Secondary Logo

Journal Logo

Original Articles—Liver and Nutrition

A Human Flora-Associated Rat Model of the Breast-Fed Infant Gut

Edwards, Christine A.*; Rumney, C.; Davies, M.; Parrett, A. M.*; Dore, J.; Martin, F.; Schmitt, J.§; Stahl, B.§; Norin, E.; Midtvedt, T.; Rowland, I. R.; Heavey, P.; Köhler, H.#; Stocks, B.#; Schroten, H.#

Author Information
Journal of Pediatric Gastroenterology and Nutrition: August 2003 - Volume 37 - Issue 2 - p 168-177
  • Free


It is well established that early feeding practices, use of breast milk or formula, and the timing and pattern of weaning have important influences on disease incidence and development in early life (1,2). One area that has been little studied is the effect of diet on the colonization of the gut by microflora and its influence on the maturation of the intestinal epithelium and gut-associated immune system.

The factors that determine the colonization of the human colon are poorly understood. The critical periods of colonization are after birth and during the weaning period. The development of the colonic bacterial ecosystem at these times is important for the prevention of diarrhea in infants (1,3) and may be related to the development of respiratory (2) and allergic disease (4–6). It also may be critical for the establishment of the adult flora, and therefore may be involved in the prevention or causation of chronic diseases such as cardiovascular disease and cancer.

Investigations of the colonization of the human gut are hampered by ethical and practical problems. In vivo studies are very difficult, and detailed mechanistic studies require good models of the neonatal gut. At present, most studies of the metabolic activities of the infant microflora are performed in batch cultures of pure cultures of bacteria or neonatal feces (7,8). These are limited in their usefulness and in the applicability of results. The functions of the adult microflora, however, have been studied with several animal in vivo models, including gnotobiotic animals and human flora-associated mice and rats (9–11).

The aim of this study was to develop a model of the infant gut in order to perform detailed evaluations of infant formula and weaning food ingredients. A good animal model would enable a systematic approach to unravel the mechanisms behind gut maturation and the identification and development of foods that promote health in the children and adults. To study the effects of diet on the interaction between the colonic microflora and the gut mucosa, an animal model was needed. The colonization of the gut of children has usually been described by comparing the characteristics of feces of germ-free rats (assumed to be comparable to the sterile gut of the newborn infant) with those of individuals with conventional microflora. This method has allowed identification of several stages at which key activities associated with the microflora are established (12–15). However, the microflora of the infant rat is very different from that of the human infant.

A validated approach to study the effects of diet on the adult microflora is to use germ-free rats colonized with human fecal bacteria (11). In this study, a similar approach has been used with infant germ-free rats colonized with the fecal flora of exclusively breast-fed human infants (IHFA rats). The validity of the model has been evaluated by comparing the normal fecal flora, their metabolic functions, and subsequent interactions between pathogenic bacteria and intestinal mucosa in model animals and exclusively breast-fed human infants.


Human Flora-Associated Rat Model

The experiment was performed using two batches of 6 Fischer F344 rats. The rats were born germ free and were maintained in flexible film isolators. They were weaned at 3 to 4.5 weeks of age onto a modified infant formula milk, Aptamil (Milupa Research, Friedrichsdorf, Germany), supplemented with vitamins, minerals, and other nutrients to satisfy the nutrient requirements of the rats (renamed APTSUP;Table 1). The rats were caged in groups of three males or three females, and water was provided ad libitum. Having confirmed the germ-free status of the rats on two occasions, they were inoculated with human infant fecal flora 2 days after transfer to the milk diet.

Composition of modified infant formula fed to the infant human flora-associated rats

Human Fecal Inoculum

Fecal samples (3.27 g and 2.48 g) from two female, breast-fed infants, both 3 months old (batch 1), and fecal samples (0.41 g, 3.08 g, and 0.39 g) from three male breast-fed infants 1 week to 2 months of age (batch 2) were obtained from infants living within a 30-minute traveling distance of Yorkhill Hospital, Glasgow. The samples were processed within 1 hour of passage. For each batch of rats, the feces were combined in a bottle, flushed with oxygen-free nitrogen, and packed in ice before being flown to Heathrow airport. The container was stored at 4°C overnight. The sample was delivered to BIBRA TNO at 6:55 am, still under chill conditions. The sample was then prepared as a 10% suspension in sterile Brucella broth, an aliquot was removed for bacteriologic analysis, and the remainder was transferred to the animal unit by 8:00 am. All rats were dosed by 8:30 am with a 0.5-mL suspension applied by a 1-mL syringe into and around the mouth.

Bacterial Colonization

Five weeks later, when the rats were approximately 9 weeks old, fecal sampling commenced and continued for a total of 7 days for the enumeration of total anaerobes and Enterobacteria and for assay of bacterial products. Approximately 3 weeks after the last fecal sampling day (when the rats were approximately 13 weeks old), the rats were killed by asphyxiation with CO2. Gut contents and tissue samples were taken from all rats for the analyses described below.

