CUTS RNA Biosensor for the Real-Time Detection of TDP-43 Loss-of-Function
Curation statements for this article:-
Curated by eLife
eLife Assessment
This important study developed a new sensor for TDP-43 activity that is sensitive and robust that should strongly impact the field's ability to monitor whether TDP-43 is functional or not. The evidence, though limited to cell culture, is compelling and is the first demonstration that a GFP on/off system can be used to assess genetic TDP-43 mutants as well as loss of soluble TDP-43.
This article has been Reviewed by the following groups
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
- Evaluated articles (eLife)
Abstract
Mounting evidence implicates TDP-43 dysfunction and the accumulation of pathological cryptic exons across multiple neurodegenerative diseases, underscoring the need for accessible tools to detect and quantify TDP-43 loss-of-function (LOF). These tools are crucial for assessing potential disease contributors and exploring therapeutic candidates in TDP-43 proteinopathies. Here, we develop a sensitive and accurate real-time sensor for TDP-43 LOF: the CUTS (CFTR UNC13A TDP-43 Loss-of-Function) system. This system combines UG-rich sequences and previously reported cryptic exons regulated by TDP-43 with a reporter, enabling the tracking of TDP-43 LOF through live microscopy and RNA/protein-based assays. We show that CUTS effectively detects TDP-43 loss of function arising from mislocalization, impaired RNA binding, and pathological aggregation. Our results show the sensitivity and accuracy of the CUTS system in detecting and quantifying TDP-43 LOF, opening avenues to explore unknown TDP-43 interactions that regulate its function. In addition, by replacing the fluorescent tag in the CUTS system with the coding sequence for TDP-43, we show significant recovery of its function under TDP-43 LOF conditions, highlighting the potential utility of CUTS for self-regulating gene therapy applications. In summary, CUTS represents a platform for evaluating TDP-43 LOF in real-time and gene-replacement therapies in neurodegenerative diseases associated with TDP-43 dysfunction.
Article activity feed
-
eLife Assessment
This important study developed a new sensor for TDP-43 activity that is sensitive and robust that should strongly impact the field's ability to monitor whether TDP-43 is functional or not. The evidence, though limited to cell culture, is compelling and is the first demonstration that a GFP on/off system can be used to assess genetic TDP-43 mutants as well as loss of soluble TDP-43.
-
Reviewer #2 (Public review):
Summary:
The authors goals is to be develop a more accurate system that reports TDP-43 activity as a splicing regulator. Prior to this, most methods employed western blotting or QPCR based assays to determine whether targets of TDP-43 were up or down regulated. The problem with that is the sensitivity. This approach uses an ectopic delivered construct containing splicing elements from CFTR and UNC13A (two known splicing targets) fused to a GFP reporter. Not only does it report TDP-43 function well, but it operates at extremely sensitive TDP-43 levels, requiring only picomolar TDP-43 knockdown for detection. This reporter should supersede the use of current TDP-43 activity assays, its cost-effective, its rapid and reliable.
Strengths:
In general, the experiments are convincing and well designed. The rigor, …
Reviewer #2 (Public review):
Summary:
The authors goals is to be develop a more accurate system that reports TDP-43 activity as a splicing regulator. Prior to this, most methods employed western blotting or QPCR based assays to determine whether targets of TDP-43 were up or down regulated. The problem with that is the sensitivity. This approach uses an ectopic delivered construct containing splicing elements from CFTR and UNC13A (two known splicing targets) fused to a GFP reporter. Not only does it report TDP-43 function well, but it operates at extremely sensitive TDP-43 levels, requiring only picomolar TDP-43 knockdown for detection. This reporter should supersede the use of current TDP-43 activity assays, its cost-effective, its rapid and reliable.
Strengths:
In general, the experiments are convincing and well designed. The rigor, number of samples and statistics, and gradient of TDP-43 knockdown were all viewed as strengths. In addition, the use of multiple assays to confirm the splicing changes were viewed as complimentary (ie PCR and GFP-fluorescence) adding additional rigor. The final major strength i'll add is the very clever approach to tether TDP-43 to the loss of function cassette such that when TDP-43 is inactive it would autoregulate and induce wild-type TDP-43. This has many implications for the use of other genes, not just TDP-43, but also other protective factors that may need to be re-established upon TDP-43 loss of function.
Weaknesses:
Admittedly, one needs to initially characterize the sensor and the use of cell lines is an obvious advantage, but it begs the question of whether this will work in neurons. Additional future experiments in primary neurons will be needed. The bulk analysis of GFP-positive cells is a bit crude. As mentioned in the manuscript, flow sorting would be an easy and obvious approach to get more accurate homogenous data. This is especially relevant since the GFP signal is quite heterogenous in the image panels, for example Figure 1C, meaning the siRNA is not fully penetrant. Therefore, stating that 1% TDP-43 knockdown achieves the desired sensor regulation might be misleading. Flow sorting would provide a much more accurate quantification of how subtle changes in TDP-43 protein levels track with GFP fluorescence.
Some panels in the manuscript would benefit from additional clarity to make the data easier to visualize. For example, Figure 2D and 2G could be presented in a more clear manner, possibly split into additional graphs since there are too many outputs. Sup Figure 2A image panels would benefit from being labeled, its difficult to tell what antibodies or fluorophores were used. Same with Figure 4B.
Figure 3 is an important addition to this manuscript and in general is convincing showing that TDP-43 loss of function mutants can alter the sensor. However, there is still wild-type endogenous TDP-43 in these cells, and its unclear whether the 5FL mutant is acting as a dominant negative to deplete the total TDP-43 pool, which is what the data would suggest. This could have been clarified. Additional treatment with stressors that inactivate TDP-43 could be tested in future studies.
Overall, the authors definitely achieved their goals by developing a very sensitive readout for TDP-43 function. The results are convincing, rigorous, and support their main conclusions. There are some minor weaknesses listed above, chief of which is the use of flow sorting to improve the data analysis. But regardless, this study will have an immediate impact for those who need a rapid, reliable, and sensitive assessment of TDP-43 activity, and it will be particularly impactful once this reporter can be used in isolated primary cells (ie neurons) and in vivo in animal models. Since TDP-43 loss of function is thought to be a dominant pathological mechanism in ALS/FTD and likely many others disorders, having these type of sensors is a major boost to field and will change our ability to see sub-threshold changes in TDP-43 function that might otherwise not be possible with current approaches.
Comments on revisions:
In the revised version, most of the reviewer's comments have been appropriately addressed with the exception of 1) the use of flow sorting to improve the data analysis and 2) testing this sensor in primary neurons. The latter is the focus of an ongoing separate study. Though flow sorting would significantly strengthen this study and help others in the field to use this sensor, it is still an impactful and innovative study without it.
