As I work on my dissertation, reflecting on how few people will read it, I realize that no one will read my dissertation proposal ever again, as they are not published but only exist on the hard drives of your dissertation committee until they chose to delete it. So I am posting it here, in the hopes that it serves as an education tool on the Ph.D granting as well as the literature review process. It also places a time stamp on some of my ideas (this was written in 2013) in order to track their evolution.
“Meanwhile, have a look at my apprentice work, if you dare.” – George R. R. Martin, Dreamsongs Volume I
The conditions of overweight and obesity have many iterations of proposed etiology including those nutritional, psychological and sociocultural in origin. There now exists a general consensus in obesity research that common, or diet-induced obesity (DIO) stems from a central inflammatory condition impairing energy status sensing and regulatory neuropeptide secretion in the hypothalamus. This disturbed signaling milieu has effects peripherally and leads to positive energy balance through increased energy intake and/or decreased energy expenditure. Decades of rodent research have shown that a high fat diet (HFD) initiates and supports this obesogenic process. What is not agreed on is which type(s) of fatty acids are most effective in producing DIO; some studies focus on the saturated fatty acids (SFA) and others the polyunsaturated fatty acids (PUFA), specifically the n-6 PUFA, as being more potently obesogenic through several potential mechanisms. We want to identify the most potent obesogenic factor(s) within commonly used experimental HFDs by comparing the effects of isocaloric and isolipidic HFDs of different fatty acid profile in wild type (WT) C57BL/6J mice.
Also observed in obesity research are sex related patterns of adiposity, e.g. subcutaneous vs. visceral in women and men, respectively. These two types of obesity have different associated comorbidities; visceral is more strongly correlated to metabolic syndrome and heart disease, but it is not fully understood how origin of obesity may be dependent on sex differences in physiology. We want to understand the role of sex hormones, specifically 17β-estradiol (E2), in the adipogenic effect of HFD feeding.
Lastly, maternal nutrition has been shown to influence offspring bodyweights. Both maternal under and over nutrition seem to have negative metabolic consequences on offspring. Like adult common obesity, it is unknown whether a maternal HFD high in SFA or in n-6 PUFA would exert a stronger obesogenic influence on offspring. Therefore, my central hypothesis is that n-6 PUFA exert a greater obesogenic influence through the hypothalamic energy balance system. Differences in obesogenic effects to high n-6 PUFA diets by sex and maternal effects will also be observed. To address this hypothesis, we will determine: 1) whether a high SFA or n-6 PUFA semi-purified mouse diet produces a stronger dysregulation of the hypothalamic energy balance system as measured by increased adipose tissue, inflammatory signaling and altered central and peripheral endocrine activity; 2) whether the reproductive steroid E2 abrogates, in part, this central dysregulation and DIO and 3) whether maternal HFDs high in either SFA or n-6 PUFA produce more obese offspring with greater metabolic derangement.
AIM 1: Determine if high fat diets with different fatty acid profiles (high SFA versus high n-6 PUFA) differentially produce obesity through modulation of hypothalamic inflammation. We will test the hypothesis that a high n-6 PUFA diet will more strongly induce the altered central and peripheral states associated with DIO than a high SFA diet in intact male mice. Isocaloric and isolipidic diets will be designed with a high n-6 PUFA or SFA content. We will measure body weight and composition, food intake, glucose homeostasis, metabolic rate, protein and mRNA associated with inflammatory cytokines in the periphery and centrally and expression of metabolism related neuropeptides in the hypothalamus. The results of this aim will help explain the mechanism of DIO produced by HFD and provide evidence as to the more potent fatty acid species in causing chronic positive energy balance.
AIM 2: Determine if E2 has differential effects on the obesogenesis of high SFA or high n-6 PUFA diet in females. We will test the hypothesis that females will respond to DIO from our high SFA or high n-6 PUFA diets and the effects of hypothalamic inflammation and the dysregulation of energy homeostasis in an E2-dependent manner. We will measure hypothalamic levels of inflammatory cytokines and regulatory neuropeptide mRNA as well as body weight and composition, food intake, glucose homeostasis and metabolic rate in female mice following ovariectomy with or without E2 treatment and given high SFA or high n-6 PUFA diets. The results of this aim will contribute to our understanding of the influence of sex steroids on the differences in the hypothalamic control of energy homeostasis under the physiological challenge of DIO.
AIM 3: Determine the metabolic effects on offspring of dams fed high SFA or high n-6 PUFA diet. We will test the hypothesis that an HFD high in n-6 PUFA fed maternally will cause more metabolic disturbance and obesity in control and HFD fed offspring than an HFD high in SFA. We will measure body weight and composition, food intake, glucose homeostasis, metabolic rate, inflammatory cytokine serum levels and gene expression of neuropeptide genes and inflammatory factors in the hypothalamus. The results of this aim will increase the understanding of maternal influence on obesity and metabolism by focusing on fatty acid type, a hitherto ignored nutritional variable in trans generational obesity research.
