Effect of Probiotic Supplementation on Diet-Induced Insulin Resistance in Humans.

ABSTRACT

This current study aimed to test the hypothesis that participants undertaking 4 weeks of probiotic supplementation will have reduced high-fat-overfeeding (HFO) induced impairments in insulin sensitivity when compared with controls. Thirty-four young, healthy, male and female volunteers were assigned to either a control group (n=17) or probiotic (Lactobacillus casei Shirota [LcS]) intervention group (n=17). LcS was administered and consumed by the intervention group twice daily days 1-28 via a 65ml Yakult drink. All participants maintained their habitual diet days 1-21 and underwent HFO (65% of energy from fats and 50% kcal surplus relative to individual daily energy requirement) days 22-28. Whole body insulin sensitivity was assessed on days 1, 22 and 29 using an oral glucose tolerance test. Body mass increased by 1.72 (SD 1.33) kg in the control group and 1.47 (SD 0.79) kg in the Yakult group (P<0.05). Following HFO, fasting plasma glucose concentrations increased in both the control (5.46 (SD 0.25) vs 5.59 (SD 0.24) mmol/l) and yakult group (5.38 (SD 0.39) vs 5.76 (SD 0.72) mmol/l) (P<0.05) and fasting serum insulin concentrations increased from 12.07 (SD 4.54) vs 14.01 (SD 5.53) mmol/l for controls and 11.24 (SD 3.13) vs 14.52 (SD 3.67) mmol/l for the Yakult intervention, (P<0.05). Insulin sensitivity decreased after HFO by 28.7% (control) and 36% (Yakult) and was not rescued by probiotic supplementation (P=0.874). Fasting plasma triglycerides (TG) decreased between day 1 (0.90 (SD 0.28) mmol/l) and day 22 (0.73 (SD 0.18) mmol/l) (P<0.05) in the Yakult intervention, with no further change post HFO (0.66 (SD 0.22) mmol/l) (P=0.223)). Controls experienced no pre-experimental period change (P=0.355) in TG, but significant decreases post HFO (day 22: 0.88 (SD 0.23); day 29: 0.70 (SD 0.21) mmol/l) (P<0.05). The observed probiotic induced lowering of plasma TG suggests a potential role of probiotics in preventing hypertriglyceridemia and T2DM precursors. These results argue against the role of the gut in the initial development of insulin resistance, however with chronic over-nutrition and insulin resistance benefits may arise from gut manipulation.

INTRODUCTION

A shift in societal norms, now favoring a sedentary lifestyle, both at work and at home, in combination with increases in the consumption of energy dense foods, has contributed to the global obesity epidemic (Swinburn et al., 2011). As a societal and financial burden, obesity accounts for 2-7% of total health care costs in developed countries, and has recently become one of the largest preventable causes of disease globally (Dobbs et al., 2014). Within this context, a pressing issue is obesity’s relationship with Insulin Resistance and Type 2 Diabetes Mellitus (T2DM), and the adverse health consequences associated with these metabolic disorders.

As the co-morbidities of obesity such as Insulin Resistance and T2DM pose the biggest threats to morbidity and mortality risk, they are extensively studied conditions. Despite this, the exact pathophysiology of Insulin Resistance and T2DM remains unknown, however may mechanisms have been proposed to have an underlying role. (Després & Marette, 1999). Since research has found the main postprandial site of glucose uptake to be skeletal muscle (DeFronzo & Tripathy, 2009), it is often suggested reductions in glucose uptake at this level are involved in the development of insulin resistance, perhaps due to decreased tissue blood flow (Laakso, Edelman, Brechtel & Baron, 1990), and/or impaired GLUT4 expression and translocation (Tremblay, Lavigne, Jacques & Marette, 2001), as well as reduced glucose oxidation (Gaster & Beck-Nielsen, 2004). However, considerable research attention indicates there are, in fact, a number of other potential mechanisms and metabolically-active tissues that can be associated with the development of insulin resistance, such as: increased hepatic glucose production (Shi et al., 2006; Brøns et al., 2009), elevated circulating levels of pro-inflammatory cytokines (Boutagy, McMillan, Frisard, & Hulver, 2016; Hong, Kang, Lee & Yu, 2016), diminished pancreatic -cell function (Buchanan et al., 2002), elevated free fatty acid induced lipotoxicity, and eventually, reduced insulin secretion (Defronzo, 2004).

