BGPT: Paper Review: TDP-43 nuclear loss in FTD/ALS causes widespread alternative polyadenylation changes

Explore by Goal

Quick Explanation Copied

Bottom line

This paper argues that loss of nuclear TDP-43 in FTD/ALS is accompanied by widespread alternative polyadenylation (APA)—especially 3′UTR lengthening—mapped at single–polyA-site resolution using 3′ end-seq, and linked to changes in disease-relevant genes (e.g., NEFL, SFPQ, TMEM106B).

Long Explanation

Paper Review (Science-forward, skeptical, evidence-based)

Title: “TDP-43 nuclear loss in FTD/ALS causes widespread alternative polyadenylation changes” DOI: 10.1038/s41593-025-02049-3

Figure-of-merit snapshot (what they did → what they claim)

Correlative human evidence: neuronal nuclei sorted by TDP-43 status from FTD/ALS frontal cortex show widespread APA signals from re-analysis of RNA-seq.
Mechanistic perturbation in neurons: TDP-43 knockdown in human stem-cell-derived neurons induces widespread APA; both RNA-seq-based APA and 3′ end-seq reveal extensive polyA-site usage changes including cryptic polyA activation.
Constraint-based mechanism: they connect outcome variation to binding position and binding motif composition, plus intrinsic polyA site strength, using TDP-43-binding datasets and a deep model for polyA-site usage/strength.
Disease-gene examples: NEFL, SFPQ, TMEM106B, and ELP1/ELP3 are highlighted with RNA and protein readouts (including NMD logic for SFPQ and reporter effects for TMEM106B).

Evidence map (correlation vs causation vs functional mechanism)

Legend

Human correlation Perturbation (causal-ish) Mechanism + orthogonal readouts

Human: TDP-43-depleted neuronal nuclei associate with APA changes (RNA-seq-based).

Neuronal perturbation: TDP-43 knockdown in iNeurons induces widespread APA and cryptic polyA activation detected by 3′ end-seq.

Mechanism: binding position/strength and polyA strength predict direction/outcome; selected disease genes show RNA→protein pathway logic (including NMD).

Quantitative outcomes (from the paper’s extracted numbers)

These plots use only the aggregate counts explicitly stated in the provided full-text excerpt/data summary.

Core claim analysis (mechanism & causality)

1) APA dysregulation tracks TDP-43 loss

Strength: the paper uses both (i) human sorted-nuclei correlation and (ii) neuron perturbation (TDP-43 KD) plus an assay explicitly designed to map polyA sites at nucleotide resolution (3′ end-seq).
Critical caveat: APA calling from short-read RNA-seq is inherently sensitive to library depth, polyA reference databases, and mapping ambiguity; the authors report limited overlap between APAlyzer and QAPA (only LRFN1 and MARK3 substantial common changes), which signals a dependence on method-specific polyA site references and effective resolution.

2) Binding position + “binding strength” associate with direction of APA change

The authors observe that TDP-43-binding-site position downstream/upstream of polyA sites relates to distal-vs-proximal usage shifts upon KD, and they enrich specific GU-repeat motifs in different regions relative to polyA sites.
The paper proposes a mechanistic hypothesis involving cleavage stimulation factor 2 (CstF2) binding proximity and TDP-43 blocking (presented as a plausible mechanism, not direct biochemical demonstration).

Skeptical lens: the association-to-mechanism jump would be much stronger if the paper included orthogonal direct-binding or perturbation of polyadenylation machinery at specific loci (the excerpt provided does not show such direct biochemical follow-up).

3) Disease gene examples show RNA→protein-consistent consequences

NEFL: TDP-43 KD shifts apparent polyA usage distally, with reduced NEFL RNA and NF-L protein levels.
SFPQ: APA changes produce an alternative 3′ splice site usage linked to an NMD-sensitive isoform; NMD inhibition increases the alternative isoform, and TDP-43 KD reduces SFPQ protein.
TMEM106B: 3′UTR lengthening accompanies reduced TMEM106B dimer levels; reporter logic supports reduced translation with the long 3′UTR.

Methodological strengths & red flags (what could mislead)

Strengths

Assay choice: 3′ end-seq with stringent mis-priming filtering is conceptually well-matched to detecting premature polyadenylation events that RNA-seq APA methods may miss.
Orthogonality: they include qRT-PCR, western blots, and luciferase reporters for key examples (NEFL/SFPQ/TMEM106B).

Red flags / blind spots

Pipeline dependence: APA calling from RNA-seq differs substantially by method and polyA reference DB (limited overlap between APAlyzer and QAPA findings).
Assay filtering tradeoff: stringent mis-priming and polyA-site filtering could reduce false positives but may also eliminate low-abundance/context-specific polyA sites (risk of false negatives).
Translation mechanism not fully closed: the proposed CstF2-related blocking model is presented as plausible but is not directly tested biochemically within the excerpted content.
Model generalization: causality is shown in iNeurons, but neurons in vivo are heterogeneous (maturation state, regional identity, glial context). The paper itself suggests future work across cell types/diseases.

Reproducibility & data/code availability (from text provided)

Sequencing data: sequencing data generated in this study are available at GEO GSE252892.
Source data + code: source data and figure-generation code are available via Zenodo repositories and GitHub.

Directed “how to falsify” checklist (specific, science-grade)

Orthogonal assay falsification: show that independently mapped polyA sites (alternate 3′ end-seq library prep/analysis settings) do not reproduce the same high-level shift pattern (dominant 3′UTR lengthening and cryptic polyA activation) after TDP-43 loss.
Mechanism falsification: at a locus level (e.g., SFPQ or TMEM106B), demonstrate that blocking the predicted polyA usage change does not alter the RNA isoform stability/translation/protein output that the paper links to disease genes.
Binding-property falsification: show that altering TDP-43 binding (e.g., motif disruption or CLIP-verified binding loss) does not change polyA usage direction at predicted sites.

Relevant conceptual anchor (APA biology context)

Alternative polyadenylation is a widespread layer of post-transcriptional regulation; context-specific regulation can affect mRNA stability, localization, and translation—so the general biological plausibility of APA involvement in neurodegeneration is consistent with the established literature. For broader mechanistic context, APA as a regulatory layer is reviewed in .

Follow-up: bespoke author reviews (click through)

Feedback:

Updated: March 20, 2026