PRESENILIN 1 Mutations Causing Early-Onset Familial Alzheimer’s Disease or Familial Acne Inversa Differ in Their Effects on Genes Facilitating Energy Metabolism and Signal Transduction

Abstract

Background:

The most common cause of early-onset familial Alzheimer’s disease (EOfAD) is mutations in PRESENILIN 1 (PSEN1) allowing production of mRNAs encoding full-length, but mutant, proteins. In contrast, a single known frameshift mutation in PSEN1 causes familial acne inversa (fAI) without EOfAD. The molecular consequences of heterozygosity for these mutation types, and how they cause completely different diseases, remains largely unexplored.

Objective:

To analyze brain transcriptomes of young adult zebrafish to identify similarities and differences in the effects of heterozygosity for psen1 mutations causing EOfAD or fAI.

Methods:

RNA sequencing was performed on mRNA isolated from the brains of a single family of 6-month-old zebrafish siblings either wild type or possessing a single, heterozygous EOfAD-like or fAI-like mutation in their endogenous psen1 gene.

Results:

Both mutations downregulate genes encoding ribosomal subunits, and upregulate genes involved in inflammation. Genes involved in energy metabolism appeared significantly affected only by the EOfAD-like mutation, while genes involved in Notch, Wnt and neurotrophin signaling pathways appeared significantly affected only by the fAI-like mutation. However, investigation of direct transcriptional targets of Notch signaling revealed possible increases in γ-secretase activity due to heterozygosity for either psen1 mutation. Transcriptional adaptation due to the fAI-like frameshift mutation was evident.

Conclusion:

We observed both similar and contrasting effects on brain transcriptomes of the heterozygous EOfAD-like and fAI-like mutations. The contrasting effects may illuminate how these mutation types cause distinct diseases.

Keywords

Alzheimer’s disease acne inversa gamma-secretase Presenilin 1 RNA-seq zebrafish

INTRODUCTION

Cases of Alzheimer’s disease (AD) can be classified by age of onset and mode of inheritance. Dominant mutations in a small number of genes cause AD with an age of onset younger than 65 years (early onset familial AD, EOfAD). On a population basis, around 60%of the mutations causing EOfAD occur in one gene, PRESENILIN 1 (PSEN1) [1 –4].

PSEN1 encodes a multi-pass transmembrane protein resident in the endoplasmic reticulum, plasma membrane, endolysosomal pathway, and other membranes [5, 6]. It has nine recognized transmembrane domains [7]. A tenth transmembrane domain may exist when PSEN1 protein is in its holoprotein state [7], before it undergoes autocatalytic endoproteolysis to form the active catalytic core of γ-secretase [8], an enzyme complex consisting of PSEN1 (or PSEN2), and the proteins NCSTN, PSENEN, and APH1A (or APH1B) [9, 10].

As a locus for genetic disease, PSEN1 is truly remarkable both for the number of mutations found there, and the variety of diseases these mutations cause. Mutations have been found associated with Pick’s disease [11], dilated cardiomyopathy [12], and acne inversa [13]. However, over 300 mutations of PSEN1 are known to cause EOfAD (http://www.alzforum.org/mutations/psen-1). In total, these mutations affect 161 codons of the gene. Remarkably, the mutations are widely distributed in the PSEN1 coding sequence but are particularly common in the transmembrane domains. Only three regions of the PSEN1 protein are mostly devoid of EOfAD mutations: upstream of the first transmembrane domain; a large part of the “cytosolic loop domain” (cytosolic loop 3); and the last two thirds of the 9^th transmembrane domain together with the lumenal C-terminus (see Fig. 1).

Fig. 1

Experimental design. A) Schematic of the human PSEN1 protein adapted from https://www.alzforum.org/mutations/psen-1 with permission from FBRI LLC (Copyright © 1996–2020 FBRI LLC. All Rights Reserved. Version 3.3 –2020). Amino acid residues are color-coded as to whether they are pathogenic for Alzheimer’s disease (red) or their pathogenicity is unclear (blue). The human mutation sites P242 (fAI) and T440 (EOfAD) are indicated by black arrows. The site of the zebrafish W233fs-equivalent codon to human W244 is shown by the red arrow. Note that the human T440 codon is equivalent to the zebrafish T428 codon. The residues equivalent to those deleted in the Q96_K97del mutation of zebrafish psen1 analyzed previously are indicated by a red box. B) A fish heterozygous for the W233fs mutation (fAI-like/+) was mated with a fish heterozygous for the T428del mutation (EOfAD-like/+). The resulting family of fish contain genotypes fAI-like/+, EOfAD-like/+, EOfAD-like/fAI-like, and their wild type siblings. The pairwise comparisons performed in the RNA-seq experiment are depicted. Since the EOfAD-like/fAI-like genotype is not representative of any human disease, it was not analyzed. See online version for color.

The most common outcome of mutation of a protein sequence is either no effect or a detrimental effect on the protein’s evolved activity. Only rarely are mutations selectively advantageous so that they enhance organismal survival and reproduction. The very large number of EOfAD-causative mutations in PSEN1 and their wide distribution in the protein coding sequence is consistent with a loss-of-function. However, this cannot be a simple loss of γ-secretase activity, as EOfAD-causative mutations have never been found in the genes encoding the other components of γ-secretase complexes (other than less frequent mutations in the PSEN1 homologous gene, PSEN2, reviewed in [14]). Also, an in vitro analysis of 138 EOfAD mutations of PSEN1 published in 2017 by Sun et al. [15] found that approximately 10%of these mutations actually increased γ-secretase activity.

Currently, the most commonly discussed hypothetical mechanism addressing how EOfAD mutations of PSEN1 cause disease is that these act through “qualitative changes” to γ-secretase cleavage of the amyloid β A4 precursor protein (AβPP) to alter the length distribution of the amyloid-β (Aβ) peptides derived from it [16]. However, the comprehensive study of Sun et al. revealed no consistency in the effects of PSEN1 EOfAD mutations on Aβ length distribution. The single consistent characteristic of all EOfAD mutations in both PSEN1 and PSEN2 is that these permit production of transcripts with coding sequences containing in-frame mutations but terminated by the wild type stop codons (i.e., they still permit production of a full length protein). This phenomenon was first noted by De Strooper in 2007 [17] and described in detail by us in 2016 (the “reading frame preservation rule” [18]). The universality of this rule, and that it reflects a critical feature of the EOfAD-causative mechanism of PSEN1 mutations, is shown by the fact that the P242LfsX11 frameshift mutation of PSEN1 (hereafter referred to as P242fs) causes a completely different inherited disease, familial acne inversa (fAI, also known as hidradenitis suppurativa), without EOfAD [13]. (Recently, a frameshift mutation in PSEN1, H21PfsX2, was identified in an early-onset AD patient. However, whether the mutation is causative of EOfAD mutation is still uncertain [19]. Questionable additional claims of EOfAD-causative frameshift mutations in PSEN genes have been made and are reviewed in [18].) Critically, fAI can also be caused by mutations in NCSTN and PSENEN [13], strongly supporting that this disease is due to changes in γ-secretase activity.

Understanding the role of PSENs and their mutations is complicated by the partial functional redundancy shared by PSEN1 and PSEN2 and their complex molecular biology. For example, the PSEN1 holoprotein has been shown to have γ-secretase-independent activities required for normal lysosomal acidification [20], can form multimers [21 –24], and may interact with the HIF1α protein [25 –27] that is critical both for responses to hypoxia and for iron homeostasis (reviewed in [28]). Additionally, within γ-secretase complexes, the PSENs act to cleave at least 149 different substrates [29]. To simplify analysis, most previous investigation of PSEN activity has involved inactivation (knock-out) of PSEN1 and/or PSEN2 in cells or animals, and expression of only single forms of mutant PSEN (i.e., without simultaneous expression of wild type forms). Forced expression of PSEN genes is also usually at non-physiological levels which has unexpected regulatory feedback effects [30]. In humans, investigating PSEN’s role in AD at the molecular level is restricted to postmortem brain tissues. However, these show substantially different patterns of gene expression compared to the brains of people with mild cognitive impairment or age-matched healthy controls [31]. Since AD is thought to take decades to develop [32], we must understand the pathological effects of EOfAD mutations in young adult brains if we wish to identify preventative treatments. For this reason, we must model EOfAD mutations in animals.

The overwhelming majority of animal modelling of AD has utilized transgenic models favored for their apparent, partial reproduction of Aβ histopathology and easily discernible cognitive disabilities. However, the relationship between Aβ histopathology and cognitive change in these models is questionable [33]. Additionally, the most detailed form of molecular phenotyping currently available, transcriptome analysis, shows little consistency between the disturbed brain gene transcription of various transgenic models and limited concordance between them and human sporadic AD brain transcriptomes [34]. “Knock-in” mouse models of single EOfAD mutations (e.g., [35, 36]) make the fewest assumptions regarding the pathological mechanism(s) of AD and most closely replicate the human EOfAD genetic state (i.e., incorporating a single, dominant, endogenous mutation in the heterozygous state). However, the brain transcriptomes of these mice have never been analyzed, and interest in them waned due to their lack of Aβ histopathology and mild cognitive effects.

Analysis of mouse brain transcriptomes is complicated by strong effects on gene expression of sex [37, 38] and, potentially, litter-of-origin (i.e., due to environmental and genotypic variation) [39]. In contrast, zebrafish brain transcriptomes show only subtle influences of sex, and very large numbers of siblings can be generated from single mating event, alleviating potential litter-of-origin issues [40 –44]. In 2014, our laboratory began a program of creating knock-in models of EOfAD-like (and non-EOfAD-like) mutations in the zebrafish genes orthologous to PSEN1, PSEN2, and SORL1. In 2019 we began publishing the results of transcriptome analyses of the young adult brains of these fish [40 –47] as we attempt to establish what effect(s) all the EOfAD mutations have in common (and differentiate them from the non-EOfAD-like mutations).

