OUP user menu

Comparison of prebiotic effects of arabinoxylan oligosaccharides and inulin in a simulator of the human intestinal microbial ecosystem

Charlotte Grootaert, Pieter Van den Abbeele, Massimo Marzorati, Willem F. Broekaert, Christophe M. Courtin, Jan A. Delcour, Willy Verstraete, Tom Van de Wiele
DOI: http://dx.doi.org/10.1111/j.1574-6941.2009.00712.x 231-242 First published online: 1 August 2009


In this study, the prebiotic potential of arabinoxylan oligosaccharides (AXOS) was compared with inulin in two simulators of the human intestinal microbial ecosystem. Microbial breakdown of both oligosaccharides and short-chain fatty acid production was colon compartment specific, with ascending and transverse colon being the predominant site of inulin and AXOS degradation, respectively. Lactate levels (+5.5 mM) increased in the ascending colon during AXOS supplementation, while propionate levels (+5.1 mM) increased in the transverse colon. The concomitant decrease in lactate in the transverse colon suggests that propionate was partially formed over the acrylate pathway. Furthermore, AXOS supplementation strongly decreased butyrate in the ascending colon, this in parallel with a decrease in Roseburia spp. and Bacteroides/Prevotella/Porphyromonas (−1.4 and −2.0 log CFU) levels. Inulin treatment had moderate effects on lactate, propionate and butyrate levels. Denaturing gradient gel electrophoresis analysis revealed that inulin changed microbial metabolism by modulating the microbial community composition. In contrast, AXOS primarily affected microbial metabolism by ‘switching on’ AXOS-degrading enzymes (xylanase, arabinofuranosidase and xylosidase), without significantly affecting microbial community composition. Our results demonstrate that AXOS has a higher potency than inulin to shift part of the sugar fermentation toward the distal colon parts. Furthermore, due to its stronger propionate-stimulating effect, AXOS is a candidate prebiotic capable of lowering cholesterol and beneficially affecting fat metabolism of the host.

  • arabinoxylan oligosaccharides
  • inulin
  • prebiotic
  • Roseburia
  • Faecalibacterium prausnitzii


The intestinal microbial community is more and more regarded as an important determinant for the general health status of the human body. The gut microbiota are involved in the host's immune system (Salminen et al., 1998), modulate the etiology of colon cancer (Hughes & Rowland, 2000; Lupton, 2004), form potent estrogenic metabolites from phytoestrogens (Decroos et al., 2006; Possemiers et al., 2006) and have recently been linked to obesity-related processes by their effect on fat storage and synthesis (Delzenne & Williams, 2002; Backhed et al., 2004, 2007). Therefore, modulation of the intestinal microbial composition to improve the host's health is a point of interest. This can be done by taking up either probiotics, which are living microbial feed supplements that beneficially affect the host by improving its intestinal microbial balance (Fuller, 1989), or prebiotics, substrates improving the host health by selectively stimulating the growth and/or activity of one or a limited number of beneficial bacteria in the colon (Gibson et al., 1995).

Many prebiotics belong to the group of nondigestible oligosaccharides, which resist digestion and absorption in the human small intestine and are fermented in the large intestine. A new class of candidate prebiotics is that of the arabinoxylan oligosaccharides (AXOS). AXOS are derived from arabinoxylans, complex carbohydrates found in the cell walls of the starch endosperm and the aleurone layer (Fincher & Stone, 1986) and in pericarp tissues (Maes & Delcour, 2002) of cereals. Arabinoxylans consist of a β-(1,4)-linked d-xylose backbone substituted at the C(O)-3 and/or the C(O)-2 position(s) with monomeric α-l-arabinose side chains, to which ferulic acid can be esterified (Cleemput et al., 1993). Hydrolysis of these highly polymerized arabinoxylans leads to the formation of AXOS, which are characterized by their average degree of polymerization (avDP) and average degree of arabinose substitution (avDAS) (Swennen et al., 2005, 2006).

