Hidden RNA Errors Found at the Synapse in Frontotemporal Dementia

Deep inside the brain’s frontal cortex, where frontotemporal dementia steadily erodes personality, judgment, and language, scientists have now zoomed in on one of the smallest and most vulnerable structures in the nervous system: the synapse. A new study published in Acta Neuropathologica has delivered the most detailed molecular portrait yet of what goes wrong at these neuronal junctions in people carrying a mutation in the C9ORF72 gene, the single most common genetic cause of both frontotemporal dementia (FTD) and amyotrophic lateral sclerosis (ALS). The findings reveal an unexpected twist: faulty RNA molecules, long thought to be trapped in the cell’s nucleus, actually travel out to the synapses themselves, suggesting that the molecular sabotage driving this devastating disease reaches all the way into the brain’s communication hubs.

The research team, led by investigators at Barrow Neurological Institute in Phoenix alongside collaborators at the Translational Genomics Research Institute and City of Hope, tackled a persistent blind spot in dementia research. Most studies of diseased brain tissue grind up whole chunks of cortex, producing an average signal in which the synapse’s voice is drowned out by everything else. To overcome this, the researchers isolated synaptosomes, tiny sealed sacs of presynaptic and postsynaptic machinery that pinch off during gentle tissue disruption and can be purified by centrifugation. From frozen frontal cortex samples of ten patients with confirmed C9ORF72-linked FTD and ten neurologically normal donors, obtained through the Queen Square Brain Bank in London, they harvested these synaptic compartments and subjected them to parallel mass spectrometry proteomics and RNA sequencing.

The technical validation was rigorous. Western blots confirmed that the purified fractions were loaded with presynaptic proteins such as synaptophysin and postsynaptic markers like Homer1, while transmission electron microscopy revealed intact membrane-enclosed terminals studded with synaptic vesicles and electron-dense postsynaptic material. Principal component analyses showed that fraction identity, not random variation, was the dominant source of variance in both the protein and RNA datasets, and nuclear markers were largely absent from the synaptic fractions. In other words, the team had genuinely captured the synapse, not a contaminated soup, giving them a trustworthy window onto synaptic biology in the diseased human brain.

What that window revealed was striking. The proteomic analysis identified 1,324 proteins at altered abundance in C9ORF72-FTD synaptosomes, with 555 increased and 769 decreased relative to controls. Pathway analysis using the synapse-specific SynGO database showed that the downregulated proteins clustered heavily around core neurotransmission machinery: the synaptic vesicle cycle, chemical synaptic transmission, and trans-synaptic signaling. Upregulated proteins, by contrast, were modestly enriched for structural remodeling processes such as synapse assembly and postsynaptic actin organization, a pattern the authors interpret as evidence of widespread architectural rebuilding layered on top of failing communication machinery. Many of the altered proteins were unique to the synaptic fraction, and some even shifted in opposite directions between whole tissue and synapses, hinting at localized redistribution of proteins that whole-tissue studies would completely miss.

The RNA side of the story was just as dramatic. Sequencing of synaptosomal RNA uncovered 2,835 differentially expressed protein-coding genes, including 1,763 upregulated and 1,072 downregulated transcripts. When the researchers compared RNA changes with protein changes in the same synaptic compartments, the correlation was only moderate, with a Pearson coefficient of 0.33. Nearly forty percent of the altered genes showed discordant behavior between transcript and protein levels, a signature of broken post-transcriptional regulation inside diseased synapses. In healthy neurons, local RNA and protein abundance are tightly coordinated to keep synaptic transmission running; in C9ORF72-FTD, that coordination visibly frays.

To test whether these brain-tissue signatures could be modeled in the laboratory, the team differentiated induced pluripotent stem cells from C9ORF72-FTD patients and healthy controls into cortical neurons and matured them for one hundred days in culture. The patient-derived neurons showed clear structural deficits: the density of postsynaptic Homer1 puncta at excitatory synapses was reduced, gephyrin puncta at inhibitory synapses were diminished, and the number of fully colocalized inhibitory synapses dropped significantly, even as the corresponding presynaptic markers remained intact. Longitudinal recordings on multielectrode arrays captured a progressive decline in spontaneous network firing between days 65 and 100, indicating worsening hypoexcitability, although the synchrony of network activity was preserved. Notably, this hypoexcitability contrasts with the hyperexcitability often reported in ALS motor neuron models, underscoring that neuronal subtype and disease phenotype shape how the same genetic mutation plays out at the circuit level.