-
Reviewer #3 (Public review):
The DNA and RNA binding protein TDP-43 has been pathologically implicated in a number of neurodegenerative diseases including ALS, FTD, and AD. Normally residing in the nucleus, in TDP-43 proteinopathies, TDP-43 mislocalizes to the cytoplasm where it is found in cytoplasmic aggregates. It is thought that both loss of nuclear function and cytoplasmic gain of toxic function are contributors to disease pathogenesis in TDP-43 proteinopathies. Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function characterized by changes in gene expression and splicing of target mRNAs. However, to date, most readouts of TDP-43 loss of function events are dependent upon PCR based assays for single mRNA targets. Thus, reliable and robust assays for detection of global changes in …
Reviewer #3 (Public review):
The DNA and RNA binding protein TDP-43 has been pathologically implicated in a number of neurodegenerative diseases including ALS, FTD, and AD. Normally residing in the nucleus, in TDP-43 proteinopathies, TDP-43 mislocalizes to the cytoplasm where it is found in cytoplasmic aggregates. It is thought that both loss of nuclear function and cytoplasmic gain of toxic function are contributors to disease pathogenesis in TDP-43 proteinopathies. Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function characterized by changes in gene expression and splicing of target mRNAs. However, to date, most readouts of TDP-43 loss of function events are dependent upon PCR based assays for single mRNA targets. Thus, reliable and robust assays for detection of global changes in TDP-43 splicing events are lacking. In this manuscript, Xie, Merjane, Bergmann and colleagues describe a biosensor that reports on TDP-43 splicing function in real time. Overall, this is a well-described unique resource that would be of high interest and utility to a number of researchers validated in multiple cell types as a sensitive readout of TDP-43 loss of function. Future studies validating the utility of this biosensor in models of TDP-43 loss of function (e.g. disease iPSNs) that do not rely on TDP-43 knockdown will be of further interest.
-
Author Response:
The following is the authors’ response to the previous reviews
Public Review:
We thank the editor and reviewers for their thoughtful and constructive feedback, which has enabled us to greatly strengthen the manuscript. We apologize for the delay in resubmitting this as we were dealing with a large turnover in the lab due to trainee graduations which has We have carefully revised the text, figures, and supplementary materials in response to these comments. Below, we summarize the key revisions made followed by a point-by-point response to the reviewers’ critiques.
(1) Performed CUTS analyses in human neuronal system: In the revised manuscript, we included new data demonstrating that the CUTS system can be applied to additional cellular models, specifically neuronal cells (Figure 5, Figure S4). To address whether CUTS …
Author Response:
The following is the authors’ response to the previous reviews
Public Review:
We thank the editor and reviewers for their thoughtful and constructive feedback, which has enabled us to greatly strengthen the manuscript. We apologize for the delay in resubmitting this as we were dealing with a large turnover in the lab due to trainee graduations which has We have carefully revised the text, figures, and supplementary materials in response to these comments. Below, we summarize the key revisions made followed by a point-by-point response to the reviewers’ critiques.
(1) Performed CUTS analyses in human neuronal system: In the revised manuscript, we included new data demonstrating that the CUTS system can be applied to additional cellular models, specifically neuronal cells (Figure 5, Figure S4). To address whether CUTS functions effectively in neuronal contexts, we generated stable CUTS-expressing lines in differentiated BE(2)-C and ReN VM–derived differentiated neurons (Figure 5A-D, Figure S4 A-C). To ensure this was neuronal expression, we developed a new Tet-On3G system construct where the Tet-On3G transactivating protein is driven by the SYN1 promoter to ensure neuron-specific inducible expression for these experiments.
(2) Define the relationship between CUTS and endogenous/physiological cryptic exons inclusion: To evaluate how well the CUTS system reflects physiological cryptic exon regulation, we performed RT-PCR analysis of several cryptic exons previously reported by us and evaluated CUTS activation at the RNA level in parallel (Figure S2E) . CUTS is sensitive to low-mild reductions in TDP-43 levels, whereas the tested endogenous cryptic exons exhibit variable responses to TDP-43 knockdown.
(3) Defining stress-induced TDP-43 loss of function: We included new data demonstrating that the CUTS system can detect TDP-43 loss of function induced by acute sodium arsenite (NaAsO₂) treatment in HEK cells (Figure 3D–I). We have also tested additional stressor as part of a separate ongoing study where this work will be expanded upon (Xie et al., 2025). We selected this paradigm since TDP-43 loss of function in response to acute NaAsO₂ treatment is also supported by work from other labs(Huang et al., 2024).
(4) Implications of using a TDP-43 Loss-of-Function sensor for therapeutic applications: In the revised manuscript, we clarify that CUTS-TDP43 is auto-regulated and we highlight two potential therapeutic applications: i) TDP-43 Knockdown-and-replacement: CUTS-TDP43 provides a strategy for simultaneous depletion of pathological TDP-43 species while enabling autoregulated re-expression of wild-type TDP-43. This design mitigates the risk of supraphysiologic overexpression, a known liability in conventional replacement approaches, by restoring TDP-43 within a self-limiting regulatory network that maintains homeostatic control. ii) Aggregation-independent correction: Because CUTS is autoregulatory, it can be repurposed to regulate alternative downstream effectors, including splicing modifiers or TDP-43 functional interactors, without expressing TDP-43 itself. This approach provides a potential aggregation-independent strategy to compensate for TDP-43 loss-of-function (LOF) by restoring downstream splicing. We are evaluating this work in a follow up study (Xie et al., 2025). In these ongoing studies, we show that CUTS-regulated expression of splicing proteins in response to TDP-43 loss restored subsets of cryptic exon events (24/28 events evaluated). These findings suggest CUTS as a versatile tool for both autoregulated TDP-43 replacement and trans-regulatory therapeutic correction. We expanded on this concept in the discussion section of this revised manuscript. We also note that autoregulatory TDP-43 biosensor strategies have been proposed in related systems, including TDP-Reg, underscoring broader interest in self-regulated TDP-43 systems (Wilkins et al., 2024).
(5) Clarified mechanism of TDP-43 5FL causing strong loss of function: The TDP-43 5FL exhibits reduced RNA binding capacity, and we previously showed that the lack of RNA binding promotes aberrant homotypic phase separation of TDP-43 (Mann et al., 2019). Expression of RNA-deficient TDP-43 variant forms nuclear “anisomes” (Yu et al., 2021), which evidence suggests sequesters endogenous TDP-43 protein into insoluble structures. We expanded on this in our results section in this revised manuscript.
(6) Improved figure clarity and data presentation: To enhance clarity and organization, we maintained the main structure of the manuscript while reorganizing figures and improved data visualization. Some examples include:
Figure 1: We revised the schematic layout for greater clarity and simplicity. The figure now focuses more specifically on the CUTS data, with additional data on the UNC13A-TS and CFTR-TS moved to Figure S1. To improve readability, titles were added to all schematic panels. Visual consistency was also improved by refining the color labelling for each sensor in Figures 1C and 1D and adjusting the corresponding bar graphs accordingly.
Figure 2: We reorganized the figure to clearly distinguish between protein and mRNA analyses for greater clarity. In the revised layout, western blot quantifications of TDP-43 and CUTS (GFP) signals are shown in Figures 2D and 2E, respectively, while the corresponding qPCR analyses are presented in Figures 2H and 2I. Minor edits include removing the percentage knockdown and fold-change annotations from the graphs and incorporating these values into a mini-table in Figure S2E.
The original Figure 2D and 2G were reincorportated as reference panels in Figure S2A–B, while new graphs showing CUTS protein-level changes as a function of TDP-43 knockdown were added (Figure S2C–D). We also incorporated new data showing the behavior of endogenous cryptic exons under low siTDP-43 treatment (Figure S2E).
Figure 3: We added new data demonstrating that the application of the CUTS system in detecting TDP-43 loss of function induced by stress conditions. Specifically, we show that sodium arsenite (NaAsO₂) treatment leads to TDP-43 functional impairment detectable by CUTS and supported with endogenous cryptic exon via RT-PCR (Figure 3D-I).
Figure 5 and Figure S4: We introduced a new figure that demonstrates the effective application of the CUTS system in differentiated neuronal systems, thereby extending its usability to disease-relevant cell types.
