Skip to main content
Download PDF

Alternative cleavage and polyadenylation of genes associated with protein turnover and mitochondrial function are deregulated in Parkinson’s, Alzheimer’s and ALS disease

BMC Medical Genomics volume 12, Article number: 60 (2019) Cite this article

Abstract

Background

Transcriptome wide changes have been assessed extensively during the progression of neurodegenerative diseases. Alternative polyadenylation (APA) occurs in over 70% of human protein coding genes and it has recently been recognised as a critical regulator of gene expression during disease. However, the effect of APA in the context of neurodegenerative diseases, to date, has not been widely investigated. Dynamic Analysis of Alternative Polyadenylation from RNA-seq (DaPars) is a method by Xia and colleagues [Nat Commun. 5:5274, 2014] to investigate APA using standard RNA-seq data. Here, we employed this method to interrogate APA using publicly available RNA-seq data from Alzheimer’s disease (AD), Parkinson’s disease (PD) and Amyotrophic Lateral Sclerosis (ALS) patients and matched healthy individuals.

Results

For all three diseases, we found that APA profile changes were limited to a relative small number of genes suggesting that APA is not globally deregulated in neurodegenerative disease. However, for each disease phenotype we identified a subgroup of genes that showed disease-specific deregulation of APA. Whilst the affected genes differ between the RNA-seq datasets, in each cohort we identified an overrepresentation of genes that are associated with protein turnover pathways and mitochondrial function.

Conclusions

Our findings, while drawn from a relatively small sample size, suggest that deregulation of APA may play a significant role in neurodegeneration by altering the expression of genes including UBR1 and OGDHL in AD, LONP1 in PD and UCHL1 in ALS. This report thus provides important novel insights into how APA can shape neurodegenerative disease characteristic transcriptomes.

Peer Review reports

Background

Neurodegeneration is defined as the progressive loss of neurons in the central nervous system (CNS) and the brain. While neurodegenerative diseases can be inherited, these familial cases account for fewer than 10% for the three most common diseases: Alzheimer’s Disease (AD), Parkinson’s Disease (PD) and Amyotrophic Lateral Sclerosis (ALS). Interestingly, whilst the cause of disease in sporadic patients is largely unknown, both the sporadic (non-inherited) and familial cases of a particular neurodegenerative disease often exhibit the same symptoms.

Despite AD, PD and ALS exhibiting very different symptoms and affecting different parts of the CNS, many similarities have been identified at the cellular level, that contribute to neuronal loss. This includes compromised mitochondrial function and protein aggregation to form plaques or inclusion bodies that impair neuronal function [1]. Protein aggregates, often consisting of misfolded proteins, are usually degraded via the ubiquitin proteasome pathway (UPP) or through the autophagy pathway. Deregulation of these pathways is associated with neurodegeneration. However, it is unclear if deregulation of protein degradation pathways is a cause or consequence of neurodegeneration [2].

Transcripts encoding all metazoan protein coding genes, apart from replication dependant histone genes, are uniformly processed at the 3’end in a process known as cleavage and polyadenylation. Cleavage occurs at the poly(A) site, after recognition of the poly(A) signals in the pre-mRNA that are located in the 3′ untranslated regions (UTR) and the 3’flanking regions. Over 70% of mammalian genes undergo alternative polyadenylation (APA), where alternative poly(A) sites are utilised [3]. The regulatory powers of APA reside in the production of mRNA isoforms that differ in the lengths of their 3’UTRs. 3’UTRs harbour a plethora of regulatory elements that provide targets for RNA binding proteins or miRNAs which in turn can mediate stability, translatability or localisation of the respective transcript isoforms [4, 5]. Therefore, utilisation of alternative poly(A) sites through APA can post-transcriptionally regulate gene expression.

The relative frequencies between isoforms with long or short UTR length present in a cell represents the cellular APA profile. If cells enter a particular state or if they receive specific cues, the profiles can be changed to ensure adapted and adequate expression of the affected genes. Mechanistically, this can be achieved by favouring one poly(A) site over another at the point of co-transcriptional cleavage, known as active APA. The relative frequencies between the long and short isoforms in a cell can also be changed at the post-transcriptional level by selectively destabilising one isoform over another in the cytoplasm, known as passive APA [4].

Alterations in the transcriptome of patients with neurodegenerative diseases has been investigated with RNA-seq to assess changes in gene expression [6], splicing [7], and changes in miRNAs [8, 9] and lncRNAs expression [10]. However, there have been few studies that globally assess changes of APA profiles in neurodegenerative tissues or cells. Individual gene analysis has identified APA changes in genes associated with neurodegeneration, for example MAPT in AD [11, 12], SNCA in PD [13] and TARDBP in ALS [14]. However, to date transcriptome-wide APA profile changes have not been assessed for both AD and PD, and there has been only one such study focussing on ALS [15].

APA profiles are generally established using specific protocols that select and sequence only the very 3’ends of mRNAs (5). Recently a bioinformatics pipeline has been developed that enables APA profiles analysis from existing standard RNA-seq data sets [16]. This method, Dynamic Analysis of Alternative Polyadenylation from RNA-seq (DaPars), thus enables de novo identification and analysis of dynamic poly(A) site changes from any newly generated or deposited RNA-Seq data set.

The assessment of APA changes in the affected regions of neurodegenerative disease patients through wet-lab experiments is complicated by the scarcity of post-mortem patient RNA. However, using DaPars, APA can be investigated from standard RNA-seq data from existing studies. In this report, we use DaPars to compare APA profiles from sequenced RNA samples isolated from AD, PD and ALS patient with their respective control samples at a global scale. Using this approach, we identified individual genes that have previously been associated with the respective diseases and undergo disease specific APA changes. In addition, we find commonality between the diseases by showing that the genes which undergo disease-specific APA profile changes are associated with mitochondrial function and protein catabolism. As these processes are directly linked to neurodegeneration, our findings suggest that altered APA profiles may be a significant contributor to establish a transcriptome characteristic for a neurodegenerative state.

Results

APA in Alzheimer’s disease

Dysregulated RNA processing in AD has been identified in isolated cases, such as the extracellular aggregation of U1snRNP, a factor associated with regulation of splicing and polyadenylation, in AD brains [17]. Although certain genes associated with AD such as COX-2 [18], MAPT [11, 12] and APP [19] utilise different 3’UTRs, there have been no genome-wide studies investigating the role of APA in AD.

To address this issue, we used an RNA-seq dataset from a study on transcriptomics of 4 Late Onset AD (LOAD) patients and 4 control individuals [10] and the DaPars pipeline to identify differences of UTR length in AD compared to healthy individuals. The RNA samples used had been deep sequenced on the Illumina platform to generate single end raw FASTQ files that were deposited on the Sequence Read Archive (GEO Accession number GSE24565). FASTQ files were groomed and aligned to the hg19 genome using TopHat using the Galaxy platform (https://usegalaxy.org/). The aligned BAM files were then converted to Bedgraph files for input into the DaPars script. Between 164.8–188.6 million reads per sample were subjected to DaPars analysis.

The RNA from this dataset originated from the hippocampus, the region of the brain important for memory, which is one of the first damaged regions in AD. As the patient and control samples were not accurately age-matched, each of the 4 LOAD patients were compared to each of four different control samples yielding a total of 16 comparisons. Between 7223 and 8419 APA events were identified and of those 0.5–3.3% of genes did undergo statistically significant APA changes when the LOAD and control samples were compared (Fig. 1a). Whilst we identified significant AD specific APA changes, no trend toward either 3’UTR shortening or lengthening was evident in the patient RNA samples.

