(C) PLOS One
This story was originally published by PLOS One and is unaltered.
 . . . . . . . . . .



A quantitative cell-based reporter links TDP-43 aggregation and dysfunction to define pathogenic mechanisms [1]

['Lohany Dias Mamede', 'Edward Doisy Department Of Biochemistry', 'Molecular Biology', 'Saint Louis University School Of Medicine', 'St. Louis', 'Missouri', 'United States Of America', 'Miwei Hu', 'Department Of Neurology', 'Washington University In St. Louis']

Date: 2026-03 

TDP-43 pathology is a hallmark of fatal neurodegenerative disorders, including amyotrophic lateral sclerosis (ALS), frontotemporal dementia (FTD), and limbic-predominant age-related TDP-43-encephalopathy (LATE). In affected patients, cytoplasmic TDP-43 aggregates are accompanied by disruption of its normal nuclear localization and function. Because TDP-43 is an RNA binding protein that controls transcript processing, including repression of cryptic exon splicing, its loss leads to dysregulation of gene expression. Despite its central significance in disease, the connection between TDP-43 aggregation and dysfunction remains poorly understood, and models to study the underlying mechanisms are limited. Here, we characterize a robust and quantitative cell-based reporter that captures both aggregation and the resulting loss of function. Using this human biosensor cell line, we show that aggregation initiated by prion-like seeding drives progressive depletion of nuclear TDP-43 and induces signature features of diminished TDP-43 activity, such as increased DNA damage and activation of cryptic exon splicing. We find that aggregate seeding also induces cryptic exon splicing in human neurons implying that this pathological link extends to disease-relevant models. The seeding model provides a platform for dissecting mechanisms that underlie TDP-43 pathology and for identifying factors that modulate the aggregation-to-dysfunction transition. Our data shows that aggregate seeding impacts TDP-43 autoregulation, initiating a toxic feed-forward mechanism that disrupts TDP-43 homeostasis. Furthermore, reducing ataxin-2 levels decreases aggregation and restores TDP-43 activity. Together, these findings reveal a molecularly guided strategy to directly impact TDP-43 activity by decreasing its misfolding and aggregation, highlighting approaches to prevent TDP-43 dysfunction and mitigate toxicity under pathological conditions.

Funding: This work was supported in part by National Institutes of Health (NIH) National Institute for Neurological Disorders and Stroke (NINDS) and National Institute on Aging (NIA) grants R01 NS114289 (to Y.M.A.) ( https://www.ninds.nih.gov/ , https://www.nia.nih.gov/) ; the Department of Defense CDMRP/ALSRP W81XWH-20-1-0241 (to Y.M.A.) ( https://cdmrp.health.mil/alsrp/default.aspx) ; Hamon Charitable Foundation (to M.I.D); Lyle Rakers Foundation (to T.M.M.). Healthy Aging and Senile Dementia [P01 AG003991], Alzheimer’s Disease Research Center [P30 AG066444], Adult Children Study [P01 AG026276] to the Knight Alzheimer Disease Research Center at Washington University in St. Louis, Neuropathology Core. The sponsors did not play a role in the study design, data collection and analysis, decision to publish or preparation of the manuscript.

Experimental methods to recapitulate TDP-43 aggregation and loss of function in the same model have been challenging to develop and remain rare. To induce TDP-43 aggregation, common models rely on overexpression of wild-type or mutant TDP-43 or chemical treatment to induce oxidative stress or other proteotoxic conditions. These models trigger generalized disruption of cell function and undermine insight into disease-relevant processes. Furthermore, these experimental conditions rarely lead to TDP-43 depletion, and their effect on TDP-43 function has not been widely reported. Other assays to investigate TDP-43 dysfunction rely on knocking-out/down TDP-43 expression, however, these conditions lack the contribution of TDP-43 mislocalization and aggregation. Thus, to establish cellular models combining both TDP-43 aggregation and loss of protein function mechanisms that overcome the limitations of previous methods, we explored systems based on TDP-43 prion-like aggregate seeding. Using FTD brain-derived seeds to induce aggregation in a FRET-based reporter cell line, we find that seeding greatly impacts TDP-43 localization and leads to gradual loss of function. Our model recapitulates key features of TDP-43 pathology and provides evidence that aggregation, mislocalization and loss of function mechanisms are interconnected. Furthermore, we identify processes that become altered upon aggregate seeding and identify key factors that modulate both aggregation and dysfunction.

