Selection of a Bifidobacterium strain to complement resistant starch in a synbiotic yoghurt

Authors


Dr R.G. Crittenden VTT Biotechnology, PO Box 1500, FIN-02044 VTT, Finland (e-mail: ross.crittenden@vtt.fi).

Abstract

R.G. CRITTENDEN, L.F. MORRIS, M.L. HARVEY, L.T. TRAN, H.L. MITCHELL AND M.J. PLAYNE. 2001.

Aims: To employ an in vitro screening regime to select a probiotic Bifidobacterium strain to complement resistant starch (Hi-maize™) in a synbiotic yoghurt.

Methods and Results: Of 40 Bifidobacterium isolates examined, only B. lactis Lafti™ B94 possessed all of the required characteristics. This isolate hydrolysed Hi-maize™, survived well in conditions simulating passage through the gastrointestinal tract and possessed technological properties suitable for yoghurt manufacture. It grew well at temperatures up to 45°C, and grew to a high cell yield in an industrial growth medium. In addition to resistant starch, the organism was able to utilize a range of prebiotics including inulin, and fructo-, galacto-, soybean- and xylo-oligosaccharides. Pulse field gel electrophoresis of restriction enzyme cut chromosomal DNA revealed that B. lactis Lafti™ B94 was very closely related to the B. lactis Type Strain (DSM 10140), and to the commercial strains B. lactis Bb-12 and B. lactis DS 920. However, B. lactis Lafti™ B94 was the only one of these isolates that could hydrolyse Hi-maize™. This phenotypic difference did not appear to be due to the presence of plasmid encoded amylase. Bifidobacterium lactis Lafti™ B94 survived without substantial loss of viability in synbiotic yoghurt containing Hi-maize™ during storage at 4°C for six weeks.

Conclusions:Bifidobacterium lactis Lafti™ B94 is a promising new yoghurt culture that warrants further investigation to assess its probiotic potential.

Significance and Impact of the Study:In vitro screening procedures can be used to integrate complementary probiotic and prebiotic ingredients for new synbiotic functional food products.

INTRODUCTION

In addition to traditional starter cultures, probiotic bacteria are now often included in yoghurts with the aim of contributing to the health and well-being of consumers through the maintenance of an advantageous balance of intestinal bacterial populations. The bacteria used as probiotics are predominantly selected from the genera Lactobacillus and Bifidobacterium, both of which form part of the normal human intestinal microbiota (Mitsuoka 1978,1982; Tannock 1995).

A second approach to increasing the number of probiotic bacteria in the intestinal microbiota is through the use of prebiotics. Prebiotics are non-digestible dietary components that selectively stimulate the growth and/or activity of indigenous probiotic bacteria in the intestinal tract (Gibson and Roberfroid 1995). The prebiotics identified thus far have been carbohydrates such as lactulose, inulin, and various oligosaccharides (Crittenden 1999). Consumption of these non-digestible ingredients has been demonstrated to alter intestinal bacterial populations, in particular promoting the proliferation of bifidobacteria (Fuller and Gibson 1997; Roberfroid et al. 1998; Van Loo et al. 1999). Recent studies performed in vitro (Wang et al. 1999), in rodents (Kleessen et al. 1997; Brown et al. 1998) including rats associated with human gut microflora (Silvi et al. 1999), and in pigs (Brown et al. 1997), have indicated that resistant starch can also stimulate the proliferation of bifidobacteria in the intestinal tract, and may therefore have potential to act as a prebiotic in humans.

The obvious potential for synergy between probiotics and prebiotics has lead to the development of foods containing combinations of these ingredients. Such products have been generically termed ‘synbiotics’ (Roberfroid 1998; Ziemer and Gibson 1998). In the current investigation, a synbiotic yoghurt was envisaged that would contain resistant starch as the prebiotic ingredient, and a probiotic Bifidobacterium strain that could benefit in vivo from the inclusion of this fermentable substrate. The resistant starch to be used was Hi-maize™, a high-amylose maize starch that passes through to the colon relatively undigested due to its crystalline structure and granular conformation.