Progression of obesity through the past century
The first few decades of the American 20th century saw a virtual eradication of the malnutrition related disease, such as entero and bronchial infection, that was responsible for the deaths of approximately one third of all live births in the 19th century (1). Temporarily subsequent with this health advance was the appearance and progressive increase of overweight and obesity. Body weight as a premium-adjusting criterion could be found on life insurance actuarial sheets in the 1930s; by the 1960s government reports were appearing on the link between obesity and cardiovascular disease (2). Overweight and obesity prevalence continued to accelerate until the turn of the 21st century when the figures, although still increasing, tapered off in their rate of increase (3, 4). Currently, obesity prevalence exceeds one third in most age groups, with overweight at more than 50% (4–6). This increased adiposity has brought with it the “metabolic syndrome” and its associated comorbidities such as diabetes mellitus and heart disease (7). Correlation of bodyweight versus mortality show that a BMI of 30-35, considered moderately obese, is associated with a 2-4 year reduction in lifespan; severe obesity, a BMI of 40-45, reduces expected lifespan by 8-10 years (8). Despite the considerable resources brought to bear against this phenomenon, sustained weight loss through diet, exercise or both has been largely ineffective at the clinical and population levels (9–11). Misconceptions of the etiology and biological nature of obesity may underlie the apparent failure of its treatment. Obesity was largely considered a psychological condition of insufficient self-control and adipocytes were thought of as inert storage depots before the characterization of the “obese,” or ob, gene leptin in 1994 (12). Although now fashionable to think of white adipose tissue as an endocrine organ (13), and with many proposed physiological causes of obesity (14), the inability to reverse the trend suggests the absence of information, or the presence of misinformation, fundamental to its understanding.
Diet-induced obesity animal models
Most basic energy balance research is performed on animals with a relative minority being clinical human experimentation. The most widely used animal models are rodents with either a genetic (ob, db, Zucker, Tubby etc.) or environmental (DIO) obesity causing factor (15). This proposal is concerned with DIO rodent models of “common” human obesity rather than the rarer monogenic obesity occasionally seen in notable clinical cases.
Within DIO research, HFDs are the most commonly used to induce obesity (15). These diets can be so widely varied in both fatty acid contribution to kilocalories (kcals) and fatty acid profile as to render results between studies practically incomparable (16, 17). Although there has been a move away from supplementation of chow diets with extra fat and towards semi-purified HFDs of a more standard composition (17) there are still serious, and maybe even less tractable, problems interpreting experimental HFD data. One of the most common sources of fat in semi-purified HFDs is lard. Although the fatty acid profile of ruminant animal tissues is influenced by diet (18) the fat of pigs, like humans, is almost wholly representative of their dietary intake (19). Unless otherwise noted by a manufacturing company or lab, the fatty acid profile of lard and other fat sources is reported using the values from a foods database such as the USDA’s. The USDA database currently reports lard as having just under 18% of its fat as PUFA (20). Independent analysis in 2011-2012 by Research Diets, a provider of semi-purified rodent experimental diets, found the PUFA content of their lard to be approximately 26% (21). The difference is almost entirely from linoleic acid (LA) at the expense of oleic and palmitic acids, causing the relative contribution of kcals from LA to double in Research Diets’ D12451 and D12492 HFDs as well as HFDs from other companies using these Open Source formulas (21, 22). This makes it likely that many rodent DIO studies have been under-reporting the PUFA content and over-reporting the MUFA and SFA content of their semi-purified experimental HFDs. This misreporting has important implications as more and more articles describe distinct fatty acid species as having a particular pharmacological effect in the generally agreed upon mechanism of DIO obesity, hypothalamic inflammation (HI) (23–27).
Hypothalamic Inflammation as a Molecular Mechanism of Diet-Induced Obesity
The hypothalamus of the brain is the main energy balance center (28) and ubiquitous in DIO rodent models is neuroinflammation of that brain region (29–31). Seated atop a fenestrated capillary bed that bypasses the blood-brain barrier, the Arcuate nucleus (ARC) of the mediobasal hypothalamus contains two important populations of neurons for energy balance, pro-opiomelanocortin/cocaine and amphetamine related transcript (POMC/CART) producing and neuropeptide Y/Agouti-related peptide (NPY/AgRP) producing neurons (32). The anorexigenic POMC/CART and orexigenic NPY/AgRP neurons receive neural, peptide (insulin, leptin, adiponectin etc.) and nutrient (circulating glucose and free fatty acids) borne energy status information. In addition to being well situated for monitoring the periphery these neuronal populations have crosstalk with others, innervating and being innervated by neurons of the lateral, ventromedial, paraventricular and dorsomedial nuclei of the hypothalamus (32). The whole of these neuronal interactions with the periphery and with each other, deemed the “first order” energy sensing system, integrates energy status for relay downstream. The first order system acts agonistically or antagonistically towards the “second order” energy sensing system of melanocortin receptors 3 and 4 (MC3R and MC4R) (33). The stimulation or inhibition of that system suppresses food intake and increases energy expenditure or vice versa (34–36). Inflammatory signaling can influence parts of these energy-sensing complexes.
Hypothalamic inflammation has been suggested, in different studies, to have a variety of causes. Activation of pattern recognition receptors (PRR) (24) of the innate immune system and related reactive oxygen species (ROS) production (37) have been shown sufficient to initiate HI. Increased cytokine/chemokine activity can impair insulin signaling to POMC and NPY neurons through interference with the signaling cascade of insulin receptor substrate (IRS) to phosphoinositide-3-kinase (PI3K) (38). Similarly, both insulin and leptin signaling are blocked by expression of suppressor of cytokine signaling 3 (SOCS3) and its inhibition of signal transducer and activator of transcription 3 (STAT3), a downstream effector of both hormones (39, 40). Resistance to these signals of adiposity negative feedback progressively increase the defended body weight, consistent with the difficulty observed in maintaining weight loss achieved through diet and exercise (9, 41, 42). Reversal of the HI induced obese phenotype has been shown from anti-inflammatory drug treatment (24, 43) and also dietary fatty acid manipulation (44). Deletion of the MyD88 gene that couples PRR binding to nuclear transcription factor kappa-β (Nf-κβ) activation, a potent inflammatory signaling mechanism, protected mice against leptin resistance and DIO (45).
Given this remarkable advancement of knowledge in the central control of energy balance it is particularly important to avoid errors of interpretation through poorly understood experimental design. For example, some studies have shown evidence of a blunting of anorectic hypothalamic signaling from free SFAs, in particular palmitate, but these involve intracerebral cannulation and pharmacological administration of fatty acids that may not be physiologically equivalent to dietary administration (24, 45). It is very common in manuscripts for the fatty acid profile of the diet to not be reported, but simple mathematical mistakes in reporting and mistakes of misrepresentation are also common.
