Mouse Model LOADs on the Risk Factors

The search for a good model of late-onset Alzheimer’s disease continues. Researchers led by Adrian Oblak of the Indiana University School of Medicine in Indianapolis and Gareth Howell at the Jackson Laboratory (JAX), Bar Harbor, Maine, have characterized LOAD2 mice, which carry the human ApoE4 gene, the R47H AD risk variant of the TREM2 gene, and have a humanized Aβ sequence in their amyloid precursor protein gene. In a preprint uploaded to bioRxiv on December 20, they reported that 18-month-old mice had no amyloid plaques but did have mild neurodegeneration after a lifetime on a high-fat diet. They detected other changes seen in early AD, such as increased plasma NfL and impaired memory. “What surprised us was the number of phenotypes that required genes, age, and the environment to interact before manifesting,” Howell said. He and Oblak envision using LOAD2 mice fed a fatty diet as a model of preclinical AD.

“The study emphasizes the effective modeling of predementia stages, which are often overlooked due to a focus on dramatic changes,” wrote Maud Gratuze, Aix-Marseille University, France (comment below). “The strategic combination of genetic risk factors and dietary stressors provides insights into age-related neurodegeneration, cognitive deficits, and imaging abnormalities, establishing LOAD2+HFD as a valuable resource for investigating AD.”

Oblak, Howell, and colleagues are involved in the Model Organism Development and Evaluation for Late-Onset Alzheimer’s disease (MODEL-AD) consortium, creating new mouse strains that carry combinations of genetic risk factors for LOAD. LOAD1 mice, homozygous for APOE4 and TREM2 R47H, mustered transcriptomic changes in AD-related pathways in the brain, but no frank AD pathology (Kotredes et al., 2021). To inch the needle toward AD, the scientists added the humanized APP gene in LOAD2, and fed the mice lots of fat and sugar, mimicking the Western diet, since obesity and cardiovascular disease are both risk factors for LOAD (Apr 2021 conference news; Nov 2022 news).

Takaomi Saido, RIKEN Center for Brain Science, Japan, thinks this is a logical approach that will inform on the relationship between LOAD risk factors and AD pathogenesis (comment below). Christian Haass of the German Center for Neurodegenerative Diseases in Munich disagreed. He sees LOAD2 as a model of TREM2 dysfunction in the presence of APOE4 and environmental risk factors rather than of LOAD.

Co-first authors Kevin Kotredes and Ravi Pandey began feeding the animals normal chow or the high-fat diet (HFD) at 2 months old. They analyzed learning and memory, brain metabolism and perfusion, atrophy, gene and protein expression, blood biomarkers, and neuropathology at 4, 12, and 18 months.

Learning faltered in aged LOAD2 mice, especially after eating a fatty diet. At 12 months old, they took a few days longer than wild-type mice to learn which of two shapes shown on a touchscreen would reward them with a sip of sweet water. This touchscreen assay was designed to mimic a task from the CANTAB battery of digital cognitive tests for memory, attention, and executive function (Horner et al., 2013). Over the almost three-week learning period, the LOAD2+HFD mice only picked the right shape about one-third of the time, meaning they did no better than guessing, whereas wild-type mice chose correctly with about 65 percent of the time after a week of training.

The high-fat diet also disrupted neurovascular coupling. By 12 months, glucose metabolism in the entorhinal and perirhinal cortices had ticked up in female mice according to FDG PET, yet perfusion had dropped, as judged by 64Cu-PTSM PET. At 18 months, both glucose metabolism and blood perfusion were up in males and females, which is consistent with hypermetabolism and hyperemia seen in prodromal AD, wrote the authors. At this age, the brain had shrunk by about 5 percent. Areas most affected included the entorhinal, piriform, and cingulate cortices, regions that accumulate neurofibrillary tangles and activate microglia in the early stages of AD (image below; Dec 2023 news; Sep 2021 news). The authors did not look for tangles in these mice, however.

Atrophy with Age. Various brain regions atrophied in male (left) and female (right) LOAD2 mice on a high-fat diet. Cooler colors indicate more atrophy. [Courtesy of Kotredes et al., 2024.]

At a molecular level, bulk RNA-Seq of whole brain pointed to two sets of dysregulate genes. One included immune response genes. The other comprised those plus genes involved in lipid metabolism, oxidative stress, and apoptosis. The latter tightly correlated with the high-fat diet, while both strongly associated with age. To the authors, this meant that age drove inflammatory changes, while diet influenced other AD-related pathways. Along these lines, LOAD2-HFD mice recapitulated some proteomic changes seen in cortical tissue from people who had had LOAD, including fewer synaptic proteins, suggesting poor neuronal health (Feb 2022 news). Plasma TNFα and neurofilament light ticked up by 12 months in LOAD2+HFD mice, further signs of inflammation and neurodegeneration.

