Manipulation of colonic bacteria and volatile fatty acid production by dietary high amylose maize (amylomaize) starch granules

Authors


Correspondence to: Xin Wang, School of Medicine, Mater Adult Hospital, Raymond Tec. South Bank, QLD 4010, Australia (e-mail: xin.wang@mailbox.uq.edu.au).

Abstract

Aims: To study the effects of amylomaize starch and modified (carboxymethylated and acetylated) amylomaize starches on the composition of colonic bacteria and the production of volatile fatty acids, in mice.
Methods and Results: Balb/c mice were fed with experimental diets containing various amount of amylomaize and modified amylomaize starches. Colonic bacterial populations and short-chain fatty acids were monitored. Results showed that the increases in indigenous bifidobacteria were detected in mice fed all starches tested; however, the highest numbers were observed in the group fed with 40% unmodified amylomaize starch. The starch type influenced the populations of indigenous Lactobacillus, Bacteroides and coliforms. High Lactobacillus numbers were achieved in the colon of mice fed with high concentration of amylomaize starch. Acetylated amylomaize starch significantly reduced the population of coliforms. In addition, orally dosed amylomaize utilizing bifidobacteria reached their highest levels when fed together with amylomaize or carboxymethylated amylomaize starch and in both cases butyrate levels were markedly increased.
Conclusions: These results indicate that different amylomaize starches could generate desirable variation in gut microflora and that particular starches may be used to selectively modify gut function.
Significance and Impact of Study: Amylomaize starch appeared to enhance the desirable composition of colonic bacteria in mice, and suggested it possessed the potential prebiotic properties. Therefore, resistant starch and its chemical derivatives may exert beneficial impacts to the human colon.

Introduction

In excess of 90% of cells associated with the human body are ascribed to the complex and diverse community of bacteria which reside in the human large intestine (Savage 1977). The majority of those bacteria are the obligate anaerobes, including Bacteroides, Eubacterium, Bifidobacterium, Fusobacterium, Clostridium and anaerobic cocci (Macfarlane et al. 1991). It has been well established that the composition and metabolic activity of these bacteria directly affect the status of colonic function, and consequently influence the health of the host. Bifidobacterium and Lactobacilli are recognized as being beneficial because they may enhance the systemic resistance to pathogen infection, produce enzymes and vitamins and modulate the host immunological response (Yamazaki et al. 1982; Gibson et al. 1995; Takahashi et al. 1998). Consequently, selected live Bifidobacterium and Lactobacillus strains have been orally administered in order to increase the population of these bacteria in the large intestine (Conway 1996). Another approach to increase bifidobacteria levels has been to ingest oligosaccharides, which are largely undegraded in the upper regions of the digestive tract. When the oligosaccharides reach the large bowel they may be selectively utilized by beneficial microbes (Gibson et al. 1995). Such carbohydrates are referred to as prebiotics (Gibson and Roberfroid 1995).

Although the microecosystem of the large intestine in adults is considered fairly stable over a long period of time, there are both host and environmental factors which can alter its homeostasis. Diet is one of the most important factors directly impacting on the nature and balance of the microbiological community (Conway 1996). Hudson and Parr (1924) reported that rats fed a diet high in carbohydrates, had a lower pH in the colon than those fed a high protein diet, suggesting that the level of short-chain fatty acids in the colon of rats increased as the result of the increasing activity of saccharolytic bacteria. A later study carried out by Chung et al. (1977) showed that the populations of coliforms and clostridia increased in rats fed a high-protein diet, whereas there was an increase in lactobacilli in rats fed a high-carbohydrate diet. In human studies, a diet consisting of high levels of beef caused an increase in the faecal population of Bacteroides and some species of Clostridium, whereas high levels of brown rice in the diet increased the Bifidobacterium adolescentis, Ruminococcus gnavus and Peptostreptococcus magnus (Benno and Mitsuoka 1991). The effect of dietary components on the colonic microbial population can be ascribed to the selective enrichment of certain bacterial species. According to this principle, a range of oligosaccharides such as oligofructose and oligogalactose have been developed and added into the diet in order to stimulate the growth of indigenous bifidobacteria in the human colon (Gibson et al. 1995). Competition for the limiting nutrients and adhesion sites on the intestinal mucosa is likely to be important in controlling actual bacterial colonization of the colon. Less competitive species would be expected to be eliminated from the system (Gibson and Wang 1994).