Critical Evaluation of HFD Literature
One example of reporting error, Kim et al. showed through genetic deletion that Tlr-4 was necessary for the development of comorbidities to DIO such as insulin resistance (38) in response to HFD feeding. The DIO is described as being induced by a “high saturated fat diet” described in the supplementary methods sections as having “60% saturated fat” content. The diet used in their experiments, Research Diets’ D12492, obtains 60% of its kcals as fat from lard and soybean oil; the percentages of the fat portion are listed as 32% SFA, 35.9% MUFA and 32% PUFA (46). This means that at most the kcals from SFA is 20%, the same as the PUFA content and less than that from MUFA, and that at the currently reported levels the kcal contribution from LA alone is nearly as much as (91% of) that from all saturated species combined. Milanski et al. fed a 35% (by weight) fat diet to Wistar rats to induce hypothalamic inflammation, making a point to report a 5% saturated fat (also by weight) content (24). The title, “Saturated Fatty Acids Produce an Inflammatory Response Predominantly through the Activation of TLR4 Signaling in Hypothalamus: Implications for the Pathogenesis of Obesity,” would lead one to believe that the saturated fat content of the diet would be more than 14% of the fat portion, leaving 86% for unsaturated species. Why the resulting weight gain and metabolic disturbances are attributed to the saturated and not the unsaturated fats is not explained.
In another study Davis et al. fed two high fat diets, a “high fat control (HFC)” using soybean oil and a “high fat palmitate (HFP)” using lard and purified palmitate and looked at how deletion of the Tlr-4 gene in mice affected obesity (25). It can be assumed that the lard used is higher than reported in PUFA and lower than reported in MUFA and SFA, but by using some purified palmitate and comparing it to pure soybean oil it is almost certain their HFP diet contains more palmitate and less PUFA than their HFC. The real problem in this paper is the focus on the obesity in the WT animals from the HFP over the HFC, which was, respectively as reported, less. Tlr-4 deletion greatly reduced body weight and metabolic syndrome markers in mice fed HFP and did not in mice fed HFC, consistent with their title asserting this deletion selectively protects against obesity from SFA. They did not seem to find it interesting that the soybean oil diet made the WT mice heavier than the lard with palmitate in the first place. Another group came to a similar conclusion about insulin resistance caused by infusion of lard or soybean oil into blood. After knocking out Tlr-4, insulin resistance from lard oil (fatty acid profile not provided) was abrogated but not that from soybean oil (26). This paper discusses an inconsistency with another study showing that Tlr-4 knockout did have an insulin sensitizing effect on animals infused with an oil product similar to soybean oil, but fails to comment on the fact that the soybean oil does cause insulin resistance, perhaps by another mechanism.
Akoum et al. in a similar study report their HFD fatty acid profile, using D12492 before 2011 and therefore misreporting linoleic, palmitic and oleic acids (47). Interestingly, even at reported levels, the HFD supposedly high in SFA contains more PUFA and LA than the plant oil diet, which makes up for the lower SFA and PUFA with MUFA. The discussion, not surprisingly, mentions only the SFA and MUFA as experimental dietary variables.
There seems to be a prescientific conclusion that SFA causes obesity and the metabolic syndrome and that PUFA or MUFA does not, leading investigators to use poorly controlled experimental diets and even ignore and/or misreport their own findings while discussing their results. These studies and others like them are cited and used as evidence, seemingly from their titles alone, that saturated and not unsaturated fats cause obesity and insulin resistance.
A small but growing number of DIO studies are designing and reporting their HFDs in a more careful manner, in contrast to those previously mentioned. The results obtained of obesity from different fatty acids are also largely in contrast to conventional wisdom and previous findings. In 2012 Alvheim et al. showed an 8% LA HFD to be more obesogenic than a 1% LA HFD of identical total fat content. The 7% difference in LA was made up for by SFA from coconut oil (48). Another study primarily aimed at cardiac outcomes and using HFD fed mice, high SFA vs. high n-6 PUFA, showed that mice fed the high PUFA diet both gained more weight and had greater feed efficiency than mice fed the high SFA diet (49). As long ago as 1993 it was shown that when using tallow, a more reliably saturated animal fat than lard, male Wistar rats gained less fat than on isocaloric/isolipidic diets using olive oil; the highest gaining group were rats fed safflower oil (50). These studies and their findings, although using more clear and accurate methods for their HFDs, have not made significant traction in the obesity research world. There appears to be a pre-scientific conclusion on the obesity effects of the different fatty acids that may be related to misconceptions about the epidemiology of dietary change in the 20th century.
What We Are Modeling: Dietary Fat Intake and the “Obesity Epidemic”
Dietary fat intake in the USA may have increased during the 20th century. Although it is the general consensus that it has, there are some estimates showing a mild decrease, at least from the middle to the end of the century (51). If it has increased, the increase is not equal or even close to the degree widely used in DIO models where control vs. experimental diets derive ~10% and 40-60% kcals from fat, respectively. This possible increase has also been accompanied by a decreased intake of saturated fat (52). The reduction of saturated fat consumption has in part been offset calorically by an increase of ~800% in caloric contribution from LA alone. Some of the largest changes in caloric contribution to the American diet during the 20th century came from soybean oil (+123,810%, ~0.006 to ~8% of kcals), poultry (+426%) and non-soybean plant oils (+250%) with beef, lamb and dairy all seeing double-digit percentage decreases (53). These changes greatly surpass, on a percentage basis, the increase in sucrose intake (~60%) over this same time period. In addition to this change in fatty acid source, altered animal husbandry practices have resulted in greater tissue representation of n-6 PUFA in beef cattle, especially when viewed as a ratio to n-3 PUFA (18). The increase is greater in pigs due to their ungulate physiology. Thus, a large percentage increase in LA, moderate decrease in SFA and stable or slight increase in total fat intake are the changes in 20th century dietary fat based on food availability and intake questionnaire data (52–54). These changes occurred concomitant with the so-called obesity epidemic.
