Resistant starch modifies gut microflora and microbial metabolism in human flora-associated rats inoculated with faeces from Italian and UK donors


Prof. Ian Rowland, Northern Ireland Centre for Diet and Health, University of Ulster, Coleraine, BT 52 1SA, UK (e-mail:


The effect of sucrose and resistant starch (‘CrystaLean’– a retrograded, amylose starch) on human gut microflora and associated parameters was studied in human flora-associated (HFA) rats, colonized with microfloras from UK or Italian subjects, to determine whether such floras were affected differently by dietary carbohydrates. Consumption of the resistant starch diet resulted in significant changes in four of the seven main groups of bacteria enumerated. In both the UK and Italian flora-associated rats, numbers of lactobacilli and bifidobacteria were increased 10–100-fold, and there was a concomitant decrease in enterobacteria when compared with sucrose-fed rats. The induced changes in caecal microflora of both HFA rat groups were reflected in changes in bacterial enzyme activities and caecal ammonia concentration. Although it had little effect on caecal short-chain fatty acid concentration, CrystaLean markedly increased the proportion of n-butyric acid in both rat groups and was associated with a significant increase in cell proliferation in the proximal colon of the Italian flora-associated rats. CrystaLean appeared to play a protective role in the colon environment, lowering caecal ammonia concentration, caecal pH and β-glucuronidase activity.

Starch is a dietary component considered to have an important role in colonic physiology and functions, and a potential protective role against colorectal cancer (Cassidy et al. 1994). It is known that a portion of dietary starch resists digestion by pancreatic amylase in the small intestine and thus reaches the colon. The amount of this undigested starch is called resistant starch (Englyst & Cummings 1985). At least three types of resistant starch are recognized: RS1, starch entrapped within a food matrix preventing access of amylase to the starch; RS2, starch with a granular structure resistant to digestion; RS3, retrograded starch formed by food processing (Englyst & Cummings 1985, 1986, 1987; Englyst et al. 1992). CrystaLean starch, a highly retrograded maltodextrin, used in the present study, falls into the RS3 category.

Resistant starch can be fermented by human gut microflora, providing a source of carbon and energy for the 400–500 species of bacteria present in this anaerobic environment (Englyst & Macfarlane 1986) and thus potentially altering the composition of the microflora and its metabolic activities. A recent study in which potato starch (RS2) was fed to rats demonstrated an increase in the intestinal population of bifidobacteria, lactobacilli, streptococci and enterobacteria (Kleessen et al. 1997). The fermentation of carbohydrates by anaerobic bacteria yields short-chain fatty acids (SCFA; primarily acetic, propionic and butyric acids) which can lower the pH in the lumen (Cummings 1984; Macfarlane & Cummings 1991). This acidification can affect the balance of the bacterial species, bacterial metabolic activity and product formation (Mallett et al. 1989) as well as affecting epithelial proliferation (Newmark & Lupton 1990).

A previous study (Rowland et al. 1998) on human flora-associated (HFA) rats fed on diets containing four different carbohydrates, suggested that a diet containing CrystaLean gave the highest protection against DNA damage in the colonic mucosa induced by the colon carcinogen 1,2-dimethylhydrazine. The aim of the present study was to compare two extremes of dietary carbohydrate source for their effect on the gut microflora. The investigations were based on determinations in rat caecal contents of the principal gut bacterial groups, bacterial β-glucuronidase and β-glucosidase activities, caecal ammonia and short-chain fatty acid concentration, caecal size and pH, together with measurements of colonic epithelial proliferation.

Sucrose was chosen as an absorbable simple carbohydrate and compared with a mixture of a digestible starch and the resistant starch ‘CrystaLean’.

The study was performed in human flora-associated (HFA) rats, which provide greater relevance to man than using conventional flora rats. Human flora-associated rats, obtained by colonizing germ-free rats with human faecal bacteria, develop and maintain a flora with bacteriological and metabolic characteristics similar to that of the mature human colonic microflora (Mallett et al. 1987; Rumney & Rowland 1992; Hirayama et al. 1995). Furthermore, the flora of the HFA rats has been shown to respond to fibre supplementation of the diet in a manner similar to that of the human and distinct from that of the conventional rat (Mallett et al. 1987).

An additional aim of the study was to determine whether microfloras derived from individuals of different countries respond in the same way to dietary modification. It has been reported that populations of different countries exhibit differences in bacterial composition of their colonic microflora (Benno et al. 1986; Moore & Moore 1995).

