10 GENERAL DISCUSSION
10.2 METHODOLOGICAL CONSIDERATIONS
This is particularlyvalid in the developed countries, where most dyslipidemias are hyperlipidemias and the causes are often related to the diet and lifestyle habits.
The potential effect of Bacteroides as an emerging probiotic to affect satiety and other metabolic parameters was also studied in vivo. However, this approach was not fully supported in this study. The observed increased fecal heat value of Bacteroides-fed animals indicated that more energy was passing through the body without being utilized. This could support the initial hypothesis regarding the linkage between the high Bacteroidetes abundance and low body weight. This effect may be mediated through changes in the composition of the gut microbiota or through microbial metabolites. Unfortunately, a full gut microbiota analysis was not conducted as part of this study.
The observed lower postprandial plasma lactate concentration after the use of polydextrose could also be associated with a lower post-meal inflammatory status. In obese, low-grade inflammation is one of the main attributes to MetS, and therefore, new means to help in the attenuation of the inflammatory status are highly needed. Given the potential satiety-enhancing benefit that followed the consumption of polydextrose, its use could benefit the overweight and obese individuals. Further clinical studies with suitable population should be warranted to study this approach.
10.2 METHODOLOGICAL CONSIDERATIONS
The metabolic impact of different bioactive food ingredients was assessed in this study. Methodologically, the study designs allowed for the elucidation of mechanisms of individual supplements, and in addition, their combined effects were explored. The impact of supplemental betaine, alone or together with dietary polydextrose, was evaluated in adipose tissue in Studies I and II, respectively. The rationale to combine betaine and polydextrose into same study design (Study II) was to see if these ingredients could complement each other in ameliorating the metabolic status of experimental animals undergoing a high-fat stress. Polydextrose was known for its potential to ameliorate the postprandial metabolic state, namely weight gain and triglyceride response, and previous studies with betaine had shown promising effects on the inflammatory responses. However, as a result, polydextrose had only minor effects on adipose tissue metabolism, and no clear synergy between the treatments was seen in Study II. Furthermore, the combined effects of polydextrose or lactitol and administered Bacteroides species were looked at in Study III. This study hypothesized that B. thetaiotaomicron might function as an emerging probiotic affecting the weight gain in high-fat-diet-fed rats and that there could be potential synbiotic effects with lactitol or polydextrose. Eventually, the separate administration of B. thetaiotaomicron had only a minor or no effect on the metabolism of the rats. This led to believe that the metabolic changes detected by the combination treatments (PDX or lactitol in combination with B. thetaiotaomicron) were in fact mediated by the two carbohydrates.
The treatment dosages used in separate models were largely based on the available literature and the previous knowledge from working with the ingredients in question. The betaine concentrations used in Study I (50, 250 and 500 μmol/l) corresponded to the average concentration of betaine in human plasma and the approximate peak concentrations measured after single doses of 1 and 3 g of betaine, respectively (Schwab et al., 2006). The betaine dose and route used in Study II, 1% (w/v) given ad libitum in drinking water, were in line with the previous literature (Song et al., 2007, Kwon et al., 2009, Wang et al., 2010). This dose corresponded approximately to 30 mg of betaine per animal per day based on the average amount of water consumed ad libitum per animal. The initial polydextrose dose (6.66% w/v) corresponded to 0.2 g per animal per day, but it needed to be adjusted to 3.33 % (w/v), as the mice did not tolerate well the high amount of polydextrose. The Study II was part of an animal trial reported by Pekkinen and co-workers (Pekkinen et al., 2013). The polydextrose (2 g/animal/day) and lactitol (1.6–1.8 g/animal/day) doses selected for Study III were based on previous knowledge and studies with similar supplementations (Peuranen et al., 2004, Gee & Johnson, 2005, Tiihonen et al., 2008). The amount of B.
thetaiotaomicron (1010 bacteria/animal/day) was somewhat higher than what has typically been used in preclinical rat studies investigating the effects of probiotics (Tiihonen et al., 2008). The polydextrose dose used in the clinical Study IV (15 g) was carefully selected based on previous knowledge and available literature (Hull et al., 2012, Astbury et al., 2013).