Fig. 1
figure 1

UBR1 and OGDHL show significant APA changes in LOAD. a Bar chart showing mean percentage of significant 3’UTR lengthening (red bars) and 3’UTR shortening (green bars) events between each LOAD patient and control hippocampus derived RNA samples through DaPars analysis (PDUI > 0.25, Fisher’s exact Test, p < 0.05). The error bars indicate standard deviation from comparisons of 4 control samples. The mean number (n) of identified APA events for each LOAD patient are indicated underneath the chart. b Table showing the enrichment of biological processes identified with Gene Ontology Analysis using the PANTHER platform in genes exhibiting 3’UTR lengthening (upper three rows) and 3’UTR shortening (bottom three rows). c List of ten gene examples that show 3’UTR lengthening (red) or 3’UTR shortening (green) in LOAD samples compared to two or more control samples. d Genome browser view of UBR1 as an example gene that shows 3’UTR shortening in LOAD (blue tracks) compared to control (CTRL, green tracks). Annotated miRNA target sites in the UBR1 3’UTR are indicated by yellow boxes below the gene structure. e Genome browser view of OGDHL showing 3’UTR lengthening in LOAD (blue tracks) compared to control (CTRL, green tracks). In d & e, the proximal and distal poly (A) sites are shown as orange boxes and the length of the alternative UTR (aUTR) is indicated below the gene structure. The length of the genome browser windows shown is indicated above in kilo bases (kb) between the two arrows

Full size image

We next interrogated the genes that showed altered UTR lengths in AD by subjecting them to a Gene Ontology (GO) analysis using the PANTHER Platform. This showed that the genes exhibiting 3’UTR lengthening in AD compared to healthy controls were enriched in mitochondrial pathways (Fig. 1b), while an enrichment in genes associated with protein catabolic processes was identified in those genes that had shorter 3’UTRs in AD derived samples. Interestingly, both biological processes identified by the GO analysis are closely linked to neurodegeneration. Therefore, changes in gene regulation through APA of these pathways may be a contributing factor to AD pathology.

We next isolated APA events that occurred in 2 or more LOAD patient samples compared to controls. Using this approach, we identified 10 genes that showed 3’UTR lengthening and 10 genes that showed 3’UTR shortening in 2 or more LOAD samples (Fig. 1c and Additional file 1: Figure S1). The ubiquitin ligase UBR1, is an example that showed 3’UTR shortening in AD in two patients versus control comparisons (Fig. 1d). Whilst UBR1, to the best of our knowledge, has not previously been directly implicated in AD, it is an E3 ubiquitin ligase of the N-end rule pathway [20] and mutations in the gene are linked to the Johanson-Blizzard syndrome which is characterised by a variety of features including varied degrees of cognitive impairment [21]. In addition, as protein turnover and degradation in general is associated with AD, altered expression of genes that are part of the UPP pathway such as UBR1, could contribute to dysregulation of protein turnover. Interestingly, the alternative long 3’UTR of the UBR1 gene contains target sites for miR-26 and miR-128-3. MiR-26 has been shown to be upregulated in the temporal cortex of AD patients [22], while miR-128-3 was upregulated in the hippocampus of AD patients [23]. The upregulation of miR-26 and 128 in AD patients could result in the destabilisation of the longer UBR1 UTR isoform and may thus provide a mechanistic explanation for the observed overrepresentation of transcripts with short 3’UTRs in AD patients.

Of the ten genes that showed 3’UTR lengthening in more than one LOAD-control comparison, OGDHL (Fig. 1e) is the most notable example. OGDHL encodes a brain- specific isoenzyme for oxoglutarate dehydrogenase that functions in the mitochondrial Krebs cycle. Downregulation of this protein, which may be aided by the UTR lengthening, has been observed in an AD mouse model [24] where its decreased expression can contribute to reduced ATP production.

We conclude from these results that whilst there is little overlap between APA events in LOAD and control comparisons, we nevertheless identified reoccurring APA events in a small number of genes that have previously been linked to neurodegeneration.

Although AD typically begins through degradation of the hippocampus in the temporal lobe region of the brain, other brain regions can be affected in AD [7]. We thus expanded our analysis of APA in AD by including an additional RNA-seq data set generated using RNA isolated from the frontal and temporal lobes of AD patients and control samples [7]. Notably, the read depth of this RNA-seq dataset was 10-fold lower than that of the previous dataset with just 13–15 million reads per sample. The samples were mapped and treated as described above. The samples were subsequently interrogated for significant UTR length changes by pumping them through the DaPars pipeline using the same parameters as described above.

The comparison of APA using RNA extracted from the Frontal Lobe identified 701 genes that showed changes in UTR length in AD compared to the control samples and 7.0% of these were statistically significant. Comparisons of RNA from the temporal Lobe in this data set identified 271 APA events of which 14.7% were statistically significant. Unlike in the previous data set, the genes that did show significant APA, tended to lengthen their 3’UTRs in both the temporal and frontal lobes (Fig. 2a). However, the overlap of genes showing significant APA in both brain regions was limited to 14 genes (Fig. 2b). This indicates that APA is differentially regulated between the temporal and frontal lobes in AD.

Fig. 2
figure 2

APA Regulation in Different Regions of the Brain in AD. a Significant 3’UTR lengthening (* = p < 0.01, Fisher’s exact Test) is observed in the frontal lobe and temporal lobe regions of the brain using DaPars analysis (PDUI > 0.25, Fisher’s exact Test < 0.05). Red and green bars indicate number of identified UTR-APA lengthening and shortening events respectively. The number of identified APA events in the comparison are indicated by n below the graph. b Venn diagram highlighting that different genes undergo APA in different regions of the brain and 14 genes show the same movements in both frontal (purple) and temporal lobe (blue). c Table to highlight the biological processes that were identified through Gene Ontology Analysis using the PANTHER platform in the genes that showed 3’UTR lengthening in either cohort. d Genome browser view of the 3’UTR lengthening in BIN1 in AD (AD, blue tracks) compared to control (C, green tracks). e. 3’UTR shortening in VAMP2 in AD (blue tracks) compared to control (C, green tracks). In d & e, the proximal and distal PAS’s are shown as orange boxes and the length of the alternative UTR (aUTR) is indicated below the gene structure. The length of the genome browser windows shown is indicated above in kilo bases (kb) between the two arrows

Full size image

To characterise the cohort of genes that show AD patient-specific UTR lengthening further, we subjected them to a GO analysis using the PANTHER platform, which showed an enrichment of synaptic plasticity regulation, modulation of synaptic transmission and regulation of neuron apoptotic processes (Fig. 2c). This suggests that APA mediated regulation of the transcriptome could affect synaptic activity, which may be a contributing factor for altered neuronal function as observed in AD. A notable gene that undergoes lengthening in both the frontal and temporal lobes of AD patients compared to the controls is BIN1. The BIN1 gene, whose protein is involved in synaptic vesicle endocytosis, showed 3’UTR lengthening in the temporal and frontal lobe samples but not in the whole brain sample, qualifying it as a localised APA change (Fig. 2d). Importantly, this gene has been identified as a risk locus for AD [25] and increased expression has been linked to modulating tau pathology in LOAD [26]. The lengthening of the 3’UTR may contribute to increased expression of BIN1 protein and could so be a contributing factor for AD.

We identified six genes that showed 3’UTR shortening, and only one of these, VAMP2, showed APA in both the frontal and temporal lobe region of the brain (Fig. 2e). However, VAMP2 encodes a protein that is involved in neurotransmitter release during the fusion of synaptic vesicles to the pre-synaptic membrane. Most interestingly, decreased protein expression of VAMP2 is seen during progression of AD [27], which may contribute to AD pathology.

