National Academies Press: OpenBook

Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop (2018)

Chapter: 5 Experimental Reproducibility Using Gnotobiotic Animal Models

« Previous: 4 Modeling Human Microbiota in Animal Systems
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

5

Experimental Reproducibility Using Gnotobiotic Animal Models

The second day of the workshop focused on challenges stemming from working with gnotobiotic animal models, both with regard to experimental reproducibility and in building and operating facilities supporting the use of gnotobiotic animals. The speakers who addressed experimental reproducibility were Andrew Macpherson, professor of medicine and director of gastroenterology at the University Hospital of Bern; Craig Franklin, professor of veterinary pathology and director of the Mutant Mouse Resource and Research Center at the University of Missouri; Aldons Lusis, professor of microbiology, human genetics, and medicine at the University of California, Los Angeles; Jeremiah Faith, assistant professor at the Immunology Institute and the Institute for Genomics and Multiscale Biology at the Icahn School of Medicine at Mount Sinai; Gary Wu, the Ferdinand G. Weisbrod professor of medicine at the University of Pennsylvania’s Perelman School of Medicine; and Alexander Chervonsky, professor of pathology and chair of the committee on immunology at The University of Chicago.

CREATING STABILIZED MICROBIOMES IN LABORATORY ANIMALS

One of the most difficult aspects of animal husbandry, said Macpherson, is controlling or standardizing the animal microbiome, in part because it is still not well understood. Consequently, it is difficult for researchers to “make robust measurements, to look at the underlying biology, to design suitable controls in our experiments, and to reduce the number of animals [used],” Macpherson stated. Given the reduced genetic variance of inbred strains, environmental microbiota and experimental manipulations predominantly influence the phenotypic variance. Macpherson’s main question, and the focus of his talk, was how much further this variance could be reduced through the application of gnotobiology.

Reducing microbiome-associated phenotypic variability is challenging for two reasons, said Macpherson. First, while the gut and skin microbiomes of a typical human remain relatively constant over a period of weeks, across individuals they vary tremendously. Second, experiments in germ-free animals have

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

shown that manipulating a host’s microbiome creates a plethora of strong effects throughout the body. Therefore, a single microbiome cannot serve as the exemplar for a species.

Different experimental models (see Table 5-1) are useful for characterizing and ultimately minimizing microbiome-associated phenotypic variability, said Macpherson. His research has largely focused on bottom-up models, using axenic animals or those with simple microbiotas, but noted that these models illuminate only portions of the biology of the complex microbiomes studied using top-down models (i.e., models with complex microbiotas).

An important limitation of top-down models is their inability to precisely define microbial consortia and the resulting ambiguity in assigning phenotypic effects to particular species within those consortia. However, Macpherson feels that the biggest disadvantage of these models is the lack of reproducible results.

The reduced phenotypic variability of inbred animals allows experiments to reach statistical significance using smaller numbers of animals than when using outbred strains. Using different inbred strains in an appropriately powered factorial design can compensate for the loss of genetic traits. Standardization, however, extends beyond genetics given that the microbiota also varies greatly from facility to facility. Breeding littermate controls is one approach to help clarify whether the microbiome or host genetics are contributing to phenotypic traits (Stappenbeck and Virgin, 2016), and can be applied in all animal facilities, Macpherson said. He added that researchers should be aware that raising mice in vivaria protected from natural environmental pathogens would change the immunological maturity of the mice.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

Attempts to create inbred mouse strains with standardized microbiota have not been successful (Pang et al., 2012). One approach is to transfer isogenic embryos into a second isogenic strain with the required microbiota so that the pups produced will have acquired the defined microbiota of the second strain. Another approach is to gavage an isogenic strain with organisms from a pure culture or a fecal sample, or even add a colonized animal to a cage. A reversible approach would be to gavage a mouse with bacteria that cannot survive within its digestive system so that—in time—it becomes germ-free (Hapfelmeier et al., 2010).

The best-known diverse, standardized consortium is the altered Schaedler flora, which consists of eight microorganisms derived from mice (Wymore Brand et al., 2015). Strains are first inoculated, and then a microbial consortium that complements the model’s pathways and metabolomics must be carefully chosen. Macpherson posited that isobiotic strains should, ideally, “be stable over generations on open source diets” to ensure reproducibility. The criteria for creating an isobiotic strain are that its microbiome is stable across multiple generations, that the members of the consortium can be cultured and have published genomic sequences, and that the original germ-free mouse comes from an open-source stock so that other laboratories can regenerate the microbiota. Such criteria would enable transfer of the microbiome to mice of different genetic backgrounds and to different institutional animal facilities.