For probiotics to be effective in the colon, they must first remain viable both in the food product and during passage through the gastrointestinal tract. The acidic and protease-rich environment of the stomach, and the inhibitory effects of bile acids secreted into the duodenum, are possibly the major impediments to the survival of ingested bacteria during intestinal transit. A number of in vitro models have been used by researchers to predict the survivability of Bifidobacterium strains during intestinal transit (Clark et al. 1993; Clark and Martin 1994; Lankaputhra and Shah 1995; Marteau et al. 1997; Mustapha et al. 1997; Charteris et al. 1998). Additionally, in vitro characterization of the technological attributes of strains can be used to evaluate their potential to be successfully produced commercially and applied in food matrices. The current investigation details the selection, using in vitro models, of a Bifidobacterium isolate to complement Hi-maize™ resistant starch in a synbiotic yoghurt.

MATERIALS AND METHODS

Micro-organisms and culture conditions

Forty Bifidobacterium isolates were selected for screening on the basis that they were of human or dairy origin. These organisms and their origins are listed in Table 1. Additionally, B. breve LMG 10737, previously reported to contain a plasmid (Bourget et al. 1993), was used as a positive control organism in plasmid extraction experiments. The cultures were revived from glycerol stocks frozen at −80°C by subculturing twice into Reinforced Clostridial Medium broth (RCM; Amyl Media, Australia) using a 2% (v/v) inoculum. Unless otherwise stated, all cultures were grown on RCM, at 37°C, under an anaerobic atmosphere consisting of 10% (v/v) CO2, 10% (v/v) H2 and 80% (v/v) N2.

Table 1. Bifidobacterium isolates screened for the ability to hydrolyse Hi-maize™ and to grow well in Reinforced Clostridial Medium (RCM) broth Thumbnail image of

Strain identification

The bifidobacteria used in this investigation were identified at the genus level by assaying for the diagnostic enzyme fructose-6-phosphate phosphoketolase (Scardovi and Trovatelli 1965). The method used was as described by Biavati et al. (1992). Acid-tolerant isolates were compared at the strain level using pulse field gel electrophoresis of restriction enzyme fragmented chromosomal DNA (PFGE). High molecular weight chromosomal DNA was prepared in agarose blocks as per Hillier and Davidson (1995), with Bifidobacterium cultures grown anaerobically in 6 ml RCM. The method was modified in that no chloramphenicol was added, and the cells were harvested from the entire 6 ml sample rather than from just 1 ml. The DNA was cut using 100 U ml−1 of the restriction enzyme XbaI (Boehringer Mannheim), prepared as per the manufacturer’s instructions. Digested samples were electrophoresed through 1·2% (w/v) agarose gels in 0·5× TBE (45 mmol l−1 Tris–HCl, 45 mmol l−1 boric acid, 1 mmol l−1 EDTA, pH 8·0) using a contour clamped homogeneous electric field system (Chu et al. 1986) ‘CHEF-DRIII’ (Bio-Rad, Hercules, CA, USA). Samples were run at 6 V cm−1 for 22 h at 14°C, with an angle of 120° and ramped switch times from 1 to 20 s. Gels were stained for 45 min in distilled water containing 0·5 mg l−1 ethidium bromide, and de-stained overnight in distilled water (Sambrook et al. 1989).

Plasmid extraction

Plasmid DNA was extracted from Bifidobacterium isolates using the plasmid profiling technique of Tannock et al. (1990), with the exception of substituting Reinforced Clostridial Broth, modified to contain no agar, for the BBM broth. Electrophoresis was carried out with 1% agarose gels in 0·5× TBE. Samples were run at 3 V cm−1 for 1 h and then stained with ethidium bromide as described previously.

Hydrolysis of resistant starch

To test for the ability to hydrolyse resistant starch, the bifidobacteria were grown on agar plates containing Hi-maize™ as the sole carbon source. The growth medium contained 5 g l−1 Hi-maize™, 10 g l−1 beef extract, 3 g l−1 yeast extract, 10 g l−1 pancreatic digest of casein, 5 g l−1 NaCl and 10 g l−1 agar. The pH of the medium was 6·8. Bifidobacterium isolates were inoculated in triplicate onto the surface of the agar plates by adding 5 μl RCM broth cultures at single points. This produced bacterial colonies of uniform size. The plates were incubated anaerobically for 3 days at 37°C. Following incubation, each plate was flooded with a solution containing 10 g l−1 iodine plus 20 g l−1 potassium iodide. This solution stained the Hi-maize™ starch dark blue. A clear zone was evident around colonies able to hydrolyse the substrate.