These results show that different genes exhibit significant APA in different regions of the AD brain, but all show a significant trend to 3’UTR lengthening. Altered protein expression of BIN1 and VAMP2 have been associated with AD [26, 27]. In some patients, this may be at least partially caused by APA. Therefore, our analysis suggests a novel mechanism of BIN1 and VAMP2 gene regulation that may be disrupted in AD patients.

APA in Parkinson’s disease

Parkinson’s Disease (PD) is characterised by bradykinesia, tremors and stiffness of movement, due to a loss of dopaminergic neurons in the substantia nigra region of the brain. These neurons use dopamine as their neurotransmitter, which has reduced levels in PD. Further, genes involved in dopamine metabolism show altered expression in PD [28]. PD is also characterised by the accumulation of cytoplasmic protein aggregates, mostly composed of insoluble α-synuclein forming structures called Lewy bodies [29]. Although 90% of PD cases are sporadic, much of the research has been focussed on familial cases of the disease. Sporadic PD (S-PD) patients exhibit the same symptoms, and in some cases, mutations have been seen in the same causative genes as in familial PD (F-PD). These genes have been studied extensively (Reviewed in [28, 30]).

To date, only one study has assessed the role of APA in PD. Rhinn and colleagues showed an increased usage of the distal poly(A) site of SNCA, a gene crucial to PD pathogenesis, in the brains of PD patients. These longer isoforms produced proteins that were localised to the mitochondria and were more likely to aggregate suggesting a critical role of APA in α-synuclein regulation [13].

To assess APA changes in S-PD patients, an RNA-seq dataset from three S-PD and control patients was subjected to DaPars analysis. The RNA had been isolated from midbrain dopaminergic neurons derived from S-PD patient iPSCs. These dopaminergic neurons showed signs of oxidative stress and altered neuronal activity, as observed in the PD disease state. The paired-end FASTQ files were groomed and mapped to hg19 using TopHat on the Galaxy platform, resulting in 49.4–54.8 million reads per sample.

APA analysis was conducted on combined biological replicates of each of the three S-PD patient with three control RNA samples from derived dopaminergic neurons. The biological replicate bedgraph files were combined and DaPars analysis was conducted on the mean of the reads from the two replicates. The UTR length changes in each patient sample was compared with each control sample and the mean of the total number of APA events per S-PD vs control comparison was calculated. Less than 0.2% of genes were identified to show significant APA changes between patient and control samples (Fig. 3a). This suggests that APA is unlikely to be a major contributor to transcriptome changes that have been observed in S-PD [31]. Furthermore, we observed great variation between the number of APA events identified between the different comparisons as highlighted by the standard deviation and there was very little overlap between the genes exhibiting APA between different comparisons (Fig. 3b). Taken together, these results suggest that few APA changes can be specifically correlated to S-PD.

Fig. 3
figure 3

APA Regulation in sPD. a Bar chart showing the mean percentage of 3’UTR lengthening (red) and shortening (green) APA events in SP-D compared to each of the three control samples identified with DaPars analysis (PDUI > 0.25, Fisher’s exact Test p < 0.05). Error bars show standard deviation between the 3 comparisons. The mean number of APA events identified for each comparison is indicated by n below the chart. The percent of statistically significant APA events are labelled. b Venn diagrams showing the overlaps between the identified APA events for each of the S-PD vs control samples. Each shade of green represents a comparison to a different control sample C1, C2 or C3. c Genome browser view depicting the 3’UTR lengthening in CHURC1 gene that shows lengthening in each S-PD sample (blue tracks) compared to control samples (C, green tracks). The proximal and distal poly(A) sites are shown as orange boxes and the length of the alternative UTR (aUTR) is indicated below the gene structure. The length of the genome browser windows shown is indicated above in kilo bases (kb) between the two arrows

Full size image

An exception to the above statement is the CHURC1 gene which showed 3’UTR lengthening in each of the S-PD patients compared to at least 2 of the control samples (Fig. 3c). Interestingly, CHURC1 encodes a transcriptional activator that is involved in neuronal development, but it has, to the best of our knowledge, so far not been associated with neurodegeneration.

Transcriptomic changes between familial (F-PD) and sporadic PD (S-PD) cases have also been assessed through RNA-seq [32]. In this study, induced pluripotent cells (iPSCs) were derived and differentiated to midbrain dopaminergic neurons from fibroblasts isolated from a familial PD patient with the N370S mutation in the GBA gene, an S-PD patient and two healthy controls. We used DaPars on this RNA-seq dataset to evaluate potential UTR length changes in PD in the familial and sporadic patients compared to the healthy individuals. Biological replicates each having between 25.8–37.2 million reads, were subjected to DaPars analysis as described above.

There was no clear trend to toward either 3’UTR shortening or lengthening apparent and overall fewer than 0.6% of APA events proved significant (Fig. 4a), again indicating that global APA changes do not occur in PD. However, GO analysis of all of the genes that show 3’UTR shortening presented an enrichment in mitochondrial organisation (Fig. 4b). Altered regulation of these genes through shorter 3’UTRs may have a role in mediating mitochondrial dysfunction, which is an established phenotype in PD. Genes showing 3’UTR lengthening were associated with ‘macromolecular complex subunit organisation’, including genes associated with protein turnover (Fig. 4b). Aggregation of misfolded proteins is an important hallmark of PD pathogenesis, therefore altered regulation of these genes through 3’UTR lengthening may contribute to PD pathology.

Fig. 4
figure 4

APA Regulation in F-PD compared to S-PD. a Bar chart to show the percentage of 3’UTR lengthening (red) and 3’UTR shortening (green) events in Familial-PD (F-PD) and Sporadic-PD (S-PD) compared to the two healthy controls (C1, C2) identified by DaPars analysis (PDUI > 0.25, Fisher’s exact Test p < 0.05). The total number of identified APA events are indicated by n beneath the chart. b A table showing the biological processes identified through GO analysis on the PANTHER platform for those genes that undergo significant APA either in S-PD an or F-PD. c Genome browser view of LONP1 for which 3’UTR shortening was observed in S-PD (blue track) but not F-PD (purple track) when compared to the control (green track). d Genome browser view of the 3’UTR lengthening observed in NELFA in F-PD (F-PD, purple track) compared to control (C, green track) but not in S-PD (S-PD, blue track). c & d, the proximal and distal poly(A) sites are shown as orange boxes and the length of the alternative UTR (aUTR) is indicated below the gene structure. The length of the genome browser windows shown is indicated above in kilo bases (kb) between the two arrows

Full size image

A prominent candidate gene showing 3’UTR shortening in S-PD, but not F-PD, is LONP1 (Fig. 4c), which encodes a mitochondrial matrix protease that mediates degradation of misfolded or oxidatively damaged proteins. Given the association of misfolded protein aggregation in PD, including in the mitochondria [33], alterations in the regulation of the LONP1 could affect mitochondrial function. In particular, LONP1 knockdown has been shown to cause an increase in PINK1 accumulation, which increased mitophagy [34]. PINK1 is associated with F-PD along with Parkin, which together regulate the removal of dysfunctional mitochondria. Consequently, alterations in LONP1 protein levels in the mitochondrial matrix, may contribute to PD pathogenesis. The mechanisms by which LONP1 3’UTR length is dysregulated in PD remains unclear and no RBP sites or regulatory elements in the alternative UTR (aUTR) that could influence transcript stability, protein output or localization were not identified by scanning these sequences.

