Complex Microbiota in Laboratory Rodents: Management Considerations

Abstract

Our bodies and those of our animal research subjects are colonized by bacterial communities that occupy virtually every organ system, including many previously considered sterile. These bacteria reside as complex communities that are collectively referred to as microbiota. Prior to the turn of the century, characterization of these communities was limited by a reliance on culture of organisms on a battery of selective media. It was recognized that the vast majority of microbes, especially those occupying unique niches of the body such as the anaerobic environment of the intestinal tract, were uncultivatable. However, with the onset and advancement of next-generation sequencing technology, we are now capable of characterizing these complex communities without the need to cultivate, and this has resulted in an explosion of information and new challenges in interpreting data generated about, and in the context of, these complex communities. We have long known that these microbial communities often exist in an intricate balance that, if disrupted (ie, dysbiosis), can lead to disease or increased susceptibility to disease. Because of many functional redundancies, the makeup of these colonies can vary dramatically within healthy individuals [1]. However, there is growing evidence that subtle differences can alter the phenotype of various animal models, which may translate to the varying susceptibility to disease seen in the human population. In this manuscript, we discuss how to include complex microbiota as a consideration in experimental design and model reproducibility and how to exploit the extensive variation that exists in contemporary rodent research colonies. Our focus will be the intestinal or gut microbiota (GM), but it should be recognized that microbial communities exist in many other body compartments and these too likely influence health and disease [2, 3]. Much like host genetics, can we one day harness the vast genetic capacity of the microbes we live with in ways that will benefit human and animal health?

Factors Influencing Rodent GM

Rodents in research settings can be housed in a variety of environments ranging from open-topped caging to ventilated rack housing to pristine gnotobiotic facilities where stringent procedures are in place to maintain either bacteria-free conditions or strictly defined bacterial populations. Most rodents are housed in nongnotobiotic settings where there is often limited knowledge about the microbiota that resides in and on them. Recent reports have shown that a number of factors can result in subtle changes in microbiota. Factors that can modulate GM include, but are not limited to, producer/vendor (including different facilities within a single producer), diet, bedding type, housing enclosure, shipping, therapeutic intervention, water decontamination method, and rederivation [4–9]. Rodent GM may also undergo some “institutionalization” after being housed in a specific facility for a short period of time [10]. Moreover, the genetics and sex of the animal influence the composition of GM [11]. Because many of these factors differ among institutions, there is likely great variation in the GM of contemporary rodent research colonies. The Mutant Mouse Resource and Research Center (MMRRC) is an NIH-funded, 4- institution consortium that serves as a repository for mutant mice. The MMRRCs recruit novel mutant rodents from all over the world, cryopreserve their germplasm, and distribute these valuable models to the biomedical research community. In 2013, the MMRRC at the University of Missouri began banking fecal samples from all models submitted to their repository [12]. Analysis of 96 samples collected during that time frame confirmed the notion that the GM of nongnotobiotic contemporary rodent colonies varies greatly (Figure 1). Notably, the “effect size” of GM-modulating factors is difficult to determine without targeted phenotypic studies in the model of interest and likely varies greatly with some factors having more profound effects than others [8]. Thus, one cannot assume that a GM shift associated with a particular factor will result in modulation of a phenotype.

Figure 1.

(A) Stacked bar chart showing marked GM variation in fecal samples of mice submitted to the MU MMRRC since 2015. (B) Variation in the Firmicutes/Bacteroidetes ratio, a common GM outcomes measure, in the same fecal samples. (C) Variation in richness of the same fecal samples. Brackets demonstrate the richness of the 2 extremes found in producer colonies.

