Sunteți pe pagina 1din 12

Journal of Functional Foods 16 (2015) 211–222

Available online at www.sciencedirect.com

ScienceDirect

j o u r n a l h o m e p a g e : w w w. e l s e v i e r. c o m / l o c a t e / j ff

Addition of acacia gum to a FOS/inulin blend


improves its fermentation profile in the Simulator of
the Human Intestinal Microbial Ecosystem (SHIME®)

Massimo Marzorati a,*, Bingcai Qin b,1, Falk Hildebrand c,d,1, Abby Klosterbuer e,
Zamzam Roughead e, Claudia Roessle f, Florence Rochat f, Jeroen Raes c,d,g,h,
Sam Possemiers a,i
a
Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Ghent, Belgium
b
BGI-Shenzhen, Shenzhen, China
c
VIB Department of Structural Biology, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium
d
Department of Bioscience Engineering, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium
e
Nestlé R&D Center Minneapolis, 12500 Whitewater Drive, Minnetonka, MN, USA
f
Nestlé Research Centre, 1000 Lausanne, Switzerland
g
Department of Microbiology and Immunology, REGA Institute, KU Leuven, Herestraat 49, 3000 Leuven, Belgium
h
VIB Center for the Biology of Disease, KU Leuven, Herestraat 49, 3000 Leuven, Belgium
i
ProDigest BVBA, Technologiepark 3, 9052 Ghent, Belgium

A R T I C L E I N F O A B S T R A C T

Article history: The Simulator of the Human Intestinal Microbial Ecosystem (SHIME®) was used to assess the
Received 11 November 2014 impact of the partial substitution of fructooligosaccharides (FOS) and inulin by acacia gum in
Received in revised form 21 April a fibre blend. Both blends were well fermented and the presence of acacia gum modified the
2015 intestinal fermentation of the blend from a boosted fermentation into a gradual one in the
Accepted 22 April 2015 complete colon. Confirmation of the gradual fermentation was obtained by analysis of the pH
Available online 15 May 2015 profile, measurement of the residual acacia gum fractions in the colon and by the increased
expression of specific catabolic enzymes. Both blends increased the total amount of short chain
Keywords: fatty acids (SCFAs; +42 mmol/L), propionate (+26 mmol/L) and butyrate (+9 mmol/L) and showed
Colonic fermentation bifidogenic properties. Metagenomic Illumina sequencing confirmed that the blends exerted
Gastrointestinal tract a diverse modulating activity in the different areas of the colon. Long-term repeated admin-
SHIME® istration of the prebiotic blends is needed to reach gradual changes in the gut microbial
Acacia gum community structure and activity. The partial substitution of FOS and inulin by acacia gum
Prebiotic fibre slowed the speed of the fermentation of the blend. In humans, this may translate into ben-
efits for gut health, allowing for incorporation of higher amounts of fibre into the diet.
© 2015 Elsevier Ltd. All rights reserved.

Chemical compounds: (+)-Arabinogalactan (PubChem CID: 24847856); Inulin (PubChem CID: 16219508).
* Corresponding author. Laboratory of Microbial Ecology and Technology (LabMET), Ghent University, Coupure Links 653, 9000 Ghent, Belgium.
Tel.: +32 9 2645976; fax: +32 9 2646248.
E-mail address: massimo.marzorati@ugent.be (M. Marzorati).
1
Authors equally contributed.
Abbreviations: SHIME , Simulator of the Human Intestinal Microbial Ecosystem; FOS , fructooligosaccharides; SCFA, short chain fatty
acid; BCFA, branched chain fatty acid; AC, TC, DC, ascending, transverse and descending colon; GIT, gastrointestinal tract; MW, molecu-
lar weight; SEC, size exclusion chromatography; AG, acacia gum; qPCR, quantitative polymerase chain reaction; KO, KEGG orthologue
group; OG, orthologue group
http://dx.doi.org/10.1016/j.jff.2015.04.039
1756-4646/© 2015 Elsevier Ltd. All rights reserved.
212 Journal of Functional Foods 16 (2015) 211–222

alter the fermentation profile of the blend. In particular, ad-


1. Introduction dition of slowly fermented AG was expected to extend the
duration of fermentation relative to the other blend, and this
The gastrointestinal tract (GIT) hosts the most complex mi- would be reflected by a different modulation of the microbiota
crobial community in the human body (Maccaferri, Biagi, & in the different areas of the colon (i.e. ascending, transverse
Brigidi, 2011). Biomedical research has shown that the balance and descending colon) without impacting the potential pre-
of this additional organ within our body is of key importance biotic activity of the blend. In order to prove this, we analysed
to maintain a healthy status and that several diseases are fre- the changes in microbiota activity and composition in each simu-
quently correlated with a dysbiosis (Blottiere, de Vos, Ehrlich, lated colon compartment during a three-week treatment period
& Dore, 2013). In this respect, there is a great interest in de- with the 2 blends by means of metabolic analyses (i.e. short
veloping new products (e.g. prebiotic fibres) with the aim of chain fatty acids – SCFA, ammonium, pH profiling, enzymatic
modulating the gut microbiota in order to induce a positive activities, size exclusion chromatography) and molecular
health effect (e.g. increased or decreased production of health techniques (i.e. quantitative PCR, denaturing gradient gel elec-
related bacterial metabolites, growth of health promoting bac- trophoresis, in depth metagenomic illumina sequencing).
teria, decrease in intestinal pathogens, or immune modulation)
(Al-Sheraji et al., 2013) [www.efsa.europa.eu/en/supporting/
pub/136e.htm]. This is especially important in the clinical 2. Materials and methods
setting, where patients may have altered gut microbiota due
to stress, dietary changes, or medication. Prebiotics are com- 2.1. Products
monly added to products for supplemental or complete enteral
The nutritional products used in this study were composed of
nutrition, and identifying fibres or fibre blends that exert health
30% fat, 20% proteins and 50% carbohydrates, including a blend
benefits without compromising gastrointestinal tolerance is of
of fermentable fibres. These nutritional products only dif-
great interest for this population. In order to bring a new product
fered in the composition of the fibre blend. Blend 1 (Nestlé
with a specific health claim on the market, evidence of its effect
Health Science, Vevey, Switzerland) contained FOS and inulin
must be shown in vivo and, when possible, the mechanism of
(blended in proportion of 70:30%) while in Blend 2 (Nestlé Health
action should be explained [http://www.efsa.europa.eu/en/nda/
Science) part of the fructans present in Blend 1 were re-
ndaguidelines.htm].
placed by AG (41% FOS, 41% acacia gum, 18% inulin). The
Most of the in vivo studies on the gut microbiota are based
nutritional products were available in servings containing 3.3 g
on the analysis of faecal samples due to the fact that they can
of the fibre blends.
be easily collected with a non-invasive approach. However, while
faecal communities may be representative for processes oc-
2.2. Simulator of the human intestinal microbial
curring in the distal part of the colon, they represent, at the
ecosystem (SHIME®)
very best, an average of the microbiota composition and
functionalities (both luminal and mucosal) of the whole GIT The reactor setup was adapted from the SHIME® (ProDigest and
(Durban et al., 2011; Lee et al., 2013). In this respect, elucida- Ghent University, Ghent, Belgium), representing the GIT of the
tion of the mechanism of action of a fibre or a mix of fibres adult human, as described by Molly, Woestyne, and Verstraete
in the different areas of the GIT may become a very difficult (1993). The SHIME consists of a succession of five reactors simu-
task, making use of clinical trials with ileostomy volunteers lating stomach, small intestine, ascending, transverse and
or animal models with humanized microbiota. descending colon. The system has been inoculated with the
Well-designed in vitro technologies – aimed at simulating faecal microbiota from a healthy 75-year-old volunteer who had
the main physiological and microbiological processes occur- no history of antibiotic treatment in the 6 months prior to the
ring in the GIT – coupled with metabolic and molecular analyses faecal sample collection. Specific conditions (e.g. retention time,
can be an interesting toolbox to complement in vivo trials and pH…) and feed composition have been previously described
produce data supporting the mechanism of action of a spe- in Possemiers, Verthe, Uyttendaele, and Verstraete (2004). For
cific product. these experiments, a TWINSHIME® setup was used by oper-
The Simulator of the Human Intestinal Microbial Ecosys- ating two systems in parallel at the same time in order to obtain
tem (SHIME®) is a continuous model of the GIT that allows in- identical environmental conditions for both systems (SHIME
depth longitudinal studies of the biological activity of selected 1 = Blend 1; SHIME 2 = Blend 2). The experimental setup of the
molecules in the gut under representative environmental con- SHIME run included a two-week start-up period, a two-week
ditions and under long-term repeated administration conditions control period and a three-week treatment period, as previ-
(Marzorati et al., 2013; Possemiers, Marzorati, Verstraete, & Van ously described (Van de Wiele et al., 2004). During the treatment
de Wiele, 2010; Sanchez et al., 2009; Van de Wiele, Boon, period, 3.0 g/L of the starch in the feed was removed from the
Possemiers, Jacobs, & Verstraete, 2004). This model was used diet – to partially compensate for the additional amount of
to describe the behavior in the GIT of two mixtures of prebi- carbon sources – and replaced with 2 servings of the fibre blends
otic fibres: a blend containing fructo-oligosaccharides (FOS) and per day.
inulin (short chain, linear molecules) and a blend in which part
of the fructans present in the first blend were replaced by acacia 2.3. Analytical techniques
gum (AG), a complex fibre shown in previous studies to be slowly
fermented (Cherbut, Michel, & Lecannu, 2003). The hypothesis Samples for analytical techniques (a total of 16 mL) were col-
was that partial substitution of FOS and inulin with AG would lected three times per experimental week (i.e. Monday,
Journal of Functional Foods 16 (2015) 211–222 213