A) The HEK293 FRET cell line stably expressing the TDP-43 C-terminal domain (CTD, amino acids 262-414) fused to mClover3 and mRuby3 FRET pairs was used as a reporter of TDP-43 aggregation upon seeding with FTD extract, prepared as in S1A Fig . Image created in BioRender, Ayala, Y. (2026) https://BioRender.com/kt5mypg . B) de novo CTD aggregation in HEK293 FRET cells treated with FTD seeds visualized by confocal fluorescence microscopy. As control, cells were treated with neurologically unaffected tissue extract. Phosphorylated TDP-43 (pTDP-43) and p62, shown in (C) , were used as markers of pathological TDP-43 aggregates. DAPI staining was used to highlight nuclei. D) High resolution images of aggregates in cells treated with FTD seeds as z-stacks with DAPI overlay generated with stimulated emission depletion (STED) microscopy. The white box corresponds to the enlarged region. Scale bar for microscopy images, 10 μm. E) Flow cytometry to quantify FRET-positive and FRET-negative HEK293 FRET cells six days post-seeding. The raw data can be found in Mendeley Data, V1, https://doi.org/10.17632/yh788t3bxy.1 Percent over total of FRET-positive cells and FRET median fluorescence intensity (MFI) from the representative experiments are shown. F) Average percent FRET-positive cells treated with FTD seeds and control extract quantified by flow cytometry. (G) Calculated integrated FRET density equal to the product of MFI and %FRET-positive cells. SEM in both (F) and (G) , n > 3, *p = 0.028, Mann-Whitney test. The data underlying this figure can be found in S1 Data .

The C-terminal domain (CTD) of TDP-43 plays a critical role in aggregation and aggregate propagation, as demonstrated by the highly enhanced seeding efficiency of exposed fibrillar CTD fragments generated through limited proteolysis of full-length aggregates [ 30 ]. The CTD is a mostly disordered domain referred to as the prion-like domain for its sequence similarity to yeast prion proteins [ 31 ]. Furthermore, structural studies of aggregates isolated from FTD patient brain have shown that the core of the pathological filament is formed by the central portion of the CTD, specifically amino acids 282–360 [ 32 – 34 ] ( Fig 1A ). These observations strongly implicate a central role of this region during misfolding and transmission of disease-relevant aggregates.

TDP-43 pathology is a hallmark of neurotoxicity and neurodegeneration in almost all amyotrophic lateral sclerosis (ALS) and ∼50% of frontotemporal lobar degeneration (FTLD-TDP) and the ensuing frontotemporal dementia (FTD) [ 1 , 13 ]. TDP-43 inclusions also characterize an Alzheimer’s disease (AD)-associated disorder, limbic-predominant age-related TDP-43 encephalopathy (LATE), that affects nearly a quarter of individuals over 85 years of age [ 14 ]. In addition, TDP-43 is a comorbid feature of greater than 50% of AD cases, correlated with more aggressive memory loss and brain atrophy [ 15 – 18 ]. Similarly, TDP-43 aggregates are major components upon traumatic brain injury and chronic traumatic encephalopathy [ 19 , 20 ]. A common mechanism thought to contribute to pathogenesis in TDP-43 proteinopathies and other neurodegenerative disorders is the propagation of protein toxicity through prion-like aggregate seeding. In this process, misfolded protein serves as a template to induce the aggregation of its soluble species, creating a self-perpetuating cycle. Experimental evidence demonstrates that aggregates generated from purified TDP-43 are seeding-competent in cell culture models [ 21 , 22 ]. Additionally, insoluble extracts derived from ALS and FTD CNS tissue initiate de novo aggregation in cellular models and iPSC-derived cerebral organoids [ 23 , 24 ]. Additional compelling evidence for TDP-43 seeding activity comes from TDP-43 transgenic mouse models in which injection of FTD-derived extracts enriched in phosphorylated TDP-43 (pTDP-43) leads to the spread of pathology [ 25 ]. These observations align with clinicopathological evidence showing the progressive spread of neuronal loss and TDP-43 pathology in the CNS of ALS and FTD patients [ 14 , 17 , 26 – 29 ]. Hence, these studies strongly support the model that TDP-43 aggregates propagate toxicity and thereby contribute to neurodegeneration. The mechanisms of proteopathic seed uptake, propagation and the consequences of aggregate seeding on TDP-43 cellular function remain poorly understood. Further elucidation of these processes will highlight underlying disease mechanisms and identify potential therapeutic strategies.

The Transactive Response DNA binding protein (TDP-43) is an essential RNA binding protein that regulates gene expression through RNA processing and is strongly linked to multiple neurodegenerative disorders, collectively referred to as TDP-43 proteinopathies. TDP-43 pathology is characterized by protein aggregation, most often in the cytoplasm of neurons and glial cells, and depletion from its normal nuclear distribution [ 1 ]. TDP-43 is an RNA binding protein most well-characterized for regulating alternative splicing and alternative polyadenylation [ 2 – 6 ]. Recent findings have underscored the involvement of loss of TDP-43 function in disease, as dysregulation of TDP-43 target genes is observed in patients affected by TDP-43 proteinopathies [ 7 – 12 ]. Together, these observations strongly suggest that TDP-43 misfolding and aggregation are correlated with the loss of protein function. However, the mechanisms connecting these two major components of TDP-43 pathology remain unclear.

