Many genetic variants predispose to Alzheimer’s disease, but exactly how remains a mystery. In the October 8 Nature Genetics, researchers led by Philip De Jager at Columbia University and Towfique Raj at Icahn School of Medicine at Mount Sinai, both in New York, and David Bennett at Rush University in Chicago implicate mRNA splicing. The authors analyzed gene-expression patterns in 450 postmortem brains from AD patients and controls. The amount of plaques and tangles correlated with alternative and mis-splicing of a specific set of transcripts. Many of these genes had not been associated with AD previously. The authors re-created some of the splicing changes by exposing cultured cells to high levels of phosphorylated tau, suggesting pathology bungles RNA processing. In a separate analysis, other splicing patterns correlated with genetic variants, in particular those in the known AD genes CLU, PICALM, and PTK2B. “For those genes, I think we’ve identified the proximal mechanism by which they alter cell function and eventually lead to dementia,” De Jager told Alzforum.
- AD brains accumulate a set of alternatively spliced transcripts.
- These come from known AD genes as well as new ones.
- Some GWAS hits promote disease by causing splicing errors.
Others hailed the findings. “This is the most comprehensive analysis of RNA splicing in AD patients to date. It provides convincing evidence that splicing plays a role in the disease,” Junmin Peng at St. Jude Children’s Research Hospital in Memphis, Tennessee, told Alzforum.
Immature RNA transcripts are processed by spliceosomes in the nucleus, where introns are snipped out to form the mature message. For some genes, alternative splicing produces more than one type of mRNA. Previous proteomics work by Peng and researchers at Emory University suggested there might be splicing problems in AD brain, because they found protein aggregates of spliceosome components (Sep 2013 news). In addition, older studies had reported mis-splicing of amyloid precursor protein and tau in AD brain (Rockenstein et al., 1995; Buée et al., 2000). However, it was unclear what role, if any, these changes played in pathogenesis.
Comparing Transcripts, Genes, and Pathology. The authors correlated RNA-Seq results with genetic variations and neuropathology to identify factors that influence splicing in AD. [Courtesy of Raj et al., Nature Genetics.]
De Jager and colleagues set out to catalog the splicing changes in AD. First author Raj mined RNA sequencing data from two longitudinal aging studies, the Religious Orders Study (ROS) and the Memory and Aging Project (MAP), both in Chicago. In the former, the 243 participants had died at an average age of 88, and in the latter, the 207 participants had died at an average of 89. In both cohorts, about 40 percent of participants were diagnosed with AD during life, and 60 percent turned out to have AD pathology in the brain at autopsy. Raj and colleagues found numerous splicing changes in the dorsolateral prefrontal cortices of AD brains. For some transcripts, they saw alterations in the relative levels of different isoforms, and for others they saw mis-splicing, where introns were retained in the mature mRNAs. Overall, splicing changes in 84 genes correlated with a diagnosis of AD, while splicing changes in a subset of 67 genes were associated with specific neuropathologies such as neuritic plaques, tau tangles, and amyloid load. For most of these 84 genes, transcript levels were no different between AD and control brains, suggesting they would have been missed in studies of gene expression.
Among the mis-spliced forms, retention of an intron in the phosphofructokinase (PFKP) gene associated most strongly with AD. This isoform independently correlated with diagnosis and each type of pathology. The cancer gene NDRG2 had the next-strongest association, correlating with both amyloid and tau. Known AD genes, like APP, PICALM, and CLU, popped up in the analysis as well. The authors replicated the findings using RNA expression data from 301 brains in the Mount Sinai Brain Bank. They found similar splicing changes in 52 of the 84 genes, with a similarly strong association with AD as in the ROS/MAP cohorts.
What drives these splicing changes? One clue came from an in vitro experiment in which the authors overexpressed tau in human neurons derived from induced pluripotent stem cells. These neurons accumulated phosphorylated tau and reproduced the most notable AD splicing changes, including alterations in APP, PICALM, and NDRG2 mRNA. De Jager noted that in a separate study, his group implicated tau pathology in altering chromatin packing and gene expression (Klein et al., 2018). “I think the changes in the epigenome propagate through the spliceosome as well,” De Jager suggested. He thinks tau pathology is probably the main driver of splicing abnormalities, and plans to explore this further in iPSC lines generated from 50 of the ROS/MAP participants in this study. This will allow him to directly test if the splicing pattern caused by tau overexpression in cells replicates that seen in vivo.
While these data suggested splicing changes are a general feature of AD, the authors wondered whether specific genetic risk factors could also promote mis-splicing or alternative splicing. They analyzed genetic data from their cohort, and found around 9,000 single-nucleotide polymorphisms in 3,000 genes that associated with altered splicing. They dubbed these SNPs splicing quantitative trait loci (sQTLs). The authors had previously tied about a third of these sQTLs to epigenetic alterations such as methylation and histone acetylation, again suggesting that regulation of gene expression might affect splicing as well (Ng et al., 2017).