In vitro gastric model

An in vitro model simulating the acidic and protease-rich environment of the stomach was used to estimate the ability of the bifidobacteria to survive upon ingestion. Bifidobacteria, grown in RCM until they were in stationary phase, were resuspended in 1 g l−1 bacteriological peptone. Equal numbers of bacteria were then added to both the in vitro gastric model and to a control tube at a final concentration of approximately 1 × 108 cfu ml−1. The solution in the control tubes consisted of 6 ml 0·1 mol l−1 phosphate buffer, pH 6·5, containing 1 g l−1 bacteriological peptone. The in vitro gastric model contained 6 ml simulated gastric juice consisting of 0·1 mol l−1 HCl/KCl buffer pH 2·0 containing 500 U ml−1 pepsin A and 1·0 g l−1 bacteriological peptone. The cells in both the gastric model and the control tube were incubated for 105 min at 37°C. Samples were taken from both the gastric model and the control tube at time zero and after 105 min, and the viable count of bacteria was measured by plating serial 10-fold dilutions, in duplicate, onto RCM agar plates.

Effect of bile on growth and survival

The effect of bile was investigated on both the growth and survival of bifidobacteria that were able to survive well in the gastric model. To simulate the sequence of environmental stresses the bacteria would face in the gastrointestinal tract, the bacteria were first treated in the in vitro gastric model described previously, followed immediately by exposure to a physiological concentration of bile. Once cells were treated in the in vitro gastric model, they were transferred to a 6 ml solution containing 3·0 g l−1 ox bile (Oxoid) and 1·0 g l−1 bacteriological peptone in 0·1 mol l−1 phosphate buffer, pH 6·5. The cell concentration in this solution was approximately 1 × 108 cfu ml−1, and the bacteria were incubated with the bile for 3 h at 37°C. An equal number of the bacteria were simultaneously incubated in a control tube containing 0·1 mol l−1 phosphate buffer, pH 6·5, plus 1 g l−1 bacteriological peptone and incubated at 37°C. Samples were taken from both the intestinal model and the control tube at time zero, after acid treatment and after exposure to bile. The viable count of bacteria in the samples was measured by plating serial 10-fold dilutions, in duplicate, onto RCM agar plates.

In addition to survival, the effect of bile on the growth rate and yield of bifidobacteria was investigated. The bacteria were grown at 37°C in both RCM broth and RCM broth containing 3·0 g l−1 ox bile. The growth curves of the bacteria in the presence and absence of bile were compared by measuring the optical density (O.D.) of the cultures at 600 nm. Additionally, the ability of the bacteria to grow at higher bile concentrations was examined. Bifidobacteria were grown in RCM broth containing ox bile at concentrations ranging from 0 to 100 g l−1 in 10 g l−1 increments. The bacteria were grown in 200 μl volumes in a 96-well microplate at 37°C. The O.D. of cultures at 600 nm was measured after 48 h.

Effect of temperature on growth

The effect of temperature on the growth rate and yield of bifidobacteria was investigated. Cultures were grown in RCM at 37, 40, 45 and 50°C in 9 ml culture tubes. The growth of the cultures was monitored by periodically measuring their O.D. at 600 nm.

Yield in industrial growth medium

Fermentations were conducted using a whey-based industrial growth medium (WMP–DSM Food Specialties Australia) to determine cell yield produced under commercial conditions. Anaerobic fermentations (1 litre) were conducted in duplicate. The temperature was maintained at 37°C and the pH was controlled at 6·0 using 26% (v/v) ammonia. A 2·5 cm magnetic stirring bar in each vessel, rotated at approximately 40 rev min−1, was used to provide homogeneous nutrient distribution and to assist pH control. Each fermenter was inoculated from a 24-h RCM broth culture (0·3% v/v inoculum). Samples were taken periodically during the fermentation and analysed for viable counts using duplicate RCM agar plates.

