Aggregates of α-synuclein are the hallmark of synucleinopathies such as Parkinson’s disease and dementia with Lewy bodies, but could other protein inclusions contribute to pathology? Yes, say scientists led by Joseph Mazzulli at Northwestern University in Chicago. In the May 17 Neuron, they described how two RNA-binding proteins—one called non-POU domain-containing octamer-binding protein, the other called splicing factor proline- and glutamine-rich—accumulate in iPSC-derived neurons from a person with PD.

  • In PD and DLB, RNA-binding proteins NONO and SFPQ aggregate in the nucleus.
  • Aberrant mRNA editing then traps transcripts in the nucleus.
  • Neurons die as synaptic and mitochondrial proteins peter out.

The authors report that NONO and SPFQ aggregates unleashed an often-overlooked type of RNA editing, that is, the conversion of adenosine to inosine. When bound to NONO/SFPQ aggregates, the A-to-I-edited mRNAs got trapped in the nucleus and their translation stalled, leading to neurodegeneration. Most of the affected transcripts encode axonal, synaptic, and mitochondrial proteins. The scientists also found puncta of NONO/SFPQ and edited mRNA in cortical neurons from people with DLB, indicating that this pathological cascade happens in vivo as well.

David Sulzer of Columbia University in New York called this study well-done and surprising. “It should encourage more research into the consequences of the localization of synuclein within the nucleus, which has long been noted but remains poorly understood,” he wrote (comment below).

a-Synuclein accumulation disrupts proteostasis by slowing protein degradation (Jun 2006 news; Jun 2011 news; Mar 2014 conference news). Curious if other proteins aggregate as a result, first author Nandkishore Belur extracted insoluble proteins from midbrain neurons made of iPSCs from a person carrying the A53T synuclein mutation. NONO and SFPQ were the most abundant aggregates. These RNA-binding proteins form so-called paraspeckles in the nucleus. They bind RNA to turn down translation when cells are under duress. Paraspeckles work similarly to stress granules in the cytosol, which can incorporate other proteins linked to neurodegeneration, such as FUS and TDP43 (Feb 2024 news; Jul 2010 news).

Unlike the RNA-binding proteins FUS and TDP43, NONO and SFPQ aggregated in nuclei in the A53T neurons. Ditto in cortical tissue from four people who had had DLB (image below). Each had an average of 30-fold more of the insoluble RNA-binding proteins in nuclei than did age- and sex-matched controls (image below).

Going Nuclear. In cortical tissue from healthy people (top), RNA-binding proteins SFPQ (red) and NONO (green) form puncta in nuclei (blue). In DLB cortex (bottom), the aggregates were larger and more plentiful. [Courtesy of Belur et al., Neuron, 2024.]

Confocal microscopy of iPSC neurons and brain tissue showed NONO and SFPQ aggregating both separately and together with half of the nuclear inclusions co-localizing with α-synuclein. In vitro, recombinant α-synuclein induced soluble, recombinant SFPQ to aggregate into Thioflavin T-positive pellets, suggesting that nuclear α-synuclein might spur NONO and SFPQ to misfold and accumulate.

What might happen when NONO/SFPQ aggregate? The authors believe this drives RNA editing. Belur found that the A53T neurons had 21 percent more adenosine-to-inosine edits across 3,100 susceptible nucleosides than did control neurons. What role A-to-I editing serves is unclear, but it can alter translation, and possibly protein function, because the ribosome machinery reads inosine as guanosine (reviewed by Yang et al., 2021). A-to-I editing occurs mainly on retrotransposon sequences called Alu-repeat regions, so-called for the restriction endonuclease that recognizes and cuts those elements. Alu-repeats are unique to primates.

Toxic Cycle. In the neuronal nucleus, α-synuclein oligomers trigger NONO and SFPQ to aggregate (1), which decreases the amount of soluble SFPQ (2). This prevents ADAR3 transcription, ultimately increasing A-to-I RNA editing of mRNAs containing retrotransposon Alu elements (3). The inosine-containing RNAs bind tighter to NONO and SFPQ (4), which fuels their aggregation and prevents transcript transport to the cytosol (5). This decreases corresponding protein levels, namely synaptic and mitochondrial ones, leading to neurodegeneration (6). [Courtesy of Belur et al., Neuron, 2024.]