GM and Model Reproducibility

These findings raise the question: does such variation in GM among rodent colonies contribute to variability and, more importantly, reproducibility of rodent models? This comes at a time when several recent reports have raised awareness about poor reproducibility of mouse models [13,14]. Moreover, there are frequent anecdotal reports about changes in model phenotypes associated with events such as an investigator moving from 1 institution to another. The latter is especially prevalent in fickle models of immune-mediated disease such as Type 1 diabetes, multiple sclerosis, inflammatory bowel disease, and arthritis. Similarly, disease models may show variation in disease expression over time within the same laboratory. Proposed solutions to poor reproducibility between laboratories generally focus on rigid experimental design and thorough reporting of experimental conditions so that experiments can be optimally replicated. Standards such as the ARRIVE guidelines are promoted to ensure that a standardized list of potential environmental and experimental variables is considered [15]. Curiously, many of these considerations are also those that have been shown to modulate the GM.

To assess whether differences that exist in GM may be contributing to phenotype inconsistencies, we and others have generated several rodent models of disease colonized with differing complex GM originating from common rodent producers. As an example, the PIRC rat model of colon cancer was derived by transferring PIRC embryos into surrogate mothers that harbored differing complex microbiota [16]. Because rodents inherit their microbiota primarily from their mothers, these rats were isogenic for the PIRC mutation but harbored differing GM. Rats that harbored 1 particular “enterotype” obtained from a Lewis parent developed fewer tumors compared with rats that harbored one obtained from a F344 parent. Of note, 2 rats colonized with the protective GM developed no colonic tumors, a phenomenon not previously seen in this model. Similarly, embryos from IL-10 knockout mice, a common model of inflammatory bowel disease, were used to derive isogenic mice with differing GM [17]. Severity of disease varied significantly, dependent solely on the intestinal microbiota they inherited, providing further proof-of-concept that the differing microbiota present in contemporary rodent colonies can contribute to phenotypic differences.

Identifying Target Species in Complex GM Studies

Such studies highlight the differences in these complex communities that can contribute to model phenotypes and, by extrapolation, susceptibility to certain diseases. Through such studies, it is also possible to identify target species for further investigation. A classic example of this is that of segmented filamentous bacterium (SFB). Ivanov et al found that mice in their studies differed in development of mucosal Th17 immunity and recognized that differences were associated with different sources of mice. Through an elegant series of experiments, they identified SFB as a putative culprit and subsequently performed proof-of-concept experiments to confirm its role in promotion of Th17 mucosal immunity [18,19]. Following this, several investigators set out to assess whether the presence of SFB altered animal model phenotypes and found that indeed, the phenotypes of models of multiple sclerosis [20], type 1 diabetes [21], inflammatory bowel disease [22], and autoimmune arthritis [23] were modulated by the presence or absence of SFB. This bacterium has subsequently been considered a keystone species in the assessment of GM modulation of disease. Of note, many other bacteria are capable of inducing Th17 mucosal immunity [24], suggesting we have only scratched the surface in our understanding of how commensal bacteria stimulate and direct mucosal immunity.

Similarly, rodent helicobacters, most notably Helicobacter hepaticus and Helicobacter bilis, have been shown to serve as provocateurs of low-grade inflammation against “commensal” gut microbes [25,26]. While nonpathogenic as the sole colonizer of mono-associated mice, even in mutant strains used to model IBD [27,28], colonization by these bacteria has proven to be critical to optimization and reproducibility of specific pathogen-free murine models of inflammatory bowel disease and intestinal cancer [25,29–36]. What effects helicobacter and/or SFB colonization have on other models of disease remains unknown, but given their prevalence in contemporary rodent colonies [37–39], they are prime candidates for phenotype confounders and should be considered when lack of reproducibility arises.

