Increased unedited Alu RNA patterns found in cortex extracellular vesicles in Alzheimer's disease resemble hippocampus vasculature Alu RNA editing patterns but not cortex Alu RNA editing patterns

Abstract

Background

Endogenous Alu RNAs form double-stranded RNAs recognized by double-stranded RNA sensors and activate IRF and NF-kB transcriptional paths and innate immunity. Deamination of adenosines to inosines by the ADAR family of enzymes, a process termed A-to-I editing, disrupts double-stranded RNA structure and prevents innate immune activation. Innate immune activation is observed in Alzheimer's disease, the most common form of dementia. We have previously reported loss of A-to-I editing in hippocampus vasculature, but no change in cortex or cortex vasculature, associated with Alzheimer's disease.

Objective

Here, we investigated the status of Alu RNA A-to-I editing in cortex extracellular vesicles in Alzheimer's disease.

Methods

We used existing RNA-seq data sets and the SPRINT software package to determine levels of Alu RNA A-to-I editing in cortex extracellular vesicles in Alzheimer's disease and control groups and compared these editing profiles to those found in both total cortex and hippocampus vasculature.

Results

We find substantial loss of Alu A-to-I editing in cortex extracellular vesicles in Alzheimer's disease. By measuring editing patterns on a gene-by-gene basis, we determined that editing patterns in cortex extracellular vesicles resemble editing patterns in hippocampus vasculature rather than total cortex.

Conclusions

We conclude that hippocampus vasculature unedited Alu RNAs are packaged in extracellular vesicles, travel to the cortex, deliver their cargo and stimulate innate immunity and alter other basic biological processes contributing to Alzheimer's disease progression.

Keywords

Alu RNA a-to-I editing Alzheimer's disease double-stranded RNA extracellular vesicles inflammation neurodegeneration

Introduction

Deamination of adenosines in double-stranded RNA to yield inosines catalyzed by the adenosine deaminase acting on RNA (ADAR) family of enzymes is a common post-transcriptional modification of RNA in eukaryotic cells.^1–6 This process is commonly referred to as ‘A-to-I editing’. In humans, the most frequent editing sites are found in RNAs transcribed from Alu retrotransposons.^7–9 Alu genomic elements are primate specific and arose from insertion into the genome of a head-to-tail fusion of 7SL RNA.¹⁰ About 1 million copies of Alu genomic sequences, which are about 300 nucleotides in length, exist in the human genome, thus Alu elements make up about 10% of the human genome.¹¹ RNAs containing Alu elements are transcribed by RNA polymerase 2 as part of pre-mRNAs or by RNA polymerase 3 as part of their normal life cycle. Alu RNAs transcribed from non-coding genome space, gene introns, intergenic space, 3’ and 5’ untranslated regions (UTR) are the most common sites of A-to-I editing while editing of Alu RNAs in exons is relatively less frequent.^12,13

In general terms RNA A-to-I editing affects several biological processes.^7,8,14 Editing of RNAs transcribed from introns can produce alternative splicing of mRNAs. Further, looping of Alu containing enhancers to promoters can alter promoter-enhancer selectivity and impact expression of cognate genes.¹⁵ Editing of Alu containing elements in 3’ UTRs may alter microRNA binding sights that may also affect mRNA stability.¹⁶ Editing may also affect additional functions encoded by nucleotide sequences within the 5’ UTR such as ribosome recruitment to the mRNA and start codon choices.¹⁷

Since Alu elements arose from a head-to-tail fusion of 7SL RNA,^11–13 Alu RNAs can form double-stranded RNA (dsRNA) structures capable of activating host dsRNA sensors including RIG-I, MDA5, and TLR3.^18–25 In addition, Alu genomic elements inserted close in the genome in opposite orientations are common and when transcribed, exist as two Alu RNA elements joined by a linker sequence, also referred to as Alu inverted repeats (IR/Alu), and IR/Alu RNAs are potent activators of the host dsRNA sensor, MDA5.^18–20 Numerous lines of evidence indicate that A-to-I editing of Alu RNAs reduces their dsRNA character and prevents activation of dsRNA sensors. Activation of dsRNA sensors causes innate immune activation resulting from induction of interferon-stimulated genes (ISGs), cytokines, and other inflammatory mediators because of activation of transcription factors, interferon regulatory factors (IRF) and nuclear factor-kappa B (NF-kB). In humans, inactivating mutations in ADAR result in one form of Aicardi-Goutières syndrome, characterized by chronic innate immune activation resulting in severe encephalopathy that can lead to death or existence in a vegetative state.²⁶

Extracellular vesicles (EVs), also referred to as exosomes are small vesicles (30–150 nanometers in diameter) produced in the endosomal compartment and secreted by most eukaryotic cells.^27–31 EVs contain proteins, nucleic acids, lipids, and metabolites and are released into a variety of biological fluids, including blood, urine, saliva, breast milk, synovial fluid, amniotic liquid, and cerebrospinal fluid. As such, they mediate both near and long-distance intercellular communication in both health and disease and affect a variety of aspects of cell biology.