Bacteriologic Analysis

Conventional Methodology

Fecal samples from rats were assessed for bacterial colonization by standard serial dilution techniques and were growth monitored on agar plates including Beerens, Rogosa, MacConkey Wilkins Chalgren, Mannitol salt and azide, and blood with crystal violet and were incubated under the appropriate anaerobic or aerobic conditions at 37°C.

Molecular Techniques

Extraction of Total RNA from Frozen Fecal Samples for Quantitative Dot-Blot Assessment of the Composition of the Microflora

All lab ware and solutions were prepared RNase-free using conventional techniques. Total RNA extractions from frozen fecal material were performed as described by Doré et al. (16). Mechanical disruption on a reciprocating shaker (Mini-beadbeater; Biospec Products, Bartlesville, OK) using 0.1 g of baked Zirconium beads (0.1 mm diameter) was applied to extract total ribonucleic acids from 0.2-g samples in 2.2-mL screw cap tubes (Sarstedt, Orsay, France) containing 100 μL SDS 20%, 400 μL phosphate buffer (12.3 mM Na2HPO4/87.7 mM NaH2PO4) pH 6.0, 400 μL sodium acetate 50 mM/EDTA 10 mM pH 5.1, and 400 μL phenol (Aquaphenol; Appligene, Illkirch, France), equilibrated with the acetate/EDTA buffer. Beating followed by heating at 60°C for 2 minutes each was repeated twice. Two phenol:chloroform:isoamylalcohol (100:24:1) and chloroform-isoamyl alcohol (24:1) extractions were sequentially performed using ReadyRed (Interchim, Montluçon, France). Total RNA was precipitated overnight at −20°C with sodium acetate 3M (0.1 volume) and cold absolute ethanol (2 volumes). After three washes in 80% (v/v) cold ethanol, dry pellets were resuspended in diethylpyrocarbonate-treated water. Quantification of nucleic acid extracts was determined by hybridization with the universal oligonucleotide probe Univ1390 (17). The reference was a serial dilution of standard Escherichia coli RNA extract (Roche, Meylan, France).

Probes and Labeling

Synthetic High Pressure Liquid Chromatography (HPLC) purified oligonucleotide probes (Appligene, Illkirch, France; Cybergene, St. Malo, France) were 5´-end 32P labeled with [γ-32P] adenosine tri-phosphate (ATP) (Amersham, Les Ulis, France) and polynucleotide kinase (Gibco-BRL, Eragny, France) as described by Doré et al. (16). The efficiency of labeling was assessed by thin-layer chromatography using polyethylene-imine cellulose plates and 0.5 M ammonium carbonate as mobile phase.

Total RNA Dot-Blot Hybridization

Dot-blot membranes were prepared as previously described (16) using 200 ng total RNA in triplicate from frozen fecal samples and serial twofold dilutions ranging from 250 to 1.95 ng total RNA from Bacteroides vulgatus ATCC 8482, Eubacterium siraeum ATCC 29066 or Fusobacterium prausnitzii ATCC 27766, Bifidobacterium longum ATCC 15707, Lactobacillus acidophilus ATCC 4356, Ruminococcus productus ATCC 27340, and the E. coli rRNA standard (Roche, Meylan, France). Seven identical replicates of hybridization membranes were prepared. Hybridizations were performed overnight at 42°C followed by stringent washing performed at the half-denaturation temperature for the duplex 16S rDNA-RNA as previously described (18). For each set of seven membranes, one membrane was hybridized with the probe recognizing all members of the domain bacteria, and the others with one of the specific oligonucleotide probes.

The probes used to characterize the microflora (19) targeted the following phylogenetic assemblages:Bacteroides spp (Bacteroides, Prevotella, Porphyromonas); enterics group (essentially E. coli and phylogenetically related microorganisms); Lactobacillus group (Lactobacillus, Leuconostoc, Weisiella, Enterococcus, Streptococcus, Lactococcus); genus Bifidobacterium (all known gut species); Clostridium leptum subgroup (including Clostridium leptum, C. sporosphaeroides, C. cellulosi, Eubacterium desmolans, E. siraeum, Ruminococcus flavefaciens, Fusobacterium prausnitzii); Clostridium coccoides group (including Clostridium nexile, C. coccoides, C. clostridiiforme, C. celerecrescens, C. xylanolyticum, Coprococcus eutactus, Eubacterium rectale, E. eligens, E. formicigenerans, E. hadrum, E. ventriosum, E. uniforme, E. cellulosolvens, Ruminococcus productus). Hybridization signals were quantified using an Instant Imager (Packard Instruments, Rungis, France). The abundance of groups was expressed as percent of total bacterial 16S rRNA quantified with the bacterial domain probe and is given as means from triplicate measurements.

Short-Chain Fatty Acid and Lactate Analysis

Short-chain fatty acids in cecal, and where enough material was available, in fecal and colonic contents, were measured by capillary column gas liquid chromatography of acidified ether extracts (20). Lactate was measured by packed column after methylation and ether extraction (21)

Preparation of Sample for Mucin and Tryptic Activity Determination

Feces were homogenized thoroughly with double the amount of saline, kept at +4°C for 2 hours and thereafter centrifuged at 35000 g for 30 minutes. Aliquots of the supernatant were used for mucin and tryptic activity analyses and for determination of cholesterol conversion.