Materials and methods

Human flora-associated rats

Two pooled faecal suspensions were prepared as follows. Complete faecal samples were collected from six healthy human adults (three UK and three Italian) consuming a normal omnivorous diet and with no recent history (within 3 months) of gastrointestinal disturbance or antibiotic use. The Italian samples were transported from Italy in a sealed plastic bag under anaerobic conditions (Anaero Gen Sachet, Unipath Ltd, Basingstoke, UK) surrounded by ice packs. The samples were processed within 12 h of being passed. Both the Italian and the UK samples were transferred immediately to an anaerobic cabinet (Don Whitley Scientific Limited, Shipley, UK) and homogenized. Approximately 4 g of each sample were removed to a sterile tube, suspended at a concentration of 100 g l−1 (w/v) in anaerobic saline, and equal volumes of each suspension were mixed to provide the two pooled faecal suspensions. Samples of each pooled suspension were removed for bacteriological analysis and the remaining suspensions were then placed in a sterile, glass, screw-capped bottle to maintain anaerobic conditions and transferred with minimum delay to the germ-free isolators prior to dosing the rats.

Thirty-one germ-free Fisher (F344/N) rats (6–8-week-old, bred at BIBRA) were orally dosed with 1 ml of either of the two different pooled faecal suspensions to obtain HFA rats. The germ-free status of the rats was confirmed prior to inoculation with human flora by microscopic examination of faeces and by incubation of faecal samples in appropriate media under various incubation conditions. One week following the inoculation, colonization by the human flora was confirmed by microscopic examination of Gram-stained faecal material. The animals were maintained in flexible-film isolators throughout the experimental period until necropsy.

Diet and drinking water

Human flora-associated rats were given free access to tap water and to a non-purified rodent diet (R. & M. No.3 diet, Special Diet Service Ltd, Witham, UK). All water, diet and other materials taken into the isolators were sterilized by exposure to 5 Mrad of γ– radiation from a 60Co source. The rats were allowed 1 week for the flora to establish before being placed on the experimental diets for the following 4 weeks.

The experimental diets shown in Table 1 were a modification of AIN (American Institute for Nutrition) 76 in which the calcium concentration was reduced (based on that used by Caderni et al. 1991).

Table 1.  Composition of the diets for HFA rats
Ingredient  g kg−1 of diet
  • The diets were sterilized by irradiation before use.

  • *

    Obtained from Laboratorio Dottori Piccioni-Gessate, Milano, Italy.

  • Calcium free casein (‘Caseina latticu pura’).

Sucrose (beet)*4600
Globzeta starch*0310
CrystaLean starch0150
Corn oil231231
(+0·85 g l−1α-tocopherol acetate)
AIN mineral mix (low Ca)4040
AIN GF vitamins1212
Choline bitartrate22

Experimental design

The rats were divided into four groups as follows.

Group 1– (two males, five females), sucrose diet, colonized with UK faecal suspension (‘UK-HFA rats’);

Group 2– (three males, five females), CrystaLean diet, colonized with UK faecal suspension;

Group 3– (two males, six females), sucrose diet, colonized with Italian faecal suspension (‘Italian-HFA rats’);

Group 4– (three males, five females), CrystaLean diet, colonized with Italian faecal suspension.

The animals were maintained on the diets for 4 weeks.

Two hours before being killed, each rat was given an intraperitoneal injection of vincristine hydrochloride (Sigma) at a dose level of 1 mg kg−1 body weight to arrest metaphase in the cells of colon crypts (Goodlad & Wright 1982). The rats were killed by cervical dislocation and the colon and the caecum were removed. The caecal contents were weighed and the pH recorded before being used for analysis of gut microflora and determination of bacterial enzyme activities.

The colon was used for measuring epithelial proliferation.

