NEW YORK (GenomeWeb) – Researchers aiming to improve preclinical research into late-onset Alzheimer's disease have released several new mouse models that harbor genetic abnormalities known to be associated with the disease.
The Model Organism Development and Evaluation for Late-onset AD (MODEL-AD) project launched in 2016 with a five-year, $25 million grant from the National Institute on Aging (NIA). The program, which involves Indiana University School of Medicine, The Jackson Laboratory, Sage Bionetworks, and the University of California, Irvine, aims to advance more accurate preclinical animal models for late-onset AD research that can be used to identify biomarkers and advance drugs for the disease.
To date, the available preclinical mouse models have all harbored mutations in APP, PSEN1, or PSEN2 genes, known to cause early-onset AD. However, only 2 percent of AD patients have familial mutations for early-onset AD, where symptoms like dementia can start when people are in their 30s and 40s.
Around 98 percent of the more than 5.7 million Alzheimer's patients in the US have the late-onset onset form of the disease, which is diagnosed after age 65. While APOE4 carrier status and TREM2 variants are associated with increased risk for late-onset AD, there are no known causative genetic markers. "We don't know what causes late-onset Alzheimer's, so we've never been able to model it," said Michael Sasner, a research scientist at The Jackson Laboratory and one of the leaders of the MODEL-AD program.
"We're testing drugs on mouse models that reflect only 2 percent of the patients," Sasner said. "Not surprisingly, that hasn't worked very well. There is a huge burden from Alzheimer's, societally and economically, and we don't have any good treatments yet."
Drugmakers have been thwarted again and again in their efforts to develop an Alzheimer's drug that meaningfully slows cognitive decline. Citing the expense of these failures, Pfizer notably announced earlier this year that it would end its neuroscience research efforts. And just last week, Biogen unveiled data on an experimental drug that appeared to stave of cognitive decline in certain patients, but the company's shares tanked after healthcare providers and investors questioned the statistical analysis and design of the study and wanted to see data from later-stage studies.
In the absence of effective treatments, the number of Alzheimer's patients could balloon to 13.8 million by 2050. A new report published by the Alzheimer's Association this year estimated that the current cost of caring for patients with AD and other dementias in the US is $277 billion and could increase to $1.1 trillion by 2050.
The MODEL-AD program is trying to improve researchers' chances of advancing an effective drug by creating more accurate preclinical models. At a conference last week, researchers from this effort described their efforts to create mouse models that more accurately represent that pathophysiology of late-onset AD.
The project is divided into different groups. The bioinformatics and data management core prioritizes the variants of interest, which the Disease Modeling Project uses to create the new mouse models. The preclinical testing core then uses the models with Alzheimer's-like phenotypes to evaluate the potential efficacy and safety of a drug based on a variety of parameters, such as exposure levels of the drug in tissues and if the drug is hitting the intended target.
The researchers have so far created homozygous models that expressed humanized APOE4 and the R47H allele of Trem2. Then, to increase the chance that the mice will display Alzheimer's-like phenotypes, Sasner and his colleagues engineered in other commonly occurring variants associated with late-onset AD, including changes in ABCA7, PLCG2, IL1RAP, and CEACAM1. There are also mice with APOE3 and APOE2 variants that will serve as controls, and the models include male and female mice to enable evaluation of sex differences.
Researchers can now create mouse models with multiple genetic risk variants and study their role in a complex disease like Alzheimer's thanks to advancements in gene editing, , the availability of large genomic databases, and newer computational methods.
"For some of these genes, [like APOE4] we have to put in a big chunk of the human gene," Sasner said. "But for some genes the mouse and human genes are so similar that we can change one base and have a change in the amino acid that can be disease-causing."
For example, in mice that already harbored APOE4 and Trem2*R47Hf, researchers used CRISPR to delete exon 32 in ABCA7, which has been associated with increases in Aβ peptides that comprise amyloid plaques in Alzheimer's patients' brains. They also engineered mice with a common ABCA7 variant since studies suggest that this variant doesn't occur with rare loss-of-function variants on the same haplotype. Researchers can now compare these mouse models and study the mechanism of these pathways.
The researchers also used CRISPR to create a mouse model where IL1RAP was knocked out. IL1RAP and APOE4 together explain more than 10 percent of the variance in amyloid accumulation in Alzheimer's patients, and the mouse model will allow researchers to study the role microglia play in amyloid accumulation during aging. There is also a model with a PLCG2 missense variant associated with Alzheiemer's risk, and a CEACAM1 knock-out model.
Sasner and his colleagues have six-month data on these models, but will be aging them for more than a year. During that time, researchers will subject these models to a battery of histological, transcriptomics, metabolomics, and biomarker evaluations, to ensure that the most clinically relevant models end up in preclinical testing.
The mouse models undergo an initial phenotyping screen, where only those models with similar transcriptomic profiles to late-onset AD patients move on to the next, deep phenotyping phase where they are validated against clinical measures. Only the clinically validated models are then to be used for preclinical testing of drugs.
In this way, the researchers hope to establish best practices for preclinical drug testing on animal models for late-onset AD. "This is a very different paradigm than just seeing if a mouse can remember, which may or may not be relevant to human cognition," Sasner said.
The MODEL-AD project is releasing these models before they've aged to display an Alzheimer's-like phenotype, hoping that researchers will be able to use the genetically characterized mice to conduct basic biology research and gain insights into certain biomarkers and drug targets.
By the fall or next spring, the group hopes to have data from models that are a year old, which can be deeply phenotyped. Sasner's group plans to create more than 40 new models for late-onset AD, of which 24 will undergo initial phenotyping screens, and at least eight will graduate to deep phenotype.
The standards, animal models, and data coming out of this effort will be broadly available to the research community across academia and industry. "A lot of these models have been restricted to academics only," Sasner said. "Our mandate is to make these [models] available to everyone, including pharmas and biotech."
MODEL-AD's advisory board also includes experts from academia and industry. Pharmas have been eager to get their hands on these models, Sasner said, but they've also been providing feedback to the project on the additional types of models they need. "[Pharmas] are really excited we're not just doing cognition assays," he said. "They want to see quantifiable biomarkers and omics' changes."
MODEL-AD represents the kind of collaborative, "big science" project that the National Institute on Aging is trying to encourage, where the output is widely shared. "We need big data and big science to cure something like Alzheimer's disease," Sasner said.