A well as the above-mentioned mechanisms, recent research suggests the gut microbiota (the species of bacteria living within the gut) may play an important role in the development of insulin resistance, through influences on gut permeability (Diamant, Blaak, & de Vos, 2011). Impairment in gut barrier function has been proposed to increase the translocation of gut-derived endotoxin (Lipopolysaccharide [LPS]) into the circulation resulting in the activation of pro-inflammatory pathways and the release of pro-inflammatory cytokines (Caricilli & Saad, 2013). This pathophysiology is highlighted in cross-sectional studies of type 2 diabetics (Blaak et al., 2011), supporting a potential role of the gut microbiota in the development of Insulin Resistance.

Balance of the main compositional elements of the gut microbiota (ratio of Bacteroidetes:Firmicutes) is key in the maintenance of a healthy gut environment (Clemente, Ursell, Parfrey, & Knight, 2012). Cross-sectional research in obese subjects has demonstrated bacterial phyla changes reflecting a reduction in Bacteroidetes and proportional increase in Firmicutes when compared with lean individuals (Brugman et al., 2006; Valiquette, Sirard & Laupland, 2014). Intervention studies have demonstrated that the gut microbiota is sensitive to nutrient shifts resulting from dietary change (Turnbaughet al., 2006) with high-fat feeding interventions demonstrating decreases in Bifidobacterium quantity and increases in LPS (Murphy, Velazquez, & Herbert, 2015). These perturbations suggest that alongside the development of metabolic dysfunction there are accompanying changes in the microbiota phyla which may contribute to the development of insulin resistance. Clear associations between probiotic supplementation and Bifidobacterium levels have been demonstrated (Bird & Conlon, 2015), suggesting potential benefits of probiotic supplementation, a concept which is supported by the significant positive correlation which Larsen et al., (2010) demonstrated between Bifidobacterium levels and glucose tolerance. This manipulation of the gut microbiota (through prebiotic use) has also been associated with reductions in the negative impact of high-fat overfeeding with respect to metabolic endotoxaemia and intestinal permeability (Cani et al., 2009). Further research, focusing on probiotic supplementation (Lactobacillus casei Shirota [LcS]), has demonstrated reductions in high-fat over-eating-induced impairments in insulin sensitivity in mice (Gangarapu, Yildiz, Ince, & Baysal, 2014) and humans (Hulston, Churnside & Venables, 2015).

Whilst research has suggested probiotics are beneficial to metabolic health (Hulston et al., 2015), there are still many unknowns, and a number of limitations with the existing research. Current human research has been conducted in limited samples, and lacks specific measures to determine whether high fat feeding increases gut permeability and contributes to the development of systemic inflammation, therefore preventing confirmation LcS supplementations effect on physiological outcomes. With this in mind, the current study aims to address these issues through mechanistic measures of physiological variables, on a large sample, to provide objective evidence to support it’s claims. These developments are needed to investigate the possibility of a causal link between gut microbiota composition and insulin sensitivity, with the aim of facilitating the development of probiotic supplementation interventions as a valid intervention for T2DM and Insulin Resistance treatment, by reducing the effects of high-fat diets and obesity on insulin sensitivity.

This study therefore, aims to test the hypothesis that participants in the intervention group undertaking 4 weeks of probiotic (LcS) supplementation will have reduced impairment in insulin sensitivity, which has been induced by short-term, high-fat, overfeeding when compared with controls, in young healthy males and females. This study will develop upon previous research done by Hulston et al. (2015) by including objective measures of physiological variables, in an attempt to establish a causal link between gut microbiota composition and insulin sensitivity in humans.

METHODS

SUBJECTS

The current study was completed on 34 young, healthy, male and female volunteers, (28 males and 6 females), aged 18-30. Participant inclusion criteria included having: a normal BMI (18.5 kg/m² – 24.9 kg/m²), no known cardiovascular or metabolic diseases, an omnivorous diet not restricted by a calorie-controlled diet, and a schedule that allows 3 lab visits during a 4-week study period. Participant exclusion criteria included: taking regular medication or having been on antibiotics in the past 3 months, having given blood, used probiotic/prebiotic supplements within the last 3 months, were unwilling to restrict their consumption of dairy products, have known food intolerances or allergies which would prevent them from eating the high-fat diet, were pregnant, or participating in another clinical trial. The study was approved by the Loughborough University Ethical Subcommittee for Human Participants. Participants were informed of the procedures and possible risks before written informed consent was acquired.

PRE-SCREENING

Prior to starting the study, participants visited the laboratory and underwent a pre-screening protocol, which included a re-explanation of study requirements and experimental procedures and the assessment of baseline anthropometric characteristics. Height (to the nearest mm) and body mass (to the nearest 10 g) was recorded to allow calculation of BMI (kg/m²) and for later use in estimating resting energy expenditure (REE) and dietary planning.