Estradiol and Energy Balance
Obesity does indeed present different patterns of fat distribution. These patterns differ in their association with risk factors for disease (7, 43, 55) as well as prevalence between sexes (56, 57). Hormones important to energy balance such as leptin are known to differ in level and activity by sex both in adolescence (58) and adulthood (59) and both obesity (60) and its health damaging comorbidities (61) appear to be sexually dimorphic. The sex steroid 17β-estradiol (E2) is important in many aspects of energy balance and is shown to be protective against insulin resistance (62). These effects of E2 occur through peripheral actions and central mechanisms. E2, specifically through actions in the brain, is known to suppress feeding and fat accumulation and augment energy expenditure and activity. The key brain regions mediating E2’s effects on energy homeostasis are the hypothalamus and the hindbrain (63–66). A decrease in circulating estrogens due to ovariectomy or menopause is associated with positive weight gain in both rodent models and humans, which is attenuated by E2 replacement (67). What is not well known, although poorly controlled experiments have attempted the question (47), is how the different fatty acids affect DIO in the context of E2 signaling.
Maternal Over Nutrition and Offspring Obesity
The current idea of intergenerational links between nutrition and health began with the Barker “fetal/infant origins of disease” hypothesis. Stated simply: a poor nutritional environment during pregnancy, lactation and early infancy predisposes those offspring, whose nutritional environment is now more abundant, to chronic diseases later in life such as heart disease, metabolic derangements such as Type II Diabetes Mellitus, and obesity (68). Attempts at modeling this obesity predisposition through maternal under nutrition have had mixed results. They do, however, suggest an altered development of the hypothalamic appetite regulating system attributed to a reduced postnatal leptin surge and “catch-up” weight gain (69–71). These findings, although interesting, are largely inapplicable to the Western obesity epidemic characterized by nutritional abundance as opposed to food restriction. This over nutritional maternal environment is most often studied in rodent DIO models. Some studies show, similar to maternal under nutrition, lower birth weight in treated offspring than control offspring followed by a catch-up weight gain and eventual obesity and insulin resistance (72, 73). Higher birth and adult weights in offspring of DIO dams compared to control offspring has also been shown (74). Although there is much research on maternal HFD influencing offspring obesity and hypothalamic gene expression, few studies have looked specifically at different types of HFD’s or for inflammatory genes in the hypothalamus. There is even some evidence of a link between LA in maternal diet and E2 signaling and metabolism in offspring (75).
Figure 1: Over 12 weeks body weight gain in wild type male C57 mice fed either a low-fat control (Research Diets D12450B) or high-fat (Research Diets D12451) diet is different. High-fat diet fed males are substantially heavier than low-fat controls.
Figure 2: (A) Initial body weights were similar in all groups. (B) At weeks 5 and 10, respectively, percent weight gain in the 22.5% and 15% groups exceeded the 1% group. (C) Cumulative weight gain at 12 weeks was higher in 15% and 22.5% than 1%; the 1%, 15% and 22.5% groups exceeded controls at 3, 2 and 1 weeks, respectively. (D) All experimental groups had lower lean and higher fat mass as a percent of body weight than controls but were similar to each other; 22.5% had slightly higher lean mass than 15%. Statistical Analysis: (A, C & D) Data were analyzed by a one-way ANOVA with Bonferroni’s multiple comparisons test. (B) Data were analyzed by a repeated measures, two-way ANOVA with Bonferroni’s multiple comparisons test. * or a=p<0.05; ****=p
It is well established that high-fat diet feeding induces greater weight gain in C57 mice than low-fat diet feeding; this is shown in Figure 1. In an attempt to determine the most obesogenic fatty acids, weight gain was compared over the same time frame in the same strain of mice on the 3 experimental diets, 1%, 15% and 22.5% LA. It is clear in Figure 2 that all 3 HFDs result in greater weight gain than control, but that the higher LA diets do cause higher weight gain than the 1%. Percent body composition, however, is still similar between all HFD groups. Energy intake (Figure 3) may partially explain the increased weight gain, with the 22.5% group having slightly higher intake than the other 2 HFDs, but does not explain why the 15% group gained more weight than the 1%. Even with the small differences in food intake this was not a big enough factor to make the calculated feeding efficiency different between any HFD groups. Looking at other aspects of metabolic syndrome, such as glucose metabolism, there were many similarities between HFD groups and one interesting difference. Fasting BG was the same, as well as glucose disposal after intraperitoneal injected glucose bolus. The only difference in glucose metabolism was that the 22.5% group, alone, had reduced blood glucose suppression compared to all other groups. Strikingly, the 1% and 15% LA HFD groups had a similar response to exogenous glucose as the low-fat controls. Perhaps fasting glucose and GTT did not appear different because the 22.5% groups were pushing their endogenous insulin production harder. We also measured blood triglycerides at sacrifice, which were higher in the lower LA diets. This is consistent with the known property of saturated fat to increase circulating lipoproteins, a significant compartment for triglycerides. Similarly to most of the glucose metabolism assays, oxygen consumption and carbon dioxide production as measured by indirect calorimetry was similar between the HFD groups. The only observed in a 24-hour metabolic chamber was that the 22.5% group had less nighttime activity than the other groups. This may be related to their higher weight, either causative of or caused by since at this time they had reached their higher body weights.
