Adipose tissue metabolic and inflammatory responses to a mixed meal in lean, overweight and obese men

Thirty men aged between 35 and 55 years were recruited by local advertisement and following preliminary anthropometric assessment and were classified according to waist circumference as lean ≤94 cm, overweight 94–102 cm and obese ≥102 cm [6]. Recruitment was on a first-come first-served basis and took place between September 2011 and April 2012 until there were 10 participants in each waist circumference category. Participants attended one main trial in the Physiology Laboratories at the University of Bath, in which blood and adipose tissue samples were obtained before and 6 h after consumption of a mixed meal with further blood samples taken at regular intervals throughout. Primary outcome measures included examination of changes in adipose tissue gene expression and secretion of selected cytokines from adipose explants after the meal according to levels of adiposity (and postprandial metabolic responses in blood). This trial forms part of a larger investigation in which immune cells present in adipose tissue were characterized and their relationships with baseline adipose tissue inflammation examined, as has been previously reported [7]. All procedures followed were in accordance with the protocol reviewed and approved by the National Health Service South West (Southmead) Research Ethics Committee (11/SW/0193) and have therefore been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. This trial is registered at ClinicalTrials.gov (ID: NCT02416843), and all participants gave their written informed consent prior to their inclusion in the study. Individuals were excluded from participation if they had a history of diabetes, cardiovascular disease or dyslipidaemia, were taking any medications known to interfere with immune function or lipid/carbohydrate metabolism, if they smoked or had not been weight stable for >3 months (i.e. weight change >3 %; [8]). In addition, participants were excluded if they reported food intolerances/allergies to any component of the meal (e.g. dairy or wheat) and habitually performed more than 6 h of vigorous-intensity physical activity or 10 h of moderate-intensity physical activity per week, assessed via a self-report questionnaire (as these individuals would be unlikely to fit the meal standardization procedure). Participant characteristics within the categories lean, overweight and obese based on measurements of waist circumference are given in Table 1.

Participant waist circumference, body mass and height were measured and body mass index calculated [BMI; mass (kg)/height² (m²)]. Participants were fitted with a combined heart rate and accelerometry monitor (Actiheart) for a period of 9 consecutive days to determine habitual physical activity level (PAL; total energy expenditure/basal metabolic rate [10]). Participants were asked to maintain their normal lifestyle habits/routines during this period with the first 2 days of activity monitoring being excluded from analysis to account for potential reactivity [11].

Participants were asked to refrain from performing any strenuous physical activity for 48 h and consuming alcohol/caffeine for 24 h prior to testing. Trial days were scheduled so participants had been free from any self-reported illness for a minimum of 2 weeks in order to reduce immune system disturbance. Participants arrived in the laboratory in the morning after an overnight fast (minimum 10 h) and after consuming 1 pint of water upon waking. Body mass (post-void) was determined using a digital balance (Tanita Corp., Japan) with participants wearing lightweight shorts. Dual-energy X-ray absorptiometry (DEXA; Discovery, Hologic, Bedford, UK) was used to estimate lean and fat mass. Central adipose tissue (abdominal subcutaneous and visceral adipose tissue) was estimated from a central region between L1 and L4 which has previously been shown to be comparable to estimates of central adipose tissue measured by computerized tomography (CT) [12] and correlates with measures of metabolic health [13]. Fat mass index (FMI) was calculated using the equation FMI = total fat mass (kg)/height² (m²) and interpreted using ranges that match the WHO BMI classifications [14]. Indirect calorimetry was used to estimate resting metabolic rate (RMR) after participants had rested supine in bed for 10 min [15]. This value was used to adjust estimates of total energy expenditure and PAL and to calculate the energy requirements for the test meal given to participants.

Total RNA was extracted from whole adipose tissue using the RNeasy mini kit (Qiagen, Crawley, UK). Samples were quantified (Qubit 2.0 fluorimeter, Life Technologies, Paisley, UK) and 2 μg reverse transcribed to cDNA using a high-capacity reverse transcription kit (Applied Biosystems, Warrington, UK). Real-time PCR was performed using a StepOne™ (Applied Biosystems) using pre-designed primers and probes obtained from Applied Biosystems for measurement of glucose transporter type 4 (GLUT4; Hs00168966_m1), insulin receptor substrate 2 (IRS2; Hs00275843_s1), hormone-sensitive lipase (HSL; Hs00193510_m1), leptin (Hs00174877_m1), adiponectin (Hs00605917_m1), monocyte chemoattractant protein 1 (MCP-1; Hs00234140_m1), interferon gamma-induced protein 10 (IP-10; Hs01124251_g1), interleukin 6 (IL-6; Hs00985639_m1), interleukin 8 (IL-8; Hs99999034_m1), interleukin 10 (IL-10; Hs00961619_m1), interleukin 1 receptor antagonist (IL-1Ra; Hs00893626_m1), tumour necrosis factor alpha (TNF-α; Hs99999043_m1), interleukin 1 beta (IL-1β; Hs01555410_m1) and interleukin 18 (IL-18; Hs00155517_m1) expression. Adipose tissue expression of GCSF, MIP-1b and IFN-ɣ was also measured, but data are not shown because they were only detectable in 4–8 individuals. Peptidylprolyl isomerase A (PPIA/cyclophilin A) was used as an endogenous control [20]. Results were analysed using the comparative Ct method and expression normalized to an internal calibrator specific to each gene using the formula \(2^{{ - {\Delta \Delta }C_{\text{T}} }}\) (where ∆∆C _T is [(C _T gene of interest − C _T PPIA) − lowest ∆C _T for gene of interest], and statistical analysis was performed on LN-transformed values [21].