Until recently, starch has been considered mainly as an energy and carbon source in the human diet; however, it can also have properties similar to dietary fibre. Resistant starch is largely undegraded in the small intestine and is a major part of the carbohydrate available in the human colon for bacterial fermentation, which results in the production of short-chain fatty acids and gases (Cummings et al. 1990; Brown 1996). These end products of fermentation can exert significant physiological effects on the host (Kritchevsky 1995). In previous in vitro experiments, it has been demonstrated that a wide range of colonic bacteria, including Bacteroides, Bifidobacterium, Eubacterium, Clostridium and Propionibacterium can utilize amylopectin (waxy starch) and soluble starch. However, only a few species of Bifidobacterium and Clostridium have the capacity to degrade amylomaize starch (Wang et al. 1999b). Other workers have shown a proliferation of indigenous bifidobacteria and lactobacilli in rats fed a resistant starch-based diet (Kleessen et al. 1997; Silvi et al. 1999). Those observations led us to study the effects of amylomaize starch and modified amylomaize starches on the composition of colonic bacteria, and the production of volatile fatty acids, in mice.

Materials and methods

Growth substrates and chemicals

Amylomaize starch (Culture-ProTM 958N) and amylopectin maize starch (waxy starch) were obtained from Penford Australia Limited. The former contained more than 80% amylose and 33·4% total dietary fibre. The latter contained 98% amylopectin and less than 2% amylose (Brown 1993). Other modified amylomaize starch granules included carboxymethylated amylomaize starch (carboxyl value 1·0%) and acetylated amylomaize starch (acetyl value 4·0%) were prepared using Culture-ProTM 958N by Penford Australia Limited as the base starch used for modification. Unless otherwise specified, all other chemicals were purchased from Sigma Chemical Company, St.Louis, USA.

Animals

Female mice (Balb/C, SPF) were purchased from BHD animal supply (Sydney, Australia) at the age of 6 weeks, and maintained in house for 4–6 weeks prior to the commencement of the experiments. All mice were fed ad libitum with standard rodent feed (Lab-feed, Sydney, Australia), sterilized by radiation, and were given tap water (which was sterilized by autoclaving at 121°C for 15 min).

Effect of dietary starch on mouse colonic bacteria

To investigate the effect of dietary starch on the populations of colonic bacteria, mice were randomly divided into five groups of six mice and were housed one group per cage. Freshly void faecal pellets from each mouse were collected for bacterial analysis at d 3 and d 1 prior to commencement of experimental diets. The experimental diets were based on the AIN 76 high carbohydrate diet (Reeves et al. 1993). Waxy maize starch, amylomaize starch, carboxymethylated amylomaize starch and acetylated amylomaize starch were used as the major carbon source and were added at a concentration of 400 g kg−1 into the experimental diets designated as A, B, C and E, respectively. Other ingredients included (g kg−1): casein 200; canola oil 25; sunflower oil 25; sucrose 150; wheat bran 100; gelatin 20; mineral mix 63; vitamin mix 13 (Vitamin and Mineral premix, Naturally Pty Ltd, Mittagong, NSW, Australia); dl-methionine 2; and choline chloride 2. The animals were fed the experimental diet for four consecutive weeks, and two freshly void faecal specimens were collected from individual mice at d 25 and d 27 for analysis of Bifidobacterium, Lactobacillus and coliforms. Individual body weights were also recorded. The mice that were continuously fed with the standard rodent feed were used as the control group and referred to as group D.