GENERAL STUDY DESIGN

After anthropometric assessment, participants were pair matched as closely as possible and then randomly assigned to either a control group (n=17) or probiotic intervention group (n=17). The probiotic used in the current study intervention was LcS, as previously used by Hulston et al. (2015).  It was administered in the form of a 65ml Yakult drink, and consumed by the intervention group twice daily, from days 1-28. Probiotic consumption started after the first experimental trial (assessment of glycaemic control) that took place in the morning of day 1 (see Figure 1 below). Four weeks of probiotic supplementation was used as previous research has demonstrated this time period effectively alters gut microbiota composition (Matsumoto et al., 2010).

Between days 1-21, all participants maintained their habitual diets, recording food and drink intake on 3 days each week. On days 22-28, both groups consumed a high-fat (65% of energy from fats) diet which overfed by 50% in terms of the participants daily energy requirement. On days 1, 22 and 29 a fasted oral glucose tolerance test (OGTT) was completed, to determine the effect of high-fat overfeeding on insulin sensitivity (see Figure 1 for study protocol overview).

EXPERIMENTAL PROCEDURES

Participants arrived at the laboratory between 07:00 and 09:00 on days 1, 22 and 29, after a minimum overnight fast of 10hrs. Anthropometric measures were taken upon arrival and repeat-visit informed consent gained on days 22 and 29. A 20-gauge catheter (Venflon; BD, Plymouth, UK) was inserted into an antecubital vein in the forearm to allow repeated blood sampling throughout the OGTT. An initial 15ml blood sample was taken before participants consumed a 75g glucose drink (82.5g of dextrose monohydrate dissolved in 300ml of water), after which subsequent 10ml blood samples were taken at 15, 30, 45, 60, 90 and 120min post glucose ingestion. Blood samples were split evenly between vacutainer tubes containing either EDTA or a clotting catalyst (Becton Dickinson). EDTA tubes were stored on ice until centrifugation, whereas serum tubes were left at room temperature until clotting was complete. Both sets of blood tubes were then centrifuged for 15 min at 2,500g and 4°C before being transferred to Eppendorfs for storage at -20C until analysis.

DIET RECORDS, ANALYSIS AND COMPLIANCE DURING OVERFEEDING

Participants completed standardised diet records detailing types and amounts of food and drink consumed on 3 days each week for the initial 3 weeks (pre-experimental period). Participants were given instructions on how to complete the forms and digital scales to weigh their food. However, due to the inherent lack of validity associated with the under-reporting of energy intake (Dhurandhar et al., 2015), it was decided that energy requirements would be estimated from a standardised equation. REE was estimated using the equation of Mifflin et al. (1990), and the result subsequently multiplied by a standard correction for physical activity (1.6 for females or 1.7 for males) to determine total daily energy requirements. During the high-fat diet period (days 22-29) participants consumed 150% of the energy requirement derived from the equations/ process described above, with 65% of the energy being provided from fats. All food was purchased and prepared by the research team, and participants were given instructions on when to eat and how to cook the foods. Participants were instructed about the need for diet adherence and told; to eat all the food provided, not to consume foods which are not provided as part of the experimental diet, and in the event that they could not consume any food, they were asked to bring the uneaten food back into the lab. This allowed necessary adjustments to be made.

BLOOD ANALYSIS

Plasma was analysed with spectrophotometry assays using a semi-automatic analyser (Pentra, 400; Horiba Medical, Northampton, UK) to determine concentrations of glucose, triglycerides (TG) and C-Reactive protein (CRP). The Coefficient of Variation (CV) for glucose, TG and CRP was <1%, <4% and <2% respectively.

Serum insulin concentrations were determined using an enzyme linked immunosorbent assay (ELISA) (EIA-2935; DRG Instruments GmbH). The CV of this method was <4%.

CALCULATIONS

The Matsuda Insulin Sensitivity Index was used to determine whole-body insulin sensitivity from plasma glucose and serum insulin concentrations, with 10,000 representing a constant that allows numbers ranging between 1 and 12 and correction of the non-linear distribution of values achieved through the square root conversion (Matsuda & DeFronzo, 1999).

The area under the curve for plasma glucose and serum insulin was calculated using the trapezoid rule.

The degree of insulin resistance was determined from fasted levels of plasma glucose and serum insulin using the homeostatic model assessment (HOMA) method (Matthews et al., 1985)

STATISTICAL ANALYSIS

The primary outcome of the current study was whole body insulin sensitivity, determined from glucose and insulin concentrations attained during an OGTT.

