Introduction and Academic Background

Frontotemporal Dementia (FTD) and Amyotrophic Lateral Sclerosis (ALS) are two severe neurodegenerative diseases whose pathogenic mechanisms remain incompletely understood. In recent years, the RNA-binding protein TDP-43 (TAR DNA-binding protein 43) has been recognized as playing a central pathological role in both diseases: in the neurons of patients, TDP-43 is lost from the nucleus—translocating from the nucleus to the cytoplasm and abnormally accumulating in the brain and spinal cord—triggering downstream molecular dysfunctions. One key known function of TDP-43 is the negative regulation of “cryptic exon” inclusion, thereby protecting proper mRNA splicing and maintaining gene expression stability. However, whether TDP-43 participates in other RNA processing events, especially polyadenylation at the mRNA 3’ end, and whether its loss of function directly causes other RNA processing abnormalities, has lacked systematic research evidence so far.

Polyadenylation is one of the main RNA processing events following the transcription of the vast majority of human genes, affecting mRNA stability, export, and translation. In over 60% of genes, “alternative polyadenylation” (APA) occurs, meaning mRNA can be cleaved and tailed at different sites, generating 3’ untranslated regions (3’UTRs) of different lengths, thus regulating RNA stability, localization, and protein expression. Early studies detected TDP-43 binding not only in gene introns but also enriched in the 3’UTR, hinting it may participate in APA regulation. Yet, whether TDP-43 dysfunction in FTD/ALS directly induces abnormal APA, and whether APA abnormalities impact the expression of disease-relevant genes, remains an open question.

This study directly addresses these questions, aiming to reveal the landscape of neuronal APA changes upon TDP-43 loss-of-function and their molecular pathological significance, filling this key knowledge gap.

Paper Source and Author Information

This study, entitled “tdp-43 nuclear loss in ftd/als causes widespread alternative polyadenylation changes,” was published in the international top-tier journal Nature Neuroscience, November 2025, volume 28. The authors include Yi Zeng, Anastasiia Lovchykova, Tetsuya Akiyama, and others, primarily from Stanford University School of Medicine, Macquarie University, Mayo Clinic, and the Chan Zuckerberg Biohub. This work fully demonstrates the forefront level of interdisciplinary neurogenetics and molecular neurobiology.

Detailed Research Workflow

1. Research Subjects and Main Experimental Workflow

(1) Clinical Samples and Cell Models

The study first included postmortem brain tissue samples from FTD/ALS patients, and used human embryonic stem cell (hESC)-derived neuron models (iNeurons), as well as standard laboratory cell lines such as HEK293T cells. Neurons were efficiently differentiated by forced expression of NGN2 (a novel neurogenic transcription factor), enhancing the disease relevance and experimental controllability of the models.

(2) TDP-43 Loss/Mutation Manipulations

Two approaches were used to impair or alter TDP-43 function: first, gene knockdown via shRNA; second, introduction of known disease-causing mutants (K263E, M337V). Some experiments also analyzed publicly available data from patient-induced pluripotent stem cell-derived motor neurons, supporting the generality and disease relevance of the results.

(3) RNA Sequencing and High-Resolution Polyadenylation Site Detection

The team first used conventional RNA sequencing (RNA-seq) to analyze APA changes, applying two mainstream APA analysis software tools, Apalyzer and QAPA, which reference different polyadenylation site databases (PolyA_DB3 or PolyAsite 2.0) for quantitative analysis. Subsequently, the study innovatively employed the “3’ End-Seq” technique to comprehensively map polyadenylation sites (PolyA sites) at single nucleotide resolution, and self-developed filtering algorithms were used to remove false positive signals arising from random priming during reverse transcription, ensuring that captured APA events were precise and highly reliable.