Loss of Alu RNA A-to-I editing in blood occurs as a result of relapsing remitting multiple sclerosis (MS) and is accompanied by accumulation of Alu dsRNAs and induction of IRF and NF-kB regulated genes.^25,32 Loss of Alu RNA A-to-I editing also results from COVID-19 disease (blood and lung) and severe influenza disease (blood) and in cell culture following infection with SARS-CoV-2.^33,34 Thus, loss of endogenous Alu RNA A-to-I editing may contribute to both acute and chronic inflammation observed in these conditions.

Alzheimer's disease (AD) is a progressive disease and the most common form of dementia.³⁵ Its causes are incompletely understood but AD is associated with accumulation of aggregated misfolded proteins including amyloid-β plaques and neurofibrillary tangles. We find loss of Alu RNA A-to-I editing in hippocampus vasculature (HPCvasc) associated with AD.³⁶ In contrast, AD-dependent changes in editing are not observed in prefrontal cortex (CTX) or CTX vasculature. Here, we examined levels of A-to-I edited Alu RNAs in EVs isolated from CTX and find reduced levels of Alu RNA editing in AD samples compared to normal cognition (NCI) samples. Gene-specific editing patterns in CTX EVs are more like patterns observed in HPCvasc than CTX leading us to conclude that CTX EVs may have arisen from HPCvasc and travelled to CTX. CTX EVs carrying reduced levels of edited Alu RNAs may fuse with cells of the CTX to activate dsRNA sensors to create an inflammatory state or affect other cellular processes in CTX.

Methods

RNA-Seq FASTQ files

RNA-seq FASTQ files measuring RNA levels in EVs isolated from CTX used here have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO Series accession number GSE197505: AD (N = 8), and age and gender-matched controls (normal cognition, NCI) (N = 10).³⁷ RNA-seq FASTQ files measuring RNA levels in HPCvasc are accessible through accession number GSE163577 and have been previously described.³⁸ RNA-seq FASTQ files measuring RNA levels in total pre-frontal CTX are accessible via accession number GSE53699 and have been previously described: AD group; N = 8, and NCI group; N = 8.³⁹

Briefly, EVs were purified from pre-frontal CTX from postmortem brain tissues using standard procedures in accordance with the Minimal Information for Studies of Extracellular Vesicles 2018 guidelines, total RNA was purified and subjected to RNA sequencing using standard procedures. The human brain tissues in this study were provided by the National Human Brain Bank for Development and Function, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing, China, from neuropathologically classified individuals with AD or NCI at the time of death and brain donation.³⁷

Analysis of RNA A-to-I editing

Our methods for analysis of RNA A-to-I editing have been described.^33–35 Briefly, bulk RNA-seq FASTQ files were trimmed, and we used the following workflow to identify endogenous RNA A-to-I–editing sites from paired FASTQ sequencing files. The main identification tool was a python-based package called the SPRINT toolkit accepting sequence files and producing text files with the following information for each edit site: (1) genomic location; (2) type of edit (e.g., A-to-G; T-to-C), strand (+ or −); (3) number of edits per site and total number of reads per site.^40,41 Mathematica programs were developed to synthesize data: numbers of samples in groups with shared editing sites, mean numbers of total reads and edits for each editing site, editing sites common and unique to groups, and editing sites per gene. This information was tied to an Alu database to annotate each site: gene locations (intronic, noncoding RNA, intergenic, 3′ and 5′ untranslated regions (UTRs), exons) and if editing sites were within Alu or non-Alu elements.⁴² To avoid sequencing errors, editing sites were only included if total reads were greater than 5 and edit/read ratios were greater than 0.05. To create genome-wide A-to-I editing indices (AEI), we identified all A-to-I editing sites present in one study group and summed edit/read ratios for all editing sites across the genome among study groups. Gene expression levels from RNA-seq FASTQ files were determined using the DESeq2 R package as described.⁴³ Both raw p-values and p-values after adjusting for multiple testing, false discovery rate (FDR), were determined.

Results

AD-associated loss of a-to-i editing of Alu RNAs in pre-frontal cortex EVs

We obtained RNA-seq FASTQ files derived from a study that comprised isolation of EVs from frontal CTX of brain from individuals with normal cognition (NCI, N = 10) or AD (N = 8) at the time of death and brain donation (GSE197505).³⁶ Groups were age and gender matched and did not differ in the post-mortem time between brain harvest and RNA processing. We used the SPRINT software package to determine levels of RNA A-to-I editing among samples using RNA-seq data. Briefly, this package scans the genome for A-to-G mismatches compared to the reference genome after removing known single nucleotide polymorphisms and reports proportion of edits to total reads at each edited site. We required the proportion of edits/reads to exceed 0.05 and the number of total reads to exceed 5 to guard against sequencing errors. We also identified the number of times a given nucleotide was edited in each sample from separate groups to examine both rare and more common editing sites. To express these data, we calculated both total number of edited sites per group and created an A-to-I editing index (AEI) by multiplying average proportion of edits/reads at each site times the total number of samples in which a given site was edited in the NCI group or the AD group.