Macpherson elaborated on the ideal consortium: it should be regenerated from pure cultures, have known microbial metabolic pathways, and not cause abnormalities in clinical chemistry, hematology, histology, body composition, development, or fecundity. Moreover, it should express representative metabolic and immunological profiles, induce pathogen resistance and inflammatory response, and be relatively stable under aseptic husbandry conditions in individually ventilated cages. He and his team have designed the Stable Defined Moderately Diverse Mouse Microbiota 2 (sDMDMm2) consortium, which has been successfully transferred to other institutions and remains reasonably stable over time if the host animals are consistently fed a standardized diet. If the diet is changed, both representation and transcriptomic signatures of the consortium members change significantly.

In conclusion, Macpherson stated that “one microbiota is insufficient for a scientific community,” but he believes that significant progress can be achieved by analyzing a microbiota that is standardized across multiple institutions. Since it is possible to maintain defined microbial consortia and isobiotic strains with reasonable stability over time, experiments can be designed with small sample sizes and/or involving multiple institutions. He stressed that no one isobiotic model should be considered exclusive. Instead, researchers should view using isobiotic models in a manner that matches particular situations to particular models. This targeted use, he believes, is statistically more powerful and also

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

contributes to the Three Rs,1 as it reduces the number of animals necessary to perform an experiment.

COMPLEX GNOTOBIOLOGY: AN EMERGING PARADIGM IN THE ERA OF NEXT-GENERATION SEQUENCING

The Mutant Mouse Resource and Research Center that Franklin directs collects the mutant mice that other investigators generate. These mice come into the center with a particular microbiome, but their microbiomes are likely to be different when the center distributes them to other investigators. “Is that problematic for the phenotypes of these models that we have collected and are distributing? Intuitively, we thought that it could be,” said Franklin.

The problem with trying to answer that question rests with the fact that the animals that come into the facility are not gnotobiotic and have complex communities of microbes. Next-generation sequencing and advances in bioinformatics provide the opportunity to better characterize these complex communities, though it is not possible yet to take all of the sequence data down to the level of the operational taxonomic unit, said Franklin. In addition, sequence data, even with the help of new predictive statistical tools, cannot provide information on functionality and overall phenotype. He added that, while studies with classical gnotobiotic animals with defined communities provide important insights about function, it might be important to take what these studies show and see if the results hold true in animals with a complex microbial community with other bacteria, viruses, fungi, and organisms.

Franklin and his colleagues began exploring the issue of complex microbiota by looking at whether the gut microbial communities vary in contemporary rodent colonies. “We knew that animals produced at Jackson Labs and Taconic were probably different, but we wanted to look beyond those,” said Franklin. Principal component analysis comparing the microbiota of animals from Jackson Laboratories and Harlan Sprague Dawley, another supplier, showed differences in certain families of bacteria (Ericsson et al., 2015b) that were small between animals from the same vendor versus between animals obtained from the two different vendors.

He and his colleagues also looked at microbiota during the first weeks of a mouse’s life and found that diversity is low at week 1, increases dramatically in week 2, and by week 3 it is similar to that of an adult animal. It would be useful, said Franklin, to know the physiological changes during this period and how these may impact conditions later in life.

___________________

1 The Three Rs [3Rs] refer to the concepts of replacement, refinement, and reduction. Russell, W. M. S., and R. L. Burch. 1959. The Principles of Humane Experimental Technique. London: Methuen and Co.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

In another study, he and his colleagues examined the effect of diet, bedding, and housing on the mouse microbiome. While there was some variation among most of the combinations, when the mice were housed in static micro-isolators with Aspen bedding there was a marked difference in cecal, but not fecal, microbiota. “We rely on feces, but there is a lot going on upstream that we may not be detecting by only focusing on feces,” said Franklin.

