R.A. Rastall Food Microbial Sciences Unit, School of Food Biosciences, The University of Reading, PO Box 226, Whiteknights, Reading RG6 6AP, UK (e-mail: R.A.Rastall@afnovell.reading.ac.uk).
Aims: Comparison of in vitro fermentation properties of commercial prebiotic oligosaccharides.
Methods and Results: Populations of predominant gut bacterial groups were monitored over 24 h of batch culture through fluorescent in-situ hybridization. Short-chain fatty acid and gas production were also measured. All prebiotics increased the numbers of bifidobacteria and most decreased clostridia. Xylo-oligosaccharides and lactulose produced the highest increases in numbers of bifidobacteria whilst fructo-oligosaccharides produced the highest populations of lactobacilli. Galacto-oligosaccharides (GOS) resulted in the largest decreases in numbers of clostridia. Short-chain fatty acid generation was highest on lactulose and GOS. Gas production was lowest on isomalto-oligosaccharides and highest on inulin.
Conclusions: The oligosaccharides differed in their fermentation characteristics. Isomalto-oligosaccharides and GOS were effective at increasing numbers of bifidobacteria and lactate whilst generating the least gas.
Significance and Impact of the Study: The study provides comparative data on the properties of commercial prebiotics, allowing targeting of dietary intervention for particular applications and blending of oligosaccharides to enhance overall functionality.
There is currently much interest in the concept of actively managing the colonic microflora with the aim of improving host health. This is traditionally attempted by the consumption of live microbial food supplements, known as ‘probiotics’. An alternative approach is the consumption of food ingredients known as prebiotics. A prebiotic is defined as ‘a non-digestible food ingredient that beneficially affects the host by selectively stimulating the growth and/or activity of one or a limited number of bacteria in the colon, and thus improves host health’ (Gibson and Roberfroid 1995). The usual target species for such a dietary intervention are bifidobacteria and lactobacilli. These organisms have been reported to have health-promoting properties such as inhibition of exogenous pathogens. Many studies have now confirmed that prebiotics are a valid approach to the dietary manipulation of the colonic microflora (Gibson et al. 1995; Bouhnik et al. 1997; Kleesen et al. 1997).
In addition to the desirable effect of increased bifidobacteria and lactobacilli, short-chain fatty acids (SCFA) are produced as the end products of oligosaccharide fermentation. The profile of such SCFA varies between oligosaccharides and contains greater or lesser quantities of propionate and butyrate, both SCFA with positive implications in the prevention of colon cancer (Cummings 1981). Regular ingestion of prebiotic oligosaccharides can, however, sometimes give rise to excessive gas production, which is an undesirable side-effect. Although a substrate may be prebiotic in that it is selectively metabolized by bifidobacteria and lactobacilli, it may also be fermentable to a lesser degree by gas-producing organisms such as clostridia (Stone-Dorshow and Levitt 1987). It is, therefore, desirable that the prebiotic is fermented by bifidobacteria and lactobacilli with a very high degree of selectivity.
Several oligosaccharides have been reported to possess prebiotic activity (Hayakawa et al. 1990; Okazaki et al. 1990; Kohmoto et al. 1991; Kumemura et al. 1992; Gibson et al. 1995; Bouhnik et al. 1997; Kleesen et al. 1997) although the data often arise from single studies where the test oligosaccharide is compared with a placebo in either in vitro studies or human trials. It is difficult to compare the efficacy of different oligosaccharides as prebiotics because the studies vary greatly in terms of methodology, form of substrate, dose, duration, number of subjects and measurements taken. There is, therefore, a need for comparative studies of the activity of prebiotic oligosaccharides using a standard protocol.
There are a number of methods currently in use to determine the prebiotic properties of a test substrate, from pure culture studies with a range of intestinal bacteria, fermentation vessels containing mixed faecal cultures, to animal models such as human flora-associated (HFA) rats and human trials. For a rapid comparative evaluation, batch fermentation vessels inoculated with faecal slurries are useful as they represent the diverse gut microflora. In addition, several can be set up simultaneously, they can be used on a small scale for screening novel substrates which are only available in small quantities and they are completed within 24 h. These have been used previously in studies of oligosaccharides (Wang and Gibson 1993; Michel et al. 1998; Olano-Martin et al. 2000). Until recently, growth in such mixed fermentations was measured by colony counts on selective agars. This approach, however, suffers from several serious drawbacks, specifically, it is time-consuming, labour intensive, of low resolution and unable to account for unculturable organisms (Liesack and Stackebrandt 1992). As a result, molecular techniques such as fluorescent in situ hybridization (FISH) have been developed whereby genus-specific fluorescently labelled oligonucleotide probes are hybridized to bacterial rRNA extracted from a faecal or a fermentation sample. The resultant fluorescently labelled cells are then enumerated by fluorescence microscopy (Langendijk et al. 1995; O’Sullivan 1999).