To initiate these studies, we used the SPRINT software package to determine the total number of expressed Alu RNA elements (reads ≥ 1) and the average number reads of all expressed Alu RNA elements in NCI and AD groups. We did not find these two measurements to be substantially different between the two groups (Table 1). We also employed the DESeq2 software package to determine total RNA counts in the two groups. Again, we did not find this measurement to be substantially different between NCI and AD groups. Thus, total number of Alu RNA elements detected and average expression levels of Alu RNA elements were similar in NCI and AD groups. Further, levels of total RNA detected by DESeq2 analysis were similar between NCI and AD groups.

Table 1.

Number and expression levels of RNA Alu elements in CTX EVs from NCI and AD groups determined from RNA-seq data.

	Total # Alu	Alu RNA reads	Total RNA counts
Group	RNA elements	avg ± std.dev.	avg ± std.dev.
NCI	63,153	8.7 ± 10.1	1.86X10⁷± 0.34X10⁷
AD	56,090	7.6 ± 9.1	2.37 × 10⁷± 0.62 × 10⁷

Total # of Alu RNA elements and average ± standard deviation Alu RNA reads determined using the SPRINT software package in NCI and AD groups. Total RNA counts determined by DESeq2.

We found reduced numbers of total A-to-I edited sites in EVs isolated from the AD group compared to the NCI group (Figure 1(a)). We also found reduced overall AEI, which also takes into count the frequency of individual editing sites, in the AD group compared to the NCI group (Figure 1(b)). Overall, the majority of editing sites were found in introns followed by intergenic space, 3’ UTRs, noncoding RNAs (ncRNA), 5’ UTRs, and synonymous (syn) and nonsynonymous edits found in exons (Figure 1(c), Supplemental File 1). Loss of editing in the AD group compared to the NCI group was similar in RNAs transcribed from these different genomic regions. We conclude from these analyses that overall levels of A-to-I edited Alu RNAs in EVs were reduced in the AD group compared to the NCI group.

Figure 1.

Reduced Alu RNA A-to-I editing in AD compared to NCI CTX EVs. (a) Number of editing sites (X-axis) found in CTX EV RNAs obtained from individuals with NCI (N = 10) or AD (N = 8) at the time of death and brain donation determined by SPRINT analysis of RNA-Seq data. Y-axis is the proportion of edits/reads at each site, p < 0.0001, Welch's t-test. (b) As in (a), except AEI was calculated for each editing site, p < 0.0001, Welch's t-test. (c) Genomic locations of Alu RNA A-to-I editing sites.

As a second approach to our analysis, we determined frequencies that individual editing sites were detected in two or more EV samples in NCI and AD groups. Overall, number of individual editing sites we detected declined as frequency of presence of editing sites increased from two to ten. Overall, we found loss of editing in the AD group compared to the NCI group at all frequencies (Figure 2(a)). This was also the case when we calculated AEI (Figure 2(b)). Thus, loss of editing was found when we examined editing sites that are rare in frequency and editing sites that are common in frequency in AD compared to NCI groups.

Figure 2.

Number of times a given Alu RNA A-to-I editing site was detected in NCI or AD CTX EVs. (a) Frequency (≥2, X-axis) of editing sites in CTX EVs from NCI or AD samples determined from RNA-seq data, p < 0.0001, 2-way ANOVA. (b) As in (a), except AEI was calculated, p < 0.0001, 2-way ANOVA.

A-to-I editing patterns in CTX EVs resemble HPCvasc editing patterns

We determined AEI per gene in CTX EV samples from NCI and AD groups. For this analysis we included all editing sites traversing the 5’ UTR to the 3’ UTR including both exons and introns. We identified those genes that exhibited high levels of editing in EVs in the NCI group but were completely unedited in the AD group (Figure 3(a)). We have previously reported that A-to-I editing in total hippocampus vasculature (HPCvasc) is reduced in the AD group compared to the NCI group.³⁶ In contrast, we did not find differences in A-to-I editing in total frontal cortex between NCI and AD groups. Therefore, we compared editing patterns of these same genes in HPCvasc samples between NCI and AD groups. We found that editing patterns of these genes were uniformly lower in the HPCvasc AD group compared to the NCI group (Figure 3(b)). We performed a similar comparison of editing in frontal cortex. Here, we did not detect markedly different editing patterns of this group of genes when NCI and AD groups were compared (Figure 3(c)). In fact, most of these genes were un-edited in CTX samples in both NCI and AD groups. We conclude that the pattern of A-to-I editing of this group of genes in CTX EVs was similar to the editing pattern found in total HPCvasc and dissimilar from the editing pattern of these genes in total CTX.

