesis of pro-inflammatory and pro-resolving lipid mediators 140. ARA and DHA are acted upon by COX and lipoxygenase enzymes, with ARA giving rise to pro-inflammatory mediators such as PG (notably PGE2) and leukotrienes and DHA giving rise to specialized proresolving lipid mediators 42. It is generally appreciated that n-3 PUFA have anti-inflammatory properties in the periphery 144, 145. The mechanism by which n-3 PUFA are anti-inflammatory, however, has yet to be determined. One suggested mechanism includes the increased availability of n-3 PUFA precursors for specialized pro-resolving mediator production. The potent actions of specialized pro-resolving lipid mediators have been studied in peripheral immune cells 42, 48. It has recently been shown that supplementation with fish oil for as little as 5 days produces significant increases in plasma levels of specialized proresolving lipid mediators and their precursors, suggesting that diet modifies the body s inflammatory environment 146. It is important to note, however, that there is variation between studies in the detection of specialized pro-resolving lipid mediators and their precursors at baseline and following n-3 PUFA supplementation While studies of specialized pro-resolving lipid mediators in the brain are limited, it is known from postmortem brain samples that patients with Alzheimer s disease have lower hippocampal levels of protectin D Likewise, lipoxin A4 and resolvin D1 levels in the cerebrospinal fluid are positively correlated to Mini-Mental State Examination 22

36 scores 151, which supports a protective role for these mediators in human neurological disease. In general, in vitro evidence points to immunomodulatory effects of DHA and EPA in immortalized microglia and astrocyte cultures, with lower levels of inflammatory markers in response to stimulation with IFN-γ, LPS or A 152, 153. EPA and DHA lower markers of M1 microglial activation and improve phagocytosis of A in microglia cultures, while DHA also increases M2 microglia markers, pointing to a pro-resolving effect 39. Cultures of human glial cells produce various DHA-derived specialized proresolving lipid mediators in response to stimulation, suggesting that specialized proresolving lipid mediators may play a role in the brain 154. Protectin D1 decreases A induced apoptosis in human neuronal cell cultures 150. In a co-culture of human neuronal and glial cells, protectin D1 decreases inflammatory markers COX-2 and TNF-α, increases peroxisome proliferator-activated receptor (PPAR) γ, and protects neurons from A -induced cell death 155. Together, these results support a role for n-3 PUFA and their associated specialized pro-resolving lipid mediators in modulating elements of the neuroinflammatory environment. Given the complexity of the interaction between different cell types in neuroinflammation, along with the potential modifying role of the peripheral immune system, animal models provide some advantages over cell culture models to study the interaction between dietary n-3 PUFA and inflammation in the brain. Diet is capable of changing the plasma concentrations of n-3 PUFA and specialized pro-resolving lipid mediators 146, and the levels of these components in the plasma are often used as a basis to select treatment doses in cell culture systems. It is not clear, however, how much diet, 23

37 particularly in the short term, can influence brain composition of n-3 PUFA and specialized pro-resolving lipid mediators, and thus how relevant these doses may be to the brain. Moreover, a recent paper that established an adult microglial signature based on expression of 239 genes found that two of the most commonly used microglial cell lines, Bv2 and N9, do not express this signature, bringing into question the generalizability of work with these and other cell lines to the brain 156. In this review, we will summarize the evidence for the role of n-3 PUFA in modulating neuroinflammation in animal models by updating and adding to our previous review, published in Results n-3 PUFA and neuroinflammation in ischemia or ischemia/reperfusion Inflammation is a key component of stroke injury. We identified 16 studies (summarized in Table 2-1) that have investigated the role of n-3 PUFA in controlling inflammation following ischemia and ischemia and reperfusion. Three studies have investigated chronic effects of n-3 PUFA. Lalancette-Hébert and colleagues utilized the Toll-like receptor 2-fluc-GFP transgenic mouse, a mouse that is transgenically modified to be bioluminescent under Toll-like receptor 2 activation, supplemented with DHA (0.7% of total diet weight) for 12 weeks. Compared to the low n-3 PUFA control, DHA supplementation decreases infarct size, microglial activation (as indicated by bioluminescence) and COX-2, IL-6 and IL-1 protein expression following 1 hr middle cerebral artery occlusion. This was correlated with increased striatal DHA

38 Table 2-1: Summary of studies investigating the effects of n-3 PUFA in ischemia and ischemia/reperfusion models Authors (year) Black et al. (1984) Marcheselli et al. (2003) Yang et al. (2007) Pan et al. (2009) Belayev et al. (2009) Zhang et al. (2010) Injury Model Species PUFA Treatment(s) Comparison Treatment Ischemia/ reperfusion Ischemia/ reperfusion Ischemia/ reperfusion Ischemia/ reperfusion Ischemia/ reperfusion Ischemia (immature Mongolian Gerbils C57Bl/6 mice Sprague Dawley rats Sprague Dawley rats Sprague Dawley rats Sprague Dawley a) mg EPA i.v. b) mg EPA i.v. a) μg 10,17Sdihydroxy-DHA i.c.v. b) μg DHA i.c.v per kg body weight i.p.: a) 250 nmol (71 μg) STA b) 500 nmol (142 μg) STA c) 250 nmol (76 μg) ARA d) 500 nmol (152 μg) ARA e) 250 nmol (82 μg) DHA f) 500 nmol (164 μg) DHA per kg body weight i.p.: a) 100 nmol (33 μg) DHA b) 500 nmol (164 μg) DHA mg LNA i.v. vehicle i.c.v. saline i.p. saline i.p. Treatment Duration/ Time point 135 min infusion 5 min prior ischemia 3 or 48 hr continuous infusion post-injury 60 min postreperfusion 1 h (single), 3 d (single), or 6 weeks # (daily) prior to ischemia 14 mg/kg DHA i.v. saline i.v. 60 min postreperfusion Chow + EPA + DHA (15mg/g of diet) Low n-3 (0.5%) From day 2 of pregnancy Brain n-3 PUFA Not reported Not reported Not reported Not reported Not reported CX total DHA and Non- Inflammatory Outcome a cerebral blow flow a,b brain edema b infarct size d,f infarct size b infarct size b BBB permeability b apoptosis b # oxidative stress b lipid peroxidation infarct size Sensorimotor Inflammatory Outcome a,b brain PGF2, PGE2, TXB2, 6-ketoPGF1 a,b CX and HIP leukocyte/neutrophil accumulation a,b HIP NF- B protein a,b HIP COX-2 mrna d,f CX leukocyte/neutrophil accumulation d,f CX COX-2 mrna b CX leukocyte/neutrophil accumulation b CX IL-6 protein CX GFAP protein CX COX-2, inos, TNF-α, IL- 1α, IL-1, IL-6 mrna 25

39 brain) rats (pups) to 7 days post surgery (PND 14) EPA score (FF) learning and memory (MWM) infarct size CX, ST Iba-1 protein Belayev et al. (2011) Lalancette- Hébert et al. (2011) Okabe et al. (2011) Bazan et al. (2012) Eady et al. (2012) Ischemia/ reperfusion Ischemia/ reperfusion Ischemia/ reperfusion Ischemia/ reperfusion Ischemia/ Reperfusion Sprague Dawley rats TLR2-fluc- GFP transgenic mice (C57Bl/6) Mongolian Gerbils Sprague Dawley rats Sprague Dawley rats 5 mg/kg DHA i.v. saline i.v. Time postischemia: a) 3h b) 4h c) 5h d) 6h DHA supplemented (0.7% n-3 PUFA total diet) low n-3 PUFA (0.03% n-3 PUFA) 500 mg/kg EPA i.p. Saline i.p. 4 weeks prior to surgery 333 g/kg AT-NPD1 i.v. saline i.v. 60 min postreperfusion 5 mg/kg DHA i.v. Saline i.v. 1 hr post reperfusion a PN NPD1 and 17-HDHA a,b,c infarct size a,b,c sensorimotor score (PRT, FPT) a CX and ST GFAP protein a CX and ST CD68 protein (microglia) 12 weeks ST DHA infarct size TLR2 promoter induction (microglial activation) brain IL-1, IL-6, COX-2 protein brain TNF- protein Not reported Not reported Not reported CA1 neuronal survival oxidative stress Learning and memory (8ARM) infarct size sensorimotor score (PRT, FPT) infarct size sensorimotor score (PRT, FPT) pakt HIP Iba-1 protein CX GFAP protein CX CD68 protein SCX GFAP and CD68 protein CX GFAP protein CX CD68 protein Eady et al. Ischemia/ Sprague Per kg body weight i.v.: Saline i.v. or 1 hr post Not a,b,c,d a,b,c,d CX, ST GFAP protein 26

40 (2012) Reperfusion Dawley rats a) 5 mg DHA b) 5 mg DHA g Alb c) 5 mg DHA g Alb d) 5 mg DHA g Alb 0.63 g/kg Alb reperfusion reported sensorimotor score (PRT, FPT) a,b,c,d infarct size a,b,c,d new neurons c,d CX, ST CD68 protein Chang et al. (2013) Ischemia Sprague Dawley rats Daily 500 nmol (164 g)/kg i.p. DHA Saline i.p. 3 days prior to surgery Not reported neurological score infarct size apoptotic signals oxidative stress CX CD68, CD45 (macrophage), Ly6g (neutrophil), CD3 (lymphocyte), CD11 (microglia) protein CX MPO activity CX TNF-α, IL-1, CCR2, IL- 6, MCP-1 mrna Eady et al. (2014) Luo et al. (2014) Ischemia/ Reperfusion in aged rats (18 mo) Ischemia/ reperfusion Sprague Dawley rats Fat-1 mice Per kg body weigh i.v.: a) 5 mg DHA b) 5 mg DHA g Alb Fat-1 mice were placed on 10% corn oil Saline or 0.63 g/kg Alb WT were placed on 10% corn oil 1 hr post reperfusion Not reported Not reported HIP total DHA and n-3 DPA HIP RvD1 (following ischemia) a,b sensorimotor score (PRT, FPT) b edema b infarct size a,b new neurons learning and memory (MWM) apoptosis cell death GPCR 120 expression a,b CX, b ST GFAP protein b CX, ST CD68 protein HIP NF-κB, TNF-, IL-1β, IL-6, MCP-1, GFAP, Iba-1 protein Zendedel et al. (2014) Ischemia/ reperfusion Wistar rats Per kg body weight i.v. 140 mg DHA mg EPA saline i.v. or Lipofundin MCT 1 and 12hr post ischemia Not reported hypoxic marker axonic, brain IL-1, TNF-, Arg1, NLRP3 mrna brain Trem2 mrna 27

41 dendritic marker neurological score infarct size 17-HDHA, 17-hydroxy-DHA (DHA derivative); 8ARM, 8-arm radial maze, Aβ, amyloid-β; Alb, albumin; pakt, phosphorylated protein kinase B; ARA, arachidonic acid; AT-NPD1, aspirin triggered NPD1; BBB, blood brain barrier; C-C motif chemokine; CD, cluster of differentiation; COX, cyclooxygenase; CX, cortex; DHA, docosahexaenoic acid; DPAn-3, docosapentaenoic acid; EPA, eicosapentaenoic acid; FF, foot fault/grid walking; FPT, forelimb placing test; GFAP, glial fibrillary acidic protein; HIP, hippocampus; IL, interleukin; inos, nitric oxide synthase; LNA, linoleic acid; Ly6G, lymphocyte antigen 6; MCP, monocyte chemotactic protein; MPO, myeloperoxidase; MWM, Morris water maze; NF- B, nuclear factor-κb; NLRP3, NLR family pyrin domain containing 3; NPD, neuroprotectin D; PG, prostaglandin; PN, penumbra; PND, post natal day; PRT, postural reflex test; PUFA, polyunsaturated fatty acid; STA, stearic acid; ST, striatum; TLR2, Toll-like receptor 2; TNF, tumor necrosis factor; TX, thromboxane; WT, wildtype; a,b,c,d,e,f, indicates treatment group represented in outcome columns (brain n-3 PUFA, primary outcome, inflammatory outcome) 28

42 experience a lower level of microglial activation and expression of COX-2, TNF-, and IL-1 mrna following ischemic brain injury 159. Finally, Luo et al. utilized the fat-1 transgenic mouse to test the chronic effect of DHA on inflammation following ischemia/reperfusion. The fat-1 mouse endogenously converts n-6 to n-3 PUFA, leading to high brain DHA concentrations 104. Following 20 min occlusion of the common carotid artery, the fat-1 mouse had reduced hippocampal TNF-, IL-1, glial fibrillary acidic protein (GFAP), and ionized calcium-binding adaptor molecule 1 (Iba-1) protein levels compared to its wildtype control after 7 days of reperfusion 160. Sub-chronic studies have replicated the anti-inflammatory effects of n-3 PUFA observed in chronic exposure. Rats given daily DHA injections (500 nmol [164 g]/kg/day i.p.) for 3 days prior to brain ischemia have diminished infarct sizes at 3 days post-ischemia (without reperfusion), accompanied by reduced microglial markers, neutrophil and macrophage infiltration, and lower TNF-, IL-1 and IL-6 mrna levels 161. Daily EPA injections (500 mg/kg i.p., 4 weeks) prior to ischemia/reperfusion also result in neuroprotection in gerbils, increasing neuronal survival in the hippocampus and reducing microglial activation 162. Similarly, 6 weeks of daily DHA injections (500 nmol [164 g]/kg/day i.p.) prior to ischemia/reperfusion injury lowers infarct size, while also reducing leukocyte infiltration and IL-6 levels compared to control 163. This effect is dose-dependent, as 100 nmol (33 g)/kg/day DHA is ineffective at lowering leukocyte infiltration and IL-6 protein. Similar results were reported with single acute injections of 100 nmol (33 g) and 500 nmol (164 g)/kg DHA either 1 hr or 3 days prior to the ischemic injury

43 Six acute studies have demonstrated similar anti-inflammatory effects of n-3 PUFA on ischemia/reperfusion models, even when administered post-occlusion. Three separate studies report that i.v. injection of DHA (5 to 14 mg) 1 hr after 2 hr of ischemia in rats results in a decreased infarct size 7 days post reperfusion and improves neurological scores, while increasing GFAP protein expression, an astrocytic marker, in the cortex This is also accompanied by decreases in CD68 protein, a marker of microglia and macrophages 165, 166. A fourth study found that i.v. infusions (1hr and 12hr post occlusion) of 21.5 mg/kg or 32.5 mg/kg of DHA and EPA, respectively, attenuates ischemia/reperfusion-induced IL-1, TNF-, and nucleotide binding domain and leucine rich containing protein 3, a protein involved in IL-1 processing, increased expression. Treatment also reduced the microglial M1 phenotype mrna marker Arg Complexing DHA to albumin appears to have an additive effect, where infusion of 5 mg of DHA complexed with 0.63g of albumin causes a greater decrease in the microglia/macrophage marker CD68 than DHA alone in both young 168 and aged rats 169. The mechanism by which n-3 PUFA may provide protection in these models is not agreed upon. One suggested mechanism is through the enzymatic conversion of n-3 PUFA to specialized pro-resolving lipid mediators, including resolvins and protectins 42. Within the mouse brain, ischemia/reperfusion injury induces resolvin D1 production, with higher levels in the fat-1 mouse, which is protected against ischemia/reperfusion injury, compared to its wildtype littermate 160. Further, acute injection of 5 mg of DHA 3 hrs post-ischemia increases production of protectin D1 and its precursor compared to the saline control 165. Direct i.v. infusion of 333 g/kg of the stereoisomer of protectin D1, aspirin-triggered protectin D1, 1 hr post-reperfusion produces similar anti-inflammatory 30

44 effects as previous n-3 PUFA studies, increasing GFAP and reducing CD68 protein while reducing the infarct size 170. Not all studies, however, show anti-inflammatory effects of n-3 PUFA in ischemic and ischemia/reperfusion models. Black and colleagues reported that 135 min of i.v. infusion of 833 g EPA 5 min prior to ischemia does not reduce the concentrations of pro-inflammatory mediators including PGE Moreover, a second study reported an increase in infarct size following administration of 500 nmol (164 g) of DHA i.p. 1 hr after reperfusion along with increased leukocyte infiltration and COX-2 mrna expression 24 hr post reperfusion 172. Similar results were obtained following injection of ARA, but not stearic acid 172. The authors argue that the injection of PUFA, including DHA and ARA, results in increased oxidative damage following ischemic injuries. When looking at all the studies together, evidence appears to point to a neuroprotective effect of endogenously synthesized n-3 PUFA, dietary n-3 PUFA, or n-3 PUFA injection to reduce the inflammation related to animal models of ischemia and ischemia/reperfusion injury. There are contradicting results regarding the effects of postischemia treatment, as one study 172 suggests possible damaging effects of n-3 PUFA n-3 PUFA and neuroinflammation in spinal cord injury Microglia and astrocyte activation is a major component of the pathophysiology of spinal cord injury. During spinal cord injury, additional immune cells, leukocytes, are recruited from the blood to the site of the injury where they release various proinflammatory lipid mediators and cytokines, exacerbating the innate inflammatory 31

