Rodent models to study the relationships between mammals and their bacterial inhabitants

Abstract

Laboratory rodents have been instrumental in helping researchers to unravel the complex interactions that mammals have with their microbial commensals. Progress in defining these interactions has also been possible thanks to the development of culture-independent methods for describing the microbiota associated to body surfaces. Understanding the mechanisms that govern this relationship at the molecular, cellular, and ecological levels is central to both health and disease. The present review of rodent models commonly used to investigate microbial-host “conversations” is focused on those complex bacterial communities residing in the lower gut. Although many types of pathology have been studied using gnotobiotic animals, only the models relevant to commensal bacteria will be described.

Introduction

Studies on host-microbe interactions have largely focused on understanding the molecular mechanisms of pathogenesis of infectious diseases, which is not surprising given the considerable public health and commercial interest in developing tools for the diagnosis of infections, as well as improving vaccination, pharmacologic, and antibiotic treatments. However, we now know that the acquisition of pathogenic bacteria does not always cause disease and that microbes classified as non-pathogenic can also cause disease in certain susceptible hosts. This is because the outcome of the bacteria-host interaction depends on the context of the communication with the specific host. In normal individuals, for instance, large numbers of microbes are found on most surfaces of the body, like the skin,^1,2 the oral cavity,^3,4 and the gastrointestinal tract,^5,6 forming stable communities without causing disease. These communities are normally referred to as commensals (derived from Latin “cum mensa,” meaning “eating at the same table”). In particular, the intestinal commensals, which are the focus of this review, have been demonstrated to play a pivotal role in aspects of host nutrition⁷ and physiology such as development of adaptive lymphoid tissue,⁸ innate immune response,^9,10 healing following mucosal injury,¹¹ development of intestinal angiogenesis,¹² energy extraction and storage,¹³ and pathogenesis of autoimmune¹⁴ and metabolic diseases.¹⁵

Most of the current knowledge of how mammals, including humans, interact with their microbial commensals has been obtained by way of animal experimentation. Although the critical role of microbes in human health was suggested as early as the 1880s,^16,17 it was not until recently that appropriate laboratory tools have been developed to directly characterize the fundamental mechanisms underpinning bacteria-host communication. Probably the two most significant achievements in facilitating our understanding of these interactions were the development of nucleic acid-based techniques for analyzing complex bacterial communities, and advances in germ-free technology to manipulate the composition of the microbial environment in experimental animals.

Like in other life science disciplines such as drug discovery, preclinical studies, and toxicology, where animal models have been a mainstay of basic and applied research, rodent models have had a central place in revealing key features of the bacterial communities associated with vertebrates. The present review of rodent models commonly used to investigate microbial-host “conversations” is limited in scope to those complex bacterial communities residing in the lower gut. Although many types of pathology have been studied using gnotobiotic animals, only the models relevant to commensal bacteria will be described.

Gnotobiotic and Germ-Free Animals

The use of germ-free technology for investigating the interactions between the host and its associated microbiota has evolved substantially since the first conference on germ-free life in 1939.¹⁸ By controlling the microbial composition of the environment in which the animals are reared, scientists have been able to obtain information about how microorganisms affect the normal physiological functions of the host.

Over the past decades, the terms “gnotobiotic”, “axenic” and “germ-free”, have been occasionally (and unfortunately) used interchangeably. “Gnotobiotic” (from the Greek gnosis meaning “knowledge”, and bios meaning “life”) was originally used to describe the biological status of animals used in germ-free research.¹⁹ The terms “axenic” or “germ-free” (GF) refer to animals devoid of any other contaminating organism (also from Greek roots; a, “without”, xenikos, “foreign”).²⁰ Strictly speaking, the definition of GF requires the absolute absence of any form of life other than the subject animal, which is technically unrealistic. It may be more reasonable, however, to say that the animal is GF within the limitations of the current tests for microbial contaminants. As pointed out by Robert Fitzgerald, an animal may be demonstrably free of detectable known bacteria, yeast, fungi, and protozoa, but unless specific tests were also made for new viruses and rickettsiae, one would not be justified in concluding that the animal is, in fact, germ-free.²¹ Perhaps because there will never be an unequivocal answer to the question of whether a GF animal is indeed free of every microorganism, the term gnotobiotic becomes more appropriate, meaning that they have known, or completely defined microbiota. Nevertheless, accurately defining the composition of associated microorganisms is not without limitations, as will be discussed in the next section.