Neuropathology dovetailed with the marker and omics data. Eighteen-month-old LOAD2+HFD mice had slightly fewer cortical neurons and fewer GFAP-positive astrocytes than controls on normal chow. Yet the amount of Iba1-positive microglia remained unchanged, perhaps because APOE4 and the TREM2 variant suppressed microglial responses, suggested Howell.

Lastly, LOAD2+HFD mice had no amyloid plaques, but they did have insoluble Aβ42 in their brains, albeit much less soluble Aβ40 and Aβ42 “The drastic reduction in soluble and insoluble Aβ40, coupled with the absence of plaques, challenges the hypothesis that humanized Aβ would favor amyloid pathology, as initially hypothesized by the team,” noted Gratuze. “These unexpected findings underscore the complexity of AD pathophysiology and highlight the need for nuanced investigations.” Howell noted that many things conspire to cause plaque deposition in LOAD. “We are trying to hit the right pathways that elevate levels of Aβ in these mice and we haven’t gotten the correct combination yet.”—Chelsea Weidman Burke

Comments

  1. According to Alzforum, there are 204 murine models of AD, showing the complexity of developing a model that mimics AD perfectly. The choice of the best mouse model depends on the specific questions researchers aim to address; there is no universally perfect model. This study emphasizes the effective modeling of predementia stages in mice, often overlooked due to a focus on dramatic changes.

    The MODEL-AD initiative leads the creation of innovative mouse models, incorporating humanized genetic risk factors for a more accurate replication of Late-Onset Alzheimer's Disease (LOAD) compared to traditional transgenic models. This recently developed LOAD2 model, featuring APOE4, Trem2*R47H, and humanized Aβ, is a notable addition to this pursuit. In this study, mice were exposed to either a control diet or a high-fat/high-sugar diet (LOAD2+HFD) from the age of 2 months.

    By the 18th month, LOAD2+HFD mice exhibited significant cortical neuron loss, elevated insoluble brain Aβ42, increased plasma neurofilament light chain (NfL), and noteworthy alterations in gene/protein expression related to lipid metabolism and synaptic function. In vivo imaging revealed age-dependent reductions in brain region volume and in neurovascular coupling, alongside deficits in acquiring touchscreen-based cognitive tasks.

    The comprehensive characterization of LOAD2+HFD mice underscores the model's relevance for preclinical studies targeting LOAD features, independent of amyloid and tau pathologies. The strategic combination of genetic risk factors and dietary stressors provides insights into age-related neurodegeneration, cognitive deficits, and imaging abnormalities, establishing LOAD2+HFD as a valuable resource for investigating AD. The observed age-related neurodegeneration, cognitive deficits, and plasma NfL elevation position LOAD2+HFD as a promising tool for studying predementia phases and identifying therapeutic targets.

    However, some considerations emerge when using this model. The absence of tau pathology assessment is a notable limitation, given its significance in AD. While the study doesn't aim to replicate neurofibrillary tangles, evaluating tau phosphorylation could provide a more comprehensive understanding of the model's neuropathological landscape.

    The absence of hippocampal-dependent spatial working memory deficit as seen in AD patients raises concerns about the model's representation of primary AD symptoms. Nevertheless, the study's focus on mimicking presymptomatic phases is commendable, addressing a gap in available models.

    The drastic reduction in soluble and insoluble Aβ40, coupled with the absence of plaques, challenges the hypothesis that humanized Aβ would favor amyloid pathology, as initially hypothesized by the team. The unexpected findings underscore the complexity of AD pathophysiology and highlight the need for nuanced investigations.

    A noteworthy concern is the potential for the observed pathologies in the model to represent obesity-associated neurodegeneration pattern mimicking AD instead of AD pathologies themselves. Indeed, this model, under drastic HFD, induced sustained obesity. Excess weight has been related to brain atrophy and cognitive decline. Reports show that obesity is linked with AD-related changes, such as cerebrovascular damage or Aβ accumulation. Obesity-related gray matter atrophy resembles that of AD; however, those changes are often independent from AD.

    Moreover, the variability in markers such as Iba1 and GFAP underscores inter-individual variations. While this mirrors the diversity observed in human AD patients, this variability might create conceptual limitations, particularly in light of evolving European Union directives seeking to strongly reduce in vivo experimentation.