Very little information is available on the prebiotic potency of AXOS. Thus far, the prebiotic properties of mainly high-molecular-weight arabinoxylans have been investigated as reviewed by Grootaert et al. (2007). Fermentation of arabinoxylans by human intestinal bacteria is related to an increase in butyric acid (Grasten et al., 2003) and propionic acid production (Amrein et al., 2003; Hopkins et al., 2003), which may, respectively, have protective effects against colon cancer and cholesterol-lowering effects. Furthermore, arabinoxylans are effective against type II diabetes by decreasing postprandial glucose levels and insulin responses (Lu et al., 2004) and increased postprandial ghrelin levels in the blood (Mohlig et al., 2005). Interestingly, the size of arabinoxylans in terms of molecular weight is an important parameter for prebiotic effects, as Hughes et al. (2007) discovered more such pronounced effects in vitro for lower-molecular-weight (66 kDa) arabinoxylans. Additionally, xylanase-pretreated wheat arabinoxylans stimulated colonic bifidobacteria in a three-stage continuous model of the human gut (Vardakou et al., 2007). Yet, this xylanase-pretreated arabinoxylan was not characterized and was only supplemented to the reactor for a couple of days. Prebiotic indices obtained from batch cultures with enzyme-treated wheat arabinoxylans were higher than those of untreated arabinoxylans and comparable to those obtained from commercial fructo-oligosaccharides, suggesting the release of arabinoxylan oligomers (Vardakou et al., 2008). Finally, there are indications that nonsubstituted AXOS, the xylo-oligosaccharides, have stronger bifidogenic properties than fructo-oligosaccharides and that both types of oligosaccharides may have a positive influence on the risk of colon cancer, as they markedly reduce the number of aberrant crypt foci in the colon of 1,2-dimethylhydrazine-treated rats (Hsu et al., 2004).

AXOS have been enzymatically produced on a large scale and were separated and characterized in terms of their avDP and avDAS (Swennen et al., 2005, 2006). Because of their current availability and structural diversity, it would be interesting to explore their prebiotic potential in depth. In this study, we compared the prebiotic properties of AXOS with avDP 15 and avDAS 0.26 with those of chicory root inulin as a reference prebiotic in a twin setup of the simulator of the human intestinal microbial ecosystem (SHIME).

Materials and methods

Oligosaccharide preparations

A commercial preparation of chicory root inulin (Fibruline Instant), with a degree of polymerization ranging between 3 and 60, with an avDP of 20, was kindly provided by COSUCRA (Warcoing, Belgium). AXOS with an avDP of 15 and an avDS of 0.27 (hereafter referred to as AXOS-15-0.27) was prepared as described earlier (Swennen et al., 2006). Briefly, commercial wheat bran was treated with α-amylase and protease to hydrolyze starch and proteins. Thereafter, the suspension was boiled, filtered and the residue washed with water. The washed residue was resuspended in water and treated twice with a Bacillus subtilis endoxylanase preparation (Grindamyl H640; Danisco, Copenhagen, Denmark). After filtration and subsequent heat inactivation of the enzyme, the supernatant was consecutively fed into an evaporator and spray drier. The product had a polymeric arabinose and xylose content (calculated as 0.88 of the sum of arabinose and xylose released upon hydrolysis) of 72.3% (dry weight).

Simulation of the human gastrointestinal tract

The experimental setup consisted of two identical units of the SHIME operating in parallel, further referred to as Twin-SHIME. Each SHIME unit consists of five double-jacketed vessels maintained at a temperature of 37 °C, respectively, simulating the stomach and duodenum, small intestine (jejunum and ileum), ascending, transverse and descending colon. Both SHIME units were inoculated with a fecal sample from a healthy adult volunteer (age 22) who had no history of antibiotic treatment in the 6 months before the sample delivery. More information on the isolation of the bacterial inoculum, the composition of the feed and technical details of the operation of the Twin-SHIME has been specified before (Molly et al., 1993; Possemiers et al., 2008). The Twin-SHIME run consisted of a 2-week basal period during which basal feed medium was supplemented to the reactor, a 3-week treatment period during which either inulin or AXOS-15-0.26 was supplemented to the respective SHIME units and, finally, a 2-week washout period without inulin or AXOS addition. It is important to note that, during the treatment periods, part of the starch (3 g) was replaced by an equal amount of either inulin or AXOS in the feed for the respective SHIME units. This way, the colon microbiota received the same amount of carbohydrates throughout the entire SHIME run.