Moreover, bacterial species within the GM exist as a syntrophy, with each member interacting synergistically or antagonistically with multiple other species via cross-feeding networks. Thus, while the identification of single agents associated with phenotypic differences remains a feasible first step in mechanistic studies, future challenges will likely center on a better understanding of the complex interactions that occur among bacterial communities and how these interactions can be optimally identified and studied. Methods to assess the role of complex communities in host health and disease are rapidly developing and will surely change the paradigm with which we view host–microbial interactions. One change is a new appreciation of the need for robust longitudinal data to develop accurate models of GM communities, incorporating their inherently dynamic nature. As an example, traditional correlation analyses based on a single time-point have been used to identify groups of co-occurring, or mutually exclusive, microbes that are interpreted to be mutually beneficial or in competition, respectively. New models based on longitudinal data suggest that interpretation of static data is highly problematic [40]. Similarly, correlation analyses between the GM and some phenotypic outcomes measure are under constant development and refinement [41] in several laboratories.

Methods to Transfer Complex GM

Cohousing

While next-generation sequencing allows us to better characterize complex microbial communities that exist in our rodent colonies and associate differences in GM with phenotypic changes, these studies remain correlative by nature, similar to most human GM studies. To further pinpoint cause and effect, one must show that transfer of GM results in transfer of phenotype. This has been accomplished using a number of strategies. Perhaps the simplest is cohousing of mice that possess a desired phenotype with isogenic mice that lack the phenotype, thus relying on coprophagy for the transfer of GM. If important bacteria successfully transfer, the phenotype may also transfer. However, complex GMs are often stable, and colonization resistance may prevent key members of the GM from colonizing the intended recipients [42]. Indeed, co-housing usually results in a hybridization of GM that may not result in phenotype change. Thus, negative results of such studies must be interpreted with caution and include careful characterization of GM in recipient mice to assess the extent of transfer.

Bedding-Feces Transfer

A similar approach is to transfer feces or soiled bedding from mice with the desired phenotype to those that lack the phenotype. Anecdotally, this scenario has occurred when investigators move to a new facility and lose their model phenotype. In situations where mice remain with collaborators from the originating institution, bedding or feces can be collected and transferred to the new facility. This approach is also likely reliant on coprophagy and has met with mixed results likely because, as in co-housing, transfer of GM is incomplete, and colonization of key bacterial species or combinations of species may not be achieved in recipient mice.

Fecal Transplantation to Antibiotic-Treated Mice

A more direct approach to transfer complex GM is fecal transplantation. In this method, recipient mice can be pretreated with broad spectrum antibiotic cocktails in an attempt to markedly reduce endogenous microbiota. Recipient mice can then be gavaged with slurry prepared from donor feces or other intestinal samples containing the desired GM. While pretreatment with antibiotics abrogates much of the colonization resistance associated with co-housing and exposure to feces/soiled bedding, many members of the endogenous microbiota are not eliminated and reestablish colonization levels once antibiotics are removed. This, coupled with an incomplete transfer of donor microbiota, results in a hybridized microbiota. Transfer can be markedly limited, especially when attempting to transfer GM of low richness (number of microbial taxa present) to recipients initially colonized by high richness GM [43].

Fecal Transplantation to Gnotobiotic Mice

Given the above, it is far more ideal to render mice germfree by classical techniques and use these as recipients of GM transplantation. While this approach requires specialized facilities and as a result is more expensive, it can be considered one of the gold standards by which complex microbiota transfer can be achieved. As with any technique, transfer may not be complete, especially if obligate anaerobes are lost during preparation of the transfer material [44]. To optimize transfer of GM with fecal transplantation, one can also consider serial gavage [45], although the latter has not been extensively assessed for efficacy in complex GM transfer studies.

Derivation by Embryo Transfer

To transfer GM by the most natural means possible, mice can be derived by embryo transfer using surrogate dams possessing the desired microbiota [46]. This results in neonates receiving many key elements of GM acquisition, including exposure to vaginal microbiota during delivery, oral microbiota of the dam during neonatal care, and cutaneous and mammary microbiota during nursing. Similarly, offspring conceived following embryo transfer share the maternal bloodstream while in utero and thus undergo early stages of ontogeny under the influence of any maternal gut-derived factors in peripheral circulation. Moreover, complex GM acquired in this fashion matures and develops as oral tolerance is naturally occurring in the host, which does not occur with exposure to donor GM at later stages of life. While more expensive and labor intensive than other methods, this is arguably a second gold standard for comprehensive transfer of GM.