Our previous analyses of an EOfAD-like mutation in the zebrafish psen1 gene, Q96_K97del, revealed very significant effects on the expression of genes involved in mitochondrial function, lysosomal acidification, and iron homeostasis [46, 47]. Although Q96_K97del follows the reading-frame preservation rule of EOfAD mutations in PSEN1, it is not an exact equivalent of any human EOfAD-causative mutation. Consequently, in this study we aimed to generate and analyze an additional, exactly equivalent, model of a human PSEN1 EOfAD mutation. For technical reasons, the T428 codon of zebrafish psen1 (equivalent to the T440 codon of human PSEN1) was predicted to be readily targetable using CRISPR-Cas9 technology, and we subsequently deleted this codon in the zebrafish gene. This generated a zebrafish model of the human EOfAD mutation PSEN1^T440del [48]. This mutation was identified in a Japanese man classified as displaying a mixed dementia phenotype (variant AD with spastic paraparesis, Parkinson’s disease, and dementia with Lewy bodies). The T440del mutation of human PSEN1 mutation greatly reduces the intrinsic γ-secretase activity of the protein [15], similar to some other EOfAD mutations such as C410Y and L435F [15, 49]. This loss of intrinsic γ-secretase activity inhibits auto-catalytic endoproteolysis leading to accumulation of the holoprotein [49, 50].

To understand how reading-frame preserving and frameshift mutations can cause completely different diseases, we also generated a frameshift mutation in zebrafish psen1, W233fs, very similar to the fAI-causative P242fs mutation of human PSEN1. We then performed an RNA-seq analysis with high read depth and large sample numbers to compare the brain transcriptomes of fish from a single family of young adult siblings heterozygous for either of the psen1^T428del or psen1^W233fs mutations, or wild type. We observed subtle, and mostly distinct, effects of the two mutations. In particular, changes in the fAI-like brain transcriptomes implied significant effects on Notch, Wnt, neurotrophin, and Toll-like receptor signaling, while changes in the EOfAD-like brain transcriptomes implied effects on oxidative phosphorylation similar to those previously seen for EOfAD-like mutations in psen1 [46], psen2 [41, 44], and sorl1 [40, 42].

MATERIALS AND METHODS

Zebrafish husbandry and animal ethics

All zebrafish (Tübingen strain) used in this study were maintained in a recirculating water system on a 14 h light/10 h dark cycle, fed dry food in the morning and live brine shrimp in the afternoon. All zebrafish work was conducted under the auspices of the Animal Ethics Committee (permit numbers S-2017-089 and S-2017-073) and the Institutional Biosafety Committee of the University of Adelaide.

CRISPR-Cas9 genome editing

To mutate zebrafish psen1, we used the Alt-R® CRISPR-Cas9 system (Integrated DNA Technologies, Coralville, IA, USA). To generate the T428del mutation (EOfAD-like) in exon 11 of psen1, we used a custom-designed crRNA recognizing the sequence 5’ CTCCCCATCTCCATAACCTT 3’ and a PAM of CGG. For the W233fs mutation (fAI-like), the crRNA was designed to recognize the sequence 5’ GATGAGCCATGCGGTCCACT 3’ in exon 6 of psen1, with a PAM sequence of CGG. We aimed to generate exact equivalents of the human P242fs mutation causing fAI, and the T440del mutation causing EOfAD by homology directed repair (HDR). For the P242fs mutation, we used a plasmid DNA template as described in [51] (synthesized by Biomatik, Kitchener, Ontario, Canada). For the T440del mutation, we used an antisense, asymmetric single-stranded oligonucleotide with phosphorothioate modifications (synthesized by Merck, Kenilworth, NJ, USA) as described in [52] (HDR template DNA sequences are given in the Supplementary Material).

Each crRNA was annealed with an equal amount of Alt-R^® CRISPR-Cas9 tracrRNA (IDT) in nuclease free duplex buffer (IDT) by heating at 95°C for 5 min, then allowed to cool to room temperature, giving sgRNA solutions of 33μM (assuming complete heteroduplex formation of the RNA molecules). Then, 1μL of the sgRNA solution was incubated with 1μL of Alt-R^® S.p.Cas9 Nuclease 3NLS (IDT) at 64μM at 37°C for 10 min to form ribonucleoprotein (RNP) complexes. The final concentration for the linear ssODN for the T428del mutation was 1μM, and the final concentration of the plasmid DNA for the W233fs mutation was 25 ng/μL. Approximately 2–5 nL of RNP complexes in solution with the respective template DNAs were injected into Tübingen strain zebrafish embryos at the one cell stage. The procedures followed for testing of the mutagenesis efficiencies of CRISPR-Cas9 systems using allele-specific polymerase chain reactions and T7 endonuclease I assays, and the breeding strategy to isolate the mutations of interest, are described in [42, 53].

RNA-seq raw data generation and processing

We performed RNA-seq on a family of zebrafish as described in Fig. 1. Total RNA (with genomic DNA depleted by DNaseI treatment) was isolated from the brains of n = 4 fish per genotype and sex as described in [42]. Then, 500 ng of total RNA (RIN_e > 9) was delivered to the South Australian Genomics Centre (SAGC, Adelaide, Australia) for polyA+ library construction (with unique molecular identifiers (UMIs)) and RNA-sequencing using the Illumina Novaseq S1 2x100 SE platform.

The raw fastq files from SAGC were provided as 100 bp paired end reads as well as an index file containing the UMIs for each read (over two Novaseq lanes which were subsequently merged). The merged raw data was processed using a developed Nextflow [54] RNA-seq workflow (see https://github.com/sagc-bioinformatics/sagc-rnaseq-nf). Briefly, UMIs were added to headers of each read using fastp (v0.20.1). Alignment of the reads to the zebrafish genome (GRCz11, Ensembl release 98) was performed using STAR (v2.5.3a). Then, reads which contained the same UMI (i.e., PCR duplicates) were deduplicated using the dedup function of umi_tools (version 1.0.1). Finally, the gene-level counts matrix was generated using featureCounts from the Subread (version 2.0.1) package.

Differential gene expression

Statistical analysis of the RNA-seq data was performed using R [55]. Since lowly expressed genes are considered uninformative for differential expression analysis, we omitted genes with less than 0.1 counts per million (CPM) (following the 10/minimum library size in millions rule described in [56]). Library sizes after omitting the lowly expressed genes ranged between 61 and 110 million reads. These were normalized using the trimmed mean of M-values (TMM) method [57]. To test for differential expression of genes due to heterozygosity for the T428del or W233fs mutation, we used a generalized linear model and likelihood ratio tests using edgeR [58, 59]. A design matrix was specified with the wild type genotype as the intercept, and the T428del/+ and W233fs/+ genotypes as the coefficients. We considered a gene to be differentially expressed (DE) due to each psen1 mutant genotype if the FDR adjusted p-value was less than 0.05. We assessed whether a bias was observed for a gene to be considered DE due to its length or GC content. However, a considerable bias was not observed (Supplementary Figure 4).

Enrichment analysis

We tested for over-representation of gene ontology (GO) terms within the DE gene lists using goseq [60], using the average transcript length per gene to calculate the probability weighting function (PWF). We considered a GO term to be significantly over-represented within the DE gene lists relative to all detectable genes in the RNA-seq experiment if the FDR-adjusted p-value generated by goseq was less than 0.05.

We also performed enrichment analysis on the entire list of detectable genes by calculating the harmonic mean p-value from the raw p-values calculated from fry [61], camera [62] and fgsea [63, 64] as described in [42]. To test for changes to gene expression in a broad range of biological processes, we used the KEGG [65] gene sets obtained from MSigDB [66] using the msigdbr package [67]. We also used msigdbr to obtain gene sets which contain genes that show changed expression in response to changes in the Notch signalling pathway (NGUYEN_NOTCH1_TARGETS_UP and NGUYEN_NOTCH1_TARGETS_DN, NOTCH_DN.V1_UP, NOTCH_DN.V1_DN, and RYAN_MANTLE_CELL_LYMPHOMA_NOTCH_DIRECT_UP). We also tested for evidence of possible iron dyshomeostasis using gene sets containing genes encoding transcripts which contain iron-responsive elements in their untranslated regions (described in [46]).

Comparison of the T428del and Q96_K97del mutations in psen1

Isolation of the zebrafish Q96_K97del mutation in zebrafish psen1 and analysis of its effects on zebrafish brain transcriptomes have been described previously [46, 47]. That dataset is comprised of brain RNA-seq data for fish heterozygous for the Q96_K97del mutation and their wild type siblings, at 6 months old (young adult) and 24 months old (aged), and under normoxia or hypoxia treatment (n = 4 fish per genotype, age and treatment). In the analysis presented here, we performed enrichment analysis using the methods described above on the entire dataset but presented the results for the pairwise comparison between 6-month-old Q96_K97del/+ fish and wild type fish under normoxia.

To obtain a broader comparison on the effects of the Q96_K97del and T428del mutations, we performed adaptive, elastic-net sparse PCA (AES-PCA) [68] as implemented in the pathwayPCA package [69]. For this analysis, we utilized the HALLMARK [70] gene sets from MSigDB to generate the pathway collection. The pathway principal components (PCs) were calculated only on the gene expression data from samples heterozygous for an EOfAD-like mutation (Q96_K97del or T428del) and their wild type siblings under normoxia at 6 months of age. Then, the categorical effect of genotype was tested for association with the pathway PCs using a permutation-based regression model as described in [69].

Data availability

The paired end fastq files and the output of featureCounts have been deposited in GEO under accession number GSE164466. Code to reproduce this analysis can be found at https://github.com/karissa-b/psen1_EOfAD_fAI_6m_RNA-seq.