Figure 3: Control diet contained 3.85 kcals/g and all 3 experimental HFDs contained 4.7 kcals/g. Weekly food weighing yielded food intake per cage (3 mice/cage) in grams, which was multiplied by the kcals/g for that diet to produce weekly energy intake in kcals. Feeding efficiency is calculated by dividing the gram change in body weights per cage by the weekly intake in kcals per cage.
(A & B) Weekly energy intake was elevated in the 22.5% group compared to control. Intake between the other groups was not different. (C) Feeding efficiency was the same between the 3 HFD groups but was lower in the control group. Statistical Analysis: (A) Data were analyzed by a repeated measures, two-way ANOVA with Bonferroni’s multiple comparisons test. (B & C) Data were analyzed by a one-way ANOVA with Bonferroni’s multiple comparisons test. *=p<0.05; ****=p<0.0001
Figure 4: (A) Fasting (5 hr) glucose was elevated in all HFD groups but not different between them. (B) Serum triglycerides were higher in the 1% group compared to control, 15% and 22.5%, consistent with saturated fat elevating blood lipoproteins. Statistical analysis: Data were analyzed by a one-way ANOVA with Bonferroni’s multiple comparisons test. Comparisons are between experimental groups and control except for capped lines that represent comparisons between 2 experimental groups. *=p<0.05; **=p<0.01; ***=p<0.001; ****=p<0.0001.
Figure 5: Glucose tolerance did not differ between experimental HFD groups from a glucose challenge (A & B), but glucose levels were not as sensitive to an insulin challenge (C & D) in the 22.5% group compared to all other treatments. Statistical Analysis: (A & C) Data were analyzed by a repeated measures, two-way ANOVA with Bonferroni’s multiple comparisons test. a=p<0.05; b=p<0.01; c=p<0.001; d=p
Figure 6: (A) Daytime and nighttime patterns of O2 volume was similar in all groups except for 1% V.O2, which was lower than the controls. (B & C) CO2 production was suppressed in experimental groups indicating a change in energy substrate utilization. (D) Nighttime activity was reduced in the 22.5% group. Statistical Analysis: Data were analyzed by a two-way ANOVA with Bonferroni’s multiple comparisons test. a=p<0.05; b=p<0.01; c=p<0.001; d=p
Gene expression (Figures 7 & 8) in the Arcuate nucleus of the hypothalamus had many differences between the groups, many of which were unexpected. Tlr4 expression was reduced as LA content became higher, which may be down regulation in response to inflammation. This is supported by a similar expression in the 1% group as in the control group. The inflammatory cytokine Tnf-α, however, showed opposite results with highest expression in the control and 1% groups and lowest in the 22.5%. The down stream signaler of inflammatory receptor activity Traf6 showed reduced expression in the higher LA groups as well, but the related gene Irak was not different between any groups. The down stream signaler of insulin and leptin activation, Stat3, was upregulated in the higher LA groups compared to control and 1%, which may suggest insulin and/or leptin resistance. Socs3, an antagonist of Stat3 signaling, was not statistically different between groups.
Figure 7: Inflammatory genes: receptors and signaling genes. Diets higher in LA suppressed ARC Tlr4 (A), TNF-alpha (B) and Traf6 (C) but not Irak1 (D) mRNA. (E) Stat3 expression was increased in the highest LA group. (F) Socs3 was not different between groups. Statistical analysis: Data were analyzed by a one-way ANOVA with Bonferroni’s multiple comparisons test. Capped lines indicate comparisons between 2 groups. a=p<0.05; b=p<0.01
The hypothalamic neuropeptides that influence feeding and metabolic rate had unexpected expression results as well. The anorexigenic Pomc and Cart, usually expressed in unison, were higher (Pomc) in the 22.5% group but lower (Cart) in that same group. This seems to be a mixed message for energy balance. The orexigenic neuropeptides Npy and Agrp, however, had a more unified expression pattern in the 22.5% group; both were increased compared to the 1% and 15% groups. The control group, interestingly, behaved at times like the 1% and 15% groups (Pomc & Agrp) and other times like the 22.5% group (Cart and Npy).
Figure 8: Arcuate anorexigenic/orexigenic genes: Although the 1% and 15% groups were not different, the 22.5% group had increased Pomc (A) expression compared to control. (B). Cart expression was similar between 22.5% and control but elevated in 1% and 15%. (C) Npy expression was suppressed in 1% & 15% group but not in the 22.5% compared to controls. (D) AgRP expression was elevated in the 22.5% group compared to all other diets. Statistical analysis: Data were analyzed by a one-way ANOVA with Bonferroni’s multiple comparisons test. Capped lines indicate comparisons between 2 groups. a=p<0.05; b=p<0.01; c=p<0.001; d=p<0.0001.
AIM 1: Determine relative efficacy of high SFA versus high n-6 PUFA fatty acid profile diets in producing hypothalamic inflammation and obesity.
Rationale: High-fat diets with fatty acid compositions ranging from high SFA and low LA to low SFA and high LA were designed to address confusion in the scientific literature as to what fatty acids make a greater contribution to obesity. This question of fatty acid contribution to weight gain will be joined to the proposed positive energy balance mechanism of hypothalamic inflammation by pairing weight, food intake and body composition measurements of animals on these diets with molecular measurements of inflammatory mRNA and protein in the hypothalami of the same.