Dose effect of dietary amylomaize starch on mouse colonic bacteria

Mice were divided into six groups of six mice. The basic ingredients of the diet were similar to the previous experiment except different concentrations of amylomaize starch were used. This ranged from 0% (group A), 8% (group B), 16·7% (group C), 30% (group D) to 40% (group E). Since the amylomaize starch contained 33·4% dietary fibre, glucose and cellulose were used as substitutes proportionally to replace the high amylomaize starch. Variations in concentration of carbohydrates used are presented in Table 2. Two faecal specimens were collected before commencement of the experimental diet, then six faecal samples were collected during the experiment at d 3, 7, 11, 15, 20 and 25 for bacteriological analysis. The control group, which was continuously fed ad libitum with standard rodent feed is group F.

Table 2.  Variation in the type and concentration of carbohydrates used in the amylomaize starch dose study
GroupsTotal carbohydrates
(g kg−1)
Amylomaize
starch (g kg−1)
Cellulose
(g kg−1)
Glucose
(g kg−1)
% of resistant starch
in complete diet
  1. Amylomaize starch contains at least 33·4% of dietary fibre.

A40001202800
B40080962242·6
C400167701635·5
D40030030709·9
E4004000013·2
FStandard rodent feed

Effect of dietary starch on faecal bacteria and other parameters of mice dosed with Bifidobacterium sp. Lafti B8

Four groups of six mice were used to investigate the effects of coadministration of resistant starch with Bifidobacterium cells with amylolytic activity. The control group, D, was continuously fed control rodent feed ad libitum thoughout the whole experimental period. The composition of the experimental diet was the same as used in the previous section, with waxy starch (400 g kg−1) used in diet A, amylomaize starch (400 g kg−1) used in diet B, and carboxymethylated amylomaize starch (400 g kg−1) used in diet C. At d 1, mice were changed to the experimental diets and at the same time 200 µl of overnight cultures of the probiotic Bifidobacterium sp. Lafti B8 (2·0 × 108) were orogastrically administrated to all of the mice in groups A, B, C and D. The experimental diet was fed for 8 d and all mice were dosed with the similar number of the probiotic bacteria each day. Freshly voided faecal samples from individual mice were collected daily during the feeding of the experimental diet for immediate bacteriological analysis, and samples stored at −20°C for analysis of short-chain fatty acids and starch content.

Bacteriological analysis of faecal specimens

Two or three faecal pellets from individual mice were collected into Eppendorf tubes. A 10% (w/v) suspension was made with half strength of Wilkin-Chalgren anaerobe broth (Oxoid) on the basis of faecal weight. Samples were then gently homogenized using the automatic pipette tip and serially diluted to 10−8 with the same diluent. Aliquots (10 µl) from six dilutions were dropped onto the selective agar plates in triplicate. RCP agar plates (Reinforced Clostridial agar, Oxoid, UK, with 5 mg l−1 of propionic acid, final pH 5·2) were used to enumerate the total numbers of Bifidobacterium. Rogosa agar (Oxoid) was used for enumeration of Lactobacillus. MacConkey no. 3 agar (Oxoid) was used to enumerate coliforms. MBA (Modified Bacteroides agar) was used to qualify the Bacteroides (Macfarlane et al. 1991). All of the plates containing RCP, Rogosa and MBA were incubated at 37°C for 72 h in an anaerobic chamber (Mark 3 Workstation, DW Scientific). MacConkey plates were incubated anaerobically at 37°C for 24 h. Bifidobacterium isolates were confirmed by demonstrating the presence of fructose-6-phosphate phosphoketolase(F-6-PP) in crude cell extracts (Scardovi and Trovatelli 1965). Lactobacillus, Bacteroides and coliforms were identified on the basis of colony morphology, Gram stain, cell morphology and the catalase reaction. Uncertain strains were further identified by measuring the volatile fatty acid after growth in PYG broth. Amylose agar (Wang et al. 1999b) with I/K2I were used to detect bacteria which could hydrolyse amylose. Amylolytic Bifidobacterium isolates were confirmed using the F-6-PP reaction.

