Differential transcript usage unravels gene expression alterations in Alzheimer’s disease human brains

Abstract

Alzheimer’s disease (AD) is the leading cause of dementia in aging individuals. Yet, the pathophysiological processes involved in AD onset and progression are still poorly understood. Among numerous strategies, a comprehensive overview of gene expression alterations in the diseased brain could contribute for a better understanding of the AD pathology. In this work, we probed the differential expression of genes in different brain regions of healthy and AD adult subjects using data from three large transcriptomic studies: Mayo Clinic, Mount Sinai Brain Bank (MSBB), and ROSMAP. Using a combination of differential expression of gene and isoform switch analyses, we provide a detailed landscape of gene expression alterations in the temporal and frontal lobes, harboring brain areas affected at early and late stages of the AD pathology, respectively. Next, we took advantage of an indirect approach to assign the complex gene expression changes revealed in bulk RNAseq to individual cell types/subtypes of the adult brain. This strategy allowed us to identify previously overlooked gene expression changes in the brain of AD patients. Among these alterations, we show isoform switches in the AD causal gene amyloid-beta precursor protein (APP) and the risk gene bridging integrator 1 (BIN1), which could have important functional consequences in neuronal cells. Altogether, our work proposes a novel integrative strategy to analyze RNAseq data in AD and other neurodegenerative diseases based on both gene/transcript expression and regional/cell-type specificities.

The Alzheimer susceptibility gene BIN1 induces isoform-dependent neurotoxicity through early endosome defects

“…This suggests that an increase in BIN1iso1 may be deleterious and contribute to early phases of the disease through early endosome enlargement. It implies an increase in BIN1iso1 levels in the disease, which remains highly debated with isoform-independent or isoform-related increased or decreased expression reported (Adams et al, 2016; Chapuis et al, 2013; Glennon et al, 2013; Holler et al, 2014; De Rossi et al, 2016; Taga et al, 2019, Marques-Coelho et al 2021). Moreover, most of the expression studies do not take into account disease progression and it is almost impossible to determine whether these expression variations may be a cause or a consequence of the disease development as previously mentioned.…”

“…However, the deregulation of BIN1 expression in the brain of AD cases is still highly debated. Some results indicate that overall BIN1 expression is increased (Chapuis et al, 2013), or decreased (Glennon et al, 2013; Marques-Coelho et al, 2021), whereas more complex patterns have been also reported with a decrease in BIN1iso1 and a concomitant increase in BIN1iso9 expression (Holler et al, 2014). In addition, according to the pattern of expression, it is not clear if the observed variations of BIN1 expression are a cause or a consequence of the neurodegenerative process.…”

“…In the brain, a complex expression pattern is also observed at the cellular level. BIN1 expression is mainly observed in oligodendrocytes, microglial cells and neurons (Adams et al, 2016; Marques-Coelho et al, 2021; De Rossi et al, 2016). However, while neurons express high molecular weight isoforms including BIN1iso1, glial cells express lower molecular weight isoforms such as BIN1iso9 (De Rossi et al, 2016; Zhou et al, 2014).…”

“…Coding mutations in BIN1 causes autosomal recessive centronuclear myopathy (Nicot et al, 2007) whereas AD-associated BIN1 variants are non-coding and likely regulate BIN1 expression. In the brain, BIN1 is mainly expressed in oligodendrocytes, microglial cells and neurons (Adams et al, 2016;Marques-Coelho et al, 2021;De Rossi et al, 2016). BIN1 has more than 10 isoforms with BIN1 isoform1 (BIN1iso1) and isoform8 (BIN1iso8) respectively expressed in the brain and skeletal muscles whereas isoform9 (BIN1iso9) is ubiquitously expressed (GTEx portal, www.gtexportal.org).…”

Predicting gene regulatory networks from multi-omics to link genetic risk variants and neuroimmunology to Alzheimer’s disease phenotypes

“…These datasets thus enable finding genes and transcriptional activities that associate with AD phenotypes from the populations, providing molecular mechanistic insights into AD. For example, differential expression analyses for the temporal lobes and frontal lobes found many key Differentially Expressed genes (DEGs) in AD, including ABCA1 and 2, C1R and C1S, VGF, REST, GAD1 and 2, SST, and CALB1 10 . Further, gene co-expression network analysis such as WGCNA has been widely applied to these population data to identify various gene co-expression modules 11 .…”