“Gnotobiotic” has also been proposed to describe GF animals deliberately inoculated with one or several microbial species (“ex-GF”: mono-associated, bi-associated, etc.).²² The definition can be further stretched to include animals that lack one or more types of microorganisms but harbour an otherwise normal complex microbiota. For instance, using a combination of gnotobiotic technology and antibiotic treatment, it was possible to derive a colony of Balb/c mice that did not harbor lactobacilli in their gastrointestinal tracts but retained a complex collection of microbes, functionally equivalent to those of their conventional counterparts.²³ Animals that are guaranteed to exclude particular pathogens are called specific pathogen-free animals, and although not strictly gnotobiotic, this animals have been derived from GF ancestors and are normally kept under meticulous barrier conditions. In contrast to gnotobiotic animals, animals carrying the full (undefined) burden of microorganisms usually associated with their species are described as “conventional”.²⁴

The first GF animals were successfully produced at the end of the nineteenth century. Using aseptic Cesarean section, Nuttal and Thierfelder generated GF guinea pigs and maintained them for two weeks under axenic conditions.²⁵ The rearing of GF rodents through successive generations in axenic conditions, however, was not achieved until the 1940s by Reyniers and coworkers at the University of Notre Dame²⁶ and by Gustafsson and coworkers at Lund University.^7,27,28 The approach to generate the original GF progenitor involved hand feeding pups with an artificial diet in a sterile isolator until maturity, after which a breeding GF colony was established from these progenitors.²⁹ The process was hampered by considerable logistical and technical challenges: it required an understanding of the composition of rodent milk, the development of a suitable dietary substitute for the pups, a method to sterilize the diet without affecting its nutritional value, and devising methods for hand-rearing the pups.³⁰

Nowadays, colonies of GF rodents are generated and established through two experimental procedures. Some laboratories perform a Caesarean section on conventionally raised pregnant females at term (whose timing must be carefully calculated). Mothers are euthanized, their bodies passed through a germicidal bath, and the pups delivered inside a GF isolator (usually Trexler-type plastic isolators under positive pressure). After resuscitation, pups are placed with a GF foster mother that has recently delivered her litter, with the expectation she will accept the new pups. The GF status of the breeding colony must be confirmed and repeatedly tested. Another method, involves embryo transfer at the two-cell stage using a pseudo-pregnant GF female as recipient.^31,32 The donor female is superovulated by injecting gonadotrophin, and then mated. A few hours later, oviducts are dissected and embryos flushed out under the microscope. Fertilized two-cell stage embryos are washed with antibiotic-containing medium, and transferred into the oviducts of a GF recipient female that had been mated with a vasectomized GF male. Again, the GF conditions of the colony must subsequently be confirmed. This is generally achieved by culturing fecal pellets and skin swabs in universal culture media under both aerobic and anaerobic conditions, complemented with PCR on feces using bacterial-specific primers.

Perhaps the simplest strategy to understand microbial-host functional conversations is to study a particular host function in GF conditions, and then evaluate the consequences of adding a single or defined population of bacteria to the GF animal. Alternatively, the impact on a given host function could be investigated during the conventionalization of a GF animal. However, one has to be mindful that GF animals are functionally and physically immature in many physiological systems (immune and non-immune), which challenges comparisons of results obtained in GF conditions to those in natural settings. For example, mice raised in GF conditions show an immature intestinal pattern of high sialyltransferase and low fucosyltransferase activities relative to conventional mice;³³ the content of intestinal IgA-secreting plasma cells is reduced in GF animals³⁴ as is the size and number of Peyer’s patches.³⁵ These differences are also observed systemically due to soluble bacterial structures being absent in GF animals.³⁶ A comprehensive summary of the multiple defects in structure and function of different organs in GF animals can be found elsewhere.^28,32 In most cases, the underpinning mechanisms of these alterations are not fully understood.

In addition, conventional animals have acquired the experience of living with their microbial communities since they were born³⁷ (or maybe even in utero^38-40), so it is expected that the various systems that are affected directly or indirectly by this life-long history of commensalism would respond differently if the microbial stimulus is removed. In other words, because some functions in GF rodents have not achieved a sufficient level of development through life, simply inoculating bacteria into adult GF animals during a short period of time may reveal only partially how the microbes impact the physiology of the host.

Notwithstanding these limitations, gnotobiotic animals have been instrumental for researchers in understanding aspects of the mechanisms behind the assembly of the complex bacterial community, its implications in numerous diseases, and the evaluation of potential therapeutic solutions.

