Confounding Variables in Transgenic Mouse Models of Taupathy

Abstract

Tau is the main axonal microtubule-associated protein in the mammalian central nervous system (CNS) and functions primarily in microtubule assembly and axonal transport. However, tau also has diverse other functions ranging from postsynaptic receptor scaffolding to chromatin protection and organization. Tauopathies are a family of over 20 neurodegenerative diseases characterized by tau pathology in the brain, which is an indication of tau dysfunction and dysregulation. Importantly, tauopathies include Alzheimer’s disease and frontotemporal dementia, which are among the leading causes of early-onset dementia and are growing in prevalence due to an increasingly older population. To date, no disease-modifying therapies exist for tauopathies in part because the complex and multifunctional nature of tau biology makes it difficult to understand its role in disease pathogenesis. Mouse modeling is an invaluable research tool frequently employed to test hypotheses about the role of tau in tauopathies, but has unfortunately yielded inconsistent results and extreme phenotypic variability. ‘Conventional’ techniques have historically been used to generate such mouse models, involving random insertion of synthetic mini-gene constructs into the mouse genome. In the current work, we have employed new ‘targeted’ genome engineering techniques to precisely insert a human tau (hTau) transgene into a predetermined, non-disruptive locus. We hypothesized that a targeted model harboring the pathogenic P301L hTau mutation would develop a disease-like phenotype and thus could be used as a positive control in comparison to genetically matched mice with different hTau modifications. Unlike the widely used conventionally-made tauopathy model rTg4510, which harbors a P301L hTau transgene, our targeted equivalent (rT2/T2) did not quickly develop a robust tauopathy-like phenotype. We found this discrepancy could be explained by a large transgene insertion-deletion (TgINDEL) mutation that disrupts and dysregulates Fgf14, a gene important for neuronal function. A tet-transactivator (tTA) transgene, required to activate hTau expression, also caused a TgINDEL that disrupts another five genes in this mouse line, four of which have prominent forebrain expression. Comparing this model to rT2/T2 mice revealed that the two TgINDELs allow for acceleration of the rTg4510 phenotype, including neurodegeneration and tau histopathology. In addition, we found that extreme overexpression of hTau, or ‘hyperexpression,’ is necessary to drive the severe phenotype. We also generated a targeted transgenic mouse overexpressing non-mutant (NM) hTau, with the intention of using it as a genetically-matched negative control in comparison to the P301L line. Surprisingly, NM hTau was associated with a developmental phenotype, characterized by reduced clearance and hyperphosphorylation of hTau. We found that disrupted mitochondrial dynamics and elevated oxidative stress were involved in NM hTau pathogenicity, underlying abnormal brain development and cognitive deficits. Interestingly, these phenotypes were lost with the P301L mutation, suggesting it conferred a beneficial loss of function during development. We speculate the molecular and behavioral differences between the NM and P301L hTau lines are driven by hTau-microtubule binding, which was found to be reduced in the P301L line. Together, this work identifies and characterizes confounding variables in mouse models of tauopathy including genomic disruption, transgene hyperexpression, and overexpression of transgenes during early postnatal development. We take initial steps to improve mouse transgenesis approaches and find genetic matching to be advantageous for probing molecular phenotypes of different forms of tau. We assert that it is important to eliminate confounding variables and use proper controls where possible to ensure phenotypes are relevant to the disease of interest. These findings are an example of a broader problem affecting neurodegenerative disease research, which can be avoided in future studies with more rigorous genomic characterization of mouse models as well as careful consideration of the timing and level of transgene expression as appropriate for the hypothesis in question.