SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References

Obesity has recently been linked to the composition of human microbiota and the production of short chain fatty acids (SCFAs). However, these findings rely on experimental studies carried out using rather small and defined groups of volunteers or model animals. Our aim was to evaluate differences within the human intestinal microbiota and fecal SCFA concentration of lean and obese subjects. A total of 98 subjects volunteered to take part in this study. The BMI in kg/m2 of 30 volunteers was within the lean range, 35 were overweight and 33 were obese. The fecal microbiota was characterized by real-time PCR analyses. With the primers used herein we were able to cover 82.3% (interquartile range of 68.3–91.4%) of the total microbiota detectable with a universal primer. In addition, the concentration of SCFA was evaluated. The total amount of SCFA was higher in the obese subject group (P = 0.024) than in the lean subject group. The proportion of individual SCFA changed in favor of propionate in overweight (P = 0.019) and obese subjects (P = 0.028). The most abundant bacterial groups in faeces of lean and obese subjects belonged to the phyla Firmicutes and Bacteroidetes. The ratio of Firmicutes to Bacteroidetes changed in favor of the Bacteroidetes in overweight (P = 0.001) and obese subjects (P = 0.005). Our results are in line with previous reports suggesting that SCFA metabolism might play a considerable role in obesity. However, our results contradict previous reports with regard to the contribution of various bacterial groups to the development of obesity and this issue remains controversial.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References

Obesity is a complex disease which is not completely understood and which is considered by the World Health Organization to be a global epidemic with more than 1 billion adults overweight—one third of these being clinically obese (1). Reasons for obesity can be given in the increased consumption of more energy-dense, nutrient-poor foods containing high levels of sugar and saturated fats in combination with reduced physical activity. However, it has been suggested that several other factors contribute to the development of obesity, such as genetic factors (e.g., Prader–Willi syndrome) (2), underlying illnesses (e.g., hypothyroidism) (3), medications (e.g., atypical antipsychotic) (4,5), high glycaemic diets (6,7), stress (8,9), smoking cessation (10,11), virus infections (12,13,14) and recently also bacteria (15,16).

The microbiota of the human gastrointestinal tract has been studied extensively due to its role both in disease causation and in maintaining gut health. One of the important activities of the large intestinal microbiota is to break down substrates such as resistant starch and dietary fibre, which are not completely hydrolysed by host enzymes in the small intestine (17,18,19). The main fermentation products ensuing from this fibre breakdown are the short chain fatty acids (SCFAs) acetate, propionate, and butyrate (20), which can be utilised for lipid or glucose de novo synthesis (21). The bacterial SCFAs thus provide an additional source of energy for the body.

Recently, it has been hypothesised that an increased ratio of Firmicutes to Bacteroidetes may make a significant contribution to the pathophysiology of obesity. By comparison with wild-type animals Ley and co-workers demonstrated that genetically obese (ob/ob) mice show an increased proportion of the Firmicutes and a decrease in Bacteroidetes, (22). A further report indicated that the proportion of Bacteroidetes 16S rRNA sequences was diminished in faeces from 12 obese human subjects (23). Furthermore, Gill and co-workers found that the human gut microbiome of two healthy subjects is enriched with many clusters of orthologous groups representing key genes in the pathway of methanogens (24). This finding led to the hypothesis that Methanobrevibacter smithii may be a therapeutic target for the reduction of energy harvest in obese humans (25,26), as M. smithii is the major representative of the human gut methanogens (27).

The purpose of this study was to investigate whether the proposed role of SCFA and microbiota composition in obesity, which was based on proof of principle experiments, can be confirmed in a larger study which did not exclude all confounding factors.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References

Volunteer recruitment

Lean and obese volunteers of both sexes were recruited from the Institute of Microecology and from the obesity consultation hours at the University of Giessen and Marburg. In total 98 volunteers (34 males and 64 females) took part in the study. All of the samples collected were analysed. The volunteers were aged 47 ± 13 year (mean ± s.e.m.: range 14–74 year). The BMI in kg/m2 of 30 volunteers was within the normal range (18.5–24.9), while 35 were overweight (25.0–29.9) and 33 were obese (≥30.0). Of the latter, 17 were classified as obesity class 1 (30–35), 11 as class 2 (35–40), and 5 as class 3 (>40). No antibiotics had been taken in the 6 months prior to the study. All participants subsisted primarily on a western diet and all volunteers provided informed, signed consent.

Collection and preparation of stool samples for analysis

From the fresh stool sample provided from each volunteer DNA (200 mg) was extracted using the Easy Mag DNA Isolation system (BioMerieux, Nuertingen, Germany) according to the manufacturer's instructions. The remaining sample was stored at −20 °C for metabolic analysis.

