There is strong interest in understanding how neurodegeneration is affected by a cellular state called senescence, in which cells stop dividing, suppress intrinsic cell-death pathways and release pro-inflammatory molecules that can harm healthy neighbours1,2. In a paper in Nature, Bussian et al.3 examine the role of senescent cells in a mouse model of a type of neurodegeneration that involves aggregation of the protein tau. They find that neuronal expression of mutant tau triggers senescence in glia, the support cells of the brain. Preventing the build-up of senescent glia can block the cognitive decline and neurodegeneration normally experienced by these mice.
Senescent cells are characterized by various molecular and gene-expression changes, including elevated levels of the cell-cycle inhibitor protein p16INK4A. Senescence can be identified by a test that stains cells blue if they harbour senescence-associated β-galactosidase (SA-β-Gal) — a form of the β-Gal enzyme that is active at pH 6 (in healthy cells, β-Gal is inactive at this pH)1,4. The cells also secrete inflammatory signalling molecules, growth factors and protease enzymes that can impair the function, and ultimately the survival, of non-senescent cells in their vicinity1,4. This trait is known as the senescence-associated secretory phenotype (SASP).
The gradual build-up of senescent cells contributes to ageing in multicellular organisms1,2. Furthermore, senescence can be induced by various cellular insults. Senescent neurons or glia have been described in people with brain injury or neurodegenerative disorders such as Parkinson’s and Alzheimer’s diseases1,2,5,6. Strategies have been developed that selectively target and eliminate senescent cells, counteracting many of the effects of ageing and age-related disorders in animal models1,7,8. But despite intense study, the exact effect of senescent cells in different contexts — including in neurodegeneration — remains unclear.
Bussian et al. set out to examine the role of senescence in neurodegeneration. They focused on the aggregation-prone neuronal protein tau, which is associated with multiple forms of neurodegeneration. For instance, a mutation in tau that changes amino-acid residue 301 from proline to serine (dubbed tauP301S) causes frontotemporal dementia9. And, when phosphorylated at abnormally high levels, tau forms structures called neurofibrillary tangles (NFTs) that are a hallmark of Alzheimer’s disease9.
The authors made use of mice that have been engineered to express human tauP301S in neurons, and so model human tau-mediated neurodegenerative diseases. They found elevated levels of various senescence-associated genes, including p16INK4A, in the brains of tauP301S-expressing mice compared with control animals. Using electron microscopy, the researchers examined which types of brain cell stained for SA-β-Gal in tauP301S mice. They observed no staining in neurons, but SA-β-Gal was detected in the two main types of glia — astrocytes and microglia. The group complemented their electron microscopy with an examination of senescence-associated gene expression in isolated brain-cell types. This, too, provided evidence of senescence in astrocytes and microglia, but not in neurons.
Importantly, Bussian and colleagues found that senescence-associated gene expression in tauP301S mice increased with age, but preceded NFT deposition and neurodegeneration. This suggests that the emergence of senescent cells could affect the latter two traits. To examine this possibility, the researchers eliminated senescent cells in the animals as they arose, by using a genetic tool that causes expression of a cell-death-promoting enzyme specifically in cells that produce p16INK4A. Removal of senescent cells prevented brain shrinkage and thinning of a cognition-related brain region called the dentate gyrus — two characteristics of tau-mediated neurodegeneration typically seen in tauP301S animals. Furthermore, cognitive function was maintained in tauP301S mice lacking senescent cells, whereas tauP301S animals in which senescent cells were retained exhibited short-term memory defects.
Perhaps more surprisingly, given that it indicates complex crosstalk between neurons and senescent glia, genetically eliminating senescent astrocytes and microglia reduced neuronal tau phosphorylation and NFT deposition. Moreover, the authors found similar effects when they treated tauP301S mice with a ‘senolytic’ compound, which triggers pharmacological removal of senescent cells. Together, Bussian and colleagues’ data clearly demonstrate that tauP301S expression in neurons can induce senescence in brain astrocytes and microglia. In turn, these senescent glia affect the ability of neurons to regulate tau phosphorylation and aggregation (Fig. 1). Whether by releasing signalling molecules that directly affect tau or through the effects of SASP factors (or both), glial senescence ultimately promotes neuronal degeneration.
Bussian and co-workers’ findings point to several avenues for future study. First, the signals from tauP301S-expressing neurons that induce senescence in glia should be defined. Similarly, the mechanisms by which senescent astrocytes and microglia signal back to neurons remain to be determined. It will also be interesting to understand whether the same glia-derived signals affect both tau pathology and neuronal survival, and whether astrocytes and microglia send the same or distinct signals. The answers to these questions are likely to have broader implications for understanding neurodegenerative diseases more generally.
Finally, the current study adds to the growing body of evidence indicating that senolytic treatments could benefit people who have a wide range of conditions1,2. Of immediate interest is whether removal of senescent cells can decrease disease severity in other animal models of neurodegeneration. The authors removed senescent cells throughout the lives of their animals, but it will also be valuable to determine whether senolytics can have beneficial effects if treatment is started once a disease has progressed to symptomatic stages — a more likely scenario in humans. Finally, it will be crucial to determine whether the processes uncovered in this paper are evolutionarily conserved in humans. If so, perhaps senolytic treatments can benefit people, as promised by this and other mouse studies1,2.
Nature 562, 503-504 (2018)