Brain transcriptome analysis of a familial Alzheimer’s disease-like mutation in the zebrafish presenilin 1 gene implies effects on energy production

Abstract

To prevent or ameliorate Alzheimer’s disease (AD) we must understand its molecular basis. AD develops over decades but detailed molecular analysis of AD brains is limited to postmortem tissue where the stresses initiating the disease may be obscured by compensatory responses and neurodegenerative processes. Rare, dominant mutations in a small number of genes, but particularly the gene PRESENILIN 1 (PSEN1), drive early onset of familial AD (EOfAD). Numerous transgenic models of AD have been constructed in mouse and other organisms, but transcriptomic analysis of these models has raised serious doubts regarding their representation of the disease state. Since we lack clarity regarding the molecular mechanism(s) underlying AD, we posit that the most valid approach is to model the human EOfAD genetic state as closely as possible. Therefore, we sought to analyse brains from zebrafish heterozygous for a single, EOfAD-like mutation in their PSEN1-orthologous gene, psen1. We previously introduced an EOfAD-like mutation (Q96_K97del) into the endogenous psen1 gene of zebrafish. Here, we analysed transcriptomes of young adult (6-month-old) entire brains from a family of heterozygous mutant and wild type sibling fish. Gene ontology (GO) analysis implies effects on mitochondria, particularly ATP synthesis, and on ATP-dependent processes including vacuolar acidification.

Background

AD is the most common form of dementia with severe personal, social, and economic impacts. Rare, familial forms of AD exist caused by autosomal dominant mutations in single genes (reviewed by [1]). The majority of these mutations occur in the gene PRESENILIN 1 (PSEN1) that encodes a multipass integral membrane protein involved in intra-membrane cleavage of numerous proteins [1].

A wide variety of transgenic models of AD have been created and studied. These are aimed at reproducing histopathologies posited to be central to the disease process, i.e. amyloid plaques and neurofibrillary tangles of the protein MAPT [2]. However, analysis of the effects on the brain transcriptome of the transgenes driving a number of these mouse models showed little concordance with transcriptomic differences between human AD brains and age-matched controls [3] (although a recent study asserts that this lack of concordance for the popular “5XFAD” transgenic mouse model is due to previous failure to analyse the effects of its transgenes in a variety of genetic backgrounds [4]). We posit that, in the absence of an understanding of the molecular mechanism(s) underlying AD, the most objective approach to modeling this disease (or, at least, modeling its genetic form, EOfAD) is to create a genetic state as similar as possible to the EOfAD state in humans. Mouse “knock-in” models of EOfAD mutations were created over a decade ago and showed subtle phenotypic effects but not the desired histopathologies (e.g. [5, 6]). However, at that time, researchers did not have access to RNA-Seq technology. To the best of our knowledge, transcriptome analysis of the EOfAD mutation knock-in mouse models was never performed.

In humans, AD is thought to develop over decades and the median survival to onset age for EOfAD mutations in human PSEN1 considered collectively is 45 years [7]. Functional MRI of human children carrying EOfAD mutations in PSEN1 has revealed differences in brain activity compared to non-carriers in individuals as young as 9 years of age [8]. Presumably therefore, heterozygosity for EOfAD mutations in PSEN1 causes early molecular changes/stresses that eventually lead to AD.

Transcriptome analysis is currently the most detailed molecular phenotypic analysis possible on cells or tissues. Here we present an initial analysis of the transcriptomic differences caused in young adult (6-month-old) zebrafish brains by the presence of an EOfAD-like mutation in the gene psen1 that is orthologous to the human PSEN1 gene. GO analysis supports very significant effects on mitochondrial function, especially synthesis of ATP, and on ATP-dependent functions such as the acidification of lysosomes that are critical for autophagy.

Materials and methods

The mutant allele, Q96_K97del, of psen1 was a byproduct identified during our introduction of the K97fs mutation into psen1 (that models the K115fs mutation of human PSEN2 – see [9] for an explanation).

Q96_K97del is a deletion of 6 nucleotides from the coding sequence of the psen1 gene. This is predicted to distort the first lumenal loop of the Psen1 protein. In this sense, it is similar to a number of EOfAD mutations of human PSEN1 [10]. Also, in common with all the widely distributed EOfAD mutations in PSEN1, (and consistent with the PRESENILIN EOfAD mutation “reading frame preservation rule” [1]), the Q96_K97del allele is predicted to encode a transcript that includes the C-terminal sequences of the wild type protein. Therefore, as a model of an EOfAD mutation, it is superior to the K97fs mutation in psen1 [9].

To generate a family of heterozygous Q96_K97del allele (i.e. psen1^Q96_K97del/+) and wild type (+/+) sibling fish, we mated a psen1^Q96_K97del/+ individual with a +/+ individual and raised the progeny from a single spawning event together in one tank. Zebrafish can live for up to 5 years but, in our laboratory, typically show greatly reduced fertility after 18 months. The fish become fertile after around 3 months of age, so we regard 6-month-old fish as equivalent to young adult humans. Therefore we analysed the transcriptomes of entire young adult, 6-month-old fish brains using poly-A enriched RNA-seq technology, and estimated gene expression from the resulting single-end 75 bp reads using the reference GRCz11 zebrafish assembly transcriptome [11, 12]. Each zebrafish brain has a mass of approximately 7 mg. Since AD is more prevalent in human females than males, and to further reduce gene expression “noise” in our analyses, we obtained brain transcriptome data from four female wild type fish and four female heterozygous mutant fish. This data has been made publicly available at the Gene Expression Omnibus (GEO, see under Availability of data and materials below).

Results

Differentially expressed genes (DE genes)

Genes differentially expressed between wild type and heterozygous mutant sibling fish were identified using moderated t-tests and a false discovery rate (FDR)-adjusted p-value cutoff of 0.05 as previously described [9, 13, 14]. In total, 251 genes were identified as differentially expressed (see Additional file 1). Of these, 105 genes showed increased expression in heterozygous mutant brains relative to wild type sibling brains while 146 genes showed decreased expression.

GO analysis

To understand the significance for brain cellular function of the differential gene expression identified in young adult heterozygous mutant brains we used the goana function [15] of the limma package of Bioconductor software [14] to identify GOs in which the DE genes were enriched at an FDR-corrected p-value of less than 0.05. Seventy-eight GOs were identified (Table 1) of which 20 addressed cellular components (CC). Remarkably, most of these CCs concerned the mitochondrion, membranes, or ATPases. Seventeen GOs addressed molecular functions (MF) and largely involved membrane transporter activity, particularly ion transport and ATPase activity coupled to such transport. Forty-one GOs addressed biological processes (BP) and involved ATP metabolism, ribonucleoside metabolism, and transmembrane transport processes including vacuolar acidification (that has previously been identified as affected by EOfAD mutations in PSEN1 [16]). Overall, our GO analysis indicates that this EOfAD-like mutation of zebrafish psen1 has very significant impacts on cellular energy metabolism and transmembrane transport processes.

Table 1 GOs enriched for genes differentially expressed between heterozygous mutant and wild type sibling fish brains