Wednesday and Friday at 1 pm), divided in aliquots of 2 mL the contigs longer than 500 bp were used to predict ORFs by
and stored at −20 °C. By collecting 3 samples every experimen- MetaGene program (Noguchi, Park, & Takagi, 2006). Mean-
tal week we basically have 3 observations in what we consider while, high quality reads from each sample were aligned against
as an experimental unit (i.e. 1 week). Analyses for SCFA, am- the Human Gut Gene Catalogue (Li et al., 2009; Qin et al., 2010)
monium, D-lactic acid and L-lactic acid quantification, effect by SOAP2 (Li et al., 2009) using the criterion of identity ≥90%.
of the treatment on pH were conducted as described by Terpend We translated the aligned results to gene relative abundance
and colleagues (Terpend, Possemiers, Daguet, & Marzorati, 2013). by counting the number of reads that mapped to the gene.
The molecular weight (MW) profiles of polysaccharides in the Based upon Human Gut Gene Catalogue’s annotation infor-
residual content were analysed by size exclusion chromatog- mation, genes were assigned to different taxonomy levels, KEGG
raphy (SEC), as previously described (Terpend et al., 2013). orthologue group (KO), KEGG pathway map (map) and eggNOG
Activity for β-glucosidase, α-galactosidase, β-galactosidase, β-D- orthologue group (OG). For the species profile, we utilized
xylosidase and α-L-arabinofuranosidase was measured by Human Gut Gene Catalogue’s annotation information and
means of colorimetric assays as described by Berg, Nord, and summed the relative abundance of genes from the same
Wadstrom (1978). Gas production has been measured in peni- species. This gross relative abundance was taken as the content
cillin flasks (in triplicate) during 48 h of simulated colonic of this species in a sample to generate the species profile of
incubation, by means of a WIKA digital pressure meter (CPH6200 the samples. The genus profile, phylum profile, KO profile, map
– S1) (LHM instrumentation, Geel, Belgium). Briefly: carbon de- profile and OG profile were constructed using the same
pleted medium (composition: Pepton 2 g/L; yeast extract 2 g/L; methods. Besides, all the redundant genes were removed from
L-cystein 0.5 g/L; NaCl 0.1 g/L; K2HPO4 40 mg/L; KH2PO4 40 mg/L; all the 12 samples, leading to a gene-set of 201,356 genes.
MgSO4 7H2O 10 mg/L; CaCl2 2H2O 6.7 mg/L) has been inocu- Taxonomic assignment of predicted genes was carried out
lated 10% with SHIME fluid collected from the ascending colon using BLASTN to assign reads to a reference genome data-
during the second week of control. Either Blend 1 or Blend 2 base at a cut-off of 95% sequence identity and >100 bp overlap.
have been added as a sole carbon source for bacteria at the This assignment was used as high confidence assignment on
final concentration of 6 g/L. Gas production has been mea- species level. As reference database we used all available com-
sured after 2, 4, 6, 24 and 48 h of incubation. plete bacterial genomes from IMG version 3.50. The genes
annotated by KEGG were assigned into KEGG modules and Seed
2.4. Molecular microbiological analyses pathways (Overbeek et al., 2005). The relative pathway abun-
dance of higher order functional categories was calculated from
Samples for molecular techniques (2 aliquots of 1 mL each) were normalized KO abundances.
collected once per experimental week at the end of the week
(i.e. Friday at 1 pm), for a total of five weeks from each simu- 2.6. Statistics
lated colon compartment. The samples were immediately
centrifuged and the pellet stored at −20 °C. Metagenomic DNA Before investigating the probability of intergroup differences,
was extracted using the CTAB protocol as described by Boon, the normality data were studied with Kolmogorov–Smirnov.
Top, Verstraete, and Siciliano (2003). Comparison of normally distributed data was performed with
Quantitative polymerase chain reaction (qPCR) for total bac- Student’s t-test for pairwise comparisons or with Wilcoxon
teria, bifidobacteria, and lactobacilli were performed as reported signed-rank test for non-normally distributed data. Statisti-
by Possemiers et al. (2006). The qPCR for the Firmicutes and cal significance was set at P < 0.05. Calculations were performed
Bacteroidetes phyla was previously described by Guo et al. using the SPSS software (version 21.0, SPSS Inc., Chicago, IL,
(2008). USA). Comparison of the effect of the Blends on the meta-
Denaturing gradient gel electrophoresis (DGGE) was used bolic parameters and on qPCR data was conducted by means
to separate the PCR products for the 16S rRNA genes of total of a longitudinal statistical analysis with a linear spline model,
bacteria, bifidobacteria and lactobacilli as described by Terpend positioning the knot on the first week of treatment (SAS, Cary,
and colleagues (Terpend et al., 2013) Finally, samples from the NC, USA). The approach is based on the creation of a complex
end of the control period (week 2) and the end of the treat- model and the subsequent removal, step by step, of different
ment period (week 3) were selected for metagenomic illumina predictors. The difference of the maximum likelihood values
sequencing. of two equations compared with the respective chi-square
allows determination whether the removed predictor had a sta-
2.5. Metagenomic illumina sequencing tistical significance or not.
The obtained DGGE patterns were analysed using the
DNA library construction was performed following the manu- BioNumerics software v.2.0 (Applied Maths, Sint-Martens-
facturer’s instruction (illumina). We constructed one paired- Latem, Belgium). A matrix of similarities for the densiometric
end (PE) library with insert size of 350 bp for each sample. Then curves of the band patterns was calculated based on the Pearson
high-throughput sequencing was performed using illumina product–moment correlation coefficient. A composite dataset
Hiseq 2000 to obtain around 240 million PE reads for all 12 was created by merging the information from the all the band
samples. The reads length for each end was 90 bp. The raw reads patterns in order to obtain a combined dendrogram – using
which contain three or more “N” or adapter contamination were UPGMA linkage – containing the information from the gels on
discarded. total bacteria, bifidobacteria and lactobacilli.
For each sample, high quality reads were assembled to obtain For illumina data, each sample was normalized by divid-
long contigs by using SOAP denovo assembler (Li et al., 2010); ing each feature (species, KO abundance, etc.) within a sample
214 Journal of Functional Foods 16 (2015) 211–222