An example of F-PD specific 3’UTR lengthening was seen in NELFA (Fig. 4d). NELFA encodes a component of the negative elongation factor complex, which has a role in transcriptional pausing and regulating Pol II mediated transcription. A target site for miR-133 family of miRNAs is found in the aUTR which may mediate the isoform expression in the cytoplasm. miR-133b was shown to be deficient in the midbrain of PD patients [35], which could explain this 3’UTR lengthening event.

These results indicate APA is not globally deregulated in either S-PD or F-PD compared to their respective controls. However, the genes that do undergo APA in S-PD or F-PD are associated with mitochondrial function and protein degradation suggesting that a potential role of APA in PD pathology.

APA in ALS

Amyotrophic Lateral Sclerosis (ALS), also known as Motor Neuron Disease, is a debilitating disease which results in death of upper and lower motor neurons which control voluntary muscles, leading to muscle stiffness, weakness and eventually loss of voluntary movement. Like AD and PD over 90% of cases are sporadic, while approximately 10% are related to inherited genetic mutations. However both familial (fALS) and sporadic (sALS) cases exhibit similar neuropathologies both fALS and sALS patient samples have shown similar dysregulation of miRNA and gene expression [9].

ALS has been closely linked to defects in RNA metabolism. Mutations in the nuclear localisation sequence (NLS) of FUS and TARDBP, two genes that encode RBPs associated with RNA processing, are linked to fALS. These mutant proteins subsequently aggregate in the cytoplasm forming inclusion bodies which contribute to loss of motor neuron function [36]. FUS, which has a role in alternative splicing [37], can also interact with the Pol II CTD to regulate phosphorylation of Ser2 [38] and can control mRNA turnover [39]. Furthermore, FUS knockdown altered mRNA expression of genes associated with mitochondrial function and increased proximal poly(A) site usage [38]. More recently, FUS was shown to affect poly(A) site usage depending on the proximity of its binding site to the poly(A) site [40]. FUS therefore functions in many aspects of mRNA regulation. TDP-43, which is encoded by the TARDBP gene, was shown to autoregulate the splicing and poly(A) site selection of its own transcript [14], although its role in global poly(A) site choice has not been investigated. Thus, the regulation of RNA metabolism by these RBPs is crucial in ALS. Both FUS and TDP-43 regulate alternative splicing of genes involved in neuronal development and neurodegeneration, but there is not a significant overlap in the genes they regulate [41].

Changes in the 3’UTR length in both sALS and a C9orf27 mutant case of ALS have been assessed globally through RNA-seq [15]. APA was shown to be altered differentially in these two patients compared to the healthy control, and different regions of the brain showed widespread different poly(A) site usage [15]. While these regions of the brain are important in ALS, it is the motor neurons whose function is mainly compromised. Here, APA changes in ALS were investigated with DaPars on an RNA-seq dataset where RNA had been isolated from motor neurons of the lumbar spinal cord which had not been fully degenerated from ALS patients [42].

To assess whether APA is deregulated in ALS we considered deposited RNA-seq data from 13 sALS and 9 control individuals for DaPars analysis. Samples with fewer than 20 million reads, which fall below the 50 million reads threshold requirements for DaPars were omitted, yielding a final 10 sALS and 8 control patients. Each of the selected sALS patient samples were compared to each of the controls, resulting in 80 comparisons. For each patient and control comparison, on average, 2.7% of genes showed significant APA, with twice as many 3’UTR lengthening than shortening events (Fig. 5a). GO analysis for UTR lengthened genes identified an enrichment for genes involved in negative regulation of neuron projection development as well as genes associated with cytoskeleton intracellular transport (Fig. 5b). Therefore, although global changes are not seen in sALS, APA may be important in the regulation of transcripts encoding proteins that contribute to neuropathology.

Fig. 5
figure 5

APA Regulation in Motor Neurons in sALS. a Bar chart showing the percentage of genes that showed significant UTR changes through DaPars analysis (PDUI > 0.25, Fisher’s exact test, p < 0.05). In total, 1.8% of genes showed 3’UTR lengthening (red), while fewer than 1% showed 3’UTR shortening (green) in all comparisons between 10 sALS and 8 control samples, showing a significant trend to 3’UTR lengthening (* = p < 0.05, Fisher’s exact Test). The data labels indicate number of APA events identified among all cross comparisons. b Table highlighting the biological processes that are enriched in the genes showing UTR lengthening identified with GO analysis using the PANTHER platform. The cohort of genes showing UTR shortening were not enriched in any biological processes. c Table of 22 genes showing 3’UTR lengthening or shortening events and occur in at least 3 different sALS vs control comparisons. d Genome browser view of the 3’UTR lengthening seen in UCHL1 gene in sALS (grey tracks) compared to control (green tracks). The proximal and distal poly(A) sites are shown as orange boxes the length of the genome browser window shown is indicated above in kilo bases (kb) between the two arrows. e Schematic indicating the position of RBPs found uniquely in the aUTR of UCHL1 are indicated by aqua boxes. The length of the alternative UTR (aUTR) is shown between the arrows below the gene structure

Full size image

To investigate the key changing genes further, 3’UTR shortening or lengthening events occurring in at least 3 different sALS vs control comparisons were selected, yielding 22 genes (Fig. 5c). Of this cohort only one gene has previously been linked to neurodegenerative diseases, UCHL1. UCHL1, which encodes an abundant neuron specific enzyme associated with the UPP, shows 3’UTR lengthening in 4 sALS compared to 4 different control samples (Fig. 5d). UCHL1 is a regulator of ubiquitin turnover as it has both hydrolase activity to remove ubiquitin [43] and ubiquitin ligase activity [44]. Its crucial role in maintaining healthy motor neurons was demonstrated in a mouse knockout model where loss of UCHL1 increased ER stress causing motor neuron degeneration [45]. Furthermore, reduced UCHL1 protein levels [46], and loss of its activity have been reported in PD and AD [47].

Analysis of the UCHL1 UTR using a CLIP database (http://lulab.life.tsinghua.edu.cn/clipdb/) identified 21 RBPs that can interact with this UTR, but only 3 of these, CstF-64, WDR33 and TARDBP, uniquely interact with the aUTR (Fig. 5e). CstF-64 promotes usage of a proximal poly(A) site, while WDR33, though crucial for cleavage and polyadenylation, does not have an established role in APA. Mutations in TARDBP, which mediates splicing, are linked to fALS cases but there is no evidence that it influences APA in genes apart from its own transcript [14]. Further, no miRNAs have been identified to target the UCHL1 UTR. It is therefore unclear how the 3’UTR lengthening event in UCHL1 may arise in sALS, yet dysregulation of this gene by APA may affect its protein output or localisation, which in turn could contribute to pathology in sALS.

Overall, few significant APA events were identified in sALS patients compared to the frontal cortex and cerebellum of the brain as were reported by Prudencio and colleagues. However, 3’UTR lengthening was identified in a number of genes including UCHL1, which is important in the UPP pathway and is associated with neurodegeneration.

Discussion

The usage of alternative splicing and alternative promoter usage has been implicated in the brain to achieve the complexity required [7]. APA has also been investigated, with UTRs globally appearing to be lengthened in the brain compared to other tissues [48]. However, the possibility of changes in poly(A) site usage in neurodegeneration has not been extensively investigated at a global level.

A number of methods and pipelines have been developed to analyse APA that rely on the sequencing of the very 3’end of mRNAs [49]. As these approaches generally require tailor- made sequencing libraries, they are less suitable for the analysis of standard RNAseq data. To address this shortcoming several approaches have been developed [50,51,52] including DaPars [16]. DaPars is an established method and uses a sophisticated algorithm that uses a regression model for the identification of the distal and predictive proximal poly(A) sites in genes from standard RNA seq data sets. Importantly, this bioinformatics method has been successfully used to investigate APA from RNAseq data associated with a number of diseases [15, 16, 53, 54].

