Inbred laboratory mouse strains, though ideal for research, fail to recapitulate the genetic diversity that often steers the course of human diseases. For better Alzheimer’s disease models, researchers led by Gareth Howell, Jackson Laboratory, Bar Harbor, Maine, have taken a dip into the deeper mouse gene pool. By breeding APP/PS1 mice with strains derived from common house mice, they claim to have created three new models that better capture AD pathology, including neuronal cell loss and vascular damage. Strain-to-strain differences stemmed mostly from the animals’ neuroimmune responses to amyloid. The study was published May 31 in PLoS Genetics.

  • Wild mice harbor more natural genetic variation than laboratory strains.
  • Wild-derived strains have more varied immune responses to amyloid.
  • They better mimic the neurodegeneration and vascular disease seen in people.

“This study complements recent and past work showing that genetic background can significantly influence the onset, extent, and progression of amyloid-associated pathology in mouse models,” said Joanna Jankowsky, Baylor College of Medicine, Houston, Texas. She thinks the animals will be useful genetic tools.

Renzo Mancuso, KU Leuven, Belgium, agreed. “I think these models could be valuable as they bring us a step closer to the genetic complexity we see in humans,” he said. But he cautioned that mice have limited genetic orthology with people and therefore will only partially address the broad genetic aspects of AD. “Having said that, it is nice to see that multiple groups are trying to tackle the issues around the lack of reliable models of AD,” he said.

Some scientists think that poor mouse models help explain the failure to discover new treatments for Alzheimer’s disease. Inbred amyloidosis strains fail to recapitulate the tau pathology and extent of neurodegeneration found in AD, and hundreds of treatments that seemed promising in mouse studies failed in the clinic. Scientists also question how genetically identical lab mice can model what occurs in people, who are so genetically diverse. Efforts to introduce more genetic diversity into AD mouse models have begun. In one recent study, also from Jackson Labs, scientists bred 5XFAD transgenic mice with a panel of 27 genetically diverse research strains, creating a spectrum of pathology and memory loss phenotypes (Dec 2018 news). 

Bad Mix. Amyloid deposits (blue) occur in the parenchyma and in blood vessels of the cortex in WSB-APP/PS1 mice (left). Affected vessels (right) leak fibrin (red). [Courtesy of Onos et al., 2019 PLoS Genetics.]

Howell’s lab took another tack. First author Kristen Onos crossed the amyloid-overexpressing APP/PS1 Black6 (B6) mice, one of the most widely used AD models, with animals derived from common house mice caught in the eastern United States, Thailand, or the Czech Republic. The wild-derived strains, called WSB, CAST, and PWK, have up to three times the number of single-nucleotide polymorphisms and genomic structural variants as B6. Added to that, the three strains carry the natural genetic variation that comes from living in the wild. B6, on the other hand, were purposely bred as pets, then research animals, starting a century ago.

Like B6-APP/PS1 mice, 8-month-old male and female WSB-, CAST-, and PWK-APP/PS1 animals all harbored cortical and hippocampal amyloid plaques. Surprisingly, the CAST- and WSB-APP/PS1 strains also had amyloid deposits in blood vessels in the brain, mimicking cerebral amyloid angiopathy (CAA). The affected vessels leaked (see image above). “These mice have microhemorrhages where the CAA occurs,” said Onos. This happens in some patients with AD, but has been difficult to model in mice, she said. The animals had evidence of neurodegeneration, too. Female WSB-APP/PS1 mice had approximately 20 percent fewer neurons in the superior cortex and CA1 region of the hippocampus than did wild-type WSB animals, while female and male CAST-APP/PS1 mice lost neurons in the hippocampus. In contrast, APP/PS1 B6 mice show only minor neuron loss around plaques at this age.

RNA sequencing of whole brain tissue from transgenic and wild-type animals revealed that strain was the biggest driver of gene-expression variation. However, co-expression analysis identified a cluster of 35 genes whose expression changed in all three strains when the APP/PS1 transgenes were introduced. Many were microglial genes involved in neuroinflammation, including the known AD risk genes Trem2 and CD33. Strains varied in baseline expression of the gene cluster, which paralleled numbers of microglia in the brain. For example, WSB and CAST had the lowest baseline expression, and fewest microglia. Of the APP/PS1 mice, PWK-APP/PS1 showed the largest boost in gene expression compared with wild-type. Despite starting out lower in number, the CAST-APP/PS1 microglia were the most proliferative among the three strains, and more of them accumulated around plaques than in other strains.