Diet, bedding, and housing are just three of the many variables that can modulate microbiota, but the important question, Franklin said, is whether those shifts matter, and he believes they do. An initial proof of concept experiment examined the effect of introducing microbiota from mice obtained from three commercial suppliers into a knockout mouse model of IBD. The results showed that different microbiota present in rodent communities might be modulating disease phenotypes (Ericsson et al., 2015a). Experiments with a rat model of colon cancer produced similar variations in disease phenotype depending on which of three commercially available rats served as the microbiota source. These experiments, Franklin noted, are identifying targets to be explored in gnotobiotic studies at some point in the future, just as banking feces could potentially allow the reconstitution of phenotypes that have disappeared over time. When investigators move from one laboratory to another, he explained, they might lose the phenotype they were studying, so they request soiled bedding from their former animal facility, add it to the cages housing their animals at their new facility, and reconstitute the phenotype. “This is indirect evidence that the microbiota is playing a role,” said Franklin.

His team tested whether banked feces can reconstitute a gut microbiome and found that feces from a low-diversity donor did not reconstitute the original microbiome when transplanted into a high-diversity recipient treated with a cocktail of four antibiotics (Ericsson et al., 2017). The results were the same, he added, regardless of whether they used fresh or frozen feces or a cecal transplant. Going from a high-diversity donor into a low-diversity recipient treated with the same antibiotic cocktail is partially successful. He also noted that, when two animals with different microbiota live in the same cage, their microbiomes hybridize, something that may be important depending on the phenotype in question. As to whether mouse gut microbiomes are translatable to the study of human immune responses, he noted that a number of investigators are exploring this issue and finding that many organisms other than bacteria can trigger responses that would be useful for studying human diseases (Baxter et al., 2014; Beura et al., 2016; Chudnovskiy et al., 2016; Reese et al., 2016; Tan et al., 2016; Weldon et al., 2015; Wu et al., 2010; Zackular et al., 2016).

As a final comment, Franklin said that improving the definition of gut microbiota could help minimize variability and reduce the number of animals needed to properly power studies. He also raised the possibility that the microbial content of feces could serve as a biomarker for non-terminal experimental end points. “Can we start looking at the feces and what is happening in some of our disease models and say it is time to shut down?” he asked. As to whether the field should move toward using a standardized microbiota in future experiments,

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

he believes the answer is no, that perhaps it would be preferable to have a collection of complex microbiomes that can be used as tools.

THE ROLE OF HOST GENETICS

To get some insights into factors that lead to microbiota variability across members of a species, including humans, Lusis and his colleagues have been studying a set of some 100 commercially available inbred strains of mice selected for their genetic diversity. The genome of the members of the Hybrid Mouse Diversity Panel has been sequenced, or at least densely genotyped (Bennett et al., 2010). The diversity in their microbiomes is similar to the diversity observed in human populations (see Figure 5-1). The variability within a particular mouse strain is small compared to the variability between strains, said Lusis, which suggests the influence of a host genetic component on the variance in microbiota. The consistency within a strain could, for example, result from maternal seeding, since all of the mice in a strain were derived initially from the same mother.

To address that question, Lusis and his colleagues analyzed the genome sequences of the strains in the Hybrid Mouse Diversity Panel to map how they are related to one another given that they were all derived from a pool of pet mice around 100 years ago. Based on the sharing of particular microbes they calculated a measure of heritability, that is, the relationship between genetic and phenotypic similarity (Org et al., 2015). The results of these calculations were surprising. “Heritability is high, unexpectedly high,” said Lusis, which means that the genetics of the mouse strain, given a common environment, accounts for about 50 percent of the variability of the microbiome in that strain. A similar study of some 1,200 monozygotic and dizygotic twins in the United Kingdom performed the same analysis and calculation and found for a number of microbial genera that heritabilities were 20 to 30 percent (Goodrich et al., 2016), which Lusis said is consistent with his results in mice. In comparison, the heritability of heart disease and type 1 diabetes are approximately 40 percent and 70 percent, respectively.

Given the presentations during the first day of the workshop, it should not be surprising that host genetics play an important role in microbiome composition, said Lusis. “We have talked about how we have adapted to the microbiome over millions of years, and that there will be factors we produce that affect the microbiome,” he said. “If those factors vary in the population, then the microbiota will vary in the population.” As an example, he and his colleagues performed a small experiment in which they compared the microbiomes of gonadectomized male and female mice to controls and found that the microbiomes shifted in composition between the matched groups. Testosterone replacement in the gonadectomized males shifted their microbiomes back to match those of the sham-treated animals.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

While genetics contributes to microbiome composition, Lusis said that diet and environment will trump genetics. In fact, an experiment looking at the effects of gene-by-diet interactions found substantial changes in microbiota within the same strains depending on whether they were fed standard mouse chow, a high-fat and high-sucrose diet, or a high-fat and high-cholesterol diet. The changes, though, are not additive or linear and are still a function of genetics, he added. “The genetics determines whether an individual, or a mouse in this case, responds by changing a lot or not at all,” said Lusis.