Growth on other prebiotic substrates

A broth medium base free of fermentable carbon sources was used to investigate the ability of the B. lactis B94 to grow on prebiotic oligosaccharides. The growth medium contained 10 g l−1 beef extract, 3 g l−1 yeast extract, 10 g l−1 pancreatic digest of casein and 5 g l−1 NaCl. The pH of the medium was 6·8. The tested carbohydrates were added to the medium at 10 g l−1. The prebiotics investigated were lactulose (Morinaga), inulin (Raftiline™, ORAFTI), fructo-oligosaccharides (Raftilose™, ORAFTI and NutraFlora™, Golden Technologies), soybean-oligosaccharides (raffinose and stachyose, Sigma), xylo-oligosaccharides (Xylo- oligo95, Suntory), isomalto-oligosaccharides (Isomalto-900, Showa-Sangyo) and galacto-oligosaccharides (Oligomate 50, Yakult). Additionally, growth of the bifidobacteria was examined on glucose, fructose, galactose, xylose, sucrose, maltose, cellobiose and lactose. The organism was grown in triplicate experiments in 9 ml culture tubes under anaerobic conditions at 37°C. The growth of the bacteria was monitored throughout the fermentations by measuring the culture O.D. at 600 nm.

Survival in yoghurt

The survival of B. lactis B94 in a synbiotic yoghurt was compared over a six-week shelf-life with a commercial B. lactis strain, DS 920, and a highly amylolytic isolate, B. adolescentis B97. A standard, reduced-fat yoghurt base was used containing 14% (w/v) milk solids non-fat (MSNF), 1·2% (w/v) fat and 6·0% (w/v) sucrose. Additionally, the yoghurt bases contained 1·0% (w/v) Hi-maize™ and 1·0% (w/v) inulin (Raftiline™). The yoghurt base mixes were heat-treated at 90°C for 10 min then cooled to 42°C before inoculation. The yoghurts were each inoculated with Streptococcus thermophilus DS 224 (DSM Food Specialties) and one of the Bifidobacterium isolates. All cultures were inoculated at approximately 1 × 108 cfu g−1 into 1·0 kg yoghurt batches.

Once inoculated, the yoghurts were incubated at 42°C, during which time the pH was monitored. The 42°C incubation was ceased when the pH dropped to 4·5. The yoghurts were then stored at 4°C for 5 weeks. Viable bifidobacteria in the yoghurts were enumerated weekly using RCM agar containing 0·3 g l−1 aniline blue (BDH) and 2·0 mg l−1 dicloxacillin (Sigma).

RESULTS

Hydrolysis of resistant starch (Hi-maize™) and growth yield in RCM

The initial selective screening tests employed were to determine whether the isolates could hydrolyse resistant starch and could grow well in RCM. Previous experience has shown that bifidobacteria that do not grow well on the nutrient rich RCM medium also fail to grow well in industrial whey-based media. Therefore, cultures failing to grow well in RCM (O.D. 600 nm cm −1 <0·5) were discarded. More than half of the Bifidobacterium isolates examined were able to hydrolyse Hi-maize™ (Table 1). Hi-maize™ hydrolysis was common to all of the B. adolescentis isolates tested but appeared to be strain-specific for B. bifidum, B. breve, B. infantis, B. longum and B. lactis. Eighteen of the 40 isolates were able both to hydrolyse the resistant starch and grow well in RCM. These isolates were selected for further characterization.

Survival in the in vitro gastric model

The 18 isolates that were able to both hydrolyse the resistant starch and grow to high yield were examined for their ability to survive in an in vitro model simulating the acidic and protease-rich environment of the human stomach. The commercially-used Bifidobacterium strains, and B. animalis NCFB 2716, were also included in the test for comparison. For all isolates, the viable count in the control remained constant throughout the period of the assay (data not shown).

Only the three B. lactis isolates and the B. animalis strain were able to survive well in the conditions simulating the human stomach (Fig. 1). Of the commercial strains, B. lactis DS 920 and B. lactis Bb-12 were both acid and protease tolerant, but the B. longum strain CSCC 5554 did not survive well in these conditions. Only one Hi-maize™ hydrolysing isolate, B. lactis B94, demonstrated no significant loss of viability in the in vitro model. It appears that B. lactis and B. animalis are considerably more acid tolerant than the other species examined.

Figure . 1.