45 response that leads to extensive tissue damage and potentially contributes to loss of function 173. Nine studies (Table 2-2) have evaluated the outcome of intravenous n-3 PUFA administration either before or following spinal cord injury on neuroinflammatory markers. It was reported that 250 nmol (82 g)/kg of DHA administered i.v. 30 min after spinal hemisection reduces lesion size, and increases neuronal survival and motor recovery, despite its lack of effect on CD68 protein levels 174. In contrast, injection of 250 nmol (76 g)/kg ARA in the same model exacerbates neuroinflammation and decreases cell viability 174. The lack of anti-inflammatory effect of acute DHA is in agreement with results reported by 2 separate studies, which find that i.v. injection of 250 nmol (76 or 82 g)/kg of either EPA or DHA 30 min post spinal compression does not decrease protein concentrations of TNF-, IL-1 and IL-6 175, 176. However, Hall et al. did observe a decrease in JT1, a marker of neutrophil infiltration, following DHA administration 175. Similar to the report of King and colleagues 174, Lim et al. show EPA administration increases cell survival and motor recovery even though there is no effect on a marker of neuroinflammation 176. While the studies above reported no effect of n-3 PUFA administered following spinal injury on inflammatory markers, other studies show reduced pro-inflammatory markers following i.v. infusion of n-3 PUFA. In separate studies, i.v. injection of 250 nmol (82 g)/kg DHA following spine compression reduced COX-2 69, GFAP, TNF- and nuclear factor kappa-light-chain-enhancer of activated B cells (NF- B) protein concentrations 177, while also decreasing activated microglial markers CD68 and Iba-1 69,

46 Table 2-2: Summary of studies investigating the effects of n-3 PUFA in spinal cord injury models Authors (year) Lang- Lazdunski et al. (2003) King et al. (2006) Huang et al. (2007) Injury Model Species PUFA Treatment(s) Comparison Treatment Ischemia SCI Hemisection SCI Compression SCI Sprague- Dawley rats Wistar rats Sprague Dawley rats 250 nmol (70 g)/kg ALA i.v. Per kg body weight i.v.: a) 250 nmol (82 g) DHA b) 250 nmol (70 g) ALA c) 250 nmol (76 g) ARA a) 250 nmol (82 g)/kg DHA i.v.+ control diet b) 250 nmol (82 g)/kg DHA i.v mg/kg/d Vehicle i.v. Vehicle i.v. 250 nmol/kg OA i.v. Saline i.v. + control diet Treatment Duration/ Time point 30min pre surgery or immediately following surgery Acute i.v. 30 min post surgery Acute i.v. 30 min post surgery 1 or 6 weeks Brain n-3 PUFA Not reported Not reported Not reported Non- Inflammatory Outcome neurological outcome apoptosis neuronal survival a,b lesion size c lesion size a,b apoptosis c apoptosis a,b neuronal survival c neuronal survival a,b oligodendrocyte survival c oligodendrocyte survival a,b RNA oxidation c RNA oxidation a,b motor recovery c motor recovery a,b spinal neuronal survival a,b spinal Inflammatory Outcome spinal NF-κB protein a,b spinal CD68 protein c spinal CD68 protein a,b spinal CD68 protein 33

47 DHA/EPA p.o. diet post surgery oligodendrocyte survival a,b spinal neuron injury a,b motor recovery a,b RNA oxidation Lim et al. (2010) Figueroa et al. (2012) Hall et al. (2012) Compression SCI Compression SCI Compression SCI Compression SCI Sprague Dawley rats Sprague- Dawley rats Sprague Dawley rats Sprague Dawley rats 250 nmol (82 g)/kg DHA i.v. 250 nmol (76 g)/kg EPA i.v. 250 nmol (82 g)/kg DHA i.v. Per kg body weight i.v.: a) 250 nmol (82 g) DHA b) 250 nmol (76 g) EPA Saline i.v. Acute i.v. 30 min post surgery Saline i.v. Acute i.v. 30 min post surgery vehicle i.v. 1 hr and 1 week prior to injury Vehicle i.v. Acute i.v. 30 min post surgery Not reported Not reported Not reported Not reported spinal lipid peroxidation spinal protein oxidation spinal neuronal survival spinal oligodendrocyte spinal neuron injury motor recovery Motor recovery axonal conductance myelin and axonal integrity cell death a,b hepatic neutrophil a plasma CRP spinal COX-2 protein spinal CD68 protein spinal GFAP, CD68 (macrophage), CD11 (microglia) protein a ventral horn JTI protein (neutrophil marker) 24hr post injury and ventrolateral white matter JTI 4hr post injury a,b spinal IL-6, IL-1β, TNF-α, 34

48 Lim et al. (2013) Lim et al. (2013) Compression SCI Compression SCI Fat-1 mice Fat-1 on 10% corn oil a) WT littermate on 10% corn oil b) WT littermate on n-3 PUFA adequate diet C57Bl/6 mice a) 500 nmol (164 g)/kg DHA i.v.+ control diet b) saline i.v mg/kg/d DHA/EPA p.o. c) 500 nmol (164 g)/kg DHA i.v mg/kg/d DHA/EPA p.o. Vehicle i.v. + control diet 12 weeks spinal PL DHA Acute i.v. 30 min post surgery 4 weeks diet post surgery Not reported cell survival Motor recovery a,c oligodendrocyte survival a,c neuronal survival a,c Motor recovery KC/GRO/CINC protein spinal Iba-1 protein spinal IL-6 protein (vs. a only) spinal IL-1β protein a,c dorsal horn Iba-1 protein a,b,c ventral horn Iba-1 protein Paterniti et al. (2014) Compression SCI CD1 mice 250 nmol (82 g)/kg DHA i.v. Saline i.v. 30 min following injury Daily injection for 9 days for motor testing Not reported Histological damage Motor recovery apoptosis spinal IκB-a, protein spinal NF-κB, GFAP, TNF-α, Iba-1, inos, nitrotyrosine protein ALA, α-linolenic acid; Alb, albumin; ARA, arachidonic acid; CD, cluster of differentiation; COX, cyclooxygenase; CRP, c-reactive protein; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; GFAP, glial fibrillary acidic protein; IkB-a, nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor alpha; IL, interleukin; INOS, nitric oxide synthase; NF- B, nuclear factor-κb; OA, oleic acid; PL, phospholipid; PUFA, polyunsaturated fatty acid; SCI, spinal cord injury; TNF, tumor necrosis factor; WT, wildtype; a,b,c indicates treatment group represented in outcome columns (brain n-3 PUFA, primary outcome, inflammatory outcome) 35

49 The effect of chronic oral administration of n-3 PUFA following spinal compression was assessed in two studies, with mixed results. Huang and colleagues reported that 400 mg/kg of DHA p.o. for 4 weeks following injury, in combination with an acute injection of 250 nmol (82 g)/kg of DHA i.v. 30 minute after the injury reduces microglial markers compared to control and appears to have an additive effect compared to injection alone 69. A second study, however, found that oral intake of 400 mg/kg of DHA alone for 4 weeks only reduces microglial activation in the ventral horn of the spinal cord with no effect on motor control, while injection of 500 nmol (164 g)/kg of DHA i.v. alone or in combination with oral DHA supplementation reduces microglial activation in both ventral and dorsal horns and increases motor recovery 178. Three studies have evaluated n-3 PUFA administration prior to spinal cord injury. While acute injection of 250 nmol (82 g)/kg of DHA in rats either 1 hr or 1 week prior to spinal compression had no effect on GFAP, CD68 and CD11 179, 250 nmol (70 g)/kg of alpha-linolenic acid i.v. 30 minute prior to spinal cord ischemia appeared to decrease NF- B staining 180. The transgenic fat-1 mouse, with higher brain DHA, had reduced spinal cord injury-induced Iba-1 and IL-6 protein increase compared to wildtype littermates 181. Taking all of these studies together, it is difficult to reach a conclusion on the antiinflammatory properties of n-3 PUFA in spinal cord injury. n-3 PUFA have variable effects on astroglial and microglial markers in spinal cord injury, despite the fact that n-3 PUFA appear to increase motor control recovery and cell survival in these models. 36

50 n-3 PUFA and neuroinflammation in aging Aging is associated with cognitive decline, as well as activated microglia 182 and reactive astrocytes 183, which release pro-inflammatory cytokines. Alzheimer s disease is associated with similar neuroinflammatory markers, as well as neuronal loss and accumulation of A plaques and neurofibrillary tangles 116. There are 9 studies evaluating the effects of n-3 PUFA on neuroinflammation induced by aging or Alzheimer pathology (Table 2-3). When comparing young vs. aged rats, 3 weeks of dietary supplementation of 10 and then 20 mg/kg of ethyl-epa in aged rats reduces the IL-1 protein concentration to levels present in young rats 184, 185, while also elevating the anti-inflammatory cytokine IL- 4 in the cortex 184. This is in agreement with the observation that supplementation of 125 mg/day of ethyl-epa for 4 weeks lowers CD40, IL-1 and IFN- 186 while elevating IL and PPAR 187 in the hippocampus of aged rats. Similarly, chronic (8 week) dietary tuna oil composed of 0.55% EPA and 0.36% DHA (% of total diet weight), prevents ageinduced elevation of hippocampal TNF- and monocytic marker CD11b protein levels, whereas GFAP and IL-1 increase despite supplementation 188. A separate study demonstrates that supplementing aged mice with 200 mg/kg/day of EPA elevates cortical DHA, EPA and n-3 docosapentaenoic acid, while n-3 docosapentaenoic acid supplementation only raises cortical n-3 docosapentaenoic acid. Despite this difference, both treatments decrease levels of major histocompatibility complex (MHC) II, a protein found on antigen presenting cells, in the hippocampus and cortex of n-3 PUFA supplemented compared to control chow groups 189. Moranis et al., however, report no effect of an n-3 PUFA adequate diet consisting of alpha-linolenic acid in aged mice on 37

51 Table 2-3: Summary of studies investigating the effects of n-3 PUFA in aging and Alzheimer s disease models Authors (year) Martin et al. (2002) Maher et al. (2004) Lynch et al. (2007) Minogue et al. (2007) Kelly et al. (2011) Injury Model Species PUFA Treatment(s) Comparison Treatment Aging (4 vs. 22 mo) Aging (4 vs. 22 mo) Aging (4 vs. 22 mo) A (i.c.v.) in aged rats (22 mo) Wistar rats Wistar rats Wistar rats Wistar rats chow supplemented with 10mg/d then 20mg/d eepa chow supplemented with 10mg/d then 20mg/d eepa chow supplemented with 125mg/d eepa chow supplemented with 125mg/d eepa Aβ (i.c.v.) Wistar rats chow supplemented with 125mg/d eepa Aging (3 vs. 22 mo) Aging (4 mo vs. 20 mo) Wistar rats Rats (Strain unspecifie d) chow supplemented with 125mg/d eepa a) Chow + EPA (200 mg/kg/d) b) chow + DPAn-3 (200 mg/kg/d) Treatment Duration/ Time point chow 3 weeks (10 mg/day) + 5 weeks (20 mg/day) chow 3 weeks (10 mg/day) + 5 weeks (20 mg/day) chow supplemente d with MUFA (isocaloric) chow supplemente d with MUFA (isocaloric) chow supplemente d with MUFA (isocaloric) chow supplemente d with MUFA (isocaloric) Chow + MUFA Brain n-3 PUFA Not reported Not reported 4 weeks Not reported 4 weeks Not reported 4 weeks Not reported 4 weeks Not reported 8 weeks a CX total DHA a CX total EPA a,b CX total Non- Inflammatory Outcome LTP apoptotic markers apoptotic markers neurotrophic factors LTP LTP a,b LTP a,b learning and memory (MWM) a,b apoptotic markers Inflammatory Outcome CX and HIP IL-1 CX IL-1 protein CX IL-1RI protein CX IL-4 protein HIP MHCII and CD40 (microglial activation) HIP IL-1, IFN- protein HIP IL-1 mrna HIP IL-4 protein and mrna HIP IL-1 protein HIP IFN-, IL-1 protein HIP PPAR protein HIP PPAR protein a,b CX MHCII protein a,b HIP MHCII mrna 38

52 Lebbadi et al. (2011) Moranis et al. (2012) Labrousse et al. (2012) Parrott et al. (2015) 3xTg-AD mice (12 vs. 20 *mo) Aging (3-5* vs mo) Aging (3 mo vs. 22 mo) TgCRND8 mice 3xTg- AD/Fat-1 mice CD1 mice C57Bl/6 mice TgCRND8 mice 3xTg-AD/Fat-1 on low n- 3 diet n-3 adequate diet (fatty acids 10.7% LA, 1.6% ALA) Control + Tuna oil (0.55% EPA,.36% DHA of total diet weight) Whole food diet containing 0.246% DHA (total diet weight) 3xTg- AD/WT on low n-3 diet n-3 deficient diet (0.1% ALA of fatty acids) Rapeseed oil, high oleic sunflower oil and palm oil (0.08% ALA of total diet weight) DPAn-3 18 month CX total DHA n-3/n-6 ratio 3-5 months CX DHA or months 8 weeks DHA, EPA a,b oxidative stress * in some AD markers depressivebehaviour (FST) * spatial memory (YM) age-induced spatial memory loss (YM) spatial memory (YM) object recognition DG c-fos, Corn oil 27 weeks spatial memory (MWM) problem solving caudate nucleus task A burden * CX GFAP protein CX IL-6 or IL-10 protein HIP CD11b, IL-6, TNF-α mrna HIP GFAP, IL-1 mrna HIP astrocyte process length HIP TNF-α mrna HIP GFAP mrna ALA, α-linolenic acid; A, amyloid beta; AD, Alzheimer s disease; CD, cluster of differentiation; DHA, docosahexaenoic acid; DPAn-3, docosapentaenoic acid; EPA, eicosapentaenoic acid; eepa, ethyl-epa; FST, forced swim test; GFAP, glial fibrillary acidic protein; HIP, hippocampus; IFN, interferon; IL, interleukin; LA, linoleic acid; LTP, long term potentiation; MHCII, major histocompability complex II; MWM, Morris water maze; MUFA, monounsaturated fatty acid; PPARγ, peroxisome proliferator activated protein γ; PUFA, polyunsaturated fatty acids; TNF, tumor necrosis factor; WT, wildtype; YM, y maze a,b,,* indicates treatment group represented in outcome columns (brain n-3 PUFA, primary outcome, inflammatory outcome) 39

53 levels of pro and anti-inflammatory cytokines (IL-6 and IL-10 respectively) and ageinduced memory deficits, despite the fact that the n-3 PUFA adequate diet increases cortical DHA and decreases depressive behaviour compared to an n-3 PUFA deficient diet 190. When challenged with A i.c.v., aged mice supplemented with 125 mg/kg ethyl- EPA have lower IL and higher PPAR compared to those on a control diet 187. Twenty month old 3xTg-AD mice, a transgenic mouse model of Alzheimer s disease, have a reduction in GFAP protein in the parieto-temporal cortex when crossed with the fat-1 mouse 191. Administration of n-3 PUFA has not always yielded positive results in Alzheimer s disease models. When supplemented with 0.246% DHA (% of total diet weight) for 27 weeks, the TgCRND8 transgenic mouse, which overexpresses two mutated forms of the amyloid precursor protein gene, demonstrates poor spatial memory in the Morris water maze and elevated TNF- gene expression in the hippocampus compared to a mouse receiving corn oil. It should be noted that DHA was delivered in a whole food diet, which also contained vitamins and phytochemicals n-3 PUFA and neuroinflammation in Parkinson s disease Parkinson s disease has a neuroinflammatory component, with evidence of activated microglia, and high pro-inflammatory cytokine and NF B levels in both postmortem human samples and in vivo animal models 193. n-3 PUFA may target neuroinflammation in Parkinson s disease models, along with other potential mechanisms, including oxidative stress and increased neurotrophic factors