RESULTS

Generation of an EOfAD-like and a fAI-like mutation in zebrafish psen1

An unsolved puzzle regarding the dominant EOfAD mutations of human PSEN1 (and PSEN2) is why these are consistently found to permit production of transcripts in which the reading frame is preserved, while heterozygosity for mutations causing frameshifts (or deleting the genes) does not cause EOfAD. To investigate this quandary in an in vivo model, we initially aimed to generate mutations in zebrafish psen1 which would be exact equivalents of the T440del and P242fs mutations using homology directed repair (HDR). While screening for the desired mutations, we identified the mutations W233fs and T428del, both likely generated by the non-homologous end joining (NHEJ) pathway of DNA repair. T428del is a 3 nucleotide deletion which, nevertheless, produces a protein-level change exactly equivalent to that observed for the human T440del mutation. Hereafter, for simplicity, we refer to the T428del mutation as “EOfAD-like”. W233fs is an indel mutation causing a frameshift in the second codon downstream of the zebrafish psen1 proline codon equivalent to human PSEN1 codon P242. This change is still within the short third lumenal loop of the Psen1 protein (see Fig. 1). Assuming no effect on splicing, the frameshift mutation results in a premature stop codon 36 codons downstream of W233 (Supplementary Figure 1). Hereafter, we refer to this mutation as “fAI-like”. The alignment of the wild type and mutant sequences in humans and zebrafish is shown in Fig. 2.

Fig. 2

A) Alignment of a region of wild type human (Hs) PSEN1 exon 7, and the same region containing the human P242fs (P242LfsX11) mutation, and the equivalent zebrafish (Dr) psen1 exon 6 wild type and W233fs sequences. B) Alignment of a region of wild type human (Hs) PSEN1 exon 12, and the same region containing the human T440del mutation, and the equivalent zebrafish (Dr) psen1 exon 11 wild type and T428del sequences.

Heterozygosity or homozygosity for neither the EOfAD-like nor the fAI-like mutation produces any obvious morphological defects. However, this is unsurprising considering that rare examples of humans homozygous for EOfAD mutations are known [71, 72] and loss of PSEN1 γ-secretase activity is, apparently, compatible with viability in zebrafish [73] and rats [74] (although not in mice [49]).

Transcriptome analysis

To investigate global changes to the brain transcriptome due to heterozygosity for the EOfAD-like or fAI-like mutations in psen1, we performed mRNA-seq on a family of fish as described in Fig. 1. The family of sibling fish were raised together in a single tank, thereby reducing sources genetic and environmental variation between individuals and allowing subtle changes to the transcriptome to be detected with minimal confounding effects.

To begin our exploration of the similarities and differences between the brain transcriptomes of the mutant and wild type fish, we first performed principal component analysis (PCA) on the gene level, log transformed counts per million (logCPM) of the zebrafish RNA-seq samples. A plot of principal component 1 (PC1) against PC2 did not show distinct clustering of samples by genotype or sex, indicating that these variables do not result in stark changes to the brain transcriptome. This is consistent with our previous observations of EOfAD-like mutations in other genes [40 –42]. Additionally, we did not observe a relationship between PC1 or PC2 with RNA-seq library size (Supplementary Figure 2). However, some separation of the EOfAD-like and fAI-like samples is observed across PC2, indicating distinct, but subtle, differences between these transcriptomes. Notably, the majority of the variation in this dataset is not captured until PC6 (Fig. 3). An additional exploration of the gene expression values was performed by principal coordinate analysis (PCoA) using the Bray-Curtis method to calculate the dissimilarity between samples (Supplementary Figure 3). Similar results were observed to the PCA, supporting that heterozygosity for either of these two mutations in psen1 does not result in widespread transcriptome changes.

Fig. 3

Principal component (PC) analysis of the gene expression values for the RNA-seq experiment. A) PC1 plotted against PC2 for each sample. Each point represents a sample and is colored according to psen1 genotype. Female (F) samples appear as circles and male (M) samples appear as triangles. B) Scree plot indicating the variance explained by each principal component. The points joined by lines indicate the cumulative variance explained by each PC. See online version for color.

Heterozygosity for the EOfAD-like or fAI-like mutations of psen1 causes only subtle effects on gene expression

Which genes are dysregulated due to heterozygosity for the EOfAD-like or the fAI-like mutations? To address this question, we performed differential gene expression analysis using a generalized linear model and likelihood ratio tests with edgeR. We observed statistical evidence for 13 genes as significantly differentially expressed (DE) due to heterozygosity for the EOfAD-like mutation, and 5 genes due to the fAI-like mutation (Fig. 4, Supplementary Table 1). Notably, psen1 was the most significantly DE gene due to heterozygosity for the fAI-like mutation (logFC = –0.8, FDR = 1.33e-78), consistent with the observation that frame-shift mutations commonly induce nonsense-mediated mRNA decay when they result in premature stop codons (reviewed in [75]). Total levels of psen1 transcripts were unchanged in EOfAD-like/+ brains (logFC = –0.0065, FDR = 1, Supplementary Figure 8). No statistically significantly DE genes were found to be shared between the comparisons of either form of heterozygous mutant to wild type or were found to be significantly overrepresented by any gene ontology (GO) terms by goseq (Supplementary Tables 3 and 4). This is not unexpected due to the relatively low number of significantly DE genes detected.

Fig. 4

Differential expression analysis. A) Mean-difference (MD) plots and B) volcano plots of changes to gene expression in EOfAD-like/+ and fAI-like/+ mutant zebrafish brains. Note that the limits of the y-axis in B) are restrained to between 0 and 10 for visualization purposes. C) Upset plot indicating the low number of genes which are significantly differentially expressed (DE) in either comparison. See online version for color.

Significant differences in gene expression between the EOfAD-like and fAI-like mutants can be detected at the pathway level

Since very few DE genes were detected in each comparison of heterozygous psen1 mutant fish to their wild type siblings, we performed enrichment analysis on all detectable genes in the RNA-seq experiment. Our method, inspired by the EGSEA framework [76], involves calculation of the harmonic mean p-value [77] from the raw p-values of three different rank-based gene set testing methods: fry [61], camera [62], and GSEA [63, 64]. Unlike EGSEA, we use the harmonic mean p-value to combine the raw p-values, as the harmonic mean p-value has been specifically shown to be robust for combining dependent p-values [77]. We performed enrichment testing using the KEGG gene sets (describing 186 biological pathways and processes) to obtain information on changes to activities for these pathways. We also tested for evidence of iron dyshomeostasis using our recently defined sets of genes containing iron-responsive elements (IREs) in the untranslated regions of their mRNAs [46]. We observed statistical evidence for 7 KEGG gene sets as significantly altered by heterozygosity for the EOfAD-like mutation and 11 KEGG gene sets as significantly altered by heterozygosity for the fAI-like mutation (Fig. 5, full results of the raw p-values from each algorithm as well as the harmonic mean p-value can be found in Supplementary Table 2). Gene sets significantly altered in the brains of both forms of heterozygous mutant included the KEGG gene sets for cytokine receptor interactions, Jak/Stat signaling, and encoding the components of the ribosomal subunits. Inspection of the leading edge genes (which can be interpreted as the core genes driving the enrichment of a gene set) showed that similar genes were driving the enrichment of the gene sets for cytokine receptor interactions and Jak/Stat signaling. Similar genes were also driving the enrichment of the KEGG_RIBOSOME gene set in both heterozygous mutants. However, the magnitude of the logFC was greater in the fAI-like/+ samples, suggesting a stronger effect (Supplementary Figures 5–7). Gene sets which were only altered significantly by heterozygosity for the EOfAD-like mutation were involved in energy metabolism (KEGG_PARKINSONS_DISEASE, KEGG_OXIDATIVE_PHOSPHORYLATION, and KEGG_CITRATE_CYCLE_TCA_CYCLE). Notably, the KEGG gene sets for Parkinson’s disease, Huntington’s disease, and for oxidative phosphorylation, share 55 leading-edge genes, implying that their enrichment is driven by, essentially, the same gene expression signal (Fig. 5). Conversely, the 10 KEGG gene sets found to be altered significantly by heterozygosity for the fAI-like mutation appear to be driven mostly by distinct gene expression signals. No IRE gene sets were observed statistically to be altered in the brains of either mutant, suggesting that iron homeostasis is unaffected (at least at 6 months of age). For additional visualizations of the changes to gene expression observed in psen1 mutant brains, see Supplementary Figures 10–13. The changes to expression of genes within the KEGG gene sets are likely not due to broad changes in the relative numbers/proportions of different cell-types in the zebrafish brain samples, since the expression of marker genes of neurons, astrocytes, oligodendrocytes and microglia was similar in all samples (Supplementary Figure 14).

Fig. 5

A) KEGG gene sets with FDR-adjusted harmonic mean p-values of < 0.05 in psen1 EOfAD-like/+ and fAI-like/+ mutant brains. The color of the cells indicates the level of significance (brighter color indicates greater statistical significance, while dark grey indicates the FDR-adjusted harmonic mean p-value > 0.05). The number within each cell is the FDR-adjusted harmonic mean p-value. B) Upset plot indicating the overlap of leading edge genes from the fgsea algorithm which drive the enrichment of gene sets significantly altered in EOfAD-like/+ and C) fAI-like/+ brains. See online version for color.

The EOfAD-like and fAI-like mutations alter expression of Notch signaling genes

Notch signaling plays a critical role in many cell differentiation events and is dependent on PSEN’s γ-secretase activity. Disturbance of Notch signaling due to decreased γ-secretase activity has been suggested to contribute to the changes in skin histology of fAI, as Notch signaling is required for normal epidermal maintenance ([78 –80] and reviewed by [81]). However, fAI has not been reported as associated with EOfAD, despite that the T440del mutant form of PSEN1 appears to have little intrinsic γ-secretase activity [15]. The expression of genes involved in the KEGG gene set for the Notch signaling pathway was observed to be highly significantly altered in the brains of fAI-like/+ mutants, but not of EOfAD/+ mutants, implying that γ-secretase activity might only be affected significantly by the frameshift, fAI-like mutation (Fig. 5). However, inspection of the logFC of genes in the KEGG_NOTCH_SIGNALING_PATHWAY gene set revealed similar patterns of changes to gene expression in both mutants (Fig. 6). Upregulation of the genes encoding the Notch and Delta receptors is observed in both mutants compared to wild type. In fAI-like/+ brains, we observe downregulation of the downstream transcriptional targets of the Notch intracellular domain (NICD), implying decreased Notch signaling (and, likely, reduced γ-secretase activity). Genes encoding repressors of Notch signaling are observed to be upregulated (i.e., dyl and numb), reinforcing this interpretation.