Survival of Bifidobacterium isolates in an in vitro model simulating conditions in the human stomach. Cells were incubated for 105 min at 37 °C in 0·1 mol l−1 HCl/KCl buffer, pH 2·0, containing 500 U ml−1 pepsin A and 1·0 g l−1 bacteriological peptone. (□), Viable count of bacteria at t=0; (▮), viable count of bacteria at t=105 min

Survival and growth in bile

The survival of the three B. lactis isolates was not dramatically affected by exposure to bile, even immediately following acid and protease treatment (Fig. 2). In comparison, the B. adolescentis strain B97 (selected for testing due to its high amylase activity and ability to inhibit intestinal pathogens — personal communication Dr Anders Henriksson, The University of New South Wales, Australia) did not survive at detectable levels after passage through the gastric model and therefore, the effect of bile on the survival of this organism was not determined. Although not substantially reducing viability, bile did significantly reduce the growth rate and yield of the B. lactis isolates and other Bifidobacterium strains examined (Fig. 3). When the B. lactis isolates were grown in higher concentrations of ox bile (up to 100 g l−1), growth was observed even at 80 g l−1, although the cell yield was substantially reduced (data not shown). These results indicate that the effect of bile on these isolates was not bacteriocidal but rather, bacteriostatic.

Figure . 2.

Survival of bifidobacteria in an in vitro model simulating conditions in the human gastrointestinal tract. Error bars represent the standard error. (□),   Viable count at time=0; (▮), viable count after passage through an in vitro stomach model. (Incubation at 37 °C for 105 min in 0·1 mol l−1 HCl/KCl buffer, pH 2·0, containing 500 U ml−1 pepsin A and g l−1 bacteriological peptone.); (░), viable count after sequential passage though an in vitro stomach model (as above) followed by incubation for 360 min at 37 °C in 3·0 g l−1 ox bile in 0·1 mol l−1 phosphate buffer, pH 6·5, containing 1·0 g l−1 bacterio-logical peptone; (▧), viable count in the control at time = 465 min. (Controls contained 0·1 mol l−1 phosphate buffer, pH 6·5, plus 1·0 g l−1 bacteriological peptone and were incubated at 37 °C)

Figure . 3.

The effect of bile on the cell yield of bifidobacteria grown in batch culture for 17 h at 37 °C. Error bars represent the standard deviation from the mean of duplicate experiments. (□), Growth in Reinforced Clostridial Medium (Control); (░), growth in Reinforced Clostridial Medium containing 3·0 g l−1 ox bile

Strain identification and the plasmid profiles

To determine the genetic similarity of the acid-tolerant Bifidobacterium isolates, the B. lactis isolates B94, Bb-12 and DS 920, and B. animalis NCFB 2716, were analysed using PFGE. The Type Strain of B. lactis (DSM 10140) and a B. breve strain were included for comparison. Interestingly, the four B. lactis isolates all produced the same PFGE pattern when the chromosomal DNA was cut using the restriction endonuclease XbaI (Fig. 4). Each of these isolates also displayed identical PFGE patterns when digested using the enzyme SpeI (data not shown). This indicates that these organisms are very closely related. Indeed, it was interesting to observe that B94 possessed a phenotypic difference, in starch hydrolysis, to the other B. lactis isolates. No plasmids could be isolated from any of the B. lactis isolates, including B94, although a plasmid was successfully isolated from the positive control organism, B. breve LMG 10737 (data not shown). It therefore appears that the amylase activity in B. lactis B94 is encoded chromosomally.

Figure . 4.

Pulse field gel electrophoresis pattern of DNA from Bifidobacterium isolates. The chromosomal DNA of the bacteria was cut using the restriction endonuclease Xba1. The size of the fragments was determined using Lambda Ladder DNA (Bio-Rad)

Effect of temperature on growth

In the manufacture of yoghurt, the temperature is often raised to 42°C during acidification for the benefit of the starter cultures (Rasic and Kurman 1978). Some bifidobacteria, for example B. longum B21 and B. adolescentis B97, have a relatively low maximum growth temperature (Fig. 5). In contrast, the B. lactis isolates, including B94, were able to grow optimally at 40°C and grew well even at 45°C. It appears that the growth and survival of these isolates would not be adversely affected by elevated temperatures used during yoghurt manufacture.

Figure . 5.