Determination of SCFA and fecal gross energy content

Human stool samples to be analysed for SCFA were freeze-dried and subsequently analysed using gas chromatograph as previously described (28). Briefly, the sample was weighed (∼80 mg dry matter) and an extraction solution (1 ml) containing oxalic acid (0.1 mol/l), sodium azide (40 mmol/l), and an internal standard (caproic acid 0.1 mmol/l) was added. The solution was extracted for 60 min on a horizontal shaker and then centrifuged (10 min at 16000 × g). Concentrations of the SCFA were determined in the supernatant using an Agilent 6890N gas chromatograph with flame ionization detection equipped with a capillary column Innowax 30 m × 530 µm ×.0.1 µm (Agilent). ChemStation software was used for data processing. Bomb calorimetry was performed on fecal samples of all lean, overweight and obese volunteers. Aliquots of fecal samples were freeze-dried for 48 hours and the gross energy content measured using an adiabatic bomb calorimeter, calorimetric cooling system and oxygen station (models C7000, C7002, and C7048, respectively, from IKA-Analysentechnik, Heitersheim, Germany). The calorimeter energy equivalent factor was determined using benzoic acid standards.

Primer selection for quantitative PCR (qPCR)

Primers were selected to recognise similar bacterial groups as defined by previously released 16S rRNA targeting probes used for fluorescence in-situ hybridisation analysis. A particular fluorescence in-situ hybridisation-defined group containing the appropriate target sequence in the ARB program (31) was selected and subsequently qPCR primers were designed with the ARB program to amplify the same group of commensal bacteria (29). Discriminating nucleotides were chosen to be at the 3'end of the primer and a specific primer was combined with a universal primer that does not exclude any members belonging to that particular group. qPCR primers for bacterial groups were stringently selected using the Primer Designer programme to avoid primer-dimer formation and yield products of 100–300 base-pairs (Table 1). The standard line was based on actual counting of cultured bacteria and correlated directly to the Ct values of the qPCR. We validated the qPCR data to actual bacterial counts as described by Barman and colleagues (30). Specificity of the different qPCR primer sets was tested elsewhere (31,32,33,34,35,36,37,38). Several of the primers mentioned have already been used in at least one other published study (39). Enumeration of predominant bacteria in the fecal samples by qPCR

Table 1. 16S rRNA gene-targeted group-specific primers used in this study
inline image

Quantitative PCR amplification and detection were carried out using the primers described in Table 1. PCR amplification and detection was performed using an ABI PRISM 7900HT sequence detection system (Applied Biosystems, Darmstadt, Germany) in optical-grade 96-well plates sealed with optical sealing tape. Each reaction mixture (25 µl) was composed of 12.5 µl of QuantiTect SYBR Green PCR Master Mix (Qiagen, Hilden, Germany), 2 µl primer mix (10 pmol/µl each), 9 µl sterile distilled H2O, and 1.5 µl stool DNA (10 ng/µl). For the negative control, 2 µl of sterile distilled H2O was added to the reaction solution instead of the template DNA solution. A standard curve was produced using the appropriate reference organism to quantify the qPCR values into number of bacteria/g. The standard curves were prepared in the same PCR assay as for the samples. The fluorescent products were detected in the last step of each cycle. A melting curve analysis was carried out following amplification to distinguish the targeted PCR product from the nontargeted PCR product. The melting curves were obtained by slow heating at temperatures from 55 to 95 °C at a rate of 0.2 °C/s, with continuous fluorescence collection. The data was analysed using the ABI Prism software.

The real-time PCRs were performed in triplicate, and average values were used for enumeration. PCR conditions were optimized based on those described in the literature (31,32,33,34,35,36,37,38). The amplification programme consisted of one cycle of 95 °C for 15 min and then 40 cycles of 95 °C for 30 s, 60 °C for 30 s, and 72 °C for 60 s. This programme was used for all qPCR-primers with the exception of the qPCR-primers for the detection of the genus Methanobrevibacter. Here, the annealing temperature was 58 °C for 30 s (38).

In all assays the amplification efficiency was higher than 90%, and the standard curve showed a linear range across at least 5 logs of DNA concentrations with a correlation coefficient >0.99. The lowest detection limits of all assays were as low as 10–100 copies of specific bacterial 16S rDNA per reaction, corresponding to 104—105 copies per gram of wet-weight faeces. The data was analysed using the ABI Prism software. Table 1 shows the primer sets for the targeted bacterial groups.