Lusis also discussed how he and his colleagues have been using host genetic variation to study host-gut microbe interactions. “If there is this big genetic component to the composition of gut microbiota, we should be able to apply genetic mapping to identify the genes responsible,” he said. A genome-wide association study of the mice fed a high-fat and high-sucrose diet identified one or two loci that correlated with the abundance of a number of microbiota genera (Org et al., 2015). He and his colleagues are now working to identify the genes controlling gut microbial content.

It is also possible, said Lusis, to look for correlations between gut microbiota and host traits to identify candidate bacteria that influence host physiology or disease. His research group measured fat gain in response to a high-fat, high-sucrose diet and found that some strains respond greatly and others very little. The strains that had high proportions of Akkermansia muciniphila, for example, tended to not gain weight in response to this diet (Parks et al., 2013). “This is not causality, just correlation, but it allows you to formulate a hypothesis that you can then test.” In fact, when they gavaged A. muciniphila into obesity-prone mice—the mice were not germ-free or treated with antibiotics—and later started them on the high-fat, high-sucrose diet, the increase in body fat, plasma lipids, and glucose metabolism was less than when the mice were gavaged with heat-killed A. muciniphila. As Patrice Cani noted in his presentation, he and his colleagues have purified a membrane protein from this bacterium that improves metabolism in obese and diabetic mice (Plovier et al., 2017).

One practical application of these findings, said Lusis, relates to how his group treats mice they purchase from their mouse supplier. “If we get mice from Jackson Laboratories, they are very different from the mice in our place,” he said. Instead of using them in experiments immediately, his research group first breeds the mice in their vivarium for a few generations.

In closing, he said that embracing diversity presents an opportunity to unravel the incredible complexity of microbial interaction with the host. “Simplifying things is the classic way that scientists have operated to dissect mechanisms, but at the same time, in terms of microbiota, I think looking at diversity is important as well,” said Lusis. He also said that it is important to look at both human and mouse populations to identify similarities and differences.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

THE ROLE OF IMMUNOLOGICAL VARIATION

In 2009, researchers published what Faith considers the key paper concerning the interactions between the microbiome and the immune system (Ivanov et al., 2009). This paper showed that colonization of the mouse intestine by segmented filamentous bacteria originally absent from that particular murine strain would trigger a robust immune response that was protective against infection by an intestinal pathogen. These experimental results were important, he said, because they showed that a specific microbe could change the immune system on top of a complex background.

Faith noted that mice from Jackson Laboratory and Taconic are frequently used in immunological studies of the microbiome which have repeatedly demonstrated that differences in the microbiome affect an animal’s immune response. For example, investigators studying the anticancer effect of the protein PD-L1 found that Jackson mice treated with this molecule produced a vigorous antitumor response, whereas Taconic mice given the same treatment produced a much smaller response (Sivan et al., 2015).

Faith and his coworkers have investigated the relative versus absolute abundance of different microbiome components. Measuring relative abundance, he said, yields an incomplete picture of how the microbiota responds to various challenges. For example, dosing a Jackson mouse with the antibiotic vancomycin eliminates the animal’s microbes, and plotting the relative abundance of different bacteria as the microbiome recovers would suggest that a particular bacterial phylum, the Firmicutes, expands rapidly in the intestines by day 7. A plot of absolute abundance shows that, in fact, the microbial community remains decimated for at least another week. However, when they repeated this experiment using Jackson mice ordered on a different day, the results were markedly different—vancomycin had little effect on the total microbial biomass, though the absolute response showed there had been a bloom of vancomycin-resistant Akkermansia. From these results, Faith concluded that variation of microbiota across animal facilities, between rooms, and across time is confounding results.