Here, RNA-seq datasets from AD, PD, and ALS were investigated with DaPars to assess alterations in APA (Additional file 1: Table S1). While widespread changes in APA were not seen in any of the diseases, examples of genes which may contribute to the disease state were identified. However, we cannot rule out that the lack of identifying prevalent disease associated changes in APA may have been limited by the small sample sizes used in this analysis. Furthermore, different APA profiles were observed in different regions of the brain, and whilst distinctly different genes were subjected to APA in the three different neurodegenerative diseases they encoded proteins functionally associated with mitochondrial function band protein catabolism.

APA in AD

Assessment of APA in the two AD datasets investigated identified different cohorts of genes regulated by APA, however in each case, 3’UTR length varied in genes encoding proteins associated with AD pathology.

In the first AD dataset, RNA isolated from the hippocampal region between LOAD patients and controls was investigated for APA changes. A large degree of variation in APA was observed between different patient and control comparisons, suggesting that there is either natural variation between samples, or changes that are of collateral nature and thus may not be due or unique to AD. Here, only genes which showed APA in at least 3 of the controls were selected for further analysis eliminating more than 80% of the identified APA events. This stringent cut off makes the APA event more likely to be linked to AD instead of other unknown factors. However, it is important to keep in mind that although similar phenotypes are seen in different AD patients, AD is a composite disease, with varying pathologies associated to each case. Therefore, the same genes may not be affected by APA in each AD sample.

The RNA-seq dataset we used was originally investigated for transcriptome changes between LOAD and control samples and identified novel lncRNAs which were proposed to contribute to AD pathology [10]. Changes in gene expression in 113 protein coding genes were recognised [10], but none of these genes were identified as showing APA in our DaPars analysis, suggesting APA does not affect the stability of the resulting APA mRNA isoforms in AD but instead may alter protein production or localisation.

With the second dataset we assessed APA changes in the temporal and frontal lobe of the brain [7]. While we observed 3’UTR lengthening in both brain regions, most APA events were region-specific. In addition, we identified little overlap between genes that are subjected to APA in the lobes and the hippocampus (Additional file 1: Figure S2), which again suggests that these different brain regions have distinct APA profiles.

In the AD brain, it has previously been shown that genes associated with neuron structure and synapse function have altered expression [7]. Interestingly, our analysis identified genes associated with synaptic function and show significant APA profile changes between control and AD; as exemplified by BIN1 and VAMP2. Whether the APA changes observed in this gene cohort are physiologically relevant is unclear. However, it is well established that APA can affect protein output and localisation [4] and changing protein expression of BIN1 [26] and VAMP2 [27] have previously been associated with AD. It is therefore plausible that 3’UTR lengths changes of BIN1 and VAMP2 in AD may impact on the final protein output of these genes which in turn may contribute to the disease state or progression.

Changes in splicing patterns have been linked to age in humans [55] and reduced expression of the nervous system specific RBP Nova1/Nova2 was linked to altered splicing in AD patients, but not cognitively normal aged patients [55]. Nova2 can also modulate poly(A) site choice by binding to a YCAY cis-element which inhibits co-transcriptional usage of a proximal poly(A) site through steric hindrance [56]. Decreased Nova expression may result in 3’UTR shortening in AD, but this was not observed on a global scale in any of these AD datasets. However, Nova binding sites were identified in the aUTR of UBR1 and VAMP2 [57], which both showed 3’UTR shortening in AD. Further functional studies will be required to investigate if reduced Nova expression affects poly(A) site choice of these transcripts. RBPs are important in regulating APA, and their function has been shown to be dysregulated in disease [58]. Altered availability of RBP binding sites through changes in 3’UTR length could thus play a significant role in the deregulation of genes in AD.

Defects in RNA metabolism have been implied in AD [17], and this analysis gives an indication that APA in relevant genes may also be dysregulated. However, genes known to be directly associated with AD, such as PS1, PS2, APP and APOE, did not show changes in poly(A) site usage in any of the datasets we interrogated, suggesting no role of APA in their regulation. However, with UBR1 and VAMP2 we did identify APA changes in physiologically relevant genes which could contribute to AD pathology.

APA in PD

There is evidence that APA occurs in PD associated genes, such as α-synuclein [13], however changes in APA in PD have not been assessed on a global level. Here, RNA-seq data from patient and control iPSCs differentiated to dopaminergic neurons from two different datasets was used to assess 3’UTR length changes in PD compared to controls.

The first dataset assessed APA in dopaminergic neurons derived from iPSCs in 3 S-PD patients. Far fewer APA events were identified compared to control samples, suggesting APA changes are unlikely to be a major causative or consequential of the PD phenotype.

Comparison between F-PD and S-PD RNA-seq data with control individuals in the second RNA-seq dataset did again not show global changes in APA. However, the relative small number of genes that show altered 3’UTR length in PD versus controls, were associated with mitochondrial organisation or alterations in macromolecular subunit organisation, i.e. protein aggregation. These are two key phenotypes seen in PD, suggesting that APA may be affecting individual genes that are directly related to the disease state.

While, evidence of APA in α-synuclein has been described in PD to influence transcript localisation, and therefore protein localisation [13], no significant differences in 3’UTR length were seen in either S-PD or F-PD (Additional file 1: Figure S3), suggesting in this case, APA does not contribute to dysregulated α-synuclein expression in PD.

In conclusion, our data suggest that whilst APA may not have a global impact on gene expression changes in PD, it nevertheless may regulate the expression of a select few physiologically relevant genes including LONP1 and NELFA that could contribute to PD pathology.

APA in ALS

Defects in RNA processing have been linked to ALS, primarily due to mutations in FUS and TDP-43 [37]. In addition, depletion of HNRNPA2B1, mutations in which are associated with ALS, was shown to promote usage of distal poly(A) sites [59].

APA changes using DaPars in ALS have previously been explored and more APA changes were observed in sALS compared to c9ALS [15]. This genome-wide study by Prudencio and colleagues assessed transcriptome-wide changes in the cerebellum and frontal cortex which are not the focal points of degeneration in ALS, although these regions may have compromised function [60]. To further investigate APA in ALS, we used a publicly available RNA-seq dataset from spinal cord motor neurons, which is a physiologically highly relevant region for ALS. We found that around 700 genes (2.7%) of genes showed significant APA changes, and an overall trend to 3’UTR lengthening was observed in patient derived RNA. Whether any of these UTR lengthening events are physiologically relevant is unclear but the affected genes are enriched for GO terms such as neuron projection development and cytoskeleton intracellular transport. Furthermore, the 3’UTR lengthening event identified in the UCHL1 gene is of particular interest as the lengthening was observed in three of the sALS versus control comparisons. Most interestingly, reduced UCHL1 protein expression has been linked to neurodegeneration [46], and the loss of this enzyme contributes to motor neuron degeneration [45].

Conclusions

This paper assesses APA changes in different neurodegeneration diseases using existing publicly available RNA-seq datasets. Although widespread APA changes were not observed, for each neurodegenerative disease we identified APA events that occur in physiologically relevant genes. In particular we identified UCHL1, BIN1 and VAMP2 as examples that show disease specific UTR lengths changes that may contribute to the pathology of the respective diseases. Furthermore, we show that whilst different genes do show APA in AD, PD and ALS, in all three pathologies genes encoding proteins associated with the UPP were overrepresented indicating that deregulation of such genes by APA may be a common pathology. Finally, there is increasing evidence showing that changes in RNA metabolism, including in splicing [17] are associated with neurodegenerative disease states, to which APA can be now added.