Considerations in Choice of Transfer Technique

The choice of transfer technique depends on many factors, including study goals and resources (both monetary and physical plant) available. For example, if the goal is to perform a pilot study to assess whether GM is impacting phenotype, cross-fostering or even co-housing may suffice with the caveat that negative results should be interpreted with caution. Conversely, if the goal of the study is to assess the role of 2 specific GMs in phenotype, gold standard techniques may be desired and the choice between these 2 would rely on resources at that institution (eg, availability of gnotobiotic facilities and/or embryo transfer expertise).

It should also be noted that even with the use of either gold standard, recipient GMs will likely differ to varying degrees from the donor. This is nicely exemplified in the manuscripts by Rosshart et al. In their 2017 manuscript [45], these investigators transferred frozen ileocecal contents to pregnant mice of the targeted genotype (C57BL/6), and in the 2019 manuscript [47], GM was established by transfer of embryos from C57BL/6 mice into wild mice harboring the targeted GM. In both cases, transfer of the majority of the GM, including dominant families of bacteria, was accomplished. However, in both situations, examination of principal component analyses reveals subtle differences in GM between donor and recipient. These differences are likely due to multiple factors. First, because host genotype can shape microbiota, if the recipient genotype is different from that of the sample donor, subtle shifts may occur and the recipient GM (even in second-generation offspring) will rarely look exactly like the donor GM with regard to the relative abundance of transferred taxa. Moreover, transfer of microbiota from other sites does not occur with the exception of embryo transfer techniques that use surrogates with the desired system-wide microbiota.

Many of these methods rely on transfer of feces to weaned or adult mice. As a result, key early-life influences of the microbiota in physiological processes such as mucosal immune system development are missed. This can be overcome by using second- or subsequent-generation mice, but these result in increased expense of animal housing and, more importantly, loss of time in performing experiments. This can also be overcome by using either gold standard approach (ie, fecal transplantation to pregnant germfree mice instead of transfer to study mice themselves [47] or embryo transfer into surrogates harboring the desired GM) [46].

A Case for Feces Banking

The above approaches rely on the availability of mice with the desired phenotype and associated microbiota or feces from such mice. However, changes in phenotypes associated with a move to another institution or change in environment/husbandry are often not predicted. In these cases, feces or animals obtained prior to the change may not be available. As the majority of mouse research colonies in the United States originated from 1 of 4 main producers, one can again use mice obtained from those producers as GM donors using an approach similar to that used in the aforementioned experiments. The latter is somewhat of a last resort approach and, if pursued, one must consider that the GM of mice from producers may have changed with time or that GM may vary within different production colonies. For example, the authors have found that even SFB colonization can be inconsistent among colonies from producers thought to be “positive” for this bacterium. Many producers have begun to assess GM within their colonies, but, at this time, variability from year to year remains likely and it is best to discuss GM status prior to procuring mice for such studies.

As a preventative measure, if changes in any GM-modulating factors are anticipated, periodic fecal banking should be considered. This consists of collecting fecal samples and preserving these at −80°C. Sufficient samples should be collected to allow for analysis as well as reconstitution of mice. Should a change in phenotype occur, samples from pre-change and post-change can be analyzed and, if differences are found, one may surmise that GM is influencing the model phenotype. Reconstitution using the above techniques can then be attempted and if the phenotype is restored, target microbes can be identified that are critical to the model. If the phenotype is not restored, one can presume that either GM is not involved or that transfer of critical microbes was insufficient. While use of frozen feces has been shown to be equivalent to fresh feces in reconstitution of GM to antibiotic-treated mice [43], additional studies are needed to optimize cryopreservation techniques, for example, by incorporation of cryoprotectants.