Metabolic activity analysis

SCFA and ammonium ion analysis were performed as described before (Nollet et al., 1997; De Boever et al., 2000). Lactate was measured using the UV-method for determination of d-lactate/l-lactate kit (R-Biopharm, Mannheim, Germany), and succinate by the UV-method for determination of succinate kit (R-Biopharm) according to the manufacturer's protocols. Short-chain fatty acids (SCFA), ammonium ion and lactate in the different colon compartments are depicted as ‘net production’ values. For the ascending colon, this net production value corresponds to the absolute concentration. For the transverse colon, the net production value is calculated by subtracting the absolute concentration in the ascending colon from the absolute concentration in the transverse colon. For the descending colon, the net production value is calculated by subtracting the absolute concentration in the transverse colon from the absolute concentration in the descending colon. α-l-Arabinofuranosidase, β-d-xylosidase and endo-1,4-β-xylanase activities were determined with p-nitrophenyl-α-l-arabinofuranoside (Sigma-Aldrich, Bornem, Belgium), p-nitrophenyl-β-d-xylopyranoside (Sigma-Aldrich) and azo-wheat-arabinoxylan (Megazyme, Bray, Ireland), respectively. For the arabinofuranosidase and xylosidase activities, 100 μL of SHIME suspension was added to 100 μL of the substrate solution (5 mM) and incubated for 120 min at 37 °C. The change in A450 nm was measured (Dr Lange ISIS 9000 MDA photometer, Berlin, Germany). The enzymatic activity was calculated using a standard curve made with p-nitrophenol. One unit is the enzyme activity needed to release 1 μmol of p-nitrophenol from the appropriate substrate (2.5 mM) in 1 min under the assay conditions. To determine the xylanase activity, 200 μL SHIME suspension was added to 50 μL acetate buffer (25 mM, pH 4.5) and 250 μL azo-wheat-arabinoxylan solution, prepared according to the manufacturer's protocol. The mixture was incubated at 40 °C for 120 min and 1.5 mL ethanol [95% (v/v)] was added to stop the reaction. The samples were centrifuged at 1500 g for 10 min, and changes in supernatant A588 nm were measured. One unit is the enzyme activity needed to release 1 μmol of xylose equivalents from azo-wheat-arabinoxylan (1% w/v) in 1 min under the assay conditions.

Microbial community study

Plate counts

To assess the effect of AXOS and inulin on the general groups of bacteria in the SHIME, plate counts were performed on specific media: brain–heart infusion agar (total aerobes and total anaerobes, Oxoid, Hampshire, UK), tryptose sulfite cycloserin agar (clostridia, Merck, NJ), raffinose Bifidobacterium agar (bifidobacteria) (Hartemink & Rombouts, 1999), LAMVAB agar (lactobacilli) (Hartemink et al., 1997), Enterococcus agar (enterococci, Difco, Lawrence) and mannitol salt agar (staphylococci, Oxoid).

Quantitave PCR (Q-PCR) analysis

Q-PCR for Roseburia spp. and Faecalibacterium prausnitzii were performed as described before (Ramirez-Farias et al., 2009), using the primers Ros-F1 and Ros-R1, and Fprau645R and FPR-2F, respectively, and the Power SYBR Green PCR Master kit (Applied Biosystems, Foster City). Faecalibacterium prausnitzii was kindly provided by Philippe Langella from the INRA institute in Jouys-en-Josas, France, and Roseburia DNA by Isabelle François, Fugeia N.V., Heverlee, Belgium. The Q-PCR for Bacteroides/Prevotella/Porphyromonas spp. was performed as described by Rinttilä (2004), using the qPCR Core kit for SYBR Green I (Eurogentec, Seraing, Belgium) and primers Bacter140f and Bacter140r. All Q-PCR were performed with an ABI PRISM SDS 7000 Sequence Detection System (Applied Biosystems, Nieuwerkerk a/d Ijssel, the Netherlands).

PCR-denaturing gradient gel electrophoresis (DGGE) analysis of microbial community

The protocol for total DNA extraction from the SHIME samples was described earlier (Boon et al., 2000). The 16S rRNA genes of all bacteria were amplified applying general primers P338f with GC-clamp and P518r on total extracted DNA (Muyzer et al., 1993). For the nested PCR, the DNA was first amplified with bifidobacterial primers Bif164f and Bif662r (Satokari et al., 2001), followed by a PCR with the general DGGE primers (Van de Wiele et al., 2004). DGGE was performed as described earlier using the Bio-Rad D Gene System (Bio-Rad, Hercules, CA) (Muyzer et al., 1993; Van de Wiele et al., 2004). The normalization and analysis of DGGE gel patterns was performed with the bionumerics software 2.0 (Applied Maths, Kortrijk, Belgium). The calculation of the similarity matrix was based on the Pearson correlation coefficient and the clustering algorithm of Ward was used to calculate dendrograms (Ward, 1963). For the identification of the bacteria, the PCR bands were cut from the gel with a clean scalpel and added to 20 μL of PCR water (Sigma-Aldrich). Sequencing of the DNA fragments was carried out by ITT Biotech-Bioservice (Bielefeld, Germany). Analysis of DNA sequences and homology searches were completed with standard DNA sequencing programs and the server of the ribosomal database project (RBD) using the sequence match algorithm. Interpretation of the molecular fingerprints in terms of range-weighted richness (Rr), dynamics (Dy) and functional organization (Fo) was performed according to Marzorati et al. (2008). Briefly, the Rr defines the carrying capacity of the system, and is related to the amount and location of the DGGE bands in the gel. The Dy reflects the specific rate of species coming to significance and is related to correlation between the fingerprints of two subsequent time points. Finally, the Fo describes the relation between the composition of the microbial community and its functionality, based on the amount of bands compared with their intensities according to a Pareto–Lorentz curve.