Bacteriological methods

At necropsy, the caecum was removed and was immediately transferred, on ice, to an anaerobic chamber. The entire caecal contents were then removed to a sterile tube, the weight and pH recorded, and a portion used to prepare a 100 g l−1 suspension in anaerobic Brucella broth (Difco) and further serially diluted to 10–8 with the same diluent. From three appropriate dilutions, a 0·1 ml aliquot was spread in duplicate onto the following selective and non-selective agars (Oxoid, unless otherwise stated): Wilkins Chalgren agar containing 5% (v/v) defibrinated horse blood (for total anaerobic and total aerobic bacteria); Kanamycin-Vancomycin agar (for bacteroides) containing, l−1, 37·0 g Brain Heart Infusion, 5·0 g Yeast Extract, 20·0 g Bacteriological agar No.1, 0·5 g cysteine hydrochloride and supplemented with hemin (0·5 g l−1), Vitamin K (5 g l−1), 5% (v/v) defibrinated horse blood, 10 ml kanamycin (10 g l−1) and 7·5 ml vancomycin (1 g l−1); Rogosa agar (for lactobacilli) containing glacial acetic acid (1·32 ml l−1); Neomycin-Nagler agar (for clostridia) containing, l−1, 40 g Neutralized bacteriological peptone, 5 g di-sodium hydrogen orthophosphate (anhydrous), 1 g potassium dihydrogen orthophosphate, 1·86 g sodium chloride, 0·2 g magnesium sulphate (heptahydrate), 2 g glucose and 25 g Bacteriological agar No.1 (egg yolk emulsion 100 ml and 10 ml 20 μg l−1 neomycin sulphate were added after autoclaving); Beerens agar (for bifidobacteria) containing, l−1, 42·5 g Columbia agar (bioMérieux), 5·0 g glucose, 0·5 g cysteine hydrochloride, 1·5 g Bacteriological agar No.1 (final concentration 15 g l−1), 5 ml propionic acid (the pH was adjusted to 5·0 with NaOH); Mannitol salt agar (for staphylococci); MacConkey agar No.3 (for enterobacteria); Azide Blood-Crystal Violet agar (for streptococci) containing, l−1, Azide blood agar base, 4 ml 0·05% aqueous crystal violet solution and 5% (v/v) defibrinated horse blood (added after autoclaving).

All plating out was performed in the anaerobic cabinet and the inoculated agar plates were incubated in the anaerobic cabinet at 37 °C for 3 d, except for Beerens agar which was incubated for 5 d, and Wilkins-Chalgren agar for aerobes which was incubated at 37 °C in air for 2 d.

Determination of bacterial enzyme activities and ammonia concentration

Samples of caecal contents from rats were homogenized in anaerobic 0·1 mol l−1 potassium phosphate buffer, pH 7·0, to provide 100 g l−1 suspension. The suspensions, in screw-capped bottles, were incubated anaerobically with p-nitrophenyl-β-d-glucopyranoside (3 mmol l−1) or p-nitrophenyl-β-d-glucuronide (3 mmol l−1) (Sigma) for assessment of β-glucuronidase and β-glucosidase activities, respectively. Release of p-nitrophenol was measured colorimetrically over time and used as the measure of enzyme activity (Wise et al. 1982). The ammonia concentration was determined after centrifugation of the caecal suspension (7000 g for 5 min) using a reagent kit (Sigma) with ammonium chloride standards (Mallet et al. 1985).

Short-chain fatty acid analysis

A sample (0·5 g) of caecal contents was spiked with caproic acid (5 μmol g−1) as internal standard and acidified with 0·5 ml 50% H2SO4. The acidic solution was then extracted with 1 ml diethyl ether and centrifuged for 10 min at 3000 g. The ether extract was injected into a gas chromatograph (Sigma 3B, Perkin-Elmer, Norwalk, CT, USA) equipped with a flame ionization detector (FID) and a glass column (6 ft × 4 mm ID, packed with GP 10% SP-1000/1% H3PO4, on 100–120 Chromosorb WAW, Supelco, Hilversum, The Netherlands). The retention times for individual fatty acids were determined by injecting each standard separately on the column. Peaks were recorded on an integrator (model 1022, Perkin Elmer).

Determination of proliferation in the colon mucosa

Samples of colonic mucosa (distal colon and proximal colon) were also taken for investigations of rates of cell proliferation using isolated colonic crypts (Goodlad & Wright 1982). The mucosa samples were treated with EDTA (1 mmol l−1) at 37 °C for 50 min. The crypts were released by aspiration through the opening of a plastic Pasteur pipette, fixed in 10% Neutral Buffered Formalin (NBF), centrifuged at 500 g and then resuspended in 4% NBF. The isolated crypt suspensions were applied to poly l-lysine coated slides, air dried, exposed to absolute ethanol (15 min), 50% aqueous ethanol (15 min) and then tap water (10 min) before being treated with HCl (1 mol l−1) at 60 °C and stained by Feulgens stain. Cell proliferation was expressed as the number of arrested metaphases (following exposure to vincristine) in isolated crypts. Ten crypts per slide were examined.