The authors tied the boost in RNA editing to SFPQ aggregation. This RNA-binding protein turns on expression of adenosine deaminase acting on RNA 3 (Hirose et al., 2014). ADAR3 is one of three deaminases. While ADAR1 and ADAR2 replace the amine group on adenosine with a carbonyl, converting it to inosine, ADAR3 lacks catalytic activity. Instead, it competes with ADAR1/2 to prevent editing (Chen et al., 2000). With SFPQ sequestered in aggregates, A53T neurons had 75 percent less ADAR3 mRNA and protein than control neurons, explaining the uptick in A-to-I editing. DLB brain lysates had 70 percent less ADAR3 than lysates from control brain.

Which RNAs are susceptible to the deaminases? Axon and synaptic transcripts are substrates of these enzymes; indeed ADAR3 is only expressed in the brain. Belur found that of the 50 most extensively edited mRNAs in A53T neurons, many encoded synaptic, axonal, and mitochondrial proteins, including the NMDA receptor subunit GRIN2D, the ALS/FTD-linked dipeptidyl peptidase like 6 (DPP6), and the AD-linked protein importer TOMM40. The neurons had a dearth of these proteins. While two-thirds of the total edited RNA in control iPSC neurons were the nucleus, almost all the edited RNA in PD neurons was there. In other words, edited RNAs did not reach the cytosol for translation.

Complicating the story further, Belur found that A-to-I edited transcripts bound to recombinant SFPQ in vitro, triggering it to aggregate. In control iPSC-derived neurons, expressing hyperactive mutants of ADAR1 and ADAR2 did the same. In contrast, treating iPSC A53T neurons with the ADAR1 inhibitor 8-azaadenosine decreased A-to-I RNA editing, NONO/SFPQ aggregates, and synapse death. Taken together, it appears as if inosine-containing RNA fuels SFPQ aggregation, suppressing ADAR3 production and, in turn, boosting A-I editing in a vicious cycle (image above).

Sulzer offered a different interpretation. “Alternatively, it is possible that these nuclear structures, as suspected for cytosolic Lewy body inclusions, may be part of a stress response that plays a protective role,” he wrote.

For his part, Mark Cookson of the NIH in Bethesda, Maryland, found the parallels between FUS and RNA stress granules in ALS/FTD and NONO-SFPQ paraspeckles in synucleinopathies particularly interesting. “I wonder if RNA dysregulation is a relatively later, common pathway for neuronal death and therefore downstream of initiating events such as a gene mutation,” he wrote (comment below).

Why haven’t these NONO/SFPQ inclusions shown up before? They do not form in mice expressing human A53T α-synuclein. Mazzulli suspects this is because A-to-I editing of the Alu elements occurs specifically in primates, which means fewer inosine-RNAs to ensnare and aggregate NONO/SFPQ. He thinks this may be why mouse models do not fully capture disease pathologies and phenotypes of synucleinopathies.

The researchers will look in PD/DLB autopsy samples to see what brain areas contain nuclear NONO/SFPQ inclusions. They are also screening small molecules in silico to find inhibitors of ADAR1 or ADAR2, which could suppress A-to-I editing and NONO/SFPQ aggregates.—Chelsea Weidman Burke

Comments

  1. This is a well-done and surprising study, indicating a new form of protein inclusion in PD and synucleinopathy patients, namely, nuclear aggregates consisting of RNA-editing proteins and/or two RNA-binding proteins. The formation of these impressively large, and apparently common, nuclear structures appears to be downstream of a step in α-synuclein handling, and while misfolded synuclein has been shown to bind hundreds of proteins, the presence of these newly discovered aggregates in genuine disease is dramatic. It should encourage more research into consequences of the localization of synuclein with the nucleus, which has long been noted, but remains poorly understood.

    The functional consequences of the aggregates are not yet clear, and while the authors have done an excellent job of determining that their function in editing RNA could potentially play a role in neuronal stress, there are additional possibilities, including effects on trafficking and/or the cytoskeleton, which could be required for the aggregate formation.

    Given their abundance, the aggregates may also be a source of neoantigens. Alternatively, it is possible that these nuclear structures, as suspected for cytosolic Lewy body inclusions, may be a stress response that plays a protective role. In any case, this report spawns exciting new questions, the mark of a good discovery, and may hold an important piece of the puzzle as to what causes these disorders.