One aim of the Study I was to assess the usefulness of the hypoxia stress model of human adipocytes in studying the effects of supplemental food ingredients. The hypoxic conditions (1% O2), which were maintained in a CO2 incubator in Study I, corresponded to the partial oxygen pressure comparable to that observed in the adipose tissue of ob/ob mice (Ye et al., 2007). The effect of betaine was examined in this cell culture model. The human preadipocytes were first differentiated into mature fat cells, which were then cultivated under hypoxic conditions. Hypoxia was used to imitate the oxygen-deprived state of enlarged fat tissue in obesity. This in vitro stress model utilized a hypoxic cultivation chamber, which proved useful when modeling the effects of circulating nutrients on adipocyte function. When ordering the commercial preadipocytes for this model, the following donor information could be selected: sex, age, BMI and the exact location of the extracted fat cells. Therefore, it was possible to perform the experiments with tissues from a specific physiological area and with the same donor background. Omental preadipocytes, instead of subcutaneous or mesenteric, were chosen for this study to investigate the effects of betaine on the abdominal obesity. Omentum is the largest peritoneal fold within the abdomen and has an immunologic function. Adipokines and fatty acids from this depot have direct access to the liver through the portal circulation, which may lead to hepatic dysfunction. Omental adipocytes were considered relevant for this study based on their role in influencing the individual’s metabolic state.
Omental adipose tissue inflammation can be associated with insulin resistance in human obesity—even in subjects with similar BMI values—and an increase in the omental fat mass may contribute to the reinforcement of the inflammatory response (Hardy et al., 2011).
For the study of obesity and associated metabolic complications, several animal models have been developed. Animal models for high-fat feeding are commonly utilized to study the consequences of obesity and related manifestations, such as T2D. Especially, C57BL/6J mice react sensitively to the effects of diet on weight, and they are prone to increased body weight, fat accumulation and disruptions in glucose metabolism when fed a high-fat diet (Surwit et al., 1995). It has also been reported that fat was the primary nutritional stimulus for the development of hyperglycemia and hyperinsulinemia in C57BL/6J mice (Surwit et al., 1995). The metabolic abnormalities of C57BL/6J mice are closely related to the progression model of human obesity, and this justified their use in Study II. The body weight gain in C57BL/6J mice can be observed already after two weeks on the high-fat diet, but the weight gain becomes more obvious after four weeks (Wang & Liao, 2012). In Study II, the C57BL/6J mice were on high-fat diet for four weeks before initiating the supplementations, which lasted additional four weeks together with the continued high-fat feeding. Different animal models can be used for different research purposes, and there are strain-wise characteristics also among DIO mice. Besides C57BL/6J, there are also other mouse strains used as models for diet-induced obesity, such as AKR/J and DBA/2J mice, which are very responsive to high-fat diets, while A/J and Balb/cJ mice tend to be more resistant (Wang & Liao, 2012).
A/J and C57BL/6J mice both function as models of some components of the human metabolic syndrome;
however, important differences between strains and sexes of these mice have been reported. For example, it is more common to use male mice in DIO studies, because they are more affected by diabetes than female mice (Wang & Liao, 2012). In Study II, male C57BL/6J mice were examined. Interestingly, increased weight has been observed in A/J mice fed the high-fat diets, and it has been reported to be mainly due to adipocyte hypertrophy, while the C57BL/6J mice seem to respond to excess fat with adipocyte hyperplasia, as well (Surwit et al., 1995). If the aim is to study overt obesity and T2D, C57BL6/J mice are a suitable choice, but for the study of e.g., severely impaired glucose tolerance and insulin resistance, A/J males could be used instead (Gallou-Kabani et al., 2007).