Significance was tested using a one-way anova test using s-plus 7.0 software. Error bars indicate the SEM. Significantly different results are indicated with a different letter (P<0.05). At least six measurements were used for the statistical analysis of the SCFA, ammonium ion, lactate and Q-PCR results, and four to seven measurements were used in case of plate counts.


Metabolic activity

The measurement of the average net production of SCFA and lactate, the major carbohydrate fermentation products in the colon compartments of the Twin-SHIME revealed striking differences between the inulin and AXOS treatments (Figs 1 and 2, lower panels). Firstly, total SCFA production strongly decreased in the ascending colon compartment during AXOS treatment, which was mainly due to a significant decrease in average acetate production of −7.0 mM. In parallel, a strong increase in lactate (+2.4 mM) was observed in this part of the colon, which continued in the washout period (+5.5 mM). In the transverse colon, a significant increase in propionate production of 5.1 mM (corresponding to an increase of 5.9%) was measured, which was concomitant with strong lactate consumption in this colon compartment. Higher levels of propionate were also found in the descending colon compartment after AXOS treatment. Significant increases in butyrate production during and/or after AXOS treatment were observed in the two distal colon compartments (Fig. 1). In contrast to AXOS, inulin treatment caused a significantly higher net SCFA production in the ascending colon compartment, due to an increase in all SCFA. In this proximal colon part, inulin treatment significantly stimulated propionate production (+4.5 mM, corresponding to +3.8%) and the butyrate production was significantly higher after inulin treatment (+1.5 mM, corresponding to +3.5%). The lactate concentration was only significantly higher in the ascending colon during the washout period. Furthermore, no significant differences in succinate concentrations could be observed (data not shown).


Average net SFCA production during the Twin-SHIME run. For both inulin- (left panels) and AXOS (right panels)-treated SHIMEs, different colon compartments (CA, CT and CD for ascending, transverse and descending colon, respectively) and different periods (grey bars for basal, white bars for treatment and black bars for washout period) are presented. Indices above data bars represent significant differences within one colon compartment in function of time. Different letters indicate significantly different results (P<0.05).


Average net ammonium ion production (upper panels) and net lactate (lower panels) production during the Twin-SHIME run. For both inulin- (left panels) and AXOS (right panels)-treated SHIMEs, the different colon compartments (CA, CT and CD for ascending, transverse and descending colon, respectively) and the different periods (grey bars for basal, white bars for treatment and black bars for washout period) are presented. Indices above data bars represent significant differences within one colon compartment in function of time. Different letters indicate significantly different results (P<0.05).

The ammonium ion production, which is a marker for proteolytic activity of the microbial population, significantly decreased during AXOS treatment in the ascending colon compartment (−3.6 mM) (Fig. 2, upper panels). The increase in ammonium during AXOS treatment (+2.3 mM) in the transverse colon was due to a significantly higher ammonium production in the first treatment week. This increase rapidly disappeared once AXOS degradation began in the second treatment week. This is also consistent with the increase of AXOS-degrading enzymes during the second treatment week (Fig. 3). No differences in ammonium ion production was observed with inulin treatment.


Enzymatic activities during the Twin-SHIME run. AXON-degrading enzymes, such as xylanase (a), xylosidase (b), and arabinofuranosidase (c) were measured in the ascending (circle symbols), transverse (triangle symbols) and descending colon compartments (square symbols) for the AXOS-treated SHIME. The full-line arrow and dotted-line arrow indicate the beginning and end of the treatment period, respectively. Different letters indicate significantly different results (P<0.05).

Analysis of the activity of AXOS-degrading enzymes in the different compartments of the twin-SHIME reactor indicated striking differences between the compartments of the same unit (Fig. 3). In the ascending compartment of the AXOS unit, no increase in xylosidase and arabinofuranosidase activities was noted as a result of AXOS treatment, and only a relatively weak transient increase of xylanase activity was observed. In marked contrast, AXOS treatment caused a strong increase in xylanase, xylosidase and arabinofuranosidase activity throughout the entire treatment period in both the transverse and descending colon compartments. This increase was reverted during the washout period, indicating that the presence of AXOS is required to induce these enzymes in the microbiota.

