Background
The interest of a prebiotic approach in the modulation of gut microbiota for the management of obesity and/or associated metabolic disorders is emerging from recent studies [
1,
2]. Prebiotics are defined as “substrates that are selectively utilized by host microorganisms conferring a health benefit” [
3]. Prebiotics such as Inulin-type fructans (ITF) have largely been studied in the context of obesity and related metabolic disorders. Among the health benefits, various studies highlighted in rodent models of obesity that ITF intake improves gut barrier function [
4], low-grade inflammation [
4,
5], fat mass accumulation [
6‐
8], glucose intolerance [
5,
6,
9], postprandial hypertriglyceridemia [
10], hepatic steatosis [
6,
7,
11,
12], and endothelial dysfunction [
13]. Intervention studies in humans also confirm the interest of ITF supplementation in the management of metabolic disorders and obesity [
14,
15]. Intriguingly, the metabolic improvement following nutritional interventions such as dieting is quite variable from one patient to the other [
16]. Recent studies pointed out individual factors, such as the initial gut microbiota composition, usual dietary fiber intake, or medications (such as metformin), that could influence the metabolic response to ITF in obese individuals [
12,
15]. In this paper, we addressed the hypothesis that physical activity (PA) is another condition that potentially affects both the gut microbiota and prebiotic efficacy. Indeed, exercise training is a well-known cost-effective intervention that can improve or prevent the development of metabolic disorders [
17,
18]. In addition, the gut microbiota composition differs in professional athletes compared to the gut microbiota from sedentary people [
19,
20]. In vivo experiments have shown that exercise alters the gut microbiota composition in normal, obese or diabetic mice, with some differences according to the metabolic status of mice [
21‐
24]. An elegant study identified in prediabetic individuals that the gut microbiota was an important factor influencing the improvement of glucose metabolism and insulin sensitivity during exercise [
25]. Indeed, responders and non-responders to exercise exhibited differences in their gut microbiota composition. Moreover, the metabolic benefit to exercise observed in the responder group was transferable into mice using fecal microbial transplantation. In addition, a previous report demonstrated that 4-weeks of inulin-propionate ester, in combination with a moderate intensity exercise training program, increased the amount of fat oxidized in overweight women. This study suggested that combining PA with inulin-propionate ester is a good strategy for improving lipids metabolism in overweight subjects [
26].
While exercise training and ITF have been shown to impact the gut microbiota independently and are each associated with beneficial health effects, we investigated in both preclinical and clinical interventions, the interaction between inulin intake as prebiotic intervention and PA in the context of obesity. Using data obtained in a cohort of obese individuals treated with inulin supplements coupled to inulin-rich vegetables versus placebo, we addressed two questions. First, does inulin intervention modify PA levels in obese patients? Second, do changes in PA level during the intervention impact on the effect of inulin on gut microbiota and metabolism? In parallel, we designed an intervention study with voluntary exercise and/or inulin supplementation in HF-fed mice to investigate the molecular mechanisms behind the modulation of metabolic and microbial outcomes linked to combined PA and inulin intervention.
Methods
Subjects
This study was approved by the “Comité d’éthique Hospitalo-facultaire de Saint-Luc”. Written informed consent was obtained from all participants before inclusion. The trial protocol was published on protocols.io (
dx.doi.org/10.17504/protocols.io.baidica6), and the trial was registered at ClinicalTrials.gov under identification number NCT03852069.
Screening and inclusion
Male and female subjects were recruited in three university hospitals in Belgium (Hôpital Erasme in Brussels, Centre Hospitalier Universitaire in Liege and Cliniques universitaires Saint-Luc in Brussels). This study was a 3-month long, multicentric, single-blind, and placebo-controlled trial. The inclusion criteria were as follow: body mass index (BMI) > 30 kg/m2, aged 18 to 65 years, Caucasian ethnicity, and the presence of at least one obesity-related metabolic disorder (including prediabetes/diabetes, dyslipidemia, hypertension, or elevated hepatic enzymes suggestive of metabolic associated fatty liver disease “MAFLD”). The exclusion criteria included the use of antibiotics, pro/prebiotics, dietary fiber supplements, or any molecules that modifies intestinal transit time within 6 weeks enrolment, pregnancy in progress or planned within 6 months, presence of psychiatric problems and/or use of antipsychotics, following a special diet (e.g., vegetarian, vegan), recent (< 6 weeks) or ongoing diets (e.g., high-protein, high-fiber diet), excessive alcohol consumption (more than 3 glasses/day), type 1 diabetes, and general dislike for vegetables.
