The findings to be reported in this and the following papers reveal that mice and other mammals normally harbor an extensive bacterial flora not only in the large intestine, b u t also in the stomach and small intestine. Although this flora plays an essential role in the development and well being of its host, its exact composition is not known. In fact, several of its most important and numerous components are usually overlooked in bacteriological studies because they have exacting g r o w t h requirements, and because their susceptibility to the antagouistic activity of other microbial species makes it difficult to cultivate them in vitro.We shall describe in the present paper: (a) the methods used in our laboratory for quantitative bacteriological studies of the gastrointestinal tract; and (b) the development of the gastrointestinal flora as observed in newborn mice until the time of weaning. Materials and MethodsExperimental Animals.--Extensive studies of the gastrointestinal flora have been carried out during the past 6 years with socalled Swiss mice of the NCS colony, maintained in our laboratory under the conditions described earlier (1-3); mice from other colonies, raised under conventional conditions, were also used for comparative studies.The animals were housed, and allowed to mate, in stainless steel cages, with autoclaved wood shavings as litter. The drinking water, given ad lib., was acidified with HC1 to control bacterial contamination as described earlier.In certain experiments, the same end was achieved by adding an organic silver compound in a dilution of 1 part of silver per million. The food also given ad lib. consisted either of pasteurized pellets (supplied by Dietrich and Gambrill, Frederick, Maryland) or of a semisynthetic complete diet containing 15 per cent casein (described in reference 4). It is worth emphasizing that both the pellets and the casein diets had been so treated as to eliminate all bacterial contaminants except a few spores.Preparation of Specimer~ for Bacteriological Examination.--Stool specimens were collected from mice on sterile paper between 9:00 a.m. and 10:30 a.m. Approximately 0.1 gm of stool was emulsified in 5 ml of sterile diluent (charcoal water) by 4 minutes agitation on a mechanl-

Colonization of the gastrointestinal tract by bacteria of the normal flora was followed by bacteriological and special histological techniques in mice from several colonies. These histological techniques were designed to preserve the intimate associations that become established between particular strains of microorganisms and the epithelium of the mucosa of certain areas of the gut. The findings were as follows: 1. The various strains of bacteria of the normal flora became established in the different areas of the guts of infant mice according to a definite time sequence. 2. The first types of bacteria that could be cultured from the gut were lactobacilli and Group N streptococci. Within the first day after birth, these bacteria colonized the entire digestive tract and formed layers on the stratified squamous epithelium of the nonsecreting portion of the stomach and of the distal esophagus. 3. The bacterial types that appeared next were coliforms and enterococci. From about the 9th to the 18th day after birth, these bacteria could be cultured in extremely high numbers from the cecum and the colon. Histological sections of those organs taken during the first 2 or 3 days of that interval revealed microcolonies of Gram-positive cocci in pairs and tiny Gram-negative rods embedded in the mucous layer of the epithelium. The microcolonies were well separated from the mixture of digesta and bacteria that occupied the center of the lumen; they may have consisted of the coliforms and enterococci mentioned above; but this possibility remains to be proved. 4. Histological sections also revealed that, at about the 12th day after birth, long, thin Gram-variable rods with tapering ends were present, side by side, with the small Gram-negative rods and Gram-positive cocci in the mucous layer. By the 15th day after birth, the fusiform bacteria formed thick layers in the mucus, and seemed to be the only bacteria remaining in that location. It has not yet been possible to enumerate these tapered rods by culture methods, but as judged by visual appearances in the histological sections, they seemed to outnumber all other bacteria in the cecum and the colon by a factor of as much as 1000. It must be stressed that these bacterial layers are readily disrupted and even washed away by conventional histological techniques; their discovery was largely due to the use of the special histological techniques described in the text. The bacteriological and histological findings described here constitute further evidence for the hypothesis that symbiotic associations exist between microorganisms and animals, and that a very large percentage of the bacteria in the gastrointestinal tract constitutes a true autochthonous flora. The constant occurrence of several distinct associations of bacteria with the special histological structures of the animal host renders obsolete the notion that the intestine constitutes a chemostat in which the bacterial populations are randomly mixed. For a full understanding of the ecology of the normal microflora, it is necessary to think of body surfaces as distinct microenvironments in which virtually pure cultures of a few species of microorganisms interact with their host and the adjacent microbial populations. Experiments based on this hypothesis are admittedly difficult to design, but on the other hand studies based on the assumption that microorganisms exist as mixtures in the gastrointestinal tract will be only of limited value and may often be misleading.