Figures 2SA and 4B were edited to include the corresponding labels on the sides of each image for clarity. Sup Figure 2A was moved to Sup Figure 3A, while Figure 4B remains in its original configuration.
We thank the reviewers again for their insightful critiques and helpful suggestions, which have enabled us to substantially improve the manuscript. Please find our detailed response to each review below:
Reviewer #1 (Public review):
Summary:
The authors create an elegant sensor for TDP -43 loss of function based on cryptic splicing of CFTR and UNC13A. The usefulness of this sensor primarily lies in its use in eventual high throughput screening and eventual in vivo models. The TDP-43 loss of function sensor was also used to express TDP-43 upon reduction of its levels.
Strengths:
The validation is convincing, the sensor was tested in models of TDP-43 loss of function, knockdown and models of TDP-43 mislocalization and aggregation. The sensor is susceptible to a minimal decrease of TDP-43 and can be used at the protein level unlike most of the tests currently employed,
Weaknesses:
Although the LOF sensor described in this study may be a primary readout for high-throughput screens, ALS/TDP-43 models typically employ primary readouts such as protein aggregation or mislocalization. The information in the two following points would assist users in making informed choices.
(1) Testing the sensor in other cell lines
We thank the reviewer for raising this important point. In agreement with this suggestion, we generated ReN VM cell lines and used a neuroblastoma cell line model (BE(2)-C) expressing the TetOn3G CUTS system under a human synapsin I (hSYN1) promoter. In this construct the transactivator protein is under the control of a neuronal specific hSYN1 promoter whereas the classical TetOn3G system uses a CMV-like promoter. Several studies have reported reduced activity or silencing of CMV and PGK-driven transgenes in neurons. Therefore, we for our neuronal experiments, we removed this promoter to generate a new version of a doxycycline-inducible CUTS system in which Tet-On 3G transactivator is now driven by the hSYN1 promoter which will express CUTS in response to doxycycline treatment. In this improved construct, we also replaced mCherry with mScarlet to enhance the fluorescent signal.
To test this neuronal-adapted system, we established stable CUTS expression in undifferentiated BE(2)-C cells, a subclone of the SK-N-BE(2) neuroblastoma line that has been used to study TDP-43–dependent splicing function(Brown et al., 2022). This model can be differentiated into neuron-like cells within 10 days, as shown in Supplementary Figure 4A. Using this model, we confirmed that TDP-43 knockdown leads to robust activation of the CUTS system (Figure 5B-E). We additionally tested this in in a stable polyclonal ReN VM cells following differentiation into cortical-like neurons (Figure 5D, Figure S4B-C).
(2) Establishing a correlation between the sensor's readout and the loss of function (LOF) in the physiological genes would be useful given that the LOF sensor is a hybrid structure and doesn't represent any physiological gene. It would be beneficial to determine if a minor decrease (e.g., 2%) in TDP-43 levels is physiologically significant for a subset of exons whose splicing is controlled by TDP43.
We agree with the reviewer that correlating the sensor’s readout with physiological TDP-43 splicing targets is essential to validate its biological relevance. To this end, we complemented our sensor expression profile with endogenous cryptic exons (CEs) sensitive to TDP-43 depletion. We tested a panel of five physiological cryptic exons regulated by TDP-43 (LRP8, EPB41L4A, ARHGAP32, HDGFL2, and ACBD3). To address the reviewer’s concerned, we performed RT-PCR on samples from the low-dose siTDP-43 experiment shown in Figure S2E.
The endogenous CEs used in the panel were selected based on our own and others’ preliminary observations. Among these, HDGFL2 showed a particularly robust increase in cryptic exon inclusion at very low siTDP-43 concentrations (38 pM), while untreated samples showed almost no CE inclusion. This finding strongly supports a direct mechanism linking mild TDP-43 reduction to loss of physiological splicing control.
(3) Considering that most TDP-LOF pathologically occurs due to aggregation and or mislocalization, and in most cases the endogenous TDP-43 gene is functional but the protein becomes non-functional, the use of the loss of function sensor as a switch to produce TDP-43 and its eventual use as gene therapy would have to contend with the fact that the protein produced may also become nonfunctional. This would eventually be easy to test in one of the aggregation modes that were used to test the sensor.. However, as the authors suggest, this is a very interesting system to deliver other genetic modifiers of TDP-43 proteinopathy in a regulated fashion and timely fashion.
We thank the reviewer for this thoughtful point and agree that in the disease-relevant context where endogenous TDP-43 is intact but TDP-43 function is lost due to mislocalization and/or aggregation, a re-supply of TDP-43 risks sequestration and loss of activity. In our manuscript, the CUTS-TDP43 module was presented as a control circuit proof-of-concept rather than a stand-alone approach: it demonstrates that CUTS can (i) sense LOF with high dynamic range and proportionality, and (ii) drive a payload under negative feedback such that total TDP-43 remains near baseline while partially rescuing a splicing readout (CFTR minigene) under knockdown conditions.
Importantly, we evaluated CUTS in aggregation/mislocalization-prone contexts: ΔNLS, 5FL, and ΔNLS+5FL variants trigger CUTS activation (ref), allowing us to quantify LOF arising from these aggregation modes. This confirms that CUTS can operate precisely in the very settings where sequestration is likely to occur.
To directly address the reviewer’s suggestion, in the revision we (i) clarify in the Discussion that CUTS-TDP43 is a circuit demonstration and not our proposed monotherapy in aggregation-dominant disease; and (ii) expand our therapeutic framing into two approaches:
Knockdown-and-replacement: concurrently deplete aggregation-prone/endogenous pathologic TDP-43 species (i.e., mutant TDP-43) while using CUTS to re-deliver wild-type TDP-43 under autoregulation. Aggregation-independent correction: use of CUTS to deliver modifiers that bypass TDP-43 sequestration (e.g., downstream effectors or splicing correctors that restore LOF consequences without expressing TDP-43 itself).
(4) I don't think the quantity of siRNA is directly proportional to the degree of TDP-43 knockdown/extent of TDP-43 loss. Therefore, to enhance the utility of the dose-response curves, I'd suggest using TDP-43 levels as the variable on the x-axis, rather than the amount of siRNA administered or even just adding a plot alongside the current plots would enable readers to quickly evaluate LOF response levels concerning the protein. While I understand that the sensitivity of Western blots for quantification might be why the authors have not created the graphs in this manner, having this information would be useful.
We appreciate the reviewer’s insightful comment. As noted, in the original version of the graph, we incorporated the percentage of TDP-43 knockdown corresponding to each siTDP-43 concentration (indicated in red text). However, we agree that this format was not easy to interpret, given the amount of information presented. To address this, we generated two new plots in which the x-axis represents TDP-43 levels (percentage of remaining protein or mRNA), and the y-axis shows the fold change in CUTS signal measured by (i) TDP-43 protein pixel intensity and (ii) TDP-43 mRNA levels, respectively. These new plots are now included as Supplementary Figures 2C–D, which allow a clearer visualization of CUTS readout in relation to actual TDP-43 levels rather than siRNA dose. As the reviewer anticipated, the reason we did not originally present the data in this format was that at low siTDP-43 concentrations, the fold change is minimal and more difficult to quantify by Western blot. Nevertheless, we have now incorporated the revised plots to strengthen the interpretation of the dose–response relationship. Additionally, we experience batch effects across siRNA lots. We believe this revised format should enhance the clarity of the result.