The secondary outcomes investigated were changes in anthropometric measurements, CRP and TG. Unless otherwise stated all data, figures and results are displayed as means  standard deviation (SD). Statistical analysis was performed using SPSS version 24.0 for Mac OS (SPSS, Inc.). To determine statistical significance in response to High-Fat Overfeeding (HFO), both over time and between conditions, a three-way (baseline vs pre-HFO vs post HFO) repeated measures ANOVA, with a between-subject variable (control vs Yakult intervention) was completed with the statistical significance level set at P<0.05. Independent samples T-tests were performed when a time*condition interaction was observed.

RESULTS

ENERGY INTAKE AND DIET COMPOSITION

Estimated energy requirements (basal and 50% overfeeding) were not significantly different between the control (basal: males 3376 kcal (SD 170); females 2354 kcal (SD 206); 50% overfeeding: males 5064 kcal (SD 255); females 3532 kcal (SD 309)) and Yakult intervention groups (basal: males 3335 kcal (SD 250); females 2258 kcal (SD 232); 50% overfeeding: males 5002 kcal (SD 376); females 3386 kcal (SD 347)) (P=0.767). Average energy requirement and provision values are displayed in Table 1.

WEIGHT GAIN AND BMI POST HIGH-FAT OVERFEEDING

Post 7 days of HFO, with average fat consumptions of between 245 to 366g, body mass had increased by 1.72 (SD 1.33) kg and 1.47 (SD 0.79) kg in the control and Yakult groups respectively (P<0.05). Significant increases in BMI were also seen to occur due to 7days HFO, with increases of 0.54 (SD 0.4) kg/m² seen in controls and 0.48 (SD 0.25) kg/m² in the Yakult group (P<0.05). Anthropometric changes are displayed in Table 2.

FASTING PLASMA GLUCOSE AND SERUM INSULIN

Fasting plasma glucose concentrations were collected from all 17 subjects in both condition. Fasting serum insulin concentrations were collected for 16 participants in the control group and 17 participants in the Yakult intervention. Fasting plasma glucose concentrations were maintained from baseline to day 22 (P=0.194), however increased following 7 days HFO (day 22 vs 29) in both conditions (5.46 (SD 0.25) mmol/l vs 5.59 (SD 0.24) mmol/l and 5.38 (SD 0.39) mmol/l vs 5.76 (SD 0.72) mmol/l for control and Yakult respectively) (P<0.05). There was no effect of condition on fasting plasma glucose concentrations (P=0.0851) with no time*time interaction (P=0.155).

Fasting serum insulin concentrations were well maintained during the pre-experimental period (days 1-22) in both conditions (P=0.877). Post 7 days HFO fasting serum insulin concentrations increased from 12.07 (SD 4.54) mmol/l to 14.01 (SD 5.53) mmol/l for controls and 11.24 (SD 3.13) mmol/l to 14.52 (SD 3.67) mmol/l for Yakult intervention, (P<0.05). No effect of condition (P=0.782) or time*condition interaction was observed (P=0.210). Fasting substrate values were collected from the
0 time-point of OGTT’s conducted on days 1, 22 and 29 (see Table 3).

FASTING PLASMA TRIGLYCERIDES

There was no significant effect of condition on fasting plasma TG concentrations (P=0.490) but a statistically significant effect of time (P<0.05) was observed. Fasting TG were significantly lowered during the pre-experimental period in the Yakult intervention (baseline: 0.90 (SD 0.28) mmol/l; day 22: 0.73 (SD 0.18) mmol/l (P<0.05)) and subsequently maintained throughout HFO with a mean post HFO value of 0.66 (SD 0.22) mmol/l (P=0.223). Fasting TG did not change during the pre-experimental period in the control group (P=0.355), however a significant decrease was observed post HFO (day 22: 0.88 (SD 0.23) mmol/l; day 29: 0.70 (SD 0.21) mmol/l) (P<0.05). A significant time*condition interaction was observed (P<0.05), with the difference in response between conditions occurring at day 22 (P<0.05), due to the earlier lowering of TG levels in the Yakult intervention (see Figure 2).

FASTING CRP

Due to technical issues fasting CRP concentrations were only attained for 29 subjects (15 control and 14 Yakult intervention). There was no significant effect of condition on fasting CRP concentrations (P=0.499) and no significant change in fasting plasma CRP concentrations between baseline and day 22 (P=0.170), however a significant effect of time was found (P<0.05) with a significant increase in fasting plasma CRP post HFO in both the control (day 22: 0.34 (SD 0.25) mmol/l vs day 29: 0.56 (SD 0.52) mmol/l) and Yakult intervention (day 22: 0.28 (SD 0.21) mmol/l vs day 29: 0.37 (SD 0.26) mmol/l) (P<0.05) (see Figure 3). No time*condition interaction (P=0.092) was demonstrated, however the current level of statistical significance suggests further analysis, with more participants, may generate a statistically significant time*condition interaction at
the level of P=0.05.