Several species of animals have been raised and made to reproduce under germfree conditions. While the animals so produced seem to have a normal fife span, they exhibit histological, anatomical, and physiological characteristics , . J

Comparative studies of albino mice obtained from various colonies have revealed that the so called indigenous (or normal) microbiota of these animals exerts a profound influence on their rate of growth, their efficiency in the utilization of food, and their resistance to infection, toxic substances, and other stressful agencies (1-4). Indeed, many attributes of mice which are characteristic of the colonies from which the animals were derived are in reality determined not by genetic endowment, but by the microbiota prevailing in the colony. The very statement of this fact illustrates the ambiguity of the phrase "normal microbiota". This phrase merely denotes a multiplicity of microbial types which happen to be associated with a given animal population, but it does not imply that such organisms are necessarily present in other populations of the same animal species. As we shall see later, the same kind of ambiguity applies to the phrase normal microbiota when applied to man.The study of the intestinal bacterial flora has necessitated the development of special bacteriological techniques, designed for the quantitative enumeration of bacterial types which are extremely abundant in vivo, but which commonly fail to grow on the usual culture media under aerobic conditions. These techniques, which are described in the preceding paper, have enabled us to follow the trend of the bacterial population in the gastrointestinal tract of mice from the time of birth (5).In the present paper, we shall focus attention on some unexpected findings concerning the composition and distribution of the indigenous flora in adult animals. We shall show in particular that (a) the whole gastrointestinal tract harbors throughout life an abundant bacterial flora, the composition of which is characteristic for each section of the tract; and (b) some bacterial species are intimately associated with the wall of the various organs.

Adult mice from seven different colonies were studied with regard to (a) the numbers and types of bacteria that could be cultivated from their stools; (b) their resistance to the lethal effect of endotoxins prepared from three strains of Gram-negative bacilli. See PDF for Structure In six of the seven colonies, the stools yielded large numbers of various types of lactobacilli, enterococci, and Gram-negative bacilli. Most animals in these colonies died within 48 hours following injection of endotoxin. The other mouse colony (NCS) has been maintained for the past three years at the Rockefeller Institute under exacting sanitary conditions; it is free of many types of common mouse pathogens. The stool flora of NCS mice yielded very large numbers of viable lactobacilli (109 per gm), representing at least three different morphological types. In contrast, it contained only few enterococci and Gram-negative bacilli (less than 106 per gm). Moreover, E. coli, Proteus sp., and Pseudomonas sp. could not be recovered from the stools under normal conditions. NCS mice proved resistant to the lethal effect of endotoxins. These characteristics of the NCS colony prevailed whether the animals were housed continuously in individual cages on wire grids, or grouped continuously in large cages with wood shavings as litter. However, the composition of the bacterial flora could be rapidly and profoundly altered by a variety of unrelated disturbances such as sudden changes in environmental temperature, crowding in cages, handling of the animals, administration of antibacterial drugs, etc. The first effect of the change was a marked decrease in the numbers of lactobacilli and commonly an increase in the numbers of Gram-negative bacilli and enterococci. When tested 3 weeks after these disturbances some NCS animals were found to have become susceptible to the lethal effects of endotoxin.

Mice delivered by Caesarian section were used to develop a new mouse colony which has been maintained in an environment protected from contact with common mouse pathogens, but not in the germ-free state. These mice, designated as NCS, were compared with animals of the same sex and age coming from the parent colony maintained under ordinary conditions. The NCS mice grew more rapidly than ordinary mice on complete diets; moreover, they continued to gain weight—although at a slower rate—when fed deficient diets which caused ordinary mice to stop growing, or to lose weight. The NCS mice proved much more susceptible than ordinary mice to certain experimental bacterial infections. In contrast, they were much more resistant than ordinary mice to the lethal effect of large doses of endotoxins. However, they responded to injection of minute amounts of these endotoxins by a marked increase in susceptibility to staphylococcal infection. Bacteriological studies revealed striking qualitative differences between the intestinal flora of NCS and ordinary mice. When NCS mice were contaminated—either by contact or by feeding—with a strain of Escherichia coli recovered from the intestine of ordinary mice, they acquired the characteristics of the latter animals with regard to weight gain on various diets, and to response to bacterial pathogens and endotoxins. NCS mice have been found well suited to the study of several nutritional, bacteriological, and immunological problems and it appears that their production on a large scale will not present unsurmountable difficulties.

The small intestinal bacterial flora of 15 patients with chronic renal insufficiency was compared with that of subjects with blind loop syndrome. 9 patients were on regular hemodialysis with high protein intake and 6 (serum creatinine 7.5 to 12.5 mg/dl) were maintained on low protein diet. The chronic renal patients harbored a greatly increased microbial flora of both anaerobes and aerobes in the duodenum and jejunum, quantitatively comparable to those in blind loop subjects. The composition did not differ significantly in the two groups. Some organisms may have the potential to metabolize substrates which reach the intestinal lumen from the diet and bile, and perhaps to generate toxic metabolites that could contribute to uremic toxicity or malabsorption.

Vaccination with living attenuated tubercle bacilli (BCG) was found to increase the resistance of mice to infection with virulent staphylococci. An even more striking protective effect could be elicited by intraperitoneal or subcutaneous injection of small amounts (0.1 mg. or more) of killed BCG cells. The killed BCG cells retained most of their protective activity after prolonged heating at acid, neutral, or basic reactions—and after extraction with acetone, methanol, and NaOH (at pH 10.5). Some protective activity could be recovered in a fraction soluble in methanol at 55°C. The protective effect against infection manifested itself in a prolongation of survival time following infection, and also in the fact that smaller numbers of staphylococci were recovered from the organs of infected mice. Both types of effects were still evident 10 weeks after vaccination. Injection by the intraperitoneal route of killed cells of BCG, or of methanol extracts of them, elicited in mice a high level of protection against intravenous injection of Myco. fortuitum. A protective effect quantitatively and qualitatively similar to that elicited by BCG, resulted from the intraperitoneal or subcutaneous injection of killed cells of Myco. fortuitum.