Fig. 6

A) Pathview [87] visualisation of the changes to gene expression in the KEGG_NOTCH_SIGNALING_PATHWAY gene set in EOfAD-like/+ mutants and (B) fAI-like/+ mutants. C) The proportion of genes with increased expression (red, z> √2) and decreased expression (blue, z< √2) in MSigDB gene sets for Notch signaling in EOfAD-like/+ and fAI-like/+ mutant brains. Gene sets which contained an FDR-adjusted harmonic mean p-value (HMP) < 0.05 appear less transparent. The FDR adjust p-values are also listed on the bars. D) The expression of psen2 is trending towards upregulation, particularly in fAI-like/+ mutants. Here, p-values were determined by Student’s unpaired t-tests. E) Heatmap indicating the logFC values for genes in the RYAN_MANTLE_CELL_LYMPHOMA_NOTCH_DIRECT_UP gene set. Genes are clustered based on their Euclidean distance and are labelled with green if they appear in the leading edge of the fgsea algorithm for each comparison of a psen1 heterozygous mutant with wild type. See online version for color.

Since the KEGG gene set for Notch signaling only contains two genes that are direct transcriptional targets of the NICD, we investigated further whether Notch signaling is perturbed in both mutants by analysis of gene sets from MSigDB containing information on genes responsive to Notch signaling in different cell lines: NGUYEN_NOTCH1_TARGETS_UP; NGUYEN_NOTCH1_TARGETS_DN; NOTCH_DN.V1_UP; NOTCH_DN.V1_DN; and RYAN_MANTLE_CELL_LYMPHOMA_NOTCH_DIRECT_UP. The NGUYEN_NOTCH1_TARGETS_UP and _DOWN gene sets consist of genes which have been observed as up- or downregulated respectively in response to a constitutively active Notch receptor in keratinocytes [82]. The NOTCH_DN.V1_UP and _DN gene sets contain genes which are up- and down-regulated respectively in response to treatment with the γ-secretase inhibitor DAPT in a T-cell acute lymphoblastic leukemia (T-ALL) cell line [83]. The RYAN_MANTLE_CELL_LYMPHOMA_NOTCH_DIRECT_UP gene set contains genes showing both increased expression upon rapid activation of Notch signaling by washout of the γ-secretase inhibitor compound E, and evidence for a NICD binding site in the promotor by ChIP-seq, in mantle cell lymphoma cell lines [84]. (Note that there is no equivalent “RYAN_MANTLE_CELL_LYMPHOMA” gene set representing genes downregulated in response to Notch signaling.) Of these 5 gene sets, statistical support was found only for changes to the expression of genes in the RYAN_MANTLE_CELL_LYMPHOMA_NOTCH_DIRECT_UP gene set, and this was found for both the EOfAD-like (p = 0.006) and the fAI-like (p = 0.008) mutants. The leading edge genes were mostly observed to be upregulated, which supports increases in Notch signaling (implying increased γ-secretase activity). Transcriptional adaptation (previously known as “genetic compensation”) might contribute to the apparent increase in Notch signaling in the frameshift, fAI-like/+ mutant brains via upregulated expression of the psen1-homologous gene, psen2 [85, 86]. Although no statistically significant differences in expression were observed for psen2 in the differential expression test using edgeR (see Supplementary Table 1), a trend toward upregulation in the fAI-like/+ mutants was observed following a simple Student’s t-test (p = 0.074, Fig. 6D). El-Brolosy et al. [85] showed that the wild type allele of a mutated gene can also be upregulated by transcriptional adaptation (where the mutation causes nonsense-medicated decay, NMD, of mutant transcripts). Inspection of the number of reads aligning to the W233 mutation site across samples indicates that the expression of the wild type psen1 allele in fAI-like/ + brains appears to be greater than 50%of the expression of the wild type psen1 allele in wild type brains (p = 0.006), providing further evidence for transcriptional adaptation due to the fAI-like mutation (Supplementary Figure 9).

Together, these results suggest that Notch signaling and, by implication, γ-secretase activity, may be enhanced in psen1 mutant brains. However, future biochemical assays should be performed to confirm this prediction.

The EOfAD-like mutation T428del has a milder phenotype than the previously studied Q96_K97del EOfAD-like mutation of psen1

The T428del mutation of psen1 is the first identified zebrafish mutation exactly equivalent (at the protein sequence level) to a characterized human EOfAD mutation. Therefore, we sought to assess the consistency of its effects with those of a previously studied EOfAD-like mutation, Q96_K97del, and to identify cellular processes affected in common by the two mutations. The Q96_K97del mutation deletes two codons in the sequence encoding the first lumenal loop of the Psen1 protein (see Fig. 1). Comparison of transcriptomes from the 6-month-old brains of Q96_K97del/+ and wild type siblings previously predicted changes to expression of genes involved in energy metabolism, iron homeostasis, and lysosomal acidification [46, 47]. To compare which cellular processes are affected by heterozygosity for the Q96_K97del mutation or the T428del mutation, we first performed enrichment analysis on the RNA-seq data previously generated by our analysis of zebrafish heterozygous for the Q96_K97del mutation relative to their wild type siblings. Here, we observed that heterozygosity for the Q96_K97del mutation results in significant alterations in 7 KEGG gene sets (at 6 months of age during normoxia, Fig. 7A). We also found statistical evidence for altered expression of genes possessing IREs in their 3’ UTRs (see IRE3_ALL in Fig. 7A), consistent with our previous finding using a different method of enrichment analysis [46]. Gene sets affected in common between the two EOfAD-like mutations in psen1 are involved in energy metabolism and protein translation (Fig. 7A). The expression of genes involved in protein degradation, and of genes containing IREs in the 3’ UTRs of their transcripts, appeared significantly altered only by the Q96_K97del mutation (Fig. 7A).

Fig. 7

A) Comparison of KEGG and IRE gene sets significantly altered by the EOfAD-like mutations T428del and Q96_K97del in 6-month-old zebrafish brains. Each cell is colored according to statistical significance, and the FDR-adjusted harmonic mean p-value is shown. Gene sets not significantly altered (FDR adjusted harmonic mean p-value > 0.05) in a comparison between a psen1 mutant zebrafish with their respective wild type siblings appear grey. B) Principal component 1 (PC1) values for the HALLMARK gene sets as calculated by AES-PCA, clustered based on their Euclidean distance in T428del/+ samples relative to their wild type siblings. C) PC1 values for the HALLMARK gene sets as calculated by AES-PCA clustered based on their Euclidean distance in Q96_K97del/+ samples relative to their wild type siblings at 6 months of age under normal oxygen conditions. See online version for color.

We also compared the effects of the two EOfAD-like mutations using adaptive, elastic-net sparse PCA (AES-PCA) as implemented in the pathwayPCA package [69]. AES-PCA allows reduction of data dimensionality and for the overall activity of predefined gene sets to be observed in a sample-specific manner [68]. To obtain a global view of the changes to gene expression between the two psen1 EOfAD-like mutations over the two datasets, we utilized the HALLMARK gene sets that encompass 50 distinct biological processes (rather than the 186 KEGG gene sets that share many genes).

The latent variables estimated by AES-PCA for the HALLMARK gene sets (i.e., the first principal components) in each dataset did not show any significant association with psen1 genotype, suggesting that changes to gene expression (measured over entire brains) are too subtle to be detected as statistically significant using this method. However, clustering of the calculated PC1 values by AES-PCA for each HALLMARK gene set in each sample and dataset revealed that samples in the Q96_K97del dataset clustered mostly according to genotype (one wild type sample did not follow the trend), supporting that heterozygosity for the Q96_K97del mutation does result in marked effects on gene expression for the HALLMARK gene sets. Conversely, clustering of PC1 values in the T428del dataset resulted in two distinct clusters of samples. However, samples did not group by genotype over the two clusters to the same extent as seen for the Q96_K97del dataset. Intriguingly, the Q96_K97del dataset had less sample numbers per genotype (n = 4), and did not have as great sequencing depth as the current RNA-seq experiment. Therefore, this supports that heterozygosity for the Q96_K97del mutation has more consistent (more severe) effects on young adult brain transcriptomes than heterozygosity for the T428del mutation (Fig. 7B,C).

DISCUSSION

In this study, we exploited transcriptome analysis of whole brains of young adult zebrafish siblings, to detect differences in molecular state between the brains of fish heterozygous for an EOfAD-like mutation or an fAI-like mutation of psen1 compared to wild type in vivo. The subtlety of the effects observed is consistent with that EOfAD is, despite its designation as “early-onset”, a disease affecting people overwhelmingly at ages older than 30 years [88]. The person reported to carry the T440del mutation of PSEN1 showed cognitive decline at 41 years [48]. (Overall, EOfAD mutations in PSEN1 show a median survival to onset of 45 years [88]). In contrast, at 6 months of age, zebrafish are only recently sexually mature. Nevertheless, since AD is thought to take decades to develop [32], it is these subtle, early changes that we must target therapeutically if we wish to arrest the pathological processes driving the progression to AD. As seen in all our previous analyses of EOfAD-like mutations [40–42 , 46], changes in expression of genes involved in oxidative phosphorylation were identified as significant. However, this was not the case for the fAI-like, frameshift mutation. Therefore, oxidative phosphorylation changes appear to be an early signature cellular stress of EOfAD. Changes to mitochondrial function have been observed in heterozygous PSEN1 mutant astrocytes [89], homozygous PSEN1 mutant neurons [90], and in neurons differentiated from human induced pluripotent stem cells (hIPSCs) from sporadic AD patients [91], supporting our findings. However, such changes are not always observed [92], possibly due to issues of experimental reproducibility between laboratories when working with hIPSCs [93].

The EOfAD-like mutation also caused highly statistically significant changes in the KEGG_PARKINSONS_DISEASE gene set (FDR-adjusted harmonic mean p-value = 7.4e-7). This gene set shares many genes with KEGG_OXIDATIVE_PHOSPHORYLATION (FDR-adjusted harmonic mean p-value = 3.4e-6) and the person carrying the PSEN1^T440del mutation modelled by zebrafish psen1^T428del initially showed symptoms of early onset parkinsonism at 34 years of age before those of cognitive decline at 41 years [48].