Statistical analysis

The results were assessed for homogeneity using Bartletts test and for normality by the Kologorov-Smirnoff test. Where necessary, the data were transformed before being subjected to analysis of variance using the MINITAB Statistical Software Package (Mininc., State College, PA, USA). Individual means were compared for statistical significance by the two-sided least significant difference test.


Faecal samples from human volunteers

The bacterial composition of the UK and Italian gut microfloras used to inoculate the HFA rats is shown in Table 2.

Table 2.  Bacterial count of UK human faecal pooled suspension and Italian human faecal pooled suspension used for colonizing the germ-free rats
 UK faecal pool* Log10 cfu g−1Italian faecal pool* wet weight faeces
  • *

    Each pool was from three different human faecal samples.

Total aerobes 9·7 8·4
Total anaerobes10·410·6
Bacteroides 9·1 9·7
Lactobacilli 5·8 7·6
Bifidobacteria 6·8 8·1
Staphylococci 5·5 6·2
Enterobacteria 8·1 7·8
Streptococci 8·1 8·6
Clostridia 9·3 8·9

As the samples were pooled, it was not possible to perform a statistical analysis on the bacteriological data. In general, there were similarities in the total numbers of bacteria (anaerobic count) and in bacteroides, enterobacteria, streptococci and clostridia. Numbers of lactobacilli and bifidobacteria were 10–100-fold higher in the pooled Italian sample in comparison with the UK sample (Table 2).

Body weights, caecal weight and caecal pH value

Consumption of the CrystaLean diet was associated with a significant increase in terminal body weight of both the UK-HFA and the Italian-HFA male rats (Table 3). The female UK-HFA rats showed a significant decrease in terminal body weight on the CrystaLean diet, while female Italian-HFA rats were unaffected. The CrystaLean diet also increased significantly the caecal weights in both groups of rats (except for the UK females). Caecal pH was consistently decreased by up to 0·9 of a unit in rats consuming the CrystaLean diet, although not all the results were statistically significant (Table 3).

Table 3.  Body weight, caecal size and caecal content pH value of HFA rats colonized with UK microflora and Italian microflora and fed on sucrose and CrystaLean diets for 4 weeks
UK microfloraItalian microflora
  1. Values marked with an asterisk are significantly different from the sucrose diet (*P < 0·05; **P < 0·01; ***P < 0·001; analysis of variance, least significant difference test).

Body weight (g)320 (14)265 (9)396 (10)***219 (23)***315 (7)215 (10)367 (16)***230 (8)
Caecal size (g)4·86 (0·36)3·31 (0·41)7·06 (0·68)***3·69 (0·42)4·03 (0·62)2·81 (0·33)6·82 (0·90)***5·68 (0·56)**
Caecal content pH6·78 (0·14)7·22 (0·37)6·66 (0·16)6·45 (0·35)**6·91 (0·04)7·15 (0·63)6·06 (0·09)***6·84 (0·14)

Bacteriological analysis

Bacterial counts of caecal content (g−1) from HFA rats colonized with UK microflora and Italian microflora and fed on sucrose diet and CrystaLean diet for 4 weeks are shown in Table 4.

Table 4.  Bacterial count per gram of caecal content of HFA rats colonized with UK microflora and Italian microflora and fed on sucrose and CrystaLean diets for 4 weeks
UK microfloraItalian microfloraStatistical analysis
  1. ‡Log10 cfu g−1 wet weight of caecal content; §frequency of occurrence = 29%∥frequency of occurrence = 13%.

  2. Values marked with an asterisk are significantly different from the sucrose diet (*P < 0·05; **P < 0·01; ***P < 0·001; analysis of variance, least significant difference test).

  3. Significant differences between sex, diet and flora (analysis of variance) are indicated by daggers (P < 0·05; ††P < 0·01; †††P < 0·001).