54 Six studies were identified that investigate the effects of n-3 PUFA on neuroinflammation in Parkinson s disease models (Table 2-4). When supplementing mice with a diet containing 0.8% ethyl-epa (% of total diet weight), Luchtman et al. observed a reduction in s.c. 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine induced increases in striatal TNF- and IFN- protein. Midbrain IL-10 was also reduced by ethyl-epa treatment, while expression of COX-2 and calcium dependent cytosolic phospholipase A2, enzymes involved in inflammatory signaling, were unaffected 195. Similarly, Meng and colleagues found no difference in striatal COX-2 or calcium dependent cytosolic phospholipase A2 mrna expression following 6 weeks of 0.8% ethyl EPA prior to i.c.v. injection of 1- methyl-4-phenylpyridinium, the active metabolite of 1-methyl-4-phenyl-1,2,3,6- tetrahydropyridine. Both studies achieved increases in brain EPA, but not brain DHA with s.c. ethyl-epa 195, 196. In a third study, fat-1 transgenic mice with raised cortical DHA, have lower levels of the astrocytic marker GFAP compared to their wildtype littermates after 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine-induced injury 197. Injecting LPS directly into the substantia nigra causes dopaminergic neuron injury and a neuroinflammatory response similar to Parkinson s disease. Feeding a diet containing 15% fish oil (% of total diet weight) to Sprague Dawley rats for 2 weeks minimizes dopaminergic injury, while also reducing monocytic marker OX42 (also known as CD11b), TNF-α and IL-1β protein 198. In the last study, the authors evaluated n- 3 PUFA supplementation in A53T α synuclein transgenic mice, a transgenic model of Parkinson s disease expressing mild symptoms. Supplementation with a diet containing 13% n-3 PUFA of total fatty acids does not affect lectin, a microglial marker, 41

55 Table 2-4: Summary of studies investigating the effects of n-3 PUFA in Parkinson s disease models Authors (year) Meng et al. (2010) Muntane et al. (2010) Bousquet et al. (2011) Luchtman et al. (2012) Ji et al. (2012) Injury Model Species PUFA Treatment(s) Comparison Treatment MPP + A53T α - synuclein transgenic mice MPTP (i.p.) MPTP-P (s.c.) SN LPS (i.c.v.) C57Bl/6 mice A53T α - synuclein transgenic mice Fat-1 transgenic mice (C57Bl/6 x C3H) C57Bl/6 mice Sprague Dawley rats Chow + 0.8% eepa Chow + 0.8% palm oil a) Low n % DHA (13% of fatty acid n-3) b) Low n-3 (0.9% of fatty acid n-3) fat-1 transgenic mice on high n-6/low n-3 diet (101.79:1 n6/n3 ratio) Control (8% of fatty acid n-3) wildtype littermates on high n- 6/low n-3 diet (101.79:1 n6/n3 ratio) chow + 0.8% eepa chow + 0.8% palm oil 15% fish oil diet (30% fish oil as EPA and DHA) 15% corn oil diet Treatment Duration/ Time point Brain n-3 PUFA 6 weeks ST/FCX Total EPA and DPA ST/FCX Total DHA 6 months from 6 months of age a total DHA and EPA Non- Inflammatory Outcome ST, FCX^ DA ST Bcl-2 mrna ST Bax, Caspase-3 mrna a,b oxidative stress a,b α- synuclein 6 months CX DHA striatal or nigral dopaminergic injury Ψ 6 weeks CX EPA, DPA n-3 CX DHA 2 weeks Not reported hypokinesia and anxiety (RT, PT, OF) learning and memory (MWM) dopaminergic injury nigral dopaminergic neuron degeneration Inflammatory Outcome ST cpla2 and COX-2 mrna a,b CX lectin (microglia) and GFAP CX GFAP striatal TNF-, IFN-γ protein midbrain IL-10 protein striatal cpla2, COX-2 mrna SN CD11b, TNF-, IL-1 p65 (NF- B subunit) protein Zhang et al. SN LPS (i.c.v.) Sprague Per kg body weight Not reported 3 days prior Not b,c rotational a,b,c SN CD11b protein 42

56 (2015) Dawley rats (route not specified): a) 25 g RvD2 b) 50 g RvD2 c) 100 g RvD2 to LPS and 27 days post reported behaviour (RT) b,c dopaminergic neurons COX; cyclooxygenase; CX, cortex; DA; dopamine; DPA, docosapentaenoic acid; DHA, docosahexaenoic acid; eepa, ethyl-epa; EPA; eicosapentaenoic acid; FCX; frontal cortex; GFAP, glial fibrillary acidic protein; IFN, interferon; IL, interleukin; LPS, lipopolysaccharide; MPP+, 1-methyl-4-phenylpyridinium; MPTP, 1- methyl-4-phenyl-1,2,3,6-tetrahydropyridine; MWM; morris water maze; NF- B, nuclear factor kappa light chain enhancer of activated B cell OF, open field; PLA, phospholipase; PT; pole test; PUFA, polyunsaturated fatty acids; RT; rotorod test; Rv, resolvin: SN, substantial nigra; ST, striatum; TNF, tumor necrosis factor a,b, indicates treatment group represented in outcome columns (brain n-3 PUFA, non-inflammatory outcome, inflammatory outcome) EPA lowered COX-2 mrna compared to palm oil in saline (control) injected animals Ψ protection from nigral dopaminergic injury was correlated to brain DHA levels (secondary analysis) ^EPA increased striatal dopamine in saline injected group 43

57 or GFAP-positive cell counts 199. Injecting 25, 50, and 100 g/kg of resolvin D2 (route of administration not specified) for 3 days prior to LPS injection followed by 27 daily injections decreased the LPS-induced increases in CD11b protein. The 25 g/kg dose, however, was ineffective at reducing apomorphine induced rotational behaviour 200. Overall, the supplementation of n-3 PUFA does not appear to affect the arachidonic cascade in Parkinson s disease models. It is does appear, however, that n-3 PUFA does reduce cytokine production and may reduce astrocyte and microglial activation n-3 PUFA and neuroinflammation with lipopolysaccharide There are five studies (Table 2-5) that evaluated the effects of n-3 PUFA on the in vivo brain response to LPS. Four of these studies administered LPS peripherally by i.p. injection. Kavanagh et al. reported that mice receiving 50 mg/day of either ethyl EPA, ethyl gamma linolenic acid (GLA) or a combination of both fatty acids for 4 weeks were protected from LPS-induced (100 g/kg i.p.) decreases in anti-inflammatory cytokines IL-4 and IL-10 in the hippocampus, while only ethyl-epa and ethyl-epa + ethyl-gamma linolenic acid attenuated the decrease in PPAR protein. None of the treatments reduced hippocampal IL-1 protein concentration 201. The lack of decrease of IL-1 from n-3 PUFA administration in this study agrees with the observation that an n-3 PUFA + n-6 PUFA diet (6% total weight made of rapeseed and peanut) fed to dams from gestation through to 8 weeks postnatal does not attenuate hippocampal IL-6 mrna response of pups to 30 mg/kg LPS i.p. compared to an n-6 PUFA-only diet (6% peanut) 202. A separate study, however, reported that 500 mg/day ethyl-epa for 4 weeks, decreases 44

58 Table 2-5: Summary of studies investigating the effects of n-3 PUFA in lipopolysaccharide models Authors (year) Kavanagh et al. (2004) Lonergan et al. (2004) Mingam et al. (2008) Injury Model Species PUFA Treatment(s) Comparison Treatment Systemic LPS (i.p.) Systemic LPS (i.p.) Systemic LPS (i.p.) Wistar rats Wistar rats CD1 mice Orr et al. LPS i.c.v. C57Bl/6 (2012) mice LPS i.c.v. C57Bl/6 mice a) chow +50 mg/d eepa b) chow + 50 mg/d egla c) chow + 50 mg/d eepa and egla chow supplemented with 500 mg/d eepa 6% peanut + rapeseed oil (n-6 + n-3) diet 8% safflower + 2% fish oil a) 40 μg i.c.v. DHA b) 1 μg i.c.v.17s-hpdha Treatment Duration/ Time point Brain n-3 PUFA chow 4 weeks Not reported chow 4 weeks Not reported 6% peanut oil (n-6) diet 10% safflower diet acsf i.c.v. Gestation + CX DHA 8 weeks post-natal 9 weeks HIP total DHA HIP FFA DHA 24 hr infusion post surgery b HIP NPD1 Non- Inflammatory Outcome a,b,c LTP LTP apoptotic markers social interaction food intake Inflammatory Outcome a,b,c HIP IL-10, IL-4 protein a,c HIP PPAR protein a,b,c HIP IL-1β protein HIP IL-1 protein HIP IL-6 mrna HIP COX-2 mrna HIP IL-1, GFAP, cpla2, CCL3, inos, mpges, RelB, CD11b, CD45, CCL2, CYBB, TNF-α mrna a,b HIP IL-1, CCL3, TNF-α, CD11b, CD45, CYBB mrna b HIP GFAP, CD11b mrna a,b HIP GFAP, cpla2, COX-2, inos, mpges, RelB, CCL2, CD11b mrna a HIP GFAP, CD11b mrna LPS i.c.v. LPS i.c.v. Fat-1 transgenic mice (C57Bl/6 x C3H) Fat-1 transgenic mice Fat-1 transgenic mice on 10% safflower diet Fat-1 transgenic mice on 10% safflower diet Wildtype littermate on 10% safflower diet (n-3 deficient) Wildtype littermate on 8% 12 weeks HIP total and FFA DHA 12 weeks (wildtype were on fish HIP total and FFA DHA HIP IL-1, GFAP, cpla2, COX-2, CCL3, INOS, mpges, RelB, CD11b, CD45, CCL2, CYBB, TNF-α mrna HIP GFAP, Iba-1, FJb protein HIP mpges mrna HIP IL-1, GFAP, cpla2, COX-2, CCL3, INOS, mpges, 45

59 Delpech et al. (2014) Systemic LPS (i.p.) (C57Bl/6 x C3H) Fat-1 transgenic mice (C57Bl/6 x C3H) Fat-1 transgenic mice on standard diet (4.8% of total fatty acids n-3 PUFA) safflower diet + 2% fish oil Wildtype littermate on standard diet (4.8% of total fatty acids n-3 PUFA) oil for only 9 weeks) 3-5 months HIP total DHA HIP total DPA n-3 and EPA learning and memory (YM) food consumption body weight loss RelB, CD11b, CD45, CCL2, CYBB, TNF-α mrna HIP COX-2, TGF-β1, mpges-1 and CX3CL1 mrna HIP IL-1β mrna 24hr post LPS HIP TNF-α, IL-6, IL-10, CX3CL1 mrna 24hr post LPS HIP CD36 and MHCII protein 24hr post LPS HIP IL-10 mrna 2hr post LPS HIP TNF-α, IL-1β and IL-6 mrna 2hr post LPS 17-HpDHA, 17-hydroperoxy-DHA; CCL, chemokine (c-c motif) ligand; acsf, artificial cerebrospinal fluid; CX3CL, chemokine (c-x3-c motif) ligand; COX, cyclooxygenase, CX, cortex; CYBB, cytochrome b-245 beta polypeptide DHA, docosahexaenoic acid; DPA, docosapentaenoic acid, eepa, ethyl-epa; egla, ethylgamma-linolenic acid ; EPA, eicosapentaenoic acid; FFA, free fatty acid; FJ, Fluoro-jade; GFAP, glial fibrillary acidic protein; HIP, hippocampus; Iba, ionized calcium-binding adapter molecule; IL, interleukin; LPS, lipopolysaccharide; LTP, long term potentiation; MHC, major histocompatibility complex n, omega; NOS, nitric oxide synthase; NP, neuroprotectin; PGES, prostaglandin E synthase PLA, phospholipase; PPAR, peroxisome proliferator activated protein, PUFA, polyunsaturated fatty acids; RelB, nuclear factor-κb subunit ; TGF, transforming growth factor; TNF, tumor necrosis factor; YM, y maze a,b indicates treatment group represented in outcome columns (brain n-3 PUFA, non-inflammatory outcome, inflammatory outcome) 46

60 hippocampal IL-1 along with apoptotic cell markers 203. Finally, 24 hr following 125 g/kg of LPS i.p., fat-1 transgenic mice have attenuated LPS-induced increases in IL-1 mrna compared to their wildtype littermates. However, fat-1 mice also have augmented LPS-induced increases in COX-2, membrane associated PGE synthase-1, transforming growth factor 1 and chemokine (c-x3-c) ligand (CX3CL) 1. The authors argued that increases in these genes reflect an anti-inflammatory phenotype, where the fat-1 mouse has a higher proportion of M2 phenotype microglia 24hr post-lps 204. The fifth study administered 5 g of LPS directly into the brain left lateral ventricle, which avoids systemic effects that i.p. injection of LPS may have 91. C57bl/6 mice supplemented with 2% fish oil (% of total diet weight) for 9 weeks had elevated hippocampal total phospholipid DHA, but showed no changes in the non-esterified fatty acid pool. Out of a panel of inflammatory markers, only hippocampal COX-2 mrna were decreased by dietary fish oil supplementation upon LPS administration. The fat-1 mouse, which has both elevated total and non-esterified DHA in the hippocampus compared to its wildtype counterpart, showed attenuated LPS-induced mrna expression of a panel of pro-inflammatory genes including IL-1, GFAP, COX-2, and CD45. When wildtype littermates are switched to a 2% fish oil diet for 9 weeks from weaning until surgery, phospholipid and non-esterified DHA reaches the same concentration as in fat-1 mice, and gene expression profiles are similar with the exception of increased hippocampal expression of membrane PGE synthase mrna 91. This suggests the possibility that the non-esterified pool may be the important pool for regulating neuroinflammation. The authors concluded the study by evaluating the effect of infusing either 40 g 47

61 of DHA or 1 g of 17S-hydroperoxy DHA (protectin D1 precursor) i.c.v. immediately following LPS injection for 24 hr. While only 17-HpDHA infusion increased protectin D1 concentrations, both treatments decreased LPS-induced pro-inflammatory markers including TNF- and IL-1. 17S-hydroperoxy DHA appears to be more potent, as 1 µg decreased CD11b and GFAP mrna expression, which 40 µg DHA was unable to do. Unlike the transgenic and feeding approaches, i.c.v. administration of DHA or 17Shydroperoxy DHA does not modulate the ARA cascade enzymes COX-2 and calcium dependent cytosolic phospholipase A n-3 PUFA and neuroinflammation in i.c.v. IL-1 We identified three studies that investigated the effect of n-3 PUFA on neuroinflammation induced by i.c.v. injection of IL-1 Table 2-6) Supplementing rats for seven weeks with 1% ethyl EPA (% of total diet weight) prior to injection of IL- 1 ng i.c.v.) reduces not only memory deficits in the Morris water maze, but also brain PGE2 compared to the control coconut oil supplementation. However, supplementation of 0.2% EPA or with 5% soybean oil is ineffective at attenuating the effects of i.c.v. IL In a comparable study, 0.5% of either ethyl EPA or ethyl gamma linolenic acid (% of total diet weight) for 7 weeks prior to IL-1 administration (15 ng i.c.v.) reduces hippocampal PGE2. This study finds EPA is more effective than gamma linolenic acid at reducing IL-1 -induced amygdaloid PGE2 concentration and elevating IL-10 while also decreasing anxiety and memory deficits 206. Finally, Taepavarapruk and Song also report anti-inflammatory properties of ethyl EPA in the i.c.v. IL-1 model (15 ng i.c.v.), where 0.8% ethyl EPA (% of total diet weight) for 7 48

62 Table 2-6: Summary of studies investigating the effects of n-3 PUFA in IL-1 models Authors (year) Song and Horrobin (2004) Song et al. (2008) Taepavarapr uk and Song (2010) Injury Model Species PUFA Treatment(s) Comparison Treatment IL-1 (i.c.v.) Wistar rats a) 5% soybean, 4.8% coconut oil + 0.2% eepa, b) 4% coconut oil + 1% eepa diet IL-1β (i.c.v.) Wistar rats a) 4.5% palm oil + 0.5% eepa b) 4.5% palm oil + 0.5% egla c) 4% palm oil + 1% IL-1β i.c.v. a amygdala PGE2 a amygdala, hypothalamus IL- Long- Evans rats 5% coconut oil diet 5% palm oil diet ARA-rich oil diet 0.8% (v/w) eepa 0.8% (v/w) palm oil Treatment Duration/ Time point Brain n-3 PUFA 7 weeks Not reported 7 weeks Not reported 7 weeks prior surgery Not reported Non- Inflammatory Outcome b memory loss (MWM) a memory loss (MWM) a anxiety (EPM) Acetylcholine release NGF Learning and memory (8ARM) Inflammatory Outcome b brain PGE2 a,b HIP PGE2 10 protein HC IL-1 mrna 8ARM, 8-arm radial maze; ARA, arachidonic acid; eepa; ethyl eicosapentaenoic acid; egla, ethyl-gamma-linolenic acid; EPM, elevated plus maze; HIP, hippocampus; IL, interleukin; MWM; morris water maze; NGF, nerve growth factor; PG, prostaglandin; PUFA, polyunsaturated fatty acids a,b,c, indicates treatment group represented in outcome columns (brain n-3 PUFA, non-inflammatory outcome, inflammatory outcome) 49