Methods

Raw sequencing data in the form of FASTQ files and alignment information was downloaded from the Sequence Read Archive (SRA). GEO Accession numbers of datasets used are outlined in Table 1. FASTQ files were groomed and filtered reads were aligned to the human genome (hg19) using TopHat using Galaxy (https://usegalaxy.org/). The aligned BAM files were converted to bedgraph files and subjected to Dynamic Analysis of Alternative Polyadenylation from RNA-Seq (DaPars) [16, 61]. The code is available at https://github.com/ZhengXia/DaPars. This python script enables de novo identification and analysis of dynamic poly(A) sites from RNA-Seq data. In DaPars, the distal poly(A) site is considered to be the default poly(A) site in the transcript, and de novo poly(A) sites are identified using a linear regression model. The read intensities are modelled as a linear combination of proximal and distal poly(A) sites at a single nucleotide resolution. The relative usage of proximal and distal poly(A) sites can be assessed by looking at the percentage of reads which use the distal instead of the proximal poly(A) site (percentage of distal poly(A) site usage index, PDUI). The PDUI is calculated in multiple data sets and the difference in PDUI between different data sets is calculated. Positive or negative PDUIs determine whether a statistically significant APA event is identified as 3’UTR lengthening or shortening respectively between two samples [16]. Genes with more than a 25% change between patient and control are highlighted as being statistically significant changes in poly(A) site usage. The output from DaPars was coupled to known poly(A) site coordinates ensuring false poly(A) sites are not identified. An additional filter in the DaPars script was added so that the predicted proximal poly(A) site was present within 250 nucleotides of a previously annotated poly(A) site, rather than 500 nucleotides as initially proposed by the authors [61]. Bedgraph files were converted to bigwig files for visualisation of the 3’UTRs on the IGV Browser.

Table 1 GEO Accession Numbers of publicly available RNA-seq data
Full size table

Abbreviations

3’UTR:

3’Untranslated Region

AD:

Alzheimer’s Disease

ALS:

Amyotrophic Lateral Sclerosis

APA:

Alternative Polyadenylation

DaPars:

Dynamic analysis of Alternative Polyadenylation from RNA-Seq

EOAD:

Early Onset Alzheimer’s Disease

GO:

Gene Ontology

LOAD:

Late Onset Alzheimer’s Disease

NLS:

Nuclear Localisation Sequence

PD:

Parkinson’s Disease

poly(A):

polyadenylation

RNA:

Ribonucleic acid

References

  1. Parakh S, Atkin JD. Protein folding alterations in amyotrophic lateral sclerosis. Brain Res. 2016;1648:633–49.

    Article  CAS  Google Scholar 

  2. Dennissen FJA, Kholod N, van Leeuwen FW. The ubiquitin proteasome system in neurodegenerative diseases: culprit, accomplice or victim? Prog Neurobiol. 2012;96:190–207.

    Article  CAS  Google Scholar 

  3. Derti A, Garrett-Engele P, MacIsaac KD, Stevens RC, Sriram S, Chen R, et al. A quantitative atlas of polyadenylation in five mammals. Genome Res. 2012;22:1173–83. https://doi.org/10.1101/gr.132563.111.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Neve J, Patel R, Wang Z, Louey A, Furger AM. Cleavage and polyadenylation: Ending the message expands gene regulation. RNA Biol. 2017;14(7):865–90. https://doi.org/10.1080/15476286.2017.1306171.

    Article  Google Scholar 

  5. Neve J, Furger A. Alternative polyadenylation: less than meets the eye? Biochem Soc Trans. 2014;42(4):1190–5.

  6. Dumitriu A, Golji J, Labadorf AT, Gao B, Beach TG, Myers RH, et al. Integrative analyses of proteomics and RNA transcriptomics implicate mitochondrial processes, protein folding pathways and GWAS loci in Parkinson disease. BMC Med Genet. 2016;9:5.

    Google Scholar 

  7. Twine NA, Janitz K, Wilkins MR, Janitz M. Whole transcriptome sequencing reveals gene expression and splicing differences in brain regions affected by Alzheimer’s disease. PLoS One. 2011;6.

    Article  CAS  Google Scholar 

  8. Lau P, Bossers K, Janky R, Salta E, Frigerio CS, Barbash S, et al. Alteration of the microRNA network during the progression of Alzheimer’s disease. EMBO Mol Med. 2013;5:1613–34.

    Article  CAS  Google Scholar 

  9. Butovsky O, Jedrychowski MP, Cialic R, Krasemann S, Murugaiyan G, Fanek Z, et al. Targeting miR-155 restores abnormal microglia and attenuates disease in SOD1 mice. Ann Neurol. 2015;77:675–99.

    Article  Google Scholar 

  10. Magistri M, Velmeshev D, Makhmutova M, Faghihi MA. Transcriptomics profiling of Alzheimer’s disease reveal neurovascular defects, altered amyloid-β homeostasis, and deregulated expression of long noncoding RNAs. J Alzheimers Dis. 2015;48:647–65.

    Article  CAS  Google Scholar 

  11. Goedert M, Wischik CM, Crowther RA, Walker JE, Klug A. Cloning and sequencing of the cDNA encoding a core protein of the paired helical filament of Alzheimer disease: identification as the microtubule-associated protein tau. Proc Natl Acad Sci. 1988;85:4051–5.

    Article  CAS  Google Scholar 

  12. Dickson JR, Kruse C, Montagna DR, Finsen B, Wolfe MS. Alternative polyadenylation and miR-34 family members regulate tau expression. J Neurochem. 2013;127:739–49. https://doi.org/10.1111/jnc.12437.

    Article  CAS  PubMed  Google Scholar 

  13. Rhinn H, Qiang L, Yamashita T, Rhee D, Zolin A, Vanti W, et al. Alternative α-synuclein transcript usage as a convergent mechanism in Parkinson’s disease pathology. Nat Commun. 2012;3:1084. https://doi.org/10.1038/ncomms2032.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Koyama A, Sugai A, Kato T, Ishihara T, Shiga A, Toyoshima Y, et al. Increased cytoplasmic TARDBP mRNA in affected spinal motor neurons in ALS caused by abnormal autoregulation of TDP-43. Nucleic Acids Res. 2016;44:5820–36. https://doi.org/10.1093/nar/gkw499.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  15. Prudencio M, Belzil VV, Batra R, Ross CA, Gendron TF, Pregent LJ, et al. Distinct brain transcriptome profiles in C9orf72-associated and sporadic ALS. Nat Neurosci. 2015;18:1175–82. https://doi.org/10.1038/nn.4065.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Xia Z, Donehower LA, Cooper TA, Neilson JR, Wheeler DA, Wagner EJ, et al. Dynamic analyses of alternative polyadenylation from RNA-seq reveal a 3′-UTR landscape across seven tumour types. Nat Commun. 2014;5:5274. https://doi.org/10.1038/ncomms6274.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Bai B, Hales CM, Chen P, Gozal Y, Dammer EB, Fritz JJ. U1 small nuclear ribonucleoprotein complex and RNA splicing alterations in Alzheimer ’ s disease. Proc Natl Acad Sci U S A. 2013;110:16562–7.

    Article  CAS  Google Scholar 

  18. Lukiw WJ, Bazan NG. Cyclooxygenase 2 RNA message abundance, stability, and hypervariability in sporadic Alzheimer neocortex. J Neurosci Res. 1997;50:937–45.