Mucin Degradation

Aliquots of 5 μL were subjected to agar-gel electrophoresis on a glass plate, using 1% agarose in sodium barbiturate buffer pH 8.6 (22). The electrophoretic separation was run for 30 minutes, and immediately afterward, the plates were fixed in Carnoy solution and dried in a warm air stream. The plates were stained with toluidine blue, periodic acidic acid (for neutral mucins), amido black solution, and Coomassie blue. Mucin patterns similar to those found in germ-free animals, healthy newborns, or both were given score 3 and a pattern similar to that of adults were given score 0. Those with a pattern in between were defined as intermediate (score 1 and 2).

Conversion of Cholesterol to Coprostanol

For determination of conversion of cholesterol to coprostanol, samples were mixed with 95% ethanol and 10 M NaOH (1:2, v/v) and placed in a water bath at 60°C for 45 minutes. The hydrolysate was extracted twice with hexane, evaporated and analyzed with gas liquid chromatography (GLC) at 290°C (GCD-104; Pye Unicam, Cambridge, UK) on a glass column packed with 3% OV-17 (Supelco Inc, Bellefonte, PA, USA) (14). The conversion ratio was expressed as the percentage of coprostanol in the total amount of cholesterol plus coprostanol present. Values less than 5% were regarded as zero, due to detection limit.

Fecal Tryptic Activity

For determination of tryptic activity, 0.1 mL of respective supernatant was added to 2.9 mL 0.1 M Tris buffer, pH 8.2, containing 4.4 g calcium chloride/L. The reaction took place at 20°C and was initiated by adding 1.0 mL 0.003 M BAPNA (N-benzoyl-DL-arginine-4-nitroanilide hydrochloride). The reaction was stopped after 10 minutes by adding 0.6 mL 5 M acetic acid. Bovine pancreas trypsin type III diluted in 2 mM hydrochloric acid was used for construction of a standard curve. All samples and standards were analyzed in parallel with blanks at 405 nm spectrophotometrically (23). After correction for blank values, tryptic activity was calculated and expressed as milligrams of tryptic activity per kilogram of sample.

β Glucosidase and β Glucuronidase Activity

The pH of the diluted fecal samples was measured, and assays for both enzymes were conducted at a buffered pH closest to that of the samples. The suspensions were incubated aerobically at 37°C with p-nitrophenyl-β-D-glucopyranoside (3 mmol−1) or p-nitrophenyl-β-D-glucuronide (3 mmol−1) (Sigma-Aldrich Co. Ltd., Poole, UK) for assessment of β-glucosidase and β-glucuronidase, respectively. Release of p-nitrophenol was measured colorimetrically over time and was used as the measure of enzyme activity (24).

Ammonia Concentration

The concentration of ammonia in feces and cecal content was determined by spectrophotometer (25). Briefly, the fecal suspensions were diluted and centrifuged at 5000 g for 5 minutes. Then, 0.5 mL supernatant was added to 0.5 mL phenol-nitroprusside solution, 0.5 mL alkaline hypochlorite solution (Sigma-Aldrich Co. Ltd.), and 3.5 mL water and left at room temperature for 40 minutes. Absorbance was measured at 570 nm with reference to a standard curve.

Production of Ammonia

The rate of production of ammonia from endogenous nitrogen sources in the feces was determined according to Wise et al. (24). The fecal samples were diluted with 0.1 M potassium phosphate buffer (at a pH similar to that of the sample) to a final concentration of 20% (w/v). The suspensions were incubated at 37°C for 15 minutes and samples (0.7 mL) removed at 3-minute intervals and kept on ice. The samples were centrifuged at 5000 g/5 minutes and ammonia determined as above.

To determine whether the concentration of nitrogenous substrates in the fecal samples were limiting ammonia production, seven samples that exhibited low ammonia production rates were selected and the incubations were repeated with 0.5 mL 0.1% l-arginine and 0.1% urea (final concentrations, 1.5 μM and 0.5 μM; Sigma-Aldrich Co. Ltd.).

Measurement of Phenol and p-Cresol

The concentration of phenol and p-cresol in feces was determined by gas chromatography. One milliliter of fecal supernatant was heated with 0.1 mL internal standard solution (20 mmol−1 p-methoxyphenol), 1.0 mL methanol, and 0.2 mL 50% (v/v) H2SO4 in a 70°C water bath for 50 minutes. Distilled water (0.2 mL) and chloroform (0.8 mL) were added after the reaction mixture had cooled. The chloroform layer containing derivatized phenolic compounds was extracted, and 5 μL were injected onto a 25 m × 0.25 mm CP-Sil 5 CB column on a Perkin Elmer auto system gas chromatograph, model 970 (Norwalk, CT). The He carrier gas, H2, and air were set at flow rates of 25 cm per second, 45 and 450 mL per minute, respectively. The injection port temperature was 250°C and the detector temperature was 300°C. Identification of peaks was made by coincidence of retention times with the standards. Quantitative assessment was made by comparison of peak areas of test samples with those of authentic phenol and p-cresol standards.