Faith’s solution to the non-standardized microbiota of a supposedly reference murine strain is to work with gnotobiotic mice and to take advantage of the fact that gnotobiotic mice housed in the same cage will maintain their defined microbiome communities, assuming the animal facility staff are well trained and follow strict handling procedures. From a practical perspective this implies that, when students leave his laboratory, they can re-create their animals in a new facility merely by taking a sample of the microbiota used to create the gnotobiotic animals and innoculating axenic animals housed at their new location. He noted that his research group has developed a robotic microbiota “manufacturing” system that can reliably assemble identical standardized microbial communities. Using this system, he and his colleagues have produced defined microbiomes from more than 50 human fecal samples (Faith et al., 2013; Goodman et al., 2011).

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

Using defined colonies derived from fecal samples from humans with or without IBD, Faith and his colleagues have demonstrated that human gut microbes transfer colitis to susceptible mice by measuring weight loss in the recipient mice. The response of the mice to different human microbiota was variable—the microbiota from some healthy humans produced symptoms of colitis in the recipient mouse, for example. This variability, said Faith, points to the need to examine samples from multiple individuals and look at the resulting distribution of immune profiles in the recipient mice. “I think the most frequent question at microbiome talks is, ‘What is normal?’” said Faith. In fact, when his group assessed the immune-stimulating properties of different human microbiota in an unchallenged mouse model, the responses fell along a spectrum.

Because every human microbiota triggers an increase in regulatory T cell production in the recipient mice, Faith and his colleagues set out to determine which bacteria were responsible for this response (Faith et al., 2014). They identified seven different bacteria that could increase regulatory T cell numbers relative to germ-free animals. Using a microbial community derived from an individual with Crohn’s disease, his group identified an immunomodulatory effector strain of Enterobacteriaceae as measured by its ability to increase Th17 cells (Ahern et al., 2014). They were able to replicate this response by adding this specific bacterium to a commercially available seven-member model community that by itself did not trigger an increase in Th17 cells.

Going forward, Faith said that these types of immunological experiments would benefit from developing standard operating procedures designed to keep animals free from known pathogens and enable keeping microbiota reasonably constant for several generations. A goal for the research community, he added, should be to create a set number of microbiotas with different properties that a laboratory could order from a vendor to create a defined human microbiota in a mouse. “Logistically, this is complicated, but I think more important than anything would be to change SOPs [standard operating procedures] and training of animal staff to be able to handle this,” said Faith.

He said he would also like to see the development of an effector strain collection of individual or groups of microbes with known function and known ability to engraft on defined microbiotas. “If we all had the same baseline communities, we would know how good each effector strain is at invading a certain number of communities,” said Faith. It would be good to know how robust a newly discovered virus is across the standard defined communities, for example, to understand how well that virus can manipulate a particular immune cell population. To make these resources available, the field needs to develop a system of government suppliers, commercial vendors, and internal standard operating procedures to standardize microbiotas worldwide.

STANDARDIZING AND CHARACTERIZING DIETS

The effects of diet on the human microbiome are difficult to analyze: humans adhere poorly to standardized dietary regimens, diet can have profound

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

impacts on host biology independent of the gut microbiota, and intensive controlled feeding experiments and large outpatient cohort studies are expensive and challenging to complete. Animal models can address each of these challenges, said Wu. Animal studies, for example, enable tight control over defined diets for long periods, while germ-free animals allow researchers to examine the effect of diet independently of the gut microbe.

It is fundamentally important, he said, to understand that diet not only shapes the composition of the microbiome but also serves as a substrate for the microbiota to produce molecules that can circulate widely throughout the body and affect distant organs (Holmes et al., 2012). For example, consumption of a particular milk fat delivers more sulfated bioacids to the gut microbiota, triggering a bloom of the sulfate-reducing bacterium Bilophila wadsworthia, which in turn stimulates the immune response that exacerbates colitis in a mouse model of IBD (Sartor, 2012). Similarly, a high-fat diet provides choline, which is metabolized to trimethylamine. As Federico Rey previously noted, the liver then converts trimethylamine into trimethylamine-N-oxide, a molecule that accelerates coronary vascular disease (Wang et al., 2011).

One human disease that diet affects is IBD and its many manifestations, and in fact, dietary modification is a first-line therapy in Europe, Japan, Israel, and some U.S. and Canadian centers for Crohn’s disease. More people would use dietary therapy, said Wu, except that these diets are monotonous, often unpalatable, and require delivery by nasal gastric tube. In addition, said Wu, “Despite their efficacy, we really do not understand how they actually work.” In that regard, his goal is to answer two fundamental questions:

  1. Does this exclusive enteral nutrition (EEN) provide something “good” for patients with IBD that is not abundant in the regular diet?
  2. Does the consumption of EEN exclude something that is “bad” for patients with IBD in the regular diet?