(5) p3 line 74: one of the reasons cited as a pitfall of using the endogenous cryptic exons exhibit variable responses to TDP-43 loss and may be cell type-specific. has the sensor been used in different cell lines?
We tested the CUTS system in differentiated neuronal models using two differentiated neuronal cell types, BE(2)C and ReN VM cells. The results are presented in Figure 5 and Figure S4 of the revised manuscript.
(6) The order of the text describing 1A and 1B is confusing. The text starts describing the TS cassettes referring to 1A using the CUTS cassettes which haven't been introduced yet as an example. I'd suggest reorganising this section. The graph, always in 1A showing readout proportional to GFP should be taken out or highlighted in the figure legend that it is theoretical.
We agree with the reviewer’s point. In the original schematic (Figure 1A), we included the CUTS system as an example to introduce the TS cassette design, since it contains the three possible sensor configurations. However, we recognize that this could be confusing. Therefore, we have removed the CUTS cassette from Figure 1A, along with the theoretical graph showing GFP readout proportional to the degree of TDP-43 LOF. In agreement with this change, we also restructured Figure 1. As the focus is the CUTS system, we have moved the Western blot and quantification of UNC13A-TS and CFTR-TS to Supplementary Figure 1.
Reviewer #2 (Public review):
Summary:
The authors goal is to develop a more accurate system that reports TDP-43 activity as a splicing regulator. Prior to this, most methods employed western blotting or QPCR-based assays to determine whether targets of TDP-43 were up or down-regulated. The problem with that is the sensitivity. This approach uses an ectopic delivered construct containing splicing elements from CFTR and UNC13A (two known splicing targets) fused to a GFP reporter. Not only does it report TDP-43 function well, but it operates at extremely sensitive TDP-43 levels, requiring only picomolar TDP-43 knockdown for detection. This reporter should supersede the use of current TDP-43 activity assays, it's cost-effective, rapid and reliable.
Strengths:
In general, the experiments are convincing and well designed. The rigor, number of samples and statistics, and gradient of TDP-43 knockdown were all viewed as strengths. In addition, the use of multiple assays to confirm the splicing changes were viewed as complimentary (ie PCR and GFPfluorescence) adding additional rigor. The final major strength I'll add is the very clever approach to tether TDP-43 to the loss of function cassette such that when TDP-43 is inactive it would autoregulate and induce wild-type TDP-43. This has many implications for the use of other genes, not just TDP-43, but also other protective factors that may need to be re-established upon TDP-43 loss of function.
Weaknesses:
(1) Admittedly, one needs to initially characterize the sensor and the use of cell lines is an obvious advantage, but it begs the question of whether this will work in neurons. Additional future experiments in primary neurons will be needed.
We thank the reviewer for highlighting the importance of validating the sensor in neuronal models, given the central role of TDP-43 dysfunction in ALS/FTD and related neurodegenerative disorders. While initial characterization in established cell lines provides experimental control and scalability, we agree that demonstrating functionality in neuronal systems is essential. To address this, we adapted the CUTS platform for neuronal application by incorporating the human synapsin-1 (hSYN1) promoter into the Tet-On 3G system to enable inducible, neuronal specific expression. We validated this configuration in differentiated BE(2)-C cells (Figures 5A-C, S4A-C), where CUTS retained robust responsiveness to TDP-43 perturbation. In parallel, we generated stable CUTS-expressing ReN VM neural progenitor cells and differentiated them for three weeks prior to functional assessment (Figures 5A-C, S4A-C). In both neuronal models, CUTS was functional and responsive to TDP-43 siRNA. We are currently optimizing promoter selection and expression paradigms for fully differentiated iPSC-derived neuronal models and will be the subject of future studies.
(2) The bulk analysis of GFP-positive cells is a bit crude. As mentioned in the manuscript, flow sorting would be an easy and obvious approach to get more accurate homogenous data. This is especially relevant since the GFP signal is quite heterogeneous in the image panels, for example, Figure 1C, meaning the siRNA is not fully penetrant. Therefore, stating that 1% TDP-43 knockdown achieves the desired sensor regulation might be misleading. Flow sorting would provide a much more accurate quantification of how subtle changes in TDP-43 protein levels track with GFP fluorescence.
We thank the reviewer for this thoughtful suggestion. We agree that flow cytometry and sorting of GFP-positive populations would provide a higher-resolution, single-cell–level relationship between TDP-43 abundance and sensor output. Such an approach would reduce heterogeneity arising from incomplete siRNA penetrance and allow more precise quantification of how incremental changes in TDP-43 protein levels track with GFP fluorescence. In the present study, our goal was to establish proof-of-principle functionality of the CUTS circuit and to demonstrate that graded TDP-43 depletion produces a proportional sensor response at the population level. While GFP signal heterogeneity is visible in imaging panels, we hypothesize that this variability likely reflects known differences in siRNA uptake and transfection efficiency rather than instability of the circuit itself. Importantly, bulk measurements consistently demonstrated dose-dependent sensor regulation across independent experiments, supporting the robustness of the system despite cellular heterogeneity. Furthermore, we were able to quantify CUTS activation in HeLa TARDBP-/- cells. We also note that CUTS was developed as a practical tool for rapid assessment of TDP-43 LOF in standard laboratory settings. Although flow cytometry increases resolution, the ability to detect functional perturbation using bulk fluorescence measurements supports the utility of the system for routine and high-throughput applications.
We agree that flow cytometry would provide a more refined analysis of the dynamic range and sensitivity of CUTS, particularly for defining thresholds such as minimal TDP-43 knockdown required for measurable activation. We plan to include this work in future studies. Specifically, we have implemented FACs sorting of CUTS-expressing cells in a parallel study in which we are conducting a CRISPR knockout screen to identify modifiers of TDP-43 splicing function. For this, we incorporate TDP-43 knockdown followed by FACs to stratify cells based on CUTS activation. This strategy enables direct evaluation of the relationship between the extent of TDP-43 LOF and CUTS sensor activation. These analyses are ongoing and provide a more quantitative analyses linking TDP-43 depletion to CUTS activation and address the reviewer’s concern regarding heterogeneity in bulk measurements. We plan to include this in a future study.
(3) Some panels in the manuscript would benefit from additional clarity to make the data easier to visualize. For example, Figure 2D and 2G could be presented in a more clear manner, possibly split into additional graphs since there are too many outputs.
We thank the reviewer for this suggestion. In response, we have split the graphs previously shown in Figures 2D and 2G to improve clarity, as we agree that these panels contained an extensive amount of data. We Specifically split Figure 2D into two separate graphs showing TDP-43 and GFP pixel intensity from Western blots on the Y-axis, plotted against low siTDP-43 treatment on the X-axis. Please see this data as Figure 2 D and Figure 2E in the new manuscript.
Furthermore, for Figure 2G we also split into graphs showing the fold change of mRNA for TDP-43 and the CUTS cryptic exon plotted against low siTDP-43 treatment on the X-axis. Please see this data as Figure 2 H and Figure 2I in the new manuscript. We have maintained the previous graphs in Supplementary Figure 2 to preserve the full dataset for reference.
(4) Sup Figure 2A image panels would benefit from being labeled, its difficult to tell what antibodies or fluorophores were used. Same with Figure 4B.
We appreciate the reviewer’s careful observation. In both figures, we are showing mCherry and GFP signals. In the revised version, we have added the corresponding labels to the side of each image for clarity. Therefore, Sup Figure 2A has been moved and is now Sup Figure 3A, while Figure 4B remains in its original configuration.