Total aerobes8·07 (0·32)9·06 (0·60)***8·41 (0·24) 8·06 (0·68)††NS†††
Total anaerobes9·87 (0·28)9·63 (0·25)*9·79 (0·22)10·00 (0·08)*NSNS††††
Bacteroides9·04 (0·26)8·70 (0·17)*9·15 (0·23) 9·29 (0·40)NSNS††
Lactobacilli5·84 (0·82)7·89 (0·45)***5·36 (1·36) 7·69 (0·24)***††††††NSNS
Bifidobacteria6·89 (1·14)7·91 (0·56)**6·76 (0·88) 8·69 (0·52)***†††NSNS
Staphlococci7·36 (0·57)8·73 (0·45)***7·95 (0·59) 7·15 (0·48)***NS††††††
Enterobacteria7·17 (0·15)6168 (0·37)**7·83 (0·30) 6·10 (0·40)***NS†††NS†††
Streptococci6·65 (0·49)§7·05 (1·17)**7·64 (0·63)∥ 7·24 (1·04)NS††
Clostridia8·77 (0·34)9·16 (0·29)8·91 (0·77) 8·85 (0·19)NSNSNSNS

In the UK-HFA rats, the CrystaLean diet significantly increased the total aerobes (P < 0·001) and decreased the total anaerobes (P < 0·05) compared with the sucrose diet. Furthermore, all the principal groups of gut bacteria, except for clostridia, were significantly influenced by the CrystaLean diet. Feeding the CrystaLean diet was associated with a significant increase in lactobacilli (100-fold increase) and in bifidobacteria and staphylococci (10-fold increase compared with the sucrose diet group). Streptococci were also significantly higher in the CrystaLean-fed rats and in addition, the percentage of carriage rose from 29% (2/7) in the sucrose group to 100% in the CrystaLean-fed rats. Bacteroides and enterobacteria, in contrast, were significantly lower in the CrystaLean-fed rats.

In the Italian-HFA rats, the CrystaLean diet significantly increased the total anaerobes (P < 0·05), but the aerobes were not significantly decreased (Table 4). Four out of seven of the main groups of gut bacteria investigated were significantly affected by the CrystaLean diet (P < 0·001). Both lactobacilli and bifidobacteria were increased 100-fold compared with those in the sucrose diet group. Staphylococci and enterobacteria were, in contrast, significantly decreased. Although CrystaLean feeding did not significantly alter the mean number of streptococci, the frequency of carriage in the rats was much higher (100%) in comparison with 13% in the sucrose group.

Although there were sex differences in some cases in both groups of rats, they were relatively minor and the pattern of response to sucrose and CrystaLean diets was similar in males and females.

The analysis of variance indicated significant overall differences in the two types of gut microflora (UK and Italian), particularly bacteroides, staphylococci, streptococci and total anaerobes. More important significant diet–flora interactions were observed in six of the nine bacterial groups: total aerobes, total anaerobes, bacteroides, staphylococci, enterobacteria and streptococci. The most marked effects were seen with the viable count of staphylococci, which were significantly increased by CrystaLean feeding in the UK-HFA rats and significantly decreased (P < 0·01) in the Italian-HFA rats.

Caecal enzyme activity and ammonia concentration

Gut microflora-associated activities (Table 5) were affected by the diet in both groups of rats. β-Glucosidase activity was significantly increased by the CrystaLean diet in both the UK-HFA rats and the Italian-HFA rats (Table 5). In contrast, β-glucuronidase activity was significantly (P < 0·001) decreased in the UK-HFA rats fed the CrystaLean diet, but there were no significant dietary differences in Italian-HFA rats. However, the β-glucuronidase activity in caecal contents of UK-HFA rats fed the sucrose diet was considerably higher than in the Italian-HFA, sucrose-fed rats.

Table 5.  Caecal enzyme activities in HFA rats colonized with UK microflora and Italian microflora and fed on sucrose and CrystaLean diets for 4 weks
UK microfloraItalian microfloraStatistical analysis
Sucrose X–(s.d) CrystaLean X–(s.d) Sucrose X–(s.d) CrystaLean X–(s.d) SexDietFloraDiet–flora interaction
  1. Values marked with an asterisk are significantly different from the soucrose diet (*P < 0·05, **P < 0·01; ***P < 0·001; analysis of variance, least significant difference test).

  2. Significant differences between sex, diet and flora (analysis of variance) are indicated by daggers (P < 0·05; ††P < 0·01; †††P < 0·001).