Animals/Diets: Forty-eight WT male C57BL6/J mice were divided into 4 (n=12) treatment groups: 1) control diet (Research Diets D12450B), 2) 1% LA HFD, 3) 15% LA HFD and 4) 22.5% LA HFD. All HFDs are based on the same AIN76 guidelines as the control diet and are identical to Research Diets’ 45% kcals from fat HFD D12451 except in fatty acid composition. The percentages of kcals from LA were achieved by using primarily hydrogenated coconut oil for fat in the first HFD and adding sufficient soybean oil to attain 1% of total kcals from LA, and then ratcheting down coconut oil while increasing soybean oil to reach the 15 and 22.5% levels in the other two HFDs. Dietary n-3 PUFA content was kept constant with added menhaden oil. After weaning, at approximately 8 weeks of age, mice were put into groups and given their experimental diets. Mice were housed in groups of 3 with ad libitum access to water and their respective diets in a climate-controlled facility with a 12/12-hour light/dark cycle.
Anatomical/Physiological Measurements: Throughout the 8 week study weekly body weight and food intake data was taken. At the end of 8 weeks body composition was determined by MRI. Metabolic rate was estimated by 24 hours of indirect calorimetry (with a 24 hour acclimation) in a metabolic chamber (Oxymax). Glucose tolerance was estimated by a glucose tolerance test (GTT) involving intraperitoneal (IP) injection of glucose in 0.9% saline in H2O (dose) followed by blood glucose readings at t=0, 15, 30, 60, 90, 120 and 180 minutes. GTTs were preceded by a 15-hour fast. Insulin tolerance test (ITT) was estimated similarly, with a 5 hour fast followed by IP injection of insulin in 0.9% saline in H2O (dose) and the same time points up to 120 minutes.
Sacrifice/Tissues Collected: At least 3 days of rest were allowed between the mice being isolated in a metabolic chamber, the GTT and ITTs and sacrifice. Mice were rendered insensible by a 0.05cc IP injection of ketamine and killed by decapitation. Blood was collected, preserved with AEBSF, centrifuged for plasma and stored in -80°C. Central (basal hypothalamus and pituitary) and peripheral (liver, soleus muscle, gonadal white adipose) tissues as well as feces were taken and either snap frozen in liquid nitrogen and stored in -80°C (for protein analysis) or fixed in RNAlater and stored in -80°C (for RNA analysis). Hypothalamic tissues for RNA analysis were microdissected for individual nuclei.
Protein and Gene Expression Assays: Plasma samples were used to determine circulating ghrelin, leptin, insulin, tumor necrosis factor alpha (TNF-α), interleukin 6 (IL-6) and monocyte chemoattractant protein 1 (MCP-1) using a Millipore Magnetic Bead immunoassay (kit number) read on a Luminex plate reader. Protein will be extracted from snap frozen samples and RNA extracted from RNAlater fixed samples for analysis of inflammatory and metabolic proteins/genes in tissues. Hypothalamic regulatory neuropeptides proopiomelanocortin (POMC), neuropeptide Y (NPY), cocaine and amphetamine regulated transcript (CART) and Agouti-related protein (AGRP), leptin receptor, insulin receptor, TNF-α, IL-6 and other genes/proteins will be measured in the hypothalamus and in individual nuclei. Inflammatory genes/proteins will also be analyzed in peripheral tissues. Microbiome will be determined from fecal samples.
Expected Results: We expect all of the HFD groups to have higher body weights and fat mass than the control group, but for increasing LA content in the diet to cause increased body weight and fat mass amongst the HFD groups. We also expect for the higher LA diet animals to be less glucose tolerant and more insulin resistant than the lower LA diet and control animals as measured by GTT and ITT. We expect circulating and tissue inflammatory and metabolic hormones (leptin and insulin) to be higher in the HFD groups with higher LA content. This would presumably lead to lower expression of hypothalamic leptin and insulin receptors in those groups and suppressed anorectic neuropeptide expression (POMC and CART) and higher orexigenic neuropeptide expression (NPY and AgRP). In short, the expected phenotype of the HFD mice is that the greater LA content of their diet, the farther from the control phenotype they will be in anatomical, physiological and molecular assay. There are multiple mechanistic reasons why PUFA in general and LA and the other n-6 PUFA in specific could contribute to obesity inducing inflammation. LA is the substrate for arachidonic acid (AA) through desaturation and elongation, which itself, through cyclooxygenases and peroxidases, is the substrate for the inflammatory eicosanoids (76, 77). These include the series 1 and 2 prostaglandins as well as some of the thromboxanes and leukotrienes (76, 77). The inflammatory effects of these molecules are well known (78). In addition to forming these inflammatory signaling molecules the n-6 PUFA also produce the pro-feeding behavior endocannabinoids. LA increases both anandamide and 2-arachidonoylglycerol through its increased total abundance and the increased ratio to n-3 PUFA in the diet (48). A more general mechanism of inflammatory induction from PUFA is due to their ability to non-enzymatically oxidize into free radicals and hydroperoxides at physiological temperature (79). Reactive oxidative species are known to cause inflammation in the brain (30).
AIM 2: Determine the effects of E2 on the obesogenesis of high SFA or high n-6 PUFA diet.
Rationale: Expanding on the first aim’s question, we now look to add the concept of sexual dimorphism in DIO to the previous concepts. Female animals with sufficient endogenous E2, or supplementary exogenous E2, have been shown to be partially or fully protected from DIO as compared to male animals or E2 insufficient female animals (80).
Animals/Diets: Ninety-six female ovariectomized WT C57BL6/J mice will be divided into 8 treatment groups (n=12). The treatment diets will be the same as in aim 1; 2 groups will be on each diet, one receiving daily E2 treatment by body weight and the other receiving only vehicle treatment. The vehicle will be coconut oil mixed with peanut butter for palatability. Mice will be housed in pairs with ad libitum access to food and water in a climate-controlled facility with a 12/12-hour light/dark cycle. Anatomical/physiological, sacrifice and molecular assays will be the same as in aim 1.