(5) Figure 3 is an important addition to this manuscript and in general is convincing showing that TDP43 loss of function mutants can alter the sensor. However, there is still wild-type endogenous TDP-43 in these cells, and it's unclear whether the 5FL mutant is acting as a dominant negative to deplete the total TDP-43 pool, which is what the data would suggest. This could have been clarified.
The TDP-43 5FL variant exhibits reduced RNA-binding capacity, and we previously demonstrated that impaired RNA binding promotes aberrant homotypic phase separation of TDP-43. Consistent with this mechanism, expression of RNA-binding–deficient TDP-43 variants induces the formation of nuclear “anisomes” which have been shown to sequester endogenous TDP-43 into insoluble fractions via dominant-negative mechanisms (Cohen et al., 2015; Keating et al., 2023; Mann et al., 2019; Yu et al., 2021). These findings support a model in which disruption of RNA engagement alters TDP-43 biophysical behavior and promotes functional depletion through self-association. We have expanded this mechanistic explanation in the Results section of the revised manuscript to better contextualize the behavior of the 5FL construct and its impact on endogenous TDP-43.
(6) Additional treatment with stressors that inactivate TDP-43 could be tested in future studies.
We appreciate this suggestion and agree with this important point. Due to the lack of methods to directly induce endogenous TDP-43 aggregation and loss of function, the use of stressors has become a partial solution to address this issue. In line with this, our group has tested several stressors in follow-up research, including sodium arsenite (NaAsO₂), puromycin, KCl, MG132, sorbitol, and tunicamycin, using HEK cells expressing the CUTS system(Xie et al., 2025). We were able to show a dose-response relationship in relative GFP intensity under these conditions, with sodium arsenite showing the strongest effect, consistent with previous reports(Huang et al., 2024). To provide additional relevant findings in the current manuscript, we expanded this analysis by testing sodium arsenite in the CUTS system while also including endogenous cryptic exons. We therefore added a new figure showing the effect of sodium arsenite on the CUTS system, including GFP intensity measurements, qPCR using CUTS cryptic exon primers, and three endogenous cryptic exon reporters (ATG4B, GPSM2, and KCNQ2).
Overall, the authors definitely achieved their goals by developing a very sensitive readout for TDP-43 function. The results are convincing, rigorous, and support their main conclusions. There are some minor weaknesses listed above, chief of which is the use of flow sorting to improve the data analysis. But regardless, this study will have an immediate impact for those who need a rapid, reliable, and sensitive assessment of TDP-43 activity, and it will be particularly impactful once this reporter can be used in isolated primary cells (ie neurons) and in vivo in animal models. Since TDP-43 loss of function is thought to be a dominant pathological mechanism in ALS/FTD and likely many other disorders, having these types of sensors is a major boost to the field and will change our ability to see sub-threshold changes in TDP-43 function that might otherwise not be possible with current approaches.
(7) Regarding the methods, they seem a bit sparse and would benefit from additional detail. For example, I do not see a section in the methods where microscopy images were quantified (%GFP positive cells for example). This information is important and is lacking in the current form.
We thank the reviewers, and we add the following information in the method section: For live imaging quantification, we measured the mean GFP signal intensity for each group. The values were averaged, and the fold change was calculated and plotted. For immunofluorescent imaging, we first created maximum intensity projection images. We then applied masks to the GFP, mCherry, and Hoechst signals. By overlapping the GFP and mCherry signals, we identified the number of GFP-positive cells. Similarly, by overlapping the mCherry signal with the Hoechst mask, we identified the CUTS-expressing cells. We then calculated the ratio of GFPpositive cells to CUTS-expressing cells and plotted it as a percentage of GFP-positive cells. All analyses were performed using the Nikon NIS software. This information is included in the methods of the revised manuscript.
Reviewer #3 (Public review):
The DNA and RNA binding protein TDP-43 has been pathologically implicated in a number of neurodegenerative diseases including ALS, FTD, and AD. Normally residing in the nucleus, in TDP-43 proteinopathies, TDP-43 mislocalizes to the cytoplasm where it is found in cytoplasmic aggregates. It is thought that both loss of nuclear function and cytoplasmic gain of toxic function are contributors to disease pathogenesis in TDP-43 proteinopathies. Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function characterized by changes in gene expression and splicing of target mRNAs. However, to date, most readouts of TDP-43 loss of function events are dependent upon PCR-based assays for single mRNA targets. Thus, reliable and robust assays for detection of global changes in TDP-43 splicing events are lacking. In this manuscript, Xie, Merjane, Bergmann and colleagues describe a biosensor that reports on TDP-43 splicing function in real time. Overall, this is a well described unique resource that would be of high interest and utility to a number of researchers. Nonetheless, a couple of points should be addressed by the authors to enhance the overall utility and applicability of this biosensor.
(1) While the rationale for selecting UNC13A CE as the reporting CE species is understood given the relevance to disease, could the authors please comment on whether other CE sequences would behave similarly or as robustly? This is particularly critical given the multitude of different splicing changes that can occur as a result of TDP-43 loss of function (ie cryptic exons of differing sensitivity, skiptic exons, premature polyadenylation).
We thank the reviewer for this question regarding generalizability beyond the UNC13A CE. While UNC13A was selected due to its strong disease relevance and well-characterized sensitivity to TDP-43 loss-of-function (LOF), our platform is not intrinsically restricted to this sequence. In the manuscript, we directly compared three architectures: UNC13A-TS, CFTR-TS, and the combined CUTS sensor incorporating additional UG motif optimization. Under matched conditions in stable HEK293 lines, CUTS demonstrated superior specificity and sensitivity, exhibiting near-zero baseline activity and a proportional, log-linear response across low-dose siTDP43 (38–1200 pM) (Figures 1–2). Importantly, this head-to-head comparison demonstrates that sensor performance can be engineered and optimized beyond a single CE species.
TDP-43 LOF is known to induce a spectrum of RNA processing defects, including cryptic exons with differing sensitivities and cell-type dependence, premature polyadenylation events (e.g., STMN2), and, under conditions of excess nuclear TDP-43, exon skipping (“skiptic exons”). This diversity supports the concept in which alternative CE elements, or other TDP-43 regulated RNAs, can be incorporated into the same sensor backbone and tuned for specific biological scenarios (cell type, specific stress responses, etc...). Consistent with this, the recently described TDP-REG system (Wilkins et al., 2024) designed and AI-generated de novo CE sequences to express reporters or gene payloads, and screened multiple candidates to identify the appropriate RNA elements required for this response. These findings demonstrate that CE sequences beyond UNC13A can serve as robust TDP-43 sensing elements when optimized. Our results complement this work by demonstrating that CUTS achieves tight baseline control and a steep dynamic range (>110,000-fold induction over baseline in HEK293 cells), while maintaining compatibility across both non-neuronal and neuronal model systems, as shown in the revised manuscript.
In the revised manuscript, we show direct comparisons indicating that CUTS outperforms single-CE sensors such as UNC13A-TS and CFTR-TS under identical conditions. This supports independent work from other groups that alternative CE sequences can be engineered into effective sensors, depending on their paradigm and model systems. We have clarified this in the revised Discussion and now note that CUTS is adaptable to alternative CE inserts.
(3) Could the authors provide evidence of the utility of their biosensor in disease relevant systems that do not rely on TDP-43 KD? For example, does this biosensor report on TDP-43 loss of function in C9orf72 iPSNs in a time-dependent manner? Alternatively, groups have modeled TDP-43 proteinopathy in wildtype iPSNs via MG132 treatment.