Critically, synaptosomes isolated from these cultured neurons reproduced key molecular pathways seen in the patient brain. The iPSC-derived synaptic proteome contained 962 differentially abundant proteins, and its RNA profile showed 402 differentially expressed protein-coding genes. When the researchers integrated all four datasets, brain protein, brain RNA, iPSC protein, and iPSC RNA, a set of conserved pathways emerged: synapse organization, synapse assembly, regulation of postsynaptic neurotransmitter receptor levels, and trans-synaptic signaling. Among the consistently downregulated genes were members of the neuronal pentraxin family, NPTX1 and NPTXR, which are known to support excitatory synapse formation and have previously been flagged as synaptic biomarkers in the cerebrospinal fluid of genetic FTD patients.

The most surprising discovery, however, came from a targeted search for cryptic exons. When the RNA-binding protein TDP-43 is lost from the nucleus, a hallmark of nearly all FTD and ALS cases, it stops repressing aberrant splice sites, and strange pseudo-exons get stitched into mature transcripts, often destroying the proteins they encode. Cryptic exons in genes like UNC13A and STMN2 are well-documented markers of this process, but they were assumed to be a nuclear phenomenon. The new data show otherwise. In patient synaptosomes, cryptic exon inclusion in STMN2 was significantly increased, detected in eight of ten C9ORF72-FTD cases, and cryptic transcripts of the synaptic scaffolding gene KALRN were also present in a subset of cases. Strikingly, UNC13A cryptic exons were abundant in whole-tissue homogenates but essentially absent from the synaptic fraction, with STMN2 showing the highest synaptic enrichment ratio. This transcript-specific pattern suggests that some aberrantly spliced RNAs are selectively trafficked or retained at synapses rather than randomly redistributed.

The implications ripple outward. STMN2, also known as stathmin-2, is normally transported to distal neuronal compartments where it supports axonal maintenance and repair; KALRN encodes a scaffolding protein central to synaptic signaling and actin remodeling. If cryptic versions of these RNAs accumulate at synapses, they could disrupt local protein synthesis or RNA homeostasis in ways that directly undermine synaptic function, potentially linking TDP-43 loss of function to the synapse loss observed in living patients through neuroimaging and in postmortem tissue. The authors caution that the presence of these transcripts at the synapse does not yet prove they are locally translated or actively toxic, and future work with spatial transcriptomics and high-resolution imaging will be needed to pin down their exact sub-synaptic location and fate.

Limitations remain, as they do in any study of end-stage human disease. The postmortem synaptosomes represent surviving synapses rather than those already eliminated, so some observed changes may reflect compensatory responses rather than primary causes of degeneration. The iPSC cultures, meanwhile, are simplified neuron monocultures lacking the glial cells that shape synapses in the living brain, and they did not show TDP-43 mislocalization or cryptic exons, highlighting gaps in the model. Even so, the convergence of findings across independent systems lends strong support to a reframed view of C9ORF72-FTD: not simply a disease of dying neurons, but a disease in which the synapse, the brain’s most information-dense structure, becomes a site of RNA-level corruption. If synaptic RNA misprocessing proves to be an early event, it could open a new frontier for therapies aimed at preserving synaptic integrity before the damage becomes irreversible.

Subject of Research: Synaptosome multi-omics of C9ORF72-linked frontotemporal dementia reveals synaptic protein and RNA dysregulation and synaptic localization of TDP-43-associated cryptic exon transcripts

Article Title: Integrative synaptosome multi-omics reveals disrupted synapse organization and localized cryptic transcripts in C9ORF72-frontotemporal dementia

Article References: Spillman, A. M., Alsop, E. B., Gittings, L. M., Garcia-Mansfield, K., Piras, I., Bonfitto, A., Martinez, M. N., Sharma, R., Preller, K. R., Huentelman, M., Pirrotte, P., Van Keuren-Jensen, K., & Sattler, R. (2026). Integrative synaptosome multi-omics reveals disrupted synapse organization and localized cryptic transcripts in C9ORF72-frontotemporal dementia. Acta Neuropathologica, 152(1), Article 41. https://doi.org/10.1007/s00401-026-03084-5

Image Credits: AI Generated

DOI: 10.1007/s00401-026-03084-5

Keywords: frontotemporal dementia, C9ORF72, ALS, synaptosome, TDP-43, cryptic exons, proteomics, transcriptomics, iPSC cortical neurons, synaptic dysfunction, STMN2, UNC13A