ORAL GLUCOSE TOLERENCE TEST

When analysing plasma glucose and serum insulin responses to an OGTT, AUC was calculated and used. HFO did not cause a significant change in plasma glucose AUC in both the control (baseline: 841 (SD 136); day 22: 835 (SD 114); day 29: 862 (SD 113) mmol/l) (Figure 4:(b)), and Yakult intervention (baseline: 810 (SD 120); day 22: 801 (SD 154); day 29: 834 (SD 122) mmol/l) (Figure 4:(d)), with analysis showing no significant effect of time (P=0.285) or condition (P=0.431) and no time*condition interaction (P=0.986). Glucose concentrations throughout the OGTT can be seen in Figure 4:(a) (control) and 4:(c) (Yakult).

Serum insulin AUC was not significantly different between groups (P=0.284) with no observable time*condition interaction (P=0.350). Statistical significance was however reached for the effect of time on serum insulin concentrations (P<0.05), with no difference day 1 vs day 22 (P=0.151) but a significant increase in serum insulin AUC post HFO in both the control group (baseline: 6653 (SD 2861); day 22: 7455 (SD 3733); day 29: 8417 (SD 3561) mmol/l) (Figure 5:(b)), and Yakult intervention (baseline: 6614 (SD 1987); day 22: 6111 (SD 2528); day 29: 7302 (SD 2106) mmol/l) (Figure 5:(d)) (P<0.05). Serum insulin concentrations throughout the OGTT can be seen in Figure 5:(a) (control) and 5:(c) (Yakult).

HOMA IR

Insulin resistance, determined by the HOMA-IR was not significantly different between conditions (P=0.769), with no time*condition interaction observed (P=0.241). There was however, a significant effect of time with IR increasing post HFO in both the control group (day 22: 2.95 (SD 1.06); day 29: 3.49 (SD 1.44)) and Yakult intervention (day 22: 2.68 (SD 0.74); day 29: 3.78 (SD 0.67)), (P<0.05).

INSULIN SENSITIVITY INDEX

Post 7days HFO whole body insulin sensitivity (assessed by the Matsuda ISI) was impaired in both the control group (decreasing by 28.7% from 4.6 (SD 3.2) to 3.3 (SD 1.8) (P<0.05) and the Yakult intervention (decreasing by 36% from 5.1 (SD 2.8) to 3.3 (SD 1. 5) (P<0.05) (see Figure 6). No significant effect of condition was observed (P=0.874) and no time*condition interaction (P=0.457).

DISCUSSION

The main finding of the current study was that HFO negatively affects multiple aspects of glycemic control, including increased fasting plasma glucose and serum insulin concentrations and decreased in insulin sensitivity, which were not rescued by 4wks of supplementation with LcS. This finding is largely congruent with previously published literature, which demonstrates reductions in insulin sensitivity and increases in postprandial insulin AUC after HFO (Hulston et al., 2015; Parry, Smith, Corbett, Woods & Hulston, 2017; Parry, Woods, Hodson, & Hulston, 2017), however the inability of LcS supplementation to prevent diet induced insulin resistance contradicts the original findings of Hulston et al. (2015) and suggests that after short-term HFO the initial mechanism responsible for impairments in insulin sensitivity may not originate from diet induced changes in the gut microbiota, as originally hypothesised.

It is important to note that the current study did not demonstrate any differences in weight gain, between conditions, after HFO. This is of interest as it has previously been demonstrated that supplementation with particular Lactobacillus species has resulted in weight gain in humans and mice (Million et al., 2012) which, if correct, would argue against probiotics use as a therapy for T2DM, due to weight status being a known correlate of the disease. Although the current finding appears to contradict this theory, supplementation only involved LcS and so further investigation, with a wide range of Lactobacillus strains, is required to test the validity of this finding.

In the current study we chose to use 1 week HFO as research in both humans and rodents has shown that short-term HFO lasting between 4 days and 4 weeks is sufficient to induce insulin resistance (Bielohuby et al., 2013; Wiedemann, Wueest, Item, Schoenle, & Konrad, 2013). Our dietary manipulation period of 1 week was sufficient enough to allow dietary-induced changes in glycaemic control to occur, without being too long that risk of compliance became an issue.