Bacterial Adhesion to Intestinal Epithelial Cells

The ability of the gut contents and feces to inhibit the adhesion of pathogenic bacteria to epithelial cells was determined in a series of in vitro studies. Caco-2 cells (ACCC, Rockville, MD, USA), passage 30 to 45, were grown in modified Minimum Essential Medium Eagle (Sigma, St. Louis, MO, USA) in 6-well plates at 37°C and 5% CO2. Confluent monolayers were used in the following experiments after 11 to 14 days.

Enteropathogenic E. coli (EPEC DEF 46, kindly provided by Prof. Karch, Wuerzburg, Germany) was grown on nutrient blood (NB) agar. After an 18-hour incubation period at 37°C, bacteria were harvested with ice-cold 20 mM sodium borate buffer at pH 9.0 and washed twice with the same buffer. One milliliter of this suspension was labeled with 200 μg fluorescein isothiocyanate (Sigma Chemicals, St. Louis, MO) and surplus fluorescein isothiocyanate was removed by washing several times. Bacterial concentration was adjusted to 4 × 108 bacteria/mL. One gram of each fecal sample (5 rats 12 weeks old, 21 breast-fed infants 1–3 months old) was homogenized and diluted in 15 mL of 0.9% NaCl. After centrifugation to remove solid components, the supernatant was filtered sterile. Bacteria and stool supernatants, or phosphate buffered saline as a control, were preincubated together (250 μl each) for 15 minutes in an ice bath under gentle shaking. The Caco-2 cells were washed several times with phosphate buffered saline and bovine serum albumin to remove medium completely. The cells were then incubated with the bacteria and inhibitor solution for 1 hour in an ice bath in the darkness under gentle shaking. After the incubation period, the cells were washed three times with phosphate buffered saline and bovine serum albumin and fixated with paraformaldehyde 1%. Bacterial adhesion was determined with a fluorescence microscope by counting 20 randomly chosen fields. All assays were performed at least in duplicate.

In each individual experiment, the percentage of bacteria incubated with phosphate buffered saline only adhering to the monolayer were regarded as 100%, and subsequently the bacterial adhesion after pretreatment with the feces expressed according to this standard in percent.

Validation Against Infant Samples

The model was validated by comparing the measurements above with those obtained from fecal samples of 36 exclusively breast-fed infants younger than 3 months of age (23 females, 13 males) with a median age of 6 weeks (range, 1–12 weeks). Again, samples were from infants in Glasgow and all samples were processed and stored at −20°C or −70°C within 1 hour of passage before analysis as appropriate. The samples were processed and assayed in an identical manner to the rat fecal samples.

Comparison of the ammonia concentration and production, phenol and p-cresol concentration, and production of β-glucuronidase and β-glucosidase were compared in infant feces and rat colonic contents by Mann-Whitney test because the data was not normally distributed. Inhibition of bacterial adhesion to Caco2 cells by the rat cecal contents and human feces were compared by Student t test two sample unequal variance.


Viability of Human Fecal Samples before Inoculation of Rats

Enumeration of the major bacterial groups in the inoculum showed good survival of the component bacteria (Table 2), and the rats showed good association with fecal bacteria 5 days after inoculation.

Bacteria present in human infant fecal inoculum used for human flora-associated rats, measured at time of inoculation

Body and Tissue Weights of Rats

The animals all grew well in a comparable way with that of conventional animals (Table 3). There was a high degree of interindividual variation, with some very high weights of cecal contents as would be expected on liquid diets, but also some quite low amounts. There was very little colonic contents in many rats.

Body weight, weight gain, cecal and colonic content, and fecal weight and pH of infant human flora-associated rats colonized with breast-fed infant microflora

Bacterial Colonization of Rat Gut


RNA yields were always excellent. All rats had their fecal and cecal flora (collected, respectively, 6 and 9 weeks after association with a breast-fed infant's fecal flora) dominated by an association of bacteria belonging to the bifidobacteria, Lactobacillus group, and enterics (Fig. 1). In fecal samples, the corresponding proportions were 18% to 89% rRNA for Bifidobacterium, 5% to 33% for the Lactobacillus group, and 1% to 26% for the enterics. Only one rat had more than 80% rRNA for the Bacteroides group in the cecal sample. The overall additivity of probe signals indicated that the microflora of the rat samples was completely accounted for by the probes used.

FIG. 1.
FIG. 1.:
rRNA probes analysis of bacterial populations of (A) cecal contents (B) feces of infant human flora-associated (IHFA) rats.

Short-Chain Fatty Acids

The cecal and fecal short-chain fatty acids were dominated by acetic and lactic acid in a similar manner to the feces of the breast-fed infants (Tables 4 and 5). There were very low concentrations of propionic and n-butyric acids. The fecal and cecal short-chain fatty acid values were very similar.

