About our research interests
Understanding heterogeneity of clinical and pathological correlations in the Alzheimer’s disease (AD) continuum. AD is pathologically defined by the presence of sufficient amyloid plaques and neurofibrillary tangles in the brain but there is substantial heterogeneity in clinico-pathological correlations due to variability in comorbid pathologies, genetic influence, education attainments, etc. In collaboration with Profs. Rebecca Betensky and Jing Qian, our early analyses of the NACC autopsy cohort have tried to addressed this heterogeneity and revealed that (1) neuritic amyloid plaques, neurofibrillary tangles, moderate and severe cerebral amyloid angiopathy, severe ischemic small vessel disease, and hippocampal sclerosis all independently correlate with the degree of antemortem cognitive impairment; (2) a subset of patients (approx. 15%) diagnosed with probable mild-to-moderate AD, who would meet criteria for enrollment in anti-amyloid-β clinical trials, actually have none or only sparse neuritic amyloid plaques; and (3) Thal amyloid phases do not independently contribute to antemortem cognition. These findings have furthered our understanding of the heterogeneity of pathological substrates underlying age-related cognitive impairment.
Deciphering the complexity and heterogeneity of glial responses in AD. APOEε4 remains the strongest genetic risk factor for AD and APOEε2 the strongest genetic protective factor after the several dozen risk loci identified by GWAS in the last decade. Compared to the APOEε3 allele, APOEε4 and APOEε2 clearly influence the odds of developing AD and its age of symptom onset, but whether they also impact the rate of clinical progression after symptom onset was controversial. Moreover, these APOE isoforms were thought to operate through Abεta aggregation and clearance, but not through tau neurofibrillary tangles. In collaboration with Profs. Rebecca Betensky and Jing Qian, we have investigated the associations of APOE genotype with both postmortem AD neuropathological changes and cognitive trajectories during life. We observed that (1) the allele APOEε4 is associated with more neuritic plaques and cerebral amyloid angiopathy, whereas the APOEε2 allele is associated with fewer tangles and plaques; (2) neither APOE allele is independently associated with antemortem cognitive performance but each impacts antemortem cognition (ε2 is protective and ε4 detrimental) through its effects on AD pathology; (3) APOEε4 accelerates cognitive decline and APOEε2 slows it down in subjects who end up having moderate or high AD neuropathological changes at autopsy. In collaboration with Dr. Sudeshna Das, we have investigated the differences in expression of microglia- and astrocyte-predominant genes across APOE genotypes in public datasets and observed a pro-inflammatory and phagocytic bias of microglia in APOEε4 carriers, which is present before neuritic plaques develop and remains independent of the severity of AD neuropathological changes. These observations have contributed to broaden the focus of APOE research from an Aβ-centric view to other potential mechanisms such as tau-induced neurodegeneration and glial responses.
Reactive astrocytes and microglia decorate neuritic amyloid plaques in the AD brain but whether they have neurotoxic or neuroprotective effects remains controversial. Our early stereology-based quantitative neuropathological studies provided insights about the interactions of reactive glia and both plaques and tangles over the course of the disease. Reactive astrocytes and microglia accrue linearly over the course of AD, both paralleling the extent of tangles and diverging from plaque deposition, which remains relatively stable throughout the clinical phase of the disease. This increase in reactive glia occurs in the proximity of both plaques and tangles. However, only the number of reactive (GFAP+) astrocytes and microglia (MHC2+), but not their total numbers – reactive plus homeostatic cells – significantly differs between AD and age-matched non-demented individuals, indicating that a phenotypic change but not proliferation underlies glial responses in AD.
In recent years, it is becoming clear that astrocyte and microglial responses are complex and that a combination of markers and transcriptomics are required to obtain a complete picture of their functional changes. We have characterized the postmortem brain expression of 18 kDa translocator protein (TSPO), which has been used for almost 20 years for “activated microglia PET imaging,” and found that it is surprisingly similar in postmortem AD and age-matched control brains and that it is not restricted to microglia, but also expressed by astrocytes, endothelial cells, and vascular smooth muscle cells. We have also demonstrated the context-dependent nature of astrocyte reaction, with notable differences between mouse models of acute CNS injury and neurodegenerative disease, by meta-analyzing published astrocyte-specific transcriptomic datasets from these models. Lastly, neuroinflammation emerged as one of the main pan-neurodegenerative pathways in a meta-analysis of 60 AD, LBD and ALS-FTD microarray datasets comprising 2,600 samples, further suggesting common glial responses across diseases. All these studies have broadened the scope of glia research in AD from Aβ plaques to tau neurofibrillary tangles, established that proliferation is not a feature of glial reactions in AD, and helped unveil the complexity and heterogeneity of glial reactions in AD and other CNS diseases.