Composition of the microbial community

To evaluate whether the above differences in metabolic activities were due to shifts in the composition of the microbial community, plate counts, Q-PCR and DGGE analysis were performed. Table 1 provides an overview of the plate counts and Q-PCR results for each colon compartment and for the different treatments. Inulin treatment significantly increased the amount of total anaerobes in all colon vessels, and gave a significant increase of about 0.5 log CFU in bifidobacteria levels at the end of the SHIME-run. This was not the case for AXOS. After treatment with AXOS, lactic acid-producing bacteria levels, such as enterococci, decreased by 0.5–1.0 log CFU in all colon compartments and lactobacilli decreased by 0.5 log CFU in the last two colon compartments. Strikingly, typical arabinoxylans degraders and butyrate producers such as Roseburia spp. and Bacteroides/Prevotella/Porphyromonas spp. decreased strongly in the ascending colon compartment during and after AXOS treatment, with the lowest concentrations in the washout period (−1.4 log CFU, significant, and −2.0 log CFU compared with the basal period, for Roseburia and Bacteroides, respectively).

View this table:

Average plate count measurements and Q-PCR results (± SEM), expressed in log CFU mL−1, for the different microbial groups, oligosaccharides (inulin, AXOS), SHIME compartments and periods

Bacterial groupsPeriodInulinAXOS
Plate counts
Total aerobesBasal7.9 ± 0.127.5 ± 0.147.5 ± 0.138.2 ± 0.247.8 ± 0.317.6 ± 0.24
Treatment7.9 ± 0.097.6 ± 0.107.8 ± 0.278.4 ± 0.288.3 ± 0.338.0 ± 0.32
Washout8.2 ± 0.197.7 ± 0.137.7 ± 0.057.8 ± 0.107.7 ± 0.107.5 ± 0.06
Total anaerobesBasal8.1 ± 0.11a7.8 ± 0.12a7.6 ± 0.15a8.2 ± 0.11a8.0 ± 0.097.6 ± 0.07
Treatment8.5 ± 0.13b8.0 ± 0.16ab7.7 ± 0.12ab8.3 ± 0.24ab8.3 ± 0.318.1 ± 0.31
Washout8.9 ± 0.07c8.5 ± 0.19b8.2 ± 0.17b7.7 ± 0.14b8.3 ± 0.388.2 ± 0.36
ClostridiaBasal6.9 ± 0.466.7 ± 0.446.6 ± 0.477.3 ± 0.287.2 ± 0.267.0 ± 0.30
Treatment7.5 ± 0.216.9 ± 0.266.7 ± 0.367.3 ± 0.517.4 ± 0.616.7 ± 0.54
Washout5.8 ± 0.976.2 ± 0.646.2 ± 0.718.9*6.6 ± 1.716.6 ± 0.77
BifidobacteriaBasal6.7 ± 0.17a6.4 ± 0.236.3 ± 0.287.1 ± 0.096.8 ± 0.166.5 ± 0.16
Treatment7.1 ± 0.18ab6.7 ± 0.096.5 ± 0.126.6 ± 0.216.7 ± 0.176.4 ± 0.18
Washout7.3 ± 0.19b6.8 ± 0.106.8 ± 0.196.6 ± 0.186.4 ± 0.146.2 ± 0.07
EnterococciBasal6.8 ± 0.056.4 ± 0.276.4 ± 0.156.6 ± 0.056.4 ± 0.16a6.2 ± 0.09a
Treatment6.9 ± 0.096.6 ± 0.116.4 ± 0.096.3 ± 0.316.3 ± 0.07a6.0 ± 0.08a
Washout6.6 ± 0.336.7 ± 0.106.6 ± 0.085.3 ± 0.535.7 ± 0.09b5.5 ± 0.14b
StaphylococciBasal7.0 ± 0.076.4 ± 0.266.5 ± 0.127.2 ± 0.156.6 ± 0.226.3 ± 0.18
Treatment7.0 ± 0.106.7 ± 0.046.3 ± 0.106.9 ± 0.196.8 ± 0.176.1 ± 0.19
Washout7.0 ± 0.116.7 ± 0.176.6 ± 0.146.4 ± 0.436.3 ± 0.255.8 ± 0.39
LactobacilliBasal4.3 ± 0.043.7 ± 0.363.8 ± 0.313.9 ± 0.173.6 ± 0.45a4.1 ± 0.15a
Treatment4.2 ± 0.113.4 ± 0.143.7 ± 0.104.5 ± 0.113.5 ± 0.12a3.7 ± 0.07b
Washout4.0 ± 0.153.4 ± 0.173.7 ± 0.094.4 ± 0.192.9 ± 0.11b3.5 ± 0.12b
Roseburia spp.Basal10.6 ± 0.0210.2 ± 0.019.8 ± 0.0310.3 ± 0.03a9.8 ± 0.19ab9.4 ± 0.14
Treatment10.7 ± 0.1010.6 ± 0.109.8 ± 0.099.4 ± 0.45ab9.9 ± 0.60a9.4 ± 0.56
Washout10.4 ± 0.1310.1 ± 0.349.6 ± 0.218.9 ± 0.02b8.2 ± 0.05b7.8 ± 0.26
Bacteroides/Prevotella/PorphyromonasBasal11.7 ± 0.1211.5 ± 0.1311.6 ± 0.0111.7 ± 0.3010.5 ± 0.8811.3 ± 0.11
Treatment11.7 ± 0.1211.9 ± 0.2911.4 ± 0.1011.2 ± 0.3011.2 ± 0.5111.7 ± 0.04
Washout11.9 ± 0.1110.8 ± 0.5711.3 ± 0.069.7 ± 0.4311.5 ± 0.2111.6 ± 0.02
Faecalibacterium prausnitziiBasal8.1 ± 0.128.0 ± 0.048.2 ± 0.188.4 ± 0.518.4 ± 0.487.9 ± 0.02
Treatment7.8 ± 0.318.0 ± 0.287.9 ± 0.176.9 ± 0.128.0 ± 0.077.6 ± 0.11
Washout8.8 ± 0.148.5 ± 0.268.1 ± 0.056.8 ± 0.117.8 ± 0.178.1 ± 0.15
  • * Only one measurement available.