The present study compared the in vitro fermentation of a range of prebiotics by faecal bacteria as determined by FISH. The batch culture method was adapted from Olano-Martin et al. (2000). The oligosaccharide was added at a concentration of 1% (w/v) and an initial pH of 7·0, which was not controlled, was used for the fermentations, as in similar previous studies (Michel et al. 1998; Olano-Martin et al. 2000). Samples were removed initially, after 5 h fermentation to determine the immediate effects of the oligosaccharides and after 24 h to observe the overall effect of the oligosaccharide on the bacterial composition (Wang and Gibson 1993; Michel et al. 1998; Olano-Martin et al. 2000). Gas and SCFA production were also determined.
MATERIALS AND METHODS
The following commercial preparations of prebiotic oligosaccharides were investigated: Raftilose P95 (Orafti, Tienen, Belgium), Raftiline LS (Orafti), lactulose (LACT; Sigma, Poole, UK), xylo-oligosaccharides (XOS; Suntory, Osaka, Japan), Oligomate 55 (Yakult, Tokyo, Japan) and soybean oligosaccharides (SOS; Calpis, Tokyo, Japan). Isomalto-oligosaccharides (IMO) were prepared by controlled hydrolysis of dextran as previously described (Mountzouris et al. 1998). The compositions of these products are listed in Table 1. All other chemicals were obtained from Sigma unless otherwise stated.
Table 1. Structure and composition of oligosaccharides used in the present study
Static batch culture fermentations
The method used was described by Olano-Martin et al. (2000). Each substrate (0·5 g) was weighed into autoclaved 50-ml serum bottles and 45 ml autoclaved nutrient medium was added by filter sterilization to each bottle. This medium contained (g l−1): peptone water, 2; yeast extract, 2; NaCl, 0·1; K2HPO4, 0·04; KH2PO4, 0·04; MgSO4.7H2O, 0·01; CaCl2.6H2O, 0·01; NaHCO3, 2; hemin (dissolved in a few drops of 1 mol l−1 NaOH), 0·05; cysteine.HCl, 0·5; bile salts, 0·5; Tween 80, 2 and 10 μl vitamin K1. The medium was adjusted to pH 7·0 using 1 mol l−1 HCl. The bottles containing the hot medium were placed in an anaerobic cabinet (10% H2 : 10% CO2 : 80% N2; Don Whitley Scientific, Shipley, West Yorkshire, UK) at 37°C overnight to prereduce the media. A 10% (w/v) faecal slurry was prepared using fresh faeces from a healthy donor (who had not taken antibiotics for 3 months beforehand) and prereduced 0·1 mol l−1 phosphate buffer (pH 7·0) and this was mixed in a stomacher for 2 min. The serum bottles were inoculated with 5 ml of the slurry, mixed and capped and sterile needles placed in the caps to act as vents during fermentation. Samples were removed from the fermenters at 0, 5 and 24 h fermentation for enumeration of bacteria and SCFA analysis. The fermentation experiments were performed in triplicate for each substrate using faecal inoculum from three volunteers.
Oligonucleotide probes for fluorescent in situ hybridization
Genus-specific 16S rRNA-targeted oligonucleotide probes labelled with the fluorescent dye Cy 3, that had previously been designed and validated, were used for enumerating Bifidobacterium (Bif 164; Langendijk et al. 1995), Bacteroides (Bac 303; Manz et al. 1991), Lactobacillus/Enterococcus (Laboratory 158; Harmsen et al. 1999), Clostridium subgrp. histolyticum (His 150; Franks et al. 1998), Streptococcus (Str 493; Franks et al. 1998) and Escherichia coli (EC 1531; Poulsen et al. 1995). To obtain total cell counts, the nucleic acid stain 4′, 6-diamidino-2-phenylindole (DAPI) was added to each sample. Cells were fixed on a slide and those stained with DAPI or hybridized with probe were enumerated under a microscope as described below.