In contrast to this EOfAD-like mutation, the fAI-like mutation apparently caused highly statistically significant changes in Notch signaling (FDR-adjusted harmonic mean p-value = 1.55e-15) and changes in other signal transduction pathways such as those involving Wnt and neurotrophins, as might be expected from changes in γ-secretase activity. Also notable was enrichment for the KEGG_TOLL_LIKE_RECEPTOR_SIGNALLING_PATHWAY gene set since acne inversa is a chronic inflammatory skin disorder and, in humans, increased expression of Toll-like receptor 2 has been noted in acne inversa lesions [94].

Both the EOfAD-like and fAI-like mutations caused highly statistically significant changes in the gene sets KEGG_CYTOKINE_CYTOKINE_RECEPTOR_INTERACTION and KEGG_RIBOSOME (FDR-adjusted harmonic mean p-value < 0.002 in all cases). The former gene set reflects that both mutations appear to affect inflammation that is a characteristic of the pathologies of both EOfAD [95] and fAI (reviewed in [96]). Like oxidative phosphorylation, we have also observed effects on ribosomal protein genes sets for every EOfAD-like mutation we have studied [40–42 , 44]. This may be due protein synthesis consuming a large proportion of cells’ energy budgets [97] and requiring amino acid precursors that can be sourced from lysosomes. Recently, Bordi et al. [98] noted that mTOR is highly activated in fibroblasts from people with Down syndrome (DS, trisomy 21). DS individuals commonly develop EOfAD due to overexpression of the AβPP gene (that is resident on human chromosome 21). The consequent increased expression of AβPP’s β-CTF/C99 fragment (generated by β-secretase cleavage of AβPP without γ-secretase cleavage) affects endolysosomal pathway acidification [99] in a similar manner to EOfAD mutations of PSEN1 [100]. The mTOR protein is localized at lysosomes in the mTORC1 and mTORC2 protein complexes (reviewed in [101, 102]) and monitors the energy and nutrient status of cells (reviewed in [103]). It is important for regulating ribosomal activity, partly by regulating transcription of ribosome components (reviewed in [104, 105]). Therefore, one explanation for the consistent enrichment for transcripts of the KEGG_RIBOSOME gene set we see in EOfAD mutant brains may be mTOR activation due to effects on lysosomal acidification and/or the energy status of cells. We did not observe any significant changes to the expression of genes involved in the mTOR signaling pathway in the analyses described in this paper (the FDR-adjusted harmonic mean p-value for the KEGG_MTOR_SIGNALING_PATHWAY was 0.7 for each comparison of the psen1 mutant fish to their wild type siblings, see Supplementary Table 2). However, these changes could be undetectably subtle in young adult brains and/or occurring at the protein level and therefore not observable in bulk RNA-seq data. (Statistically significant enrichment for genes in the HALLMARK_PI3K_AKT_MTOR_SIGNALING gene set was seen previously for the normoxic, 6-month-old brains of fish heterozygous for the more severe EOfAD-like mutation Q96_K97del when compared to wild type siblings [46].)

While the brain transcriptome alterations caused by the EOfAD-like and fAI-like mutations are subtle (as illustrated by the lack of tight clustering of samples in the principal component analysis in Fig. 3, and the low number of significantly differentially expressed genes), we are reassured in their overall veracity by their similarity to the results of a parallel analysis of sibling brain transcriptomes from 6-month-old zebrafish heterozygous for either a frameshift or a frame-preserving mutation in the zebrafish psen2 gene relative to wild type [41]. In that similarly structured (but less statistically powered) experiment, only the frameshift mutation significantly affected the KEGG_NOTCH_SIGNALLING gene set while only the frame-preserving, EOfAD-like mutation significantly affected the KEGG_OXIDATIVE_PHOSPHORYLATION gene set. Both psen2 genotypes affected the KEGG_RIBOSOME gene set, but in overall opposite directions (the frameshift mutation largely upregulated these genes while the frame-preserving mutation did the opposite).

Transcriptome analysis can reveal a great deal of data on differences in gene transcript levels between different genotypes or treatments. However, interpreting changes in cellular state from this information is not straight forward. Are any changes seen direct molecular effects of a mutation/treatment (e.g., the direct, downstream effects of a change in γ-secretase activity) or homeostatic responses as cells/tissues adjust their internal states to promote survival? For example, in the KEGG_NOTCH_SIGNALING_PATHWAY gene set shown in Fig. 6B, more pathway components are upregulated than are downregulated. However, the direct transcriptional targets of Notch signalling (her4.2 and heyl) are downregulated, as might be expected from reduced expression of wild type, catalytically-competent Psen1 protein. The upregulation of other components of the pathway may represent homeostatic responses attempting to restore normal levels of Notch signaling.

Only two Notch downstream transcriptional target genes are described in the KEGG_NOTCH_SIGNALING_PATHWAY gene set. Therefore, in an effort to assess more generally the effects of the EOfAD-like and fAI-like mutations on γ-secretase activity, we also analyzed additional sets of genes previously identified (in various systems) as direct transcriptional targets of γ-secretase-dependent signaling. One of these sets, encompassing genes identified as Notch signaling targets by both γ-secretase inhibitor responses and binding of the Notch intracellular domain to chromatin, revealed apparent upregulation of Notch signaling in both the EOfAD-like and fAI-like heterozygous mutant brains relative to wild type siblings. In both EOfAD-like/+ and fAI-like/+ mutant brains, the most highly ranked genes in terms of differential expression tended to be upregulated (although some highly ranked genes were downregulated in fAI-like/+ brains). The idea that putative low levels of a form of Psen1 protein truncated in the third lumenal loop domain could increase Notch signaling is not unexpected, as we previously observed an implied upregulation of Notch signaling in zebrafish embryos with forced expression of the fAI-causative P242Lfs allele of human PSEN1 [106]. However, a widespread assumption within AD research is that EOfAD-like mutations of PSEN1 decrease γ-secretase cleavage of AβPP [107], possibly through a dominant negative mechanism [21]. This assumption conflicts with the observation of Sun et al. [15] that approximately 10%of the 138 EOfAD mutations of human PSEN1 they studied actually increased γ-secretase cleavage of AβPP’s β-CTF/C99 fragment (in experiments examining the activities of the mutant proteins in isolation from wild type protein). Zhou et al. [108] also observed increased γ-secretase activity (cleavage of AβPP’s β-CTF/C99 fragment) due to an EOfAD mutation of PSEN1 (S365A, this replicated Sun et al.’s finding for this mutation).

It is important to note that mutations of PSEN1 need not cause similar effects on Notch and AβPP cleavage [106 , 110]. The transmembrane domains of the Notch receptor and the AβPP’s C99 fragment have different conformations [111]. Therefore, changes in the conformation of PSEN1 within γ-secretase due to a mutation may differentially affect Notch and C99 cleavage. Indeed, in our previously mentioned study of forced expression of the human PSEN1^P242Lfs allele in zebrafish embryos, increased apparent Notch signaling was observed without change in AβPP processing [106]. Conversely, Zhang et al. [110] showed that transgenic expression of an EOfAD mutation S169del in PSEN1 under the control of the Thy1 brain specific promotor altered the processing of AβPP in vivo without affecting Notch signaling. Notably, both of these studies did not use the PSEN1 gene’s own promoter to express mutant forms of this gene, and so the effects seen may be distorted by gene/protein over-expression.

Unfortunately, the direct transcriptional targets of the intracellular domain of AβPP (AICD) have not been characterized to the same extent as those of NICD (reviewed in [112]). This constrains transcriptome analysis for detection of differential effects on γ-secretase cleavage of AβPP caused by the EOfAD-like and fAI-like mutations. Exploration of the effects of the EOfAD-like or fAI-like mutations in psen1 on AβPP fragments is hindered in zebrafish due to a lack of antibodies specific to zebrafish-orthologous AβPP proteins (Appa and Appb [113]), and because the β-secretase cleavage site of AβPP may not be conserved in either of the zebrafish AβPP proteins [114]. Investigation of the effects of these mutations on γ-secretase cleavage of AβPP may proceed faster using knock-in mouse models and exploiting the many molecular tools available for that organism. However, the subtlety of the gene expression effects we have observed in young adult heterozygous brains means that their physiological consequences may also be subtle and thus challenging to quantify.

In conclusion, we have performed the first direct comparison of an EOfAD-like and a fAI-like mutation of presenilin 1 in an in vivo model. Both forms of mutation cause apparent changes in inflammation, downregulate expression of genes encoding the components of the ribosome subunits, and potentially affect γ-secretase activity as supported by altered expression of Notch signaling pathway transcriptional target genes. We see that changes to mitochondrial function are a specific, common characteristic of EOfAD-like mutations while the fAI-like mutation specifically affects important signal transduction pathways. These differential effects on brain transcriptomes give insight into how reading-frame preserving mutations in PSEN1 cause EOfAD while frameshift mutations do not.

Footnotes

ACKNOWLEDGMENTS

The authors would like to thank Dr. Nhi Hin for providing the Q96_K97del gene expression values and the zebrafish IRE gene sets. We also would like to thank Dr. Jimmy Breen for his assistance with using the Nextflow pipeline. The authors thank Dr Giuseppe Verdile for critical reading of the manuscript.

This work was supported with supercomputing resources provided by the Phoenix HPC service at the University of Adelaide and by grants GNT1061006 and GNT1126422 from the National Health and Medical Research Council of Australia (NHMRC). KB was supported by an Australian Government Research Training Program Scholarship and by funds from the Carthew Family Charity Trust. YD was supported by an Adelaide Graduate Research Scholarship from the University of Adelaide. MN was supported by funds from the grants listed above. ML is an academic employee of the University of Adelaide.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Janssen

, Beck

, Campbell

, Dickinson

, Fox

, Harvey

, Houlden

, Rossor

, Collinge

(2003) Early onset familial Alzheimer’s disease: Mutation frequency in 31 families. Neurology 60, 235–239.