What about AD-specific variants? Numerous variants found in previous AD genome-wide association studies (GWAS) turned up in the 9,000 sQTLs, hinting they might affect splicing. To confirm this, the authors correlated whole transcriptome patterns with AD diagnosis using data from the International Genomics of Alzheimer’s Project. They inferred gene expression in IGAP from SNP data, using the ROS/MAP transcriptome data set as a model. With this method, they found 21 genes with splicing changes that associated with the disease. These included the known AD genes PICALM, CLU, SPI1, CR1, PTK2B, and MTCH2. For PICALM, CLU, and PTK2B, the authors identified a specific isoform that may account for the gene’s effect on AD risk. For example, for PTK2B they found that an AD-associated G to A change in an intron caused it to be retained more frequently in the mature transcript. The authors also confirmed the previously reported association between full-length CD33 and AD (Aug 2013 news; Raj et al., 2014).
The transcriptome-wide association analysis also turned up several genes not previously linked to AD, although the IGAP data had suggested as much: AP2A1, AP2A2, FUS, MAP1B, and TBC1D7. A meta-analysis of IGAP and U.K. Brain Bank data, again imputing gene expression from SNPs, identified three more genes: the known AD gene ABCA7 and new genes RHBDF1 and VPS53. In addition, many of the IGAP genes replicated in the UKBB dataset.
These new genes code for proteins that interact with products of known AD genes. In particular, many are involved in endocytosis and lysosomal degradation, pathways often implicated in AD (see image above). AP2A1 and MAP1B are hub proteins in these networks, and are expressed at low levels in AD brains, the authors noted (Jul 2017 news). Mis-spliced mRNAs may lead to the production of aberrant proteins that disrupt these processes, they contend.
“I’m excited about this analytic approach,” Asa Abeliovich, founder and CEO of Prevail Therapeutics in New York City, told Alzforum. He believes that cataloging splicing changes will be a useful tool for identifying new genes associated with AD. “This is a resource that will be tremendously helpful for the field.”
De Jager suggested that the data may offer targets for therapeutic intervention at either the transcript or protein level. Directly targeting splicing machinery might prevent or mitigate AD pathology, he said. He also wants to screen for compounds that reverse or block the effect of tau pathology on splicing in vitro. “We’re excited to leverage this information to see where it leads us in terms of drug development, and also to find out how these variants affect biological function to cause AD,” he said.—Madolyn Bowman Rogers
- Proteomics Point to RNA Splicing Defects in Alzheimer’s
- Protective Microglial Gene Variant Promotes Phagocytosis
- Alzheimer’s Proteomics Treasure Trove?
- Rockenstein EM, McConlogue L, Tan H, Power M, Masliah E, Mucke L. Levels and alternative splicing of amyloid beta protein precursor (APP) transcripts in brains of APP transgenic mice and humans with Alzheimer's disease. J Biol Chem. 1995 Nov 24;270(47):28257-67. PubMed.
- Buée L, Bussière T, Buée-Scherrer V, Delacourte A, Hof PR. Tau protein isoforms, phosphorylation and role in neurodegenerative disorders. Brain Res Brain Res Rev. 2000 Aug;33(1):95-130. PubMed.
- Klein HU, McCabe C, Gjoneska E, Sullivan SE, Kaskow BJ, Tang A, Smith RV, Xu J, Pfenning AR, Bernstein BE, Meissner A, Schneider JA, Mostafavi S, Tsai LH, Young-Pearse TL, Bennett DA, De Jager PL. Epigenome-wide study uncovers tau pathology-driven changes of chromatin organization in the aging human brain . bioRχiv. 2018 Feb 28.
- Ng B, White CC, Klein HU, Sieberts SK, McCabe C, Patrick E, Xu J, Yu L, Gaiteri C, Bennett DA, Mostafavi S, De Jager PL. An xQTL map integrates the genetic architecture of the human brain's transcriptome and epigenome. Nat Neurosci. 2017 Oct;20(10):1418-1426. Epub 2017 Sep 4 PubMed.
- Raj T, Ryan KJ, Replogle JM, Chibnik LB, Rosenkrantz L, Tang A, Rothamel K, Stranger BE, Bennett DA, Evans DA, De Jager PL, Bradshaw EM. CD33: increased inclusion of exon 2 implicates the Ig V-set domain in Alzheimer's disease susceptibility. Hum Mol Genet. 2014 May 15;23(10):2729-36. Epub 2013 Dec 30 PubMed.
- Johnson EC, Dammer EB, Duong DM, Yin L, Thambisetty M, Troncoso JC, Lah JJ, Levey AI, Seyfried NT. Deep proteomic network analysis of Alzheimer's disease brain reveals alterations in RNA binding proteins and RNA splicing associated with disease. Mol Neurodegener. 2018 Oct 4;13(1):52. PubMed.
- Raj T, Li YI, Wong G, Humphrey J, Wang M, Ramdhani S, Wang YC, Ng B, Gupta I, Haroutunian V, Schadt EE, Young-Pearse T, Mostafavi S, Zhang B, Sklar P, Bennett DA, De Jager PL. Integrative transcriptome analyses of the aging brain implicate altered splicing in Alzheimer's disease susceptibility. Nat Genet. 2018 Nov;50(11):1584-1592. Epub 2018 Oct 8 PubMed.