  • Different letters indicate significantly different results (P<0.05).

Figure 4 presents the DGGE profiles for the different vessels and treatments. In general, the fingerprints of the ascending (CA) and the descending colon (CD) gave a clear clustering according to treatment (inulin vs. AXOS). Furthermore, in these colon compartments, a clear separation could be made between the fingerprints at the start and at the end of the SHIME-run. This was less visible in the transverse colon (CT). In the ascending colon, the microbial communities of both inulin- and AXOS-treated SHIME units initially clustered together in the basal period. However, as soon as the treatment started, the inulin- and AXOS-treated communities clustered separately. Sequencing of some bands of the general DGGE to specify the affected bacteria was unsuccessful. Yet, sequencing of a prominent band in the specific nested PCR-DGGE for bifidobacteria (data not shown) revealed the appearance of Bifidobacterium bifidum (98% similarity, 123 out of 125 bp) in the descending colon after 1 week of inulin treatment, and in all colon compartments during the washout period. This confirms earlier observations with this compound and points to prebiotic effect (Van de Wiele et al., 2004). Furthermore, an ecological interpretation of the fingerprints was conducted and the data were plotted (Fig. 5) according to Marzorati et al. (2008). The Rr varied between 44 and 127, and was thus classified as high (Rr>30). No trend in Rr variation was visible before, during or after treatment. Calculation of the Fo parameter showed that, on average, all the microbial communities analyzed had a moderate level of functional organization. This situation ensures a sufficient distribution of the digestive functions over a wide range of bacterial species and the capability to deal with sudden stress situations. Yet, during inulin treatment, the Dy index moved to high values. This was especially apparent from the transverse colon compartment. This indicated that inulin affects the composition of the microbial community more than does AXOS.


Clustering of the DGGE profiles during the Twin-SHIME run. The clustering was based on the Pearson correlation coefficient. CA, CT and CD refer to the ascending transverse and descending colon compartments, respectively. IN and AX in the names of the lanes refer to the inulin and AXOS treatment, respectively; the letters B, T and W refer to the basal, treatment and washout periods, respectively; and the numbers 1–7 refer to the corresponding week of the SHIME-run. Different letters indicate significantly different results (P<0.05).


2D-Plot of the functional organization (Fo) in function of the dynamics (Dy), as proposed by Marzorati et al. (2008). In this graph, the indication of the range-weighted richness was not included, because all the values were classified as high. Instead, indications of the colon vessels (dark squares, light squares and white triangles for the ascending, transverse and descending colon, respectively) are presented. The letters B, T and W refer to the basal, treatment and washout period, respectively. Different letters indicate significantly different results (P<0.05).


The Twin-SHIME experiment revealed striking differences between inulin and AXOS in terms of their effect on the metabolism of the intestinal microbiota and the colon region in which the effects were most prominent. A first important observation was that inulin elicited prebiotic effects primarily in the ascending colon, whereas AXOS breakdown and concomitant SCFA production took place in the more distal colon regions. The lack of significant activity of AXOS-degrading enzymes in the ascending colon vessel of the AXOS unit suggests that little AXOS degradation takes place in the proximal part of the colon. In contrast, the drastic increase in xylanase, xylosidase and arabinofuranosidase activities in the transverse colon clearly demonstrated that AXOS degradation and fermentation takes place in the more distal parts of the colon. In the descending colon vessel, there was a boost in AXOS-degrading enzyme activities during treatment, which, however, was not accompanied by increased SCFA production. This may indicate that activation of AXOS-degrading enzymes or their corresponding genes in the microbiota of the transverse colon compartment persisted during transfer of these bacteria to the descending colon compartment, or that the AXOS was already consumed before entering the descending colon.