Short-chain fatty acids in cecum and feces of infant human flora-associated rat model and feces of breast-fed infants
Molar proportion of short-chain fatty acids in cecum and feces of IHFA rat model and feces of breast-fed infants

Cecal and Fecal Ammonia, Phenol, and p -Cresol Concentrations

The concentrations of phenol, p-cresol, ammonia, and ammonia production were similar in both the infant and rat fecal samples as well as in the rat cecal samples (Table 6). Ammonia concentration and phenol and p-cresol concentration in rat cecal contents closely matched the levels seen in fecal samples from human infants. In addition, both ammonia and p-cresol concentrations in rat fecal samples also corresponded with those in infants, although there was a higher phenol concentration in the rat fecal samples when compared with the infants (P = 0.003). However, most of the individual results were in the same magnitude as the infants.

Comparison of metabolic activity in breast-fed infants and infant human flora-associated rats

β-Glucuronidase, β-Glucosidase, and Nitrate Reductase

The median and range of β-glucuronidase and β-glucosidase (GS) activities were higher in the infants compared with those levels found in fecal and cecal samples from the human fecal bacteria-associated rats (Table 7). β-glucuronidase was significantly higher in infant feces compared with fecal (P = 0.009) and cecal (P = 0.003) samples from the IHFA rats. Similarly, GS was significantly lower in the cecal contents from the IHFA rats when compared with the same activity on the feces of the breast-fed infants (P = 0.001). There was, however, no significant difference observed between the fecal samples from the infants and the fecal samples form the IHFA rats for GS activity.

Comparison of enzyme activity in breast-fed infants and infant human flora-associated rats

β-glucuronidase assays were performed at the pH of the sample. Interindividual variation between animals occurred, but in most cases, higher activity was apparent in cecal samples than in fecal samples, which may be partly attributed to the different ages (9 weeks for feces, 12 weeks for cecum) at which the two samples were taken. The pH was higher in samples taken from the female rats; this was particularly true for cecal pH. Females also showed higher β-glucuronidase levels in cecal samples than males, which agrees with the association of pH and β-glucuronidase activity observed in infant feces as in the infant samples was considerable.

The levels of GS in fecal samples often were very low or undetectable. As seen with β-glucuronidase activity, female rats had higher levels of GS in cecal samples than males. However, GS activity in samples from both sexes was much lower than the levels observed for β-glucuronidase.

Nitrate reductase was undetectable in fecal samples in all cases despite long incubation times and high sample concentrations. This possibly reflects very low activities in fecal material, although long sampling times causing problems with sample freshness may also be responsible.

Bacteria Adhesion to Mucosal Cells

The cecal content of the IHFA rats was able to inhibit the adhesion of enteropathogenic E. coli by nearly 50% (Fig. 2; P < 0.05). This was not as great as the inhibition caused by the infant fecal samples, which caused significantly greater inhibition (P < 0.05).

FIG. 2.
FIG. 2.:
Inhibition of adhesion of enteropathogenic Escherichia coli (EPEC) to Caco 2 cells by infant and infant human flora-associated rat fecal samples.

Mucin Degradation

The mucin analyses showed a pattern in between that of germ-free rats and microflora-associated characteristics. The mucin was not totally degraded by the flora, as is usually seen in conventional animals or healthy adult volunteers.

Cholesterol Conversion

The analyses of cholesterol showed a germ-free rat pattern. Conversion to coprostanol was found in only one animal.

Fecal Tryptic Activity

In the samples analyzed for fecal tryptic activity, very high levels of enzymatic activity were found. These were often unusually high but were similar to those previously seen in animals during an “establishment period.”


Previous studies of the adult human colonic microflora have used human fecal bacteria-colonized mice and rats (9–11). However, the flora of the infant gut is very different from that of the adult, and several functions associated with the adult microflora are slow to develop. For studies of the development of the infant microflora and of the effects of diet on the resident bacteria, a more relevant model is needed. This paper describes the first human flora-associated rat model of the infant gut. The bacterial flora of this model was shown to be reproducible and to mimic closely that of the breast-fed infant in terms of bacteriology, pH, and bacterial products from carbohydrate, protein, and other substrates. Since there is considerable variation in these measurements between individual human infants, it is difficult to model the system completely if only one fecal sample is used to inoculate the rats. This is partly overcome in our system by using more than one fecal sample to inoculate the rats and to use more than one batch of rats per experiment.

In a previous model (26,27), a mixture of bacteria isolated from human infants was used to inoculate germ-free mice. The mice were fed human milk and a variety of modified cows milk formulas for 14 days. This model produced a flora with large numbers of bifidobacteria as in our study, but the colonization of the mice was also dominated by clostridia and bacteroides species, which are not characteristic of our human infants. They did not use lactobacilli in their model, and again this is an important component of the flora of the human infant.

In our model, in which germ-free infant rats were associated with human infant feces, the survival of the bacterial flora was very representative of the human infants studied. In the inocula used in this study, there was a higher level of clostridia in the infant feces than previously reported (28,29). The reason for this is unclear, but is probably the result of the great variation in fecal flora of individual human infants. In the rat model, the clostridia were less dominant, reflecting the published data more closely.