Results

TDP-43 aggregate seeding triggers de novo aggregation and gradual TDP-43 mislocalization–To quantify, isolate and characterize cells affected by TDP-43 aggregates induced by intracellular seeding, we employed a human embryonic kidney cell line (HEK293) that reports TDP-43 aggregation based on Förster resonance energy transfer (FRET) [35]. This cell line stably expresses the TDP-43 C-terminal domain (CTD, amino acids 262-414) C-terminally fused to FRET pairs mClover3 and mRuby3 (HEK293FRET) (Fig 1A). To investigate TDP-43 aggregate seeding in the context of disease-derived proteopathic seeds, we isolated the insoluble fraction of autopsy brain from FTLD-TDP patients subtypes A and B [36–43]. For these studies we utilized frontal cortex tissue primarily from four different cases. We prepared sarkosyl-insoluble fractions upon sequential extraction based on previously established methods [44]. Herein, we refer to this extract as FTD seeds. These samples were enriched for total TDP-43 and a marker of pathological inclusions, phosphorylated TDP-43 (pTDP-43, Ser409/410) [45] (S1A Fig). As control, we processed brain tissue from three neurologically unaffected individuals, which did not show significant increase of pTDP-43. Treatment of HEK293FRET cells with FTD seeds induced the formation of cytoplasmic inclusions composed of mClover and mRuby-CTDs that strongly accumulated pTDP-43, as seen by immunofluorescence (Fig 1B). In addition, biochemical fractionation into RIPA-soluble and Urea-soluble fractions indicated strong pTDP-43 upregulation in cells treated with FTD seeds, compared to control extract (S1B Fig). Moreover, pTDP-43 was highly enriched in the Urea-soluble fraction, consistent with aggregate formation. We observed recruitment of the ubiquitin-binding protein 62/sequestosome 1 (p62/SQTM1) to cytoplasmic aggregates which, together with pTDP-43 accumulation, is commonly associated with TDP-43 inclusions in disease (Fig 1C) [36,46]. Intriguingly, the aggregates often showed filament-like morphology, as seen by high-resolution microscopy (Fig 1D) and confocal imaging (S2A Fig), suggesting that de novo aggregation follows a defined structural pattern upon seeding. Whether these structures resemble TDP-43 filaments in FTD patient brains remains an open question for future investigation [32–34]. Analogous cell lines expressing only mClover- or mRuby-fused CTD showed similar aggregate seeding behavior (S2B Fig). Next, we quantified the number of cells that developed aggregates upon seeding by flow cytometry and determined the ratio of FRET-positive cells relative to total. Treatment with FTD seeds resulted in an average of 14% FRET-positive cells, compared to <1% for control tissue extract (Fig 1E and 1F). This correlated to a dramatic increase in integrated FRET density, compared to control extract (Fig 1G). The large increase in this value, which measures the number of FRET-positive cells and the degree of aggregation in each cell, highlights the sensitivity of the assays to measure aggregate accumulation [47]. These findings are consistent with the previously observed response of this reporter cell line to seeding with pre-formed aggregates of purified TDP-43 and aggregates derived from a mouse model of TDP-43 pathology in skeletal muscle [24,35]. The significant increase in FRET activity upon treatment with FTD seeds, compared to control-treated cells, in which this signal is nearly absent, suggests that homotypic reversible CTD interactions, such as phase separation [48,49], do not significantly contribute to the FRET signal detected in our analyses. It is also possible that, under the conditions tested, the concentration of CTD fragments in HEK293FRET cells does not reach the threshold required for condensate formation or phase separation.

To rule out a contribution of FRET signal from interactions between mClover and mRuby proteins in the absence of the CTD, we tested stable HEK293 cells expressing the CTD fused to mClover or mRuby alone or in combination with mRuby or mClover proteins lacking the CTD, respectively (S3A Fig). These multiple cell lines were transfected with either FTD seeds, tau-positive AD extract, lipofectamine and media only controls. After 48 hours, only HEK293FRET cells expressing both the CTD-mClover and mRuby fusion proteins generated significant FRET signal upon treatment with FTD-derived extract (S3B and S3C Fig). All other cell lines showed negligible FRET values. Previous findings together with new control experiments demonstrate that the HEK293FRET reporter specifically responds to TDP-43 proteopathic seeds. Seeding with α-synuclein pre-formed fibrils fail to induce significant FRET signal [24]. Similarly, we find no significant FRET signal in cells treated with insoluble brain extract from tau-positive Alzheimer’s disease cases lacking TDP-43 pathology (S3C Fig). Together, these results indicate that de novo TDP-43 aggregation in this biosensor cell line is specifically induced by proteopathic seeds derived from tissue with TDP-43 pathology, while remaining insensitive to other protein aggregates. Furthermore, the FRET signal generated in HEK293FRET cells depends on the presence and aggregation of the CTD.