Plaque Attack. The number of microglia (white) clustered in the vicinity of amyloid plaques (circled) depends on strain. CAST mice show the most robust response. Green cells are astrocytes, while blue marks cell nuclei. [Courtesy of Onos et al., 2019, PLoS Genetics.]

This is an important finding for researchers studying Trem2 and the microglial response to plaques, wrote Mathias Jucker, University of Tübingen, Germany, to Alzforum. “The work suggests we must carefully control for genotype when doing such microglia/plaque analyses,” he said.

What about behavior? The usual lab tests for learning and memory don’t always work with wild-derived strains, which tend to be more aggressive than B6. They jump, bite, and struggle to escape. To measure working and short-term memory, the researchers fashioned a special cover for their Y maze to keep the mice from leaping out. While the transgenes did not diminish working memory in any strain, at least when tested at seven to eight months, the results for short-term memory were less conclusive. In PWK mice, the transgene did not dampen novel spatial recognition, but it did in WSB males and CAST females. For other sex/strain combinations, including B6, the wild-type mice could not complete the task, so the effect of the transgene could not be determined. These difficulties highlight the challenge of finding standard tests that diverse strains can perform, or of devising new, strain-specific tasks based on how the animals behave in the wild, the authors said.

Jankowsky agreed. “There are practical reasons that mouse geneticists have shied away from using these wild-derived strains—as the paper indicates. It is hard to determine which cognitive outcomes are feasible when baseline behaviors vary so much from the inbred strains we’re used to,” she said.

Onos said the group continues to study the mice as they age past the 8-month point. They plan to create new models based on the wild lines by replacing the APP and tau genes with human versions, and incorporating sporadic risk variants such as ApoE4 and the TREM2 R47H variant. That work will be part of the MODEL-AD collaboration, which is generating new mouse models for late-onset AD based on genetic risk alleles found in people (Jan 2017 news). 

Paul Territo of Indiana University, Indianapolis, said the wild-derived mice offer a unique opportunity to study the role of natural genetic variation in mouse models of AD. Their behavior is not a deal-breaker, he said. Territo has helped establish the preclinical testing pipeline for MODEL-AD, and he told Alzforum they are moving away from behavior as a primary readout for phenotyping and testing therapeutics. Instead, the group is using measures that are more directly translatable, including PET and MRI scans, just like those used in the clinic.—Pat McCaffrey


  1. This study complements recent and past work showing that genetic background can significantly influence the onset, extent, and progression of amyloid-associated pathology in mouse models. To the best of my knowledge, this line of investigation started in 2003 with Bruce Lamb testing the impact of different inbred strains on amyloid pathology in his APP-YAC mice (Lehman et al., 2003). Prior to that, Karen Ashe had shown that inbred strain background could influence lethality in APP transgenic lines (Carlson et al., 1997). More recently, Catherine Kaczorowski had a beautiful paper using the BXD recombinant inbred panel to show that even hemizygosity for differing alleles was sufficient to dramatically alter cognitive, biochemical, and transcriptional consequences of the 5xFAD transgenic model (Dec 2018 news; Neuner et al., 2019). I think that this paper from Gareth Howell’s lab adds to this line of work in confirming that genetic background can dramatically influence the traits we attribute to our transgenes.

    I am less sure how I see the new genetically diverse models being used in future. There are practical reasons that mouse geneticists have shied away from using these wild-derived strains—as the paper indicates, it is hard to determine which cognitive outcomes are feasible when baseline behaviors vary so much from the inbred strains we're used to. As I see it, the strongest use for the new wild-derived backgrounds will be for genetic studies. I think that the transcriptional analyses are the most striking of the outcomes tested here, and may be the most informative measure for future work on the effects of genetic background. 