Żekanowski

, Styczyńska

, Pepłońska

, Gabryelewicz

, Religa

, Ilkowski

, Kijanowska-Haładyna

, Kotapka-Minc

, Mikkelsen

, Pfeffer

, Barczak

, Łuczywek

, Wasiak

, Chodakowska-Żebrowska

, Gustaw

, Łączkowski

, Sobów

, Kuźnicki

, Barcikowska

(2003) Mutations in presenilin 1, presenilin 2 and amyloid precursor protein genes in patients with early-onset Alzheimer’s disease in Poland. Exp Neurol 184, 991–996.

Cruts

, van Duijn

, Backhovens

, Van den Broeck

, Wehnert

, Serneels

, Sherrington

, Hutton

, Hardy

, St George-Hyslop

(1998) Estimation of the genetic contribution of presenilin-1 and-2 mutations in a population-based study of presenile Alzheimer disease. Hum Mol Genet 7, 43–51.

Kamimura

, Tanahashi

, Yamanaka

, Takahashi

, Asada

, Tabira

(1998) Familial Alzheimer’s disease genes in Japanese. J Neurol Sci 160, 76–81.

Area-Gomez

, de Groof

, Boldogh

, Bird

, Gibson

, Koehler

, Yu

, Duff

, Yaffe

, Pon

, Schon

(2009) Presenilins are enriched in endoplasmic reticulum membranes associated with mitochondria. Am J Pathol 175, 1810–1816.

Schon

, Area-Gomez

(2010) Is Alzheimer’s disease a disorder of mitochondria-associated membranes? J Alzheimers Dis 20(Suppl 2), S281–292.

Laudon

, Hansson

, Melén

, Bergman

, Farmery

, Winblad

, Lendahl

, von Heijne

, Näslund

(2005) A nine-transmembrane domain topology for presenilin 1. J Biol Chem 280, 35352–35360.

Fukumori

, Fluhrer

, Steiner

, Haass

(2010) Three-amino acid spacing of presenilin endoproteolysis suggests a general stepwise cleavage of gamma-secretase-mediated intramembrane proteolysis. J Neurosci 30, 7853–7862.

Sun

, Zhao

, Yang

, Yan

, Zhou

, Xie

, Zhao

, Wu

, Li

, Shi

(2015) Structural basis of human γ-secretase assembly. Proc Natl Acad Sci U S A 112, 6003.

10.

Wolfe

(2020) Unraveling the complexity of γ-secretase. Semin Cell Dev Biol 105, 3–11.

11.

Dermaut

, Kumar-Singh

, Engelborghs

, Theuns

, Rademakers

, Saerens

, Pickut

, Peeters

, van den Broeck

, Vennekens

, Claes

, Cruts

, Cras

, Martin

, Van Broeckhoven

, De Deyn

(2004) A novel presenilin 1 mutation associated with Pick’s disease but not beta-amyloid plaques. Ann Neurol 55, 617–626.

12.

, Parks

, Kushner

, Nauman

, Burgess

, Ludwigsen

, Partain

, Nixon

, Allen

, Irwin

, Jakobs

, Litt

, Hershberger

(2006) Mutations of presenilin genes in dilated cardiomyopathy and heart failure. Am J Hum Genet 79, 1030–1039.

13.

Wang

, Yang

, Wen

, Sun

, Su

, Liu

, Ma

, Lv

, Wen

, Qu

, Chen

, Sun

, Shen

, Zhang

(2010) γ-secretase gene mutations in familial acne inversa. Science 330, 1065–1065.

14.

Jiang

, Jayadev

, Lardelli

, Newman

(2018) A review of the familial Alzheimer’s disease locus PRESENILIN 2 and its relationship to PRESENILIN 1. J Alzheimers Dis 66, 1323–1339.

15.

Sun

, Zhou

, Yang

, Shi

(2017) Analysis of 138 pathogenic mutations in presenilin-1 on the in vitro production of Aβ42 and Aβ40 peptides by γ-secretase. Proc Natl Acad Sci U S A 114, E476–E485.

16.

Szaruga

, Veugelen

, Benurwar

, Lismont

, Sepulveda-Falla

, Lleo

, Ryan

, Lashley

, Fox

, Murayama

, Gijsen

, De Strooper

, Chávez-Gutiérrez

(2015) Qualitative changes in human γ-secretase underlie familial Alzheimer’s disease. J Exp Med 212, 2003–2013.

17.

De Strooper

(2007) Loss-of-function presenilin mutations in Alzheimer disease. Talking Point on the role of presenilin mutations in Alzheimer disease. EMBO Rep 8, 141–146.

18.

Jayne

, Newman

, Verdile

, Sutherland

, Munch

, Musgrave

, Moussavi Nik

, Lardelli

(2016) Evidence for and against a pathogenic role of reduced gamma-secretase activity in familial Alzheimer’s disease. J Alzheimers Dis 52, 781–799.

19.

Wong

, Seelaar

, Melhem

, Rozemuller

AJM

, van Swieten

(2020) Genetic screening in early-onset Alzheimer’s disease identified three novel presenilin mutations. Neurobiol Aging 86, 201.e209–201.e214.

20.

Lee

, McBrayer

, Wolfe

, Haslett

, Kumar

, Sato

, Lie

, Mohan

, Coffey

, Kompella

, Mitchell

, Lloyd-Evans

, Nixon

(2015) Presenilin 1 maintains lysosomal Ca(2+) homeostasis via TRPML1 by regulating vATPase-mediated lysosome acidification. Cell Rep 12, 1430–1444.

21.

Heilig

, Gutti

, Tai

, Shen

, Kelleher

, 3rd (2013) Trans-dominant negative effects of pathogenic PSEN1 mutations on γ-secretase activity and Aβ production. J Neurosci 33, 11606–11617.

22.

Schroeter

, Ilagan

MXG

, Brunkan

, Hecimovic

, Li

Y-m

, Xu

, Lewis

, Saxena

, De Strooper

, Coonrod

, Tomita

, Iwatsubo

, Moore

, Goate

, Wolfe

, Shearman

, Kopan

(2003) A presenilin dimer at the core of the γ-secretase enzyme: Insights from parallel analysis of Notch 1 and APP proteolysis. Proc Natl Acad Sci U S A 100, 13075.

23.

Brautigam

, Moreno

, Steele

, Bogush

, Dickstein

, Kwok

JBJ

, Schofield

, Thinakaran

, Mathews

, Hof

, Gandy

, Ehrlich

(2015) Physiologically generated presenilin 1 lacking exon 8 fails to rescue brain PS1–/–phenotype and forms complexes with wildtype PS1 and nicastrin. Sci Rep 5, 17042.

24.

, Dang

, Yan

, Gong

, Wang

, Shi

(2013) Structure of a presenilin family intramembrane aspartate protease. Nature 493, 56–61.

25.

De Gasperi

, Sosa

MAG

, Dracheva

, Elder

(2010) Presenilin-1 regulates induction of hypoxia inducible factor-1α: Altered activation by a mutation associated with familial Alzheimer’s disease. Mol Neurodegener 5, 38–38.

26.

Villa

, Chiu

, Brandes

, Escorcia

, Villa

, Maguire

, Hu

, de Stanchina

, Simon

, Sisodia

, Scheinberg

, Li

(2014) Nontranscriptional role of Hif-1alpha in activation of gamma-secretase and notch signaling in breast cancer. Cell Rep 8, 1077–1092.

27.

Newman

, Nik

, Sutherland

, Hin

, Kim

, Halliday

, Jayadev

, Smith

, Laird

, Lucas

, Kittipassorn

, Peet

, Lardelli

(2020) Accelerated loss of hypoxia response in zebrafish with familial Alzheimer’s disease-like mutation of presenilin 1. Hum Mol Genet 29, 2379–2394.

28.

Lane

DJR

, Merlot

, Huang

MLH

, Bae

, Jansson

, Sahni

, Kalinowski

, Richardson

(2015) Cellular iron uptake, trafficking and metabolism: Key molecules and mechanisms and their roles in disease. Biochim Biophys Acta 1853, 1130–1144.

29.

Güner

, Lichtenthaler

(2020) The substrate repertoire of γ-secretase/presenilin. Semin Cell Dev Biol 105, 27–42.

30.

Thinakaran

, Harris

, Ratovitski

, Davenport

, Slunt

, Price

, Borchelt

, Sisodia

(1997) Evidence that levels of presenilins (PS1 and PS2) are coordinately regulated by competition for limiting cellular factors. J Biol Chem 272, 28415–28422.

31.

Berchtold

, Sabbagh

, Beach

, Kim

, Cribbs

, Cotman

(2014) Brain gene expression patterns differentiate mild cognitive impairment from normal aged and Alzheimer’s disease. Neurobiol Aging 35, 1961–1972.

32.

Villemagne

, Burnham

, Bourgeat

, Brown

, Ellis

, Salvado

, Szoeke

, Macaulay

, Martins

, Maruff

, Ames

, Rowe

, Masters

(2013) Amyloid β deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: A prospective cohort study. Lancet Neurol 12, 357–367.

33.

Foley

, Ammar

, Lee

, Mitchell

(2015) Systematic review of the relationship between amyloid-β levels and measures of transgenic mouse cognitive deficit in Alzheimer’s disease. J Alzheimers Dis 44, 787–795.

34.

Hargis

, Blalock

(2017) Transcriptional signatures of brain aging and Alzheimer’s disease: What are our rodent models telling us? Behav Brain Res 322, 311–328.

35.

Guo

, Fu

, Sopher

, Miller

, Ware

, Martin

, Mattson

(1999) Increased vulnerability of hippocampal neurons to excitotoxic necrosis in presenilin-1 mutant knock-in mice. Nat Med 5, 101–106.

36.

Kawasumi

, Chiba

, Yamada

, Miyamae-Kaneko

, Matsuoka

, Nakahara

, Tomita

, Iwatsubo

, Kato

, Aiso

, Nishimoto

, Kouyama

(2004) Targeted introduction of V642I mutation in amyloid precursor protein gene causes functional abnormality resembling early stage of Alzheimer’s disease in aged mice. Eur J Neurosci 19, 2826–2838.

37.

, Burgoyne

, Arnold

(2002) Sex differences in sex chromosome gene expression in mouse brain. Hum Mol Genet 11, 1409–1419.

38.

Bundy

, Vied

, Nowakowski

(2017) Sex differences in the molecular signature of the developing mouse hippocampus. BMC Genomics 18, 237.