We next investigated changes in the localization of endogenous TDP-43 following seeding-induced aggregation. To specifically detect endogenous TDP-43, we used an antibody recognizing an N-terminal domain (NTD) epitope absent in mClover- and mRuby-CTD constructs. We confirmed that the signal detected by the TDP-43 NTD-specific antibody does not arise from internalized FTD-derived seeds by examining HEK293 and HEK293FRET cells six and two days post-seeding, respectively. No significant cytoplasmic TDP-43 NTD-specific antibody signal was detected in these cells (S4A and S4B Fig). In addition, during our seeding experiments, cells were trypsinized and replated 24 hours post-transfection with FTD seeds or control extract to remove any non-internalized aggregates. These experiments indicate that the amount of internalized FTD seeds is insufficient to generate detectable immunofluorescence with the NTD antibody under our experimental conditions. Despite the accumulation of cytoplasmic aggregates in HEK293FRET cells three days post-treatment with FTD seeds, endogenous TDP-43 localization remained largely unaffected (Fig 2A, arrows). In contrast, by six days post-seeding, we observed colocalization of endogenous TDP-43 with CTD cytoplasmic inclusions, accompanied by marked reduction of its nuclear localization (Fig 2B). To quantify depletion of nuclear TDP-43 distribution in cells treated with FTD seeds, we measured the mean fluorescence intensity of endogenous TDP-43 in the region of interest defined as the nucleus. We found that nuclear endogenous TDP-43 decreased by approximately 70% percent in cells with cytoplasmic aggregates, compared to cells without observable aggregates (Fig 2C). These findings indicate that seeding-initiated cytoplasmic CTD aggregation sequesters endogenous TDP-43, leading to its gradual depletion from the nucleus, thus recapitulating key features of neuronal TDP-43 pathology in patients [1].

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 2. Aggregate seeding induces co-aggregation and gradual depletion of nuclear endogenous TDP-43. A) Immunofluorescence of HEK293FRET cells treated with FTD seeds or neurological unaffected tissue extract, as control, probed with a TDP-43 antibody targeting an N-terminal domain epitope to detect endogenous protein only. Cells were imaged by confocal microscopy three and six days post treatment in (B). Dashed lines delineate cell boundaries and arrows indicate cells affected by cytoplasmic aggregates. The nuclei of aggregate-affected cells treated with FTD seeds at six days post seeding are highlighted by solid yellow lines. Scale bar, 10 μm. C) Mean fluorescence intensity of nuclear endogenous TDP-43 in cells six days post treatment with FTD seeds, comparing cells with and without cytoplasmic aggregates. ImageJ was used to quantify pixel intensity per nuclear area, defined by DAPI staining. Calculated mean values, SEM, of >250 cells from n = 5 are shown, **p = 0.001, paired t test. The data underlying this figure can be found in S2 Data. https://doi.org/10.1371/journal.pbio.3003662.g002

Aggregate seeding results in loss of TDP-function in affected cells—We then examined whether the loss of nuclear TDP-43 in aggregate-affected cells correlates with loss of TDP-43 function. First, we compared the impact on genomic stability, as TDP-43 downregulation increases accumulation of DNA damage and genomic instability in human cells, including neurons [50–53]. Six days post-seeding, phosphorylated histone H2AX (γH2AX), a marker of DNA breaks, increased and was detected in 50 ± 7% of TDP-43 aggregate-positive cells (>15 γH2AX foci per nucleus) (Fig 3A). In contrast, control-treated cells showed 4% γH2AX-positive nuclei. Furthermore, immunoblotting showed a significant 2.5-fold increase in γH2AX levels upon seeding with FTD-derived extract, compared to control extract (Fig 3B). To explore whether DNA damage is linked to the seeding-induced loss of nuclear TDP-43, we quantified γH2AX and nuclear TDP-43 after treatment of stable HEK293 cells expressing mClover-CTD with FTD seeds (S4C Fig). Our results showed a significant inverse correlation between the mean fluorescence intensity of TDP-43 and γH2AX in the nucleus, whereby decreasing nuclear TDP-43 levels are associated with increasing γH2AX detection and vice versa in a linear relationship (Fig 3C). These results strongly suggest that aggregation progressively depletes functional TDP-43. Moreover, these observations imply that the link between TDP-43 dysfunction and loss of genomic integrity may be extended to conditions triggered by seeding and TDP-43 mislocalization.