Statistical analyses

All statistical analyses were performed using SPSS (SPSS, Chicago, IL). The normality of the data was checked using the nonparametric Kolmogorov–Smirnov test with Lilliefors correction. Depending on the normality of the underlying data, the ANOVA test or the Mann-Whitney-U-test was used to perform statistical analyses. The χ2-test statistic was used to analyze the proportion of volunteers harbouring methanogens. Correlations between the parameters measured were appraised by calculation of nonparametric Spearman correlation coefficients and subsequent visual inspection of scatter plots. Additionally, only the significant Spearman correlations that were also significant in a partial correlation analysis with “age” and “gender” as control variables were considered. Test results of all analyses were considered significant at an alpha of P < 0.05.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References

Changes in SCFA concentrations and fecal energy content

The major SCFA found in the stool samples were acetate, propionate, butyrate, and valerate as well as iso-valerate and iso-butyrate (Table 2). The differences in the stool SCFA concentrations between the lean, overweight and obese subjects were considerable. The mean total SCFA concentration in fecal samples of obese volunteers was by more than 20% higher in total than of lean volunteers (P = 0.024). The highest increase was seen for propionate with 41% (P = 0.002), followed by butyrate (28%, P = 0.095). A numerical increase of valerate (21%), and acetate (18%) concentrations was observed. In addition, this resulted in changes in the proportions of individual to total SCFA. The propionate proportion was thus higher in overweight (18.7%, P = 0.019) and obese (18.3%, P = 0.028) volunteers than in lean volunteers (15.9%). No considerable increase or change in the proportions of the iso-SCFA was observed.

Table 2. Mean SCFA (mmol/l) concentration and gross energy content (kJ/g) in dry feces of volunteers with different BMI
inline image

To estimate the efficiency of nutrient utilisation, gross energy content in fecal samples was determined. The mean content in fecal samples of lean, overweight and obese volunteers was noted as 21.7, 21.6, and 21.6 kJ/g dry matter, respectively, and was just numerically different between the BMI classes studied.

Quantification of the predominant bacterial groups in the stool

qPCR analyses were performed to quantify individual bacterial groups in stool samples collected from the study group. The most abundant bacterial groups in lean and obese subjects (Table 3) were the gram-positive bacteria belonging to the Clostridium leptum group and the Clostridium coccoides group as well as the gram-negative Bacteroides spp., which altogether made up a median proportion of 96.3% (interquartile range of 89–99.2%) of all detected bacteria. Species of the genera Veillonella, Bifidobacterium, Prevotella, Lactobacillus, Enterococcus, Methanobrevi-bacter, and the Eubacterium cylindroides group accounted for the remaining minor proportion of all bacteria. Some of these microbial groups were present in slightly different concentrations in faeces of lean, overweight and obese volunteers. Overweight volunteers harboured significantly higher fecal concentrations of the genus Bacteroides than lean (P = 0.002) but not than obese volunteers (P = 0.145). On the other hand, both overweight (P = 0.006) and obese volunteers (P = 0.011) exhibited lower cell numbers of the Ruminococcus flavefaciens subgroup. In addition, in obese volunteers numerically or even significantly lower fecal concentrations of the C. leptum group (P = 0.07) and of the genus Bifidobacterium (P = 0.02), respectively, were noted. The C. leptum group, which is made up to an extensive part by the Ruminococcus flavefaciens subgroup, belongs to the bacterial division of Firmicutes, which therefore accounted for a significantly smaller proportion of the total sum of all micro-organisms studied in overweight (P = 0.001) and obese volunteers (P = 0.002) compared with lean volunteers. In turn, the median proportion of the bacterial division Bacteroidetes of the total sum of species studied was higher in overweight (46.8%, P = 0.001) and obese (45.0%, P = 0.006), respectively, than in lean volunteers (22.9%), since the Bacteroidetes count remained largely unaffected by BMI. On the whole, this led to a lower ratio of Firmicutes to Bacteroidetes in overweight (P = 0.001) and obese volunteers (P = 0.005). Furthermore, species of the genus Methanobrevibacter were detected in a smaller proportion of overweight (45.7%, P = 0.155) and obese volunteers (33.3%, P = 0.017) compared to lean volunteers (63.3%). Simultaneous, obese carriers of methanogens exhibited a lower cell count of Methanobrevibacter spp. than lean carriers (P = 0.018).