    These recent studies lead me to believe that we should move toward preclinical testing of drug candidates using models expressed on outbred strain backgrounds with more genetic variation between individuals. You'd have to have a substantial effect to reach significance with the added inter-animal variability of this approach. This strategy might help to winnow out at the preclinical stage therapeutic ideas that don't work across a heterogeneous genetic landscape.


    . Genetic background regulates beta-amyloid precursor protein processing and beta-amyloid deposition in the mouse. Hum Mol Genet. 2003 Nov 15;12(22):2949-56. Epub 2003 Sep 23 PubMed.

    . Genetic modification of the phenotypes produced by amyloid precursor protein overexpression in transgenic mice. Hum Mol Genet. 1997 Oct;6(11):1951-9. PubMed.

    . Harnessing Genetic Complexity to Enhance Translatability of Alzheimer's Disease Mouse Models: A Path toward Precision Medicine. Neuron. 2019 Feb 6;101(3):399-411.e5. Epub 2018 Dec 27 PubMed.

  2. Mouse models for Alzheimer’s disease only partially recapitulate the complexity of the human disease. Indeed, despite their aggressive amyloid pathology, current mouse models develop neither neurofibrillary tangles nor neuronal loss, both major hallmarks of AD. This is an important obstacle for the translation of preclinical findings into potential clinical applications. Most of the preclinical research is carried out on the inbred mouse strain C57BL/6. These mice have a very limited numbers of genetic polymorphisms, even across several generations of breeding. On one hand, such a genetic homogeneity has helped researchers to reduce data variability and improve comparisons across different labs. However, genetic homogeneity in laboratory mice poorly matches with the human counterpart of AD, which is mostly sporadic and driven by both environment and combinations of polymorphic alleles.

    Thus, while we are learning a lot from mouse models of AD, unfortunately, this increasing knowledge may have limited impact on our understanding of the actual disease. Onos and colleagues tackled this important topic, backcrossing the most popular C57BL/6 AD models onto different wild mouse strains, characterized by a broader genetic variation. Not surprisingly, wild strains exhibited variable degrees of neuronal loss, severity of amyloid and vascular pathology and microglia activation. Interestingly, transcriptomic analysis highlighted the strain, but not the disease, as the major driver of the gene expression variability. This suggests that the disease-associated gene expression signature identified in the canonical C57BL/6 models may appear remarkably different in mice with a different genetic background.

    Nonetheless, certain genes were consistently affected across all the strains. Most of these genes are critically expressed in myeloid cells, suggesting that certain immune pathways are conserved under AD pathology regardless of the strain differences. Possibly, comparison of transcriptomic alterations amongst multiple strains, both inbred and outbred, may help discern strain-dependent biases, thus highlighting true molecular players in Alzheimer disease.

  3. Lovely contribution to the small but rapidly growing literature on the major impact that genetic background effects have on AD progression and severity. Reductionist models are a great start, but testing against genetic variation (modifiers) is essential to improve translational relevance of models.

    Just two minor tweaks to Pat McCaffery's solid review and summary of the study in PLoS Genetics:

    1. The strains that Drs. Onos, Howell, and team exploited are commonly referred to as wild strains, but at this point they have been inbred at The Jackson Laboratory for many generations—87 generations in the case of WSB/EiJ. This is a good thing—they are a now a wonderful and stable resource for this and many other studies.

    2. The statement about "three times the number of B6" needs a bit of explanation. Members of an inbred strain will be as isogenic as monozygotic human twins. But by studying the effects of AD variants on three different strains, Onos and colleagues have added back a great deal of DNA variation. These three strains will differ at well over 35 million sequence variants—as much or more than the number of common sequence variants in human populations.

Make a Comment

To make a comment you must login or register.


News Citations

  1. Missing Ingredient: New Mice Model Alzheimer’s Genetic Variability
  2. Building Better Mouse Models for Late-Onset Alzheimer’s

Research Models Citations

  1. APPswe/PSEN1dE9 (line 85)

Other Citations

  1. 5XFAD

Further Reading

No Available Further Reading

Primary Papers

  1. . Enhancing face validity of mouse models of Alzheimer's disease with natural genetic variation. PLoS Genet. 2019 May;15(5):e1008155. Epub 2019 May 31 PubMed.