Serum insulin was measured by ELISA (Mercodia, Uppsala, Sweden), and plasma glucose, serum total cholesterol, HDL cholesterol, ALT and triglycerides were measured using commercially available assay kits and analyser (Daytona Rx, Randox, Crumlin, UK). Adipose tissue secretion of IL-6, IL-8, IL-10, IL-1Ra, IP-10, GCSF, MCP-1, MIP-1b and TNF-α was measured using Luminex (Bio-Rad, Hercules, CA, USA).

All data are presented as mean ± standard error of the mean (SEM). Total area under the curve (AUC) for insulin, glucose, triglycerides and NEFA was calculated using the trapezium rule and homoeostasis model assessment for insulin resistance (HOMA-IR) calculated using the formula fasting glucose (mmol/L) × fasting insulin (mU/L)/22.5 [22]. Adipose tissue responses to the meal were compared between the three groups using mixed-model ANOVA (adiposity × time) irrespective of minor deviations from a normal distribution within the normal physiological range [23]. Greenhouse–Geisser corrections were applied to intra-individual contrasts where ε < 0.75; however, for less severe asphericity the Huynh–Feldt correction was selected [24]. Where mixed-model ANOVA identified significant interactions effects, multiple t tests were applied to identify the location of variance both within each group relative to baseline and between groups at level time points, with p values subject to a Holm–Bonferroni correction [24]. Statistical analysis was performed using SPSS version 20 (IBM, Armonk, NY, USA) using an alpha level of p ≤ 0.05. The symbols used to identify significant differences are as follows: * denotes a main effect of time, † denotes a main effect of adiposity and # denotes any adiposity × time interaction effects.

Temporal insulin and glucose responses following the meal and their respective AUCs were elevated with increasing magnitude in proportion to increased adiposity (Fig. 1a, b). No such differences were apparent, however, for the temporal responses or AUCs for NEFA or triglycerides (Fig. 1c, d).

At 6 h after the meal, significant time effects reflected an up-regulation of IL-6 (F = 14.7, p = 0.001) and MCP-1 (F = 10.7, p = 0.003) and down-regulation of IRS2 (F = 24.6, p < 0.001) gene expression across all groups relative to baseline (Fig. 2). Greater expression of IL-18 (F = 8.3, p = 0.002), IP-10 (F = 4.1, p = 0.029), IL-1Ra (F = 6.1, p = 0.006) and MCP-1 (F = 8.8, p = 0.001) and lower expression of GLUT4 (F = 9.9, p = 0.001), IRS2 (F = 5.2, p = 0.012) and HSL (F = 4.8, p = 0.016) with increased adiposity were maintained over the 6-h postprandial period monitored in subcutaneous adipose tissue (Fig. 2). No differences in magnitude or direction of these responses were identified between groups (i.e. no adiposity × time interactions were apparent).

Baseline secretion of adipokines from adipose tissue together with changes following the meal is given in Table 2. Despite changes in gene expression of IL-6 and MCP-1 at 6 h following the meal, no significant changes in secretion were found. Adiposity × time interactions were identified for IP-10 (F = 4.3, p = 0.029) and IL-10 (F = 5.2, p = 0.017); however, post hoc t tests only indicated that IL-10 secretion from the obese adipose tissue was reduced to a level similar to the overweight group at 6 h (p < 0.05). Effects of adiposity on secretion were maintained across both time points for IL-6, MCP-1, GCSF, IP-10, MIP-1b and IL-10. Due to the limited size of some adipose tissue samples obtained, there was only sufficient material to culture tissue for six lean and five overweight individuals.

Within adipose tissue, gene expression of the typically proinflammatory cytokines IL-6 and MCP-1 were increased 6 h following consumption of the meal, which supports previous findings from people with metabolic syndrome [4] and non-obese relatives of people with type 2 diabetes [5]. The magnitude of changes in IL-6, MCP-1 and IRS2 gene expression observed in the present study was equal across all groups from lean to class I obese despite poorer glucose control/poorer insulin sensitivity with increasing levels of adiposity and despite participants with increased adiposity receiving a larger meal (since the meal was given relative to RMR). This indicates that these responses may be part of a ‘normal’ postprandial response within adipose tissue, at least within 6 h following a meal. A possible explanation for the similar responses is that relative insulin resistance in obese adipose tissue (suggested by the chronic down-regulation of GLUT4, HSL and IRS2) may counteract the increased insulin response to the meal and result in the same overall effect as in lean ‘insulin-sensitive’ tissue. Furthermore, given the increased relative mass of adipose tissue in obese individuals compared with lean, exposure of each gram of adipose tissue to insulin and glucose may actually be comparable.