We thank the reviewer for this important suggestion. We agree that demonstrating CUTS responsiveness in disease-relevant models independent of artificial TDP-43 knockdown would further strengthen its translational relevance. In the current study, our primary objective was to establish the sensitivity, dynamic range, and autoregulatory properties of the CUTS circuit under controlled perturbation of TDP-43 levels. siRNA-mediated depletion provides a reliable approach to establish the relationship between graded TDP-43 LOF and the CUTS sensor sensitivity/specificity. That said, CUTS is designed to detect functional TDP-43 loss irrespective of the upstream cause. As the reviewer notes, disease-relevant systems, such as C9orf72 iPSC-derived neurons and proteotoxic stress paradigms (e.g., MG132-induced impairment of TDP-43 nuclear function), are important for future studies. We are currently evaluating CUTS in iPSC-derived neuronal models of TDP-43 proteinopathy, but are optimizing the induction system, promoters, and timing. It should be noted that C9orf72 iPSC neurons do not exhibit TDP-43 LOF using standard differentiation protocols. Regarding pharmacological stress, we have shown that acute sodium arsenite treatment can activate CUTS (Figure 3). In a concurrent study under revision, we show that MG132 similarly causes TDP-43 LOF and CUTS activation (Xie et al., 2025). Notably, none of these induce complete nuclear loss of TDP-43; instead, they show nuclear TDP-43 retention or modest mislocalization. This suggests that TDP-43 LOF may also result from nuclear redistribution and dysfunction under these stress conditions, rather than from complete nuclear loss. We look forward to presenting these ongoing studies in the future.
References
Brown A-L, Wilkins OG, Keuss MJ, Kargbo-Hill SE, Zanovello M, Lee WC, Bampton A, Lee FCY, Masino L, Qi YA, Bryce-Smith S, Gatt A, Hallegger M, Fagegaltier D, Phatnani H, NYGC ALS Consortium, Newcombe J, Gustavsson EK, Seddighi S, Reyes JF, Coon SL, Ramos D, Schiavo G, Fisher EMC, Raj T, Secrier M, Lashley T, Ule J, Buratti E, Humphrey J, Ward ME, Fratta P. 2022. TDP-43 loss and ALS-risk SNPs drive mis-splicing and depletion of UNC13A. Nature 603:131–137. doi:10.1038/s41586-022-04436-3
Cohen TJ, Hwang AW, Restrepo CR, Yuan C-X, Trojanowski JQ, Lee VMY. 2015. An acetylation switch controls TDP-43 function and aggregation propensity. Nat Commun 6:5845. doi:10.1038/ncomms6845
Huang W-P, Ellis BCS, Hodgson RE, Sanchez Avila A, Kumar V, Rayment J, Moll T, Shelkovnikova TA. 2024. Stress-induced TDP-43 nuclear condensation causes splicing loss of function and STMN2 depletion. Cell Rep 43:114421. doi:10.1016/j.celrep.2024.114421
Keating SS, Bademosi AT, San Gil R, Walker AK. 2023. Aggregation-prone TDP-43 sequesters and drives pathological transitions of free nuclear TDP-43. Cell Mol Life Sci 80:95. doi:10.1007/s00018-023-04739-2
Mann JR, Gleixner AM, Mauna JC, Gomes E, DeChellis-Marks MR, Needham PG, Copley KE, Hurtle B, Portz B, Pyles NJ, Guo L, Calder CB, Wills ZP, Pandey UB, Kofler JK, Brodsky JL, Thathiah A, Shorter J, Donnelly CJ. 2019. RNA Binding Antagonizes Neurotoxic Phase Transitions of TDP-43. Neuron 102:321-338.e8. doi:10.1016/j.neuron.2019.01.048
Wilkins OG, Chien MZYJ, Wlaschin JJ, Barattucci S, Harley P, Mattedi F, Mehta PR, Pisliakova M, Ryadnov E, Keuss MJ, Thompson D, Digby H, Knez L, Simkin RL, Diaz JA, Zanovello M, Brown A-L, Darbey A, Karda R, Fisher EMC, Cunningham TJ, Le Pichon CE, Ule J, Fratta P. 2024. Creation of de novo cryptic splicing for ALS and FTD precision medicine. Science 386:61–69. doi:10.1126/science.adk2539
Xie L, Zhu Y, Hurtle BT, Wright M, Robinson JL, Mauna JC, Brown EE, Ngo M, Bergmann CA, Xu J, Merjane J, Gleixner AM, Grigorean G, Liu F, Rossoll W, Lee EB, Kiskinis E, Chikina M, Donnelly CJ. 2025. Contextdependent Interactors Regulate TDP-43 Dysfunction in ALS/FTLD. BioRxiv. doi:10.1101/2025.04.07.646890
Yu H, Lu S, Gasior K, Singh D, Vazquez-Sanchez S, Tapia O, Toprani D, Beccari MS, Yates JR, Da Cruz S, Newby JM, Lafarga M, Gladfelter AS, Villa E, Cleveland DW. 2021. HSP70 chaperones RNA-free TDP-43 into anisotropic intranuclear liquid spherical shells. Science 371. doi:10.1126/science.abb4309.
-
-
-
eLife assessment
Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function resulting in changes in gene expression and splicing of target mRNAs. This study developed a sensitive and robust sensor for TDP-43 activity that should impact the field's ability to monitor whether TDP-43 is functional or not. Though limited to cell culture, the evidence presented is convincing and is the first demonstration that a GFP on/off system can be used to assess TDP-43 mutants as well as loss of soluble TDP-43. The findings are valuable and may represent a novel tool to investigate TDP-43-associated disease mechanisms.
-
Reviewer #1 (Public review):
Summary:
The authors create an elegant sensor for TDP -43 loss of function based on cryptic splicing of CFTR and UNC13A. The usefulness of this sensor primarily lies in its use in eventual high throughput screening and eventual in vivo models. The TDP-43 loss of function sensor was also used to express TDP-43 upon reduction of its levels.Strengths:
The validation is convincing, the sensor was tested in models of TDP-43 loss of function, knockdown and models of TDP-43 mislocalization and aggregation. The sensor is susceptible to a minimal decrease of TDP-43 and can be used at the protein level unlike most of the tests currently employed,Weaknesses:
Although the LOF sensor described in this study may be a primary readout for high-throughput screens, ALS/TDP-43 models typically employ primary readouts such as …Reviewer #1 (Public review):
Summary:
The authors create an elegant sensor for TDP -43 loss of function based on cryptic splicing of CFTR and UNC13A. The usefulness of this sensor primarily lies in its use in eventual high throughput screening and eventual in vivo models. The TDP-43 loss of function sensor was also used to express TDP-43 upon reduction of its levels.Strengths:
The validation is convincing, the sensor was tested in models of TDP-43 loss of function, knockdown and models of TDP-43 mislocalization and aggregation. The sensor is susceptible to a minimal decrease of TDP-43 and can be used at the protein level unlike most of the tests currently employed,Weaknesses:
Although the LOF sensor described in this study may be a primary readout for high-throughput screens, ALS/TDP-43 models typically employ primary readouts such as protein aggregation or mislocalization. The information in the two following points would assist users in making informed choices. 1. Testing the sensor in other cell lines 2. Establishing a correlation between the sensor's readout and the loss of function (LOF) in the physiological genes would be useful given that the LOF sensor is a hybrid structure and doesn't represent any physiological gene. It would be beneficial to determine if a minor decrease (e.g., 2%) in TDP-43 levels is physiologically significant for a subset of exons whose splicing is controlled by TDP-43.Considering that most TDP-LOF pathologically occurs due to aggregation and or mislocalization, and in most cases the endogenous TDP-43 gene is functional but the protein becomes non-functional, the use of the loss of function sensor as a switch to produce TDP-43 and its eventual use as gene therapy would have to contend with the fact that the protein produced may also become nonfunctional. This would eventually be easy to test in one of the aggregation modes that were used to test the sensor.. However, as the authors suggest, this is a very interesting system to deliver other genetic modifiers of TDP-43 proteinopathy in a regulated fashion and timely fashion.