    Article  CAS  Google Scholar 

  19. Mbella EG, Bertrand S, Huez G, Octave JN. A GG nucleotide sequence of the 3′ untranslated region of amyloid precursor protein mRNA plays a key role in the regulation of translation and the binding of proteins. MolCell Biol. 2000;20:4572–9 http://www.ncbi.nlm.nih.gov/pubmed/10848584.

    CAS  Google Scholar 

  20. Hwang C-S, Shemorry A, Auerbach D, Varshavsky A. The N-end rule pathway is mediated by a complex of the RING-type Ubr1 and HECT-type Ufd4 ubiquitin ligases. Nat Cell Biol. 2010;12:1177–85. https://doi.org/10.1038/ncb2121.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Zenker M, Mayerle J, Lerch MM, Tagariello A, Zerres K, Durie PR, et al. Deficiency of UBR1, a ubiquitin ligase of the N-end rule pathway, causes pancreatic dysfunction, malformations and mental retardation (Johanson-blizzard syndrome). Nat Genet. 2005;37:1345 https://doi.org/10.1038/ng1681.

    Article  CAS  Google Scholar 

  22. Absalon S, Kochanek DM, Raghavan V, Krichevsky AM. MiR-26b, upregulated in Alzheimer’s disease, activates cell cycle entry, tau-phosphorylation, and apoptosis in Postmitotic neurons. J Neurosci. 2013;33:14645–59.

    Article  CAS  Google Scholar 

  23. Lukiw WJ. Micro-RNA speciation in fetal, adult and Alzheimer’s disease hippocampus. Neuroreport. 2007;18:297–300.

    Article  CAS  Google Scholar 

  24. Ciavardelli D, Silvestri E, Del V a, Bomba M, De GD, Moreno M, et al. Alterations of brain and cerebellar proteomes linked to Aβ and tau pathology in a female triple-transgenic murine model of Alzheimer’s disease. Cell Death Dis. 2010;1:e90.

    Article  CAS  Google Scholar 

  25. Tan MS, Yu JT, Tan L. Bridging integrator 1 (BIN1): form, function, and Alzheimer’s disease. Trends Mol Med. 2013;19:594–603.

    Article  CAS  Google Scholar 

  26. Chapuis J, Hansmannel F, Gistelinck M, Mounier a VCC, Kolen KV, et al. Increased expression of BIN1 mediates Alzheimer genetic risk by modulating tau pathology. Mol Psychiatry. 2013;18:1225–34.

    Article  CAS  Google Scholar 

  27. Vallortigara J, Whitfield D, Quelch W, Alghamdi A, Howlett D, Hortobágyi T, et al. Decreased levels of VAMP2 and monomeric alpha-Synuclein correlate with duration of dementia. J Alzheimers Dis. 2015;50:101–10.

    Article  Google Scholar 

  28. Poewe W, Seppi K, Tanner CM, Halliday GM, Brundin P, Volkmann J, et al. Parkinson diseaseNat Rev Dis Prim. 2017;3:17013.

    Article  Google Scholar 

  29. Spillantini MG, Schmidt ML, Lee VM, Trojanowski JQ, Jakes R, Goedert M. Alpha-synuclein in Lewy bodies. Nature. 1997;388:839–40.

    Article  CAS  Google Scholar 

  30. Bogaerts V, Theuns J, Van Broeckhoven C. Genetic findings in Parkinson’s disease and translation into treatment: a leading role for mitochondria? Genes, Brain Behav. 2008;7:129–51.

    Article  CAS  Google Scholar 

  31. Glaab E, Schneider R. Neurobiology of disease comparative pathway and network analysis of brain transcriptome changes during adult aging and in Parkinson ’ s disease. Neurobiol Dis. 2015;74:1–13.

    Article  CAS  Google Scholar 

  32. Woodard CM, Campos BA, Kuo S, Nirenberg MJ, Nestor MW, Zimmer M, et al. iPS cell-derived dopamine neurons reveal differences between monozygotic twins discordant for Parkinson’s disease. Cell Rep. 2014;9:1173–82.

    Article  CAS  Google Scholar 

  33. Desideri E, Martins LM. Mitochondrial stress signalling: HTRA2 and Parkinson’s disease. Int J Cell Biol. 2012;2012.

  34. Jin SM, Youle RJ. The accumulation of misfolded proteins in the mitochondrial matrix is sensed by PINK1 to induce PARK2/Parkin-mediated mitophagy of polarized mitochondria. Autophagy. 2013;9:1750–7.

    Article  CAS  Google Scholar 

  35. Kim J, Inoue K, Ishii J, Vanti WB, Voronov SV, Murchison E, et al. A MicroRNA feedback circuit in midbrain dopamine neurons. Science. 2007;317:1220–4.

    Article  CAS  Google Scholar 

  36. Blokhuis AM, Groen EJN, Koppers M, Van Den Berg LH, Pasterkamp RJ. Protein aggregation in amyotrophic lateral sclerosis. Acta Neuropathol. 2013;125:777–94.

    Article  CAS  Google Scholar 

  37. Masuda A, ichi TJ, Ohno K. FUS-mediated regulation of alternative RNA processing in neurons: insights from global transcriptome analysis. Wiley Interdiscip Rev RNA. 2016;7:330–40.

    Article  CAS  Google Scholar 

  38. Schwartz JC, Ebmeier CC, Podell ER, Heimiller J, Taatjes DJ, Cech TR. FUS binds the CTD of RNA polymerase II and regulates its phosphorylation at Ser2. Genes Dev. 2012;26:2690–5. https://doi.org/10.1101/gad.204602.112.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Kapeli K, Pratt GA, Vu AQ, Hutt KR, Martinez FJ, Sundararaman B, et al. Distinct and shared functions of ALS-associated proteins TDP-43, FUS and TAF15 revealed by multisystem analyses. Nat Commun. 2016;7:12143.

    Article  CAS  Google Scholar 

  40. Masuda A, ichi TJ, Okuno T, Okamoto T, Ohkawara B, Ito M, et al. Position-specific binding of FUS to nascent RNA regulates mRNA length. Genes Dev. 2015;29:1045–57. https://doi.org/10.1101/gad.255737.114.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Rogelj B, Easton LE, Bogu GK, Stanton LW, Rot G, Curk T, et al. Widespread binding of FUS along nascent RNA regulates alternative splicing in the brain. Sci Rep. 2012;2:1–10.

    Article  Google Scholar 

  42. Batra R, Hutt K, Vu A, Rabin SJ, Baughn MW, Libby RT, et al. Gene Expression Signatures of Sporadic ALS Motor Neuron Populations. BioRxiv. 2016;1:038448.

    Google Scholar 

  43. Osaka H, Wang YL, Takada K, Takizawa S, Setsuie R, Li H, et al. Ubiquitin carboxy-terminal hydrolase L1 binds to and stabilizes monoubiquitin in neuron. Hum Mol Genet. 2003;12:1945–58.

    Article  CAS  Google Scholar 

  44. Liu Y, Fallon L, Lashuel HA, Liu Z, Lansbury P. The UCH-L1 gene encodes two opposing enzymatic activities that affect a-Synuclein degradation and Parkinson’s disease susceptibility. Cell. 2002;111:209–18.

    Article  CAS  Google Scholar 

  45. Jara JH, Genç B, Cox GA, Bohn MC, Roos RP, Macklis JD, et al. Corticospinal motor neurons are susceptible to increased ER stress and display profound degeneration in the absence of UCHL1 function. Cereb Cortex. 2015;25:4259–72.