Analysis of short-chain fatty acids and total starch

Faecal slurries were prepared from frozen faecal samples using 0·05 mol l−1 phosphate buffer (Na2HPO4 and NaH2PO4, pH 7·2) to yield a 10% w/v suspension. Aliquots (0·4 ml) of the slurry were mixed with 0·1 ml phosphoric acid and 16 µl of internal standard (2-ethyl butyric acid, 0·2 mol l−1) for fatty acid analysis. After centrifuging at 12 800 × g for 15 min, the supernatants were carefully removed and 1 µl aliquots were injected into agas chromatograph (Autosystem GC, Perkin Elmer, Norwalk, CT, USA) equipped with a carboxyl wax capillary column (inside diameter 0·32 mm) (Supelco, Bulletin 749D), a flame ionization detector and integrater. The mixed standard of short-chain fatty acids included acetic acid (6·67 mmol l−1), propionic acid (3·33 mmol l−1), isobutyric acid (0·333 mmol l−1), n-butyric acid (1·66 mmol l−1), isovalenic acid (0·333 mmol l−1) and n-valenic acid (0·333 mmol l−1).

The residual starch in faeces was measured using the alpha-amylase amyloglucosidase method (Megazyme, total starch assay kit, Megazyme, Co Wicklow, Ireland) in the 10% (w/v) faecal suspensions.

Statistical analysis

The results are presented as a mean ± standard deviation. Before statistical analysis of these data, the normality of data was checked by the Kolmogorov–Smirnov test. Logarithms of the counts were used to achieve normal distribution. Variance between the dietary regimes was determined by analysis of variance (single factor anova). Since the detection limit for enumeration of the bacteria was log10 3, the bacterial counts in faecal samples below the detection level were arbitrarily taken to be three in the statistical analysis.

Results

The effect of dietary starch on mouse colonic bifidobacteria

The energy levels of experimental diets are similar to the standard rodent feed since changing the diets did not lead to significant changes in mouse body weights as indicated by the average weights prior to and post feeding experimental diets (Pre. and Post.) (Table 1). The indigenous Bifidobacterium in mice faeces before changing to the experimental diets were enumerated using RCP agar plates. No colonies were observed, suggesting that the mice were either free of indigenous bifidobacteria or their numbers were below the detection level (103). In addition, when the mice were maintained on the control mouse feed throughout the entire experimental period (group D), Bifidobacterium in the faeces continued to be undetectable (Table 1). High numbers of faecal Bifidobacterium were observed in mice fed the experimental diets for 4 weeks (Table 1), and the magnitude varied for the types of starch being incorporated into the diets. The highest number (log10 8·2 cfu g−1 wet weight faeces) was found in mice which were fed amylomaize starch, whilst the lowest count was from the mice which were fed the carboxymethylated amylomaize starch diet (log10 3·5 cfu g−1 wet weight faeces) and only two of six mice in the latter group had positive isolates in faeces. It is interesting to note that waxy starch produced relatively high numbers of bifidobacteria in the faeces (log10 7·5 cfu g−1 wet weight faeces), as did acetylated amylomaize starch (log10 6·4 cfu g−1 wet weight faeces). For Lactobacillus, there was no significant difference in the counts when the mice were fed the amylomaize starch diet in comparison with the standard rodent feed; however, the numbers declined significantly for the mice fed carboxymethylated amylomaize starch and acetylated amylomaize starch diets (P < 0·01). The population of lactobacilli in mice fed the waxy starch diet was significantly lower than in mice receiving the amylomaize starch diet (P < 0·05) (Table 1). The population of coliforms varied among the test groups. A statistical significant increase (P < 0·01) was observed in the group of mice fed the carboxymethylated amylomaize starch diet, and the increase was observed in mice fed with amylomaize starch and with waxy starch diets when compared to the control group. In contrast, the count of coliforms in the group consuming the acetylated amylomaize starch diet was significantly decreased (P < 0·05). In fact, no coliforms colonies were detected from any of the six mice in that group.