  2. Belur et al. report a novel consequence of nuclear synuclein oligomers, namely the accumulation of NONO and SFPQ, leading to aberrant editing of Alu elements and the formation of novel inclusions in SNCA p.A53T ISPC lines. There is prior evidence that nuclear synuclein may affect RNA processing, notably work from the Khurana lab (Hallacli et al., 2022), showing in triplication iPSC lines that non-membrane bound synuclein can interact with P-bodies, which are involved in mRNA storage and turnover.

    Put together, these studies suggest that dysregulation of mRNA processing may be a relatively major consequence of synuclein mutations. That a consistent shift in solubility is now reported in brains from donors who had dementia with Lewy bodies suggests that a similar process may occur in vivo.

    One area that I found particularly interesting was the change in proteins that are related to other neurodegenerative diseases, notably FUS, a gene mutated in ALS-FTD (reviewed in Abramzon et al., 2020). Given that PD and ALS-FTD are clinically and pathologically distinctive, I wonder if RNA dysregulation is a relatively later common pathway for neuronal death and therefore downstream of initiating events such as a gene mutation.

    If correct, one might expect to see converging events related to RNA processing and editing across different disease groups. In this context, it is worth noting that we have generated multiple iPSC lines in a consistent genetic background including SNCA p.A53T and multiple FUS variants. It would be of great interest to examine RNA processing, editing, and translation across multiple lines to confirm the reported phenotypes and evaluate when convergence begins to emerge.

    References:

    . The Parkinson's disease protein alpha-synuclein is a modulator of processing bodies and mRNA stability. Cell. 2022 Jun 9;185(12):2035-2056.e33. PubMed.

    . The Overlapping Genetics of Amyotrophic Lateral Sclerosis and Frontotemporal Dementia. Front Neurosci. 2020;14:42. Epub 2020 Feb 5 PubMed.

  3. I very much enjoyed reading this important paper from Belur, Mazzulli, and colleagues. It reinforces the emerging function of α-synuclein in gene regulation and RNA metabolism, and a potentially important role of dysregulation of these pathways in the pathogenesis of synucleinopathies. More generally, it draws attention to how complex and widespread the consequences of protein misfolding and mislocalization are in the cell. This phenomenon was well documented by Haenig and colleagues some years ago when they showed that protein-protein interaction mapping could reveal proteins that co-aggregate in degenerative proteinopathies (Haening et al., 2020). Among such proteins, RNA-binding proteins (RBPs) are ideal candidates to undergo co-aggregation. These proteins are rich in prion domains and primed to undergo phase separation. Interestingly, Brunet, Jarosz, and colleagues very recently showed that RBPs comprise the major class of proteins that become progressively insoluble in the aging vertebrate brain (Harel et al., 2024). Like the current paper, these others powerfully exploited the growing toolbox of systems cell biology to gain a more global view of how protein misfolding perturbs the cellular proteome.

    The current paper specifically identifies proteins that become insoluble in aging iPSC-derived dopaminergic neurons as α-synuclein misfolds. As noted in the Alzforum news piece, foremost among the many proteins recovered were the RBP proteins NONO and SFPQ. Importantly, these proteins themselves then mislocalize and aggregate in the nucleus, and are sequestered away from their normal functions, leading to alteration of RNA-specific adenosine deaminase (ADAR) function. Altered RNA editing and mRNA translation ensues, which correlates with dysregulation of key neuronal proteins.

    In prior papers, we have adopted different systems-cell-biology approaches yet hit strikingly similar findings. NONO and SFPQ are among many RBPs that physically interact with a-synuclein, alter their localization when α-synuclein accumulates, and impact its toxicity when genetically manipulated (Chung et al., 2017; Khurana et al., 2017; Hallacli et al., 2022; Lam et al., 2022; Feb 2017 news; Jun 2022 news). While our own focus was drawn to mRNA stability and mRNA translation, the current paper elegantly extends this to RNA editing. I am struck by the strong overlap among specific proteins and protein classes recovered by these very different unbiased systems cell biology assays, conserved across species and cell types, all, in my mind, strengthening conviction in the biological connections.