Various animal models have shown that substances present in the human diet today, such as artificial sweeteners and emulsifiers that were not present several decades ago, alter the microbiota in a way that favors inflammation said Wu (Chassaing et al., 2015; Suez et al., 2014). While he does not claim that artificial sweeteners and dietary emulsifiers cause IBD or any type of disease, this idea is worth exploring. Work from Jeff Gordon’s group has shown in both humans and mice that diet does not have to alter microbiota composition to produce a physiological response (Faith et al., 2011; McNulty et al., 2011). For example, humans and gnotobiotic mice fed a fermented milk product with live microbes experienced no significant change in the composition of their microbiota, but there was a specific and reproducible transcriptomic signature, related to polysaccharide metabolism, seen in both humans and mice. This type of experiment, said Wu, shows that researchers can use gnotobiotic mice to mimic and understand the human response to a particular dietary intervention.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

Computational biologists, Wu noted, have begun to mine clinical metadata, including microbiome data, to predict how someone will respond to diet. One group of investigators, for example, used genomic flux modeling to predict how certain microbes would respond to diet and how that response would lead to changes in serum amino acid levels (Shoaie et al., 2015). Another group devised a machine-learning algorithm that integrates various physiological parameters, dietary habits, physical activity, and data from gut microbiota to predict personalized postprandial glycemic response to meals (Zeevi et al., 2015).

Numerous research groups, said Wu, have shown that the gut microbiome can have a significant effect on the metabolome of the host animal (Wikoff et al., 2009; Xie et al., 2013; Zheng et al., 2011). Researchers have for the most part conducted these studies using germ-free and colonized mice, and studies have yet to confirm if these results hold true in humans, he added. “I personally believe that a lot of the input, at least in the metabolome in terms of diet, is independent of the gut microbiota,” said Wu.

As Lusis noted earlier, though, diet certainly has a strong effect on mouse microbiota, and Wu and his collaborators have shown the same effect in humans in a study of 15 vegans and 16 omnivores. They have also demonstrated that the plasma and urinary metabolomes of omnivores and vegans differ to such an extreme that a computational analysis of an individual’s plasma metabolome can predict with 94 percent accuracy whether a person is a vegan or omnivore (Wu et al., 2016). Despite these huge differences, the gut microbiota composition of vegans and omnivores differed only modestly, as did the diversity of the microbiomes. One explanation could be that organisms other than bacteria—fungi, viruses, or bacteriophages, for example—could be responsible for the observed metabolomic shifts, and this is something Wu plans to examine in a future study.

These results, said Wu, seem to contradict those of a large number of studies, including one he conducted showing that a change in diet rapidly and reproducibly alters the human gut microbiome (David et al., 2014; O’Keefe et al., 2015; Wu et al., 2011). These studies, however, all involved relatively extreme dietary changes and were of relatively short duration—the vegans in his study had been so for at least six months. In addition, Wu learned from speaking with these investigators that the variability between subjects was far greater than the variability in one individual.

With regard to the challenges of studying how diet affects the mouse microbiome and translating those results to humans, Wu said the issues include the fact that mice will eat their own feces and the mouse chow diet is monotonous relative to the variability of the human diet. In addition, mouse digestive physiology and its response to diet is different from that of humans—mice, for example, are hindgut fermenters and have a large cecum, whereas humans, who are not, have a small cecum—and the response of the endogenous mouse gut microbiota to diet differs in magnitude and consistency relative to that in humans. As far as the studies themselves go, the lack of a standardized mouse chow and whether it is sterilized or not can be a cause of variance.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

THE ROLE OF BIOLOGICAL SEX

Research has shown that males and females have different sensitivities to infectious disease (Úbeda and Jansen, 2016) and responses to vaccination (Voysey et al., 2016). There are also differences in the incidence of various cancers and autoimmune diseases between the two sexes (Klein and Flanagan, 2016) (see Figure 5-2). Most studies that aim to explain these differences have focused on differences in sex hormones, overexpression of X-linked genes, and even the expression of mitochondrial genes. What these studies often overlook, Chervonsky said, is that the microbiota could be responding to one of the strongest biological stressors, sex; that there exist sex-specific microbiota; and as several speakers have noted, that the microbiota can have a marked effect on its host’s immune system.