Randomization
Following the screening step, subjects were randomly assigned to the prebiotic or placebo arm. (random sequence generated using MS Excel®, simple randomization). Randomization sequences were not revealed to the study staff.
Blinding
All participants and research staff (except dietitians who provided dietary advices and recipes books) were blinded to the treatments. One person was in charge of enrolling and assigning participants to interventions, in each center.
Outcomes
The primary outcome of this trial was the effect of the prebiotic intervention on the gut microbiota composition. Secondary outcomes of this trial concern the effect of prebiotic intervention on anthropometric and clinical parameters but also on the evaluation of physical activity.
Dietary intervention, energy, and nutrient intake
Subjects were asked to consume either16 g/day of native inulin for the prebiotic group (extracted from chicory root, Cosucra, Belgium) or 16 g/day of maltodextrin for the placebo group (Cargill, Belgium), provided in identical packaging, for a period of 3 months. Empty and unused packets were returned to measure compliance. In addition to inulin or maltodextrin intake, participants received a cookbook with recipes based on vegetables either rich (for prebiotic group) or poor (for placebo group) in fructans. Participants were thus advised to consume at least one meal a day with a recipe from this cookbook. Participants received before, and monthly during the protocol, dietary advices from a dietitian in order to reduce their energy intake during the intervention. Energy and nutrient intake were calculated from the one-week recall using the Nubel Pro program (Nubel asbl, Brussels, Belgium). A telephone follow-up was performed three times during the intervention (after 2 weeks, 1 and 2 months).
IPAQ questionnaire
At baseline and at the end of the protocol, 61 participants had a meeting with a unique person by center, in order to fill a long form of International PA Questionnaire (IPAQ). The IPAQ long form asks details about the specific types of activities undertaken within the four main domains (leisure time PA, domestic and gardening (yard) activities, work-related PA and transport-related PA). Computation of the total scores for the long form requires summation of the duration (in minutes) and frequency (days) for all the types of activities in all domains. Results allowed us to classify different levels of PA proposed by the questionnaire (total, low-moderate or high-intensity PAs). Among the 106 patients who completed the study, 61 patients filled the questionnaire for PA before and after intervention (placebo: n = 31, prebiotic: n = 30). After evaluation of PA according to this questionnaire, there were 19 participants who voluntarily increased their PA during the protocol in the maltodextrin group and 14 in the inulin group. 12 participants in the maltodextrin arm and 16 participants in the inulin arm did not change or voluntarily decreased their PA during the intervention.
Mice procedure
Forty-five mice (9-week old C57BL/6 J male) were purchased from Janvier Labs (Le Genest St Isle, France). Mice were individually housed in a controlled environment (12-h daylight cycle) with free access to food and water. The experiment was approved by and performed following the guidelines of the local ethics committee for animal care of the Health Sector of Université catholique de Louvain under the specific agreement number 2017/UCL/MD/005. Housing conditions were as specified by the Belgian Law of 29 May 2013 regarding the protection of laboratory animals (Agreement no LA 1230314). Every effort was made to minimize animal pain, suffering, and distress.
The acclimatization period lasted one week with a standard diet (Research Diet Inc., New Brunswick, NJ, USA). Two independent experiments were performed. Randomization of mice into five groups was done based on body weight to minimize baseline differences. Experimental groups consisted in one group of mice fed with a low-fat diet (LF: n = 10; 10% kcal from fat, D12450, Ssniff, Soest, Germany) and four groups of mice fed with a high-fat diet (HF, 45% kcal from fat, D12451, Ssniff, Soest, Germany). Mice fed with a HF were divided in four additional groups: (1) a group fed a HF only (HF: n = 8); (2) a group fed a HF and receiving 0.3 g/day per mouse of native inulin (Fibruline®, Cosucra, Pecq, Belgium) in the drinking water (INU: n = 9) ; (3) a group fed a HF and receiving a running wheel coupled to an electronic device for measuring the voluntary exercise performed (Ex: n = 9); and (4) a group fed a HF receiving both inulin supplementation and the running wheel (INU + Ex: n = 9). There was no exclusion of mice during the procedure. Running wheel was coupled with an odometer for recording continuously the activity of mice (kilometers and time spent in the wheel). Odometers were tested for reliability of time and distance covered. The cages were visually checked three times a week to ensure that odometers correctly recorded mice activity. To ensure a same enrichment of environment for all mice, the other groups of mice also received a locked wheel in their cage.