by the samples’ respective total sum of reads. Higher func- the “coin” R-package (Hothorn, Hornik, Van de Wiel, & Zeileis,
tional categories were calculated from a normalized KO 2006). The respective test was then used to calculate blocked
abundance matrix, but not further normalized, as described post-hoc significances, similar to the above procedure. This pro-
in Hildebrand et al. (2012). Feature abundance matrices were cedure is described in more detail in Hildebrand et al. (2013).
transformed by adding 1 to each feature and calculating log10 Non-metrical dimensional scaling (NMDS) was used to vi-
subsequently, avoiding negative infinite values for absent fea- sualize inter-sample hellinger distances, as implemented in the
tures. All statistics were calculated using R 2.14. Functional “vegan” 2.0.7 package (http://cran.r-project.org/web/packages/
abundance categories were tested for significant differences vegan/index.html). The phylogenetic clustering was calcu-
using the non-parametric Kruskal–Wallis test (p-value) that was lated from inter-sample Bray–Curtis distances from the
subsequently corrected for Multiple Testing using the transformed data matrix. These distances were clustered using
Benjamini–Hochberg (Benjamini & Hochberg, 1995) false dis- centroid hierarchical clustering as implemented in the
covery rate (q-value). Functional categories with less than 10 R-package “cluster” (http://cran.r-project.org/web/packages/
reads over all samples were excluded from this analysis to avoid cluster/index.html).
artifacts. Post-hoc statistical testing for significant differ-
ences between all combinations of tested groupings was
conducted only if the Kruskal–Wallis test p-value was <0.2. 3. Results
Wilcoxon rank-sum tests were calculated for all possible com-
binations and corrected for multiple testing using Benjamini– 3.1. Metabolic parameters
Hochberg false discovery rate (q-value). If not mentioned
otherwise, tests were considered significant if they had a SCFA profiles consisted mainly of acetate, propionate and bu-
p-value ≤0.05 and a q-value ≤0.1. The significance of blocked tyrate with small amounts of other acids such as isobutyric,
test designs was calculated using the “independence_test” of valeric, isovaleric and caproic acid. As shown in Table 1, both

Table 1 – Comparison of the effect of Blend 1 and Blend 2 on the SCFA (mmol/L), lactate (mmol/L), BCFA (mmol/L) and
ammonium (mg/L) production in the different colon compartments of the TWINSHIME (ascending – AC, transverse – TC
and descending – DC colon). The data are presented per the full control period (observations per 2 weeks = 6) and per
experiment week of treatment (observations per week = 3). Data are shown as average of the observations ± standard
deviation. Comparison of the effect of the Blends was conducted by means of a longitudinal statistical analysis with a
linear spline model, positioning the knot on the first week of treatment. Significant differences are indicated with
* (P < 0.05) close to the specific parameter.
Control Treatment – week 1 Treatment – week 2 Treatment – week 3
Blend 1 Blend 2 Blend 1 Blend 2 Blend 1 Blend 2 Blend 1 Blend 2
AC
Acetate 27.4 ± 2.0 30.3 ± 0.9 37.4 ± 2.7 34.2 ± 2.4 33.3 ± 4.5 34.8 ± 2.2 33.8 ± 2.5 35.2 ± 4.3
Propionate 18.9 ± 1.6 19.2 ± 2.1 39.1 ± 4.1 43.5 ± 10.3 45.3 ± 2.3 49.3 ± 2.4 48.1 ± 7.7 48.3 ± 6.4
Butyrate* 9.9 ± 0.1 9.8 ± 0.1 15.6 ± 3.4 12.4 ± 1.2 19.0 ± 7.2 13.7 ± 1.5 23.7 ± 5.2 13.7 ± 3.7
Valerate 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
Caproate 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
BCFA 2.9 ± 0.1 2.9 ± 0.0 1.1 ± 0.5 1.5 ± 0.6 1.4 ± 0.1 1.2 ± 0.2 1.4 ± 0.2 1.3 ± 0.3
Total 59.0 ± 3.6 62.2 ± 3.1 93.3 ± 5.5 91.7 ± 11.8 99.2 ± 5.5 98.9 ± 5.6 107.3 ± 15.7 98.6 ± 14.6
Lactate* 2.4 ± 0.5 2.5 ± 0.2 9.5 ± 4.8 3.5 ± 2.1 8.0 ± 6.5 3.5 ± 1.3 7.5 ± 6.1 7.2 ± 6.5
Ammonium 352.3 ± 19.5 358.6 ± 39.7 173.1 ± 42.0 217.8 ± 51.7 168.9 ± 18.2 183.2 ± 21.5 194.1 ± 28.1 153.4 ± 34.2
TC
Acetate 31.9 ± 1.2 32.1 ± 0 35.0 ± 1.2 36.0 ± 4.5 38.1 ± 1.5 42.1 ± 2.8 37.7 ± 0.9 41.3 ± 3.0
Propionate 20.9 ± 0.6 19.9 ± 0.9 34.8 ± 11.6 40.1 ± 17.2 50.7 ± 0.8 56.9 ± 1.4 48.3 ± 2.2 51.0 ± 2.5
Butyrate* 9.7 ± 0.4 9.9 ± 0.3 19.0 ± 10.2 14.4 ± 4.4 29.2 ± 3.3 18.3 ± 2.0 25.0 ± 1.7 15.3 ± 2.3
Valerate 0.0 ± 0.0 0.0 ± 0.0 0.3 ± 0.3 0.1 ± 0.1 0.8 ± 0.3 0.2 ± 0.0 0.6 ± 0.1 0.2 ± 0.1
Caproate 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0 0.0 ± 0.0
BCFA 3.1 ± 0.1 3.1 ± 0.1 2.2 ± 0.5 2.4 ± 0.3 2.2 ± 0.2 2.1 ± 0.2 1.8 ± 0.1 1.7 ± 0.1
Total 65.5 ± 1.4 65.1 ± 0.5 91.4 ± 22.4 92.9 ± 26.0 121.0 ± 4.9 119.6 ± 1.4 113.5 ± 4.1 109.6 ± 7.6
Lactate 3.5 ± 0.3 2.8 ± 0.2 3.9 ± 0.7 2.6 ± 0.1 3.3 ± 0.6 3.1 ± 0.4 4.4 ± 2.6 3.7 ± 2.4
Ammonium 436.4 ± 26.3 410.1 ± 28.8 302.1 ± 58.9 320.2 ± 47.1 280.6 ± 23.4 266.0 ± 17.4 257.2 ± 18.5 234.6 ± 14.5
DC
Acetate 35.7 ± 0.7 33.7 ± 0.0 39.5 ± 5.7 38.7 ± 12.3 45.3 ± 2.1 50.4 ± 4.8 44.1 ± 2.8 54.4 ± 4.9
Propionate 20.9 ± 0.8 18.0 ± 1.0 32.4 ± 12.8 33.1 ± 18.9 47.3 ± 1.2 52.4 ± 2.8 47.3 ± 1.1 49.6 ± 1.6
Butyrate* 8.7 ± 0.1 8.1 ± 0.4 15.3 ± 9.0 11.0 ± 4.2 24.8 ± 2.9 16.7 ± 1.5 23.1 ± 1 .8 15.5 ± 2.2
Valerate 1.6 ± 0.2 1.5 ± 0.5 2.9 ± 1.1 2.1 ± 1.2 4.2 ± 0.2 4.0 ± 0.2 3.1 ± 1.7 2.5 ± 1.7
Caproate 1.3 ± 0.3 1.4 ± 0.7 2.3 ± 1.3 1.1 ± 0.7 4.4 ± 0.4 2.3 ± 0.1 2.3 ± 2.2 1.1 ± 1.2
BCFA 3.3 ± 0.0 3.0 ± 0.0 2.7 ± 0.3 2.5 ± 0.1 2.6 ± 0.2 2.3 ± 0.3 2.2 ± 0.1 1.9 ± 0.1
Total 71.6 ± 0.3 65.8 ± 0.1 95.1 ± 29.5 88.5 ± 37.3 128.7 ± 0.8 128.1 ± 6.9 122.1 ± 1.6 125.0 ± 8.2
Lactate 3.7 ± 0.8 2.8 ± 0.5 3.9 ± 0.3 3.0 ± 0.5 3.2 ± 0.1 3.6 ± 0.0 3.2 ± 0.5 2.6 ± 0.5
Ammonium 456.0 ± 24.5 408.0 ± 58.5 380.7 ± 50.6 346.9 ± 3.9 348.8 ± 21.9 290.5 ± 53.1 301.7 ± 18.5 230.7 ± 17.6
Journal of Functional Foods 16 (2015) 211–222 215