FASTING PLASMA GLUCOSE AND INSULIN – ISI

The exact physiological mechanisms underpinning the observed impairment in glycaemic control remain uncertain, however there are many theories which have received considerable research attention. It is often suggested that reduced glucose uptake by skeletal muscle occurs as a result of decreased glucose transport and GLUT4 translocation following HFO (Tremblay et al., 2001). The knowledge that skeletal muscle is the main site of postprandial glucose uptake (DeFronzo & Tripathy, 2009), implies that impairments at this level would have large effects on glycaemic control, and may provide rationale for our current finding of increased fasting plasma glucose concentrations post HFO. Previous rodent research by Tremblay et al. (2001) looking at the effects of HFO has demonstrated diet-induced reductions in GLUT4 translocation to both the plasma membrane and T-tubules of the muscle in response to insulin, suggesting during the current OGTT the observed maintenance of glucose AUC may be due to the observed increase in insulin secretion, in an attempt to counter the effects of HFO and still stimulate GLUT4 translocation, but additional insulin sensitive mechanisms such as GLUTX1 (Ibberson, Uldry, & Thorens, 2000) may have been required to support the maintenance of plasma glucose. Due to the lack of direct objective measurement, the involvement of this mechanism in the current study findings remains speculative and warrants further investigation.

The suggestion that HFO induced impairments in glycaemic control result from impaired insulin sensitivity mediated glucose transport, may be due to increased serine phosphorylation of insulin receptor substrate-1 (IRS-1) which inhibits insulin activation of phosphatidylinositol (PI) 3-kinase and thus GLUT4 translocation (Zierath, Houseknecht, Gnudi, & Kahn, 1997), a mechanism which has been demonstrated to occur without alteration of IRS-1 tyrosine phosphorylation. This is consistent with the research demonstrating that impairments in IRS-1 signalling occur later in the development of insulin resistance as so highlights the possibility of a key role of PI 3-kinase mediated glucose transport in early insulin resistance.

Another potential influence on glycaemic control is oxidative stress, as after short-term HFO skeletal muscle has also been shown to demonstrate increased markers of oxidative stress (an early consequence of over-nutrition) suggesting chronic over-nutrition and the resultant sustained high levels of reactive oxygen species, may lead to impairment of mitochondrial proteins involved in metabolism, resulting in mitochondrial dysfunction and skeletal muscle TG accumulation, which usually characterises obesity and T2DM (Samocha-Bonet et al., 2012). Although evidence of the effects of oxidative stress in healthy humans is limited, current literature suggests oxidative stress positively correlates with CRP levels (Boyle et al., 2011; Cottone et al., 2006), indicating the increase in fasting CRP levels demonstrated in this study may be caused by diet-induced oxidative stress. This highlights the negative health consequences of HFO as elevated CRP is associated with an increased risk of coronary heart disease (CHD) and atherosclerosis progression, suggesting even moderate over-nutrition may lead to early manifestations of CHD development.

The current finding of increased fasting plasma glucose post HFO suggests glycaemic control may also be reduced through downregulation of glycolytic flux, due to preferential oxidation of fats as a result of their increased availability from exogenous sources. This notion is supported by the recent research of Lundsgaard et al. (2017) which reported lower respiratory exchange ratios following HFO and increased CPT1 mRNA, reflecting an enhanced capacity for mitochondrial fatty acid transport and oxidation. This is also reinforced by findings of an increase in basal PDK4 mRNA and PDH-E1 Ser³⁰⁰ phosphorylation after high-fat-feeding, indicating an attenuation of PHD activity (Lundsgaard et al., 2017). This is congruent with our current finding of increased fasting glucose concentrations post HFO and suggests the mechanisms responsible for maintaining glucose AUC during an OGTT may not be increased uptake but rather a compensatory increase in serum insulin levels, as shown in the current study, potentially through increased secretion and reduced clearance, consistent with early insulin resistance models demonstrated in obese subjects (Cavaghan, Ehrmann, & Polonsky, 2000). However, this notion can only be speculated, and further research utilising stable isotope tracers is required.

Although previous research has demonstrated reductions in fasting plasma glucose concentrations with probiotic supplementation (Ejtahed et al., 2012), this finding was not replicated in the current study or the original pilot study (Hulston et al., 2015). This suggests the benefit of probiotics on fasting glucose concentrations may only manifest in subjects with pre-existing metabolic syndrome, and not in lean populations, with normative glucose values like those used in the current and pilot study.