    In conclusion, the LOAD2+HFD mouse model offers a promising avenue for investigating LOAD features beyond traditional paradigms. The model's contribution to understanding predementia phases and potential therapeutic targets positions it as a valuable asset in the complex landscape of AD research.

    View all comments by Maud Gratuze

  2. Until a century ago, most people used to die before the age of 50 with few exceptions, suggesting our creator did not expect us to live long enough to suffer from late-onset Alzheimer’s disease (LOAD). Kotredes and colleagues have tried to generate a mouse model of LOAD by combining genetic risk factors, including Apoe4 and TREM2R47H, together with humanization of the mouse Aβ sequence. I think this is a logical approach because Apoe4 and TREM2R47H are the most influential risk factors for LOAD. I expect that humans carrying these genotypes, particularly if homozygous, would be affected by LOAD upon aging with a probability of 100 percent, so such diseased conditions would be equivalent to late-onset familial AD (LOFAD). If mice lived as long as humans, the mutant mice would develop LOAD after some decades.

    The question then is, what is the difference between FAD and LOFAD? Neuropathologically, FAD and LOFAD resemble each other in Aβ amyloidosis, neuroinflammation, tau pathology, and neurodegeneration. If these pathological features are the targets for prevention and treatment of AD in both the cases, medications effective for prevention and treatment for FAD should also be effective for LOFAD.  If so, preclinical studies would be well done using models of FAD. There is, however, great utility in the LOFAD models. Both the macroscopic and microscopic analyses (spatial transcriptomics and others) would contribute to establishment of cause-and-effect relationships between LOFAD risk factors and AD pathogenesis. I also expect these approaches to contribute to the understanding of the mechanisms underlying these relationships. Because the LOAD model alone does not exhibit the major AD pathologies, it would be interesting to crossbreed them with AppNL-F mouse model.

    View all comments by Takaomi Saido

Make a Comment

To make a comment you must login or register.

References

Research Models Citations

  1. hAbeta/APOE4/Trem2*R47H (LOAD2)
  2. Trem2 R47H KI x APOE4 (LOAD1)

Mutations Citations

  1. APOE C130R (ApoE4)
  2. TREM2 R47H

News Citations

  1. New Mouse Models Better Mimic Tauopathy, Alzheimer's
  2. Cornucopia: LOADs of New Mouse Models Available
  3. Plaques Kick Neocortical Neurons into Overdrive, Entangling Tau
  4. PET Firms Up Amyloid Cascade: Plaques, Inflammation, Tangles
  5. Proteomics Highlight Alzheimer’s Changes in Matrisome, MAPK Signaling

Paper Citations

  1. Kotredes KP, Oblak A, Pandey RS, Lin PB, Garceau D, Williams H, Uyar A, O'Rourke R, O'Rourke S, Ingraham C, Bednarycek D, Belanger M, Cope Z, Foley KE, Logsdon BA, Mangravite LM, Sukoff Rizzo SJ, Territo PR, Carter GW, Sasner M, Lamb BT, Howell GR. Uncovering Disease Mechanisms in a Novel Mouse Model Expressing Humanized APOEε4 and Trem2*R47H. Front Aging Neurosci. 2021;13:735524. Epub 2021 Oct 11 PubMed. Correction.
  2. Horner AE, Heath CJ, Hvoslef-Eide M, Kent BA, Kim CH, Nilsson SR, Alsiö J, Oomen CA, Holmes A, Saksida LM, Bussey TJ. The touchscreen operant platform for testing learning and memory in rats and mice. Nat Protoc. 2013 Oct;8(10):1961-84. PubMed.

External Citations

  1. CANTAB

Further Reading

No Available Further Reading

Primary Papers

  1. Kotredes KP, Pandey RS, Persohn S, Elderidge K, Burton CP, Miner EW, Haynes KA, Santos DF, Williams SP, Heaton N, Ingraham CM, Lloyd C, Garceau D, O'Rourke R, Herrick S, Rangel-Barajas C, Maharjan S, Wang N, Sasner M, Lamb BT, Territo PR, Sukoff Rizzo SJ, Carter GW, Howell GR, Oblak AL. Characterizing Molecular and Synaptic Signatures in mouse models of Late-Onset Alzheimer's Disease Independent of Amyloid and Tau Pathology. bioRxiv. 2023 Dec 20; PubMed.

Annotate

To make an annotation you must Login or Register.