Table 1.  The population of indigenous Bifidobacterium, Lactobacillus and coliforms in the faeces of mice continuously fed with experimental diets for 4 weeks
 A (Waxy)B (HM)C (CMC)D (Control)E (AC)
  1. Results are expressed as the mean ± standard deviation of six mice from faecal samples obtained at d 25 and d 27 after consumption of the diet, and expressed as cfu log10 g−1 wet wt faeces.

  2. The diets contained 40% carbohydrates as follows: Group A (Waxy), amylopectin (waxy) maize starch; Group B (HM), amylomaize starch; Group C (CMC), carboxymethylated amylomaize starch; Group D (Control), mouse standard feed; Group E (AC), acetylated amylomaize starch. Other ingredients are described in Materials and Methods. Values marked with asterisk are significantly different from group D (*P < 0·05; **P < 0·01).

Body weight (g) Pre.20·39 ± 1·6820·06 ± 1·8921·22 ± 2·5020·27 ± 2·0820·87 ± 8·70
Body weight (g) Post.20·82 ± 1·7020·85 ± 1·2221·26 ± 2·6320·94 ± 2·3020·89 ± 1·57
Bifidobacterium  7·5 ± 0·5**  8·2 ± 0·5**  3·5 ± 2·2< 3·0  6·4 ± 0·6*
(Positive mice)(6/6)(6/6)(2/6)(0/6)(6/6)
Lactobacillus  7·7 ± 0·4*  8·0 ± 0·1  7·5 ± 0·1**  8·1 ± 0·0 7·37 ± 0·0**
(Positive mice)(6/6)(6/6)(6/6)(6/6)(6/6)
Coliforms  4·7 ± 0·5*  5·0 ± 0·9*  6·0 ± 0·2**  3·8 ± 0·2< 3·0
(Positive mice)(6/6)(6/6)(6/6)(3/6)(0/6)

Dose effect of dietary amylomaize starch on mouse colonic bacteria

Replacement of amylomaize starch granules by glucose and cellulose (Table 2) did not significantly affect mice body weight (Table 3). Comparative populations of Bifidobacterium, Lactobacillus and coliforms before and after changing to the experimental diets are also shown in Table 3. The population of Lactobacillus was significantly influenced by the concentration of starch in the diet. The viable count of Lactobacillus in group A, which was fed with the nonstarch diet, was significantly lower (P < 0·01) than those recorded in group F, which were fed control rodent feed. However, the decline in the number of lactobacilli was progressively minimized with increasing levels of amylomaize starch in the diet. There was no significant differences in the numbers of lactobacilli between the group fed 40% amylose starch diet and control rodent feed.

Table 3.  Comparison of the populations of coliforms, Lactobacillus and Bifidobacterium in the faeces of the mice before (1 d) and 25 d after changing to the experimental diets
Experimental
groups*
% of amylose
maize starch
Body weight (g)ColiformsLactobacillusBifidobacterium
BeforeAfterBeforeAfterBeforeAfterBeforeAfter
  1. Results are mean ± standard deviation of six mice. Samples in both experimental stages are expressed as cfu log10 g−1 wet faeces.

  2. *The concentration of carbohydrates from groups of A, B, C, D and E is described in Table 2. Other ingredients in the diet are referred to the Materials and Methods. The mice in group F were fed standard rodent feed.