The modified infant formula fed to the infant rats was successful in producing a breast-fed infant-style flora, even though the same formula fed to human infants does not. The only changes made to the formula before it was fed to the rats was an increase in protein content (in the same proportion of whey and casein as in the original formula) and the addition of vitamins and minerals as dictated by the needs of the growing rats. It is not clear how these additions could affect the colonization by the breast-fed–style flora. The rats were inoculated with feces from breast-fed infants. It is possible that some function of this flora, either the number of the bacteria or the bacterial products, may have facilitated the colonization process. In this case, it would suggest that the initial few days of breast milk may determine the survival of bacteria in the gut rather than the later feeding of breast or formula milk. Infants fed both human and formula milk have a flora that demonstrates metabolism between that of exclusively breast-fed or formula-fed infants (30,31).

The pattern of short-chain fatty acids in the breast-fed infant is dominated by acetic and lactic acid in contrast to that of the formula-fed infant, who has more propionic acid and less lactic acid. In the rat model described here, the cecal and fecal short-chain fatty acids were mainly acetic and lactic acid, matching those of the breast-fed infant feces. Butyric acid production is a property of adult colonic bacteria and, despite its perceived importance in protecting the colonic mucosa butyrate, does not appear to be important in infancy (13,32).

The ammonia concentration and cresol data from the IHFA rats showed a close similarity with that observed in the breast-fed infants. There was a higher phenol concentration in the rat fecal samples when compared with the infants; however, most of the individual results were in the same magnitude as the infants. All of these parameters were substantially lower than those reported in rats associated with an adult microflora (11,33).

Ammonia is considered to be potentially toxic to the intestinal mucosa and can reduce the lifespan of these cells. Mammalian cells can tolerate only modest concentrations of ammonia because of disruptions in intracellular pH (34).

Phenol and p-cresol are formed through the metabolism of the amino acid tyrosine. Phenols are usually detoxified by either sulphate or glucuronide conjugation in the colonic mucosa and then excreted in the urine or remain unabsorbed and excreted in the feces (35). p-Cresol has been implicated as a contributing factor in hyperactivity in children (36). Animal studies have shown that small amounts of phenol and p-cresol have growth-depressing effects in young pigs (37).

β-glucuronidase acts on ingested toxicants that have been conjugated to glucuronic acid in the liver and secreted into the gut via the biliary tract, releasing potentially toxic aglycones in the colon. Similarly, β-glucosidase can hydrolyze glucoside conjugates found in plants, releasing biologically active phytochemicals. The activity of these enzymes is not affected by method of delivery, but levels are much higher in formula-fed infants compared with those exclusively breast-fed infants (38). The metabolic data from the IHFA rats showed a close similarity with that observed in the breast-fed infants. Most of the individual results were in the same magnitude as the infants. The differences observed between enzyme activity in fecal samples from breast-fed infants and fecal and cecal samples obtained from the IHFA rats may be the result of the small numbers of samples collected from the rats. Levels in both the infants and IHFA rats were lower than enzyme activity in adults (Heavey P, unpublished data) and in adult human fecal bacteria-associated rats.

In these IHFA rats, there were microbes present that partly degraded the mucin content of the intestine. Previous investigation has shown that this function is most often established during the second part of the first year of life in human infants (12). This function is associated to microbes belonging to Peptostreptococcus, Bacteroides, Bifidobacterium, and Ruminococcus groups (39), some of which were present in this IHFA rat model. The tryptic activity levels of the feces of the IHFA rats were high but were at similar levels seen during conventionalization of germ-free animals by other means (15). These levels may reflect the change in diet, but also a disturbed intestinal microflora balance from the conventional animal. Cholesterol conversion to coprostanol was established in only one animal. The microbes responsible for this metabolic function, Eubacteria are normally established later in childhood, and therefore the IHFA rats reflect the profile of these activities in the young infant. Overall, the level of these bacterial activities was lower than in adults or conventional rats, as expected in comparison with the breast-fed infant. Many of these activities reflect the actions of bacteria that are not dominant organisms in the first weeks of life. These activities can be used as a marker of the diversification of the flora.

According to the World Health Organization, infectious diarrhea is still the leading cause of death in children younger than 5 years of age in developing countries, the problem being most severe during the first year of life. It accounts for a total mortality of 3.3 million deaths per year (40). In these countries, enteropathogenic E. coli is one of the most frequently found pathogens. In Brazil for example, enteropathogenic E. coli was isolated from stools of more than 40% of infants with acute diarrhea and was associated with a mortality of 7% (41).

The pathogenesis of enteropathogenic E. coli involves a process in which enteropathogenic E. coli binds initially to intestinal epithelial cells, described as localized adherence, and resulting in a so-called attaching and effacing lesion before finally invading the cell (42). Because the bacterial–cellular adhesion is the hallmark of the beginning of the infection, blocking this initial process is considered a protective characteristic of, for example, the luminal content.

We have shown earlier that the stools of newborn breast-fed babies are capable of inhibiting the adhesion of enteropathogenic E. coli to buccal epithelial cells (43). Here we not only used an intestinal epithelial cell line resembling the epithelium lining of the small bowel, but we also used an enteropathogenic E. coli wild-type strain aiming to mimic the in vivo conditions as closely as possible. The cecal content of the animals was able to inhibit the bacterial adhesion to the intestinal epithelial cell line by nearly 50% compared with the adhesion observed with untreated enteropathogenic E. coli. This was not quite the same extent of inhibition of adhesion found with stools from breast-fed infants, but clearly demonstrates the bioactivity of the investigated animal samples. Despite the better inhibition of bacterial adhesion by the stools of breast-fed infants, we consider this a relevant anti-infectious stool function in the animal model.

