Fig. 2: Workflow depicting multi-disciplinary method to identify genes involved in hippocampal vulnerability in AD. | Nature Communications

Fig. 2: Workflow depicting multi-disciplinary method to identify genes involved in hippocampal vulnerability in AD.

From: Transcriptomic analysis to identify genes associated with selective hippocampal vulnerability in Alzheimer’s disease

Fig. 2

Controls and typical AD cases were grouped into the representative phenotype (green-black), while hippocampal sparing AD and limbic predominant AD were grouped into the extreme phenotype (red-blue). Frozen human hippocampal tissue was first dissected, followed by RNA extraction, purification, and lastly prepped for RNA-Seq. A multi-step approach was implemented for gene prioritization, which began with genes nominated from the AD literature (Step 1). Next, we examined differentially expressed genes derived from the representative phenotype and extreme phenotype (Step 2a) before further bioinformatic prioritization (Step 2b). The differentially expressed genes identified in Step 2a were further prioritized based upon process network interactions (Step 3). We then combined the 339 unique genes from Steps 1–3 to apply a translational neuropathology approach (Step 4) whereby protein-coding genes needed to exhibit monotonic directionality and associate with local neuropathologic markers (tau and amyloid-β) or global measures of AD pathology (Braak stage and Thal phase). Once genes were prioritized, biological significance was investigated first by expanding the RNA-Seq cohort to a total of n = 182 in NanoString analyses. Machine learning was applied to NanoString data using the random forest algorithm (Step 5). Random forest was applied to the phenotypic groups, in addition to controls versus all AD subtypes (depicted by green versus white bar) to identify the top five genes predictive of hippocampal vulnerability in AD as a whole. This enabled us to utilize objective classification of disease spectrum to uncover transcriptomic changes that underlie selective vulnerability of the hippocampus in AD. AD, Alzheimer’s disease; DE, differential expression; HpSp, Hippocampal sparing; Limbic, limbic predominant; RNA-Seq, RNA sequencing. Note: Numbers in boxes represent total gene number from each part of workflow. Steps 1–3 culminated in 339 unique genes, of which 19 overlapped in one or more of the steps. *In Step 4, one gene appeared in both groups.

Back to article page