63 weeks reduces IL-1 induction of IL-1 mrna, even though it does not alter acetylcholine concentrations n-3 PUFA and neuroinflammation in traumatic brain injury Traumatic brain injury is associated with an increase in pro-inflammatory cytokine production, including IL-1 and TNF-, and also is marked by increased microglial activation 208, 209. Three studies (Table 2-7) have evaluated the antineuroinflammatory properties of n-3 PUFA in traumatic brain injury models. Supplementing mice with a DHA- and EPA-enriched diet (1.5% of total diet weight) 60 days prior to controlled cortical impact decreases IL-1, IL-1 and TNF- mrna expression following the injury compared to mice fed a low n-3 PUFA control. Mice on a high n-3 PUFA diet also exhibit lower COX-2 mrna and protein concentrations following controlled cortical impact 210. A separate study found that following controlled cortical impact, CD68 protein levels are lower in mice consuming 40 mg/kg of DHA compared to no supplementation 211. Traumatic brain injury by midline fluid percussion induces cognitive impairment and motor deficits, while increasing activated microglia. The administration of 100 ng of aspirin-triggered resolvin D1 i.p. for 7 days, starting 3 days before the percussion, reduced the injury induced cognitive impairment and motor deficit, but did not reduce microglial activation. Resolvin E1, however, did reduce traumatic brain injury induced microglial activation, while not having any effects on the cognitive impairments and motor deficits

64 Table 2-7: Summary of studies investigating the effects of n-3 PUFA on traumatic brain injury models Authors (year) Mills et al. (2011) Pu et al. (2013) Harrison et al. (2015) Injury Model Species PUFA Treatment(s) Comparison Treatment Traumatic brain injury Traumatic brain injury Traumatic brain injury Sprague Dawley rats C57Bl/6 mice C57Bl/6 mice Per kg per day: a) 4 mg DHA b) 12 mg DHA c) 40 mg DHA DHA and EPA supplemented diet (15g/kg of diet) a) 100 ng AT-RvD1 i.p. b) 100 ng RvE1 i.p. No treatment Low n-3 diet (0.5% n-3) Treatment Duration/ Time point 30 days prior injury 2 months prior injury Saline i.p. Daily for 7 days starting 3 days before injury Brain n-3 PUFA Not reported total brain DHA, EPA and n- 3 DPA Not reported Non- Inflammatory Outcome c axon injury c apoptosis c learning and memory (MWM) sensorimotor control (FF, WH, CT) learning and memory (MWM) lesion volume CA3 neuronal survival myelin injury nerve conductance a motor deficit a learning and memory (OR) b sleep righting reflex Inflammatory Outcome c CD68 protein CX Iba-1 COX-2 protein CX IL-1α, IL-1β, TNF-α, COX- 2 inos mrna b activated microglia b rod microglia b ramified microglia 51

65 AT, aspirin triggered; CA, cornu ammonis, COX, cyclooxygenase; CT, cylinder test; CX, cortex, DHA, docosahexaenoic acid; DPA, docosapentaenoic acid, EPA, eicosapentaenoic acid; FF, foot fault; IL, interleukin; Iba, ionized calcium-binding adapter molecule; MWM, morris water maze; NOS, nitric oxide synthase; PUFA, polyunsaturated fatty acids; Rv, resolvin; TNF, tumor necrosis factor; WH, wire hang 52

66 n-3 PUFA and neuroinflammation in neuropathic pain Neuropathic pain is a disorder associated with a lesion of a nerve in either the peripheral or central nervous system 213, 214. Microglia are present in both acute and chronic neuropathic pain 215, while pro-inflammatory cytokines such TNF- and IL-1 are thought to modulate pain responses 214. Two studies were identified that evaluated the response of neuroinflammation following n-3 PUFA bioactive mediator treatment in a neuropathic pain model (Table 2-8). Injection of 300 ng of protectin D1 at the site of injury immediately following chronic constriction of the sciatic nerve lowered the concentration of CCL2, a chemotractant for microglia, and microglial activation in the spinal cord dorsal horn 216. Similar protection was obtained with 3 days of intrathecal injection of 100 ng of resolvin E1, a bioactive mediator derived from EPA, following injury. Resolvin E1 decreases pro-inflammatory Iba-1 and GFAP mrna expression and TNF- protein concentration n-3 PUFA and neuroinflammation in diabetes There is evidence that diabetes is linked with increased neuroinflammation, including NF B induction 218. Moreover, diabetics often experience diabetic neuropathic pain, which itself is associated with neuroinflammation including microglial activation 219. Two studies investigated whether DHA was anti-neuroinflammatory in the streptozotocin diabetic rat model (Table 2-9). Streptozotocin is a toxin that targets pancreatic beta cells, inducing diabetes. Rats gavaged with 13.3 mg/kg/day of DHA for 12 weeks prior to streptozotocin (i.p.)-induction had decreased hippocampal NF- B and 53

67 Table 2-8: Summary of studies investigating the effects of n-3 PUFA in neuropathic pain models Authors (year) Xu et al. (2013) Injury Model Species PUFA Treatment(s) Comparison Treatment Chronic constriction injury CD1 mice 300 ng NPD1 s.c. perisurgical Vehicle perisurgical Treatment Duration/ Time point 1 week prior to surgery Brain n-3 PUFA Not reported Non- Inflammatory Outcome mechanical allodynia on going pain autotomy spinal LTP axonal injury Inflammatory Outcome spinal cord dorsal horn Iba- 1 protein spinal cord dorsal horn GFAP, IL-1β and CCL2 mrna Xu et al. (2013) Chronic constriction injury CD1 mice 100 ng RvE1 i.t. Vehicle i.t. Daily for 3 days post injury Not reported mechanical allodynia heat hyperalgesia dorsal horn Iba-1 and GFAP mrna dorsal horn TNF-α protein CCL, chemokine (c-c motif) ligand; GFAP, glial fibrillary acidic protein; Iba, ionized calcium-binding adapter molecule; IL, interleukin; LTP, long term potentiation; NP, neuroprotection, PUFA, polyunsaturated fatty acids; Rv, resolvin; TNF, tumor necrosis factor 54

68 Table 2-9: Summary of studies investigating the effects of n-3 PUFA in diabetes models Authors (year) Alvarez- Nölting et al. (2012) Injury Model Species PUFA Treatment(s) Comparison Treatment STZ (i.p.) Wistar rats chow plus 13.3 mg/kg/d DHA by gavage Treatment Duration/ Time point Brain n-3 PUFA Chow # 12 weeks Not reported Non- Inflammatory Outcome blood glucose and glycated hemoglobin HIP neurogenesis HIP neuronal apoptosis HIP oxidative stress learning and memory (MWM) Inflammatory Outcome HIP NF-κB protein Jia et al. (2014) STZ (i.p.) Sprague Dawley rats 4% fish oil (1.2% EPA + DHA) chow 1 week prior to STZ and 5 week post STZ Not reported blood glucose HIP oxidative stress learning and memory (MWM) HIP TNF-α mrna, HIP pikkβ, TNF-α, NF-κB proteins HIP I Bα protein DHA, docosahexaenoic acid, EPA, eicosapentaenoic acid; HIP, hippocampus; I B, nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor alpha; IKK, inhibitor of nuclear factor kappa-b kinase subunit beta; MWM, morris water maze; NF- B, nuclear factor kappa light chain enhancer of activated B cell; STZ, PUFA, polyunsaturated fatty acids; STZ, streptozotocin; TNF, tumor necrosis factor # insulin-treated rats were also included in the study and were similar to DHA-only treated rats in all measures excluding weight, glycemic control, and oxidative stress 55

69 memory deficits despite elevated blood glucose levels, which were not impacted by DHA treatment 220. Likewise, Jia and colleagues supplemented Sprague-Dawley rats with 4% fish oil (% of total diet weight), starting 1 week prior to streptozotocin injection and continuing 5-weeks post-streptozotocin, and showed that supplementation attenuates streptozotocin-induced TNF- mrna and protein increases in the hippocampus 221. Rats receiving fish oil performed better in the Morris water maze, indicating improved memory, even though blood glucose levels remained high n-3 PUFA and neuroinflammation in other models Several studies have reported on the anti-inflammatory effects of n-3 PUFA in other models (Table 2-10). Radiotherapy is a common therapeutic strategy against brain tumors, and it is associated with cognitive dysfunction and increases in pro-inflammatory cytokines mrna such as IL-1 and TNF- 222, 223. Lynch et al. observed that 4-week supplementation of 250 or 500 mg/day of ethyl EPA attenuated the increase of proinflammatory cytokines in the hippocampus of rats, including IL-1, IL-1RI, IL-1RAcP, induced by whole body irradiation. Interestingly, the lower dose of 250 mg/day also raised IL-10 concentration 224. Olfactory bulbectomy has been proposed as a model of depression in rats 225, presenting with changes in immunity 226 and increased brain pro-inflammatory cytokines 227. Olfactory bulbectomized rats supplemented with 1% ethyl EPA (% of total diet weight) for 7 weeks demonstrated lower induction of calcium dependent cytosolic phospholipase A2 mrna expression and protein activity compared to rats supplemented 56

70 Table 2-10: Summary of studies investigating the effects of n-3 PUFA on other neuroinflammatory models Authors (year) Lynch et al. (2003) Song et al. (2009) Cupri et al. (2012) Terrando et al. (2013) Yip et al. (2013) Injury Model Species PUFA Treatment(s) Comparison Treatment Whole body irradiation Olfactory bulbectomy (depression) BAFF transgenic mice (lupus and Sjogren s syndrome) Surgically induced cognitive decline G93A-SOD1 (ALS) Wistar rats Sprague Dawley rats BAFF transgenic mice C57bl/6 mice G93A- SOD1 a) chow mg/d eepa b) chow mg/d eepa chow 1% eepa diet 1% palm oil diet n-3 supplemented diet (1.54% of fatty acids n-3 PUFA) Control (0% of fatty acids n-3 PUFA) Treatment Duration/ Time point 4 weeks prior to irradiation 100 ng AT-RvD1 i.p. Vehicle i.p. Prior to incision 300 mg/kg/d eepa and 43 mg/kg/d edha Control diet Brain n-3 PUFA Not reported 7 weeks Not reported 12 weeks Not reported From 14 to 20 weeks Not reported spinal DHA Brain DHA and EPA Non- Inflammatory Outcome a,b apoptotic markers depressivelike symptoms (MWM and OF) neurogenesis LTP plasma LXA4, IL-6, AST protein LTP memory retention (FTC) disease progression in symptomatic mice disease progression in pre- Inflammatory Outcome a,b HIP IL-1, IL-1RI, IL- 1RAcP protein a,b HIP IRAK protein phosphorylation ratio a HIP IL-10 protein hypothalamus cpla2 mrna and activity HIP CD68 protein HIP GFAP area spinal GFAP, Iba-1, and CD11b protein 57

71 Keleshian et al. (2014) NMDA induced excitotoxicity a) n-3 adequate (4.6% of diet n-3 PUFA) b) fish oil (9.4% of diet n-3 PUFA) n-3 deficient (0.2% of diet n-3 PUFA) symptomatic spinal vacuoles neuron morphology lipid peroxidation 15 weeks DHA a,b BDNF, NGF, ipla2 protein a,b ipla2 activity b body a,b IL-1, cpla2, spla2, COX- 1, COX-2 and GFAP protein a,b spla2 activity a,b cpla2 activity in saline weight ALS, amyotrophic lateral sclerosis; AST, aspartate transaminase; AT, aspirin triggered; BDNF, brain derived neurotrophic factor; COX, cyclooxygenase; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; edha; ethyl DHA; eepa, ethyl-epa; GFAP, glial fibrillary acidic protein; HIP, hippocampus; FTC, fear conditioning test; Iba-1, ionized calcium binding adaptor molecule; IL, interleukin; IkB-a, nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor alpha; IL, interleukin; IRAK, IL-1 receptor-associated kinase; LTP, long term potentiation; LX, lipoxin; MWM, Morris water maze; n, omega; NGF, nerve growth factor; OF, open field; PUFA, polyunsaturated fatty acids; PLA, phospholipase a,b indicates treatment group represented in outcome columns (brain n-3 PUFA, primary outcome, inflammatory outcome) 58

72 with 1% palm oil 228. Ethyl-EPA supplemented rats also have reduced depressive behaviour in the open field test and improved scores in the Morris water maze 228. The BAFF (B cell activating factor belonging to the TNF family) transgenic mouse is a mouse model that presents as a model of lupus. This model has been reported to have microglial activation 229. When compared to an n-3 PUFA deficient control, BAFF transgenic mice consuming a diet containing 1.54% n-3 PUFA (% of total fatty acids, combination of -linolenic acid, EPA and DHA) have improved neurogenesis and long-term potentiation, and lower hippocampal CD68 protein concentrations 230. Surgery is associated with cognitive decline 231. Administering 100 ng of the specialized pro-resolving lipid mediators aspiring-triggered resolvin D1 to mice prior to tibia fracture stabilization increases and improves memory retention compared to mice receiving vehicle control. This protection against surgery-induced cognitive decline was accompanied with an increase in GFAP labeling in the hippocampus 232. NMDA-induced excitotoxicity increases inflammatory markers in the brain, including GFAP 233. Fish oil supplemented and n-3 PUFA adequate diets appeared to decrease NMDA-induced cytosolic phospholipase A2 activity compared to diets deficient in n-3 PUFA, despite not changing cytosolic phospholipase A2 protein level. Fish oil and n-3 PUFA adequate diets also did not modify NMDA-induced GFAP, IL-1 or secretory phospholipase A2 increases in protein levels 234. In some instances, however, reducing inflammation with n-3 PUFA has been reported to worsen symptoms. In the G93A-SOD1 mouse, a model of amyotrophic lateral sclerosis, pre-symptomatic mice fed 343 mg/kg/day of n-3 PUFA for 6 weeks have lowered GFAP and microglial markers levels, but accelerated symptom progression and 59

73 increased lipid peroxidation compared to mice on control diet Conclusion In animal models, n-3 PUFA are generally associated with protection against neuroinflammation, although with varying efficacy and consistency between disease models. Herein we have summarized the growing body of literature on the modulation of neuroinflammation by n-3 PUFA in animal models of stroke, spinal cord injury, aging, Alzheimer s disease, Parkinson s disease, lipopolysaccharide and IL-1β injections, diabetes, neuropathic pain, traumatic brain injury, depression, diabetes, surgicallyinduced cognitive decline, whole body irradiation, amyotrophic lateral sclerosis and lupus. In some cases, such as stroke, the evidence for an anti-inflammatory effect is strong, where in other instances such as in spinal cord injury, the results are relatively mixed. The differences in results may be due to heterogeneity between studies. There are vast differences between the type of n-3 PUFA administered (alpha-linolenic acid vs. EPA vs. DHA), the route of administration (i.v. vs. p.o. vs. i.p. vs. i.c.v.), the dose (from g to mg), the duration of administration (acute bolus to 18 months), the inclusion of antioxidants, the inflammatory markers (microglial vs. astrocytic vs. cytokines) measured and the timing of measurement. Moreover, control diets used in the studies reviewed vary greatly in n-3 PUFA content, with some studies using an n-3 PUFA adequate diet while others use an n-3 PUFA deficient diet. Thus, it is often unclear whether the phenotype observed is related to supplementation or lack of apparent deficiency. Considering that the relationship between n-3 PUFA levels is likely not linear to their biological effects 236, 60

74 it is difficult to compare studies that use different baselines to determine the efficacy of augmenting n-3 PUFA levels. As most studies measured only a few inflammatory markers in their experiments, it is important to note that differentiation between various immune cell types is often difficult or impossible based on a single marker; microglia and peripheral macrophages share multiple known surface markers, and only techniques that can compare the origin of the cells or the relative expression of various surface markers, such as flow cytometry, are reliable for identification 118, 237. In addition, neutrophils have also been shown to express CD68 and CD11b on occasion, while macrophages can express the common neutrophil marker myeloperoxidase, highlighting the need for multiple methods of identification 238. It is also important to take into consideration that most studies evaluated the expression of neuroinflammatory marker(s) at a single time point. Different cellular markers and cytokines, however, change expression at different rates 239, 240. Evaluating only single time points may result in false negatives, and limit any conclusion to be made on the resolution of neuroinflammation. Due to the small number of null studies reported in this review, combined with the multiple variables mentioned above, it is difficult to define therapeutic dose and duration of administration when comparing null studies with studies that found a positive or negative effect of n-3 PUFA. Considering this, there does not appear to be a definite therapeutic acute dose across all studies. While a dose of 833 g was reported to be ineffective at lowering ischemia induced rise in PGE2 concentration 171, a lower dose of 20 g was successful at reducing ischemia induced myeloperoxidase activity 241. The 61