    Article  Google Scholar 

  46. Choi J, Levey AI, Weintraub ST, Rees HD, Gearing M, Chin LS, et al. Oxidative modifications and Down-regulation of ubiquitin carboxyl-terminal hydrolase L1 associated with idiopathic Parkinson’s and Alzheimer’s diseases. J Biol Chem. 2004;279:13256–64.

    Article  CAS  Google Scholar 

  47. Ristic G, Tsou W-L, Todi SV. An optimal ubiquitin-proteasome pathway in the nervous system: the role of deubiquitinating enzymes. Front Mol Neurosci. 2014;7:72.

    Article  Google Scholar 

  48. Miura P, Shenker S, Andreu-Agullo C, Westholm JO, Lai EC. Widespread and extensive lengthening of 3’UTRs in the mammalian brain. Genome Res. 2013;23:812–25.

    Article  CAS  Google Scholar 

  49. Neve J, Patel R, Wang Z, Louey A, Furger AM. Cleavage and polyadenylation: ending the message expands gene regulation. RNA Biol. 2017:1–26. https://doi.org/10.1080/15476286.2017.1306171.

    Article  Google Scholar 

  50. Grassi E, Mariella E, Lembo A, Molineris I, Provero P. Roar: detecting alternative polyadenylation with standard mRNA sequencing libraries. BMC Bioinformatics. 2016;17:423. https://doi.org/10.1186/s12859-016-1254-8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Ha KCH, Blencowe BJ, Morris Q. QAPA: a new method for the systematic analysis of alternative polyadenylation from RNA-seq data. Genome Biol. 2018;19:45. https://doi.org/10.1186/s13059-018-1414-4.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  52. Kim M, You B-H, Nam J-W. Global estimation of the 3′ untranslated region landscape using RNA sequencing. Methods. 2015;83:111–7. https://doi.org/10.1016/J.YMETH.2015.04.011.

    Article  CAS  PubMed  Google Scholar 

  53. Masamha CP, Xia Z, Yang J, Albrecht TR, Li M, Shyu A-B, et al. CFIm25 links alternative polyadenylation to glioblastoma tumour suppression. Nature. 2014;509:412–6. https://doi.org/10.1038/nature13261.

    Article  CAS  Google Scholar 

  54. Hu B, Li X, Huo Y, Yu Y, Zhang Q, Chen G, et al. Cellular responses to HSV-1 infection are linked to specific types of alterations in the host transcriptome. Sci Rep. 2016;6:28075.

    Article  CAS  Google Scholar 

  55. Tollervey JR, Wang Z, Hortobagyi T, Witten JT, Zarnack K, Kayikci M, et al. Analysis of alternative splicing associated with aging and neurodegeneration in the human brain. Genome Res. 2011:1572–82.

    Article  CAS  Google Scholar 

  56. Licatalosi DD, Mele A, Fak JJ, Ule J, Kayikci M, Chi SW, et al. HITS-CLIP yields genome-wide insights into brain alternative RNA processing. Nature. 2008;456:464–9. https://doi.org/10.1038/nature07488.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  57. Zhang C, Darnell RB. Mapping in vivo protein-RNA interactions at single-nucleotide resolution from HITS-CLIP data. Nat Biotechnol. 2011;29:607–14.

    Article  CAS  Google Scholar 

  58. Tian B, Manley JL. Alternative polyadenylation of mRNA precursors. Nat Rev Mol Cell Biol. 2016;18:18–30. https://doi.org/10.1038/nrm.2016.116.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  59. Martinez FJ, Pratt GA, Van Nostrand EL, Batra R, Huelga SC, Kapeli K, et al. Protein-RNA networks regulated by Normal and ALS-associated mutant HNRNPA2B1 in the nervous system. Neuron. 2016;92:780–95.

    Article  CAS  Google Scholar 

  60. Prell T, Grosskreutz J. The involvement of the cerebellum in amyotrophic lateral sclerosis. Amyotroph Lateral Scler Frontotemporal Degener. 2013;14:507–15.

    Article  CAS  Google Scholar 

  61. Masamha C, Xia Z, Yang J, Albrecht TR, Li M, Shyu A-B, et al. CFIm25 links alternative polyadenylation to glioblastoma tumor suppression. Nature. 2014;510:412–3.

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank all the members of the Furger lab for their discussions and contributions to this paper. We thank Jane Mellor for the critical reading of the manuscript.

Funding

RP and JN were funded by an MRC PhD studentship. CB holds an Oxford Percival Stanion Studentship. AF is supported by the BBSRC (BB/N001184/1). The funders had no role in the design, analysis and interpretation of the results.

Availability of data and materials

All the data sets we used in our analysis are publicly available and the accession numbers are detailed in the designated table (GEO Accession Numbers of publicly available RNA-seq data) in the methods section of this manuscript.

Author information

Author notes

  1. Cillian Brophy and Mark Hickling contributed equally to this work.

Authors and Affiliations

  1. Department of Biochemistry, University of Oxford, Oxford, OX1 3QU, UK

    Radhika Patel, Cillian Brophy, Mark Hickling, Jonathan Neve & André Furger

Authors
  1. Radhika Patel
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Cillian Brophy
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Mark Hickling
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Jonathan Neve
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. André Furger
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

RP did the bioinformatics analysis supported by CB, MH and JN. RP together with AF wrote the manuscript. AF and RP designed the research. The authors have read and approved the manuscript.

Corresponding author

Correspondence to André Furger.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Additional file

Additional file 1:

Figure S1. APA Heat map for the ten genes that undergo shortening of their 3’UTRs and ten genes that show 3’UTR lengthening in LOAD samples compared to two or more control samples. The gene names are indicated on the left (10 lengthening genes and ten shortening genes as per Fig. 2c) The percentage of distal poly(A) site usage index values (PDUI) range from shades of green indicating shortening and shades of red indicating Lengthening. The different comparisons between diseased (AD1–4) and controls (C1-C4) are outlined on the X-axis. Figure S2. Different genes affected by APA in different regions of the brain. A. Venn Diagram to show the overlap of genes regulated by APA in the hippocampus from the LOAD Dataset (teal) and the frontal and temporal brain region dataset (blue). B. Table to show the APA change of the 21 genes identified to show altered UTR lengths with DaPars analysis (PDUI > 0.25, Fisher’s exact Test, p < 0.05) in the two AD datasets assessed. The genes that showed 3’UTR lengthening or 3’UTR shortening in both datasets have been separated from those that showed differential APA regulation. Figure S3. SNCA does not show UTR length changes in PD. A. Genome browser view of SNCA in the first PD dataset assessed to show no change in UTR length in the three S-PD samples (blue tracks) compared to control samples (green tracks). B. Genome browser view of SNCA in the second PD dataset assessed to show no change in UTR length in S-PD (blue track) or F-PD (purple track) compared to control (green tracks). In A & B, the length of the genome browser window shown is indicated above in kilo bases (kb) between the two arrows. Table S1. Summary of all the data sets used in Figs. 1, 2, 3 4, 5. Details regarding the data sets used to in the analysis’ that lead to the data presented in Figs. 1, 2, 3 4, 5 are given. (PPTX 342 kb)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Patel, R., Brophy, C., Hickling, M. et al. Alternative cleavage and polyadenylation of genes associated with protein turnover and mitochondrial function are deregulated in Parkinson’s, Alzheimer’s and ALS disease. BMC Med Genomics 12, 60 (2019). https://doi.org/10.1186/s12920-019-0509-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12920-019-0509-4

Keywords

BMC Medical Genomics

ISSN: 1755-8794

Contact us