It has been demonstrated in this study that the infant human fecal bacteria-associated rat model developed here was feasible and that the microflora and its metabolism closely reflected that of the breast-fed infant in the first few months of life. There were very few differences between cecal and fecal samples that should help predict cecal conditions in the human infant. Greater differences would have been expected in the adult colon. This model should prove useful in the study of dietary and other factors that may determine bacterial colonization of the human infant where direct study of the child is impractical and or unethical.


All authors were involved in the design of the study and the writing of the paper. CE coordinated the project. The IHFA rat model was situated at BIBRA TNO (CR, MD). The diet for the animals was designed between BIBRA TNO (CR) and Milupa research (BS, JS). The inocula for the animals were provided by Glasgow University (CAE, AMP). Samples from the model were dispatched to the other authors for analysis. Short-chain fatty acids were measured by Glasgow (CAE, AMP); rRNA analysis by INRA (JD, FM); ammonia and bacterial enzymes by the University of Ulster (PH, IRR); mucin, tryptic activity, and coprostanol were measured by Karolinska Institute (EN, TM); tissue collection and bacterial adhesion studies were performed in Heinrich-Heine University, Dusseldorf (HS, BS, and HK).


1. Howie PW, Forsyth JS, Ogston SA, et al. Protective effect of breast-feeding against infection. Br Med J 1990; 300:11–6.
2. Wilson AC, Forsyth JS, Greene SA, et al. Relation of infant diet to childhood health seven year follow up of cohort of children in Dundee infant feeding study. Br Med J 1998; 316:21–5.
3. Isolauri E, Juntunen M, Rautanen T, et al. A human Lactobacillus strain casei sp strain GG strain promotes recovery from acute diarrhoea in children. Pediatrics 1991; 88:90–7.
4. Bjorksten B, Sepp E, Julke K, et al. Allergy development and the intestinal microflora during the first year of life. J Allergy Clin Immunol 2001; 108:516–20.
5. Kalliomaki M, Kirjavainen P, Eerola E, et al. Distinct patterns of neonatal gut microflora in infants in whom atopy was and was not developing. J Allergy Clin Immunol 2001; 107:129–34.
6. Ouwehand AC, Isolauri E, He F, et al. Differences in Bifidobacterium flora composition in allergic and healthy infants. J Allergy Clin Immunol 2001; 108:144–5.
7. Liftschitz CH, Wolin MJ, Reeds PJ. Characterisation of carbohydrate fermentation in faeces of formula fed and breast fed infants. Pediatr Res 1990; 27:165–9.
8. Parrett AM, Edwards CA. In vitro fermentation of carbohydrate by breast fed and formula fed infants. Arch Dis Child 1997; 76:249–53.
9. Narushima S, Ito K, Kuruma K, et al. Composition of cecal bile acids in ex germfree mice inoculated with human intestinal bacteria. Lipids 2000; 35:639–44.
10. Hirayama K. Ex germfree mice harboring intestinal microbiota derived from other animal species as an experimental model for ecology and metabolism of intestinal bacteria Exp Anim 1999; 48:219–27.
11. Mallett AK, Bearne CA, Rowland IR, et al. The use of rats associated with a human faecal flora as a model for studying the effects of diet on the human gut microflora. J Appl Bacteriol 1987; 63:39–45.
12. Midtvedt AC, Carlstedt-Duke B, Norin KE, et al. Development of five metabolic activities associated with the intestinal microflora of healthy infants. J Pediatr Gastroenterol Nutr 1988; 7:559–67.
13. Midtvedt AC, Midtvedt T. Production of short chain fatty acids by the intestinal microflora during the first 2 years of human life J Pediatr Gastroenterol Nutr 1992; 15:395–403.
14. Midtvedt A-C, Midtvedt T. Conversion of cholesterol to coprastanol by the intestinal microflora during the first two years of human life. J Pediatr Gastroenterol Nutr 1993; 17:161–8.
15. Midtvedt T, Carlstedt-Duke B, Hoverstad T, et al. Establishment of a biochemically active intestinal ecosystem in ex-germfree rats. Appl Environ Microbiol 1987; 53:2866–71.
16. Doré J, Sghir A, Hannequart-Gramet G, et al. Design and evaluation of a 16S rRNA-targeted oligonucleotide probe for specific detection and quantification of human faecal Bacteroides populations. Syst Appl Microbiol 1998; 21:65–71.
17. Zheng D, Alm EW, Stahl DA, et al. Characterization of universal small-subunit rRNA hybridization probes for quantitative molecular microbial ecology studies. Appl Environ Microbiol 1996; 62:4504–13.