Figure 3.

Editing patterns of hyper-edited genes in CTX EV compared to total HPCvasc and total CTX. (a) AEI/gene was determined for indicated genes in NCI CTX EVs; editing of these genes was undetectable in AD CTX EVs. (b) As in (a) except AEI/gene in HPCvasc was determined. (c) As in (a) except AEI/gene in total CTX was determined.

Studies to analyze HPCvasc gene expression in NCI and AD groups employed single cell RNA sequencing (scRNA-seq).³⁸ Investigators used this approach to identify genes selectively expressed by different cell lineages within HPCvasc including astrocytes, neurons, oligodendrocytes, endothelial cells, different myeloid cells, T cells and others. We previously compared AEI of these lineage-defining genes between NCI and AD groups to ask if differences in editing in HPCvasc may be attributed to specific lineages.³⁶ In general terms, we found that AEI of lineage defining genes was reduced in AD groups compared to NCI groups leading us to conclude that loss of editing observed in HPCvasc was not cell-type specific but rather a general feature of cell lineages present within HPCvasc. Frontal CTX contains many of these same cell lineages, astrocytes, neurons, oligodendrocytes, etc. In general terms, EVs are packaged intracellularly, released from cells and migrate across extracellular space to reach target cells or tissues. Therefore, we asked if editing patterns of lineage-defining genes found in CTX EVs resembled editing patterns of these genes in HPCvasc or total CTX in NCI and AD groups. To illustrate our approach, we determined editing patterns of astrocyte lineage defining genes in EV, HPCvasc and CTX tissues by determining AEI and expressed data as the AD/NCI ratio, log₂ (Figure 4(a)). Gene AEIs were segregated into three categories; 1) log₂ ratio < -0.4 = reduced editing in AD compared to NCI, 2) ratio −0.4 to +0.4 = no difference in editing between AD and NCI groups, and 3) ratio > 0.4 = increased editing in AD compared to NCI groups. We determined if this ratio fell in the same category in cortex EVs and HPCvasc or in cortex EVs and CTX, 1 = ‘agree’ or 0 = ‘disagree’.

Figure 4.

Editing patterns of astrocyte specific genes in CTX EVs, total CTX and HPCvasc. (a) Editing patterns (Y-axis) of indicated genes (X-axis) selectively expressed by astrocytes were determined in CTX EVs, HPCvasc and total CTX. (b) Gene AEIs were segregated into three categories; 1) log₂ ratio < −0.4 = reduced editing in AD compared to NCI, 2) ratio −0.4 to +0.4 = no difference in editing between AD and NCI groups, and 3) ratio > 0.4 = increased editing in AD compared to NCI groups. We determined if this ratio fell in the same category in cortex EVs and HPCvasc or in cortex EVs and CTX, 1 = ‘agree’ or 0 = ‘disagree’. p = 0.003 determined by chi-square analysis assuming 33% of gene AEIs fell into each category by chance.

We determined statistical significance by chi-square analysis assuming 33% of gene AEIs should fall into each category by chance (Figure 4(b)). This comparison indicated that the editing pattern of astrocyte lineage defining genes in CTX EVs was similar to the editing patterns observed in HPCvasc but not CTX editing patterns. We performed a similar analysis of genes that define other cell lineages and obtained similar results (Table 2, Supplemental File 2). These studies support the notion that editing patterns of RNA in CTX EVs are more similar to RNA editing patterns in HPCvasc than CTX and raise the possibility that CTX EVs may not have been packaged in CTX but another brain region such as HPCvasc.

Table 2.

Comparison of lineage specific gene AEIs in CTX EVs to HPCvasc or total CTX.

	HPC vasculature			Frontal cortex
Lineage	# agree	# disagree	p	# agree	# disagree	p
Astrocyte	12	7	0.003	3	16	0.14
BEC, arterial	7	0	3 E-05	1	6	0.40
BEC, capillary	18	1	3 E-09	6	13	0.14
BEC, venous	10	1	4 E-05	0	10	0.03*
Ependymal	6	2	0.01	3	5	0.8
Microglia	29	15	3 E-06	9	35	0.08
Meningeal fibroblast	13	7	0.002	6	14	0.78
Neuron	18	26	0.26	17	27	0.42
Oligodendrocyte	37	26	1 E-05	27	36	0.10
OPC	27	14	7 E-06	13	28	0.87
Pericyte	16	7	0.0002	4	19	0.11
SMC	12	4	0.0004	1	15	0.02*
T cell	15	13	0.02	7	21	0.38