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Seeding-induced aggregation increases DNA damage corresponding to decreasing nuclear TDP-43 localization. A) Immunofluorescence obtained by confocal microscopy to detect γH2AX, a marker of DNA breaks, in cells six days post treatment with control or FTD seeds. Arrow points to the accumulation of γH2AX in aggregate-affected cells. Scale bar, 10 μm. As negative control, cells were treated with neurological unaffected tissue extract. B) γH2AX levels quantified by immunoblotting in cells treated with FTD seeds and control, plotted as mean, SEM of n = 6. Mann-Whitney test, **p = 0.002. GAPDH was used as loading control. C) Mean γH2AX and nuclear endogenous TDP-43 fluorescence intensity (defined by DAPI staining) quantified from images in S4C Fig (ImageJ). Values were calculated for >50 cells per replicate for n = 3. The Pearson correlation coefficient (r) and p-value are shown. The data underlying this figure can be found in S3 Data and the original image can be found in S1 Raw Images. https://doi.org/10.1371/journal.pbio.3003662.g003

To further examine TDP-43 activity in cells affected by aggregate seeding, we measured mRNA transcript expression and processing of TDP-43-regulated targets. FRET-positive and FRET-negative cells were isolated via fluorescence activated cell sorting (FACS) at six days post-seeding with FTD seeds (Fig 4A). Cells treated with control tissue extract or non-treated control were also sorted, resulting in mostly FRET-negative cells. Based on previous studies in human cells, we selected transcripts controlled by TDP-43 through different posttranscriptional processing mechanisms. TDP-43 inhibits the expression of SMCA1 and GXYLT1 through binding to the 3’UTR and controlling the usage of alternative polyadenylation poly(A) sites [6] (Fig 4B). Using real-time quantitative PCR (qPCR), we measured mRNA expression in cells treated with FTD seeds, control-derived extracts and non-treated control. FRET-positive cells showed approximately 2-fold or greater significant increase in mRNA expression, compared to FRET-negative cells (Fig 4B). Actin B mRNA (ACTB) served as reference for a transcript that is not regulated by TDP-43 and showed no significant differences in expression between groups. Similar analysis at earlier time points (three days post-seeding) when approximately 4% of cells treated with FTD seeds were FRET-positive, showed no significant changes in SMCA1, GXYLT1 expression between FRET-positive and negative groups (S5A Fig).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. Aggregate seeding impairs TDP-43–dependent RNA processing. A) Experimental scheme to isolate FRET-positive (FRET+) and -negative (FRET-) HEK293FRET biosensor cells by FACS six days post transfection with FTD seeds, or neurologically unaffected tissue extract and compared to non-treated control. Image created in BioRender, Ayala, Y. (2026) https://BioRender.com/yy466ls. B) TDP-43 regulates 3′ UTR processing and polyadenylation choice via direct recruitment to transcripts, whereby TDP-43 binding inhibits SMC1A and GXYLT1 mRNA expression. Relative SMC1A and GXYLT1 mRNA levels were measured by quantitative real-time PCR (qPCR) and normalized to non-treated sample. ACTB served as TDP-43-independent transcript control. Graphs represent mean values, SEM of n > 4, analyzed by unpaired one-way ANOVA, *p = 0.02, **p = 0.009, ***p < 0.0002, ns, non-significant. C) Cryptic exon (CE) splicing is normally suppressed by TDP-43 in HDGFL2 and ARHGAP32. Relative CE inclusion quantified by qPCR from non-sorted HEK293FRET cells collected three days post treatment with FTD seeds and controls. Values were normalized to non-treated samples, mean SEM for n = 5, analyzed with non-parametric, unpaired t test, **p < 0.008, ns, non-significant. D) Relative HDGFL2 and ARHGAP32 CE inclusion quantified in sorted cells six days post seeding, normalized to non-treated control. Mean SEM for n = 5, ****p < 0.0001, according to unpaired one-way ANOVA. E) Immunofluorescence confocal microscopy of neurons derived from neurogenin-2 (Ngn2)-inducible human induced pluripotent stem cells (iNeurons), DIV15. iNeurons incubated with control extract or FTD seeds post-differentiation and collected after 2-3 weeks were stained for MAP2 and Tuj1 to evaluate neuronal differentiation. Scale bar, 100μm. F) MAP2AB expression was measured by qPCR. G) ARHGAP32 relative CE inclusion and relative expression of truncated STMN2 quantified by qPCR. Graphs represent mean values, SEM of n > 5. *p = 0.01, ns, non-significant, non-parametric, unpaired t test. GAPDH was used as reference in all experiments. The data underlying this figure can be found in S4 Data and the original image can be found in S1 Raw Images. https://doi.org/10.1371/journal.pbio.3003662.g004