A 016·0 ± 1·915·7 ± 1·95·1 ± 0·34·5 ± 0·17·9 ± 0·16·5 ± 0·4<3·0<3·0
B 816·0 ± 1·916·1 ± 1·67·0 ± 2·45·8 ± 1·08·3 ± 0·16·7 ± 0·2<3·0<3·0
C16·715·6 ± 1·516·4 ± 1·65·0 ± 0·24·1 ± 0·18·0 ± 0·27·3 ± 0·1<3·0<3·0
D3016·3 ± 1·717·4 ± 2·34·3 ± 0·14·1 ± 0·28·0 ± 0·17·0 ± 0·1<3·08·7 ± 1·2
E4016·0 ± 1·216·9 ± 0·75·8 ± 2·57·8 ± 0·58·2 ± 0·28·1 ± 0·1<3·08·3 ± 1·6
F_16·0 ± 0·716·3 ± 1·25·7 ± 2·36·9 ± 1·38·0 ± 0·28·0 ± 0·1<3·0<3·0

The concentration of amylomaize starch in the diets affected the number of indigenous Bifidobacterium. High levels of bifidobacteria were observed only in the groups of mice fed 30% or more amylomaize starch. The time course for development of indigenous bifidobacteria in the groups of mice fed 30% and 40% amylomaize starch is shown in Table 4. The bifidobacteria numbers increased dramatically even after 3 d of feeding. The population of coliforms fluctuated during the experimental period (Table 3), but no clear trend was apparent because of the variable error bars.

Table 4.  Development of Bifidobacterium inthe faeces of mice fed with 30% and 40%ofamylomaize starch
Duration of
experiment
30% high amylose diet40% high amylose diet
PopulationPositive micePopulationPositive mice
  1. Results are mean ± standard deviation of six mice from single faecal specimen and expressed as cfu log10 g−1 wet faeces.

Day 0<3·00/6<3·00/6
Day 36·0 ± 2·84/66·3 ± 1·66/6
Day 77·0 ± 2·56/68·5 ± 1·76/6
Day 118·1 ± 2·16/68·7 ± 1·36/6
Day 158·2 ± 1·56/68·1 ± 1·06/6
Day 207·7 ± 1·26/68·4 ± 1·76/6
Day 258·7 ± 1·36/68·3 ± 1·66/6

Effect of dietary starch on faecal bacteria and volatile fatty acid production of mice dosed with Bifidobacterium sp. Lafti B8

Bifidobacterium sp. Lafti B8 in the faeces of mice could be distinguished from indigenous bifidobacteria by the production of clear zones on amylose-containing plates. As presented in Table 5, the amylomaize starch diet and carboxymethylated amylomaize starch diet significantly increased the numbers of Bifidobacterium Lafti B8 in the faeces in comparison to the control and waxy diets. For Bacteroides, a significant increase of viable counts was detected in all of the starch-based diets, with the highest numbers observed in mice fed the carboxymethylated amylomaize starch diet.

Table 5.  Viable cells of Bifidobacterium, coliforms and Bacteroides in the faeces of mice fed with the control amylopectin (waxy) maize starch (Group A), amylomaize starch (Group B), carboxymethylated amylomaize starch (Group C) and mouse standard diet(GroupD)
 Experimental diet
Group A
(Waxy)
Group B
(HM)
Group C
(CMC)
Group D
(Control)
  1. In addition, each day 200 µl of Bifidobacterium Lafti B8 was oro-gastrically administered to all mice. Results expressed as mean ± standard deviation, using results for samples from d 3 to d 8. Samples from all animals (n=6 per group) were analysed on triplicate plates. Bacterial counts are presented as cfu log10 g−1 wet weight of faeces. Volatile fatty acids (VFA) are presented as mmol ml−1 of faecal slurry. The residual starch detectable in the faeces was quantified as mg starch per gram wet wt of faeces.

  2. Values marked with asterisks are significantly different to group D (*P < 0·05; **P < 0·01).

Bifidobacterium Lafti 8B 7·2 ± 1·0  8·0 ± 0·6*  8·0 ± 0·5* 7·2 ± 0·6
Coliform 4·6 ± 1·6  4·7 ± 1·9  5·3 ± 1·8 3·9 ± 0·9
Bacteroides 9·2 ± 0·4*  9·0 ± 0·4* 10·0 ± 0·3** 8·5 ± 0·6
VFA
 Acetate67·2 ± 13·89* 63·0 ± 9·22 52·6 ± 2·8550·6 ± 6·34
 Propionate19·4 ± 2·33** 16·0 ± 2·11 17·7 ± 3·4214·0 ± 1·34
 Butyrate 6·9 ± 2·78 27·0 ± 16·47** 26·2 ± 6·08** 8·8 ± 3·49
Starch (mg/g faeces)15·6 ± 12·3288·7 ± 8·6**167·3 ± 20·0** 4·1 ± 2·8
Total faecal output
 (g/per d per mouse)
0·090·380·440·43