Circumventing GM-Associated Phenotype Changes That Occur During Purposeful Rederivation

GM also likely changes when animals are rederived during attempts to eliminate pathogens. In some cases, this may also result in phenotypic changes. Unfortunately, reconstitution with the original GM is no longer an option because of the potential to reintroduce the unwanted pathogen. In these cases, the phenotype-associated GM can still be characterized and compared with the GM of mice obtained from different rodent producers. Because producers have been shown to have differences in GM, one can conceivably identify a producer-associated GM that is “closest” to that of the desired GM and use mice obtained from that producer as the source of GM for reconstitution using 1 of the above procedures. Notably, in determining an optimal GM, one must consider that GM may vary among different facilities/rooms of producers as well as between genotypes. With that in mind, the source of mouse, in our experience, has a much greater influence than genotype on the GM composition of inbred mice [4].

Is There an Optimal Complex GM for Rodent Modeling?

In determining an optimal GM for specific models, considerations should include the goal of the experiment. For example, in disease models, is the desire to have a GM that causes the most severe phenotype so that assessment of therapeutic or preventative measures can be optimally assessed? Or is the goal to have GM that results in mild disease so that other factors that may contribute to disease (eg, carcinogens) can be assessed? Or should one conduct experiments with GMs at both extremes? How can naturally occurring GMs of wild and pet store mice be incorporated into such studies? While an added expense, comparison of multiple GMs has many advantages. Potential target species or combinations of species may be identified for further study. Such studies may also unearth important interactions with other independent variables, such as host genetic modifiers. Collectively, differences seen among different GMs or even among animals with similar GMs may ultimately yield data important in advancement of precision medicine strategies.

Standardized Complex GM

While the above highlight how existing complex GMs can be exploited in model refinement and development, some standardization of GM may also be desired. A potential starting point for such standardization may lie with those GMs found in the major rodent producers. As described above, notable differences in GM exist among subsets of mice obtained from the 4 major suppliers of mice in the United States. Because the majority of rodents in academic and industry colonies likely arose from 1 or more of these producers, some GM standardization may already be in place. To facilitate studies that examine the role of differing GMs, the University of Missouri Mutant Mouse Resource and Research Center has developed 4 colonies of outbred mice that harbor GMs obtained from these 4 producers [46]. The GM of these colonies is being monitored quarterly and remained stable for over 20 generations. To avoid small drifts, we now employ an annual “refreshing” with GM known to mirror the original composition. These findings suggest that these colonies are amenable to use in controlled studies. Additionally, while the composition may change subtly following shipment to other institutions, those changes are minimal compared with the inter-colony differences [46]. These colonies are available for use in all of the above reconstitution methods, including co-housing, fecal transfer, cross-fostering, and surrogates for embryo transfer. Moreover, these colonies can be supplemented with specific target bacteria to assess their role in a “standardized” complex GM background. An example of how one might use these colonies and other experimental design strategies in assessing the role of complex microbiota in rodent model phenotypes is provided in Figure 2.

Figure 2.

Schematic of possible experimental design strategies to assess the role of complex gut microbiota in rodent model phenotypes.

GM Maintenance and Monitoring

In experiments where GM is of interest, consideration should also be given to monitoring GM throughout the course of the study or at least banking samples at various time points so that GM can be characterized temporally should unanticipated findings arise. Such practices also provide a safeguard for future experiments in that samples are banked as described above. GM can be maintained in colonies of rodents for many generations using appropriate husbandry techniques. These include contained housing (eg, static micro isolator or ventilated rack), changing of cages in disinfected biosafety cabinets, changing of personal protective equipment between animals with differing GMs (especially personal protective equipment such as gloves and sleeves that come in contact with animals), and use of irradiated or autoclaved feed, water, and bedding [48]. Using these common practices, the University of Missouri MMRRC has successfully maintained disparate GM in 4 colonies of outbred mice harboring differing GM for over 20 generations without evidence of GM drift or hybridization among colonies.