μmol h−1
63·02 (16·91)38·98 (11·00)***28·41 (9·89)25·76 (6·03)NS†††††
μmol h−1
2·05 (0·74)3·23 (1·41)*1·55 (0·28)4·15 (2·03)***††††††NSNS
Caecal ammonia
μmol g−1
172·90 (77·70)109·26 (21·82)*166·75 (27·60)113·7 (33·00)**NS††NSNS

The caecal ammonia concentration was significantly decreased by 30–40% in the UK-HFA and Italian-HFA rats (Table 5).

Caecal SCFA concentration and profile

The major SCFAs produced in the caecum were acetic, propionic and n-butyric acids; data for the minor, branched-chain fatty acids (i-valeric, i-butyric acids), which were present in very low concentrations, are not shown.

The caecal concentration of total SCFA was not significantly affected by the CrystaLean diet in both types of HFA rat (Table 6). However, there were marked differences in the proportions of the main SCFA. In general, the proportions of acetic and n-butyric acids were higher, and that of propionic acid, lower, in the UK-HFA rats than in the Italian-HFA rats for any given diet (Table 6). The feeding of CrystaLean was associated with a major change in the SCFA profile in both HFA groups, with a marked and highly significant decrease in proportion of propionic acid and a corresponding increase in n-butyric acid (Table 6).

Table 6.  Caecal concentration of SCFA in HFA rats colonized with UK microflora and Italian microflora and fed on sucrose and CrystaLean diets for 4 weeks
UK microfloraItalian microfloraStatistical analysis
SexDietFloraDiet–flora interaction
  1. Values marked with an asterisk are significantly different from the sucrose diet (*P < 0·05; ***P < 0·001; analysis of variance, least significant difference test).

  2. Significant differences between sex, diet and flora (analysis of variance) are indicated by daggers (P < 0·05; ††P < 0·01; †††P < 0·001).

Total SCFA mmol l−138·21 (16·32)37·08 (8·32)32·56 (6·09)36·43 (6·71)NSNSNSNS
% Acetic acid65·03 (5·67)66·36 (7·69)57·39 (4·16)62·57 (4·58)*NS††NS
% Propionic acid27·61 (5·10) 6·94 (1·93)***35·95 (4·03)18·37 (3·89)***NS††††††NS
% Butyric acid 4·02 (1·00)25·63 (8·67)*** 2·79 (0·55)18·23 (3·21)***NS††††††NS

Proliferation in the colon mucosa

Figure 1 shows the results from the proliferation study on the proximal and distal colon.

Figure 1&.

emsp;Mean number of mitoses per crypt in the proximal and distal colon portions of UK-HFA rats and Italian-HFA rats fed on two different diets for 4 weeks. Values shown are means for 6–8 rats, bars show S.E.M. Statistical comparisons are shown for sucrose and CrystaLean data within each microflora group (**P < 0·01, analysis of variance, least significant difference). (□), Proximal colon; (▪), distal colon

The carbohydrate content of the diet did not seem to affect the proliferation in the distal colon. There was only a non-significant increase in the UK-HFA rats on the CrystaLean diet compared with those on the sucrose diet. Proliferation in the proximal colonic mucosa was, however, significantly affected by the carbohydrate fed, although an increase in proliferation in response to the CrystaLean diet was observed only in the Italian-HFA rats. A comparison of the two microflora groups revealed no differences in proliferation in the distal colon, but the proximal colon of Italian HFA rats fed sucrose exhibited lower proliferation, and those fed CrystaLean, higher proliferation, than their corresponding UK-HFA group (Fig. 1).


Although it was not possible to compare the Italian and UK faecal floras on a statistical basis as the samples were pooled, overall differences and similarities did emerge. The most notable differences were seen in the lactobacilli and bifidobacteria, the counts of which were over one order of magnitude higher in the Italian flora, and in the aerobic count, which was higher in the UK flora. In contrast, there were similarities in the total number of bacteria and in bacteroides, enterobacteria, streptococci and clostridia in the total anaerobic count. It is unlikely that these differences were due to differences in bacterial survival during the period before the faecal samples were processed. The UK samples, which were processed within 2 h of being passed, had, in fact, slightly lower counts of strict anaerobes (total anaerobes, bifidobacteria and bacteroides) than the Italian samples which were processed within 12 h. This indicates that the anaerobic conditions during transport were satisfactory. The Italian samples were also kept chilled throughout the transport period, thus minimizing multiplication of the bacteria.