At the sixth week, a 24-h feces collection was performed.
At the seventh week, an oral glucose tolerance test (OGTT) was performed on 6 h fasted-mice. Glucose was administered orally (2 mg/g of body weight, 50% glucose solution), and blood glucose levels were determined using a glucose meter (Roche Diagnostics) on blood collected from the tip of the tail vein both before (30 min and 0 min) and after glucose administration (15, 30, 60, 90, 120 min). Twenty microliters of blood were sampled 30 min before and 15 min after the glucose load to assess plasma insulin concentrations.
At the eighth week, running wheels in the two exercised groups were also locked 24 h before the end of the protocol. After 6 h of fasting, mice were anesthetized using isoflurane gas (Abbot, Ottignies, Belgium). Blood from cava vein was harvested in EDTA tubes. Plasma was immediately collected after centrifugation (12 000 × g for 3 min) and stored at − 80 °C for biochemical analysis. Mice were necropsied after cervical dislocation. Liver, brown and white adipose tissues (epididymal, visceral, and subcutaneous), gastrocnemius muscles, cecal content, and intestinal tissues were dissected and immersed in liquid nitrogen before storage at − 80 °C.
Gut microbiota analysis
The raw sequencing data can be accessed in the Sequence Read Archive with the SRA accession number: PRJNA595949 (cohort subset) and PRJNA721281 (mouse subset) [
27].
Cohort subset
Stool samples were available for 59 patients (n = 11 for maltodextrin, n = 19 for maltodextrin + PA; n = 16 for inulin and n = 13 for inulin + PA). Stool samples were collected at baseline and at the end of the 3-month of intervention and stored at room temperature with a DNA stabilizer (Stratec biomolecular, Berlin, Germany) for maximum 3 days, and then transferred to − 80 °C for the analysis of the gut microbiota composition. Genomic DNA was extracted from feces using a PSP® spin stool DNA kit (Stratec biomolecular, Berlin, Germany).
Mouse subset
DNA was extracted from the mouse cecal content using a QIAamp DNA Stool Mini Kit (Qiagen, Hildren, Germany) including a bead-beating step.
For mouse and human subsets, subsequent bioinformatics and biostatistics analyses were performed
in house as previously described [
12].
Initial quality filtering of the reads was performed with the Illumina Software, yielding an average of 96943 pass-filter clusters per sample for the human cohort and 130174 for the mouse cohort. A subset of 25000 reads for the human cohort and 34000 reads for the mouse cohort were randomly selected using Mothur v1.25.0 [
28] to avoid large disparities in the number of sequences. Taxonomic prediction was performed using the nbc_tax function [
29], an implementation of the RDP Naive Bayesian Classifier algorithm [
30]. Taxonomy for significant ASV (amplicon sequence variants) was also confirmed using the EzBioCloud 16S database. Alpha-diversity and beta-diversity indexes were calculated using QIIME [
31]. PcoA plot of the beta-diversity indexes were visualized using the R software. Barplots for phylum was visualized on the GraphPad Prism version 8.0 software.
Statistical analysis
Gut microbiota analysis
For gut microbiota analysis, data are expressed as the mean percentage of relative abundance ± SEM. A Kruskal–Wallis ANOVA test was performed for detecting significant differences for taxa and ASV between groups, followed by Dunn’s multiple comparisons test, using R software (version 3.5.1) and the FSA package.
p-value of the Kruskal-Wallis test was adjusted (
q-value, significant if
q < 0.05) to control for the false discovery rate (FDR) for multiple tests according to the Benjamini and Hochberg procedure [
32]. To avoid analyzing spurious sequences, only ASV with at least one sample showing a relative abundance of more than 0.1% were kept.
In the cohort and mouse experiments, α-diversity indexes were analyzed with a two-way ANOVA, followed by Tukey’s multiple comparison test (significant if p < 0.05). For β-diversity index, a Monte-Carlo rank test was performed.