Lineage = different cell types in brain. AEI per lineage-defining gene was determined from CTX EVs, HPCvasc, and CTX RNA-seq data representing 3 independent studies. AD AEI/NCI, log₂, was determined for each gene per group. Gene AEIs were segregated into 3 categories: 1) ratio < -0.4 = reduced editing in AD compared to NCI groups, 2) ratio −0.4 to +0.4 = no difference in editing between AD and NCI groups, and 3) ratio > 0.4 = increased editing in AD compared to NCI groups. We determined if this ratio fell in the same category in CTX EVs and HPCvasc groups or in CTX EV and CTX groups = ‘agree’ or fell into one of the other categories = ‘disagree’. Numbers in the table are # of genes from each lineage that ‘agree’ or ‘disagree’ between groups. p values were determined by chi-square analysis assuming 33% of gene AEIs should fall into each category by chance. * = ‘# agree’ lower than expected by chance

Gene expression and pathways analysis of hyper-edited genes

Enhancer RNAs (eRNAs) transcribed from distal genomic enhancers regulate transcription of target genes by processes that are incompletely understood.^44–50 One view is that eRNAs participate in chromatin looping via RNA:DNA;DNA hybrids, termed R-loops, to recruit enhancers to promoters resulting in gene activation. New studies demonstrate that Alu sequences in eRNAs interact with Alu sequences in complimentary promoters of target genes and alter rates of gene transcription.¹⁵ We reasoned that A-to-I editing of Alu RNAs may affect ability of eRNAs to interact with their target promoters, either positively or negatively, and this may change transcript levels of target genes. Further, eRNAs packaged in EVs may be delivered to distal tissues and alter mRNA levels of target genes in tissues in which EVs travel to. As an initial test of this notion, we identified genes in EVs that exhibited the greatest level of editing in CTX EVs in the NCI group, hyper-edited genes, AEI > 50 (Figure 5(a)). We found that editing of these hyper-edited genes was uniformly reduced in the AD group. The vast majority of these edited Alu RNAs were derived from introns (Figure 1(c)).

Figure 5.

AEI and expression profiles of hyper-edited genes. (a) Differences in AEI of hyper-edited genes between NCI and AD EVs, Y-axis is AEI/gene and X-axis is rank, p < 0.0001 determined by 2-way ANOVA. (b) NCI and AD expression differences of hyper-edited genes in total CTX, Y-axis shows expression ratios of hyper-edited genes (a), NCI/AD, log2, X-axis is rank, p < 0.0001 determined by 2-way ANOVA. (c) Pathways analysis of hyper-edited genes determined by NIAID DAVID Functional Annotation Bioinformatics Microarray Analysis program. Input was hyper-edited genes (a) and indicated pathways were over-represented. X-axis is Benjamini p value, -log10. Dashed line = p 0.05.

We next asked if mRNA levels of these same genes were different in total CTX among NCI and AD groups reasoning that CTX EVs may deliver edited or unedited Alu-containing eRNAs to CTX and influence levels of target gene mRNAs in CTX. Overall, we found most of the genes within this group exhibited greater levels of expression in total CTX in the NCI group compared to the AD group (Figure 5(b), Supplemental File 3). We submitted this gene list to the NIAID DAVID Functional Annotation Bioinformatics Microarray Analysis program for pathways analysis.⁵¹ Pathways found to be significantly over-represented were 1) zinc finger proteins, 2) microtubules, 3) actin binding, 4) protein K48-linked ubiquitination, 5) ubiquitin conjugation, 6) isopeptide bonds, and 7) cross-inking glycyl-lysine isopeptide bonds (Figure 5(c)). Although we found that differences in expression levels of individual genes were not huge, it seems that small differences in expression of the large number of genes involved in these pathways may impact functions of these pathways.

Discussion

Our studies demonstrate elevated levels of unedited Alu RNAs in EVs isolated from CTX in AD samples compared to NCI samples. Reduced editing in AD is seen at both rare and common editing sites. Our previous studies show that reduced editing of Alu RNAs associated with AD is seen in HPCvasc but not total CTX or CTXvasc.³⁶ Our analysis of A-to-I editing of individual genes shows that gene editing patterns are more similar to editing patterns observed in HPCvasc than in CTX. Similarly, editing patterns of CTX EV cell lineage-specific genes are more similar to the editing patterns found in HPCvasc than in total CTX in both AD and NCI groups. A model consistent with these results is that EVs found in CTX are packaged with Alu RNAs from HPCvasc and travel to CTX to deliver their cargo.

