Elevated nuclear TDP-43 induces constitutive exon skipping

Abstract

Background: Cytoplasmic inclusions and loss of nuclear TDP-43 are key pathological features found in several neurodegenerative disorders, suggesting both gain- and loss-of-function mechanisms of disease. To study gain-of-function, TDP-43 overexpression has been used to generate in vitro and in vivo model systems.

Methods: We analyzed RNA-seq datasets from mouse and human neurons overexpressing TDP-43 to explore species specific splicing patterns. We explored the dynamics between TDP-43 levels and exon repression in vitro. Furthermore we analyzed human brain samples and publicly available RNA datasets to explore the relationship between exon repression and disease.

Results: Our study shows that excessive levels of nuclear TDP-43 protein lead to constitutive exon skipping that is largely species-specific. Furthermore, while aberrant exon skipping is detected in some human brains, it is not correlated with disease, unlike the incorporation of cryptic exons that occurs after loss of TDP-43.

Conclusions: Our findings emphasize the need for caution in interpreting TDP-43 overexpression data and stress the importance of controlling for exon skipping when generating models of TDP-43 proteinopathy.

Fig. 1

TDP-43 overexpression in mice leads to skipping of constitutive exons. (A) Human TDP-43 (TDP-43^WT) and TDP-43 carrying a G298S mutation (TDP-43^G298S) were expressed under the weak Thy1.2 promoter in mice. (B) TDP-43 levels in the spinal cord and cortex of transgenic mice were compared to control mice using immunoblotting. In both transgenic lines, TDP-43 protein levels were elevated at approximately 1.5x and 1.3x higher in the spinal cord and cortex, respectively (* p < 0,05, ***p < 0.001). (C) We measured the hanging time of transgenic mice compared to their littermate controls and found a reduction associated with age, indicative of a motor neuron deficit. (D) Both transgenic lines had shorter survival times compared to non-transgenic (NT) controls (NT vs. WT: p = 0.0005, NT vs. G298S: p = 0.0035), but no differences were found between the two transgenic lines (WT vs. G298S: p = 0.1260). (E-G) RNA-Seq analysis on isolated mouse spinal cords revealed several examples of exon skipping (arrows) in both transgenic lines. (H) We further validated these findings by transfecting mouse N2a cells with human TDP-43 and performing RT-PCR to test whether TDP-43 expression alone was sufficient to induce exon skipping

Fig. 2

TDP-43 overexpression induces exon repression in humans. (A) Human i3Neurons were transduced with lentivirus expressing human TDP-43 and sequenced to compare exon skipping between mouse and human. (B) RNA-Seq analysis revealed numerous genes with skipped exons that are involved in a variety of molecular pathways related to intellectual disability, synaptic activity, and mitochondrial proteins (C). We identified exons with particularly high levels of exon skipping in the genes HYOU1, NUP93, and XPNPEP1 (arrows). (D-F) When cross-referenced with datasets from transgenic mice, we found that exons repressed in humans were not repressed in mice. Using RT-PCR, we validated skipping events in i3Neurons with primers located in exons adjacent to the repressed exon (G) or primers that spanned a skipped junction (H). (I) Analysis of UG repeats in human skiptic exons revealed that TDP-43’s consensus motifs are found in both the repressed exon itself and adjacent intronic sequences. UG motifs appear slightly more frequently around the downstream 5’ splice site, but with far shorter UG repeat lengths than those found adjacent to cryptic exons [73]

Fig. 3

TDP-43 exon skipping events are found in aging human brains but do not correlate with disease. (A) Alignment of syntenic mouse (mm10) and human (hg38) genomic sequences surrounding exons repressed by TDP-43 overexpression in human cells. Constitutively spliced exons in the genes XPNPEP1, NUP93, MYBBP1A, and HYOU1 are skipped in human cells but not mouse cells when TDP-43 is overexpressed. By contrast, the exon in DDI2 is repressed in both mouse and human cells. UG motifs (yellow highlights) are slightly enriched around the 3’ and 5’ splice sites (Supplementary Fig. 5). (B) We performed RT-PCR to amplify cryptic junctions or exon-repressed junctions in human brain samples from patients with AD pathology with or without TDP-43 inclusions, frontotemporal dementia with inclusions, and control patients who did not have TDP-43 inclusions and profiled the STMN2 cryptic exon [16, 17]. RT-PCR analysis showed that exon skipping occurred in both control and disease samples. (C) Exploration of skipping events through normal human aging was performed by analyzing RNA-Seq datasets from the Genotype-Tissue Expression (GTEx) project. PSI values of skiptic exons across human brain regions are shown for the age range of 60 to 69 years old (all age ranges are available in Supplementary Fig. 7). Exon skipping is found at low levels in most of the different brain areas analyzed, with slightly higher levels in the cerebellum. (D-K) AlphaFold2 was used to model protein structures with (D-G) and without (H-K) skipped exons. Purple highlights indicate repressed exons while green highlights indicate flanking amino acid sequences

Fig. 4

Overexpression of the ΔNLS mutant TDP-43 (TDP-43^NLSm) induces exon repression when expressed at higher levels than wildtype TDP-43. QBI-293 cells with a dox-inducible cassette expressing TDP-43-GFP (iGFP-WT) or TDP-43 with a mutated NLS (iGFP-NLSm) were exposed to Dox for 0, 24, 48, 72 h in triplicate. (A) Immunoblot of TDP-43 levels after Dox induction (left) and RT-PCR for NUP93, XPNPEP1 and MYBBP1A. (B) RT-PCR quantification of iGFP-NLSm, exon repression reached ∼ 9% compared to ∼ 95% in iGFP-WT induction. (C) Total TDP-43 protein levels for WT and NLSm reached ∼ 2 times normal. (D, E) Estimated nuclear TDP-43 levels based on TDP-43^NLSm passive diffusion. (F, G) HEK-293 cells transfected with TDP-43^WT-P2A-GFP or TDP-43^NLSm-P2A-GFP were FACS sorted into seven fractions by GFP intensity. Immunoblotting of isolated nuclei showed increasing TDP-43^NLSm/TDP-43^WT ratio with higher expression. (H) RT-PCR showed progressive exon skipping with TDP-43^WT or TDP-43^NLSm overexpression, performed in duplicate. (I) With strong expression, TDP-43^WT can repress exons by ∼ 95% but TDP-43^NLSm only by ∼ 40%. Using these data, we estimate the proportion of TDP-43^NLSm in the Dox-inducible system, dotted line (D). Levels of total TDP-43^WT and total predicted TDP-43^NLSm in the nucleus are plotted together with their respective exon expression levels (E)

Fig. 5

Summary Diagram. TDP-43 is a highly autoregulated protein due to different forms of cellular toxicity when TDP-43 protein levels are either too low (cryptic exon incorporation) or too high (constitutive exon skipping, i.e. skiptic exons). Our study has demonstrated that these splicing deficits are linked specifically to nuclear TDP-43, whereas toxicity due to cytoplasmic TDP-43 remains to be fully elucidated. Future studies that avoid constitutive exon skipping (skiptic exons) may identify biomarkers for cytoplasmic-specific TDP-43 toxicity