The data suggest, therefore, that there may be differences in the major bacterial groups between Italian and UK faecal microfloras, although caution should be exercised in drawing such a conclusion due to the limited number of samples.

The germ-free rats were inoculated with pooled faecal suspensions derived from UK or Italian volunteers. We chose to use pooled samples, rather than individual faecal samples, in an attempt to avoid modelling just one individual’s microflora and to provide results of more general applicability. Although pooling could have led to an unnatural balance in microbial types, the proportions of the major groups in the faecal samples were consistent with previous reports (Rowland et al. 1985; Terada et al. 1992).

The microfloras that developed in the HFA rats were broadly similar to the inocula used; there was a similar anaerobe/aerobe ratio and all the major bacterial groups were present. The exception was streptococci, where the frequency of occurrence for the sucrose-fed rats was low (< 29%).

The statistically significant diet–flora interaction (Table 4) for six of the nine groups of organism studied strongly indicates that different human microflora may respond in different ways to dietary change. The total aerobic count was increased by CrystaLean feeding of the UK-HFA rats, but it was unchanged in the Italian-HFA rats. The total anaerobes behaved in an opposite fashion, with stimulation by CrystaLean being seen only with the Italian microflora rats. Bacteroides and streptococci were not altered by diet in Italian-HFA rats, but they were reduced or enhanced, respectively, in UK-HFA rats. Numbers of staphylococci increased in UK-HFA rats but decreased in Italian-HFA rats.

The most striking and consistent effect of diet on microflora was seen in the three bacterial groups lactobacilli, bifidobacteria and enterobacteria. In these cases, the microflora in both the UK- and Italian-HFA rats responded in the same way. There was a significant increase in lactic acid-producing bacteria and a concomitant decrease in enterobacteria, although the latter effect was more pronounced in the Italian-HFA rats. It is noteworthy that there were significant differences between male and female rats in the bacterial counts for some groups (Table 4). In particular, the numbers of lactobacilli and bifidobacteria were consistently lower, by 1–1·5 orders of magnitude, in sucrose-fed female rats (both UK and Italian flora-associated) than in males (data not shown). CrystaLean consumption, however, was associated with an increase in numbers of both these bacterial groups in both sexes.

It is interesting to compare our results with those of Kleessen et al. (1997) who studied the effect of resistant starch from potatoes on the caecal flora of rats. The feeding of potato starch, which is an RS2 type, led to increases in the intestinal population of total anaerobes, bifidobacteria, lactobacilli, streptococci and enterobacteria. It is possible therefore that RS2 and RS3 starches have different effects on the gut microflora.

It would appear from our study that CrystaLean specifically stimulated lactobacilli and bifidobacteria and in this way, behaved in a manner similar to many oligosaccharidic substances such as fructooligosaccharides and galactooligosaccharides (Terada et al. 1992; Rowland & Tanaka 1993; Wang & Gibson 1993). An increase of bifidobacteria in the intestinal tract is thought to exert a beneficial effect to the host by production of SCFA, lowering the pH in the bowel, decreasing putrefactive products, inhibiting the growth of potential pathogens, and immunopotentiation (Fuller 1989; Rowland et al. 1985; Mitsuoka 1990).

The promotion of bifidobacteria may suppress undesirable micro-organisms such as enterobacteria, many of which, including some types of Escherichia coli, have pathogenic potential (Rubio et al. 1995) In fact, in the present study, a significant fall in enterobacteria numbers occurred in both UK- and Italian-HFA rats fed the CrystaLean diet.

Further evidence that CrystaLean may beneficially modify the microflora may be gleaned from our data on the changes in metabolic activity in the gut. The enzymes investigated in the present study play an important role in the generation of toxic and carcinogenic metabolites from dietary and endogenously produced substances.

β-Glucuronidase is considered to be a major gut microflora enzyme and has been demonstrated to be inducible by glucuronide conjugates and bile flow (Robertson et al. 1982; Mallett et al. 1983). It plays an important role in the hydrolysis of xenobiotic glucuronides, leading to potential toxic consequences for the colonic mucosa (Larsen 1988; Rowland 1995). In our study, there was a decrease in β-glucuronidase in both groups of rats fed the CrystaLean diet (although statistically significant only in the UK-HFA rats).