In addition to regulating the constitutive splicing of hundreds of genes, TDP-43 inhibits or enhances splicing of cryptic exons (CEs), which constitute untranslated/intronic regions that are normally excluded from mature mRNA [7]. Recent studies show the inclusion of TDP-43 target CEs in ALS-FTD cases and AD, consistent with loss of TDP-43 function [9–12,54–57]. Thus, impaired regulation of CE splicing is viewed as a salient feature of TDP-43 dysfunction and disease, and abnormal mRNA products may be used as sensitive markers of TDP-43 pathology in patients. We investigated activation of CEs upon TDP-43 aggregate seeding in cases where TDP-43 inhibits their inclusion, HDGFL2 and ARHGAP32 (Fig 4C) [7,58]. We observed CE splicing in cells treated with FTD seeds starting at three days post-treatment, ARHGAP32 CE splicing was significantly upregulated and HDGFL2 CE splicing showed an increasing trend. These non-sorted cells showed nearly 2-fold greater CE inclusion compared to control. Strikingly, FRET-positive cells isolated six days post-seeding showed 150 and 1,000-fold greater HDGFL2 and ARHGAP32 CE inclusion, respectively, compared to FRET-negative and control-treated cells (Fig 4D). In contrast, we found no significant differences in the levels of HDGFL2 and ARHGAP32 transcripts lacking the CEs between the experimental groups (S5B Fig). The significant activation of CE inclusion starting at early timepoints after aggregate seeding, which dramatically increases in aggregate-affected cells, implies that CE regulation of these targets is highly sensitive to decreasing TDP-43 levels.

To test whether aggregate seeding induces pathological markers of TDP-43 aggregation (pTDP-43, p62) and loss of TDP-43 function in a different cell line, independent of CTD expression, we first investigated seeding in HEK293 cells without overexpression of exogenous TDP-43, mutant or fragment. At one week post transfection with FTD seeds we observed no significant cytoplasmic aggregation or loss of endogenous TDP-43 levels, compared to control. The number or size of the aggregates did not change with longer incubation times (up to six weeks) in which case cells were replated weekly to prevent overcrowding and cell death. These findings indicate that seeding-induced aggregation may not be observed upon expression of endogenous TDP-43 alone or without adding proteotoxic stress conditions, within the experimental timeline tested. Furthermore, these results suggest that misfolded TDP-43 is normally cleared efficiently by the proteasomal degradation and autophagy pathways. Disruptions of these clearance mechanisms, additional proteotoxic stress, impaired nuclear-cytoplasmic trafficking, accumulation of C-terminal fragments, or other unknown factors may be required to promote aggregation-induced nuclear depletion and dysfunction. Therefore, we tested the HEK293TDP-43NLS cell line, expressing a single copy of the nuclear localization-deficient full length mutant TDP-43NLS [21]. We previously showed that TDP-43NLS undergoes de novo aggregation following transfection with TDP-43 aggregates derived from full-length purified protein [21]. Seeding with FTD seeds increased cytoplasmic aggregation recognized by pTDP-43 [21] and p62, which were absent in control-treated cells (S6A Fig). Furthermore, we found upregulation of CE splicing, compared to control, suggesting loss of normal TDP-43 function upon aggregate seeding (S6B Fig). These results indicate that the strong link between loss of function and aggregation is not specific to the expression of CTD fragments.

Next, we investigated whether TDP-43 aggregate seeding elicits loss of function in human neurons, as a more disease-relevant model. Induced pluripotent stem cells (iPSCs) with a doxycycline-inducible neurogenin-2 transgene integrated at the AAVS1 locus were differentiated into neurons (iNeurons) [59,60] and treated with FTD seeds or control one week post-differentiation (Fig 4E). iNeurons incubated for two to three weeks after treatment with FTD seeds or control extract showed comparable MAP2AB expression (Fig 4F), indicating similar neuronal differentiation and maturation. Treatment with FTD seeds induced a significant upregulation of ARHGAP32 CE inclusion relative to control treated cells (Fig 4G), suggesting that seeding-induced TDP-43 loss of function may be triggered in human neurons. We also observed increased expression of truncated STMN2, although this change did not reach statistical significance.