-
Reviewer #2 (Public review):
Summary:
The authors goal is to develop a more accurate system that reports TDP-43 activity as a splicing regulator. Prior to this, most methods employed western blotting or QPCR-based assays to determine whether targets of TDP-43 were up or down-regulated. The problem with that is the sensitivity. This approach uses an ectopic delivered construct containing splicing elements from CFTR and UNC13A (two known splicing targets) fused to a GFP reporter. Not only does it report TDP-43 function well, but it operates at extremely sensitive TDP-43 levels, requiring only picomolar TDP-43 knockdown for detection. This reporter should supersede the use of current TDP-43 activity assays, it's cost-effective, rapid and reliable.Strengths:
In general, the experiments are convincing and well designed. The rigor, number of …Reviewer #2 (Public review):
Summary:
The authors goal is to develop a more accurate system that reports TDP-43 activity as a splicing regulator. Prior to this, most methods employed western blotting or QPCR-based assays to determine whether targets of TDP-43 were up or down-regulated. The problem with that is the sensitivity. This approach uses an ectopic delivered construct containing splicing elements from CFTR and UNC13A (two known splicing targets) fused to a GFP reporter. Not only does it report TDP-43 function well, but it operates at extremely sensitive TDP-43 levels, requiring only picomolar TDP-43 knockdown for detection. This reporter should supersede the use of current TDP-43 activity assays, it's cost-effective, rapid and reliable.Strengths:
In general, the experiments are convincing and well designed. The rigor, number of samples and statistics, and gradient of TDP-43 knockdown were all viewed as strengths. In addition, the use of multiple assays to confirm the splicing changes were viewed as complimentary (ie PCR and GFP-fluorescence) adding additional rigor. The final major strength I'll add is the very clever approach to tether TDP-43 to the loss of function cassette such that when TDP-43 is inactive it would autoregulate and induce wild-type TDP-43. This has many implications for the use of other genes, not just TDP-43, but also other protective factors that may need to be re-established upon TDP-43 loss of function.Weaknesses:
Admittedly, one needs to initially characterize the sensor and the use of cell lines is an obvious advantage, but it begs the question of whether this will work in neurons. Additional future experiments in primary neurons will be needed. The bulk analysis of GFP-positive cells is a bit crude. As mentioned in the manuscript, flow sorting would be an easy and obvious approach to get more accurate homogenous data. This is especially relevant since the GFP signal is quite heterogeneous in the image panels, for example, Figure 1C, meaning the siRNA is not fully penetrant. Therefore, stating that 1% TDP-43 knockdown achieves the desired sensor regulation might be misleading. Flow sorting would provide a much more accurate quantification of how subtle changes in TDP-43 protein levels track with GFP fluorescence.Some panels in the manuscript would benefit from additional clarity to make the data easier to visualize. For example, Figure 2D and 2G could be presented in a more clear manner, possibly split into additional graphs since there are too many outputs. Sup Figure 2A image panels would benefit from being labeled, its difficult to tell what antibodies or fluorophores were used. Same with Figure 4B.
Figure 3 is an important addition to this manuscript and in general is convincing showing that TDP-43 loss of function mutants can alter the sensor. However, there is still wild-type endogenous TDP-43 in these cells, and it's unclear whether the 5FL mutant is acting as a dominant negative to deplete the total TDP-43 pool, which is what the data would suggest. This could have been clarified. Additional treatment with stressors that inactivate TDP-43 could be tested in future studies.
Overall, the authors definitely achieved their goals by developing a very sensitive readout for TDP-43 function. The results are convincing, rigorous, and support their main conclusions. There are some minor weaknesses listed above, chief of which is the use of flow sorting to improve the data analysis. But regardless, this study will have an immediate impact for those who need a rapid, reliable, and sensitive assessment of TDP-43 activity, and it will be particularly impactful once this reporter can be used in isolated primary cells (ie neurons) and in vivo in animal models. Since TDP-43 loss of function is thought to be a dominant pathological mechanism in ALS/FTD and likely many other disorders, having these types of sensors is a major boost to the field and will change our ability to see sub-threshold changes in TDP-43 function that might otherwise not be possible with current approaches.
-
Reviewer #3 (Public review):
The DNA and RNA binding protein TDP-43 has been pathologically implicated in a number of neurodegenerative diseases including ALS, FTD, and AD. Normally residing in the nucleus, in TDP-43 proteinopathies, TDP-43 mislocalizes to the cytoplasm where it is found in cytoplasmic aggregates. It is thought that both loss of nuclear function and cytoplasmic gain of toxic function are contributors to disease pathogenesis in TDP-43 proteinopathies. Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function characterized by changes in gene expression and splicing of target mRNAs. However, to date, most readouts of TDP-43 loss of function events are dependent upon PCR-based assays for single mRNA targets. Thus, reliable and robust assays for detection of global changes in …
Reviewer #3 (Public review):
The DNA and RNA binding protein TDP-43 has been pathologically implicated in a number of neurodegenerative diseases including ALS, FTD, and AD. Normally residing in the nucleus, in TDP-43 proteinopathies, TDP-43 mislocalizes to the cytoplasm where it is found in cytoplasmic aggregates. It is thought that both loss of nuclear function and cytoplasmic gain of toxic function are contributors to disease pathogenesis in TDP-43 proteinopathies. Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function characterized by changes in gene expression and splicing of target mRNAs. However, to date, most readouts of TDP-43 loss of function events are dependent upon PCR-based assays for single mRNA targets. Thus, reliable and robust assays for detection of global changes in TDP-43 splicing events are lacking. In this manuscript, Xie, Merjane, Bergmann and colleagues describe a biosensor that reports on TDP-43 splicing function in real time. Overall, this is a well described unique resource that would be of high interest and utility to a number of researchers. Nonetheless, a couple of points should be addressed by the authors to enhance the overall utility and applicability of this biosensor.
-
Author response:
Public Reviews:
Reviewer #1 (Public review):
Summary:
The authors create an elegant sensor for TDP -43 loss of function based on cryptic splicing of CFTR and UNC13A. The usefulness of this sensor primarily lies in its use in eventual high throughput screening and eventual in vivo models. The TDP-43 loss of function sensor was also used to express TDP-43 upon reduction of its levels.
Strengths:
The validation is convincing, the sensor was tested in models of TDP-43 loss of function, knockdown and models of TDP-43 mislocalization and aggregation. The sensor is susceptible to a minimal decrease of TDP-43 and can be used at the protein level unlike most of the tests currently employed.
Weaknesses:
Although the LOF sensor described in this study may be a primary readout for high-throughput screens, ALS/TDP-43 models …
Author response:
Public Reviews:
Reviewer #1 (Public review):
Summary:
The authors create an elegant sensor for TDP -43 loss of function based on cryptic splicing of CFTR and UNC13A. The usefulness of this sensor primarily lies in its use in eventual high throughput screening and eventual in vivo models. The TDP-43 loss of function sensor was also used to express TDP-43 upon reduction of its levels.