    How important are RBP perturbations for disease risk and progression? This will need further investigation. In the current paper, strong data tied altered base editing to alterations in levels of key neuronal proteins. The impact of these perturbations on neuronal health and viability can now be more directly tested with genetic manipulation in neurons and targeted human genetic studies. As we have previously reported, cumulative mutations in RBP pathways may give rise to stronger genetic signals than individual gene variation. Finding these signals may require new statistical genetic methods. If specific mRNA targets are repeatedly implicated, these may prove to be disease-relevant targets. In his comment, Mark Cookson raises the important question of how specific these changes in RNA metabolism are in different proteinopathies, and whether these are early or late. My own hunch is that these are early and specific effects. To give one example, we have shown previously that the effects of Ataxin-2 manipulation on TDP-43 and α-synuclein toxicities are strong but diametrically opposed (Khurana et al., 2017), and the risk signal we saw for P-body genes in synucleinopathies was absent in ALS. But more needs to be done to address his important question.

    Finally, the paper nicely reminds us that our tidy view that many neurodegenerative diseases equate to the misfolding of one or a handful of proteins may be overly simplistic. Proteins fold, function, and traffic in an astonishingly crowded and harsh environment. As one protein misfolds and mislocalizes, there are corresponding shifts in many other proteins that can mislocalize, themselves misfold, or become destabilized. And, of course, these effects can extend to RNA, lipids, and other metabolites, etc. “Mixed pathologies” thus turn out to be far more extensive than in the standard neuropathological lens through which we view it! And this complexity may in turn be critical for understanding the vexing heterogeneity among patients with neurodegenerative disease.

    References:

    . Interactome Mapping Provides a Network of Neurodegenerative Disease Proteins and Uncovers Widespread Protein Aggregation in Affected Brains. Cell Rep. 2020 Aug 18;32(7):108050. PubMed.

    . Identification of protein aggregates in the aging vertebrate brain with prion-like and phase-separation properties. Cell Rep. 2024 May 24;:112787. PubMed.

    . In Situ Peroxidase Labeling and Mass-Spectrometry Connects Alpha-Synuclein Directly to Endocytic Trafficking and mRNA Metabolism in Neurons. Cell Syst. 2017 Feb 22;4(2):242-250.e4. Epub 2017 Jan 25 PubMed.

    . Genome-Scale Networks Link Neurodegenerative Disease Genes to α-Synuclein through Specific Molecular Pathways. Cell Syst. 2017 Feb 22;4(2):157-170.e14. Epub 2017 Jan 25 PubMed.

    . The Parkinson's disease protein alpha-synuclein is a modulator of processing bodies and mRNA stability. Cell. 2022 Jun 9;185(12):2035-2056.e33. PubMed.

    . Rapid iPSC inclusionopathy models shed light on formation, consequence and molecular subtype of alpha-synuclein inclusions. 2022 Nov 09 10.1101/2022.11.08.515615 (version 1) bioRxiv.

    . Genome-Scale Networks Link Neurodegenerative Disease Genes to α-Synuclein through Specific Molecular Pathways. Cell Syst. 2017 Feb 22;4(2):157-170.e14. Epub 2017 Jan 25 PubMed.

Make a Comment

To make a comment you must login or register.

References

News Citations

  1. ER-Golgi Traffic Jam Explains α-Synuclein Toxicity
  2. Feedback Loop—Molecular Mechanism for PD, Gaucher’s Connection
  3. Protecting Neurons by Ramping Up Waste Disposal?
  4. No Loitering: Pathogenic Liaisons Trap TDP-43 in the Cytoplasm
  5. Going Nuclear: First Function for FUS Mutants

Paper Citations

  1. . Adenosine-to-inosine RNA editing in neurological development and disease. RNA Biol. 2021 Jul;18(7):999-1013. Epub 2021 Jan 6 PubMed.
  2. . NEAT1 long noncoding RNA regulates transcription via protein sequestration within subnuclear bodies. Mol Biol Cell. 2014 Jan;25(1):169-83. Epub 2013 Oct 30 PubMed.
  3. . A third member of the RNA-specific adenosine deaminase gene family, ADAR3, contains both single- and double-stranded RNA binding domains. RNA. 2000 May;6(5):755-67. PubMed.

External Citations

  1. paraspeckles 

Further Reading

No Available Further Reading

Primary Papers

  1. . Nuclear aggregates of NONO/SFPQ and A-to-I-edited RNA in Parkinson's disease and dementia with Lewy bodies. Neuron. 2024 May 16; PubMed.