Studies looking at the effect of sex on autoimmune disorders, such as type 1 diabetes, have largely used non-obese diabetic (NOD) mice, a strain of NOD mice in which the incidence of diabetes is 1.3 to 4.4 times higher in females than males. However, Chervonsky and his colleagues found that this gender bias disappeared in germ-free NOD mice (Yurkovetskiy et al., 2013). Their analysis of the microbiota in post-pubertal male and female mice showed that there were marked sex differences, but these differences normalized when the males were castrated, which Chervonsky said confirmed that androgens influence gut microbiota composition. They then looked at gnotobiotic males and females colonized with the same microbiota and again found clear differences after puberty. Further analysis found that, while there are always microbiome differences between males and females, the differences change, showing that there is no male-specific microbiota signature. One conclusion, said Chervonsky, is that gender bias seen with type 1 diabetes does not depend on the specific microbial lineage. It is also possible, he noted, that the expansion of specific microbial lineages is also irrelevant to the gender bias of disease, but experiments with individual bacterial lineages showed that not all bacteria can influence a gender bias and that bacteria of very different families can affect gender bias.

One possible hypothesis to explain a connection between microbiota, autoimmunity, and sex predicts that microbes can affect the levels of hormones that reduce autoimmunity. It also infers that male microbiota should affect disease development in females. To test this hypothesis, Chervonsky and his colleagues tested the effect of colonizing mice with different bacteria and found that some bacteria induce a rise in testosterone levels, whereas others do not. They then did an experiment using NOD mice and found that mice with a wide range of testosterone levels, including very low levels, and a protective microbiota did not score high on a marker for diabetes. However, mice without a protective microbiome did develop the signs of diabetes. Moreover, protective microbiota from males transferred to female NOD mice had no protective effect in females, indicating the protective bacteria require male hormones to produce that effect. Taken together, he said, these results mean it is likely that this hypothesis is incomplete.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×

A so-called dual-signal model, in which both androgens and microbes work in concert to reduce type 1 diabetes in males, is more likely to explain gender differences, said Chervonsky. In this model, some hormones might amplify some microbes and some microbes might amplify hormone levels. “These two signals do not have to be applied simultaneously, but can be differential effectors during development,” he said. He noted that he and his colleagues are testing this model and have some early evidence to support it.

Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 29
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 30
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 31
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 32
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 33
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 34
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 35
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 36
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 37
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 38
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 39
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 40
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 41
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 42
Suggested Citation:"5 Experimental Reproducibility Using Gnotobiotic Animal Models." National Academies of Sciences, Engineering, and Medicine. 2018. Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop. Washington, DC: The National Academies Press. doi: 10.17226/24858.
×
Page 43
Next: 6 Establishing and Evolving Gnotobiotic Facilities »
Animal Models for Microbiome Research: Advancing Basic and Translational Science: Proceedings of a Workshop Get This Book
×
Buy Paperback | $45.00 Buy Ebook | $36.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

The surface of the human body and its mucous membranes are heavily colonized by microorganisms. Our understanding of the contributions that complex microbial communities make to health and disease is advancing rapidly. Most microbiome research to date has focused on the mouse as a model organism for delineating the mechanisms that shape the assembly and dynamic operations of microbial communities. However, the mouse is not a perfect surrogate for studying different aspects of the microbiome and how it responds to various environmental and host stimuli, and as a result, researchers have been conducting microbiome studies in other animals.

To examine the different animal models researchers employ in microbiome studies and to better understand the strengths and weaknesses of each of these model organisms as they relate to human and nonhuman health and disease, the Roundtable on Science and Welfare in Laboratory Animal Use of the National Academies of Sciences, Engineering, and Medicine convened a workshop in December 2016. The workshop participants explored how to improve the depth and breadth of analysis of microbial communities using various model organisms, the challenges of standardization and biological variability that are inherent in gnotobiotic animal-based research, the predictability and translatability of preclinical studies to humans, and strategies for expanding the infrastructure and tools for conducting studies in these types of models. This publication summarizes the presentations and discussions from the workshop.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    Switch between the Original Pages, where you can read the report as it appeared in print, and Text Pages for the web version, where you can highlight and search the text.

    « Back Next »
  6. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  7. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  8. ×

    View our suggested citation for this chapter.

    « Back Next »
  9. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!