CRP concentrations are often used as a proxy for inflammation and represent systemic low-grade inflammation (Cottone et al., 2006), a state which has frequently been implicated in the development of insulin resistance due to it’s association with  elevated LPS concentrations, and resultant stimulation of pro-inflammatory cytokine production (Creely et al., 2006) through it’s binding to the CD14 complex on the surface of the gut endothelium (Diamant, Blaak & de Vos, 2011). This in turn leads to defective insulin signalling through modulation of serine phosphorylation of IRS-1 (Hotamisligil et al., 1996). It has been suggested dietary-induced elevations in circulating LPS and thus development of metabolic endotoxaemia (Neves, Coelho, Couto, Leite-Moreira, & Roncon-Albuquerque, 2013) originates from increases in the Bacteriodetes:Firmicutes ratio, a physiological state often seen in obese subjects (Brugman et al., 2006), which results in increases in gut permeability and allows the translocation of LPS into the circulation (Cani et al., 2008). Bifidobacterium supplementation has been shown to have advantageous effects on gut barrier function in obese mice, with reductions in plasma LPS concentrations and thus an attenuated impact of high-fat-diet-induced metabolic endotoxaemia occurring after supplementation with prebiotics (Cani et al.,2009). As it has also been demonstrated probiotic supplementation increases gram-positive bacteria, (Conlon, & Bird, 2015) it may be possible for probiotic supplementation to facilitate gut barrier function in humans and thus prevent LPS mediated insulin resistance. However, as the increase in LPS is said to result in altered serine phosphorylation of IRS-1 (Hotamisligil et al., 1996), and this impairment is not detected in the early stages of insulin resistance (Tremblay et al., 2001), it may be that probiotic supplementation may be beneficial when over-nutrition has occurred chronically and insulin resistance has persisted, at which point impairments in gut barrier function may be responsible for the insulin resistance. As the current study demonstrates that 7 days of HFO is sufficient cause insulin sensitivity impairments and insulin resistance of up to 36% and 41% respectively, and was not able to be rescued by probiotic supplementation, our findings also support the idea that in the short term, insulin resistance does not originate from the gut, and thus is unable to be prevented by probiotic supplementation, however direct measurement of the gut microbiota through faecal samples is required to confirm this speculation.

As insulin resistance was determined using the HOMA-IR, it suggests the observed increase in insulin resistance may be a product of hepatic insulin resistance (Parente et al., 2014) due to the reliance of the equation on fasting concentrations of glucose and insulin. This notion supports the lack of intervention effect as probiotics target metabolic endotoxaemia induced IR (Cani et al., 2009), rather than hepatic IR, suggesting future research may benefit from the use direct measures of insulin resistance such as the use of the euglycaemic insulin clamp to more accurately define the level of insulin resistance.

With this in mind it is plausible to suggest that the observed increase in fasting plasma glucose may be a result of diet-induced hepatic insulin resistance rather than peripheral insulin resistance in accordance with the findings of Brøns et al. (2009) who demonstrated that following a diet protocol similar to that used in the current study, for 5 days, led to a 26% increase in hepatic glucose production, despite an accompanying increase in fasting insulin. As these results are congruent with those of the current study, it gives strength to the concept of compensatory insulin secretion to account for diet-induced hepatic steatosis and the associated hyperglycaemia and impairment in insulin signalling (Wiedemann et al., 2013). The observed increase in insulin secretion may also be a result of high-fat-diet-induced increases in gastroinhibitory peptide

(GIP) secretion, which if persistent can result in higher GIP circulating concentrations as is seen in those with T2DM (Carr et al., 2008). This implies that even short term HFO, a behaviour that reflects common practice in today’s society, can have negative consequences for healthy individuals, and may be a contributory factor to the development of insulin resistance and thus T2DM.

FASTING PLASMA TG

A significant decrease in fasting plasma TG was observed in both the control and yakult intervention group, however the respective timing of the decrease was different. Within the control group, fasting plasma TG levels decreased post HFO, an observation which receives support from existing literature (Brøns et al., 2009; Cahova, Dankova, Palenickova, Papackova, & Kazdova, 2012; Hulston et al., 2015). It is suggested this decrease is not due to HFO induced impairments in lipolysis (shown by a lack of effect on lysosomal lipase), but rather a reduced hepatic TG secretion rate (Heden et al., 2014), reductions in free fatty acid (FFA) synthesis de novo and channelling of TG derived FFA towards oxidative utilisation (Cahova et al., 2014) as a result of increased exogenous fats. These mechanisms also support our finding of increased fasting plasma glucose concentrations, which would occur as a result of reductions in glycolysis in favour of FFA oxidation. In the short term, one, or a number of these mechanisms may act as a protective response of the body to prevent hypertriglyceridemia, however if occurrence becomes chronic, these mechanisms may become saturated, not be able to be maintained, and TG concentrations will increase, with the pathophysiology reflecting states often seen with obesity and T2DM.