Association between the variation of genera or ASV significantly regulated in the cohort and significant changes in metabolic outcomes was assessed by Spearman’s correlation tests. A significance level of p < 0.05 was adopted for all analyses.
In the cohort
Clinical outcomes were analyzed using mixed model effects (with PA and inulin supplementation as fixed effects and patients and hospitals as random effects) using the JMP Pro 14 software.
Mice experiment
Variables were analyzed with a two-way ANOVA, performed in the four groups of mice fed a high-fat diet, followed by Tukey’s multiple comparisons test. Within-groups variances were compared using Bartlett’s test. If variances were significantly different between groups, values were normalized by Log transformation before proceeding to the analysis.
Discussion
Previous studies proposed ITFs as interesting prebiotics with beneficial impact on host physiology. We have recently shown that internal and external factors influence the metabolic response to inulin in the context of obesity (intrinsic gut microbiota composition or metformin use) [
12,
15]. Our current findings in obese study participants and in HF diet-fed mice suggest that increased PA enhances health outcomes of a nutritional intervention with inulin in obesity.
Previous studies performed in humans have suggested that the improvement of metabolic alterations and obesity upon inulin intervention in obesity varies between individuals. In our study, we demonstrated for the first time that several clinical improvements in obese individuals was observed mostly when PA was combined to a nutritional intervention with prebiotics supplementation, especially on the management of body weight and glucose homeostasis, but also on liver enzymes and total cholesterol. We also highlighted that the practice of PA during inulin supplementation may improve the gastrointestinal tolerance of high fermentable dietary fibers consumption. In parallel, a preclinical study corroborated the interest of inulin in mice to improve metabolic disorders, but as compared to existing data, it brought new insights into the mechanisms of action of inulin combined with PA.
Voluntary PA was not influenced by prebiotic intervention since we observed a similar change in PA upon the interventions both in obese people and HF-fed mice. This clearly suggests that inulin did not impact the motivation to perform exercise.
In the human cohort, we found a greater improvement of BMI when inulin was combined with PA. Waist/hip ratio and total cholesterol were reduced only in the group combining inulin with PA. This is in keeping with previous findings in rats fed a HF/high-sucrose diet, where the combination of treadmill training and a supplementation with fructans (a mix of inulin and oligofructose) prevented knee joint damage, usually observed in this rat model of obesity [
33]. In this same study, training and prebiotics separately had beneficial effects on serum endotoxin and leptin levels and insulin sensitivity, whereas the combination of both reduced body fat and improved serum lipids. The same team recently showed that combining PA and prebiotic supplementation had no effect on glucose tolerance but improved body mass and LDL cholesterol when the treatment starts 12 weeks after the induction of obesity [
34]. In our study, we observed decreased subcutaneous fat only by voluntary exercise in HF-fed mice. Obesity has multiple consequences, notably on the liver (with the development of metabolic dysfunction-associated fatty liver disease) [
35]. In humans, we further found improved hepatic parameters with the combination of PA and inulin, such as a reduction of both AST and gGT. Interestingly, it has been recently demonstrated that exercise can regulate liver steatosis and stiffness in men with non-alcoholic fatty liver disease, independently of weight loss [
36]. Here, we did not find a beneficial and global effect of PA on hepatic parameters, but we show that combination of PA and inulin is beneficial for the liver. In both humans and mice, we observed improved glucose tolerance and insulin sensitivity. The similar expression of hexose transporters in mouse jejunum suggests that diminished glucose absorption probably does not mediate the improved glucose tolerance.