Simplified Microbiota Models

Gnotobiotic mice colonized with a pure bacterial culture (mono-association) represent the most reductionist approach for obtaining information about host-microbe specificity, the ecological niche of that particular microbe, and mechanisms of pathogenicity, without competition from other species. Ex-GF NMRI mice mono-associated with Bacteroides thetaiotaomicron, a normal resident of the distal intestine of mice and humans, were used to demonstrate specific biochemical factors involved during bacterial colonisation.⁴¹ This simple ecosystem reproduces the cellular, spatial, and kinetic features displayed by a complete microbiota with regards to the utilization of the host epithelial fucosylated glycans as a source of energy.⁴² B. thetaiotaomicron senses the availability of fucose in the gut through the repressor FucR, and coordinates the expression of enzymes in the L-fucose pathway with those that regulate the production of fucosylated glycans in intestinal enterocytes. The authors speculated that during weaning, when nutrient accessibility becomes critical and fucose availability declines, B. thetaiotaomicron is capable of instructing the host to produce hydrolysable fucosylated glycans in order to ensure a sustained supply of carbohydrates. Similar studies by the same group have demonstrated that this organism can vary the expression pattern of its genes related to the utilization of polysaccharides as a function of the host’s diet: during transition from milk to polysaccharide-rich chow at weaning,⁴³ or after switching from a diet rich in plant polysaccharides to a diet devoid of them but rich in simple sugars.^44,45 In addition, by simultaneously profiling the relative abundance of thousands of B. thetaiotaomicron mutants under various conditions, Goodman and colleagues were able to identify genes that are critical for the establishment and persistence of this bacterium in the human gut.⁴⁶ Strictly speaking, this is not a mono-associated model, but because all mutants belong to the same strain and individual mutants can be retrieved and analyzed, the model allows for the identification of various microbial functions, including adaptation to the host, in one single strain.

The metabolic interactions that occur in the large bowel have also been explored using defined microbial communities consisting of two or more representatives of microbes naturally occurring in humans. If sequenced representatives are chosen, predictions of their functions could be made after inspecting their genomes as well as how they modulate their gene expression in response to host stimuli or diet. Studies in a two-member human microbiota model (B. thetaiotaomicron and E. rectale in NMRI mice) demonstrated the ability of intestinal microbes to adapt their environment in the presence of neighboring bacteria. For instance, B. thetaiotaomicron upregulates the expression of a variety of polysaccharide utilization loci (PUL) to broaden its niche and degrade greater variety of glycan substrates, including those derived from the host that E. rectale is not able to access. In contrast, E. rectale became more selective in its harvest of sugars and other nutrients, downregulating a significant number of genes for carbohydrate metabolism in the presence of its neighbor, but increasing the expression of selected sugar and amino acid transporters.⁴⁷ E. rectale utilizes acetate produced by B. thetaiotaomicron to generate large amounts of butyrate, which in turn is used by the intestinal epithelium. A set of key metabolic genes relevant to energy conservation is also upregulated when E. rectale encounters B. thetaiotaomicron. Interestingly, the pathway for acetate metabolism observed in this model differs significantly from that in mice colonised with B. thetaiotaomicron and Methanobrevibacter smithii, a single predominant archeal methanogen in humans.⁴⁸ In this case, there is an increased production of acetate and no diversion to butyrate, indicating a specificity of the ecological dynamics in the intestinal tract. The quest for successful strategies to manipulate energy harvest from the diet has also been directed to the role of acetogens and sulfate-reducing bacteria. Using an elegant approach in bi-colonised mice, Rey and colleagues characterized the niches of two acetogens in the mammalian gut: Blautia hydrogenotrophica and Marvinbryantia formatexigens. These microorganisms produce acetate from H₂ and CO₂ via the acetyl-CoA pathway in the distal colon, making an important contribution to the nutrition of the host.⁴⁹ The authors demonstrated through a combination of transcriptomics and mass spectrometry of metabolites that these two species occupy different niches in the intestinal tract with their own patterns of substrate utilization: B. hyrogenotrophica forages on complex oligosaccharides derived from the diet and the host, whereas M. formatexigens consumes mono- and oligo-saccharides resulting in a differential impact on energy balance.⁵⁰