Study III utilized a male DIO rat model (Wistar) for studying the metabolic effects of potentially beneficial food ingredients, polydextrose and lactitol, together with an emerging probiotic. However, there are also other rat models besides Wistar rats that are suitable for studying high-fat-diet-related consequences. Sprague-Dawley rats can also be used as a model for diet-induced obesity, although the metabolic effects caused by the high-fat diet seem to be more pronounced in Wistar rats. It has been demonstrated that high-fat diet increases weight gain, body fat mass and mesenteric adipocyte size and affects the adiponectin and leptin plasma levels, as well as decreases oral glucose tolerance in both Wistar and Sprague-Dawley rats (Marques et al., 2016). However, the majority of these effects has been more noticeable or detected earlier in Wistar rats, and this deteriorated metabolic state may be due to the observed differences in the gut microbial ecology between the two models (Marques et al., 2016).
104
The amount of fat in the diet naturally plays an important role in the development of obesity and related manifestations. Typically, obesity is induced in rodents with diets containing 40% to 45% of energy as fat; however, more severe metabolic changes can be induced by diets with even higher fat load, e.g., 60% energy as fat (Speakman, 2019). It has been largely debated if such very high-fat diets should even be used to model human obesity, since they contain much higher levels of dietary fat than what is commonly consumed by humans. The typical American or European diet contains approximately 36–
40% energy as fat, whereas a normal rodent diet contains about 10% energy as fat (Speakman, 2019).
Hence, the very high-fat-containing rodent diet (60% energy as fat) would provoke a greater distortion of the overall fat content compared to more moderate high-fat diets (e.g. 40–45% energy as fat) and consequently induce an even more exaggerated metabolic response (Speakman, 2019). Thus, it has been suggested that such studies using very to extremely high-fat diets are less congruent to human physiology than those studies that are using a 45% fat diet. The Studies II and III utilized the high-fat diets with 45%
and 42% of energy originating from fat, respectively. Thus, these models aimed to develop diet-induced obesity but not diabetes per se. The Study III utilized a nutritionally relevant high-fat diet model (42%
energy as fat), which allowed the evaluation of the ingredients in a set-up that is fairly similar to the obese human condition. However, it has been postulated that the type of fat would also define the functionality and the relevance of the high-fat diets to human physiology. Research has shown that high-fat diets based on animal fats and omega-6/-9 fatty acid-containing plant oils will lead to obesity and insulin resistance, whereas diets with large amounts of (marine) omega-3 fatty acids will not (Buettner et al., 2007). The main source of fat in the experimental high-fat diet (D12451 from Research Diets, Inc.) used in Study II was animal-based saturated fat (lard), with a basal quantity of soybean oil, which contains majority of linoleic acid, a polyunsaturated omega-6 fatty acid. According to literature, this is a relevant combination of fats for DIO purposes. In Study III, the experimental Western diet’s (829100, Special Diet Services) fat load consisted mainly of milk fat and additional corn oil (high in linoleic acid), while the main fat source in the control diet (801002, Special Diet Services) was soya oil. It should be kept in mind that the primary fat sources and uncontrollable differences in diet preparation might yield to different results, even though the high-fat diets would otherwise appear the same. In Study II, the effect of high-fat feeding on adipose tissue metabolism was largely looked at, and clear differences in the metabolite profile between low-fat (10% energy as fat) and high-fat (45% energy as fat) diets were detected. Another recent animal study revealed significant differences in metabolites from lung tissue between low- and high-fat diets (Showalter et al., 2018). However, a comparison to even more extreme high-fat diet (60% energy as fat) within the same study (Showalter et al., 2018) indicated that the diets with different magnitude of fat generated rather different metabolic responses.