Fig. 1 – pH profiles and gas production: pH profiles expressed as the difference of the consumption NaOH and HCl on a
weekly base, in the ascending (AC), transverse (TC) and descending (DC) colon compartments of the SHIME systems treated
with Blend 1 (A) and Blend 2 (B) (observation = 1). Samples were marked as: CTRL = control, TREAT = treatment; 1, 2 or 3
indicates the specific week during either the control or treatment period in which the samples were collected, (C) Δ total gas
production related to the treatment with Blend 1 and Blend 2 ± standard deviation (n = 3). Significant differences between
the two blends have been assessed by means of a Student’s two-tailed t-test and are indicated with * for P = 0.06.

blends induced an increase in the total SCFA concentration in colon for Blend 1 (with the exception of a peak in the DC during
all the simulated colon compartments (p < 0.01), indicating that the first week of treatment), acidification occurred through-
both products were well fermented in the GIT. Moreover, both out the entire simulated colon upon administration of Blend
products induced a higher concentration of propionate and 2 (Fig. 1a and 1b). These data were confirmed by the results
butyrate (Table 1) and were able to move the ratio acetate– of the SEC analysis. Blend 1 was quickly fermented in the simu-
propionate–butyrate (A/P/B) towards a higher proportion of P lated ascending colon compartment (Appendix: Supplementary
and B (data not shown). When evaluating statistical differ- Fig. S1) while, when dosing Blend 2, a residual part of the
ences in SCFA production between the two blends, the product (a peak with a MW above 200,000 g/mol correspond-
butyrogenic effect of Blend 1 was higher than that of Blend 2 ing to acacia gum) was still available for fermentation in the
(approx. +10 mmol/L), which in turn led to a higher propio- simulated distal colon (Appendix: Supplementary Fig. S2).
nate concentration (even if not statistically significant). Both Moreover, comparison of the samples from the different ex-
blends induced an increase in lactate production in the simu- perimental weeks showed that the remaining fractions of acacia
lated ascending colon compartment, with the production of gum decreased towards the end of the treatment period, in-
Blend 1 being 3.6 g/L higher than Blend 2 (p < 0.05) (Table 1). dicating a further adaptation of the microbial community
Finally, both products induced a decrease in ammonium pro- metabolism for AG.
duction during the treatment period in all the simulated colon The data related to the delta in increase/decrease (values
compartments (Table 1). of the treatment period minus values of the control period) of
The analysis of the NaOH and HCl consumption (acid and the activity of the enzymes potentially involved in the break-
base are used to maintain the colon compartments in the down of the tested blends are shown in Table 2. When dosing
correct pH range) showed that administration of both blends Blend 1, the alpha-galactosidase, alpha-L-arabinofuranosidase
induced acidification of the simulated colon reactors. However, and beta-galactosidase activities were proportionally mainly
whereas this effect was limited to the simulated ascending enhanced in the simulated AC. In contrast, Blend 2 led to a more
216 Journal of Functional Foods 16 (2015) 211–222

Table 2 – Increase or decrease of the average enzymatic activity ± standard deviation (U/mL) calculated as the delta
between the average of the treatment period (observations = 3) and the average of the control period (observations = 2) in
the ascending (AC), transverse (TC) and descending colon (DC) of the SHIME systems treated with Blends 1 and 2. No
significant differences between the two blends were observed by means of a Wilcoxon signed-rank.
Blend 1 Blend 2
AC TC DC AC TC DC
Alpha-L-arabinofuranosidase 1.6 ± 1.1 3.4 ± 1.2 2.4 ± 1.7 0.7 ± 0.5 3.9 ± 1.7 8.2 ± 4.4
Alpha-galactosidase 7 ± 1.5 7.8 ± 3.2 5.1 ± 4 4 ± 2.4 9.5 ± 3.1 7.3 ± 3.5
Beta-galactosidase 8.4 ± 2.6 −0.7 ± 12.5 2.3 ± 6 2.4 ± 2.6 3.6 ± 9.2 4.4 ± 8.6
Beta-glucosidase 0±0 0±0 0±0 0.1 ± 0 0.1 ± 0.1 0.05 ± 0
Beta-D-xylosidase 1.2 ± 2.7 0 ± 2.0 −1 ± 2.2 1 ± 2.1 2.8 ± 2.6 2.7 ± 2.4

balanced increase of the activity of all the tested enzymes (i.e. 3.2. Analysis of the microbial community composition
alpha-L-arabinofuranosidase, alpha-galactosidase, beta-
galactosidase, beta-glucosidase, beta-D-xylosidase) along the 3.2.1. Quantitative PCR and denaturing gradient gel
entire simulated colonic compartment. electrophoresis
Finally, a short-term measurement (up to 48h) of gas pro- qPCR and DGGE were used to monitor main changes in mi-
duction confirmed the difference in the fermentation profile crobial groups along the simulated colon compartments.
of the two blends, with Blend 1 leading to a higher gas pro- When comparing the control period and the last two weeks
duction as compared to Blend 2 (p = 0.06) (Fig. 1c). of treatment, administration of both blends resulted in a