Expected Results: We expect, as in aim 1, for the HFD groups to have higher bodyweights, fat masses and molecular markers of inflammation than the control diet. We also expect as the percentage of kcals from LA is increased that obesity and inflammation will also increase. We expect oil treated mice to be more obese than their E2 treated counterparts, and will look for a synergistic effect between E2 and LA.
AIM 3: Determine the metabolic effects on offspring of dams fed high SFA or high n-6 PUFA diet
Rationale: The central hypothesis of this proposal, that n-6 PUFA is more obesogenic than SFA and works through inducing hypothalamic inflammation, will be applied to the existing field of transgenerational energy balance and metabolic studies. A developing neonate is particularly vulnerable to a centrally targeted nutritional insult. The question being asked is if the effect of fatty acid profile for a short period very early in life can be resolved in offspring being maintained as adults on an identical, non-obesogenic diet.
Animals/Diets: Six female WT C57BL6/J mice will be divided into 3 treatment groups, control, 1% LA and 22.5% LA HFD. At 10 weeks of age females will be singly housed, put on their respective diets and weighed weekly. At the fifth week a WT male will be introduced for breeding. After signs of pregnancy the male will be removed until after weaning. Dams will be bred twice. Offspring mice will be weighed at postnatal day (PND) 2, 14 and 21 at which point they will be removed from their mother and split up into two cages by sex. After weaning, offspring will be fed control diet (10% kcal fat, D12450B). Offspring weight and food intake by cage will be taken weekly for 25 weeks, at which point physiological assays followed by sacrifice/tissue collection and molecular assays will be performed as in aims 1 and 2. The dam breeding scheme will be repeated with another 6 females with an identical treatment. The second set of offspring will be fed after weaning a high fat diet (45% kCal fat, D12451). All assays will remain the same.
Expected Results: We expect the offspring of the dams fed the experimental diets to follow a similar trend as the adult experiments in aims 1 and 2. That is to say that we expect both of the HFD groups to produce heavier offspring with greater inflammatory markers but for the high LA diet to do so more than the low LA diet. We expect this effect of fatty acid type in maternal diet to be more pronounced when the offspring are fed an HFD than a low fat control diet.
Potential Pitfalls and Alternatives
As shown in the preliminary results section the weight gain differences between the 3 HFD groups less than was expected. Unexpected results were also seen in weight gain in the female experiment. One potential way to address this is to analyze the final weights at day of sacrifice and see if there is more of a difference between groups. These weights have the problem of not having associated food intake data from the date of the end of the weight study until sacrifice. Another alternative analysis being looked into is comparing free water as measured in MRI between groups as evidence of inflammation or edema in lean tissue. As molecular data is collected primers for different mRNA products will be considered as results suggest looking further into molecular pathways. Protein assays will also be considered in this additive process. Both the male and female (Aims 1 and 2) experiments have a conceptual flaw in that the phenomenon of hypothalamic inflammation causing weight gain is being measured after the weight gain has taken place. One way to address this is to run another cohort of mice on the experimental diets and collect hypothalamic tissues from them before they gain weight, or gain a substantial amount of weight, versus a control group. This would translate into a shorter study, most likely no more than 2-4 weeks as compared to the 12 weeks for the full male body weight study.
1. K.-S. Lee, Perspect. Biol. Med. 50, 585–602 (2007).
2. B. Caballero, Epidemiol. Rev. 29, 1–5 (2007).
3. E. A. Finkelstein et al., Am. J. Prev. Med. 42, 563–70 (2012).
4. K. M. Flegal, M. D. Carroll, C. L. Ogden, L. R. Curtin, JAMA 303, 235–241 (2013).
5. C. L. Ogden, M. D. Carroll, B. K. Kit, K. M. Flegal, JAMA 311, 806–14 (2014).
6. B. K. Kit, C. L. Ogden, K. M. Flegal, Epidemiology of Obesity W. Ahrens, I. Pigeot, Eds. (Springer New York, New York, NY, 2014.
7. J.-P. Després, I. Lemieux, Nature 444, 881–7 (2006).
8. G. Whitlock et al., Lancet 373, 1083–96 (2009).
9. D. M. Garner, S. C. Wooley, Clin. Psychol. Rev. 11, 729–780 (1991).
10. A. L. Mark, Clin. Exp. Pharmacol. Physiol. 33, 857–862 (2006).
11. E. L. Melanson et al., Med. Sci. Sports Exerc. 45, 1600–1609 (2013).
12. Y. Zhang et al., Positional cloning of the mouse obese gene and its human homologue.Nature 372, 425–32 (1994).
13. P. Wang, E. Mariman, J. Renes, J. Keijer, The secretory function of adipocytes in the physiology of white adipose tissue.J. Cell. Physiol. 216, 3–13 (2008).
14. J. a Martinez, Proc. Nutr. Soc. 59, 337–45 (2000).
15. M. Tschöp, M. L. Heiman, J. A. Barth, Exp. Clin. Endocrinol. diabetes 109, 307–319 (2001).
16. R. Buettner et al., J. Mol. Endocrinol. 36, 485–501 (2006).
17. R. Buettner, J. Schölmerich, L. C. Bollheimer, Obesity (Silver Spring). 15, 798–808 (2007).
18. C. A. Daley, A. Abbott, P. S. Doyle, G. A. Nader, S. Larson, Nutr. J. 9 (2010).
19. J. Ruiz et al., Meat Sci. 49, 155–63 (1998).
20. pork, fresh, separable fat, cooked.