Fluorescent in situ hybridization
Static batch culture fermentations were set up as described above and samples removed for bacterial enumeration by FISH. At 0, 5 and 24 h fermentation, samples (375 μl) were removed from the batch cultures and added to 1·125 ml filtered 4% (w/v) paraformaldehyde solution (pH 7·2), mixed and stored at 4°C overnight to fix the cells. The fixed cells were washed twice in filtered 0·1 mol l−1 phosphate buffer solution (pH 7·0) and resuspended in 150 μl phosphate buffer solution (0·1 mol l−1, pH 7·0). Ethanol (150 μl) was added and the sample mixed and stored at −20°C until needed, but no longer than 3 months. The fixed cells (16 μl) were added to 64 μl prewarmed high performance liquid chromatography (HPLC) grade water and 200 μl prewarmed filtered double-strength hybridization buffer (1·8 mol l−1 NaCl, 40 mmol l−1 Tris-HCl, pH 7·2, 0·2% sodium dodecyl sulphate (w/v)) and mixed. This mixture was added to each probe (50 ng μl−1), in a ratio of 9 : 1 (v/v), mixed and placed in the hybridization oven at the appropriate temperature overnight (Bif 164, His 150 and Str 493 probes, 50°C; Bac 303 and Laboratory 158, 45°C and EC 1531, 37°C).
The hybridized sample (5–50 μl) was washed in 5 ml prewarmed, filtered, single-strength hybridization buffer (0·9 mol l−1 NaCl, 20 mmol l−1 Tris-HCl, pH 7·2) containing 20 μl DAPI solution (500 ng μl−1) for 30 min at the appropriate hybridization temperature. The sample was filtered onto a 0·2-μm pore size filter (Millipore, Watford, UK) and the filter mounted in ‘Slow fade’ (Molecular Probes, Leiden, The Netherlands) on a clean microscope slide. Cells were counted using a microscope (Leica, Wetzlar, Germany) fitted with appropriate filters for the DAPI stain (excited at 359 nm and emits at 461 nm) and the Cy3 dye (excited at 550 nm and emits at 565 nm). A minimum of 15 fields, each containing 10–100 cells, was counted for each slide.
Short-chain fatty acid analysis
The method was carried out as described by Olano-Martin et al. (2000). Samples taken from the fermenters were centrifuged (13 000 g for 15 min) to remove particulate material and 20 μl injected onto an HPLC system (Model 1350T; Bio-Rad, Hemel Hempstead, UK) attached to a u.v. detector (Knauer, Type 298.00; LC Services, Wootton, Bedfordshire, UK) at 210 nm. The column was an ion-exclusion Aminex HPX-87H (7·8 × 300 mm; Bio-Rad) maintained at 50°C with a column heater (Model 1250426; Bio-Rad). The eluent, 0·005 mol l−1 sulphuric acid in HPLC-grade water, was pumped through the column at a flow rate of 0·6 ml min−1. Data from the u.v. detector were integrated using the ‘ValueChrom™’ software package (Bio-Rad). Using external calibration curves, acetate, propionate, butyrate and lactate were quantified in the samples.
Gas production by faecal bacteria in batch fermentations
Batch culture fermenters were set up as previously described using the carbon sources already listed. These were sealed with aluminium caps and incubated at 37°C. The volume of gas produced was measured after 24 h fermentation by inserting a gas-impermeable syringe fitted with a sterile needle into the rubber cap of each vessel. This allowed the pressure build-up in the headspace to push the syringe barrel so that a volume could be recorded. The gas production experiments were performed in triplicate for each substrate using faecal samples from different volunteers.
Differences between bacterial counts and SCFA concentrations at 0, 5 and 24 h fermentation for each substrate were tested for significance using paired t-tests assuming equal variances and considering both sides of the distribution (two-tailed distribution). Differences were considered significant if P ≤ 0·05.
The changes in bacterial counts and SCFA concentrations were also compared between the substrates using paired t-tests to determine whether there were any significant differences in the effects of the substrates.
The recorded changes in selected bacterial populations with the prebiotics tested are presented in Table 2. With all of the prebiotics studied, a large significant increase in numbers of bifidobacteria was observed, although only inulin and XOS significantly increased total bacterial numbers (Table 2). Changes in numbers of lactobacilli were, however, less pronounced and more variable. The effect of the different prebiotics on clostridia was also very variable as fructo-oligosaccharides (FOS) and XOS caused a significant increase in numbers after 5 h while, after 24 h, large decreases were observed in some cases. Populations of Bacteroides showed very similar changes with all of the oligosaccharides, with a significant increase in numbers after 5 h fermentation. Escherichia coli and streptococci showed little change with most substrates although inulin caused relatively large significant increases in both groups, which were maintained by 24 h. Fructo-oligosaccharides and XOS caused initial significant increases in these groups but the numbers were not maintained.
