The importance of incorporating genetic diversity in biomedical research

A group of diversity outbred mice from The Jackson Laboratory taken in 2016. A group of diversity outbred mice from The Jackson Laboratory. 

Research using one or two strains of inbred mice fails to capture important aspects of the genetic diversity found in human populations. JAX scientists are working to make it more feasible for researchers to incorporate diverse background genetics to better model human disease in mice.

This week, a research team led by Jackson Laboratory Professor Martin Pera, Ph.D., presented an important step forward for biomedical research. Their in vitro stem cell research platform represents a new way to incorporate genetic diversity into mouse-based research, an important issue in the ongoing discussion regarding how to bridge the gaps between research discovery and clinical delivery. JAX, arguably the world leader in mouse expertise and resources, has pushed mouse genetic diversity forward for years, particularly with the Collaborative Cross (CC) inbred strains and Diversity Outbred (DO) population, with notable results. Previous advances have been with in vivo populations and platforms, however, limiting feasibility for researchers without significant vivarium space, husbandry expertise, and/or budget. The work of Pera and colleagues promises to help.

Success and limitations

A huge number of medical advances over the past century can be traced back from insights gained from mouse-based research, with recent examples including the potential for immune checkpoint inhibition (cancer immunotherapies, e.g., Keytruda) and GLP-1 receptor activation (weight loss, diabetes and cardiovascular disease treatment, e.g., Wegovy). Nonetheless, most promising discoveries made in mice over the years have failed to translate to effective human therapies. There are many factors involved, but an important one is found in our genetics.

While mice and humans obviously have genetic differences, their genomes and the functions of a vast majority of their genes are very similar. But what human patient populations have that most mice used for biomedical research don’t are different background genetics. Since the first human genome sequence was completed, it’s become very apparent that small genomic differences between people can be extremely important in the clinic. Over the years, however, most research with mice have used one or two of a few very well-characterized inbred strains, which is analogous to testing drugs and other therapies in close relatives from only one or two families. What works for them may not work for most other people, and vice-versa. In addition to research with the CC and DO diversity platforms, scientists at JAX are developing innovative ways to address the problem.

Machine vision

Obviously, research into the genetics of behavior requires living animals to exhibit said behavior. But traditional behavioral research is limited in scope, focusing on inexact assays and small numbers of mice because observing more hasn’t been humanly possible. And it still isn’t, which is where video and machine learning come in. JAX Associate Professor Vivek Kumar, Ph.D., has developed both non-intrusive video systems—human presence changes behavior—and machine learning algorithms capable of capturing and analyzing massive amounts of behavioral data, with accuracy rates matching expert human observers. And it’s already being put to good use for studies on grooming, gait, sleep, aging, and more.

The grooming study also provides an excellent proof point for why it’s important to incorporate genetically diverse mice in research. A specific mouse strain known as BTBR had been thought to be a possible model for autism spectrum disorder, in part because of a high level of repetitive behavior, in this case grooming. But, importantly, the high amount of time grooming was in comparison to the most commonly used mouse strain in research, C57BL/6J. When able to add in more strains—many more, for a total of 62—Kumar and colleagues found that BTBR, while having high levels of grooming, doesn’t stand out. In fact, there were 11 other strains with similar patterns. The results suggest that BTBR isn’t necessarily a good model strain for ASD based on its grooming patterns and other factors should be considered.

The variability of infectious disease

The COVID-19 pandemic provided a stark reminder of how variable the human response to disease can be. Clear trends emerged—the elderly and those with existing comorbidities were most vulnerable—but at first researchers were unable to study the biological mechanisms underlying the variability. Re-deriving a mouse with a humanized ACE2 receptor was an important first step, as wild-type mice were unable to be infected with the initial SARS-CoV-2 variant. But disease progression in the so-called K18-hACE2 transgenic mice, made in the C57BL/6J strain, was invariably swift and lethal. To investigate why the disease was so different in different people, Professor and JAX Mammalian Genetics Scientific Director Nadia Rosenthal, Ph.D., F.Med.Sci., knew that more mice were needed, with different genetic backgrounds. For the purpose, she and her colleagues turned to eight highly genetically diverse strains, the ones originally used to develop the CC and DO mouse populations.

By crossing the K18-hACE2 mice with the other strains, the team generated an effective platform for studying COVID-19 in a range of genetic backgrounds. And sure enough, they found that the original C57BL/6J mice were the most susceptible, while others were less so. In fact, one strain known as PWK was highly resistant, showing little or no sign of disease, and three other strains had consistently different levels of disease severity between male and female mice. In addition to thoroughly characterizing the disease characteristics across the panel, the researchers compared the immune responses and other factors between the strains. Their findings demonstrated that the timing and regulation of type 1 interferon response was particularly important for control of virus replication and resolution of pro-inflammatory immune responses, providing a focal point for future mechanistic studies.

A leap forward for genetically diverse in vitro research

JAX researchers have access to a wealth of mouse resources, making studies using eight or even 62 different strains possible. Keeping so many mice “on the hoof,” so to speak, can be unfeasible for many scientists elsewhere, however, because of lack of space, expertise, budget or all three. Looking to expand genetic diversity in in vitro research, a JAX team led by Martin Pera focused on mouse embryonic stem cells (mESCs). Such cells can, in theory, be made to grow and differentiate into various cells and tissues in culture, but doing so across multiple mouse strains had proven to be a challenge. The researchers worked with the same eight strains used for the SARS-CoV-2 work detailed above but found that they were only able to induce the mESCs from one strain, known as 129, to develop into neurons using existing protocols. With some innovative thinking, they were able to create revised protocols that worked across all eight strains, as well as tweak them so they could produce a variety of different neuron types.

With the in vitro platform developed, they conducted a proof-of-concept study that followed up on previous research in human induced pluripotent stem cells (iPSCs) and 129 mESCs. In humans, dysfunction of a gene called DYRK1A is associated with neurodevelopmental disorders including autism spectrum disorders and intellectual disability. Pera and colleagues found that DYRK1A inhibition in human iPSCs interfered with the cells’ transition from a pluripotent state to differentiation into neurons. Working with the recently developed mESCs, they observed important differences in the different cell lines with DYRK1A inhibition. Some strains were unaffected and able to differentiate into neural progenitor cells, including 129. One strain in particular, C57BL/6J, was strongly inhibited, mirroring the human iPSC response. Additional work indicated that C57BL/6J likely represents the best strain to work with for in vivo modeling of DYRK1A-related human conditions and diseases. The study provides strong evidence that the in vitro platform can be used to assess background genetic effects, characterize strain differences, and efficiently match mouse and human findings to better model human disease in vivo.