Fig. 2 – Quantitative PCR results: qPCR data from each bacterial group presented per experimental week in each colon
compartment. (A) Total bacteria; (B) Bacteroidetes; (C) Firmicutes; (D) Bifidobacteria; (E) Lactobacilli. Samples were marked
as: CTRL = control, TREAT = treatment; 1, 2 or 3 indicates the specific week during either the control or treatment period in
which the samples were collected. For each bar is indicated the standard error of the technical replicate of the qPCR (n = 3).
* = the difference from the average of the control is statistically significant according to a t-test (P < 0.05) (n ≥ 6). To compare
the effect of the 2 Blends, a linear spline model was applied, with the knot positioned on the first week of treatment.
Samples with different letters are statistically different.
Journal of Functional Foods 16 (2015) 211–222 217

statistically significant increase of lactobacilli (p < 0.01) and the knot on the first week of treatment, in order to take into
bifidobacteria (p < 0.05) in all colon compartments (Fig. 2). Ad- account the extent of delay in the treatment effect.
ministration of both blends increased the counts of the number Both blends showed a bifidogenic effect. When comparing
of total bacteria and bacteria belonging to the phylum of the effect of the blends, the increase in bifidobacteria induced
Firmicutes. However, when considering the first experimen- by Blend 1 was statistically higher than Blend 2. Conversely,
tal weeks, Blend 2 initially led to a decrease in the 16S rRNA Blend 2 induced a higher increase in lactobacilli in the simu-
genes copy number for total bacteria in the simulated AC. This lated ascending colon as compared to Blend 1.
decrease mainly correlated with the decrease of the domi- DGGE analyses on total bacteria, bifidobacteria and lacto-
nant Firmicutes and Bacteroidetes phyla in the same colon bacilli were used to evaluate the qualitative changes that
compartment. Finally, Blend 2 induced also a statistically sig- potentially occurred within the structure of the microbial com-
nificant increase in Bacteroidetes in the simulated descending munity during the treatment period. Data are presented as a
colon (Fig. 2). cluster analysis conducted on a composite dataset of the three
To compare the effect of the 2 blends on the different bac- gels using the unweighted pair group with mathematical av-
terial groups, we applied a longitudinal statistical approach that erages (UPGMA) and distance matrices of each DGGE gel, based
allowed comparison of the different trends induced by the treat- on the Pearson correlation similarity coefficients (Fig. 3). For
ments. A linear spline model was used to fit the data positioning both blends, samples from treatment week 1 tended to cluster

100
15

20

25

30

35

40

75

80

85

95
45

50

55

60

65

70

90
CTRL2 TC
CTRL2 DC
CTRL2 AC
CTRL1 TC
CTRL1 DC
CTRL1 AC
TREAT1 TC
TREAT1 DC
TREAT1 AC
TREAT2 TC
TREAT2 DC
TREAT2 AC
TREAT3 AC
TREAT3 TC
TREAT3 DC

b
100
70

72

74

76

78

80

82

84

86

88

90

92

94

96

98

CTRL2 DC
CTRL2 AC
CTRL2 TC
CTRL1 TC
CTRL1 DC
CTRL1 AC
TREAT1 TC
TREAT1 DC
TREAT1 AC
TREAT2 AC
TREAT2 TC
TREAT2 DC
TREAT3 DC
TREAT3 AC
TREAT3 TC

Fig. 3 – DGGE dendrograms: Clustering analysis for a composite dataset, created by the combination of the information
contained within the DGGE gels for total bacteria, lactobacilli and bifidobacteria from Blend 1 (a) and Blend 2 (b) SHIME
experiments. The analysis on the composite dataset was performed based on the Pearson product–moment correlation
coefficient and the result is presented as dendrogram by using UPGMA linkage. Samples were marked as: CTRL = control,
TREAT = treatment; 1, 2 or 3 indicates the specific week during either the control or treatment period in which the samples
were collected for DNA extraction; AC, TC or DC indicates respectively ascending, transverse and descending colon.
218 Journal of Functional Foods 16 (2015) 211–222

separately from the other treatment samples and more closely Blend 1 is a common ingredient mixture in enteral nutri-
to the samples from the control period, confirming that long- tion formulas, and has been previously tested in vivo. In these
term administrations of the blends are needed to induce gradual studies, it was associated with significantly increased lacto-
changes towards the development of a new community bacilli in preadolescent cancer patients (n = 67), a bifidogenic
structure. effect in children (n = 140) following antibiotic treatment, and
an enhanced IgG antibody response to vaccination in infants
3.3. Illumina deep sequencing (Klosterbuer, Roughead, & Slavin, 2011). Due to its composi-
tion of readily fermented fructans (i.e. FOS and inulin), Blend
Analysis of the illumina data to evaluate the composition of 1 is predominantly fermented in the proximal colon
the microbial community confirmed a different modulating (Macfarlane, Gibson, & Cummings, 1992). Moreover, the single
effect of the 2 blends in the different simulated colon com- ingredients of the blend have also been positively associated
partments. The number of phylogenetically unclassified reads to biomarkers of large bowel health in rat studies (Paturi et al.,
had a decreasing trend when comparing the AC with TC and 2015). Although this blend has an established prebiotic benefit,
DC in the control period (Appendix: Supplementary Fig. S3). low amounts of fibre have to be dosed in order to minimize
Moreover, a decreasing trend was also observed in each simu- potential gastrointestinal discomfort due to rapid fermenta-
lated colon compartment when comparing control vs. tion (Barrett & Gibson, 2012). In order to extend the fermentation
treatment. Hierarchical clustering of phylogenetic families along the colon it was decided to replace part of FOS and inulin
showed a separation of treatment and control samples, as well with AG. The latter, due to its more complex structure (a poly-
as a clustering of gut sectors (Fig. 4a). Following the treat- saccharide with high MW), has been recently shown to be
ment, AC sectors were more separated from the other colon fermented more distally in the colon (Terpend et al., 2013). It
compartments and showed a closer clustering, while DC was therefore hypothesized that the addition of AG would
samples showed the least stable clustering (indicative of the extend positive physiological effects to the distal colon without
fact that the 2 blends had a different effect in the distal colon). significantly impacting the prebiotic activity of the original
The analysis of the functional genes associated to the dif- blend. Moreover, the prolonged fermentation was also sus-
ferent simulated colon compartments is shown in Fig. 4b, c and pected to beneficially impact the tolerance of the fibre blend.
d. A NMDS ordination of the KO abundances showed a clear By means of a dynamic, multi-compartment gastrointes-
separation between treated and control samples on axis 1 tinal simulator representative for the conditions of the human
(Fig. 4b). On the contrary, the simulated gut sectors are mostly ascending, transverse and descending colon, we were able to
close to each other under all conditions and show a similar show that the two blends had a different fermentation profile
distribution along axis 2 independently of the treatment. and induced different colonic region-specific effects (Fig. 5).
However, in a test blocked for the type of treatment and control, Indeed, the partial substitution of FOS and inulin with AG re-
it is possible to observe that there were differences along the sulted in a more gradual fermentation of the blend along the
simulated colon (Fig. 4c). For instance, while the cell struc- entire simulated colon. This main conclusion was supported
ture biosynthesis indicated a higher growth rate in the by several analyses conducted on the samples collected from
simulated proximal colon and in the control samples, fermen- the different areas of the simulated colon compartments. SEC
tation and amino-acid degradation increased mainly in the analysis and the analysis of the pH profiles showed that while
simulated distal colon and were higher for the samples from Blend 1 was mainly fermented in the proximal colon, Blend 2
the treatment period. Significant functional differences were was more gradually fermented, with some fractions of the
observed also when comparing control vs. treatment (analy- product being still available in the distal colon. SEC analysis
sis blocked for the gut region – Fig. 4d). Following the treatment also showed that the efficiency of bacteria using the AG added
with both blends, it was possible to observe a higher abun- in Blend 2 increased during the 3 weeks of treatment. These
dance of genes involved in fermentation (e.g. glutamate data and the adaptation property are in line with what was
degradation via hydroxyglutarate; pyruvate fermentation to described by Terpend et al. (2013) on the fermentation of AG.
ethanol; pyruvate fermentation to propionate via acrylate Moreover, the lower gas production induced by Blend 2 was
pathway; and acetyl-CoA fermentation to butyrate), protein and further evidence of the gradual fermentation of this product
lipid degradation (e.g. triacylglycerol degradation; fatty acid beta- (Fig. 1c). In line with these findings, Koecher, Thomas, and Slavin
oxidation; acetoacetate degradation to acetyl CoA) and (2015) showed – in a randomized, double-blind, crossover trial
carboxylates degradation. Finally, during the treatment with with 20 healthy subjects – that the consumption of a blend very
Blend 1 and Blend 2, glycolysis was substituted by the faster, similar to Blend 2 used in this work (i.e. a mix of FOS, inulin,
but less energy-yielding Entner–Doudoroff pathway (Fig. 4d). AG and pea hull fibre) led to less negative symptoms related
This effect was stronger in the distal colon for Blend 2. to bowel urgency and no problem with gas production and
bloating symptoms.
The SCFA data gave indications of a different but benefi-
cial fermentation profile for the two blends, especially when
4. Discussion considering the net production in the simulated TC compart-
ment (Fig. 5). When evaluating differences in SCFA production
The objective of this study was to compare the effect of long- between control and treatment, the standard deviation of the
term repeated administration of an enteral feed containing two average of the 3 samples collected during the first week of treat-
different fibre blends on the composition and metabolism of ment resulted to be high (Table 1). This relates to the adaptation
the gut microbiota. period the bacteria needed to modulate their metabolism to
Journal of Functional Foods 16 (2015) 211–222 219