39.

Barthelson

, Newman

, Lardelli

(2021) Comparative analysis of Alzheimer’s disease knock-in model brain transcriptomes implies changes to energy metabolism as a causative pathogenic stress. bioRxiv, 2021.2002.2016.431539.

40.

Barthelson

, Pederson

, Newman

, Lardelli

(2021) Brain transcriptome analysis of a protein-truncating mutation in sortilin-related receptor 1 associated With early-onset familial Alzheimer’s disease indicates early effects on mitochondrial and ribosome function. J Alzheimers Dis 79, 1105–1119.

41.

Barthelson

, Pederson

, Newman

, Jiang

, Lardelli

(2021) In-frame and frameshift mutations in zebrafish presenilin 2 affect different cellular functions in young adult brains. J Alzheimers Dis Rep 5, 395–404.

42.

Barthelson

, Pederson

, Newman

, Lardelli

(2020) Brain transcriptome analysis reveals subtle effects on mitochondrial function and iron homeostasis of mutations in the SORL1 gene implicated in early onset familial Alzheimer’s disease. Mol Brain 13, 142.

43.

Hin

, Newman

, Kaslin

, Douek

, Lumsden

, Nik

SHM

, Dong

, Zhou

X-F

, Mañucat-Tan

, Ludington

, Adelson

, Pederson

, Lardelli

(2020) Accelerated brain aging towards transcriptional inversion in a zebrafish model of the K115fs mutation of human PSEN2. PLoS One 15, e0227258.

44.

Jiang

, Pederson

, Newman

, Dong

, Barthelson

, Lardelli

(2020) Transcriptome analysis indicates dominant effects on ribosome and mitochondrial function of a premature termination codon mutation in the zebrafish gene psen2. PLoS One 15, e0232559.

45.

Dong

, Newman

, Pederson

, Barthelson

, Hin

, Lardelli

(2021) Transcriptome analyses of 7-day-old zebrafish larvae possessing a familial Alzheimer’s disease-like mutation in psen1 indicate effects on oxidative phosphorylation, ECM and MCM functions, and iron homeostasis. BMC Genomics 22, 211.

46.

Hin

, Newman

, Pederson

, Lardelli

(2020) Iron responsive element (IRE)-mediated responses to iron dyshomeostasis in Alzheimer’s disease. bioRxiv, 2020.2005.2001.071498.

47.

Newman

, Hin

, Pederson

, Lardelli

(2019) Brain transcriptome analysis of a familial Alzheimer’s disease-like mutation in the zebrafish presenilin 1 gene implies effects on energy production. Mol Brain 12, 43.

48.

Ishikawa

, Piao

Y-S

, Miyashita

, Kuwano

, Onodera

, Ohtake

, Suzuki

, Nishizawa

, Takahashi

(2005) A mutant PSEN1 causes dementia with Lewy bodies and variant Alzheimer’s disease. Ann Neurol 57, 429–434.

49.

Xia

, Watanabe

, Wu

, Lee

, Li

, Tsvetkov

, Bolshakov

, Shen

, Kelleher

, 3rd (2015) Presenilin-1 knockin mice reveal loss-of-function mechanism for familial Alzheimer’s disease. Neuron 85, 967–981.

50.

Kaneko

, Kakita

, Kasuga

, Nozaki

, Ishikawa

, Miyashita

, Kuwano

, Ito

, Iwatsubo

, Takahashi

, Nishizawa

, Onodera

, Sisodia

, Ikeuchi

(2007) Enhanced accumulation of phosphorylated α-synuclein and elevated β-amyloid 42/40 ratio caused by expression of the presenilin-1 ΔT440 mutant associated with familial Lewy body disease and variant Alzheimer’s disease. J Neurosci 27, 13092.

51.

Zhang

, Zhang

, Ge

(2018) An efficient platform for generating somatic point mutations with germline transmission in the zebrafish by CRISPR/Cas9-mediated gene editing. J Biol Chem 293, 6611–6622.

52.

Prykhozhij

, Fuller

, Steele

, Veinotte

, Razaghi

, Robitaille

, McMaster

, Shlien

, Malkin

, Berman

(2018) Optimized knock-in of point mutations in zebrafish using CRISPR/Cas9. Nucleic Acids Res 46, e102–e102.

53.

Jiang

, Newman

, Lardelli

(2018) The zebrafish orthologue of familial Alzheimer’s disease gene PRESENILIN 2 is required for normal adult melanotic skin pigmentation. PLoS One 13, e0206155.

54.

Di Tommaso

, Chatzou

, Floden

, Barja

, Palumbo

, Notredame

(2017) Nextflow enables reproducible computational workflows. Nat Biotechnol 35, 316–319.

55.

R Core Team (2019) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria.

56.

Chen

, Lun

, Smyth

(2016) From reads to genes to pathways: Differential expression analysis of RNA-Seq experiments using Rsubread and the edgeR quasi-likelihood pipeline [version 2; peer review: 5 approved]. F1000Res 5, 1438.

57.

Robinson

, Oshlack

(2010) A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol 11, R25.

58.

McCarthy

, Chen

, Smyth

(2012) Differential expression analysis of multifactor RNA-Seq experiments with respect to biological variation. Nucleic Acids Res 40, 4288–4297.

59.

Robinson

, McCarthy

, Smyth

(2009) edgeR: A Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26, 139–140.

60.

Young

, Wakefield

, Smyth

, Oshlack

(2010) Gene ontology analysis for RNA-seq: Accounting for selection bias. Genome Biol 11, R14.

61.

, Lim

, Vaillant

, Asselin-Labat

M-L

, Visvader

, Smyth

(2010) ROAST: Rotation gene set tests for complex microarray experiments. Bioinformatics 26, 2176–2182.

62.

, Smyth

(2012) Camera: A competitive gene set test accounting for inter-gene correlation. Nucleic Acids Res 40, e133–e133.

63.

Sergushichev

(2016) An algorithm for fast preranked gene set enrichment analysis using cumulative statistic calculation. bioRxiv, 060012.

64.

Subramanian

, Tamayo

, Mootha

, Mukherjee

, Ebert

, Gillette

, Paulovich

, Pomeroy

, Golub

, Lander

, Mesirov

(2005) Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A 102, 15545.

65.

Kanehisa

, Goto

(2000) KEGG: Kyoto encyclopedia of genes and genomes. Nucleic Acids Res 28, 27–30.

66.

Liberzon

(2014) A description of the Molecular Signatures Database (MSigDB) Web site. Methods Mol Biol 1150, 153–160.

67.

Dolgalev

(2020) msigdbr: MSigDB Gene Sets for Multiple Organisms in a Tidy Data Format. R package version 7.2.1. https://CRAN.R-project.org/package=msigdbr.

68.

Chen

(2011) Adaptive elastic-net sparse principal component analysis for pathway association testing. Stat Appl Genet Mol Bio 10, 48.

69.

Odom

, Ban

, Liu

, Sun

, Pico

, Zhang

, Wang

, Chen

(2019) pathwayPCA: An R package for integrative pathway analysis with modern PCA methodology and gene selection. bioRxiv, 615435.

70.

Liberzon

, Birger

, Thorvaldsdóttir

, Ghandi

, Mesirov Jill

, Tamayo

(2015) The molecular signatures database hallmark gene set collection. Cell Syst 1, 417–425.

71.

Kosik

, Muñoz

, Lopez

, Arcila

, García

, Madrigal

, Moreno

, Ríos Romenets

, Lopez

, Gutierrez

, Langbaum

, Cho

, Suliman

, Tariot

, Ho

, Reiman

, Lopera

(2015) Homozygosity of the autosomal dominant Alzheimer disease presenilin 1 E280A mutation. Neurology 84, 206–208.

72.

Parker

, Mozaffar

, Messmore

, Deignan

, Kimonis

, Ringman

(2019) Homozygosity for the A431E mutation in PSEN1 presenting with a relatively aggressive phenotype. Neurosci Lett 699, 195–198.

73.

Sundvik

, Chen

Y-C

, Panula

(2013) Presenilin1 regulates histamine neuron development and behavior in zebrafish, Danio rerio. J Neurosci 33, 1589.

74.

Tambini

, D’Adamio

(2020) Knock-in rats with homozygous PSEN1(L435F) Alzheimer mutation are viable and show selective γ-secretase activity loss causing low Aβ40/42 and high Aβ43. J Biol Chem 295, 7442–7451.

75.

Nickless

, Bailis

, You

(2017) Control of gene expression through the nonsense-mediated RNA decay pathway. Cell Biosci 7, 26.

76.

Alhamdoosh

, Law

, Tian

, Sheridan

, Ng

, Ritchie

(2017) Easy and efficient ensemble gene set testing with EGSEA. F1000Res 6, 2010–2010.

77.

Wilson

(2019) The harmonic mean-value for combining dependent tests. Proc Natl Acad Sci U S A 116, 1195.

78.

Pan

, Lin

, Tian

, Cheng

, Gridley

, Shen

, Kopan

(2004) gamma-secretase functions through Notch signaling to maintain skin appendages but is not required for their patterning or initial morphogenesis. Dev Cell 7, 731–743.

79.

Blanpain

, Lowry

, Pasolli

, Fuchs

(2006) Canonical notch signaling functions as a commitment switch in the epidermal lineage. Genes Dev 20, 3022–3035.

80.

Kopan

, Ilagan

(2009) The canonical Notch signaling pathway: Unfolding the activation mechanism. Cell 137, 216–233.

81.

Nowell

, Radtke

(2013) Cutaneous Notch signaling in health and disease. Cold Spring Harb Perspect 3, a017772.

82.

Nguyen

B-C

, Lefort

, Mandinova

, Antonini

, Devgan

, Della Gatta

, Koster

, Zhang

, Wang

, di Vignano

, Kitajewski

, Chiorino

, Roop

, Missero

, Dotto

(2006) Cross-regulation between Notch and p63 in keratinocyte commitment to differentiation. Genes Dev 20, 1028–1042.

83.