Two lines of evidence support our notion that EVs isolated from CTX develop in HPCvasc and travel to CTX. First, genes that have high levels of editing in NCI CTX EVs but no detectable levels of editing in AD CTX EVs display a similarly high editing level in NCI HPCvasc and a reduced or undetectable editing level in AD HPCvasc. In contrast, these same genes are largely unedited in both NCI CTX and AD CTX, suggesting RNAs in CTX EVs do not come from CTX but rather come from HPCvasc. Second, editing patterns of cell lineage specific genes observed in CTX EVs, both from NCI and AD samples, look more like the gene editing patterns observed in HPCvasc than gene editing patterns observed in CTX. Further studies will be required to determine if RNAs from EVs located in other brain regions resemble HPCvasc editing patterns or exhibit unique or region-specific RNA editing patterns.

Neuroinflammation is implicated in the pathogenesis of AD.^52–55 Loss of editing of Alu RNAs results in activation of IRF and NF-kB transcriptional paths and innate immune responses. Activation of these same transcriptional paths and induction of IRF and NF-kB responsive genes is also observed in AD.^53–55 Thus, packaging of unedited Alu RNAs in CTX EVs associated with AD may contribute to the neuroinflammation observed in AD.

As outlined above, editing of Alu elements within eRNAs may affect eRNA function that, in turn, will affect gene transcript levels.¹⁵ Therefore, it seems that a second possibility is that CTX EVs deliver unedited, in the case of AD, and edited, in the case of NCI, eRNAs to CTX that may alter gene expression programs. In support of this idea, we find that hyper-edited genes in CTX EVs display uniformly reduced editing in AD samples compared to NCI samples. Expression of this group of genes in CTX is largely reduced in AD samples compared to NCI samples. Pathway analysis shows that these genes fall into groups whose functions have been implicated in AD pathogenesis, such as microtubule and actin function,⁵⁶ ubiquitination,⁵⁷ and others.⁵⁸ Interestingly, our studies in multiple sclerosis result in a similar finding²⁵; reduced editing on a per gene basis is associated with reduced gene expression raising the possibility that this may be a more general finding.

In general terms, EVs are released from most eukaryotic cells and carry nucleic acids, proteins, lipids, and various other cellular products as cargo.^27–31 EVs can deliver this cargo to either adjacent cells or cells at a distance and this is thought to represent a form of intercellular communication. Our results support the idea that CTX EVs may come from the HPCvasc given the similar A-to-I gene editing patterns that are observed. Loss of volume or atrophy of the HPC is an early sign of AD.^59,60 Further, breakdown of the blood-brain barrier in the HPC is not only an early sign of AD but is also associated with mild cognitive impairment.^60–64 Although somewhat speculative, it seems that accumulation of unedited Alu RNAs in HPCvasc, their packaging in EVs, and spreading to other regions of the brain may contribute to initiation and progression of AD-associated cognitive decline.

Footnotes

Acknowledgments

The authors have no acknowledgments to report.

ORCID iDs

Thomas M Aune

Philip S Crooke

Author contributions

Philip Crooke, 3rd (Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Validation; Writing – original draft; Writing – review & editing); John Tossberg (Data curation; Investigation; Validation; Writing – review & editing); Thomas Martin Aune (Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Project administration; Writing – original draft; Writing – review & editing).

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by grants from the NIH (R21AI144193 and R21AI164107 to T.M. Aune.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability

Supplemental data will be deposited in PubMed Central.

Supplemental material

Supplemental material for this article is available online.

References

Bazak

Haviv

Barak

, et al. A-to-I RNA editing occurs at over a hundred million genomic sites, located in a majority of human genes. Genome Res 2014; 24: 365–376.

Porath

Knisbacher

Eisenberg

, et al. Massive A-to-I RNA editing is common across the Metazoa and correlates with dsRNA abundance. Genome Biol 2017; 18: 185.

Reich

Bass

. Mapping the dsRNA world. Cold Spring Harb Perspect Biol 2019; 11: a035352.

Sakurai

Ueda

Yano

, et al. A biochemical landscape of A-to-I RNA editing in the human brain transcriptome. Genome Res 2014; 24: 522–534.

Picardi

Horner

Pesole

. Single-cell transcriptomics reveals specific RNA editing signatures in the human brain. RNA 2017; 23: 860–865.

Nishikura

. A-to-I editing of coding and non-coding RNAs by ADARs. Nat Rev Mol Cell Biol 2016; 17: 83–96.

Athanasiadis

Rich

Maas

. Widespread A-to-I RNA editing of alu-containing mRNAs in the human transcriptome. PLoS Biol 2004; 2: e391.

Deininger

. Alu elements: know the SINEs. Genome Biol 2011; 12: 236.

Bennett

Keller

Mills

, et al. Active alu retrotransposons in the human genome. Genome Res 2008; 18: 1875–1883.

10.

Jurka

Smith

. A fundamental division in the alu family of repeated sequences. Proc Natl Acad Sci U S A 1988; 85: 4775–4778.