A step up in complexity, the association of GF animals with 5–15 species provides a more complex yet simple enough model to investigate host-microbe and microbe-microbe interactions. A simplified human intestinal microbiota consisting of seven bacterial species harbored in gnotobiotic rats⁵¹ showed metabolic functions comparable to conventional rats with respect to previously proposed mucosa-associated characteristics:⁵² production of short-chain fatty acids, conversion of bilirubin to urobilinogen, degradation of mucins and β-aspartylglycine, and inactivation of trypsin. Genomes of the selected bacterial community are publicly available, which has facilitated (as already mentioned) further studies at the molecular level. This approach can also be used to develop predictive models to speculate on the effect of various perturbations in the composition of the bacterial community. For instance, changes in species abundance and microbial gene expression in response to different diets were studied in a model community of 10 sequenced human intestinal bacteria in gnotobiotic mice. Transcript levels were used to develop a statistical model to identify dietary factors responsible for the changes in the microbial community and explain the interrelationship between diet and the structure of the gut microbiome.⁵³

Occasionally, standard methods of inoculation of bacteria into GF animals do not result in a complete colonization of a microbial set, even for a relatively simple 10-member community. However, it has been reported that almost the whole community could be successfully established for up to 70 days when single bacterial strains were inoculated into individual animals followed by grouping the animals to exchange their microbiotas.⁵⁴ Taking advantage of the coprophagic habits of the rodents, a high level of microbial colonisation was achieved. Moreover, when GF animals were introduced into the colony after a few weeks, they quickly acquired a similar microbiota to that of the donors. This suggests that the transferred microbiota had already achieved a significant level of stability and adaptation to the rodent gut environment. These observations suggest that the assembly of microbial communities is governed, among other factors, by niche-related deterministic processes.

The Specific Pathogen-Free Animal

There is considerable evidence that infections and general wellbeing in laboratory animals influence a variety of biological parameters that in turn significantly affect the outcomes of scientific experiments. Over the years, governmental, academic, and professional organizations have recommended programs for health monitoring of breeding colonies with the intention to harmonize procedures,^55-57 and although local circumstances and historical practices may affect how these recommendations are actually applied, almost all commercial breeders are currently able to supply laboratory rodents with certified specific pathogen-free (SPF) status. This label is used only to indicate that the colony from which they originated tested negative for certain pathogens or perhaps opportunistic agents that are known to result in subclinical infections. SPF rodents are produced in barrier rooms in uncovered cages, and because of their exposure to microorganisms in the environment (air, food, humans, litter), they soon become colonized with commensal bacteria, the diversity of which is yet to be accurately defined.³¹ The inadequate characterization of the “normal” microbial community structure in SPF rodents has considerable implications for the relevance of such animals as a standard model for the investigation of bacteria-host interactions.

SPF rodents currently available have been derived from previously GF ancestors that were associated with a few bacterial isolates originating from the feces of a healthy mouse. The original experiment was performed by Russell Schaedler and his colleagues during the mid-1960s.^58,59 They used pure cultures of four bacterial species (lactobacilli, anaerobic group N streptococci, bacteroides, and coliform bacilli) to inoculate GF mice in their laboratory at the Rockefeller Institute. The animals were maintained in plastic isolators where they were given food contaminated with the bacterial isolates. The experiment was initially conceived to assess the consequences of associating fecal isolates with GF mice, but given that the animals became protected against acquisition of opportunistic bacteria, the researchers subsequently supplied animal breeders with this “cocktail” of microorganisms for use in colonizing their rodent colonies.⁶⁰ A few other attempts to include extremely oxygen-sensitive fusiform (EOS) bacteria in defined microbiotas were later used for gnotobiotic studies. EOS bacteria constitute the predominant microbiota of mice but are technically challenging to manipulate in the laboratory.⁶¹ One of those microbiotas, the altered Schaedler flora (ASF), reached great popularity in the late 70s and 80s when the National Cancer Institute decided to standardize the microbiota used in their rodent colonies and those of their contractors.⁶² The ASF consisted of eight bacterial species: Lactobacillus acidophilus, Lactobacillus salivarius, Bacteroides distasonis, four extremely oxygen sensitive bacteria, and one spiral-shaped bacterium. Over the years, breeders of mice and rats⁶³ around the world have adopted the ASF, and kept the animals for generations under strict barrier conditions to maintain their SPF status. However, animals harboring this microbiota have not evolved to a widespread model to study microbial-host interactions. Advances in ongoing genome projects coupled with current ease in cultivating strict anaerobes would certainly overcome previous technical hurdles to work with members of this microbial cocktail, although to date, this cocktail has not been deposited in any official culture repository.