Table 3. Fecal microbiota composition of lean (n = 30), overweight (n = 35), and obese subjects (n = 33)
inline image

An analysis of correlations between the parameters measured and the BMI supports the results obtained from the comparison of the different BMI groups as significant correlations between the BMI and propionate (P = 0.003), the propionate proportion of SCFA (P = 0.005), and the concentration of Bifidobacterium (P = 0.003) and Methanobrevibacter (P < 0.005) were detectable which even remain significant following correction to take into consideration the influence of age and gender using partial correlation analysis. Within the subgroup of obese participants no additional dependencies between the BMI and other parameters could be detected.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References

In recent years the importance of the gut microbiota to health has been widely acknowledged. Recent reports of a possible correlation between the human gut microbiota and obesity has placed the focus on new aspects of the significance of the human microbiota to wellbeing (16,22,26,40). It has been assumed that the proportional representation of the Firmicutes and Bacteroidetes may play a role in the development of obesity (22,40). Members of both groups produce SCFA from dietary compounds which escape digestion in the small intestine, thus supplying the host with an additional amount of energy. Normal colonic epithelia derive 60–70% of their energy supply from SCFA, particularly butyrate (41). Propionate is largely taken up by the liver and is a good precursor for gluconeogenesis, liponeogenesis and protein synthesis (42,43). Acetate enters the peripheral circulation to be metabolised by peripheral tissues and is a substrate for cholesterol synthesis (21). Within our study group total SCFA content increased significantly from 84.60 mmol/l in lean to 103.87 mmol/l in obese subjects. Additionally, the propionate concentration increased significantly from lean to obese subjects as well as its proportion of total SCFA.

High fecal concentrations of total or individual SCFA might also be the result of increased microbial production, shifts in microbial cross feeding patterns, low mucosal absorption or even the rate of transit alone. Nevertheless, it is known that changes in concentration and proportion of individual SCFA are concurrent with changes in bacterial groups (44,45,46). In order to examine the importance of the microbial composition on these effects, a panel of quantitative PCR (qPCR) primers for major groups of the microbiota, which account for up to 90% of the total cells, was employed (32,44). Known propionate producers belong to the genera Bacteroides and Prevotella, whose joint cell numbers were more numerous in overweight (P = 0.001) volunteers. It is worthy of mention that the proportion of Bacteroidetes significantly increased in overweight and obese subjects.

Our results are in accordance with two reports by Duncan and co-workers, who showed that despite weight loss there was no change in the relative counts of the Bacteroides spp. or the percentage of Firmicutes (44,47), thus assuming that not the ratio of Firmicutes and Bacteroidetes is important but rather the amount of SCFA produced. When carbohydrate intake was lowered in their study, the acetate proportion increased, butyrate decreased and propionate remained unaffected (order unchanged) (44). Subsequently, leaner people had a higher ratio of acetate to butyrate and propionate. The respective ratio was also higher—though only slightly—in lean volunteers of our study. The increase of the proportion of Bacteroides spp. which are the major representatives of the Bacteroidetes, from lean to obese subjects is in concordance with reports by Collado and co-workers (48) but in contrast to earlier reports which linked an increased ratio of Firmicutes to Bacteroidetes and high amounts of SCFA to obesity (16,22,23). As previously reported by others, we also measured higher SCFA concentrations. On the other hand, we were unable to identify any correlation of obesity with higher proportions of Firmicutes or higher concentrations of methanogenic Archaea in obese volunteers. Both observations have been made in proof of principle experiments and are proposed to be one driving factor of obesity (16,22,23,40). It seems that under field conditions with unrestricted human volunteers as in our study, other lifestyle-related factors, such as intensity and regularity of exercise as well as total daily energy intake, might be much more important in causing and maintaining obesity. These factors, as possible confounding variables, were excluded in previously reported well-defined experiments using in part even genetically-modified animals or strictly-controlled human studies. In our opinion, this could be the most probable explanation for the results obtained in our study which are principally different and in contradiction to those obtained in the afore-mentioned studies, although methodological differences in DNA extraction protocols as well as primer design may have caused additional variation. Therefore, owing to fundamental differences in “proof of principle experiments” and field studies and the contrasting results reported here, it seems premature to suggest a certain type of gut microbiome to be a biomarker for obesity in humans or to identify methanogens as a potential therapeutic target as done in the past (16). Many more clinical studies are needed here to establish a consistent theory on the extent of the influence of intestinal microbiota on obesity. Interestingly, the reduction of Bacteroidetes and bifidobacteria has an influence on obesity (49).

We see it rather difficult to draw definite conclusions on the importance of various bacterial groups in obesity, as not all influencing parameters such as diet, genetic background, habitation, and overall fitness were taken into account.

Acknowledgments

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References

We are grateful to Matthias Filter for support in statistical matters, to Jeffrey Gordon for the critical reading of the manuscript and to Lorraine Herfort for syntax revision.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Disclosure
  8. Acknowledgments
  9. References