21. M. A. Pellizzon, Lard fatty acid profile update – Research Diets, Inc (2012).
22. I. Research Diets, Typical Composition of Lard Used by Research Diets , Inc . Research Diets , Inc . (2012).
23. C. M. Mayer, D. D. Belsham, Endocrinology 151, 576–85 (2010).
24. M. Milanski et al., 29, 359–370 (2009).
25. J. E. Davis, N. K. Gabler, J. Walker-Daniels, M. E. Spurlock, Obesity (Silver Spring). 16, 1248–55 (2008).
26. W. L. Holland et al., J. Clin. Invest. 121, 1858–1870 (2011).
27. A. P. Arruda et al., Endocrinology 152, 1314–26 (2011).
28. J. K. Elmquist, C. F. Elias, C. B. Saper, Neuron 22, 221–232 (1999).
29. L. A. Velloso, 2025, 151–158 (2009).
30. D. Cai, T. Liu, Ann. N. Y. Acad. Sci. 1243, E1–39 (2011).
31. D. Cai, T. Liu, 4, 98–115.
32. A. Sahu, Endocrinology 145, 2613–2620 (2004).
33. S. H. Bates et al., Nature 421, 856–859 (2003).
34. C. F. Elias et al., Neuron 23, 775–786 (1999).
35. A. Abizaid, Q. Gao, T. L. Horvath, Neuron 51, 691–702 (2006).
36. A. P. Coll, I. S. Farooqi, S. O. Rahilly, , 251–262 (2007).
37. A. Manuscript, E. Imbalance, Cell 135, 61–73 (2008).
38. F. Kim et al., , 1589–1596 (2007).
39. J. P. Thaler, M. W. Schwartz, Endocrinology 151, 4109–15 (2010).
40. A. L. Rodrigues et al., J. Nutr. Biochem. 22, 109–17 (2011).
41. R. R. Wing, S. Phelan, Am. J. Clin. Nutr. 82, 222S–225S (2005).
42. N. a King et al., Obesity (Silver Spring). 15, 1373–83 (2007).
43. M. Boaz, L. Lisy, G. Zandman-Goddard, J. Wainstein, Diabet. Med. 26, 708–13 (2009).
44. R. Pauli et al., 7 (2012), doi:10.1371/journal.pone.0030571.
45. A. Kleinridders et al., Cell Metab. 10, 249–259 (2009).
46. W. Lagakos, Typical Fatty Acid Composition of Fats used by Research Diets , Inc . (2013).
47. S. El Akoum, V. Lamontagne, I. Cloutier, J. Tanguay, Diabetol. Metab. Syndr. 3, 34 (2011).
48. A. R. Alvheim et al., Obesity 20, 1984–1994 (2012).
49. Y. S. Diniz et al., Nutrition 20, 230–234 (2004).
50. D. A. Pan, L. H. Storlien, J. Nutr. , 512–519 (1993).
51. L. Madsen, R. Koefoed, K. Kristiansen, 1740, 266–286 (2005).
52. A. M. Stephen, N. J. Wald, Am. J. Clin. Nutr. 52, 457–469 (1990).
53. T. L. Blasbalg, J. R. Hibbeln, C. E. Ramsden, S. F. Majchrzak, R. R. Rawlings, Am. J. Clin. Nutr. 93, 950–962 (2011).
54. G. Taubes, Science (80-. ). 291, 1–16 (2001).
55. S. Gesta, Y.-H. Tseng, C. R. Kahn, Cell 131, 242–256 (2007).
56. S. Lemieux et al., Diabetologia 37, 757–764 (1994).
57. O. T. Ayonrinde et al., Hepatology 53, 800–809 (2011).
58. J. N. Roemmich et al., Am. J. Physiol. 275, E543–E551 (1998).
59. A. Kennedy et al., 82 (1997).
60. L. M. Brown, L. Gent, K. Davis, D. J. Clegg, Brain Res. 1350, 77–85 (2010).
61. J. A. P. Da Silva, S. H. Peers, M. Perretti, D. A. Willoughby, J. Endocrinol. 136, 389–397 (1993).
62. E. Riant et al., Endocrinology 150, 2109–2117 (2009).
63. J. Santollo, A.-M. Torregrossa, L. A. Eckel, Horm. Behav. 60, 86–93 (2011).
64. L. Asarian, N. Geary, Endocrinology 148, 5656–5666 (2007).
65. S. Thammacharoen, T. A. Lutz, N. Geary, L. Asarian, Endocrinology 149, 1609–1617 (2008).
66. S. Musatov et al., Proc. Natl. Acad. Sci. U. S. A. 104, 2501–2506 (2007).
67. T. W. Abel, N. E. Rance, Mol. Brain Res. 69, 202–208 (1999).
68. D. J. Barker, BMJ Br. Med. J. 301, 1111 (1990).
69. F. Delahaye et al., Endocrinology 149, 470–475 (2008).
70. S. P. Sébert et al., Endocrinology 150, 634–641 (2009).
71. R. L. Cripps et al., Clin. Sci. 117, 85–93 (2009).
72. G. J. Howie, D. M. Sloboda, T. Kamal, M. H. Vickers, J. Physiol. 587, 905–915 (2009).
73. G. J. Howie, D. M. Sloboda, C. M. Reynolds, M. H. Vickers, J. Nutr. Metab. (2013), doi:10.1155/2013/517384.
74. A.-M. Samuelsson et al., Hypertension 51, 383–392 (2008).
75. A. Cabanes, S. de Assis, J. Gustafsson, L. Hilakivi-Clarke, Dev. Neurosci. 22, 488–493 (2000).
76. L. M. Dunbar, J. M. Bailey, J. Biol. Chem. 250, 1152–1153 (1975).
77. L. S. Wolfe, J. Neurochem. 38, 1–14 (1982).
78. S. L. Tilley, T. M. Coffman, B. H. Koller, J. Clin. Invest. 108, 15–23 (2001).
79. I. S. Young, J. Mceneny, Biochem. Soc. Trans. 29, 358–362 (2001).
80. K. J. Mamounis, J. A. Yang, A. Yasrebi, T. A. Roepke, Steroids 81, 88–98 (2014).