11.

Kriegs

Churakov

Jurka

, et al. Evolutionary history of 7SL RNA derived SINEs in supraprimates. Trends Genet 2007; 23: 158–161.

12.

Mighell

Markham

Robinson

. Alu sequences. FEBS Lett 1997; 417: 1–5.

13.

Ullu

Tschudi

. Alu sequences are processed 7sl RNA genes. Nature 1984; 312: 171–172.

14.

Eisenberg

Levanon

. A-to-I RNA editing—Immune protector and transcriptome diversifier. Nat Rev Genet 2018; 19: 473–490.

15.

Liang

Cao

, et al. Complementary Alu sequences mediate enhancer–promoter selectivity. Nature 2023; 619: 868–875.

16.

Marceca

Distefano

Tomasello

, et al. MiREDiBase: a manually curated database of validated and putative editing events in microRNAs. Sci Data 2021; 8: 199.

17.

Nishikura

. Functions and regulation of RNA Editing by ADAR deaminases. Annu Rev Biochem 2010; 79: 321–349.

18.

Ahmad

Yang

, et al. Breaching self-tolerance to alu duplex RNA underlies MDA5-mediated inflammation. Cell 2018; 172: 797–810.e13.

19.

Lamers

van den Hoogen

Haagmans

. ADAR1: “editor-in-chief” of cytoplasmic innate immunity. Front Immunol 2019; 10: 1763.

20.

Liddicoat

Piskol

Chalk

, et al. RNA Editing by ADAR1 prevents MDA5 sensing of endogenous dsRNA as nonself. Science 2015; 349: 1115–1120.

21.

Portal

Pavet

Erb

, et al. Human cells contain natural double-stranded RNAs with potential regulatory functions. Nat Struct Mol Biol 2015; 22: 89–97.

22.

Samuel

. Adenosine deaminase acting on RNA (ADAR1), a suppressor of double-stranded RNA-triggered innate immune responses. J Biol Chem 2019; 294: 1710–1720.

23.

Sharma

Fitzgerald

Cancro

, et al. Nucleic acid-sensing receptors: rheostats of autoimmunity and autoinflammation. J Immunol 2015; 195: 3507–3512.

24.

Yang

Deng

Zhu

, et al. Adenosine deaminase acting on RNA 1 limits RIG-I RNA detection and suppresses IFN production responding to viral and endogenous RNAs. J Immunol 2014; 193: 3436–3445.

25.

Tossberg

Heinrich

Farley

, et al. Adenosine-to-Inosine RNA editing of Alu double-stranded RNAs is markedly decreased in multiple sclerosis and unedited Alu dsRNAs are potent activators of proinflammatory transcriptional responses. J Immunol 2020; 205: 2606–2617.

26.

Rice

Kasher

Forte

GMA

, et al. Mutations in ADAR1 cause aicardi-goutières syndrome associated with a type I interferon signature. Nat Genet 2012; 44: 1243–1248.

27.

You

Ikezu

. Emerging roles of extracellular vesicles in neurodegenerative disorders. Neurobiol Dis 2019; 130: 04512.

28.

Mulcahy

Pink

Carter

. Routes and mechanisms of extracellular vesicle uptake. J Extracell Vesicles 2014; 3: 24641.

29.

van Niel

D’Angelo

Raposo

. Shedding light on the cell biology of extracellular vesicles. Nat Rev Mol Cell Biol 2018; 19: 213–228.

30.

Doyle

Wang

. Overview of extracellular vesicles, their origin, composition, purpose, and methods for exosome isolation and analysis. Cells 2019; 8: 727.

31.

Ramos-Zaldívar

Polakovicova

Salas-Huenuleo

, et al. Extracellular vesicles through the blood–brain barrier: a review. Fluids Barriers CNS 2022; 19: 60.

32.

Heinrich

Purcell

Pruijssers

, et al. Endogenous double-stranded Alu RNA elements stimulate IFN-responses in relapsing remitting multiple sclerosis. J Autoimmun 2019; 100: 40–51.

33.

Crooke

3rd Tossberg

Porter

, et al. Cutting edge: reduced adenosine-to-inosine editing of endogenous Alu RNAs in severe COVID-19 disease. J Immunol 2021; 206: 1691–1696.

34.

Crooke

3rd Tossberg

Porter

, et al. Reduced A-to-I editing of endogenous Alu RNAs in lung after SARS-CoV-2 infection. Curr Res Immunol 2021; 2: 52–59.

35.

Knopman

Amieva

Petersen

, et al. Alzheimer disease. Nat Rev Dis Primers 2021; 7: 33.

36.

Crooke

3rd Tossberg

Heinrich

, et al. Reduced RNA adenosine-to-inosine editing in hippocampus vasculature associated with Alzheimer’s disease. Brain Commun 2022; 4: fcac238.