18. Raskin L, Stromley JM, Rittmann BE, et al. Group-specific 16S rRNA hybridization probes to describe natural communities of methanogens. Appl Environ Microbiol 1994; 60:1232–40.
19. Sghir A, Gramet G, Suau A, et al. Quantification of bacterial groups within human fecal flora by oligonucleotide probe hybridization. Appl Environ Microbiol 2000; 66:2263–6.
20. Spiller GA, Chernoff MC, Hill RA, et al. Effect of purified cellulose, pectin, and a low residue diet on fecal volatile fatty acids, transit time and fecal weight in humans. Am J Clin Nutr 1980; 33:754–9.
21. Holdeman LV, Moore WEC. Anaerobic Laboratory Manual. 2nd ed. Blacksburg VA, USA: Virginia Polytechnic Institute; 1974.
22. Gustafsson BE, Carlstedt Duke B. Intestinal water-soluble mucins in germfree, ex-germfree and conventional animals Acta Path Microbiol Immunol Scand (B) 1984; 92:247–52.
23. Norin KE, Gustafsson BE, Midtvedt T. Strain differences in faecal tryptic activity of germ-free and conventional rats. Lab Anim 1986; 20:67–9.
24. Wise A, Mallet AK, Rowland IR. Dietary fibre, bacterial metabolism and toxicity of nitrate in rat. Xenobiotica 1982; 12:111–8.
25. Solorzano L. Determination of ammonia in natural waters by the phenol-hypochlorite method. Limnol Oceanogr 1969; 14:799–801.
26. Hentges DJ, Marsh WW, Petschow BW, et al. Influence of infant diets on the ecology of the intestinal tract of human flora-associated mice. J Pediatr Gastroenterol Nutr 1992; 14:146–52.
27. Hentges DJ, Marsh WW, Petschow BW, et al. Influence of a human milk diet on colonisation resistance mechanisms against Salmonella typhimurium in human faecal bacteria-associated mice. Microbial Ecol Health Dis 1995; 8:139–49.
28. Orrhage K, Nord CE. Factors controlling the bacterial colonisation of the intestine in breast fed infants. Acta Paediatr Suppl 1999; 88:47–57.
29. Sepp E, Naaber P, Voor T, et al. Development of intestinal microflora during the first month in Estonian and Swedish infants. Microbial Ecol Health Dis 2000; 12:22–6.
30. Bullen CL, Tearle PV, Willis AT. Bifidobacteria in the intestinal tract of infants: an in vivo study J Med Microbiol 1976; 9:325–33.
31. Parrett AM, Farley K, Fletcher A, et al. Comparison of faecal short chain fatty acids in breast fed, formula fed and mixed fed neonates Proc Nutr Soc 2001; 48:A.
32. Edwards CA, Parrett AM, Balmer SE, et al. Faecal short chain fatty acids in breast fed and formula fed babies. Acta Paediatr 1994; 83:459–62.
33. Djouzi Z, Andrieux C. Compared effects of three oligosaccharides on metabolism of intestinal microflora in rats inoculated with a human faecal flora. Br J Nutr 1997; 78:313–24.
34. Visek WJ. Diet and cell growth modulation by ammonia. Am J Clin Nutr 1978; 31:S216–20.
35. Ramakrishna BS, Gee D, Weiss A, et al. Estimation of phenolic conjugation by colonic mucosa. J Clin Pathol 1989; 42:620–3.
36. Adams RF, Murray KE, Earl JW. High levels of faecal p-cresol in a group of hyperactive children. Lancet 1985; ii:1313–8.
37. Yokoyama MT, Tabori C, Miller ER, et al. The effects of antibiotics in the weanling pig diet on growth and the excretion of volatile phenolic and aromatic bacterial metabolites. Am J Clin Nutr 1982; 35:1417–24.
38. Gronlund MM, Salminen S, Mykkanen H, et al. Development of intestinal bacterial enzymes in infants—relationship to mode of delivery and type of feeding. APMIS 1999; 107:655–60.
39. Carlstedt-Duke B, Midtvedt T, Nord CE, et al. Isolation and characterisation of a mucin degrading strain of Pepto Streptococcus from rat intestinal tract. Acta Path Microbiol Immun Scand (B) 1986; 94:293–300.
40. Ben C, Martins J, Zoya I, et al. The magnitude of the global problem of diarrhoeal disease: a ten year update. Bull WHO 1992; 70:705–14.
41. Fagundes-Neto U, Scaletsky IC. The gut at war: the consequences of enteropathogenic Escherichia coli infection as a factor of diarrhea and malnutrition. Sao Paulo Med J 2000; 118:21–9.
42. Donnenberg MS, Kaper JB. Enteropathogenic Escherichia coli. Infect Immun 1992; 60:3953–61.
43. Schroten H, Lethen A, Hanisch FG, et al. Inhibition of adhesion of S-fimbriated Escherichia coli to epithelial cells by meconium and feces of breast-fed and formula-fed newborns. Mucins are the major inhibitory component. J Pediatr Gastroenterol Nutr 1992; 15:150–8.

Infant human flora-associated rats; Intestinal bacteria; Infant microflora; Bacteria; Infants

© 2003 Lippincott Williams & Wilkins, Inc.