Aggregate seeding disrupts TDP-43 homeostasis through its impact on autoregulation—We next investigated whether TDP-43 functional defects triggered by aggregate seeding impact TDP-43 autoregulation, a critical process controlling TDP-43 homeostasis. TDP-43 regulates its own mRNA (TARDBP) and protein levels through a negative feedback loop that requires TDP-43 recruitment to the TARDBP 3’UTR within exon 6 [5,61](Fig 5A). This binding downregulates TARDBP mRNA expression by activating alternative poly(A) site usage, splicing and mRNA sequestration, which, combined, decrease protein expression [5,61,62]. Low TDP-43 levels enhance usage of the proximal poly(A) site (pA 1 ) and inhibits removal of alternative introns 6, 7 and 8 through splicing, overall increasing Short 3’UTR transcript expression and protein synthesis [62]. On the other hand, excess TDP-43 protein increases the accumulation of multiple transcript isoforms that are targets of nonsense-mediated decay (NMD), as well as the Long 3’UTR which is a product of distal poly(A) site (pA 4 ) selection that becomes sequestered in nuclear particles [5,62–64]. Maintaining physiological TDP-43 proteostasis through this evolutionarily conserved process is important for cell function as both diminished TDP-43 and elevated protein levels result in misregulated target expression. Moreover, abnormally increased TDP-43 concentration disrupts self-assembly and promotes aggregation ([65] for review). To determine whether aggregate seeding alters TDP-43 autoregulation, we compared TARDBP mRNA expression in FRET-positive and FRET-negative HEK293FRET cells treated with FTD seeds. We also sorted cells treated with control tissue extract as well as non-treated control. We found greater than 2-fold significant upregulation of TARDBP mRNA expression in FRET-positive, compared to FRET-negative cells treated with FTD seeds by qPCR (Fig 5B). This significant increase was also observed relative to control cells, which were almost all FRET-negative. Similar results were obtained using primers for qPCR that are common to Short and Long 3’UTR mRNA isoforms, Ex 2–3, and primers for Short 3’UTR (Ex 6) that do not recognize the NMD-sensitive isoforms. This primer pair captures the TDP-43 3’UTR binding region (TDPBR) upstream of pA 1 . In contrast, primers detecting the region closely upstream of the polyA 4 (distal), specific for the Long 3′ UTR isoform, did not show significant differences in expression. These results suggest that aggregate seeding specifically upregulates pA 1 usage and inhibits exon 6 splicing, thereby increasing expression of the Short 3’UTR isoform, consistent with a loss of active TDP-43. This expression profile is consistent with that observed in FTD/ALS patient-derived cells devoid of nuclear TDP-43 [8,62] and previous evidence showing specific upregulation of Short 3’UTR upon TDP-43 knockdown [62]. We next examined whether upregulation of the Short 3’UTR transcript in cells accumulating cytoplasmic aggregates correlates with increased TDP-43 protein expression. Unexpectedly, we detected no significant differences in protein levels between FRET-positive and FRET-negative sorted cells five days post-seeding (S7 Fig). The discrepancy between elevated Short 3’UTR transcript levels and unchanged protein expression may be due to activation of protein clearance mechanisms specifically in seeding-impacted cells, resulting in enhanced degradation of newly synthesized TDP-43. This hypothesis is supported by the recent report by Rummens and colleagues, which showed upregulation of genes involved in proteasomal and autophagosomal pathways in cells exposed to TDP-43 aggregate seeds [66].

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Aggregate seeding impacts TDP-43 autoregulation. A) Autoregulation is mediated by TDP-43 protein binding to the TDP-43 Binding Region (TDPBR, red) within the 3’UTR of its own transcript, upstream of the proximal polyadenylation site (pA 1 ) and near exon 6 splice sites (dashed lines indicate splicing events). TDP-43 recruitment promotes distal polyadenylation site (pA 4 ) usage and exon 6 splicing to generate isoforms which become targets for nonsense-mediated decay (NMD). In the absence of TDP-43 binding, the Short 3’UTR isoform is upregulated, resulting in increased protein expression. Indicated are the positions of primers for qPCR assays. Image created in BioRender, Ayala, Y. (2026) https://BioRender.com/pcvor00. B) Relative endogenous TARDBP mRNA expression quantified by qPCR in FACS-sorted FRET-positive (FRET+) and negative (FRET-) cells, six days after treatment with FTD seeds. GAPDH was used as reference. For each replicate, values were normalized to non-treated sample, mean and SEM are shown for n = 5. ****p < 0.0001, ns = non-significant, according to unpaired one-way ANOVA. The data underlying this figure can be found in S5 Data. https://doi.org/10.1371/journal.pbio.3003662.g005