The effect of temperature on the growth of bifidobacteria. The optical density represents growth after 48 h in RCM broth. The results presented are the means of three experiments ± standard deviation. (▮), Bifidobacterium lactis B94; (░), B. lactis DS 920; (□), B. longum B21; (▒), B. adolescentis B97

Cell yield in laboratory scale fermentations

The cell yield of B. lactis B94 when grown in a whey-based growth medium was compared with that produced by the commercially-produced strain, B. lactis DS 920, under laboratory conditions. Both organisms reached stationary phase at 24 h with comparable cell yields. At 24 h, B. lactis DS 920 produced an average yield of 5·5 × 109 cfu ml−1 (S.D. ± 1·0 × 109), while B. lactis B94 yielded 6·4 × 109 cfu ml−1 (S.D. ± 1·3 × 109). Therefore, growth under industrial conditions does not appear to be an impediment to the commercial use of B. lactis B94.

Utilization of prebiotics and other carbohydrates

Bifidobacterium lactis B94 was grown using a number of different prebiotic carbohydrates as the carbon source (Fig. 6). Growth was not observed in the medium when a carbon source was omitted, confirming that the base medium was carbon limited for this strain. Bifidobacterium lactis B94 was able to grow on a diverse range of carbon sources, including most of the prebiotic oligosaccharides. However, no growth was observed on the prebiotic lactulose. The final biomass yield varied considerably between fermented carbon sources. Interestingly, the bacterium grew very poorly on the monosaccharides fructose, galactose and xylose, in comparison with its growth on disaccharides and oligosaccharides composed predominantly of these sugar moieties.

Figure . 6.

The effect of carbon source on the cell yield of Bifidobacterium lactis B94. The results presented represent the mean of three experiments ± standard deviation

Survival in yoghurt

The survival of B. lactis B94 was compared with that of a commercial B. lactis isolate (DS 920) and a strongly amylolytic strain (B. adolescentis B97) in a synbiotic yoghurt. The yoghurt contained both resistant starch and inulin (Raftiline™) as prebiotics. Both B. lactis isolates survived well in the yoghurt, maintaining viable counts greater than 7 × 107 cfu g−1 throughout the 5-week storage period. However, the B. adolescentis strain did not remain viable in this yoghurt. Its viable counts dropped steadily to below the detection limit of 102 cfu g−1 after 4 weeks. This strain also failed to survive in the in vitro stomach model described previously and is evidently sensitive to low pH. The pH in the yoghurts dropped from 4·5 at the start of storage down to 4·1 after 5 weeks.

DISCUSSION

The concept of synbiotics offers the potential for increased potency of functional foods by exploiting the synergy between prebiotic and probiotic ingredients. In the current investigation, a probiotic Bifidobacterium isolate was specifically selected to complement a predefined prebiotic (Hi-maize™) for a synbiotic yoghurt. Although many of the 40 Bifidobacterium isolates examined were able to hydrolyse the resistant starch, only one organism, B. lactis B94, passed the stringent selection criteria imposed. PFGE analysis revealed that this isolate was very closely related to the Type Strain of B. lactis (DSM 10140) and to the commercial organisms B. lactis Bb-12 and B. lactis DS 920. It is interesting that a phenotypic difference, in this case amylase activity, was observed between isolates with the same PFGE pattern. In a preliminary study, no plasmids were detected in B. lactis B94. It is therefore likely that the amylase gene in this organism is located on the chromosome. Use of different restriction enzymes, or combinations of enzymes, may demonstrate strain differences.

The B. lactis isolates, and to a lesser degree B. animalis NCFB 2716, were the only organisms that survived well in the gastric model. The viable counts of the B. angulatum, B. adolescentis, B. breve and B. longum isolates all decreased by several orders of magnitude. Similar results have been observed for the majority of Bifidobacterium species and strains trialed using in vitro gastric models (Clark et al. 1993; Lankaputhra and Shah 1995; Marteau et al. 1997; Charteris et al. 1998; Zavaglia et al. 1998). It therefore appears that tolerance to low pH for extended periods is not a common trait amongst Bifidobacterium strains. Conditions in the human stomach, including pH, residence time and protein levels, that can influence survival (Conway et al. 1987) vary considerably with diet and throughout the day. The quantitative levels of bifidobacterial survival observed using the static in vitro models employed in this and other studies only serve, at best, as a crude guide to anticipated survival rates in vivo. These models are nonetheless useful tools for the selection of strains most likely to survive in vivo. Bifidobacterium lactis Bb-12, which survived well in the intestinal models used in this investigation, has been demonstrated to survive intestinal transit in humans (Hove et al. 1994; Fukushima et al. 1997; Mattila-Sandholm et al. 2000). It is reasonable to predict that due to its genetic similarity to Bb-12 and its comparable level of survival in the gastric model, B. lactis B94 would also survive gastric transit in humans.