It has been previously demonstrated that exercise can modulate gut microbiota composition in rodents [
23,
24,
37‐
40]. Six weeks of treadmill running modified gut microbiota in normal and diabetic mice, but changes differed according to the metabolic status of mice [
22]. In our mouse model, the index of Bray-Curtis distance shows that PA did not much alter overall gut microbiota composition, compared to the prebiotic-induced changes. Univariate analyses neither pointed to differences for taxa levels between HF and PA groups. This negligible impact of PA on gut microbiota is consistent with Lamoureux et al. who demonstrated that neither voluntary nor forced exercise contributed to microbiome changes [
41]. The authors discussed that these results contradicting previous reports of significant impact of exercise on gut microbiota in animals and humans [
20,
40,
42,
43] could be due to dietary or housing differences. Our data demonstrate that the impact of PA on gut microbiota depends on the diet as highlighted by the greater effects of PA combined with prebiotic supplementation compared to prebiotics alone. The effects of PA on gut microbiota and metabolic parameters are bigger when the diet contains prebiotic, compared to a HF diet poor in dietary fibers. Thus, PA can change gut transit and fermentation profiles and, therefore, diet plays a crucial role in mediating exercise effects on the gut microbiota. This can explain the lesser impact of PA in mice fed a diet poor in fibers, compared to inulin-supplemented mice. Recently, Liu et al. demonstrated that exercise induces a modest but distinguishable effect on the gut microbiota in metformin-naïve prediabetic individuals [
25]. The response to high-intensity training was heterogenous in prediabetic subjects, associated with changes in insulin sensitivity and glucose metabolism. The microbiome of responders was associated with higher carbohydrate fermentation whereas that from non-responders produced metabolically detrimental compounds. Interestingly, fecal material transfer from responder individuals into antibiotic-pretreated mice mimicked the effect of exercise on glucose homeostasis. This provides support that fermentation features are important for mediating the effect of PA on metabolism.
In our study, we hypothesized that PA modifies gut bacterial fermentation and therefore influences the response to prebiotics. Indeed, we found a higher cecal content in inulin plus PA mice versus inulin-treated mice. Exercise can alter transit time and therefore regulate the substrate availability in the colon. Consistent with this, in HF-fed mice PA increased feces number and weight in the group receiving inulin. Unexpectedly, we observed in the obese cohort that gastrointestinal symptoms caused by inulin intake were reduced with PA. It is well established that gastrointestinal symptoms can appear during acute strenuous exercise [
44]. In our cohort, the low- and moderate-intensity exercise conferred protective effects on the gastrointestinal tract with reduced cramps and bloating. PA may thus be considered in high dose prebiotic dietary interventions.
We examined whether combining PA with inulin changed the fermentation profile in obese individuals. Indeed, the stimulation of
Bifidobacterium growth by inulin was more important when participants increased their PA during the protocol. Curiously, such a higher increase in
Bifidobacterium genus was not found in HF-fed mice. Consistently, it has been previously demonstrated that inulin did not regulate all
Bifidobacterium strains in the same way [
45]. Since human and mice exhibit different strains of
Bifidobacterium, this can explain the different bifidogenic effects induced by exercise combined with inulin supplementation in mice model versus human. Some other bacteria can be differently regulated by the prebiotic, according to PA. For instance, in mice, inulin decreased
Oscillibacter and
Alistipes more so when mice exercised. These bacterial regulations induced by ITF intake have been previously shown in a model of HF diet supplemented with gluten in mice [
46]. However, it is interesting to observe that these ITF effects can be amplified with voluntary exercise. Inulin increased
Prevotella, especially in combination with voluntary exercise. Some studies reported an increased in
Prevotella abundance with dietary fibers in mice [
46,
47]. An elegant study in humans reported that the improvement of glucose metabolism by a dietary fiber (barley kernel) is associated with an induction of
Prevotella [
48]. Consistent with this finding, we found improved glucose tolerance in mice on inulin plus PA in which we detected the higher increase of
Prevotella. Our study highlighted that combining physical activity with a nutritional intervention based on prebiotic supplementation really improved clinical and metabolic parameters observed during obesity. However, the limitation if this exploratory work is that the study is not initially designed for evaluating the impact of PA on prebiotic supplementation since no advices were given to the participants in order to increase their PA. Well-powered studies are thus needed to evaluate the potential beneficial effects of inulin with PA on the metabolic outcomes in participants who follow a specific program of PA during the intervention in a larger cohort. In addition, the cohort included obese subjects with various associated metabolic alterations and diseases (hypertension, dyslipidemia, type 2 diabetes…). Due to number of participants within each group, it was not possible to evaluate and consider the effects of all these variables in this subcohort. Future investigations should focus on the effect of PA associated with prebiotic supplementation in each specific condition to evaluate the impact of combining these two strategies on specific metabolic condition. Finally, another limitation is the fact that we could not predict which bacterial functions and/or metabolites is involved in the metabolic improvements induced by PA combined to inulin supplementation. This can be considered as a perspective of our work to investigate in specific obesity-associated metabolic alterations.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.