75 same issue arises with duration of n-3 PUFA treatment. While 20 months of n-3 PUFA adequate diet administration did not mitigate the aging induced decrease in IL , the consumption of ethyl EPA for 4 weeks was able to mitigate the LPS induced decrease in IL It may be possible to suggest, however, that DHA is more potent than EPA. Out of 3 studies that evaluated the anti-inflammatory effects of EPA alone 162, 171, 176, only one reported anti-inflammatory properties of EPA 162. Due to the small number of studies, however, more studies are needed to establish the higher potency of DHA. From the data presented in this review, it is also hard to determine whether n-3 PUFA always act directly on neuroinflammatory pathways. It is possible that n-3 PUFA reduced the injury directly, such as by decreasing infarct size or neuronal cell death, and attenuated the neuroinflammatory response that accompanies such injuries as a consequence. Studies infusing IL-1 and LPS, however, directly activate neuroinflammatory pathways with minimal injury. Consistent with the direct antiinflammatory properties of n-3 PUFA observed in cell culture 39, n-3 PUFA reduced neuroinflammation in response to i.c.v. injection of both LPS and IL-1 indicating that n-3 PUFA may have the ability to impact neuroinflammatory pathways separate from their modulation of non-inflammatory pathways. The mechanism by which n-3 PUFA convey their anti-inflammatory properties is not clear. One hypothesis is that enzymatic metabolism of n-3 PUFA to bioactive mediators is primarily responsible for the anti-inflammatory effect of increased tissue n-3 PUFA levels. As reported above, these bioactive lipid mediators are sufficient to reduce neuroinflammation in stroke, i.c.v. LPS, neuropathic pain models, and surgery-induced cognitive decline. Future studies evaluating the mechanism of n-3 PUFA in 62

76 neuroinflammation are warranted. N-3 PUFA administration has been tested in multiple clinical trials for various neurological and psychiatric disorders. At best, these have yielded mixed results It is unknown, however, whether neuroinflammation was actually targeted and lowered in these trials. With the link between neuroinflammation and both neurological and psychiatric disorders 117, 120, it would be important to correlate any reduction of neuroinflammation to symptoms. With the development of translocator protein 18 kda, a receptor highly expressed in activated microglia, ligands for positron emission tomography imaging, neuroinflammation can now be imaged in vivo in humans 246. Since n-3 PUFA modulate microglia markers such as Iba1 as described above, future clinical studies in humans could image neuroinflammation following n-3 PUFA treatment and relate translocator protein 18 kda binding with the reduction with symptoms. Overall, this review summarizes all of the known literature on n-3 PUFA and neuroinflammation. N-3 PUFA appear to target brain inflammation signaling in a variety of animal models. However, the mechanism by which n-3 PUFA are antineuroinflammatory and whether the neuroinflammatory effects observed are direct effects or secondary have yet to be determined Acknowledgments This work was supported by grants from the Natural Sciences and Engineering Research Council of Canada and the Canadian Institutes of Health Research to RPB. MOT received a studentship from the Natural Sciences and Engineering Research Council of Canada and RPB holds a Canada Research Chair in Brain Lipid Metabolism. 63

77 Chapter 3: Objectives and Hypotheses 64

78 3.1 Objectives 1. To measure the rat neurolipidome and to determine the effect of using microwave fixation on the neurolipidome. 2. To develop a self-resolving model of neuroinflammation following i.c.v. LPS injection. 3. To determine if increasing brain DHA increases the resolution of neuroinflammation following i.c.v. LPS injection Hypotheses 1. Bioactive mediators will be increased due to ischemia and will be inhibited by microwave fixation. 2. DHA derived bioactive mediators will be correlated with the resolution of neuroinflammation following i.c.v. LPS injection. 3. Increasing brain DHA will increase the resolution of neuroinflammation following i.c.v. LPS injection. 65

79 Chapter 4: High-resolution lipidomics coupled with rapid fixation reveals novel ischemia-induced signaling in the rat neurolipidome Adapted from: Marc-Olivier Trépanier, Michael Eiden, Delphine Morin-Rivron, Richard P. Bazinet, Mojgan Masoodi (Submitted to Journal of Neurochemistry) Contribution: Along with RPB, I helped design the study. I collected all brain samples and travelled to Lausanne, Switzerland to perform the lipid extraction and learn the mass spectrometry techniques. Along with MM, I performed the statistical analysis. I wrote the first draft of the manuscript, with some technical help from MM. 66

80 4.1. Abstract The field of lipidomics has evolved vastly since its creation 15 years ago. Advancements in mass spectrometry have allowed for the identification of hundreds of intact lipids and lipid mediators. However, due to the release of fatty acids from the phospholipid membrane in the brain caused by ischemia, identifying the neurolipidome has been challenging. Microwave fixation has been shown to reduce the ischemiainduced release of several lipid mediators. Therefore, this study aimed to develop a method combining high-resolution tandem mass spectrometry, high-energy head-focused microwave fixation and statistical modeling, allowing for the measurement of intact lipids and lipid mediators in order to eliminate the ischemia-induced release of fatty acids and identify the rat neurolipidome. In this study, we demonstrated the ischemia-induced production of bioactive lipid mediators, and the reduction of variability by using microwave fixation in combination with liquid chromatography with tandem mass spectrometry. We have also illustrated for the first time the microwave inhibition of alterations of intact lipid species due to ischemia. While many phospholipid species were unchanged by ischemia, other intact lipid classes such as lysophospholipids were increased due to ischemia. 67

81 4.2 Introduction The field of lipidomics has vastly expanded since its emergence approximately 15 years ago Due to the diversity of lipid species, as well as the wide range of their concentrations, it is essential to use different lipidomic approaches to produce a global profile of structurally and functionally diverse lipids to understand lipid metabolism and signaling in brain tissue biology. Lipidomic approaches have previously been reported in various tissues, in either normal or pathological state, including adipose 250, nasal washes 251 and brain 24, 252. Mass spectrometry-based lipidomics is currently the most commonly used tool for profiling lipid species. There are currently two main analytical approaches which in parallel provide a comprehensive, global profiling of the brain tissue lipidome. Firstly, there is the lipidomics of intact lipids, including shotgun lipidomics (developed by Han and Gross) 249 and liquid chromatography with tandem mass spectrometry-based lipidomics, which aims at the rapid identification of hundreds of molecular lipids across multiple structural classes. The second approach, lipidomics of lipid-signaling molecules, captures a wide range of low-abundance lipid species within a class or specific pathway. It is essential to combine these two approaches to capture and study the lipidome and related lipid signaling in the brain. Chromatographic separation, solid-phase extraction and liquid-liquid fractionation can greatly improve the recovery of lowabundance lipid species. In brain tissue, due to the high concentration of lipids, fractionation of different classes of lipids is essential prior to analysis of lipid mediators. Brain tissue is capable of generating a wide range of signaling molecules. Arachidonic acid (ARA), for example, is an important component of mammalian cell 68

82 membrane phospholipids. Depending on the enzyme that cleaves ARA from the membrane, ARA can present in the forms of non-esterified ARA, 2-arachidonyl glycerol and arachidonyl ethanolamine. Non-esterified ARA, a signaling molecule in itself, can also be further metabolized by a multitude of enzymes, generating a vast array of bioactive mediators such as the pro-inflammatory prostaglandins (PG) and pro-resolving lipoxins 253. The identification and quantification of fatty acid-, fatty acyl glycerol- and fatty acyl ethanolamide-related bioactive lipids is a challenging task. This is mainly due to the large number of bioactive lipids with similar chemical properties which are produced within the same cascade and are part of a complex regulatory network. Thus, they have to be measured simultaneously to assess the biochemical processes being studied. This task is further complicated by the presence of a large number of isomers of bioactive lipids with very similar physiochemical properties but diverse biological functions. Therefore, the comprehensive study of lipids requires a highly sensitive and selective analytical method. We have developed a lipidomics platform using high-mass accuracy and massresolution mass spectrometry, which allows us to identify the wide range of bioactive lipids in biological systems 254. High-resolution mass spectrometry allows the separation of many isobars by measuring the accurate mass/charge (m/z) ratios of the lipid species and by computing the elemental formulae when combined with tandem mass spectrometry. This simplifies the identification and structural elucidation of unknown lipid mediators 254,

83 The challenges of brain tissue lipidomics include the development of relevant analytical methodologies as well as bioinformatics tools for determination of alterations in lipid metabolic pathways and signaling during disease progress. Although currently available databases such as LIPID MAPS and METLIN provide valuable tools for the identification of lipid species in brain tissue, the development of sophisticated bioinformatics tools such as lipid prediction for database-independent approaches, theoretical databases and search algorithms is critical for the identification of unknown lipid species 254, 256, 257. The measurement of artifacts poses another challenge in the attempt to describe the brain lipidome. It has been known for some time that the lipid profile of the brain drastically changes under hypoxic conditions 258. Coined the Bazan Effect 259, Bazan and colleagues demonstrated that ischemia releases various fatty acid from the phospholipid membrane into the non-esterified fatty acid pool due in part to increased glutaminergic transmission , resulting in increased activity of phospholipases 263. While multiple fatty acids are affected by ischemia, ARA is the main component of that release 258. The released fatty acids are then metabolized and increase the low basal concentrations of the downstream mediators 264, 265. This release of non-esterified fatty acids from the phospholipid membrane causes complications when applying lipidomics approaches to measure lipid species in the brain, and especially low abundance downstream mediators. In order to minimize the Bazan Effect and eliminate the increased variability in the data, microwave fixation has been applied as a method of euthanasia 266. In short, a focused high-energy microwave beam is aimed at the top of the head, denaturing all proteins involved in the release of fatty acids from the phospholipid membrane as well as 70

84 the proteins involved in downstream metabolism, effectively fixing the brain in its current state. Previous studies utilizing microwave-fixation have used microwaves that generate approximately 3 to 13 kw beams for 1.6 sec up to 3.5 sec , which allows for rapid and humane euthanasia and further reduces the potential confounding effects of ischemia. In this study, we used of high-energy head-focused microwave fixation in combination with multiple mass spectrometry methods in order to characterize and to illustrate the effect of ischemia on the rat neurolipidome (Figure 4-1). The rat neurolipidome was assessed using high-energy head-focused microwave fixation and compared to the neurolipidome following CO2 asphyxiation. Furthermore, we measured the rat neurolipidome following LPS-induced inflammatory signals independent of ischemia Methods Subjects The present study was conducted in accordance to the standards of the Canadian Council for Animal Care and was approved by Animal Care Committee of the Faculty of Medicine at the University of Toronto (protocol number ). Two month old Long Evans rats were purchased from Charles Rivers (La Prairie, Qc). Animals were housed 3 per cage in a vivarium on a 12 hr light/dark cycle and maintained at a temperature of 21 C. Water and food were available ad libidum. The rat chow (Teklad Global, % Protein Rodent Diet; Envigo, Madison, WI, U.S.A.) composition consisted of 189 g/kg protein, 60 g/kg fat, 554 g/kg carbohydrates, 38 g/kg fiber, 59 g/kg 71

85 ash, and 100 g/kg moisture. The diet fat composition (in percent of total fatty acids) was palmitate (18.5%), stearate (2.8%), oleate (18.5%), linoleate (54.8%), and α-linolenate 72

86 Figure 4-1. Flow of methods in Chapter 4 CO 2 CO 2 + MW LPS CO 2 asphyxiation 5 min Head focused High energy Microwave fixation 13.5kW, 1.6s MW LPS CO2+MW CO2 Brain homogenization SPE column MTBE extraction LC-MS/MS Directinfusion MS/MS Mediators Intact lipid MW 73

87 Treatment groups Following the acclimatization period, the animals were separated into 4 groups based on the proposed euthanasia method; 1) microwave fixation (MW, n=10), 2) CO2 asphyxiation (CO2, n=10), 3) CO2 asphyxiation followed by microwave fixation (CO2+MW, n=9) and 4) lipopolysaccharide (LPS, Sigma Aldrich, St-Louis, MO, USA) i.p. (1 mg/kg, 1mg/ml in 0.9% saline solution) 3 hr prior to microwave fixation (LPS, n=9). Group 3 served as a positive control to ensure that microwave fixation did not denature lipid species, while group 4 served to identify lipid mediators that have previously been shown not be present at basal levels and need an inflammatory insult to increase its production Microwave fixation In order to perform high-energy head-focused microwave fixation, conscious, unanesthetized animals are placed into an animal restrainer and immediately inserted into the microwave (Cober Electronics Inc., Norwalk, CT, model S15P Vivostat). Following insertion of the animal restrainer into the microwave, a single microwave beam (13.5 kw, 1.6s, 2450 MHz) is aimed directly at the top of the head. For CO2 asphyxiation group, animals were placed in a CO2 tank and left inside the tank for 5 minutes following the end of respiration. Once euthanized, the heads of all 4 groups were cut off and placed on ice for 5 minutes. Brains were excised following the 5 minutes wait period and flash frozen in 74

88 liquid nitrogen for 15 seconds. Brains were placed in glass vials and the vials were filled with N2 gas and stored at -80 C until analysis. (5.6%) 236. The acclimatization period was 2 weeks, and the animals were handled every 2 nd day to reduce stress during experimental procedures Brain preparation Once all brains had been collected and were ready to analyze, they were kept on dry ice to keep frozen, along with tubes to be used. Frozen brains were then inserted into tissuetube TM (Covaris Ltd., Brighton, U.K.) and attached to the cooled tubes. TissueTUBE TM containing frozen brains were quickly inserted in a cryoprep TM CP02 impactor (Covaris Ltd., Brighton, U.K.) and received 1 to 3 calibrated impacts in order to get brain into powder form. TissueTUBE TM were inverted, transferring the brain powder into attached cooled tubes and quickly returned to dry ice to avoid thawing. Approximately 100 mg of crushed brain was weighed and transferred into a new frozen Eppendorf tube to maintain brain frozen Lipid extraction 100 mg of whole brain tissue was homogenized in 1 ml of ammonium bicarbonate buffer (concentration: 150 mm of ammonium bicarbonate in water) using a Tissue Lyser (Qiagen AG, Switzerland) at a speed of 25 Hertz for 2.5 min. 150 l of the homogenate was collected for intact lipid analysis, leaving 850 l for bioactive mediator analysis. 75

89 20 l of the 150 l homogenate was further diluted with 160 l of ammonium bicarbonate buffer using Hamilton Robot and 810 l of MTBE /methanol (7/2 v/v) containing internal standard was added to this mixture. The internal standard mixture contained: lysophasphatidylglycerol 17:1, lysophosphatidic acid 17:0, phosphatidylcholine 17:0/17:0, phosphatidylserine 17:0/17:0, phosphatidylglycerol 17:0/17:0, phosphatidic acid 17:0/17:0, lysophposphatidylinositol 13:0, lysophosphatidylserine 13:0, lysophosphatidylcholine 12:0, lysophosphatidylethanolamine, cholesteryl D6, diacylglycerol 17:0/17:0, triacylglycerol 17:0/17:0/17:0, ceramide 18:1;2/17:0, sphingomyelin 18:1;2/ 12:0, phosphatidylethanolamine 17:0/17:0, cholesteryl ester 20:0, phosphatidylinositol 16:0/16:0. The solution was mixed at 700 rpm, 15 min at 4 C using a ThermoMixer C (Eppendorf AG, Hamburg, Germany) and then centrifuged at 3000 g for 5 min. 100 l of the organic phase was transferred to a 96-well plate, and dried in a speed vacuum concentrator. Lipid extract was reconstituted in 40 l of 7.5 mm ammonium acetate in chloroform/methanol/propanol (1:2:4, V/V/V). All liquid handling steps were performed using a Hamilton STAR robotic platform with the Anti Droplet Control feature for organic solvents pipetting as described previously 271. The remaining 850 l of homogenate was used for bioactive mediator analysis. 150 l of 100% methanol was added to the remaining homogenate to bring the volume to 1 ml. The mixture was spun at approximately g (5430 R centrifuge, FA HS rotor) (Eppendorf AG, Hamburg, Germany) for 5 min at 4 C. The supernatant was removed into a new glass tube on ice. One ml of 15% methanol was added to the pellet and homogenized in a Tissue Lyser (25 Hz, 2.5 min). The homogenate was spun 76

