Part III: Best Practices for Study Design
Microbiome Research Best PracticesThe Navigating the Challenges of Studying the Microbiome in Mouse Models workshop, sponsored by Taconic Biosciences, brought together leading academic and industry researchers to identify the key considerations, strengths, and weaknesses for studying the microbiome in mouse models.
The panelists and attendees discussed these topics in detail and the main findings are summarized here.
Key findings of the workshop:
- The Role of In Vivo Studies
Advances in bioinformatics, high-throughput sequencing, and proteomics have dramatically accelerated our understanding of the composition of the microbiome. This wealth of information has generated many insights into how specific microbiome characteristics are associated with health and disease.
Despite these advances, a fundamental limitation of these studies is the inability to identify, confirm, and understand the mechanistic basis of these microbiome linkages with disease. Association studies cannot identify whether microbiome links are truly causative, or are secondary to other pathological processes. We need a better understanding of how the microbiome interacts with the body, and how these processes become perturbed in disease, as well as establishing the causal role of the microbiome in specific conditions.
A key benefit of in vivo studies is the ability to test specific hypotheses regarding the microbiome-host relationship in a controlled and tractable manner. These mechanistic data are critical to identifying new disease mechanisms and targets for therapeutic intervention.
Clinical studies in human populations are messy; varying environmental, dietary, and genetic factors pose a challenge in identifying true causal influences. In contrast, animal models allow for tightly controlled studies in which confounding environmental and genetic factors are eliminated.
Furthermore, the ability to control and sample the microbiome in an animal model, as well as to control host genetics, makes in vivo modeling an extremely powerful tool for understanding host-microbiome interactions and the influence of the microbiome in health and pathological processes.
A vital tool for in vivo microbiome research is the germ-free animal. Germ-free animals offer a clean slate into which specific microbes or microbial communities can be engrafted, making it possible to study how the microbiota regulates health and disease. Comparing germ-free animals to those with defined or disease-associated microbiota is generally considered definitive proof for a microbiome-driven etiology for disease.
By far, the most common and well-used germ-free animal is the mouse, and gnotobiotic facilities for the research and breeding of germ-free mice are key to advancing microbiome knowledge. Mice benefit from relatively low maintenance costs, a size suitable for gnotobiotic husbandry and breeding, and the wealth of genetically-modified models. The mouse is reasonably like a human in terms of gross anatomy and physiology, although there are differences that should be considered (see item 3).
Other animal models include the germ-free zebrafish, fruit fly (Drosophila), and the pig. The zebrafish and Drosophila models benefit from low-cost generation and maintenance, easily-modified genetics, a rich resource of available models and experimental techniques. However, these models do have limited translatability given their substantial differences to humans in terms of anatomy, physiology, genetics, environment, and diet.
Perhaps the most translatable model is the germ-free pig, which is very similar to humans in many respects; however, the expense of generating and maintaining germ-free pigs, their size and lifecycle, and the lack of genetic models make germ-free pigs unsuitable for most studies.
- Integrating of in vivo studies with in vitro models and -omics data
Microbiome research is complex and interdisciplinary; no single approach or methodology can fully probe how the microbiome contributes to health and disease. Microbiome studies benefit by taking advantage of various models and methodologies; rather than competing with each other, -omics data and in vivo and in vitro studies are powerful, complementary tools for understanding the microbiome.
Results from sequencing data may lead to specific hypotheses, which are then tested in vivo and/or in vitro; these in turn may lead to improved -omics approaches, and so on, with refinement and greater knowledge gained with each turn. This convergence of technologies and approaches is necessary to drive microbiome research further.
- Mouse models as translational tools
Mice and humans are separated by some 65 million years of evolution, but are remarkably similar at the level of fundamental biology. This made mouse models very powerful in research and discovery over the past century.
However, there are differences in anatomy and physiology that are important considerations for microbiome research:These behavioral and anatomic differences aside, there are substantial similarities between the human and mouse microbiome at a gross level1.
- Perhaps the most important difference is diet and gastrointestinal tract anatomy1. The largely grain-based diet of common laboratory rodent chows differs widely from the varied food of humans. This, combined with an enlarged cecum and relatively longer colon in mice, leads to different energy sources and niches for microbiota.
- A second, often overlooked, factor is the difference between the human and mouse immune systems2,3; the immune system, in particular secretory IgA, plays an important role in shaping the microbiome.
- A final consideration is that mice are notoriously coprophagic, which is generally not the case in humans. Coprophagia, or the consumption of feces, acts to transmit microbiomes between individuals, and autocoprophagia likely affects the microbiome within an individual. Although not overtly coprophagic, humans are exposed to (and consume) microbes from a variety of sources, including fecal matter, other humans, animals, and the environment.
It is important to note that there is no such thing as a "human" or "mouse" microbiome, per se. Laboratory mice can have widely different microbiomes depending on where and how they are sourced and housed, and there are substantial differences between vendors, barrier facilities and institutions4.
Laboratory mice also have different microbiomes from wild mice, which are exposed to environmental influences5 and pathogens. Human microbiomes also differ widely6.