a) b)
0.5

0.15
AC1_TREAT
0.4

0.10
AC2_CTRL
AC1_CTRL AC2_TREAT
0.3

NMDS D2 (27.79%)
0.05
0.2

TC1_TREAT

0.00
TC2_TREAT
TC1_CTRL
TC2_CTRL

−0.05
0.1

DC1_CTRL DC2_TREAT
DC1_TREAT

−0.10
0.0

AC2_TREAT
TC2_TREAT

DC2_TREAT
AC2_CTRL

AC1_CTRL

TC1_CTRL

DC1_CTRL

DC1_TREAT

TC1_TREAT

AC1_TREAT
TC2_CTRL

DC2_CTRL

−0.15
DC2_CTRL
−0.15 −0.10 −0.05 0.00 0.05 0.10 0.15
NMDS D1 (55.24%)

c) cell−structure−biosynthesis fermentation amino−acid−degradation


p= 0.02237 p= 0.01608 p= 0.0186
0.0044

AC
TC
0.0020

DC
log10 Read Count

0.0040
0.0038

0.0016

0.0030
0.0032

d1

d2
d2

d1
d1

L
d2

TR

TR
en

TR
en
en
en
en
en
Bl

Bl

Bl
Bl

Bl

C
Bl

C
d) fatty−acid−and−lipid−degradation
p= 0.0221
protein−degradation
p= 0.01413
fermentation
p= 0.01574

Blend1
0.00055

Blend2
0.0020

CONTROL
log10 Read Count

4e−05
0.00035 0.00045

2e−05

0.0016
0e+00
AC

TC

AC

TC

AC

TC

C
D

carboxylates−deg glycolysis−variants MC0098:Entner−Doudoroff pathway I


p= 0.01246 p= 0.02788 p= 0.01246
3.0e−05
0.00075
log10 Read Count
0.0010

1.5e−05
0.00065
0.0008

0.0e+00

AC

TC

C
AC

TC

C
AC

TC

D
D
D

Fig. 4 – 16S rRNA-targeted Illumina sequencing data on samples from the end of control and treatment periods:
(a) hierarchical clustering based on phylogenetic families and (b) NMDS (Non-metrical dimensional scaling) ordination of
the KO abundances for samples belonging to ascending (AC), transverse (TC) and descending (DC) colon compartments of
the SHIME treated with Blend 1 and Blend 2 during the last week of control (CTRL) and the last week of treatment (TREAT)
period. Each colon compartment is indicated with numbers 1 or 2 depending on which Blend was tested. D1 and D2 refer to
the first 2 dimensions of the NMDS analysis. (c) Selected significant functional differences between AC, TC and DC in a test
blocked for the type of treatment.(d) Significant functional differences between Blends 1 and 2 and the control samples, in a
test blocked for the colon region.
220 Journal of Functional Foods 16 (2015) 211–222

clustering of phylogenetic families showed that the TC and DC


sectors of the SHIME treated with Blend 2 tended to cluster
much closer following the treatment as compared to the re-
spective control period, indicative of a modulation of the
microbiota induced by the treatment itself. Moreover, when
comparing the 2 SHIME systems, the least stable clustering was
associated with the DC sectors. This was again indicative of
the fact that the two blends had a differential effect in the distal
colon. DGGE analyses showed that long-term repeated admin-
istrations of the blends are needed to reach gradual changes
in the gut microbial community structure. The increased con-
centration of Bacteroidetes in the simulated DC upon Blend 2
administration further confirmed the more gradual fermen-
tation of AG in the distal colon. Bacteroidetes are considered
as very important saccharolytic fermenting bacteria, as a large
part of the proteins codified by this phylum goes to breaking
down polysaccharides and metabolizing their sugars (Karlsson,
Ussery, Nielsen, & Nookaew, 2011). Also bacteria belonging to
the phylum of Firmicutes – frequent users of the metabolic in-
termediates produced by the metabolism of Bacteroidetes –
Fig. 5 – Schematic summary of the blends effect in the showed a similar trend. Both blends had a positive effect on
colon: Summary of the effects that Blend 1 and Blend 2 bifidobacteria and lactobacilli – two genera indicated as po-
induced in the different areas of the simulated colon with tentially beneficial bacteria – showing that the new composition
respect to: pH, ammonium (NH3), positive net production of of Blend 2 supported this potential health-promoting effect
acetate (A), propionate (P) and butyrate (B) (the relative (Ventura et al., 2007).
letter is present in a colon section only if the treatment The aim of this work was to show that the structural char-
induced a positive net production of the specific SCFA), acteristics of AG (i.e. a polysaccharide with high MW) were
glycolysis (GL), Entner–Doudoroff pathway (ED), linked to a distal fermentation of the product. In this type of
bifidobacteria (Bif), lactobacilli (Lactos) and residual amount studies, the relevance of the data in terms of potential
of substrate to be fermented (*). interindividual variability (i.e. different effect of the test product
due to a different composition of the gut microbiota) is always
questionable. In this respect, we can conclude that the distal
the new nutritional environment. In parallel with the above, fermentation of AG appears to be independent of the tested
a decrease in ammonium production was observed during the microbiota as we were able to show this specific fermenta-
treatment with both blends, indicative of the fact that both tion pathway working with a mix of faecal samples from three
products were well-fermented by bacteria. different donors (Terpend et al., 2013), with a new indepen-
In this study, an illumina metagenomic approach has been dent donor in this study and also with a third donor as shown
used for the first time in a SHIME experiment. The analysis of in Appendix: Supplementary Fig. S4 (unpublished results).
the functional genes at DNA level (i.e. potentiality of the mi- Moreover, most studies on gut microbiota are based on the
crobial metabolic activity) confirmed that differences occurred analysis of faecal samples because they are easily collected in
between the two blends in the simulated colon compart- a non-invasive manner (Durban et al., 2011). However, the rep-
ments. Among them, the most interesting effect is that related resentativeness of these samples for the microbial processes
to the switch between the energy pathways of glycolysis and occurring along the colon may be questioned (e.g. SCFA that
Entner–Doudoroff (ED). ED is commonly used by bacteria when are produced in the ascending colon are readily absorbed or
carbon sources are not limiting in a system for carbohydrate used in cross-feeding reactions). This work shows that the use
metabolism: it is faster even if less energy-yielding (1 ATP/ of in vitro models provides a tool for looking at potential changes
glucose vs. 2 ATP/glucose of the glycolysis) (Conway, 1992). For in other colonic areas that may not be or are poorly repre-
both blends, glycolysis decreased while the ED pathway in- sented by a faecal sample (i.e. proximal colon). This approach
creased and this effect was colon-sector specific (Fig. 4d). More represents an interesting preliminary or complementary tool
specifically, with Blend 1 – which was fermented more proxi- to in vivo trials in order to elucidate the potential mechanism
mally – the decrease in genes involved in glycolysis was stronger of action of a given product.
in the simulated AC. On the contrary, with Blend 2, the de-
crease in glycolysis was more balanced and extended also to
the simulated DC. An opposite and complementary effect was
observed for the genes involved in the ED pathway (i.e. the in- 5. Conclusions
crease for Blend 2 in the distal colon was bigger than that induced
by Blend 1). This work suggests that inulin/FOS and acacia gum may exert
The analyses on the microbiota (qPCR, DGGE and illumina) an interesting complementary effect. In fact, the partial re-
further confirmed a different modulating effect of the 2 blends placement of FOS and inulin by acacia gum changed the
in the simulated colon compartments. In fact, the hierarchical intestinal fermentation profile from boost fermentation in the
Journal of Functional Foods 16 (2015) 211–222 221