The Human Flora-Associated (HFA) Rodent Model

The inoculation of GF mice or rats with fecal suspensions originating from human donors was conceived as a strategy to circumvent the variability introduced by environmental and genetic factors in human studies, or when ethical or practical reasons limited the study of the gastrointestinal communities directly on human volunteers. Investigators had hoped that HFA rodents would mimic the microbiota of the human intestinal tract, therefore being a more relevant model than their conventional counterparts for predicting the situation in humans. However, much controversy has been generated over the adequacy of HFA animals as surrogates for studying the ecology and metabolism of the human microbiota, with valid arguments on each side of the debate.⁶⁴

The likelihood of successfully transferring human fecal bacteria into recipient rodents is questionable. Early reports indicated that the composition of the intestinal microbiota of HFA animals was similar to that of donor human inocula when classical microbiology techniques were employed.^65,66 We now know that microbial cultures largely underestimate the complexity and size of intestinal microbial communities;^67,68 therefore, conclusions from those observations may be biased, as they merely consider a subset of the full microbial load administered to the animals. But even with these technical limitations, scientists have noticed, in some cases, that not all members of the initial inocula could be implanted, in particular bifidobacteria and lactobacilli.^69-71 Recent publications adopt a more critical perspective by concluding that the composition of microbial communities from human donors resembles that of their corresponding HFA rodents only at the predominant species level.^72,73 Difficulties in achieving an exact match in microbial profiles between feces from recipient animals and human donors are not unexpected: reciprocal inoculation experiments in related vertebrates show that the host environment plays an important role in determining the microbial makeup.⁷⁴ This was challenged, however, when scientists from the Gordon group published the results of a transplantation study of human intestinal communities into GF animals.⁷⁵ Using multiplex pyrosequencing of the bacterial 16S rRNA genes and statistical models to compare the degree of similarity of the fecal bacterial communities, the authors demonstrated that human fecal microbiotas were successfully transplanted to GF mice with a significant preservation of their structure and diversity, even when the starting material was frozen feces. All bacterial phyla, 11 out of 12 bacterial classes, and 88% (58/66) of the genera detected in the donor sample were detected among the recipient mice, and this structure was stable for up to one month. Interestingly, the humanized mouse microbiota could be transmitted to a second generation of mice without a significant reduction in diversity.

From a metabolic standpoint, it may also be questioned whether the enzymatic activities tested or the metabolite profiles assessed represent valid readouts to determine a successful transfer of microbial activity from human donors to rodents. A limited set of gastrointestinal enzyme activities (β-glucosidase, β-glucuronidase, nitrate reductase, nitroreductase) are generally measured, and levels of some putrefactive products or short chain fatty acid are reported.^76-78 In any case, bacterial metabolism in the intestine of HFA mice reflected that of human feces only with respect to some metabolic activities, probably due to changes in the bacterial composition or a different intestinal environment in the recipient animals compared with that in the donors. Recent data indicate that it is possible to cluster microbial communities based on their gene content and infer pathways involved in their metabolism to better compare transmissibility of their function.⁷⁵ One report suggests that it is possible to reproduce the functions and composition of the human gut microbiota with remarkable similarity out of its readily culturable members using anaerobic culture conditions. When transplanted to gnotobiotic mice, the complete fecal set was comparable to the cultured community in its colonisation dynamics, distribution and responsiveness to dietary changes.⁷⁹ It would be interesting to investigate how the metabolic functions of transplanted microbiotas compare with those developed by the microbiotas that co-evolved with their hosts.

Diet can have a profound effect on the resulting bacterial composition following inoculation of mice with fecal slurries. Since rodents and humans have different dietary habits, it should come as no surprise to find that failure to stabilize human-derived microbiotas in mice could be due to substantially different nutrient profiles. Standard chow diets normally administered to rodents are low in fat but rich in complex polysaccharides generally originating from plants; a stark difference to typical Western-like human diets, high in fat and simple sugars. Studies feeding rodents with human diets indicate that mice and rats show high adaptability to changes in their diets, although not without major impact on their metabolism.^80,81 Consequently, feeding HFA rodents with human diets as a strategy to stabilize their human-derived microbiota should be measured against how well the animals are able to maintain a normal metabolism.