This variation in the inherent mouse and human microbiomes can present a challenge when studying and modeling disease. Efforts to control for and minimize microbiome differences should be considered in all mouse studies, regardless of their primary goal.
- Key factors affecting the mouse microbiome
There are numerous factors which affect the murine microbiome, with increasing impetus to control for these in any study.
When using mouse models for any study (whether or not your primary focus is the microbiome), you should consider these factors:
- Source of mice — Conventional mice, bred and raised in different locations, will have a different microbiome. This is true between different institutions and animal vendors. Animal vendors breed and raise their animals under different conditions and in different locations. The inherent microbiome in a mouse purchased from a vendor may differ widely depending the bioexclusion practices in place and the environment in which it was raised. Validate studies for mice sourced from a single location (e.g. barrier location in the case of animal vendors) and use those mice exclusively for such studies.
- Maternal transmission — Mice "inherit" their microbiomes in a similar fashion to all placental mammals, including humans: through vaginal delivery and close maternal association throughout the neonatal period (vertical transmission). Maternal vaginal, skin, mammary, fecal, and oral microbiomes are efficiently transmitted to offspring, and mouse pups tend to resemble their dam in terms of their microbiomes once they have matured. This early inoculation is critical for immune system development and microbiome structure. All things being equal, maternal transmission is the single biggest factor determining the microbiome of a mouse. This can often be a huge confounding factor for studies involving mice derived from breeders with different microbiomes.
- Cage effects — Individual microbiomes tend to drift over time as a function of other factors covered in this section. In the case of individually-ventilated cages (IVCs), where there is little chance for horizontal transmission between cages, the microbiome of mice in a single cage may drift over time, such that they diverge from others in the colony. In contrast, the microbiome of individuals within a single cage tends to remain similar, due to coprophagia and other routes of horizontal transmission. Over the lifespan of an individual mouse, these cage effects may be minor; however, as is often the case with small colonies of genetically-modified mice, these changes may accumulate over generations through maternal (vertical) transmission and founder effects.
- Diet — Laboratory rodent chows are made primarily from various grain-based sources, such as wheat, corn, oats, alfalfa, and soy bean. These sources provide a wide range of fermentable dietary fibers and other bioactive compounds that affect the microbiome. The composition of these chows (grain source and fiber types) may change from lot to lot, and seasonal variations are inevitable. The microbial content of these diets will also vary based on source locations and processing methods; however, sterilized irradiated or autoclaved chow will mitigate effects of viable microbiota. Purified diets, in contrast, are made of highly purified standardized ingredients. Although these may minimize diet effects on the microbiome, the relative lack of fermentable fiber in most purified diets (the majority use non-fermentable cellulose as a fiber source) may have an impact on microorganisms. While not typically considered "diet", mice will consume cage bedding, and differential effects on the microbiome between wood chip, paper, and corncob bedding cannot be excluded.
- Water — Mice are provided water, either through bottles or automated watering systems; this water is typically processed through reverse osmosis, and may be either autoclaved or acidified to prevent bacterial contamination. Contaminated, acidified, or chlorinated water can affect the mouse microbiome.
- Health status — Exposure to pathogens or various commensal bacteria (e.g. segmented filamentous bacteria) will affect the immune system and subsequent microbiome of a mouse. Treatment with antibiotics or other immunomodulatory or bioactive compounds may also affect the microbiome.
- Housing — The caging used to house mice can affect microbial exposure and spread throughout a colony. Open-topped wire-lidded cages vs. IVCs differ greatly in their biocontainment and bioexclusion capabilities. Specially-designed IVCs or gnotobiotic isolators will offer the greatest protection, allowing for the maintenance of either the germ-free or gnotobiotic state with a defined microflora.
- Stress — Stress has a great impact on the microbiome, although the reasons behind this are unclear. Transportation, housing conditions, or disease may impact the microbiome through this mechanism.
- Considerations for study design The microbiome is a key factor influencing the performance and reliability of animal models, which is itself an increasingly important area of research.
These considerations for study design were proposed:
- All mouse models — Even if the primary objective is not to study the microbiome directly, it is likely influencing study outcomes. Close attention should be paid to minimizing microbiome influences on study outcomes, and also to validate protocols on a specific microbiome background. When this is not practical, consider co-housing mice to minimize microbiome differences prior to study onset.
- Fecal microbiota transplant studies (FMT) — Transferring the microbiome of one organism (e.g. a mouse or human patient) to a recipient mouse often involves FMT, or the transfer of fecal microbiota as a donor microbiome source. For this to be successful, the recipient mouse needs to be either germ-free or have its own microbiome depleted with antibiotic cocktails; be aware there are substantial differences between these methods7. Germ-free mice are efficiently and reliably engrafted by an FMT, whereas this is less reliable in antibiotic-treated mice. Antibiotic-treated mice are also affected by off-target effects of drug cocktails. Both models lack the priming effect of the microbiome on the immune system in early life; however, inoculated germ-free mice can be bred under gnotobiotic conditions to generate offspring that have been associated with specific microbiomes since birth.