37.

Luo

Liu

, et al. Long RNA Profiles of human brain extracellular vesicles provide new insights into the pathogenesis of Alzheimer’s disease. Aging Dis 2023; 14: 229–244.

38.

Yang

Vest

Kern

, et al. A human brain vascular atlas reveals diverse mediators of Alzheimer’s risk. Nature 2022; 603: 885–892.

39.

Scheckel

Drapeau

Frias

, et al. Regulatory consequences of neuronal ELAV-like protein binding to coding and non-coding RNAs in human brain. Elife 2016; 5: e10421.

40.

Zhang

Yan

, et al. SPRINT: an SNP-free toolkit for identifying RNA editing sites. Bioinformatics 2017; 33: 3538–3548.

41.

Zhang

Lin

, et al. Evaluation and comparison of computational tools for RNA-Seq isoform quantification. BMC Genomics 2017; 18: 583.

42.

Dagan

Sorek

Sharon

, et al. A database of alu elements incorporated within protein-coding genes. Nucleic Acids Res 2004; 32: D489–D492.

43.

Love

Huber

Anders

. Moderated estimation of fold change and dispersion for RNA-Seq data with DESeq2. Genome Biol 2014; 15: 550.

44.

Cao

Xue

. Enhancer RNA: biogenesis, function, and regulation. Essays Biochem 2020; 64: 883–894.

45.

Lam

MTY

Cho

Lesch

, et al. Rev-Erbs repress macrophage gene expression by inhibiting enhancer-directed transcription. Nature 2013; 498: 511–515.

46.

Notari

, et al. Functional roles of enhancer RNAs for oestrogen-dependent transcriptional activation. Nature 2013; 498: 516–520.

47.

Schaukowitch

Joo

J-Y

Liu

, et al. Enhancer RNA facilitates NELF release from immediate early genes. Mol Cell 2014; 56: 29–42.

48.

Lai

Orom

Cesaroni

, et al. Activating RNAs associate with Mediator to enhance chromatin architecture and transcription. Nature 2013; 494: 497–501.

49.

Zhou

Chen

, et al. GRID-seq reveals the global RNA–chromatin interactome. Nat Biotechnol 2017; 35: 940–950.

50.

Henninger

Oksuz

Shrinivas

, et al. RNA-mediated feedback control of transcriptional condensates. Cell 2021; 184: 207–225.e24.

51.

Huang

Sherman

Lempicki

. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 2009; 4: 44–57.

52.

Hodson

. Alzheimer’s disease. Nature 2018; 559: S1.

53.

Heppner

Ransohoff

Becher

. Immune attack: the role of inflammation in Alzheimer disease. Nat Rev Neurosci 2015; 16: 358–372.

54.

Roy

Cao

. Antiviral immune response in Alzheimer’s disease: connecting the dots. Front Neurosci 2020; 14: 577744.

55.

Roy

Wang

Wan

, et al. Type I interferon response drives neuroinflammation and synapse loss in Alzheimer disease. J Clin Invest 2020; 130: 1912–1930.

56.

Penzes

VanLeeuwen

J-E

. Impaired regulation of synaptic actin cytoskeleton in Alzheimer’s disease. Brain Res Rev 2011; 67: 184–192.

57.

Upadhya

Hegde

. Role of the ubiquitin proteasome system in Alzheimer's disease. BMC Biochemistry 2007; 8: S12.

58.

Nemes

Devreese

Steinert

, et al. Cross-linking of ubiquitin, HSP27, parkin and synuclein by γ-glutamyl-ε-lysine bonds in Alzheimer's neurofibrillary tangles. FASEB J 2004; 18: 1135–1137.

59.

Rao

Ganaraja

Murlimanju

, et al. Hippocampus and its involvement in Alzheimer’s disease: a review. 3 Biotech 2022; 12: 55.

60.

Gao

Liu

Wang

, et al. Proteomic analysis of human hippocampal subfields provides new insights into the pathogenesis of Alzheimer’s disease and the role of glial cells. Brain Pathol 2022; 32: e13407.

61.

Montagne

Barnes

Sweeney

, et al. Blood-brain barrier breakdown in the aging human hippocampus. Neuron 2015; 85: 296–302.

62.

Obermeier

Daneman

Ransohoff

. Development, maintenance and disruption of the blood-brain barrier. Nat Med 2013; 19: 1584–1596.

63.

Iadecola

. The pathobiology of vascular dementia. Neuron 2013; 80: 844–866.

64.

Maruszak

Thuret

. Why looking at the whole hippocampus is not enough—A critical role for anteroposterior axis, subfield and activation analyses to enhance predictive value of hippocampal changes for Alzheimer’s disease diagnosis. Front Cell Neurosci 2014; 8: 95.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

8.84 MB