The meal used in this study was relatively high in both carbohydrates and fats. However, it cannot be determined from this study whether either of these or another factor related to the ingestion of either of these meal components (e.g. insulin) is the main stimulus for the increased gene expression of inflammatory cytokines in adipose tissue. Certainly, there is good evidence that insulin infusion alone increases IL-6 and TNF-α expression in subcutaneous adipose tissue from lean men [25, 26] and expression of IL-6, MCP-1 and IL-1β in rodents [27]. The effect of 3 h of hyperglycaemia alone (in the absence of increased insulin) on adipose tissue gene expression has also been examined [28], and although IL-6 and MCP-1 were not reported to be affected [28], this does not exclude glucose as a stimulus for the gene expression changes observed since different sampling time points were used. Assuming that the meal was indeed the key stimulus for changes in adipose tissue gene expression, another component to consider is triglycerides which, like the changes in adipose tissue gene expression, showed no differences in magnitudes of response with increased levels of adiposity. High-fat meal challenges have been shown to induce inflammation in adipose tissue [4, 29], and the type of fat (i.e. saturated or unsaturated) is not thought to be important in determining the extent of postprandial inflammation in adipose tissue [4]. It is possible, however, that adipose tissue may respond to glucose, insulin and triglycerides (and other associated stimuli not examined in this study, e.g. LPS) independent of each other, and adipose tissue sampling at different time points may have produced a different pattern of results. Although the exact stimuli would be difficult to isolate, understanding the dynamics of postprandial inflammatory responses in adipose tissue and how they relate to glucose control may ultimately provide insight regarding the aetiology of obesity-related insulin resistance.

Cells within the adipose tissue stromavascular fraction, in particular macrophages, are likely to be the major sources of adipokines such as IL-6 and MCP-1 [30‐32], although mature adipocytes may also make a substantive contribution [33]. Each component of the meal and the associated insulin response have been shown in vitro to affect inflammation in both macrophages and adipocytes. For example, in experiments using (murine) 3T3-L1 adipocytes, insulin can dramatically enhance MCP-1 expression from these cells [34]. MCP-1 may be an important factor enhancing macrophage recruitment to adipose tissue (via repeated stimulation), as is observed in adipose tissue of obese individuals [35‐38], and can also induce insulin resistance in adipocytes [34]. In vitro work has shown that IL-6 production can be induced in macrophages by high levels of glucose uptake via GLUT1 [39] and by fatty acids and lipopolysaccharides produced by gram-negative bacteria present within the intestine following ingestion of triglycerides [40‐42]. Although chronic elevation of IL-6 is typically regarded as detrimental to insulin resistance and cardiovascular health [43], acute oscillations in IL-6, as seen for example following exercise, may be important in maintaining insulin sensitivity in muscle post-exercise [44, 45]. The implications of a postprandial increase in IL-6 gene expression within adipose tissue and its time course may therefore warrant further investigation.

Cytokine secretion from adipose tissue remained similar in samples taken before and 6 h after the meal, which could be interpreted as evidence that adipose tissue has ‘protective’ mechanisms to prevent acute increases in the secretion of these cytokines in response to changes in gene expression (e.g. into the circulation). There appears to be large variation in postprandial secretion between participants, and this may reflect the wide ranges in adipose tissue responsiveness to the meal components and/or insulin and differences in relative proportions of adipocytes/non-adipocyte cells in adipose tissue. These results are limited to some extent by the small amount of tissue samples and data available for lean and overweight participants. Arterio-venous sampling may instead give a better indication of adipose tissue secretion over the entire duration following the meal and any effects of adiposity.

Based on this study design, we cannot be absolutely certain that the meal directly caused the reported time-related changes. Ideally, we would have been able to include some kind of comparator trial in order to rule out other explanations such as circadian rhythms. Such comparisons have been made previously in rats, which demonstrated that gene expression of inflammatory markers IL-6 and NF-KB was significantly increased (at 2 h) specifically following a high-fat meal compared with water [29]. However, even with such a study design it is difficult to conclude that the changes are due to feeding because of course extended fasting per se also represents a physiological challenge. Furthermore, it must be recognized that interpretation based on just two time points may not necessarily provide a full and valid reflection of the time course of events in vivo over the entire postprandial period. For example, there may have been insufficient time for synthesis of RNA transcripts/proteins for secretion into the circulation by the 6-h time point, or some proteins may already have been released from storage vesicles. In this context, a number of changes in gene expression have been identified at 4 h following lipid challenges [4] and after only 3 h of hyperglycaemia [28] and 2 h of hyperinsulinaemia [26, 27]. Repeated adipose biopsies to generate a time course would be useful but very challenging. It would also be interesting to investigate changes in adipose tissue following subsequent food intake, since blood glucose and insulin responses are known to be much lower after repeated feeding (i.e. Staub–Traugott effect) [46‐48]. This could also help identify whether glucose and insulin are indeed key drivers of the postprandial adipose tissue responses.