TDP-43 aggregates specifically colocalize with modulators of aggregate seeding—Our findings showing strong co-localization of endogenous TDP-43 with cytoplasmic aggregates upon seeding (Fig 2B) led us to explore whether this recruitment is specific to TDP-43 and whether these inclusions act as sinks for proteins that interact with TDP-43 under normal and pathological conditions. We focused on RNA binding proteins with disordered or prion-like domains that interact with TDP-43 during RNA processing, stress granule assembly, or disease-associated conditions: hnRNP A2/B1, hnRNP C1/C2, hnRNP H1, FUS, matrin 3, SFPQ, G3BP1, TIAR, HSPB1, and ataxin-2 [67–71]. Detection of these proteins in non-treated control cells is shown in S10 Fig. Six days after treatment with FTD seeds we observed no significant colocalization of most proteins tested (Figs 6, S8A and S8B). These findings suggest that seeding-induced aggregates in HEK293FRET cells do not recruit proteins indiscriminately, even if these interact with TDP-43 under physiological conditions or during stress granule assembly. In contrast, ataxin-2 (ATXN2) and the small heat shock protein-1 (HSPB1) strongly colocalized with de novo cytoplasmic aggregates (Figs 7A and S8C arrows).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. Protein partners that normally bind TDP-43 are not recruited to seeding-induced aggregates. Representative immunofluorescence images, obtained by confocal microscopy, of HEK293FRET cells treated with FTD seeds detecting RNA binding proteins that form complexes with TDP-43 under physiological conditions. Neurological unaffected tissue extract was used as control. Images of non-treated control are shown in S10 Fig. Scale, 10 μm, n > 3. DAPI staining was used to highlight nuclei. https://doi.org/10.1371/journal.pbio.3003662.g006

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. Ataxin-2 specifically colocalizes with seeding-induced aggregates promoting their accumulation and loss of TDP-43 function. A) Immunofluorescence of HEK293FRET cells treated with FTD seeds and probed with ataxin-2 antibody after six days, obtained by confocal microscopy. Neurological unaffected tissue extract was used as control. Images of non-treated control are shown in S10 Fig. Ataxin-2 colocalization with TDP-43 CTD aggregates is denoted by arrows. Scale bar, 10 μm, n = 3. B) Two days after treatment with FTD seeds, HEK293FRET cells were transfected with siRNA to down-regulate ATXN2 or non-targeting siRNA, as negative control (S9A-C Fig). FRET-positive cells (% of total) quantified by flow cytometry and integrated FRET density are shown for n = 5. **p = 0.009, *p = 0.03, Mann-Whitney test. C) The mean fluorescence intensity of endogenous TDP-43 levels in the nucleus in cells treated with FTD seeds was quantified upon treatment with ATXN2 and control siRNA (S9E Fig). Nuclei were defined by DAPI staining of >380 cells from three independent experiments. Mean, ***p = 0.0008, Mann-Whitney test. D) Relative HDGFL2 and ARHGAP32 CE inclusion measured by qPCR, using GAPDH as reference. For each replicate, values were normalized to non-treated sample. Mean and SEM for n = 4 analyzed by one-way ANOVA, *p < 0.02, **p < 0.001, ns = non-significant. The data underlying this figure can be found in S6 Data. https://doi.org/10.1371/journal.pbio.3003662.g007

Previous reports showed that ATXN2 alters TDP-43 aggregation and toxicity in disease-relevant models [72,73]. This and our observations of ataxin-2 colocalization with TDP-43 cytoplasmic aggregates (Fig 7A) led us to examine whether ATXN2 expression influences seeding-induced TDP-43 aggregation. We knocked-down (KD) ATXN2 expression by siRNA in HEK293FRET cells, following treatment with FTD seeds, achieving approximately 50% reduction or greater in ATXN2 mRNA and protein levels, compared to non-targeting control siRNA (S9A-C Fig). Decreasing ATXN2 expression significantly reduced the accumulation of FRET-positive cells and integrated FRET density six days after treatment with FTD seeds (Fig 7B). ATXN2 KD did not significantly change the number of FRET-positive cells upon treatment with control tissue extract, which accounted for <1% of all cells (S9D Fig). Additionally, we found that ATXN2 KD significantly increased nuclear endogenous TDP-43 [21] in cells that were treated with FTD seeds, compared to control siRNA treatment (Fig 7C). These results suggest that reducing ataxin-2 levels by 50% or greater suppresses the loss of nuclear TDP-43 triggered by aggregation. To determine whether ATXN2 downregulation suppresses aggregation-induced TDP-43 loss of function, we measured ARHGAP32 and HDGFL2 CE inclusion, following ATXN2 KD. Six days post-transfection with FTD seeds, CE splicing significantly decreased by greater than 25% for both transcripts upon ATXN2 KD, compared to siRNA control (Fig 7D). In contrast, ATXN2 downregulation did not significantly alter CE splicing in cells treated with control extract. Together, our results suggest that ataxin-2 promotes seeding-induced TDP-43 aggregation as well as the ensuing loss of TDP-43 function. These observations are consistent with previous studies showing that ATXN2 enhances TDP-43 aggregation and neurotoxicity in human cells and animal models [72,74], and suggest that ataxin-2 potentiates the link between TDP-43 aggregation and depletion of cellular function. Based on our findings, specific protein recruitment to de novo TDP-43 aggregates in this system may indicate interactions that contribute to TDP-43 pathology.

[END]
---
[1] Url: https://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.3003662

Published and (C) by PLOS One
Content appears here under this condition or license: Creative Commons - Attribution BY 4.0.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/