Investigating astrocyte dysfunction in mouse models of Alzheimer’s disease
Every 3 seconds someone in the world is diagnosed with dementia. This staggering fact is compounded by another, 60-80% of these patients are suffering from the same disease, Alzheimer’s disease (AD). Despite this, we currently have no disease modifying treatments. AD is characterized by an accumulation of extracellular amyloid-beta (Aβ) plaques, intracellular tau neurofibrillary tangles and neuronal loss. Additional pathological features include synapse loss and an increase in reactive astrocytes and microglia, particularly around Aβ plaques. Aβ accumulation is thought to occur early in the disease process and trigger downstream pathology. For this reason, many therapeutics in clinical trials have targeted Aβ accumulation. However, thus far there has been little success, highlighting that pharmaceutical companies need to explore new therapeutic strategies. Genome wide association studies have illustrated the importance of the innate immune system in AD. Astrocytes and microglia are the main cells in the brain involved in innate immunity. In this study, I investigate the impact of Aβ on astrocytes, in the hope that a better understanding of how these cells are affected may lead to novel therapeutic targets. We have used two mouse models of familial AD which develop amyloid plaque pathology similar to that seen in AD; the well characterized transgenic APPswe/PSEN1dE9 (APP/PS1) mouse model and the less well characterized knock-in APPNLF mouse model. Firstly, we used array tomography and immunohistochemistry to quantify the synapse loss and astrogliosis around plaques in 12-month APPNLF mice. This demonstrated that synapse loss and astrogliosis in the APPNLF mouse is comparable to other mouse models of AD, but is generally confined to within 10μm of the plaque core. Next, we used translating ribosome affinity purification (TRAP) followed by bulk ribonucleic acid-sequencing (RNA-seq) to assess how astrocytic gene expression changed with increasing Aβ pathology in 6, 12 and 18-month APP/PS1 and APPNLF neocortical astrocytes. Amyloidopathy in APP/PS1 astrocytes exacerbated age dependent gene changes in astrocytes, as well as inducing a profile which resembled acutely induced reactive astrocytes. Thus, highlighting an overlap of acute and chronic reactive astrocyte signatures. APP/PS1 astrocytes also showed overlapping up and down regulation of genes with those changed in astrocytes in the MAPTP301S tauopathy model, suggesting that the two core proteinopathies in AD induce elements of similar changes in astrocyte phenotype. Pathways related to protein degradation and inflammation appeared to be upregulated due to the AD proteinopathies, whilst pathways related to mitochondrial function and protein synthesis were downregulated. Notably, genes induced in APP/PS1 astrocytes were enriched for genes induced in human post-mortem AD astrocytes, indicating that the results obtained in this mouse model are somewhat translatable to the human condition. The APPNLF astrocyte bulk RNA-seq showed few differentially expressed genes at the ages tested. However, genes that were upregulated in the APP/PS1 astrocytes demonstrated a positive fold change in the APPNLF astrocytes. This might indicate overlap of astrocytic response to amyloidopathy in both models, but that astrocyte pathology is slower to develop in the APPNLF mouse model, compared to the APP/PS1 mouse model, mirroring the slower development of plaque pathology. It is also possible that some APPNLF astrocytes displayed phenotypes similar to APP/PS1 astrocytes, but the gene expression was obscured when conducting bulk RNA-seq. To investigate the heterogeneity of astrocytic response to Aβ pathology, we conducted single cell RNA-seq on astrocytes from the neocortex of APP/PS1 and APPNLF mice. This revealed an astrocyte phenotypic state present in both mouse models of amyloidopathy, which was largely absent in 12-month WT mice, but which began to become apparent in 18-month WT astrocytes. Genes involved in oxidative phosphorylation, protein synthesis, the unfolded protein response, and nuclear factor erythroid 2-related factor 2 (NRF2)-mediated oxidative stress response were expressed higher in this pathology associated cluster of astrocytes, compared to astrocytes in other clusters. However, genes involved in synaptogenesis were lowly expressed in the pathology associated cluster of astrocytes compared to other astrocytes, potentially linking Aβ induced astrocyte dysfunction with the reduction in synapses seen in AD. This analysis highlights the benefits of single cell gene expression analysis in identifying heterogeneity of astrocyte response to amyloidopathy, which was lost in the bulk translatome analysis of astrocytes from mouse models of AD. These investigations add to our growing understanding of mouse models of AD and astrocyte dysfunction in AD. Amyloidopathy appears to induce both neuroprotective and neurotoxic phenotypic alterations in astrocytes. Designing therapeutic strategies which enhance neuroprotective functions of astrocytes, such as NRF2-mediated antioxidant signaling, whilst reducing neurotoxic overactivation of inflammatory responses and the unfolded protein response, may improve AD pathogenesis. Further research defining phenotypic states of astrocytes in AD will enable the design of astrocyte-state targeted therapeutics, which will likely demonstrate superior results to broader astrocyte targeted therapeutics.