It may be less debateable though, that the simulation of human gastrointestinal conditions in HFA mice represents a suitable and reliable approach for the investigation of colonisation resistance against pathogenic bacteria,^82,83 impact of the consumption of toxic compounds,⁸⁴ or carcinogens,⁸⁵ and the efficacy of therapeutic drugs. HFA animals have been used to assess the effects of antibiotics on human intestinal microbiota,^86,87 and the risks associated with DNA transfer from food-borne genetically modified microorganisms.⁸⁸ Colon cancer biomarkers have also been studied in HFA animals.⁸⁹ The characterization of the microbiota configuration and its variations along the length of the gut, even when the gastrointestinal tract of rodents is not exactly the same as humans, is an example of a practical use of the model.⁷⁵

Despite its limitations, the HFA mice model continues to stand as a useful tool for studying the ecosystem and metabolism of the human microbiota in conditions similar to those of the human intestinal tract. It is a useful substitute for human volunteers, especially when it is difficult to control for genetic, environmental, dietary, and statistical factors that usually challenge conclusions from clinical studies.

The Microbiota of Conventionally Raised and Feral Animal

While gnotobiotic mice are colonised with simple, defined collections of microorganisms, conventional animals carry a full, usually undefined, community of microbes associated with their species. It has been discussed previously that even the most strict barrier conditions cannot prevent SPF animals from acquiring environmental bacteria, and that microbiological monitoring of these animals only aim at documenting that they are free of pathogens without providing an inventory of the actual microbial population. Given that the environmental conditions under which SPF animals are raised are considerably different to those used for conventional mice, it is highly likely that the microbiota that becomes adapted through successive generations in each type of animal is also different. Similarly, one could speculate that the microbiota of conventional laboratory mice do not truly represent the “normal” biota of mice. Husbandry is, therefore, an important consideration for host-microbe interactions.^90-92

The composition of the fecal bacteria in feral or “wild” mice was determined by Wilson et al.⁹³ using 16S rRNA gene sequence analysis, and compared with that of SPF animals. Wild animals harbored a much larger proportion of bacteroides and lactobacilli, whereas the majority of sequences found in laboratory mice belonged to anaerobic clostridia. Although the study was not sufficiently powered to justify a broad conclusion (only 2 feral and 3 laboratory mice were used), these observations suggest that foreign microorganisms may have replaced the indigenous biota that co-evolved with mice in nature.

Understanding the evolutionary processes by which mammals have been interacting with their bacterial commensals and the factors that affect community makeup have been the aims of large programs, like the multinational MetaHIT^94,95 initiative and the Human Microbiome Project^96,97 supported by NIH. Although reviewing the scientific evidence that supports these concepts is out of the scope of this review, it is important to recognize their implications for the selection of an animal model for studying host-bacteria interactions. If we accept that mammals have co-evolved with microbes and therefore did not need to develop functions that are provided by bacteria, including the ability to extract energy and nutrients from the diet,⁹⁸ it is fair to assume that the physiological functions of laboratory mice are likely to differ from those naturally occurring in feral mice. Of course, this may be an oversimplified conclusion as we now know that even distantly related vertebrates share similar microbiotas, at least at a shallow phylogenetic level,³⁷ and that considerable functional redundancy of the gut microbiota has been reported.^99-101 Nevertheless, as more sophisticated metagenomic tools are becoming readily available, it may be possible to detect even subtle functional differences in the host-microbe relationship, suggesting that selecting the appropriate model is, after all, not a trivial decision.

The identification and differentiation of “autochthonous” (resident) microbes from “allochthonous” (transient) microbial members of the intestinal community have been challenging tasks. The concept of autochthony has been extensively reviewed,^102,103 and raises fundamental questions concerning the ecological role of a given species in the complex intestinal environment. For instance, autochthonous species are specialized to occupy a defined physical niche and form stable populations during long periods of time, whereas transient species may behave more unpredictably depending on the endogenous or exogenous factors, such as the diet of the host. Native microorganisms can develop a remarkable host specialization, which has implications on the model of choice if the aim is to study these microorganisms in their native habitat: several reports indicate that some species of lactobacilli are not native to the human large bowel,^68,104,105 originating probably from ingested foods or other regions of the intestinal tract.¹⁰⁶ This is not the case for lactobacilli in rodents, where they can form stable and specialized communities; therefore, having a different impact on the host.

Final Remarks

Human health could be thought of as the collective property of human-associated microbiota, and experimental tools to decorticate the mechanisms that govern this interaction are becoming increasingly necessary. Gnotobiology coupled with molecular genetics provide an excellent technology to create and manipulate bacterial ecosystems to investigate fundamental questions about us and our intestinal symbionts. Although one single model may not suffice to unravel the complexity of these interactions, thorough consideration of the limitations inherent to each model will certainly allow us to articulate the right questions.