Translatability of Murine GM

With any animal model, we must always be cognizant of the translatability to the human condition. This is also true when considering GM. While intestinal physiology is similar between rodents and humans, there are also stark differences. Most notably, rodents possess a cecum where abundant hindgut fermentation occurs, and these animals rely on coprophagy for acquisition of these fermentation products. Moreover, rodents and humans have coevolved with their microbes and, while quite similar both functionally and at the taxonomic level of genus, they rarely share the same species and strains [49]. Rodents in the biomedical research community also live in comparably pristine environments compared with even the most hygienic humans, and our success in eliminating pathogens over several decades has likely decreased the richness of their microbiota to a level much lower than that of their human counterparts [50]. However, with animal models, there is always the half of the glass that is full, and we stand to glean a great deal about host–microbial interactions from our rodent models.

To optimize translatability, at least 3 approaches may be considered: (1) humanization of the GM through fecal or intestinal content transplants; (2) use of wild or pet store mice that are exposed to a much more diverse environment, which correlates with a richer GM; and (3) comparison of functionality of complex communities rather than comparison of species composition. The latter involves the use of tools such as metabolomics or transcriptomics [51], and an extensive review is beyond the scope of this manuscript. Humanizing the GM of rodents has grown in popularity and has resulted in many exciting results. One notable example from the early studies by Turnbaugh et al [52–54] elegantly showed that colonization of mice with GM from obese individuals resulted in weight gain in mice, whereas colonization with GM from lean individuals did not, and that diet influenced the phenotype of this weight gain. However, studies using this strategy also come with caveats, most notably the fact that transplantation of microbiota from humans to mice is incomplete (not all species transfer), and murine immune systems may not fully develop compared with mice reconstituted with murine GM [55].

In 2016, work from the laboratories of David Masopust and Stephen Jameson resulted in a paradigm shift in how we should consider the translatability of the GM of biomedical research rodents [56]. They showed that co-housing of laboratory mice with pet store mice resulted in marked changes in the composition of innate and adaptive immune system components (eg, CD8+ T cells) and that these changes resulted in enhanced resistance to infection by Listeria monocytogenes and to the immune response to lymphocytic choriomeningitis virus infection. They surmised that a laboratory mouse has limited antigen exposure, resulting in an immune system more closely resembling that of a human infant, whereas those exposed to a more antigen-laden pet store mouse GM had an immune system more similar to that of an adult human. Similarly, Rosshart et al surveyed hundreds of wild mice and showed marked differences in GM compared with laboratory mouse GM. Transfer of this GM to laboratory mice rendered them remarkably resistant to a normally lethal influenza infection. Moreover, colonization of laboratory mice with wild mouse GM resulted in a marked decrease in tumors in an AOM/DSS model of colon cancer [45]. Recent studies by this laboratory have provided further proof-of-concept of improved translatability due to a wild mouse GM. Laboratory mice born to wild mice via embryo transfer showed a dramatic change in their response to anti-TNF-α and TNF-α receptor fusion protein in an LPS model of septic shock. Mice with laboratory GM showed increased survival rates with both treatments, whereas “wildling” mice (those colonized by GM from wild mice) did not [47]. The latter recapitulated findings in humans, which have shown minimal or even adverse effects of TNF-α inhibition in the treatment of septic shock.

While these studies are very provocative, housing of pet store or wild mice presents new challenges to the laboratory animal and biomedical communities. Mice purchased from pet stores or obtained from the wild can have marked variation in their GM as well as exposure to and infection with viral, parasitic, and fungal agents, resulting in potential inconsistencies in sequential studies or studies performed by different laboratories. This is being addressed by producers who have expanded their services of creating germfree or reduced flora mice to making mice with characterized GM, such as the wild mouse GM established by Rosshart et al. [45] Housing of pet store or wild mice also present the laboratory animal veterinarian with challenges in containment and biosecurity so that inadvertent contamination of other rodent colonies housed at their institution is avoided.