Acetate, propionate and butyrate are the major short-chain fatty acids measurable in the faeces, while isovalerate, isobutyrate and n-valerate were too low to be detected. The concentration of acetate and propionate was slightly higher in all mice fed with experimental diets, in comparison with the mice fed with the control diet. When the mice were fed the amylomaize starch diet or carboxymethylated amylomaize starch diet the concentration of butyrate was 26·9 and 26·2 mmol ml−1, respectively. This was significantly higher than the levels in the waxy diet and the control mouse diet, 6·78 and 8·80 mmol ml−1, respectively.

The digestibility of starches in the mouse digestive tract (Table 5) varied. Waxy maize starch was almost completely hydrolysed, with only 15·6 mg of starch (g−1) detectable in the faeces, in contrast to the carboxymethylated amylomaize starch which was relatively resistant to digestion, with 167·3 mg of starch detected per gram faeces. Amylomaize starch demonstrated the highest resistance to digestion in the mouse digestive tract with 288·7 mg of starch per gram of faeces. Amylomaize starch and carboxymethylated amylomaize starch diets produced similar levels of dry faecal output as standard rodent feed, whereas the completely digested waxy starch resulted in five times lower faeces per day per mouse (Table 5).

Discussion

The work presented in this paper shows that amylomaize starch sustains good growth of colonic bifidobacteria and induces an elevation in faecal butyrate levels. The degree of stimulation varied depending on the chemical modification of the starch and its concentration in the diet. This promotion of indigenous bifidobacteria is consistent with other studies which have demonstrated that some forms of resistant starch in the diet have the capacity to sustain good growth of indigenous bifidobacteria in the colon of rats or pigs (Brown et al. 1997; Kleessen et al. 1997; Silvi et al. 1999).

It was generally accepted that mouse is a good animal model for studying interactions between the gut microbes and the host, since the mouse is the best studied of the intestinal ecosystems of monogastric animals. Although there are some anatomical differences in the gastrointestinal tracts of mice and man (Tannock 1997), the faecal bacteria populations of the major groups of bacteria were similar (Finegold et al. 1983;Tannock 1997) and hence it is suggested that the mouse could be considered as an animal model to study the dietary impact on the population of colonic bacteria. From previous invitro experiments, it is known that the fermentation patterns of amylopectin (waxy) starch is markedly different to that of amylomaize starch. The former can be degraded by a range of colonic bacteria including Bacteroides, Eubacterium, Propionicbacterium, Bifidobacterium and Clostridium, whereas the latter is hydrolysed by only amylolytic bifidobacteria and clostridia (Wang et al. 1999a,b). Hence it was proposed that amylomaize starch might have prebiotic properties in term of stimulating the growth of indigenous bifidobacteria. In current experiments, the effect of amylomaize starch on the composition and population of indigenous colonic bacteria has been investigated using mouse model. Since the control mice fed a standard mouse diet had no detectable bifidobacteria, the high levels of bifidobacteria in starch-fed mice confirms the in vitro studies, implying the bifidogenic effect of amylomaize starch. Although it was not possible to detect bifidobacteria in the faeces of mice fed the standard rodent feed, they must have been present in numbers that were under the detection level.

Among the starches tested, unmodified amylomaize starch was the most effective substrate in term of stimulation of the growth of bifidobacteria. It seems that chemical modifications significantly influence the bacterial utilization patterns. Notwithstanding, all starches tested are able to sustain the good growth of bacteroides in the colon of mice; carboxymethylated amylomaize starch resulted in the highest levels of Bacteroides (Table 5). Furthermore, acetylated amylomaize starch promoted growth of indigenous bifidobacteria more effectively than did carboxymethylated amylomaize starch (Table 1).