β-Glucosidase activity, by contrast, was significantly increased in both groups of CrystaLean-fed rats. The increase could be a consequence of the stimulation of bifidobacteria and lactobacilli, which posses high levels of β-glucosidase, as reported previously (Saito et al. 1992; Rowland & Tanaka 1993). It is noteworthy that the activity of β-glucuronidase was higher in female rats than in males, although the degree of stimulation of the enzyme activity by CrystaLean was similar in both sexes.

With regard to the caecal ammonia concentration, we observed a significant reduction in both UK-HFA and Italian-HFA rats fed the starch diet. This is consistent with other animal (Mallett et al. 1988; Levrat et al. 1991) and human (Birkett et al. 1996; Heijnen et al. 1997) studies, which have reported a reduction in caecal ammonia concentrations after supplementing the diet with resistant starch. Ammonia is a product of deamination of peptides by the caecal microflora (Vince et al. 1983; Macfarlane & Cummings 1991) and it has been thought to play a role in carcinogenesis and other disease processes (Visek 1978; Lin & Visek 1991).

The lack of effect of CrystaLean on the total SCFA concentration in the caecum was unexpected, given the known differences in digestibility of sucrose and the resistant starch in the upper gut. However, it is clear that total amount SCFA in the gut was higher in rats on the CrystaLean diet as the caecal weight of the CrystaLean-fed rats was significantly greater than that of the sucrose-fed animals. The effects of CrystaLean on the SCFA profile in the caecum is consistent with previous studies in humans, which usually show an increase in the proportion of n-butyric acid in faeces after consumption of resistant starch (Scheppach et al. 1988; Van Munster et al. 1994; Phillips et al. 1995; Cummings et al. 1996). In the present study, the increase in n-butyric acid was very large, probably because SCFAs were being measured in the caecum rather than in the faeces, hence circumventing problems of SCFA absorption and utilization in the colon. It should also be noted that the increase in n-butyric acid was almost exclusively at the expense of propionic acid rather than acetic acid.

The pattern of SCFA produced in the CrystaLean-fed rats did not appear to be consistent with the changes in bacterial flora, namely, an increase in numbers of lactobacilli and bifidobacteria. These two groups produce predominantly acetic acid, rather than butyric acid, from carbohydrate (Holdeman & Moore 1973). The most likely explanation of this discrepancy is that the resistant starch induced changes in other gut microbial groups that were not detected by the bacteriological methods used. In particular, groups such as the fusobacteria and eubacteria, for which no selective media are available, do contain species (e.g. Fusobacterium nucleatum, Eubacterium multiforme, Eu. limosum) that produce n-butyric acid (Holdeman & Moore 1973). It is also feasible that although total numbers of clostridia in the gut were not significantly affected by CrystaLean, the resistant starch diet may have favoured certain species, such as Clostridium butyricum, which exhibit enhanced butyric acid production compared with other species.

Epidemiological studies and experimental data have correlated colon proliferation with risk of colon cancer (Lipkin 1988) and a low rate of epithelial proliferation is thought by some to be protective (Becker 1981). The results from the proliferation study showed that the distal colon had a higher rate of mucosal cell proliferation that the proximal colon. The dietary effects, however, were inconsistent between the UK- and Italian-HFA animals. CrystaLean slightly increased proliferation in the distal colon of the UK-HFA rats. This is consistent with a previous study (Rowland et al. 1998) although in the latter, the increase was larger and statistically significant. In the proximal colon, the CrystaLean diet was associated with a significant increase in mucosal proliferation in the Italian-HFA rats, but not in the UK-HFA animals.

This study demonstrated that a resistant starch could markedly modify the human gut microflora, particularly stimulating lactic acid bacteria and decreasing potentially pathogenic types such as enterobacteria. Furthermore, the concomitant decrease in caecal ammonia concentration, caecal pH value and β-glucuronidase activity, suggests that the RS diet may have beneficial effects related to reduced formation of potentially toxic bacterial metabolites in the colon. The type and extent of microflora modification and its potential associated health benefits is, however, somewhat dependent on the host’s initial flora, suggesting that further work may be required to establish the extent of diet–microflora interactions.


The authors thank James Coutts and Antonia Davies for their excellent technical assistance, and Rob Gilbert and Kate Green for their assistance with the HFA rats. This study was supported by grants EC-AIR-2-CT94–0933 and EC-AIR-1-CT94–7122.


  • *

    Present address: Northern Ireland Centre for Diet and Health, University of Ulster, Coleraine, UK.