90 (25000g, 5 min, 4 C) and the supernatant was added to the glass tube. One ml of 15% methanol was used to make a final volume of 3 ml. Extraction of lipid mediators from the brain tissue was performed according to our published protocol 254 with slight modifications, outlined as follows: internal standards PGB2-d4 (40 ng), 12-hydroxyeicosatetraenoic acid-d8 and arachidonyl ethanolamine-d8 (Cayman Chemicals, Ann Arbor, MI, USA) were added to the homogenized brain in 15% (v/v) methanol in water. The cartridges (Strata-X 33 u Polymeric Reversed phase 60 mg /3 ml) were washed with methanol (3 ml) followed by water (3 ml) prior to loading the homogenate (3 ml). The cartridges were then washed with 15% methanol in water (3 ml) and lipid mediators were eluted in methanol (3 ml) and collected in glass tubes. The organic solvent was evaporated using a fine stream of nitrogen and the remaining residue was re-dissolved in ethanol (100 l) and stored at 20ºC awaiting analysis Mass spectrometry analysis Lipidomics analysis of intact lipids was performed using a QExactive mass spectrometer (Thermo Fisher Scientific) equipped with a TriVersa NanoMate ion source (Advion Biosciences) as described previously 271. The data were acquired in both positive and negative mode using a resolving power of 140,000 in full scan and 17,500 in tandem mass spectrometry mode. Scan m/z range from 200 to 1,000. The lipidomics analysis of bioactive lipid mediators was performed as previously described 272 on an LTQ Elite (Thermo Scientific) linear ion trap-orbitrap mass spectrometer using a heated electrospray ionization source in both negative and positive ionization mode. Chromatographic analyses were performed using a A I-Class ultra 77

91 performance liquid chromatography system (Waters Corporation, Milford, MA, USA) combining a binary pump, a FTN autosampler and a column oven. The autosampler temperature was set at 4 C. Ten l out of the 100 l sample was injected onto a chromatographic column. For the negative mode, the bioactive lipids were separated on a C18 reversed-phase liquid chromatography column (Phenomenex Luna, 3 m particles, mm) using a linear mobile phase gradient (A, 0.02% glacial acetic acid in water; B, 0.02% glacial acetic acid in acetonitrile) at a rate of 0.5 ml/min. Starting conditions consisted of 35% B and were maintained for 1 minute. The gradient was then increased to 95% B over 12 min, remained there for 2 min and finally was returned to the initial conditions for 2 min to allow equilibration. For the positive mode, the bioactive lipids were separated on a C18 reversed-phase liquid chromatography column (Phenomenex Kinetex-XB-C18, 2.6 m particles, mm) using a gradient (A: 10 mm ammonium acetate+ 0.1% formic acid; B: ACN: H2O: formic acid (90:10:0.1)+ 10 mm ammonium acetate) at 0.5 ml/min. Starting conditions consisted of 35% B and were maintained for 4 min. The gradient was then increased to 95% B over 6 min, maintained for 2 min and finally returned to the initial conditions for 2 min to allow equilibration. Capillary and source heater temperatures were set to 325 C and 50 C, respectively, and spray voltage was adjusted to 4,000 V. A resolving power of 120,000 was used in full scan and 1,500 in tandem mass spectrometry mode. Scan m/z ranges of 150 to 500 (mass spectrometry) and 50 to 500 (tandem mass spectrometry) were used. Method development and validation, along with identification process, bioinformatics and related software have been described previously 254, 271,

92 Data analysis Univariate statistical analysis using one-way analysis of variance (ANOVA) was performed on log transformed concentrations due to unequal variance. For protectin D1, 17-hydroxy DHA, PGE2 and thromboxane B2, samples below the detection limits were removed from the analysis. Since protectin D1 was not detected in either the MW and LPS groups, a t-test was performed on the log transformed protectin D1 concentration of the CO2 and CO2+MW groups. Differences in variability of concentrations were measured by Bartlett s test 273. For unsupervised multivariate statistical analysis, we used hierarchical cluster analysis (HCA) using Ward s algorithm. Supervised analysis was performed using Partial Least Squares Discriminant Analysis (PLS-DA), where repeated stratified crossvalidation was used for model validation. All multivariate data analyses were performed using the programming language R using custom-built scripts as well as the 'pls' and 'pheatmap' packages. We calculated pairwise correlations between all variables in order to visualize a correlation network in The BioLayout Express 3D software. The Fruchterman-Reingold algorithm was used to generate the layout and only edges with a pairwise correlation of higher than 0.85 were considered. To further clean up the data, only cliques with more than 10 connected members were considered. 79

93 4.4. Results In order to investigate the changes between the different treatment groups, we first used unsupervised concepts of multivariate statistical analysis. HCA using Ward s algorithm was used to detect clusters in the data sets from lipid mediators and intact lipids. In the case of the of the lipid mediators (Figure 4-2a), the CO2 asphyxiation group showed strong differences and clustered together compared to the other treatment groups. (Figure 4-2a). The CO2+MW samples also grouped together (with the exception of a single sample) and again showed a similar tendency to have elevated metabolite concentrations compared to the MW group, albeit with much lower levels of metabolite concentration compared to the CO2 group (Figure 4-2a). For example, PGE2 concentration was 522-fold higher in the CO2 group compared to the MW (Figure 4-2b). In the CO2+MW group, PGE2 was also increased compared to the MW, but only by 59 fold. This would indicate that microwave fixation is a crucial sample preprocessing step in order to mitigate the increased mediator concentration measured following ischemia. Injecting LPS 3 hours prior to microwave fixation increased the production of PGE2 by 19 fold compared to the MW (Figure 4-2b). Most mediators, such as thromboxane B2, arachidonyl ethanolamide, 12- hydroxyeicosatetraenoic acid, and 17-hydroxy DHA, showed similar increases in the CO2 and CO2+MW groups, but no effect of LPS (Figure 4-2c,d,e,f). Some mediators, such as protectin D1, were detected in the CO2 and CO2+MW groups, but were below detection limits in the other two groups (Figure 4-2g). Furthermore, variability between groups was 80

94 Figure 4-2. Microwave fixation inhibits ischemia-induced production of bioactive lipid mediators A B PGE 2 (pmol/mg of brain tissue) A CO 2 B CO 2 +MW C LPS D MW C TXB 2 (pmol/mg of brain tissue) A CO 2 B CO 2 +MW C LPS C MW H D F AEA (pmol/mg of brain tissue) 17-HdoHE (pmol/mg of brain tissue) A CO 2 A CO 2 B CO 2 +MW B CO 2 +MW C LPS C LPS C MW C MW E G 12-HETE (pmol/mg of brain tissue) PD1 (pmol/mg of brain tissue) A CO 2 A CO 2 B CO 2 +MW B CO 2 +MW C LPS N.D. LPS C MW N.D. MW Heat map representation illustrates that CO2 asphyxiation and CO2+MW clusters separately from other groups (A). Brain concentrations (n=9-10, ± SEM) of prostaglandin (PG) E2 (B), thromboxane (TX) B2 (C), arachidonyl ethanolamide (AEA) (D), 12-hydroxyeicosatetraenoic acid (HETE) (E), 17-hydroxy DHA (HdoHE) (F), and protectin D1 (PD1) (G). Bar labeled with different superscripts identifies significant differences identified by one-way ANOVA and Tukey s post hoc test of log-transformed concentrations (p<0.05). PLS-DA was performed to elucidate the metabolites driving the separation. A 77.2% overall predictive accuracy was achieved (H). 81

95 significantly different from one another, with the MW and LPS groups (aside for PGE2) having the lowest variability. To further elucidate what metabolites are driving the separation between the groups, we performed a supervised multivariate analysis using PLS-DA (Figure 4-2h). To test the group separation, we performed repeated stratified cross-validation and evaluated the predictive performance on the respective hold out data sets. Separation of the four phenotypic groups on the lipid mediator data set results in an overall prediction accuracy of 77.2% (Tables 4-1). Inspecting the class-based prediction statistics (Tables 4-2) adds more detail to the predictive performance. The lipid mediators data sets for the CO2 group could be predicted with a balanced accuracy of 94%. Also the CO2+MW group had strong predictive performance, reaching more than 89% balanced accuracy. In addition, and not clearly observable in the heat map representation, the supervised analysis showed that the MW group showed decent classification performance (87% balanced accuracy) (Tables 4-2). Similar to lipid mediators, the CO2 group showed very distinct differences in intact lipid concentrations and formed a separate cluster (Figure 4-3a). In contrast to the previous observation with lipid mediators, the CO2+MW, the MW and LPS group could not be clearly separated from each other in the unsupervised inspection of the intact lipid data, which is indicative of high intra-group variation. There was not any significant difference in phospholipid levels upon 5 minutes of hypoxic-ischemia. For example, PI 38:4 showed no differences across all groups (Figure 4-3b). The release of non-esterified ARA, however, caused a 535% increase in lysophposphatidylinositol 18:0 82

96 Table 4-1: Confusion matrix for PLS-DA calculated for lipid mediators. CO2 CO2+MW LPS MW CO CO2+MW LPS MW True values in columns, predicted values in rows. Overall prediction accuracy: 77.2%. Values represent percentages of table totals obtained from repeated stratified cross-validation. Table 4-2: Class-based prediction statistics for PLS-DA calculated for lipid meditators. CO2 CO2+MW LPS MW Sensitivity 88% 83% 40% 95% Specificity 100% 96% 95% 79% Pos. Pred. Value 100% 86% 69% 62% Neg. Pred. Value 96% 95% 84% 98% Balanced Acc. 94% 89% 67% 87% 83

97 A Figure 4-3. Microwave fixation inhibits ischemia-induced changes of intact lipids 3000 B H D F PI 38:4 (pmol/mg of brain tissue) TAG 54:6 (pmol/mg of brain tissue) Cer 36; 1:2 (pmol/mg of brain of tissue) A CO2 CO 2 CO 2 A CO 2 +MW AB CO2+MW CO 2 +MW LPS LPS B LPS MW MW B MW B B B Heat map representation illustrates that CO2 asphyxiation group clusters separately from other groups (A). Brain concentrations relative to internal standard (n=9-10, ± SEM) of phosphatidylinositol (PI) 38:4 (B), lysophosphatidylinositol (LPI) 18:0 (C), triacylglycerol (TAG) 54:6 (D), diacylglycerol (DAG) 38:4 (E), ceramide (Cer) 36:1;2 (F), sphingomyelin (SM) 36:1;2 (G). Bars labeled with different superscripts identify significant differences identified by one-way ANOVA and Tukey s post hoc test of log-transformed concentrations (p<0.05). PLS-DA was performed to elucidate the metabolites driving the separation. A 63.6% overall predictive accuracy was achieved (H). C E G LPI 18:0 (pmol/mg of brain tissue) DAG 38:4 (pmol/mg of brain tissue) SM 36:1;2 (pmol/mg of brain tissue) 1000 A CO 2 A CO2 CO 2 CO 2 +MW B CO2+MW CO 2 +MW B LPS B LPS LPS C MW B MW MW C 84

98 compared to MW (Figure 4-3c). While the release of fatty acids caused no change in the phospholipid pool (Figure 4-3b), smaller pools such as triacylglycerols, such as triacylglycerol 54:6, were decreased in the hypoxic group compared to the MW (Figure 4-3d). Reciprocal increases in diacylglycerol, for example diacylglycerol 38:4, were observed (Figure 4-3e). Ischemia also increased the production of ceramides (Figure 4-3f) while other species, such as sphingomyelin, showed no differences between groups (Figure 4-3g). An overall prediction accuracy of 63.6% (Table 4-3) could be achieved using PLS-DA (see Figure 4-3h showing PLS components 1 and 2). Also here, class-based prediction statistics yielded a very high balanced accuracy for CO2 and CO2+MW (100% and 94% respectively, Table 4-4). Unlike lipid mediators, however, the classification performance for the MW group using intact lipids was poor. The LPS group also had poor classification performance throughout the data sets. To further investigate the classification outcomes, we inspected which variables contributed most to the group separation observed. Tables 4-5 and 4-6 show the top 20 most important variables for the separation of the respective groups in descending order. The values shown are derived from the regression coefficients of the underlying PLS-DA model across the number of PLS components chosen. The importance values are calculated separately for each class and have been scaled between 0 and 100 for better interpretability. High values indicate a high contribution of the given variable for the discrimination. Finally, in order to elucidate the relationship between the intact lipids and the bioactive lipids, we calculated correlation networks within the four phenotypic 85

99 Table 4-3: Confusion matrix for PLS-DA calculated for intact lipids. CO2 CO2+MW LPS MW CO CO2+MW LPS MW True values in columns, predicted values in rows. Overall prediction accuracy: 63.9%. Values represent percentages of table totals obtained from repeated stratified cross-validation. Table 4-4: Class-based prediction statistics for PLS-DA calculated for intact lipids. CO2 CO2+MW LPS MW Sensitivity 100% 89% 32% 34% Specificity 100% 98% 74% 80% Pos. Pred. Value 100% 94% 27% 38% Neg. Pred. Value 100% 97% 78% 77% Balanced Acc. 100% 94% 53% 57% 86

100 Table 4-5: Top 20 lipid mediators in PLS-DA discrimination of the four phenotypic groups CO2 CO2+MW LPS MW 5-oxo-ETE HETE ,15-DHET TXB HETE HDoHE HEPE PGE HDoHE HETE HDoHE PGF2a HEDE keto-F1a PGD ,17-DiHDoHE HDoHE HDoHE* HDoHE ,9-DHET DHET, dihydroxyeicosatrienoic acid; DiHDoHE, protectin D1; ETE, eicosatetraenoic acid; HDoHE, hydroxydocosahexaenoic acid; HEDE, hydroxyeicosadeinoic acid; HEPE, hydroxyeicosapentaenoic acid; HETE, hydroxyeicosatetraenoic acid, PG, prostaglandin; TX, thromboxane 87

101 Table 4-6: Top 20 intact lipids in PLS-DA discrimination of the four phenotypic groups CO2 CO2+MW LPS MW LPI 22: LPI 20: LPI 18: LPC 12: TAG 58: PC 38: LPI 18: LPC 12: Cer 42:3; DAG 36: DAG 36: DAG 34: Cer 36:2; DAG 38: DAG 38: DAG 36: DAG 40: Cer 40:2; DAG 38: Cer 38:2; Cer, ceramide; DAG, diacylglycerol; LPI, lysophosphatidylinositol; LPC, lysophosphatidylcholine; PC, phosphatidylcholine; TAG, triacylglycerol 88

102 groups. To visualize potential relationships between variables, we constructed correlation networks. For better interpretability, we excluded edges with correlation of less than Furthermore, nodes with less than 3 connections were omitted from the network. In addition, cliques of fewer than 10 members were removed leaving only big clusters in the network. Figure 4-4 shows a correlation network for the CO2 asphyxiation group visualized in the BioLayout Express3D software. Lipid mediators are colored in red and are clustered almost exclusively together in the top right hand corner of the plot. The intact lipids they share the strongest interaction with are lipids of the lysophosphatidylethanolamine and lysophosphatidylcholine class. Lysophposphatidylinositol and diacylglycerol formed separate unique clusters, unrelated to the lipid mediators Discussion The purpose of this study was 1) to describe the rat brain lipidome using a method utilizing multiple mass spectrometry approaches and 2) to describe the effect of ischemia on the rat brain lipidome by utilizing high-energy head-focus microwave fixation. The major findings of this study were that ischemia induces an increase in the production of bioactive mediators from lipids released from the phospholipid membrane. Head-focused high-energy microwave fixation reduces this production, while the effects of LPS were much smaller than the effect of ischemia. Following ischemia, ARA is released from the phospholipid membrane in the brain as unesterified ARA, a phenomenon known as the Bazan effect. This unesterified ARA becomes available for metabolism into mediators. It has previously been demonstrated 89

103 Figure 4-4. Correlation network between lipid mediators and intact lipids in the CO2 group. Correlation analysis shows that lipid mediator (red) cluster almost exclusively together and has strongest network connection with lysophosphatidylcholine (light blue) and lysophosphatidylethanolamine (purple). 90

104 270, 274, that ischemia results in a 4- to 20-fold increase in PGE2 concentration in the cortex 275. In this study, we report a similar, though much higher magnitude, increase in PGE2 production. PGE2 concentration was 522-fold higher in the CO2 group compared to the MW group. The higher PGE2 increase in our study may be a result of longer hypoxic periods compared to other studies 270. Similarly, CO2+MW had an increase in PGE2 of 59 fold compared the MW group. Although the production of PGE2 in the CO2+MW group is lower than in the CO2 group, it is still elevated compared to the MW group, suggesting that that microwave fixation does not degrade PGE2. Similar results are seen with the other bioactive mediators measured in this study. The differences between the CO2 and CO2+MW groups may be explained by the differences in ischemia time between the two groups. In the CO2+MW group, animals were exposed to CO2 for 5 minutes prior to microwave fixation. The CO2 group was also exposed to 5 minutes of CO2, however, ischemia continued while the head was placed on ice for 5 minutes (in order to be consistent with the other 3 groups) and for another few minutes in order to remove the brain and place the brain in liquid N2. It is possible that the higher ischemia time in the CO2 group resulted in higher production of bioactive lipid mediators 258, 268. A similar pattern has been reported for other bioactive lipid mediators such as PGD2 270, 276, thromboxane B2 270, 17-hydroxy DHA 277 and arachidonyl ethanolamide 267. Results with these mediators are all in agreement with the results in this study. There were some mediators, such as protectin D1, which were detectable in the CO2 group, but 91