(4) Functional Validation and Extension Experiments

The study further evaluated how APA events affect RNA levels, protein expression, and related signaling stability through qRT-PCR, competitive PCR, chemical inhibitor treatment (NMD inhibition), western blotting, and luciferase reporter assays. Pulse-chase experiments (using Actinomycin D) were used in some cases to assay mRNA half-life.

2. Data Analysis Methods and Software Algorithms

• RNA-seq Analysis: Expression quantification and differential analysis were performed using Salmon and DESeq2, while STAR was used for sequence alignment, and Leafcutter detected cryptic exon splicing events.
• APA Analysis:
  • RNA-seq-based APA events were analyzed and compared using Apalyzer and QAPA;
  • For 3’ End-Seq data, a self-developed LAPA long-read APA analysis algorithm was combined with a multi-layer filtering strategy (removal of false positives, minimum site usage of ≥5%, average read count ≥10, inter-experimental reproducibility ≥0.75), to ensure high-quality detection of APA events.
  • StringTie was used to assist transcript assembly to help map new PolyA sites.
• Polyadenylation Signal Analysis: Upstream characteristic sequences (AAUAAA, ATTAAA, etc.) were screened to compare in depth the signal types and distribution differences between known database and newly discovered PolyA sites.
• TDP-43 Binding Site Analysis: By cross-referencing the CLIP-Seq database (POSTAR3), the distance and strength of TDP-43 binding sites (GU content, hexamer analysis using the Transite package) were analyzed, in relation to APA regulation.
• PolyA Site Strength Prediction: The deep learning model Aparent2 (a residual neural network) was used to quantify PolyA site “strength,” represented by the log odds ratio metric.

Major Research Findings Explained

1. TDP-43 Dysfunction Causes Widespread APA Abnormalities in Neurons

• In FTD/ALS brain samples, nuclear TDP-43 loss closely correlated with massive abnormal APA events. Dozens of genes (such as LRFN1, MARK3, SYN2, etc.) showed changes in polyadenylation site usage, and in 34 genes, APA alterations directly accompanied changes in gene expression, indicating regulatory effects on protein levels and cellular functions.
• In human neuronal cell models, TDP-43 knockdown induced APA changes in over three thousand genes, with nearly eight thousand sites displaying significantly changed usage. The vast majority of events represented mRNA 3’UTR “lengthening”—which usually increases RNA stability or alters translation—but there were also “shortening” or “premature termination” polyadenylation events directly impacting protein production.

2. Innovative 3’ End-Seq Reveals the Complete Landscape of APA Regulation

• The 3’ End-Seq method not only enabled high-precision polyadenylation site detection but also, for the first time, identified large numbers of “cryptic/variant” PolyA sites and premature polyadenylation events, which truncate RNAs or proteins and impair cellular function.
• 404 sites became “cryptic polyadenylation sites” after TDP-43 KD, of which 152 sites caused premature transcript termination, severely interfering with functional genes.
• Polyadenylation and splicing abnormalities appeared coupled in some genes (such as ARHGAP32, STMN2), suggesting a coordinated mechanism modulating RNA stability and protein production.

3. TDP-43 Binding Patterns and PolyA Site Strength Determine APA Changes

• About 70% of genes with APA changes had actual TDP-43 binding sites; the position of these binding sites relative to the PolyA site determined the direction of APA regulation (closer sites suppressed usage, more distant sites enhanced usage).
• GU-rich binding sites (high-affinity TDP-43 sites) were distributed upstream of APA sites, and when the APA site was weak, it was even more susceptible to TDP-43 perturbation, suggesting that decreases in TDP-43 levels would first “release” weak binding sites, allowing more PolyA sites to be used.
• Using the Aparent2 algorithm, it was found that the newly activated distal polyadenylation sites were “stronger” than the original proximal sites, explaining the notably increased 3’UTR length.