Amylomaize starch granules have been shown to be resistant to degradation in the small intestine (Muir et al. 1995) and therefore would be expected to be available in the colon as a substrate for the colonic bacteria. The effect of stimulation of indigenous bifidobacteria in the faeces of mice was clearly observed when mice were fed with a diet containing 30% or more of amylomaize starch. It is interesting to note that 40% of waxy starch in the diet also stimulated the growth of indigenous bifidobacteria. Although it was believed that waxy starch would be fully degraded in the small intestine (Topping et al. 1997) the high level of inclusion in the diet may result in waxy starch reaching the colon of the mice in this study. In fact a small amount of starch residue was detected in the faeces of mice fed with 40% waxy starch (Table 5).

Alteration of the mouse diet from the standard rodent diet to experimental formula did not significantly affect theanimal body weights but had a major effect on the population of indigenous Lactobacillus (Table 3). The number of Lactobacillus detected was lowest (log10 6·5 cfu g−1 wet weight faeces) when the mice were fed experimental diet with only glucose and cellulose added as the carbohydrates source. Increasing levels of amylomaize starch increased the numbers of Lactobacillus detected. The diet containing 40% of amylomaize starch produced Lactobacillus levels similar to those achieved using the standard rodent feed (Table 3). This observation is different from other studies which indicated that high resistant starch diets led to a stimulation of Lactobacillus numbers over those observed with standard diets (Brown et al. 1997; Kleessen et al. 1997; Silvi et al. 1999). Many factors in diet can contribute to a decrease in the number of Lactobacillus in faeces; however, since glucose may be completely utilized in small intestine, the lack of fermentable carbohydrates in the colon may account for the low Lactobacillus numbers.

The mice bifidobacterial strains stimulated by the dietary amylomaize starch were not able to efficiently hydrolyse amylomaize starch, but they can degrade amylopectin, resulting in production of clear zones on agar plates which contain amylopectin. This enabled the enumeration of the added probiotic Bifidobacterium Lafti B8 from indigenous bifidobacteria by means of detection of clear zones on agar plates containing amylomaize starch. Results presented in Table 5 demonstrated that amylomaize starch could increase the survival of probiotic bifidobacteria. This effect of the amylomaize starch is consistent with the elevated faecal bifidobacteria detected when this starch was used in the feed of pigs orally dosed with Bifidobacterium longum (Brown et al. 1997).

The experimental errors in coliform numbers were quite high; however, it is interesting to note that mice fed acetylated amylomaize starch had no detectable coliforms. This indicates that a particular modified starch may be identified that can be used to minimize coliform counts.

Clinical studies have shown that ingestion of amylomaize starch can elevate the level of butyrate in the large bowel (Noakes et al. 1996). Elevation of butyrate was evident in the present study in mice fed amylomaize starch or carboxymethylated amylomaize starch (Table 5). Since bifidobacteria produce relative lower amounts of butyrate (Barcenilla et al. 2000), other bacteria such as clostridia, eubacteria and fusobacteria must be involved. Indeed in vitro experiments have demonstrated that several strains of Fusobacterium, Clostridium and Eubacterium are able to utilize soluble starch and amylopectin (Wang et al. 1999b), which suggests that those bacteria possess the starch-degrading enzymes. Although those bacteria can not hydrolyse amylomaize starch in vitro, carbohydrate fragment products of primary starch-degrading bacteria may feed other butyrate-producing bacteria and so result in increased butyrate. Detailed differentiation of the microbial community using molecular ecological techniques would be required in order to identify the bacterial species involved in the amylomaize starch degradation and component(s) contributing to the elevation in butyrate.

Acknowledgements

The authors thank the CRC for Food Industry Innovation for financial support and Dr Anders Henriksson for valuable discussions.

Ancillary