Table 2. Bacterial populations* in static batch culture fermentations with the various prebiotics
The significant differences between the substrates in terms of their effects on bacterial populations are recorded in Table 3. Initially, inulin caused a significantly larger increase in the total count than most of the other substrates as well as a larger increase in numbers of streptococci than lactulose and SOS. Also, between 0 and 24 h, inulin gave a significantly smaller increase in numbers of bifidobacteria than IMO or SOS lactulose, while FOS showed a significantly larger increase in numbers of streptococci than galacto-oligosaccharides (GOS) or SOS (Table 3).
Table 3. Significant differences between substrates on basis of changes in bacterial populations between time points
The levels of SCFA measured in the fermentations are presented in Table 4. Significant increases in lactate occurred by 24 h with all substrates except inulin, while all of the substrates effected a significant increase in acetate. Propionate concentrations were significantly increased with all substrates except IMO, with inulin and lactulose generating the highest concentrations. Although none of the substrates caused significant increases in butyrate, the highest concentrations were recorded on FOS and inulin as previously reported in studies with rats (Campbell et al. 1997; Djouzi and Andrieux 1997) and humans (Gibson et al. 1995). Negative controls, i.e. batch cultures containing no additional carbohydrate, have been reported (Michel et al. 1998) to produce minimal total SCFA concentrations (5 mmol l−1) after 24 h fermentation. This study indicated a lack of fermentation when no carbohydrate was added to faecal slurries, although bacterial populations were not monitored.
Table 4. Short-chain fatty acid and lactate production by prebiotic fermentations*
Significant differences between the substrates in terms of their effects on SCFA and lactate concentrations are recorded in Table 5. Initially, the rate of lactate production divided the substrates into three separate groups (Table 5): FOS and inulin which produced the lowest lactate concentrations; XOS, lactulose and IMO which produced slightly larger amounts of lactate and GOS and SOS which produced significantly larger lactate concentrations (Table 5). Similar differences were also recorded between the substrates from 5 to 24 h fermentation. Fructo-oligosaccharide, inulin and IMO tended to effect smaller increases while lactulose gave a significantly larger increase in lactate than the other substrates. Overall, between 0 and 24 h fermentation, FOS and IMO produced significantly lower lactate concentrations than the other substrates, while lactulose produced a significantly larger increase in lactate (Table 5). Acetate production showed fewer significant differences between substrates, as only inulin produced significantly smaller acetate concentrations than lactulose, GOS and SOS between 5 and 24 h fermentation (Table 5). Substrates also differed in terms of propionate production. Initially, FOS and inulin produced significantly lower concentrations of propionate than GOS and SOS whereas, between 5 and 24 h fermentation, FOS and lactulose fermentations had a significantly higher rate of propionate production than GOS and SOS (Table 5).
Table 5. Significant differences between substrates on basis of changes in short-chain fatty acid and lactate concentrations between time points
Gas evolution data are presented in Fig. 1. The lowest gas production was recorded with IMO, followed by GOS and then SOS. Inulin, lactulose, FOS and XOS produced the highest levels of gas.
The view taken in this study is that, whilst it is not feasible to fully characterize the changes occurring in the colonic microflora, it is possible to monitor the populations of selected species believed to be indicative of the state of health of the colon. It must, of course, be recognized that our knowledge of this ecosystem is far from complete (Liesack and Stackebrandt 1992).
From the data presented in Table 2, it is possible to identify the best prebiotic for a desired microflora change. If a maximal increase in numbers of bifidobacteria is desired, then XOS or lactulose would appear to be the best carbon sources, maximal increases in numbers of lactobacilli were recorded on FOS (P=0·08), whereas the maximal decrease in numbers of clostridia was recorded on GOS, although the change was not significant.
The production of SCFA and gas is an ensemble product of the activities of the total microflora present in the fermentations, much of which has not been characterized and only selected genera were monitored in this study. It has also been reported that SCFA are only produced in appreciable levels in the presence of added carbohydrate, a situation that would be expected to pertain in the gut (Wang and Gibson 1993). However, there are some correlations between production of metabolites and the changes seen in the microflora. For example, in the cases of IMO and inulin, an inverse relationship existed between an increase in the bifidobacterial population and production of gas.