proximal colon into a more gradual fermentation in the com- Durban, A., Abellan, J. J., Jimenez-Hernandez, N., Ponce, M.,
plete colon. Blend 2 extended the benefits of fermentation (e.g. Ponce, J., Sala, T., D’Auria, G., Latorre, A., & Moya, A. (2011).
decreased ammonium production, acidification, SCFA produc- Assessing gut microbial diversity from feces and rectal
mucosa. Microbial Ecology, 61(1), 123–133. doi:10.1007/
tion) to the distal colon, and at the same time attenuated gas
s00248-010-9738-y.
production. Further in vivo studies are needed to link the outcome Guo, X., Xia, X., Tang, R., Zhou, J., Zhao, H., & Wang, K. (2008).
of this study to the possibility of a greater provision of fibre in Development of a real-time PCR method for Firmicutes and
enteral products, with potentially reduced risk of intolerance. Bacteroidetes in faeces and its application to quantify
intestinal population of obese and lean pigs. Letters in Applied
Microbiology, 47(5), 367–373. doi:10.1111/j.1472-765X.2008
.02408.x.
Acknowledgments Hildebrand, F., Ebersbach, T., Nielsen, H. B., Li, X., Sonne, S. B.,
Bertalan, M., Dimitrov, P., Madsen, L., Qin, J., Wang, J., Raes, J.,
Kristiansen, K., & Licht, T. R. (2012). A comparative analysis of
The authors would like to thank Tim Lacoere for his support
the intestinal metagenomes present in guinea pigs (Cavia
on graphical design, and Eva Ackermann and Terri Born for their porcellus) and humans (Homo sapiens). BMC Genomics, 13,
work on preparing the fibre samples. doi:10.1186/1471-2164-13-514.
M.M., F.H. and J.R. are funded by the Research Foundation Hildebrand, F., Nguyen, T. L., Brinkman, B., Yunta, R. G., Cauwe,
Flanders (FWO, Belgium). F.H. and J.R. are also funded by the B., Vandenabeele, P., Liston, A., & Raes, J. (2013).
Brussels Institute for Research and Innovation, the EU Frame- Inflammation-associated enterotypes, host genotype, cage
and inter-individual effects drive gut microbiota variation in
work 7 programme (MetaCardis Health-F4-2012-305312), VIB,
common laboratory mice. Genome Biology, 14(1), doi:10.1186/
the REGA institute and KU Leuven. Support for this study was
gb-2013-14-1-r4.
provided by Nestlé Health Science S.A. (Vevey, Switzerland). Hothorn, T., Hornik, K., Van de Wiel, M. A., & Zeileis, A. (2006). A
Lego system for conditional inference. The American
Statistician, 60(3), 257–263. doi:10.1198/000313006x118430.
Karlsson, F. H., Ussery, D. W., Nielsen, J., & Nookaew, I. (2011). A
Appendix: Supplementary material closer look at bacteroides: Phylogenetic relationship and
genomic implications of a life in the human gut. Microbial
Supplementary data to this article can be found online at Ecology, 61(3), 473–485. doi:10.1007/s00248-010-9796-1.
Klosterbuer, A., Roughead, Z. F., & Slavin, J. (2011). Benefits of
doi:10.1016/j.jff.2015.04.039.
dietary fiber in clinical nutrition. Nutrition in Clinical Practice,
26(5), 625–635. doi:10.1177/0884533611416126.
Koecher, K. J., Thomas, W., & Slavin, J. L. (2015). Healthy
REFERENCES
subjects experience bowel changes on enteral diets:
Addition of a fiber blend attenuates stool weight and gut
bacteria decreases without changes in gas. Journal of
Al-Sheraji, S. H., Ismail, A., Manap, M. Y., Mustafa, S., Yusof, R. M., Parenteral and Enteral Nutrition, 39(3), 337–343. doi:10.1177/
& Hassan, F. A. (2013). Prebiotics as functional foods: A review. 0148607113510523.
[Review]. Journal of Functional Foods, 5(4), 1542–1553. Lee, S. M., Donaldson, G. P., Mikulski, Z., Boyajian, S., Ley, K., &
doi:10.1016/j.jff.2013.08.009. Mazmanian, S. K. (2013). Bacterial colonization factors control
Barrett, J. S., & Gibson, P. R. (2012). Fermentable oligosaccharides, specificity and stability of the gut microbiota. Nature,
disaccharides, monosaccharides and polyols (FODMAPs) and 501(7467), 426. doi:10.1038/nature12447.
nonallergic food intolerance: FODMAPs or food chemicals? Li, R., Yu, C., Li, Y., Lam, T.-W., Yiu, S.-M., Kristiansen, K., & Wang,
Therapeutic Advances in Gastroenterology, 5(4), 261–268. J. (2009). SOAP2: An improved ultrafast tool for short read
Benjamini, Y., & Hochberg, Y. (1995). Controlling the false alignment. Bioinformatics (Oxford, England), 25(15), 1966–1967.
discovery rate – A practical and powerful approach to doi:10.1093/bioinformatics/btp336.
multiple testing. Journal of the Royal Statistical Society: Series B Li, R., Zhu, H., Ruan, J., Qian, W., Fang, X., Shi, Z., Li, Y., Li, S., Shan,
(Methodological), 57(1), 289–300. G., Kristiansen, K., Li, S., Yang, H., Wang, J., & Wang, J. (2010).
Berg, J. O., Nord, C. E., & Wadstrom, T. (1978). Formation of De novo assembly of human genomes with massively parallel
glycosidases in batch and continuous cultures of Bacteroides short read sequencing. Genome Research, 20(2), 265–272.
fragilis. Applied and Environmental Microbiology, 35, 269–273. doi:10.1101/gr.097261.109.
Blottiere, H. M., de Vos, W. M., Ehrlich, S. D., & Dore, J. (2013). Maccaferri, S., Biagi, E., & Brigidi, P. (2011). Metagenomics: Key to
Human intestinal metagenomics: State of the art and future. human gut microbiota. Digestive Diseases, 29(6), 525–530.
Current Opinion in Microbiology, 16(3), 232–239. doi:10.1016/ doi:10.1159/000332966.
j.mib.2013.06.006. Macfarlane, G. T., Gibson, G. R., & Cummings, J. H. (1992).
Boon, N., Top, E. M., Verstraete, W., & Siciliano, S. D. (2003). Comparison of fermentation reactions in different regions
Bioaugmentation as a tool to protect the structure and of the human colon. Journal of Applied Bacteriology, 72(1), 57–
function of an activated sludge microbial community against 64.
a 3-chloroaniline shock load. Applied and Environmental Marzorati, M., Maignien, L., Verhelst, A., Luta, G., Sinnott, R.,
Microbiology, 69(3), 1511–1520. Kerckhof, F. M., Boon, N., Van de Wiele, T., & Possemiers, S.
Cherbut, C., Michel, C., & Lecannu, G. (2003). The prebiotic (2013). Barcoded pyrosequencing analysis of the microbial
characteristics of fructooligosaccharides are necessary for community in a simulator of the human gastrointestinal tract
reduction of TNBS-induced colitis in rats. Journal of Nutrition, showed a colon region-specific microbiota modulation for two
133(1), 21–27. plant-derived polysaccharide blends. Antonie Van Leeuwenhoek,
Conway, T. (1992). The Entner-Doudoroff pathway – History, 103(2), 409–420. doi:10.1007/s10482-012-9821-0.
physiology and molecular-biology. FEMS Microbiology Letters, Molly, K., Woestyne, M. V., & Verstraete, W. (1993). Development
103(1), 1–28. of a 5-step multichamber reactor as a simulation of the
222 Journal of Functional Foods 16 (2015) 211–222