4. Key Functional Gene APA Changes Affect Disease Markers and Pathogenic Mechanisms

• ELP1, ELP3, ELP6 (Translation Elongation Factor Related): After TDP-43 KD, genes such as ELP1 showed significant 3’UTR lengthening, with upregulation of mRNA and protein expression, suggesting tRNA modification abnormalities may participate in ALS/FTD pathogenesis.
• NEFL (Neurofilament Light Chain Protein): TDP-43 KD caused a shift of PolyA site usage from proximal to distal locations, with decreases in both mRNA and NF-L protein levels, directly impacting long-term neuronal stability and the interpretation of disease biomarkers.
• SFPQ (RNA-binding Protein): TDP-43 KD activated distal PolyA sites and downstream splice sites, forming new mRNA isoforms targeted by NMD (nonsense-mediated decay), leading to accelerated degradation and decreased protein expression. The new isoform was validated to be elevated in a large number of FTLD-TDP patient brain samples, proving disease relevance.
• TMEM106B (FTD Risk Gene): TDP-43 KD led to an increased 3’UTR length of TMEM106B and decreased protein dimer expression. Luciferase reporter experiments further confirmed that 3’UTR lengthening lowers protein translation efficiency and increases mRNA stability. Notably, recent discoveries of Alu element insertions in the TMEM106B 3’UTR closely correlate with APA events and may affect dimer formation and amyloid fibril deposition.

Conclusions and Research Significance

This study systematically demonstrates that loss of TDP-43 function not only induces cryptic exon splicing abnormalities but also causes widespread alternative polyadenylation (APA) changes in thousands of neuronal genes, impacting the expression stability and function of key intracellular proteins. These findings greatly enrich our understanding of the molecular pathology of ALS/FTD. The innovative 3’ End-Seq technique enabled high-resolution and accurate detection of polyadenylation events—something that conventional RNA-seq could not fully achieve. This study proposes that the manner in which TDP-43 regulates PolyA sites—its binding strength and the intrinsic polyadenylation signal strength—jointly drive APA changes and ultimately dictate expression profile shifts in disease-relevant genes.

Moreover, the results highlight APA abnormalities as a novel class of biomarkers and potential therapeutic targets in TDP-43 proteinopathies: for example, strategies using antisense oligonucleotides to correct aberrant APA events may, in future, operate alongside existing therapies aimed at correcting cryptic exon splicing, thus broadening interventional options for FTD/ALS.

Key Highlights and Innovative Value

• First systematic revelation of the APA landscape following TDP-43 dysfunction, and validation of the direct impact of polyadenylation abnormalities on the protein expression of neurologically relevant marker genes.
• Innovative application of 3’ End-Seq, achieving de novo coverage of polyadenylation events at single-nucleotide resolution, and capturing cryptic and premature termination events.
• Algorithmic innovations: Integration of custom filtering strategies with deep learning models (Aparent2) achieves highly accurate event classification.
• Discovery of a quantitative relationship between TDP-43 binding affinity and APA regulation, revealing unique molecular mechanisms governing polyadenylation regulation and expanding the theoretical foundation of RNA biology.
• Provides clear guidance for understanding the molecular mechanisms and clinical applications in neurodegenerative diseases; represents a breakthrough for new therapeutic targets.

Other Noteworthy Information

The study also incorporates recent genetic findings—such as the impact of Alu element insertions in the TMEM106B gene 3’UTR on disease risk—providing new directions for future research on the interplay between genetic variation and APA regulation. Additionally, all data, algorithms, and methods in this study have been thoroughly published and shared, offering valuable technical references for peers in the field to advance related research.

Summary

This study, published in Nature Neuroscience, provides a comprehensive decoding of the far-reaching effects of TDP-43 loss on neuronal APA regulation through rigorous experimental design, innovative methods, and in-depth data analysis. It further reveals new molecular pathological mechanisms underlying neurodegenerative diseases such as ALS and FTD. Its scientific and clinical value is extremely high, laying a solid foundation for future basic research and the development of clinical therapeutic targets for these related diseases.