105 were below the detection limit of our equipment in the MW group. This is in agreement with the earlier reports, which failed to detect protectin D1 following head-focused microwave fixation 277, 278. However, it should be noted that Farias and colleague did not measure protectin D1 following 5 minutes of ischemia, which differs from the CO2 group in this study 277. The reason for this disparity is unclear. In this study, we expand on this current list of mediators with several new mediators, including several hydroxyeicosatetraenoic acids and hydroxy DHA, all of which show the same pattern as the other mediators. In order to stimulate the production of some mediators, LPS was injected systematically 3 hours prior to microwave fixation, as described in previous studies 268. Only PGE2 was increased by LPS injection. LPS injection resulted in a 19 fold increase in PGE2 production compared to the MW group. Despite being a significant increase compared to the MW group, this is a much smaller effect than the ischemia effect on PGE2 production. It is therefore possible that this effect of LPS would not be detected in hypoxic animals due to the greater production of PGE2 and the increased variability in concentration, although this was not tested in this study. It should be noted that no other group received any vehicle injection, which could affect interpretation. It is not believed, however, that vehicle injection would result in the release of inflammatory mediators. The increase in bioactive mediators in ischemia can be explained by the Bazan effect, where phospholipases are activated and release lipids from the phospholipid membrane 258. To this date, the effect of microwave-fixation has yet to be reported on intact lipids. Although generally no changes in phospholipid species were observed due to their high concentrations, increases in lysophospholipids, such as 92

106 lysophposphatidylinositol 18:0 were detected in the CO2 group compared to the MW group. This is in agreement with the phospholipases increasing cleavage of ARA in ischemic environments, resulting in higher lysophospholipid production. Due to the smaller size of their pool, the release of fatty acids from triacylglycerols caused by ischemia, reducing triacylglycerol concentration, was inhibited in the MW group. It had previously shown that increasing ischemic time reduced triacylglyceride concentration 258. In parallel, this results in an increase in diacylglycerol. In agreement with this result, diacylglycerol has previously been shown to be reduced by freezing fixation in in vitro neuronal culture 279 and when ischemia time is reduced in vivo 280. Interestingly, degradation and production could be occurring simultaneously in hypoxic brains. It has previously been reported that while ischemic brains had 60 times more 2-arachidonyl glycerol 30 minutes following death, exogenously infused labeled 2- arachidonyl glycerol had decreased by approximately 99% 281. This suggest that although fatty acids are increasing through their release from the phospholipid in ischemia, degradation of fatty acids also appears to be active at a slower rate. It should be noted that these new lipidomic approaches of mass spectrometry have identified novel odd chain fatty acids, which had not previously been measured with older techniques. Targeted studies using standards for these odd chain fatty acids should be used in the future to determine whether these fatty acids truly exists or are only artifacts. This study was the first to attempt to compare the effect of ischemia on the neurolipidome to the neurolipidome of animals euthanized by microwave fixation using a lipidomic approach. Overall, the ischemia-induced neurolipidome is clearly distinct from 93

107 that of the microwave fixed neurolipidome. With the changes induced by ischemia, we were able to determine which lipids are tightly related to one another. Bioactive mediators are closely related to lysophosphatidylcholine and lysophosphatidylethanolamine, suggesting a possible source for these mediators. In summary, we demonstrated a systematic approach to assess the lipidome of the rat brain. While bioactive lipids were all decreased due to microwave fixation, only specific intact lipid pools were either increased or decreased by this fixation method. Moreover, this effect of ischemia is much larger than that of LPS injection and this effect is attenuated with the use of microwave-fixation. The use of microwave fixation decreases variability in measurements, allowing for increased sensitivity in accessing small differences between experimental groups. This study demonstrates the need to consider the effect of ischemia when measuring lipid profile of non-microwaved brain tissue, more specifically postmortem brain tissue. It questions the interpretation that can be achieved with these results Acknowledgements MOT holds a studentship from the Natural Sciences and Engineering Research Council of Canada. RPB acknowledges funding from the Canadian Institutes of Health Research (grant # ) and the Natural Health Science and Research Council of Canada (grant # ), and holds a Canada Research Chair in Brain Lipid Metabolism. MM and DM are supported by the Nestlé Institute of Health Sciences. 94

108 4.7. Author contributions MOT was responsible for sample collection. MOT and DM performed sample preparation and lipid extraction. MOT, MM and RPB designed the study. MM generated and interpreted the data. MM and MOT performed statistical analyses and MM performed data modeling. MOT and MM wrote the manuscript. All the authors contributed to writing the manuscript and approved it Conflict of interest statement There is no competing financial interest 95

109 Chapter 5: N-3 polyunsaturated fatty acids mediate small changes in the resolution of neuroinflammation following intracerebroventricular lipopolysaccharide injection independent of pro-resolving lipid mediators Marc-Olivier Trépanier, Kathryn E. Hopperton, Vanessa Giuliano, Ali Salahpour, Mojgan Masoodi, Richard P. Bazinet Contribution: Along with RPB, I helped design the study. With KEH, I maintained and fed the mouse colony. I performed all the surgeries and collected all samples analyzed in this study. I performed most of the immunohistochemistry presented in this study, along with VG. I travelled to Lausanne to perform the lipid extraction for mass spectrometry analysis. I conducted the Y-maze test and TLDA. I performed the statistical analysis and wrote the first draft of the manuscript. 96

110 5.1. Abstract Resolution of inflammation in the periphery was once thought to be a passive process, but new research now suggests it is an active process mediated by specialized pro-resolving lipid mediators derived from omega-3 polyunsaturated fatty acids (n-3 PUFA). However, this has yet to be illustrated in neuroinflammation. The purpose of this study was 1) to develop a self-resolving model of neuroinflammation and 2) to test whether increasing brain docosahexaenoic acid (DHA) affects the resolution of neuroinflammation. C57Bl/6 mice (Experiment 1) and the fat-1 mice and their wildtype littermates, fed either fish oil or safflower oil (Experiment 2), received lipopolysaccharide (LPS) in the left lateral ventricle. Animals were then euthanized at various time points for immunohistochemistry, gene expression, and lipidomic analysis. They were also tested in the Y-Maze. In Experiment 1, peak microglial activation was observed at 5 days post-surgery and the resolution index was 10 days. Of the approximately 350 genes significantly changed over the 28 days post LPS injection, 130 were uniquely changed at 3 days post injection. While cytokine expression peaks at 24hr post injection, microglial marker expression peaks at 3 days. No changes were observed in the phospholipid and bioactive mediator pools. However, a few lysophospholipid species were decreased at 24hr post surgery. LPS-treated animals did not show deficits in spontaneous alternation performance in the Y-maze at 7 days post LPS injection. When brain DHA is increased (Experiment 2), microglial cell density resolves slightly faster. In terms of gene expression, only COX-2 mrna expression is affected by 97

111 increasing brain DHA. In conclusion, resolution of neuroinflammation appears to be independent of specialized pro-resolving lipid mediators. Increasing brain DHA has a small effect in this model. This model may be more appropriate for a pharmaceutical approach. 98

112 5.2. Introduction Neuroinflammation is a characteristic of many neurological and psychiatric disorders. In vivo and postmortem studies have both reported increased neuroinflammation in Alzheimer s disease, Parkinson s disease, and schizophrenia 116, 117. Growing evidence is suggesting a potential causal effect of neuroinflammation in the progression of the pathogenesis of neurological and psychiatric disorders 116, 120. The brain is an immunologically privileged tissue, blocking most peripheral immune cells from entry 121. It contains its own resident immune cell in the microglia. Microglia survey the environment, and communicate with other glia and neurons. Following insults or tissue damage, microglia are activated to an M1 phenotype, releasing pro-inflammatory cytokines such as tumor necrosis (TNF)- and interleukin (IL)-1 119, 123. These signals activate astrocytes to release more pro-inflammatory cytokines including IL-1. When chronic inflammation persists, neuronal death ensues. Microglia, however, can be also be activated by IL-4 and IL-13 to a M2 phenotype, which releases anti-inflammatory cytokine IL-10 and growth factors such as insulin growth factor and transforming growth factor 119, 123. Classically, it was thought that inflammation dissipated passively. It is becoming clear, however, that the resolution of inflammation is an active process 42, 282. Resolution of inflammation in the periphery is driven by specialized pro-resolving lipid mediators derived from the enzymatic oxygenation of polyunsaturated fatty acids (PUFA) 42, 44. Specialized pro-resolving lipid mediators are considered both anti-inflammatory and proresolving. In the periphery, specialized pro-resolving lipid mediators actively return the inflamed tissue to homeostasis by blocking neutrophil entry and activating the 99

113 recruitment of macrophages to the tissue to repair and clear debris. While omega (n)-6 PUFA are typically considered pro-inflammatory due to the production of prostaglandins and leukotrienes, they also produce the specialized pro-resolving mediator lipoxins. Through the enzymatic activity of lipoxygenase, n-3 PUFA also produce specialized proresolving lipid mediators including protectins, resolvins, and maresins. Due to the differences between inflammation in the periphery and neuroinflammation, it is unknown, however, whether resolution of neuroinflammation utilizes the same mechanism. In the brain, n-3 PUFA make up approximately 10% of all lipids 278, 283. Docosahexaenoic acid (DHA) is the most abundant n-3 PUFA in the brain and is involved in regulating neuronal and glial structure, while also producing signaling molecules. Due to its abundance and multiple functions in the brain, it is not surprising that a link between n-3 PUFA and both neurological and psychiatric disease has been proposed and investigated 242, Observational studies have suggested a protective role of n-3 PUFA in multiple brain disorders, such as Alzheimer s disease and depression 139, 288, , 290, 291. The results from clinical trials, however, are conflicting with only a few studies pointing to a protective effect 242, 243, 292. There have been several mechanisms proposed for the protective effects of n-3 PUFA in neurological and psychiatric disorders. These include anti-apoptotic, neurotrophic, and anti-oxidative mechanisms 293. Another potential mechanism of n-3 PUFA involves their anti-neuroinflammatory actions 293. N-3 PUFA have antiinflammatory properties in a multitude of disease models including stroke, spinal cord injury, Alzheimer s disease and Parkinson s disease (For review, see Chapter ). 100

114 Increased brain DHA, either through dietary intervention or in the fat-1 mouse, has decreased pro-inflammatory gene expression 24 hours following i.c.v. injection of lipopolysaccharide (LPS). Moreover, i.c.v. injection of 17S-hydroperoxyDHA, a precursor of protectin D1, had a more potent effect than DHA itself, suggesting that some or all of the anti-neuroinflammatory effects of DHA may be mediated by its metabolism to protectin D1 91. This is consistent with the anti-neuroinflammatory effects of protectin D1, aspirin-triggered resolvin D1, resolvin E1, and resolvin D2 in stroke 170, 241, Parkinson s 295, traumatic brain injury 212 and neuropathic pain models 216, 217. Despite the fact that animal studies have generally pointed to antineuroinflammatory properties of n-3 PUFA, not much is known regarding their effects on resolution, as most studies have evaluated only a few pro-inflammatory markers at one time point 294. It is therefore possible that the effects of n-3 PUFA may have been missed if the wrong marker or time point was chosen. The goal of this study was first to develop a self-resolving model of neuroinflammation following i.c.v. LPS over 28 days utilizing microarray and lipidomic approaches (Experiment 1). Once developed, the second goal of this study was to determine whether resolution of neuroinflammation is influenced by increasing brain DHA (Experiment 2) Methods The present experiments were conducted in accordance with the standards of the Canadian Council on Animal Care and were approved by the Animal Care Committee of the Faculty of Medicine of the University of Toronto. Animals were housed 1-4 per cage 101

115 in our animal facility where temperature (21 C) and light (14/10 light/dark cycle) were controlled. Food and water were available ad libitum Diets Animals were fed one of 3 diets, 1) rodent chow (Teklad Global Diets, Envigo, Madison, WI), 2) 10% safflower oil (D , Research diets, New Brunswick, NJ), or 3) 8% safflower and 2% menhaden oil (D , Research Diets, New Brunswick, NJ). Diet fatty acid composition was confirmed in triplicate by gas chromatography flame ionization detection and is presented in Table 5-1. The rodent chow contained 18.9% oleic acid, 56.4% linoleic acid and 6.5% -linolenic acid as a % of total fatty acids. Eicosapentaenoic acid (EPA) was not detected by gas chromatography flame ionization detection in the rodent chow diet, while a trace amount of DHA was detected. Gas chromatography with mass spectrometry, however, was not able to detect any DHA in the rodent chow diet. The safflower diet contained 13.3% oleic acid, 67.3% linoleic acid and 0.2% - linolenic acid as a percent of total fatty acids. Similar to rodent chow, a trace amount of DHA was measured by gas chromatography flame ionization detection. Gas chromatography with mass spectrometry confirmed that the DHA percent composition was 0.01%. The trace amount of EPA is also believed to be artifact, although it was not confirmed by gas chromatography with mass spectrometry. The fish oil diet was composed of 5.0% oleic acid, 58.8% linoleic acid, and 0.43% 102

116 Table 5-1: Percent of total fatty acids of the 3 experimental diets. Fish oil Safflower oil Chow C12: C14: C14: C16: C16: C18: C18:1 n C18:1 n C18:2 n C18:3 n C18:3 n C20: C20:1 n C20:2 n C20:3 n C20:4 n C20:3 n C20:5 n C22: C22:1 n C22:5 n C22:5 n C22:6 n-3* N.D C24:1 n *composition was confirmed by gas chromatography with mass spectrometry 103

117 -linolenic acid as a percent of total fatty acids. It was composed of 1.5% and 1.4% EPA and DHA respectively Subjects C57Bl/6 male mice were ordered at 10 weeks of age from Charles River (Saint Constant, Qc). Animals were fed rodent chow and allowed to acclimatize for 2 weeks. Fat-1 male mice (C57Bl/6 X C3H background) 102 were graciously donated by Dr. David Ma (University of Guelph) for breeding purposes. Subject mice were obtained by mating male fat-1 mice with C57Bl/6 females from Charles River (Saint Constant, Qc, Canada). Females were fed safflower oil diet two weeks prior to being placed in harems. Pups were genotyped at 2-3 weeks of age prior to weaning as described before 91. F1 male progeny were used as experimental subjects. At 3 weeks of age, wildtype pups were weaned and either maintained on safflower diet or placed on the fish oil diet. Fat-1 pups were placed only on safflower diet as previous work in our laboratory has shown that fish oil does not further increase brain DHA Intracerebroventricular LPS injections At 12 weeks of age, subjects were anesthetized by isofluorane (3% induction, 2% maintenance). The head was secured in a stereotaxic apparatus (Stoelting, Wood Dale, IL, USA) and 150 l of 0.03% sensorcaine was injected s.c. at the incision site. Following the incision and exposing the skull, a small hole was drilled (-1.0 mm medial/lateral, -0.5 mm anterior/posterior). A 33g needle was lowered into the left 104

118 ventricle (-2.4 mm dorsal/ventral) and LPS (5 g in 5 l, E.coli serotype 055:B5, Sigma Aldrich, St-Louis, MO, USA) was infused over 5 minutes by electronic pump (Stoelting, Wood Dale, IL, USA). The needle remained in the left ventricle for 25 minutes post infusion to ensure LPS diffused within the ventricle. The needle was removed slowly to avoid infusate backflow. The skull was closed with bone wax and sutured. Mice were euthanized at 4hr, 8hr, 12hr, 1, 2, 3, 5, 7, 14, and 28 days following surgeries. Nonsurgery animals were used throughout the study as controls. The accuracy of the LPS injection was sporadically checked with Evan s blue injection Immunohistochemistry For immunohistochemistry, mice were anesthetized by i.p. injection of Avertin (20 ml/g, 250 mg 2,2,2 tribromoethanol, 0.5 ml 2-methyl-2-butanol, 20 ml dh2o, Sigma Aldrich, St-Louis, MO, USA) and were euthanized by transcardiac perfusion at 4hr, 12hr, 1, 3, 5, 7, 14 and 28 days (n = 6-10 per group). Approximately 12 ml phosphate buffered saline was infused by peristaltic pump (GE Healthcare, Mississauga, ON, Canada) followed by 18 ml of 4% paraformaldehyde. Brains were post-fixed overnight in 4% paraformaldehyde and dehydrated in 30% sucrose on the following day. Brains remained in sucrose until frozen in cryostat sectioning medium prior to slicing. Brains were sectioned in 40 m slices in a Leica cryostat (CM 1510S, Concord, ON) Anti-ionized calcium-binding adapter molecule (Iba) 1 was utilized as a microglia marker. Slices were quenched with 0.5% sodium borohydride and washed with 3 phosphate-buffered saline washes. Slices were then blocked for 2 hr in a blocking solution (10% normal goat serum, 0.75% bovine serum albumin, 0.1% Triton-x). Anti- 105