In a similar batch culture study (Olano-Martin et al. 2000), glucose, used as a non-prebiotic control, was rapidly fermented within 6 h and increased the total bacterial count and bifidobacterial population. The total count remained at 7·2 log cfu ml−1 while counts of bifidobacteria decreased to 4·8 log cfu ml−1 between 6 and 24 h and immediate decreases were observed in the bacteroides and clostridia groups (Olano-Martin et al. 2000). These data suggested that glucose could not support bifidobacteria, unlike the substrates tested in the present study (Table 2), and caused increases in other bacterial groups that were not enumerated. Glucose fermentation produced no propionate or butyrate although high concentrations of lactate (44 mmol l−1) and acetate (55 mmol l−1) were produced (Olano-Martin et al. 2000). These may have resulted from lactic acid fermentation from bacteria other than bifidobacteria, such as streptococci, which were not enumerated.
Results from the present study allow comparisons of the properties of the prebiotics in terms of bacterial growth and metabolite production. In general, galactose-containing oligosaccharides, namely lactulose, GOS and SOS, were ostensibly more effective than the fructose-containing inulin and FOS in terms of increasing numbers of bifidobacteria, lactate production and generating low gas volumes. This is supported by the statistical comparisons between the substrates in terms of changes in bacterial populations (Table 3) and SCFA concentrations (Table 5) It must be borne in mind, however, that gas evolution in vivo will undoubtedly be linked to dosage. This has previously been suggested in a comparative pure culture study (Jaskari et al. 1998) where galactose-containing raffinose was more stimulatory to bifidobacteria than FOS. Also, in a study with HFA rats (Djouzi and Andrieux 1997), GOS increased bifidobacteria populations to a greater extent than FOS. None of the commercial oligosaccharides are pure, hence the data from in vitro models may be skewed by fermentable carbohydrate impurities that would not reach the colon. A human volunteer trial comparing these substrates would help to confirm this.
Previous comparative studies of prebiotics have compared two or three substrates and observed SCFA production and/or bacterial populations using selective plates. Comparative studies in vitro (Wolf et al. 1994) and in rats (Campbell et al. 1997) revealed a greater increase in numbers of bifidobacteria and more lactate production with XOS than FOS, as recorded in the present study. A batch culture study comparing FOS and inulin (Wang and Gibson 1993) recorded a much higher lactate production, greater selectivity for bifidobacteria and lactobacilli and lower gas production with FOS.
When all the data are considered, IMO and GOS appear to be effective prebiotics as they increased numbers of bifidobacteria with little effect on the other groups (Table 2), significantly increased lactate production (Table 4) and produced the lowest gas volumes (Fig. 1).
A clear correlation exists between the increase in bifidobacteria during 24 h batch fermentations and their initial population (Fig. 2). Figure 2 also illustrates the large variation between the three volunteers, relating to the large standard deviations (Table 2). The standard deviations of the mean bifidobacteria populations are greatest for the inoculum, smaller for the 5-h samples and smaller still for the 24-h samples for all the seven substrates. This suggests that the bifidobacterial population reaches a maximum after 24 h of fermentation irrespective of the initial population in the volunteer’s faecal sample. This phenomenon was not seen with any other bacterial group enumerated.
This in vitro phenomenon reflects the in vivo situation as previous human volunteer trials with a range of different oligosaccharides have shown the same inverse correlation (Fig. 3). This relationship has been observed previously for FOS (Hidaka et al. 1986; Roberfroid et al. 1998). Prebiotics may, therefore, be particularly useful for targeting certain groups of individuals, such as the elderly, who have diminished bifidobacteria populations. The comparative data presented here may also facilitate the rational blending of prebiotic oligosaccharides to extend their functionality as food ingredients.
The experiments carried out in this study, whilst allowing a preliminary comparison of the prebiotic properties of the selected oligosaccharides, will be extended by studying the fermentation of these molecules under more controlled conditions and in gut models. In addition, the physicochemical and organoleptic properties of these prebiotics will be compared. The oligosaccharides will ultimately be compared in a human volunteer trial using a large number of volunteers to generate a more definitive comparison.
The authors would like to acknowledge receipt of a BBSRC/Convatec CASE studentship for Catherine Rycroft.