Multi-compartmental approach is nowadays commonly used in preclinical research. This kind of complex approach can offer advantage in finding associations in multiple organs and hence may strengthen the outcomes and improve biological interpretation in intervention studies (Yde et al., 2014). Multi-compartmental approaches enable the identification of key metabolites in various tissues and thus can reveal the benefits of a dietary or supplementary intervention. Study II utilized a multicompartmental Principal Component Analysis (PCA), which is a tool to investigate structural differences and similarities in multi-block data across different tissues. In order to comprehensively characterize the whole-body metabolic profile, multiple analytical techniques and extraction solvents are typically needed (Sundekilde et al., 2020). A nontargeted metabolite profiling, as utilized in Study II, is commonly used to semi-quantitatively measure metabolic responses to dietary stimuli. Liquid chromatography (LC) coupled to mass spectrometry (LC-MS) has several advantages, such as sensitivity and good biomarker identification. The LC-MS method with tandem mass spectrometry (LC-MS/MS) based on electrospray ionization (ESI) is frequently used in determining the quantity of drugs, metabolites and biomarkers in different biological matrices (Xia & Jemal, 2009). The hydrophilic interaction liquid chromatography (HILIC) (Gama et al., 2012) can improve the sensitivity in MS due to increased ionization efficiency that results from the high organic content of the solvents used (Kivilompolo et al., 2013). Another separation technology commonly utilized in nontargeted metabolite profiling is Reverse Phase (RP) chromatography, which can cover a wider range of metabolites than HILIC but is simultaneously limited to mostly separating semi-polar compounds (Sandra & Sandra, 2013). RP prefers lipids, and HILIC prefers water-soluble metabolites. The proper identification of lipids requires the use of both ionization
modes (neg and pos), as was the case in Study II. The field of metabolomics using biochemical analytical techniques like LC-MS in human biofluids is rapidly evolving in obesity and diabetes research. A recent publication elaborated the regulation between different biological levels in systems biology by utilizing a DIO animal model to elucidate the biological response of the entire organism to identify molecular mechanisms of high-fat-diet-induced obesity (Sundekilde et al., 2020). By using metabolomics, new biomarkers can be recognized to identify the metabolic health status of subjects at risk, e.g., with high BMI. Also, metabolically healthy and unhealthy obesity, and different diabetic states, can be better characterized by using metabolomics tools in the forthcoming preclinical and clinical studies.
The method of gavaging a bacterial inoculum, as was done in Study III, has been demonstrated as an efficient way to ensure colonization (Samuel & Gordon, 2006). However, it has been shown that a conventional rat model does not necessarily predict well the pro- and prebiotic treatment responses in humans, and the responses are heavily dependent on the baseline abundance of certain gut microbes, such as lactic acid bacteria and/or bifidobacteria (Tiihonen et al., 2008). Germ-free animals inoculated with a human microbiota might be a more relevant model for studying the effects of pro- and prebiotics. In addition, a broader evaluation of the gut microbiota composition would have complemented the analyses conducted in Study III, in terms of the effects potentially mediated via the microbial activity.
The Study IV utilized a well-known, acute, high-energy-containing model (Ahotupa et al., 2010) to evaluate the postprandial effect of a fatty meal (4293 kJ, 36% energy from fat) and supplemental polydextrose (60 kJ) on satiety hormone levels and subjective feelings of appetite. The study meal contained a standard commercial hamburger (2071 kJ), french fries (1423 kJ) and a carbonated drink (799 kJ). The macronutrient composition of the meals in both groups is expressed in Table 6. The subjective feelings of appetite are commonly measured using visual analogue scales (VAS) in controlled clinical trials, as was done in Study IV. However, the methods used to analyze VAS during the satiation (pre- to post-meal) and satiety (post-meal to subsequent meal) periods have varied broadly. This has made it somewhat difficult to compare results amongst independent studies testing the same products, as seen in a recent systematic review and meta-analysis (Ibarra et al., 2016). During the analysis phase of Study IV, a new methodology was created to analyze VAS during both satiation and satiety periods. This new methodology expresses VAS results as incremental areas under the curve (iAUC) separately for satiation and satiety periods. This method allowed us also to compare VAS results from different polydextrose studies in a subsequent meta-analysis (Ibarra et al., 2016).