Strengths:
The validation is convincing, the sensor was tested in models of TDP-43 loss of function, knockdown and models of TDP-43 mislocalization and aggregation. The sensor is susceptible to a minimal decrease of TDP-43 and can be used at the protein level unlike most of the tests currently employed.
Weaknesses:
Although the LOF sensor described in this study may be a primary readout for high-throughput screens, ALS/TDP-43 models typically employ primary readouts such as protein aggregation or mislocalization. The information in the two following points would assist users in making informed choices. 1. Testing the sensor in other cell lines 2. Establishing a correlation between the sensor's readout and the loss of function (LOF) in the physiological genes would be useful given that the LOF sensor is a hybrid structure and doesn't represent any physiological gene. It would be beneficial to determine if a minor decrease (e.g., 2%) in TDP-43 levels is physiologically significant for a subset of exons whose splicing is controlled by TDP-43.
Considering that most TDP-LOF pathologically occurs due to aggregation and or mislocalization, and in most cases the endogenous TDP-43 gene is functional but the protein becomes non-functional, the use of the loss of function sensor as a switch to produce TDP-43 and its eventual use as gene therapy would have to contend with the fact that the protein produced may also become nonfunctional. This would eventually be easy to test in one of the aggregation modes that were used to test the sensor.. However, as the authors suggest, this is a very interesting system to deliver other genetic modifiers of TDP-43 proteinopathy in a regulated fashion and timely fashion.
We thank Reviewer #1 for their detailed feedback. In response, we will investigate the function of CUTS in neuronal cells and evaluate how a modest reduction in TDP-43 levels affects the splicing of physiologically relevant TDP-43-regulated cryptic exons within these cells (eg. STMN2, UNC13A, etc…).
Reviewer #2 (Public review):
Summary:
The authors goal is to develop a more accurate system that reports TDP-43 activity as a splicing regulator. Prior to this, most methods employed western blotting or QPCR-based assays to determine whether targets of TDP-43 were up or down-regulated. The problem with that is the sensitivity. This approach uses an ectopic delivered construct containing splicing elements from CFTR and UNC13A (two known splicing targets) fused to a GFP reporter. Not only does it report TDP-43 function well, but it operates at extremely sensitive TDP-43 levels, requiring only picomolar TDP-43 knockdown for detection. This reporter should supersede the use of current TDP-43 activity assays, it's cost-effective, rapid and reliable.
Strengths:
In general, the experiments are convincing and well designed. The rigor, number of samples and statistics, and gradient of TDP-43 knockdown were all viewed as strengths. In addition, the use of multiple assays to confirm the splicing changes were viewed as complimentary (ie PCR and GFP-fluorescence) adding additional rigor. The final major strength I'll add is the very clever approach to tether TDP-43 to the loss of function cassette such that when TDP-43 is inactive it would autoregulate and induce wild-type TDP-43. This has many implications for the use of other genes, not just TDP-43, but also other protective factors that may need to be re-established upon TDP-43 loss of function.
Weaknesses:
Admittedly, one needs to initially characterize the sensor and the use of cell lines is an obvious advantage, but it begs the question of whether this will work in neurons. Additional future experiments in primary neurons will be needed. The bulk analysis of GFP-positive cells is a bit crude. As mentioned in the manuscript, flow sorting would be an easy and obvious approach to get more accurate homogenous data. This is especially relevant since the GFP signal is quite heterogeneous in the image panels, for example, Figure 1C, meaning the siRNA is not fully penetrant. Therefore, stating that 1% TDP-43 knockdown achieves the desired sensor regulation might be misleading. Flow sorting would provide a much more accurate quantification of how subtle changes in TDP-43 protein levels track with GFP fluorescence.
Some panels in the manuscript would benefit from additional clarity to make the data easier to visualize. For example, Figure 2D and 2G could be presented in a more clear manner, possibly split into additional graphs since there are too many outputs. Sup Figure 2A image panels would benefit from being labeled, its difficult to tell what antibodies or fluorophores were used. Same with Figure 4B.
Figure 3 is an important addition to this manuscript and in general is convincing showing that TDP-43 loss of function mutants can alter the sensor. However, there is still wild-type endogenous TDP-43 in these cells, and it's unclear whether the 5FL mutant is acting as a dominant negative to deplete the total TDP-43 pool, which is what the data would suggest. This could have been clarified. Additional treatment with stressors that inactivate TDP-43 could be tested in future studies.
Overall, the authors definitely achieved their goals by developing a very sensitive readout for TDP-43 function. The results are convincing, rigorous, and support their main conclusions. There are some minor weaknesses listed above, chief of which is the use of flow sorting to improve the data analysis. But regardless, this study will have an immediate impact for those who need a rapid, reliable, and sensitive assessment of TDP-43 activity, and it will be particularly impactful once this reporter can be used in isolated primary cells (ie neurons) and in vivo in animal models. Since TDP-43 loss of function is thought to be a dominant pathological mechanism in ALS/FTD and likely many other disorders, having these types of sensors is a major boost to the field and will change our ability to see sub-threshold changes in TDP-43 function that might otherwise not be possible with current approaches.
We thank Reviewer #2 for their constructive evaluation of our study. In response, we will assess CUTS in human neuronal cells, as also recommended by Reviewer #1. Additionally, we will incorporate an analysis of CUTS using flow cytometry to provide quantitative measurements of GFP signal. We agree that investigating how CUTS responds to stressors affecting TDP-43 function would be a valuable addition (eg. MG132), and we will include this data in the revisions to the study.
We also appreciate the feedback on our figures and will work to enhance their clarity, incorporating the Reviewer’s suggestions. Specifically, we will split Figure 2D and 2G into multiple plots and ensure clearer labeling of the image panels in Figures 2A and 4B.
Regarding the comment on the 5FL data, we believe this occurrence can be explained by existing literature, and we will address this directly in the discussion section of the manuscript.
Reviewer #3 (Public review):
The DNA and RNA binding protein TDP-43 has been pathologically implicated in a number of neurodegenerative diseases including ALS, FTD, and AD. Normally residing in the nucleus, in TDP-43 proteinopathies, TDP-43 mislocalizes to the cytoplasm where it is found in cytoplasmic aggregates. It is thought that both loss of nuclear function and cytoplasmic gain of toxic function are contributors to disease pathogenesis in TDP-43 proteinopathies. Recent studies have demonstrated that depletion of nuclear TDP-43 leads to loss of its nuclear function characterized by changes in gene expression and splicing of target mRNAs. However, to date, most readouts of TDP-43 loss of function events are dependent upon PCR-based assays for single mRNA targets. Thus, reliable and robust assays for detection of global changes in TDP-43 splicing events are lacking. In this manuscript, Xie, Merjane, Bergmann and colleagues describe a biosensor that reports on TDP-43 splicing function in real time. Overall, this is a well described unique resource that would be of high interest and utility to a number of researchers. Nonetheless, a couple of points should be addressed by the authors to enhance the overall utility and applicability of this biosensor.
We thank Reviewer #3 for their time and thoughtful assessment of our manuscript. We will address all their recommendations, including expanding the discussion on the CE sequences utilized in the CUTS sensor and exploring the potential utility of the CUTS sensor in alternative disease-relevant systems.
-
-