119 Iba (1:2000, Wako Chemicals, Richmond, VA, USA) was applied overnight. Slices were labeled with Alexa Fluor 680 (1:2000, Life Technologies, Burlington, ON, Canada) for 1 hr the next day. Slices were imaged on a LI-COR Odyssey (settings: Resolution, 21; Quality, Highest; Intensity, 4; Lincoln, NE). Optical density of images was recorded using ImageJ (version 1.46R. Bethesda, MD). Differences from baseline were calculated by comparing subjects receiving LPS to non-surgery animals analyzed on the same day to avoid methodological variation. Cell counting was conducted using an epi-fluorescence microscopy in order to see possible regional differences. Iba1 reactive cells (secondary Alexa Fluor 568, Life Technologies, Burlington, ON, Canada) were counted as described previously 296 using Nikon Elements software (NIS-Elements Basic Research, version 3.10) at 10X magnification. Using automatic exposure, the fluorescent intensity thresholds limits were automatically determined and were set to fall within the linear range. Counts were completed by a blind observer Genetic expression analysis For gene expression analysis, animals were euthanized by CO2 asphyxiation at 8 hr, 1, 2, 3, 7, 14 and 28 days following LPS surgery (n=8 per group). The left hippocampus was dissected and frozen by liquid N2. For the microarray analysis for Experiment 1, RNA was extracted using an Agencourt RNAdvance Tissue Kit (Beckman Coulter, Inc.). The quality of total RNA was checked using the BioAnalyzer 2100 with Total RNA Nano kit (Agilent Technologies, Santa Clara, CA). Quantification was done using the Quant-iT RiboGreen RNA Assay Kit assay (Life Technologies, Inc.). 300ng of RNA was reversed transcribed 106

120 and 750 ng of crna was loaded unto a MouseRef-8 v2.0 Expression BeadChip (Illumina, San Diego, CA), which contains approximately 25,600 novel transcripts and measures over 19,100 separate genes. For the Taqman Low Density Array analysis in Experiment 2, total RNA was extracted using Trizol (Thermo Fisher Scientific, Waltham, MA) following the manufacturer s instructions. Samples were stored at -80 C. RNA quantity and quality (OD230/260, OD260/280) were measured by a Nanodrop 1000 Spectrophotometer (Nanodrop Technologies, Wilmington, DE, USA). Random samples were sent for analysis by BioAnalyzer 2100 to confirm RNA quality. RNA was reverse transcribed using a High Capacity cdna Reverse Transcription kit (Thermo Fisher Scientific, Waltham, MA). 150 ng of cdna in 50 l of RNase free water was combined with 50 l Taqman Fast Advance Mastermix (Thermo Fisher Scientific, Waltham, MA) in Taqman Low Density Array wells as instructed by the manufacturer s instructions (Thermo Fisher Scientific, Waltham, MA). Plates were custom designed with 45 separate assays including, microglial markers Iba1 (assay ID Mm _g1), translocator protein (TSPO, assay ID Mm _m1), Cluster of Differentiation 86 (CD86, assay ID Mm _m1), CD68 (assay ID Mm _m1), arginase 1 (arg1, Mm _m1), triggering receptor on myeloid cell 2 (Trem2, assay ID Mm _g1), CD11b (Mm _m1), and CD206 (Mm _m1), cytokines and chemokines IL-10 (assay ID Mm _m1), IL-1 (assay ID Mm _m1), TNF (assay ID Mm _m1), chemokine (c-c motif) ligand 5 (CCL5, assay ID Mm _m1), and chemokine (c-x-c motif) ligand 1 (CXCL1, assay ID Mm _m1), astrocytic markers glial fibrillary acidic protein (GFAP, 107

121 assay ID Mm _m1) and S100 calcium-binding protein b (S100b, assay ID Mm _m1), arachidonic cascade markers cyclooxygenase-2 (COX-2, assay ID Mm _m1), prostaglandin E synthase (PTGES, assay ID Mm _m1), cytosolic phospholipase A2 (cpla2, assay ID Mm _m1), nuclear factor kappalight-chain-enhancer of activated b cell (NF- B) pathway markers NFKB1 (assay ID Mm _m1), transcription factor p65 (Rela, Mm _m1), and NF- B inhibitor, alpha (I B, assay ID Mm _m1), fatty acid and specialized proresolving mediator receptors free fatty acid receptor 4 (GPR120, assay ID Mm _m1), chemokine like receptor 1 (ChemR23, Mm _s1), and peroxisome proliferator-activated receptor gamma (PPARg, assay ID, Mm _m1), fatty acid metabolising enzymes cytochrome p450 1b1 (cyp1b1, assay ID Mm _m1), 5-lipoxygenase (ALOX5, assay ID Mm _m1), ALOX12 (assay ID Mm _m1), ALOX15 (assay ID Mm _m1), BBB marker matrix metallopeptidase 9 (MMP9, assay ID Mm _m1) and other makers identified in the microarray of Experiment 1 including S100a8 (assay ID Mm _g1), S100a9 (assay ID Mm _m1), intracellular adhesion molecule (ICAM-1, ID Mm _m1), chitinase like 1 (Chil1, assay ID Mm _m1), vascular endothelial growth factor A (VEGFA assay ID Mm _m1), lipocalin 2 (LCN2, assay ID Mm _m1), serum amyloid A3 (SAA3, assay ID Mm _m1), interferon-induced guanylate-binding protein 2 (GBP2, assay ID Mm _g1), interferon-induced protein with tetratricopeptide repeats 3 (IFIT3, assay ID Mm _s1), interferon-induced transmembrane protein 3 (IFITM3, assay ID Mm _s1), Fas (assay ID Mm _m1),

122 oligoadenylate synthase-like 2 (OASL2, assay ID Mm _m1), tissue inhibitor of metalloproteinase (TIMP1, assay ID Mm _m1), serine peptidase inhibitor clade A, member 3N (SERPINA3N, assay ID Mm _m1), lysozyme 1 (Lyz1, assay ID Mm _m1), and inducible nitric oxide synthase 2 (NOS2, assay ID Mm _m1). Plates were analyzed on a ViiA 7 real time PCR machine (Thermo Fisher Scientific, Waltham, MA) Lipidomic analysis To measure lipid changes following i.c.v. LPS, we utilized our lipidomic approach developed in Chapter 4 in order to eliminate the ischemia-induced changes to the neurolipidome. Following LPS surgery, animals were euthanized at 1, 3, 7, 14, and 28 days post surgery (n = 6 per group). Animals were gently inserted in the mouse restrainer and placed inside the microwave (Cober Electronics Inc., Norwalk, CT, model S15P Vivostat) where a single high-energy microwave beam was focused directly on top of the skull (approximately 1,900 J, 0.5 kw, 2,450 MHz). Brains were quickly excised and the left hippocampus was dissected from the remainder of the brain. The left hippocampus was flash frozen by liquid N2 and stored at -80 C until analysis Bioactive mediator extraction The whole left hippocampus (approximately 12 mg of tissue) was homogenized in 1 ml of 15% methanol by Tissue Lyser (Qiagen AG, Switzerland) at a speed of 25 Hertz for 2.5 min. 100 l of the homogenate was collected for intact lipid analysis. 109

123 The remaining homogenate was used for bioactive mediator analysis using a method similar to Chapter l of 100% methanol was added to the remaining homogenate and spun at approximately 25,000 g (5430 R centrifuge, FA HS rotor) (Eppendorf AG, Hamburg, Germany) for 5 min at 4 C. Supernatant was removed into new glass tubes on ice. One ml of 15% methanol was added to the pellet and homogenized by Tissue Lyser (25 Hz, 2.5 min). The homogenate was spun (25,000g, 5 min, 4 C.) and supernatant was added to the glass tube. One ml of 15% methanol was used to make a final volume of 3 ml. Extraction of lipid mediators from the brain tissue was performed according to a previously published protocol 254 with slight modifications outlined as follows: internal standards PGB2-d4 (40 ng), 12-hydroxyeicosatetraenoic acid-d8 and arachidonyl ethanolamide-d8 (Cayman Chemicals, Ann Arbor, MI, USA) were added to the homogenized brain in 15% (v/v) methanol in water. The cartridges (Strata-X 33 u Polymeric Reversed phase 60 mg /3 ml) were washed with methanol (3 ml) followed by water (3 ml) prior to loading the homogenate (3 ml). The cartridges were then washed with 15% methanol in water (3 ml) and lipid mediators were eluted in methanol (3 ml) and collected in glass tubes. The organic solvent was evaporated using a fine stream of nitrogen and the remaining residue was re-dissolved in ethanol (100 μl) and stored at 20ºC prior to analysis Extraction of intact lipids from the brain The remaining 100 l of the homogenate used for intact lipid analysis was further diluted with 160 l of ammonium bicarbonate buffer using a Hamilton Robot and 810 l 110

124 of MTBE /methanol (7/2 v/v) containing internal standard which was added to this mixture. The internal standard mixture contained: lysophasphatidylglycerol 17:1, lysophosphatidic acid 17:0, phosphatidylcholine 17:0/17:0, phosphatidylserine 17:0/17:0, phosphatidylglycerol 17:0/17:0, phosphatidic acid 17:0/17:0, lysophposphatidylinositol 13:0, lysophosphatidylserine 13:0, lysophosphatidylcholine 12:0, lysophosphatidylethanolamine, cholesteryl D6, diacylglycerol 17:0/17:0, triacylglycerol 17:0/17:0/17:0, ceramide 18:1;2/17:0, sphingomyelin 18:1;2/ 12:0, phosphatidylethanolamine 17:0/17:0, cholesteryl ester 20:0, phosphatidylinositol 16:0/16:0. The solution was mixed at 700 rpm, 15 min at 4 C using a ThermoMixer C (Eppendorf AG, Hamburg, Germany) and then centrifuged at 3,000 rcf for 5 min. 100 l of the organic phase was transferred to a 96-well plate, and dried in a speed vacuum concentrator. Lipid extract was reconstituted in 40 µl of 7.5 mm ammonium acetate in chloroform/methanol/propanol (1:2:4, V/V/V). All liquid handling steps were performed using a Hamilton STAR robotic platform with the Anti Droplet Control feature for organic solvents pipetting as described previously Mass spectrometry analysis Lipidomic analysis of intact lipids was performed using a QExactive mass spectrometer (Thermo Fisher Scientific) equipped with a TriVersa NanoMate ion source (Advion Biosciences) as described previously 271. The data were acquired in both positive and negative mode using resolving power of 140,000 in full scan and 17,500 in tandem mass spectrometry mode. Scan mass charge ratio (m/z) range from 200 to 1,

125 Lipidomics analysis of bioactive lipid mediators was performed on an LTQ Elite (Thermo Scientific) linear ion trap-orbitrap mass spectrometer using a heated electrospray ionization source in both negative and positive ionization mode. Capillary and source heater temperatures were set to 325 C and 50 C, respectively, and spray voltage was adjusted to 4,000 V. A resolving power of 120,000 was used in full scan and 1,500 in tandem mass spectrometry mode. Scan m/z ranges of 150 to 500 (mass spectrometry) and 50 to 500 (tandem mass spectrometry) were used Total lipid extraction For lipid analysis, total lipids were extracted from the rest of brain (from brains used for gene expression analysis) into 6 ml of chloroform / methanol (2:1 v/v), using 1.75 ml of 0.88% KCl to separate the aqueous phase. Non-esterified heptadecanoic acid (Nu Chek Prep, Elysian, MN) in hexane was added as an internal standard. Brains were homogenized by glass homogenizer. This was followed by a wash with 4 ml of chloroform. The total lipid extract was then dried under nitrogen and reconstituted in 4 ml of hexane. Ten percent of total lipids were then methylated in 14% methanolic BF3 (2 ml) and hexane (2 ml) at 100 C for 1 hour. The samples were allowed to cool at room temperature for 10 minutes and then centrifuged at 1460 rpm for 10 minutes following the addition of deionized water (2 ml). The upper hexane layer was extracted, dried under nitrogen and reconstituted in 1 ml of hexane 112

126 Fatty acid methyl ester analysis by gas-chromatography for Experiment 2 Fatty acid methyl esters were analyzed on a Varian-430 gas chromatograph (Varian, Lake Forest, CA, USA) equipped with an Agilent capillary column (DB-23; 30 m x 0.25 mm i.d. x 0.25 µm film thickness, Santa Clara, Ca). One μl of fatty acid methyl esters was injected in splitless mode. The carrier gas was helium, set to a constant flow rate of 0.7 ml/min. The injector and detector ports were set at 250 o C. Fatty acid methyl esters were eluted using a temperature program set initially at 50 o C for 2 minutes, followed by a ramp-up at 20 o C/min to 170 o C, a hold at 170 o C for 1 minute, and an increase of 3 o C/min to 212 o C and a hold at 212 o C for 5 minutes. Peaks were confirmed by identifying the retention times of authentic fatty acid methyl ester standards of known composition (Nu-Chek Prep, Elysian, MN). Fatty acid concentrations (nmol/g of brain tissue) were calculated by proportional comparisons of the gas chromatography peak areas with that of the heptadecanoic acid internal standard Y-maze Seven days following surgery on mice naïve to the test (n=19), working memory was measured using a standard Y-maze. Non-surgery mice served as controls (n=18). The Y-maze consisted of 3 arms of the same length meeting at 120 at the centre (38X7.6X12.7 cm 3, San Diego Instruments, San Diego, CA). Each arm was defined as a zone (Zone A, B or C), from the end of the arm up to 5 cm from the centre of the maze. Subjects were introduced in zone A and were allowed to navigate for 8 minutes. Movement was recorded using the video-based tracking software Biobserve Viewer2. Spontaneous alternations were recorded when the subject entered each arm of the maze 113

127 consecutively before entering an arm previously entered (Either ABC, ACB and CAB). The spontaneous alternations performance was calculated by the following equation: total spontaneous alternation/(total arm entries -2)X Statistics Results are expressed as mean standard error of the mean (SEM). For the immunohistochemistry, differences between groups were measured by one-way ANOVA (Experiment 1) and two-way ANOVA (Experiment 2), with Tukey s post hoc analysis. Modified from the definition by Serhan et al. 54, the resolution index (Ri) was defined as the time between the time of maximal microglial activation to the time of 50% maximal inflammation of the untreated group (WT safflower fed mice, WTSO). Linear regression from the maximal inflammation to the return to baseline was performed and the slope and x-intercepts were calculated as further resolution indices. For the microarray analysis, a one-way ANOVA was performed on Log2 transformed, quantile normalized data, followed by a Benjamini-Hochberg correction with a cutoff of 0.01 to identify the number of genes significantly expressed differently compared to non-surgery controls. Specific pro-inflammatory and lipid metabolism genes of interest were selected for analysis by one-way ANOVA and Tukey s post hoc analysis to evaluate the time course of their expression. Differences between time points for different lipid species were evaluated by one-way ANOVA and Tukey s post hoc analysis. The difference between the 2 groups in the Y-maze was analyzed by Student s t-test. A two-way ANOVA was utilized to detect any differences between treatment groups and time in respect to delta CT measured by Taqman low-density array. No differences in analysis were observed 114

128 between the three reference genes (PGK1, beta-actin, and 18S). Data were presented as fold change of baseline using PGK1 as the reference gene Results Experiment Microglial activation peaked by 5 days and resolved by 21 days, independent of neutrophil and macrophage infiltration In order to define resolution of neuroinflammation, C57Bl/6 mice were euthanized at various time points following i.c.v. LPS surgery. LPS was directly injected in the left lateral ventricle in order to minimize systemic inflammation created by the injection of LPS in the periphery. We observed an initial increase in Iba1 labeling by immunohistochemistry at 24 hours following LPS injection. Microglia labeling continued to increase up to 5 days (Tmax) and was reduced by half (T50) at day 15 (Figure 5-1A-E). In order to define resolution, the Ri was calculated to be 10 days. We also calculated the slope of the fitted line from the maximal point (Day 5) to the return to baseline (Day 21). The slope was calculated to be Fold change/day, while the x-intercept was calculated to be days (Figure 5-1A insert). 115

129 B A Fold Change From Baseline T max = 5d R i = 10d Fold Change From Baseline Days T 50 = 15d Days Slope = x-intercept = C D E Figure 5-1. Time course of Iba1 optical density in the hippocampus in the C57Bl/6 mouse following i.c.v. LPS. Maximal microglial activation was found at day 5. Microglial activation was reduced by half at day 15, and the resolution index was calculated to be 10 (n=6-10, SEM) (A). Examples of Iba1 labeling as measured by LI-COR imaging representing nonsurgery (B), 3 days post surgery (C), 7 days post surgery (D), and 28 post surgery animals (E) are shown above. Linear regression from time of maximal optical density to return to baseline (insert) illustrates an alternative resolution index. 116