The decrease in fasting plasma TG in the Yakult intervention group was observed pre HFO, with concentrations subsequently maintained throughout the over-feeding period, suggesting the presence of a lower bound threshold for TG levels, and highlighting the possibility probiotic supplementation may be able to reduce fasting plasma TG levels, irrespective of diet. If a true effect, this finding could prove invaluable to health solutions due to the strong links between elevated TG levels and CHD risk, independent of cholesterol levels (Morrison & Hokanson, 2009). The mechanisms underpinning the probiotic induced lowering of TG levels remains unclear as only recently have studies with human subjects started to investigate the benefits of probiotic supplementation on secondary physiological markers. Apolipoprotein A-V (apoA-V) has been implicated in the manipulation of plasma TG, in both humans and mice, potentially through enhancement of lipoprotein lipase (LPL) mediated TG hydrolysis (Nilsson, Heeren, Olivecrona & Merkel, 2011). Recent research has demonstrated 12 weeks of probiotic supplementation with Lactobacillus plantarum and Lactobacillus curvatus significantly increased apoA-V levels, with a negative correlation observed between apoA-V levels and plasma TG concentrations (Ahn et al., 2015). This suggests that the TG lowering effects of LcS supplementation demonstrated in the current study, may be mediated by probiotic induced increases in apoA-V concentrations and a resultant increase in LPL mediated TG hydrolysis, however due to the lack of objective measures in this area in the current study the exact mechanisms responsible remain speculation at best. The difference in timing of reductions in plasma TG between the control and probiotic supplementation group suggests the reductions were achieved via different mechanisms, however what these exact mechanisms are remains unknown, and requires further investigation using stable isotope tracers.

FASTING PLASMA CRP

The lack of intervention effect on insulin sensitivity and thus the proposed hypothesis that early development of high-fat-diet-induced insulin resistance is due to mechanisms at the liver and muscle rather than metabolic endotoxaemia also receives support from the current finding that after 7 days of HFO, systemic low grade inflammation, as shown by CRP concentrations, was increased in both the Yakult and control group, with no attenuation due to probiotic supplementation. This suggests initially with HFO the pro-inflammatory environment observed in the current study does not originate from the gut and may be instead may be explained by a diet-induced increase in tumor necrosis factor- (TNF-) expression, which has been linked to increases in CRP production, irrespective of gut permeability (Borst & Conover, 2005; Peairs & Rankin, 2008), although this remains speculation. Although statistical significance was not reached in the current study, there was a tendency for a time*condition interaction for fasting CRP concentrations, with Yakult appearing to slightly reduce fasting CRP levels pre HFO and an attenuate the increase in CRP post HFO. This observed effect may be due to the fact fasting CRP measurements were only attained for a reduced number in the current sample, or it may imply that the contribution of the gut microbiota to the systemic low-grade inflammation, although not dominant after 1 week of HFO and therefore not the main mediator of insulin resistance, may become a crucial factor with prolonged over-nutrition. Further research with larger populations and a greater variation of dietary manipulation lengths is required to discover a true effect.

CONCLUSION

To conclude, the current study demonstrated that high-fat-diet-induced impairments in insulin sensitivity were not attenuated by 4wks of probiotic supplementation with LcS in human participants, suggesting the gut microbiota and it’s compositional changes may not be the main cause of the observed insulin resistance in the short term. The inflammatory marker concentrations observed in the current study, and the tendency for a time*condition interaction, suggests with chronic insulin resistance the gut may be implicated in the pathophysiology. This suggests future research with longer dietary manipulation is needed to investigate the time-course development of insulin resistance and the mechanisms implicated in this. A considerable probiotic induced lowering of plasma TG was observed prior to HFO in the Yakult intervention group, suggesting a potential role of probiotics in preventing hypertriglyceridemia, a condition frequently associated with T2DM (Sniderman, Scantlebury, & Cianflone, 2001), and warranting further investigation to confirm this effect, and determine the underlying mechanisms involved.

ACKNOWLEDGEMENTS

The current study was hypothesised and developed by the lead investigator Dr Carl Hulston with financial support received via an educational grant from Yakult UK limited. Collection and analysis of blood samples was completed by Dr Carl Hulston and Dr Rachel Woods. Francesca Bolt assisted with the preparation of the diet intervention and participant recruitment.

I would like to thank Dr Carl Hulston who hypothesised and developed the current study. I would also like to thank Dr Rachel Woods for completing the collection of blood samples and for completing blood analysis with Dr Carl Hulston. I would like to thank Francesca Bolt for her assistance with the preparation of the diet intervention and participant recruitment.