Dohda

, Maljukova

, Liu

, Heyman

, Grandér

, Brodin

, Sangfelt

, Lendahl

(2007) Notch signaling induces SKP2 expression and promotes reduction of p27Kip1 in T-cell acute lymphoblastic leukemia cell lines. Exp Cell Res 313, 3141–3152.

84.

Ryan

RJH

, Petrovic

, Rausch

, Zhou

, Lareau

, Kluk

, Christie

, Lee

, Tarjan

, Guo

, Donohue

LKH

, Gillespie

, Nardi

, Hochberg

, Blacklow

, Weinstock

, Faryabi

, Bernstein

, Aster

, Pear

(2017) A B cell regulome links notch to downstream oncogenic pathways in small B cell lymphomas. Cell Rep 21, 784–797.

85.

El-Brolosy

, Kontarakis

, Rossi

, Kuenne

, Günther

, Fukuda

, Kikhi

, Boezio

GLM

, Takacs

, Lai

S-L

, Fukuda

, Gerri

, Giraldez

, Stainier

DYR

(2019) Genetic compensation triggered by mutant mRNA degradation. Nature 568, 193–197.

86.

Rossi

, Kontarakis

, Gerri

, Nolte

, Hölper

, Krüger

, Stainier

DYR

(2015) Genetic compensation induced by deleterious mutations but not gene knockdowns. Nature 524, 230–233.

87.

Luo

, Pant

, Bhavnasi

, Blanchard

Jr. , Brouwer

(2017) Pathview Web: User friendly pathway visualization and data integration. Nucleic Acids Res 45, W501–W508.

88.

Ryman

, Acosta-Baena

, Aisen

, Bird

, Danek

, Fox

, Goate

, Frommelt

, Ghetti

, Langbaum

JBS

, Lopera

, Martins

, Masters

, Mayeux

, McDade

, Moreno

, Reiman

, Ringman

, Salloway

, Schofield

, Sperling

, Tariot

, Xiong

, Morris

, Bateman

(2014) Symptom onset in autosomal dominant Alzheimer disease: A systematic review and meta-analysis. Neurology 83, 253–260.

89.

Oksanen

, Petersen

, Naumenko

, Puttonen

, Lehtonen

, Gubert Olivé

, Shakirzyanova

, Leskelä

, Sarajärvi

, Viitanen

, Rinne

, Hiltunen

, Haapasalo

, Giniatullin

, Tavi

, Zhang

S-C

, Kanninen

, Hämäläinen

, Koistinaho

(2017) PSEN1 mutant iPSC-derived model reveals severe astrocyte pathology in Alzheimer’s disease. Stem Cell Rep 9, 1885–1897.

90.

Martín-Maestro

, Sproul

, Martinez

, Paquet

, Gerges

, Noggle

, Starkov

(2019) Autophagy induction by bexarotene promotes mitophagy in presenilin 1 familial Alzheimer’s disease iPSC-derived neural stem cells. Mol Neurobiol 56, 8220–8236.

91.

Birnbaum

, Wanner

, Gietl

, Saake

, Kündig

, Hock

, Nitsch

, Tackenberg

(2018) Oxidative stress and altered mitochondrial protein expression in the absence of amyloid-β and tau pathology in iPSC-derived neurons from sporadic Alzheimer’s disease patients. Stem Cell Res 27, 121–130.

92.

Kwart

, Gregg

, Scheckel

, Murphy

, Paquet

, Duffield

, Fak

, Olsen

, Darnell

, Tessier-Lavigne

(2019) A large panel of isogenic APP and PSEN1 mutant human iPSC neurons reveals shared endosomal abnormalities mediated by APP β-CTFs, not Aβ. Neuron 104, 256–270.e255.

93.

Volpato

, Smith

, Sandor

, Ried

, Baud

, Handel

, Newey

, Wessely

, Attar

, Whiteley

, Chintawar

, Verheyen

, Barta

, Lako

, Armstrong

, Muschet

, Artati

, Cusulin

, Christensen

, Patsch

, Sharma

, Nicod

, Brownjohn

, Stubbs

, Heywood

, Gissen

, De Filippis

, Janssen

, Reinhardt

, Adamski

, Royaux

, Peeters

, Terstappen

, Graf

, Livesey

, Akerman

, Mills

, Bowden

, Nicholson

, Webber

, Cader

, Lakics

(2018) Reproducibility of molecular phenotypes after long-term differentiation to human iPSC-derived neurons: A multi-site omics study. Stem Cell Rep 11, 897–911.

94.

Hunger

, Surovy

, Hassan

, Braathen

, Yawalkar

(2008) Toll-like receptor 2 is highly expressed in lesions of acne inversa and colocalizes with C-type lectin receptor. Br J Dermatol 158, 691–697.

95.

Otani

, Shichita

(2020) Cerebral sterile inflammation in neurodegenerative diseases. Inflamm Regen 40, 28.

96.

Zouboulis

, Benhadou

, Byrd

, Chandran

, Giamarellos-Bourboulis

, Fabbrocini

, Frew

, Fujita

, González-López

, Guillem

, Gulliver

WPF

, Hamzavi

, Hayran

, Hórvath

, Hüe

, Hunger

, Ingram

, Jemec

GBE

, Ju

, Kimball

, Kirby

, Konstantinou

, Lowes

, MacLeod

, Martorell

, Marzano

, Matusiak

, Nassif

, Nikiphorou

, Nikolakis

, Nogueira da Costa

, Okun

, Orenstein

LAV

, Pascual

, Paus

, Perin

, Prens

, Röhn

, Szegedi

, Szepietowski

, Tzellos

, Wang

, van der Zee

(2020) What causes hidradenitis suppurativa ?—15 years after. Exp Dermatol 29, 1154–1170.

97.

Institute of Medicine (US) Committee on Military Nutrition Research. The Role of Protein and Amino Acids in Sustaining and Enhancing Performance. Washington (DC): National Academies Press (US); 1999. 5, The Energy Costs of Protein Metabolism: Lean and Mean on Uncle Sam’s Team. Available from: https://www.ncbi.nlm.nih.gov/books/NBK224633/

98.

Bordi

, Darji

, Sato

, Mellén

, Berg

, Kumar

, Jiang

, Nixon

(2019) mTOR hyperactivation in Down syndrome underlies deficits in autophagy induction, autophagosome formation, and mitophagy. Cell Death Dis 10, 563.

99.

Jiang

, Sato

, Im

, Berg

, Bordi

, Darji

, Kumar

, Mohan

, Bandyopadhyay

, Diaz

, Cuervo

, Nixon

(2019) Lysosomal dysfunction in Down syndrome is APP-dependent and mediated by APP-βCTF (C99). J Neurosci 39, 5255.

100.

Lee

, Yu

, Kumar

, Lee

, Mohan

, Peterhoff

, Wolfe

, Martinez-Vicente

, Massey

, Sovak

, Uchiyama

, Westaway

, Cuervo

, Nixon

(2010) Lysosomal proteolysis and autophagy require presenilin 1 and are disrupted by Alzheimer-related PS1 mutations. Cell 141, 1146–1158.

101.

Linke

, Fritsch

, Sukhbaatar

, Hengstschläger

, Weichhart

(2017) mTORC1 and mTORC2 as regulators of cell metabolism in immunity. FEBS Lett 591, 3089–3103.

102.

Inpanathan

, Botelho

(2019) The lysosome signaling platform: Adapting with the times. Front Cell Dev Biol 7, 113.

103.

Bond

(2016) Regulation of mTORC1 by growth factors, energy status, amino acids and mechanical stimuli at a glance. J Int Soc Sports Nutr 13, 8.

104.

Loewith

, Hall

(2011) Target of rapamycin (TOR) in nutrient signaling and growth control. Genetics 189, 1177–1201.

105.

Lempiäinen

, Shore

(2009) Growth control and ribosome biogenesis. Curr Opin Cell Biol 21, 855–863.

106.

Newman

, Wilson

, Verdile

, Lim

, Khan

, Moussavi Nik

, Pursglove

, Chapman

, Martins

, Lardelli

(2014) Differential, dominant activation and inhibition of Notch signalling and APP cleavage by truncations of PSEN1 in human disease. Hum Mol Genet 23, 602–617.

107.

Kelleher

, Shen

(2017) Presenilin-1 mutations and Alzheimer’s disease. Proc Natl Acad Sci U S A 114, 629.

108.

Zhou

, Yang

, Shi

(2017) Dominant negative effect of the loss-of-function γ-secretase mutants on the wild-type enzyme through heterooligomerization. Proc Natl Acad Sci U S A 114, 12731–12736.

109.

Chávez-Gutiérrez

, Bammens

, Benilova

, Vandersteen

, Benurwar

, Borgers

, Lismont

, Zhou

, Van Cleynenbreugel

, Esselmann

, Wiltfang

, Serneels

, Karran

, Gijsen

, Schymkowitz

, Rousseau

, Broersen

, De Strooper

(2012) The mechanism of γ-Secretase dysfunction in familial Alzheimer disease. EMBO J 31, 2261–2274.

110.

Zhang

, Cai

, Wu

, Bozorgmehr

, Wang

, Zhang

, Huang

, Guo

, Shen

, Rankin

, Tang

, Song

(2020) A presenilin-1 mutation causes Alzheimer disease without affecting Notch signaling. Mol Psychiatry 25, 603–613.

111.

Deatherage

, Lu

, Kroncke

, Ma

, Smith

, Voehler

, McFeeters

, Sanders

(2017) Structural and biochemical differences between the Notch and the amyloid precursor protein transmembrane domains. Sci Adv 3, e1602794.

112.

Bukhari

, Glotzbach

, Kolbe

, Leonhardt

, Loosse

, Müller

(2017) Small things matter: Implications of APP intracellular domain AICD nuclear signaling in the progression and pathogenesis of Alzheimer’s disease. Prog Neurobiol 156, 189–213.

113.

Musa

, Lehrach

, Russo

(2001) Distinct expression patterns of two zebrafish homologues of the human APP gene during embryonic development. Dev Genes Evol 211, 563–567.

114.

Moore

, Gillentine

, Botezatu

, Wilson

, Benson

, Langeland

(2014) Asynchronous evolutionary origins of Aβ and BACE1. Mol Biol Evol 31, 696–702.