The lack of activities of arabinoxylans-degrading enzymes in the ascending colon does not correspond to earlier observations from Vardakou et al. (2007). These authors reported an immediate increase in xylanase activity in the ascending colon upon arabinoxylans supplementation to a comparable simulator of the gastrointestinal tract. One explanation may be in the presence of glucose in the SHIME feed and the absence of glucose in the growth medium of the Vardakou et al. (2007) study. Glucose can act as a repressor to the production of arabinoxylans-degrading enzymes (Kulkarni et al., 1999), thus hindering the breakdown of AXOS, whereas the glucose-free medium used by Vardakou et al. (2007) did not inhibit xylanase production. This glucose repressive effect toward AXOS breakdown can also be expected under in vivo conditions. It has been shown that 10% of dietary starch escapes enzymatic degradation in the upper intestinal tract and enters the proximal colon region where it can be hydrolyzed and further fermented by many bacteria including Lactobacillus, Eubacterium, Bacteroides, Bifidobacterium and Escherichia species, thereby providing a source of glucose to the ascending colon (Baghurst et al., 1996; Bird et al., 2000). Although inulinase activity can also be inhibited to some extent by glucose (Beluche et al., 1980; Zhang et al., 2005), previous SHIME-runs with similar compounds always showed inulin fermentation in the ascending colon (Van de Wiele et al., 2004, 2007). Hence, repression of the enzymatic activity of AXOS-degrading enzymes may only partially explain the striking difference between fermentation of inulin and AXOS in the ascending colon compartment of this study.

Production of SCFA is generally considered as beneficial to the host because they protect the host against pathogens (Gaschott et al., 2001), induce immune responses (Saemann et al., 2000), reduce cholesterol synthesis (Berggren et al., 1996), stimulate colonic blood flow, enhance muscular contractions (Cummings et al., 1995), and may protect the colon against colon cancer development (Velasquez et al., 1997). The location of SCFA production is generally the proximal colon, the site of sugar fermentation (Gibson & Rastall, 2006). As colon cancer mainly takes place in the distal parts of the colon because of higher concentrations of more hazardous compound due to proteolysis and a higher pH, it is often the goal to extend sugar fermentation toward the distal parts of the colon (Jackson-Thompson et al., 2006; Ahmed et al., 2007). As the primary site of inulin fermentation is the ascending colon and that of AXOS fermentation is the transverse colon, we derive from these observations that inulin and AXOS are fully complementary with respect to their spatial pattern of fermentation in the colon.

The second important observation was the difference in fermentation profile between the inulin and AXOS-treated SHIME units. Similar SCFA profiles for inulin treatment were already reported before (Van de Wiele et al., 2004, 2007). AXOS treatment in particular resulted in a strong propionate production in the transverse colon. This propionate originates from two sources. Firstly, we noticed a strong lactate production in the ascending colon compartment, while lactate was consumed in the transverse colon. As one of the microbial lactate conversion pathways results in the production of propionate (Belenguer et al., 2007), we infer that lactate is converted to propionate by the acrylate pathway, carried out by bacteria belonging to clostridial cluster IX (Belenguer et al., 2007). Unfortunately, we were unsuccessful in analyzing the colon suspension for clostridial cluster IX with Q-PCR. The lactate presumably originates from feed components other than AXOS, as AXOS-degrading enzymes were still repressed in the ascending colon (Fig. 3). Secondly, the increase in propionate in the transverse colon (5.5 mmol L−1) exceeded the decrease in lactate (1 mmol L−1). This leads us to conclude that AXOS fermentation as such is the predominant contributor to propionate production in the transverse colon. We found no significant differences in succinate, an important intermediate in the alternative succinate pathway for propionate production. Therefore, a kinetic analysis of the fermentation process in the transverse colon compartment may reveal over which pathway this additional propionate is formed. When comparing these observations with literature, very few data are available for arabinoxylans oligosaccharides, but much more is known about the effects of the polymeric precursor. Stimulation of propionate production by long-chain arabinoxylans was already reported before (Amrein et al., 2003). In vivo studies with rats, fed an arabinoxylans-enriched diet, revealed a higher propionate production and a decreased cholesterol absorption (Lopez et al., 1999). Propionate indeed affects lipid and cholesterol metabolism (Todesco et al., 1991; Lin et al., 1995; Berggren et al., 1996). Therefore, Delzenne & Kok (2001) proposed that propionate production, besides other mechanisms, might explain the beneficial effect of fructans on fat metabolism in mice. Because our results indicate a stronger propionate-stimulating effect from AXOS over inulin, we conclude that AXOS may be a well-suited candidate to explore in vivo its modulating effects toward fat metabolism and cholesterol production in the host.