Is there a happy medium whereby we can increase the antigen exposure to and subsequent immune system development of laboratory mice in a more controlled and containable/biosecure fashion? The answer may lie in the strategy employed by Reese et al, who sequentially infected laboratory mice with herpesviruses, influenza, and an intestinal helminth and assessed the immune response to a yellow fever virus vaccine [57]. This targeted, experimental infection strategy resulted in immune responses similar to those seen in pet store-raised mice and, as described above, better mimicked an adult human response. It remains to be determined which specific sources of antigen (ie, bacterial, viral, etc.) are most influential on development of the immune system, and those influences are likely host mouse strain dependent. That being said, to mimic the naturally occurring effect of the wild mouse GM, perhaps “keystone” taxa such as SFB and enterohepatic Helicobacter spp. represent a logical starting point for the development of a pseudo-wild mouse GM.

Collectively, these studies highlight the challenges we face in interpretation of the translatability of many contemporary studies that use rodents colonized by GMs that functionally do not result in immune system development recapitulating that which occurs in humans.

Limitations of Complex GM Studies

The study of complex microbiota and its role in rodent model phenotypes is still in its infancy. As a result, there exist many limitations to these studies, and data should be interpreted with caution. Many contemporary studies rely on targeted amplicon sequencing using the 16S rRNA gene to determine bacterial phylogeny. This is a good low-cost starting point, but it must be recognized that whole bacterial genome shotgun sequencing improves speciation and bacterial strain assignment and, more importantly, other genes that may impact bacterial function (eg, virulence factors). Regardless of the method used to characterize the GM composition, these data must also ultimately be linked to some sort of functional assessment such as metatranscriptomics or metabolomics. Perhaps data obtained from targeted sequencing should be considered observation-generating data from which hypothesis-based mechanistic studies can arise. Other limitations include the sequence databases used for annotation (many operational taxonomic units or amplicon sequence variants can only be mapped to genus, family, or even order level) and inability to cultivate many intestinal bacteria that have emerging interest. Similarly, while the inability to cultivate certain candidate species limits our ability to test their influence in controlled, prospective studies, it is equally, if not more, challenging to selectively remove specific species from a community. There is also a need for bioinformatics tools for the study of complex interactions of bacteria as well as between microbiome data and metabolomics (or other -omics) data. It should also be recognized that with temporal studies of intestinal microbiota data are obtained from feces, and populations present in these samples do not reflect those of higher levels of the intestinal tract such as the jejunum, ileum, and even the cecum [8]. Lastly, how can we incorporate the role of the metagenome, which includes other microbes such as viruses, fungi, and parasites as well as host genome contributions? All of these present challenges that the biomedical research community must consider and address.

Conclusion

In conclusion, much of today’s biomedical research occurs in rodents that possess complex and often uncharacterized GM. These GM markedly vary and can be modulated by a number of environmental variables. Such variability may also result in differing phenotypes and in turn contribute to poor model reproducibility. However, there are many silver linings to this previously underappreciated variable, and there has been an explosion of interest in how GM differences impact our health and disease. While gnotobiotic facilities and tools have reemerged in many major research institutions, studies using nongnotobiotic settings may also yield important observations about GM and its role in health and disease. Exploiting, rather than reducing, the existing complexity of rodent GM should be considered as an important adjunct to our desire to understand how microbes contribute to our health.

Potential conflicts of interest. All authors: No reported conflicts.

Funding

The authors were supported in part by NIH U42 OD010918; Mutant Mouse Resource and Research Center (CLF and ACE) and NIH K01 OD019924; Impact of gut enterotypes and segmented filamentous bacteria on colitis-associated colorectal cancer (ACE)2019; 11(5).