human intestinal microbial ecosystem. Applied Microbiology community in a simulator of the human intestinal microbial
and Biotechnology, 39(2), 254–258. ecosystem. FEMS Microbiology Ecology, 49, 495–507.
Noguchi, H., Park, J., & Takagi, T. (2006). MetaGene: Prokaryotic Qin, J. J., Li, R. Q., Raes, J., Arumugam, M., Burgdorf, K. S.,
gene finding from environmental genome shotgun Manichanh, C., Nielsen T., Pons N., Levenez F., Yamada T.,
sequences. Nucleic Acids Research, 34(19), 5623–5630. Mende D. R., Li J., Xu J., Li S., Li D., Cao J., Wang B., Liang H.,
doi:10.1093/nar/gkl723. Zheng H., Xie Y., Tap J., Lepage P., Bertalan M., Batto J. M.,
Overbeek, R., Begley, T., Butler, R. M., Choudhuri, J. V., Chuang, H. Hansen T., Le Paslier D., Linneberg A., Nielsen H. B., Pelletier
Y., Cohoon, M., de Crécy-Lagard, V., Diaz, N., Disz, T., Edwards, E., Renault P., Sicheritz-Ponten T., Turner K., Zhu H., Yu C., Li
R., Fonstein, M., Frank, E. D., Gerdes, S., Glass, E. M., S., Jian M., Zhou Y., Li Y., Zhang X., Li S., Qin N., Yang H., Wang
Goesmann, A., Hanson, A., Iwata-Reuyl, D., Jensen, R., J., Brunak S., Doré J., Guarner F., Kristiansen K., Pedersen O.,
Jamshidi, N., Krause, L., Kubal, M., Larsen, N., Linke, B., Parkhill J., Weissenbach J., MetaHIT Consortium, Bork P.,
McHardy, A. C., Meyer, F., Neuweger, H., Olsen, G., Olson, R., Ehrlich S. D. & Wang J. (2010). A human gut microbial gene
Osterman, A., Portnoy, V., Pusch, G. D., Rodionov, D. A., catalogue established by metagenomic sequencing. [Article].
Rückert, C., Steiner, J., Stevens, R., Thiele, I., Vassieva, O., Ye, Nature, 464(7285), 59–U70. doi:10.1038/nature08821.
Y., Zagnitko, O., & Vonstein, V. (2005). The subsystems Sanchez, J. I., Marzorati, M., Grootaert, C., Baran, M., Van
approach to genome annotation and its use in the project to Craeyveld, V., Courtin, C. M., Broekaert, W. F., Delcour, J. A.,
annotate 1000 genomes. Nucleic Acids Research, 33(17), 5691– Verstraete, W., & Van de Wiele, T. (2009). Arabinoxylan-
5702. doi:10.1093/nar/gki866. oligosaccharides (AXOS) affect the protein/carbohydrate
Paturi, G., Butts, C. A., Bentley-Hewitt, K. L., Hedderley, D., fermentation balance and microbial population dynamics of
Stoklosinski, H., & Ansell, J. (2015). Differential effects of the Simulator of Human Intestinal Microbial Ecosystem.
probiotics, prebiotics, and synbiotics on gut microbiota and Microbial Biotechnology, 2, 103–113.
gene expression in rats. [Article]. Journal of Functional Foods, 13, Terpend, K., Possemiers, S., Daguet, D., & Marzorati, M. (2013).
204–213. doi:10.1016/j.jff.2014.12.034. Arabinogalactan and fructo-oligosaccharides have a different
Possemiers, S., Bolca, S., Grootaert, C., Heyerick, A., Decroos, K., fermentation profile in the Simulator of the Human Intestinal
Dhooge, W., De Keukeleire, D., Rabot, S., & Van de Wiele, T. Microbial Ecosystem (SHIME (R)). Environmental Microbiology
(2006). The prenylflavonoid isoxanthohumol from hops Reports, 5(4), 595–603. doi:10.1111/1758-2229.12056.
(Humulus lupulus L.) is activated into the potent phytoestrogen Van de Wiele, T., Boon, N., Possemiers, S., Jacobs, H., & Verstraete,
8-prenylnaringenin in vitro and in the human intestine. W. (2004). Prebiotic effects of chicory inulin in the simulator of
Journal of Nutrition, 136(7), 1862–1867. the human intestinal microbial ecosystem. FEMS Microbiology
Possemiers, S., Marzorati, M., Verstraete, W., & Van de Wiele, T. Ecology, 51, 143–153.
(2010). Bacteria and chocolate: A successful combination for Ventura, M., O’Connell-Motherway, M., Leahy, S., Moreno-Munoz,
probiotic delivery. International Journal of Food Microbiology, J. A., Fitzgerald, G. F., & van Sinderen, D. (2007). From bacterial
141(1–2), 97–103. doi:10.1016/j.ijfoodmicro.2010.03.008. genome to functionality; case bifidobacteria. International
Possemiers, S., Verthe, K., Uyttendaele, S., & Verstraete, W. (2004). Journal of Food Microbiology, 120(1–2), 2–12. doi:10.1016/
PCR-DGGE-based quantification of stability of the microbial j.ijfoodmicro.2007.06.011.

S-ar putea să vă placă și