To investigate whether the pronounced differences in the level and location of the metabolic activity for both substrates could be attributed to shifts in the composition of the microbial ecosystem, plate counts and DGGE profiles were investigated. As expected, inulin treatment resulted in higher cultivable Bifidobacterium sp. levels, more specifically B. bifidum, which is consistent with previous results (Van de Wiele et al., 2004). The higher propionate and butyrate concentrations can be explained by cooperation of bifidobacteria with acetate- and lactate-converting bacteria (Falony et al., 2006). No report of endo-1,4-β-xylanase activity has been made for bifidobacteria, although they dispose of β-xylosidases, α-l-arabinofuranosidases (Shin et al., 2003), arabinoxylans arabinofuranohydrolase-D3 (Van Laere et al., 1999; van den Broek et al., 2005) and exo-oligoxylanases (Lagaert et al., 2007). All these enzymes detach xylose and arabinose units from the ends of the chains. Therefore, bifidobacteria are more likely to grow on short-chain AXOS or xylo-oligosaccharides fragments. Growth of bifidobacteria in vitro on short-chain xylo-oligosaccharides and AXOS has previously been demonstrated (Okazaki et al., 1990; Van Laere et al., 1999; Kontula et al., 2000). Unlike our results, where no change in the amount of bifidobacteria was seen, Vardakou et al. (2007) observed a significant increase in bifidobacteria in xylanase-pretreated arabinoxylans. This may be due to a different content in xylo-oligosaccharides – a strong bifidogenic prebiotic (Hsu et al., 2004) – between AXOS in this study and the xylanase-pretreated arabinoxylans from Vardakou et al. (2007). Furthermore, AXOS treatment strongly decreased Roseburia and Bacteroides/Prevotella/Porphyromonas levels in the ascending colon compartment, which are typical arabinoxylan degraders and butyrate producers (Chassard et al., 2007, 2008). Together with the strong decrease in butyrate and the low activity of arabinoxylans-degrading enzymes in this particular compartment, these data confirm that AXOS is probably not consumed in the ascending colon.

Interpretation of the DGGE profiles according to Marzorati et al. (2008), indicated that the Rr is high, meaning that the SHIME compartments are all rich in bacterial species and functionally well organized. Changing the substrate had a pronounced effect on the dynamics of the system in case of inulin, but not at all in case of AXOS supplementation. This indicates that inulin is indeed specifically stimulating certain groups of bacteria, resulting in metabolic changes, whereas AXOS mainly changes the metabolic activity, for example, by ‘switching on’ the AXOS-degrading enzymes. Neither of the treatments resulted in strong disturbances of the microbial communities, thus ensuring functional stability over time.

The aspect of reproducibility of the SHIME runs warrants attention. When considering the clustering of the microbial communities during the basal periods of the Twin SHIME units, it was seen that both the ascending and descending colon compartments had a similarity >80%, which is a desirable characteristic for a stable SHIME (Possemiers et al., 2004). The latter was not the case for the transverse colon compartments. Therefore, a cross-over experiment may be advisable for future experiments. Yet, data analysis of the metabolic characteristics of both SHIME units showed no significant differences at the end of the basal period revealed. Hence, the microbial functionality was the same and the observed effects can be assumed to be significant.

From this study it can be concluded that compared with the reference product inulin, AXOS had a higher potential to shift a part of the beneficial sugar fermentation toward the distal colon parts. Furthermore, due to its strong propionate-stimulating effect, AXOS might be a candidate prebiotic to lower cholesterol levels and beneficially affect fat metabolism in the host. Finally, because of the different prebiotic characteristics of AXOS compared with inulin, it would be interesting to test whether a combined product containing both AXOS and inulin could provide beneficial health effects throughout the entire colon.


We thank Ellen Van Gysegem for the technical assistance, and Ellen Eeckhaut, Selin Bolca and Sam Possemiers for critical revision of this manuscript. We thank Chantal Bridonneau and Philippe Langella (INRA, Jouys-en-Josas, France) for providing F. prausnitzii and Isabelle François (Fugeia N.V., Heverlee, Belgium) for Roseburia DNA. We also thank the Instituut voor de aanmoediging van innovatie door Wetenschap en Technologie (IWT) for funding the SBO project Impaxos and for providing financial support to C.G. T.V.d.W. is a Postdoctoral Fellow of the Fund for Scientific Research (FWO) – Flanders (Belgium).


  • Editor: Julian Marchesi


View Abstract