Bile appeared to have a bacteriostatic rather than a bacteriocidal effect on the bifidobacteria investigated in this study. Little loss of viability was observed for B. lactis isolates upon exposure to physiological levels of bile, even following pre-treatment in simulated gastric juice. Additionally, all of the isolates examined were able to grow to some extent in the presence of bile, and at levels far in excess of physiological concentrations. This is in agreement with the findings of Charteris et al. (1998), who observed no significant loss of viability for five of six Bifidobacterium strains during simulated small intestinal transit. Ibrahim and Bezkorovainy (1993) also concluded that bifidobacteria were able to survive exposure to physiological and higher bile salt levels. In contrast, other researchers have observed bacteriocidal effects of bile on bifidobacteria (Clark and Martin 1994; Lankaputhra and Shah 1995; Marteau et al. 1997). Comparisons between these studies are difficult as different bile sources, concentrations, exposure periods and model designs were used. However, these studies do demonstrate inter- and intraspecies variations in bile survival among bifidobacteria, and assessment of the bile tolerance of strains with potential for use as dietary adjuncts is warranted.

Although not significantly decreasing viability, bile did dramatically reduce the specific growth rate and yield of the B. lactis isolates examined in the present study. This effect was universal among the bifidobacteria used here. Reduced growth rate and yield in bile would be expected to impede the survival of probiotics colonizing the small intestine. However, bifidobacteria colonize predominantly in the colon of humans (Mitsuoka 1982). As the bile secreted into the duodenum is almost all reabsorbed in the terminal ileum, with only a minor fraction passing through to the colon (Weisbrodt 1977), bifidobacteria surviving transit to the colon would not have to contend with continued exposure to bile. It is therefore likely that bile would not inhibit the growth of B. lactis B94 in vivo.

The gastric model used in this investigation appeared to be predictive of the survival of bifidobacteria in yoghurt. Both B. lactis DS 920 and B. lactis B94 survived well in yoghurt, whereas B. adolescentis B97, which failed to survive in the gastric model, also rapidly lost viability during storage in yoghurt.

Bifidobacterium lactis B94 was able to ferment a number of prebiotic carbohydrates in addition to resistant starch. Inulin, and fructo-, galacto-, xylo- and soybean-oligosaccharides (raffinose and stachyose), could all be included with B. lactis B94 as complementary prebiotics. However, it is not possible from the in vitro yield and growth rate data alone to determine which would provide the most selective, and effective prebiotic for this organism in vivo. It is interesting to note that B94 was able to grow better on fructo-, galacto- and xylo-oligosaccharides than on the monosaccharides (fructose, galactose and xylose, respectively) of which the more complex sugars are composed. This phenomenon has been observed for a number of Bifidobacterium strains (Gibson and Wang 1994; Hopkins et al. 1998) and suggests that these organisms have specific substrate transport mechanisms that are more efficient at transporting indigestible oligosaccharides than simple sugars. This may be a reflection of the evolutionary environment of these organisms in the colon of humans and animals, where the supply of simple sugars is limited due to adsorption in the small intestine.

The results of the current investigations demonstrate that B. lactis B94 possesses the technological characteristics for it to be successfully manufactured and incorporated into synbiotic yoghurt. The organism also appears likely to survive gastrointestinal transit. It is therefore a promising new probiotic bacterium. Efforts are currently under way to determine what health functionalities the strain may possess, including inhibition of intestinal pathogens, adhesion to intestinal mucosa and production of vitamins. Additionally, the safety of the strain is being assessed. Ultimately, in vivo trials will be required to determine whether this organism can positively influence the health of consumers and therefore fulfil the requirements of a probiotic organism.

Acknowledgements

The authors gratefully acknowledge the financial assistance for this project provided by The CRC for Food Industry Innovation, Australia. They thank Dr Anders Henriksson for the kind donation of the Bifidobacterium strains B21, B21C, Lafti™ B22, B74, B74A and B97 used in this study.

Ancillary