Review Reports - Optimizing In Vitro Efficacy Assessment of the Antisense Oligonucleotide Nusinersen in Human Cellular Models

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

General Comment:

In Spinal Muscular Atrophy (SMA), mutations or deletions of the SMN1 gene are directly implicated in the disease. Although humans have two copies of SMN (SMN1 and SMN2), SMN2 expression is severely limited by silencing due to the deletion of exon 7 during pre-mRNA splicing. This results primarily in an abnormal form of the protein (SMNΔ7), leaving only a small proportion of normal SMN, the scarcity of which determines the severity of SMA. In this regard, correcting exon 7 skipping with the antisense oligonucleotide nusinersen represents a significant advance in improving the course of SMA, and therefore, optimizing the conditions for its use is an important objective. This justifies the present study, in which the authors use well-established procedures to evaluate the optimal conditions for nusinersen use, seeking the optimal conditions for efficacy and safety of nusinersen transfer in in vitro cell line models.

Strengths

- The study objective

- The choice of cell lines (HEK293 and GM03813)

- The detailed description of the procedures and treatments

- The analyses to evaluate the effect of nusinersen on SMN transcription and protein

- The identification of the ED50 using dose-response curves

Weaknesses

There are issues that need to be addressed, which we list below in order of appearance in the manuscript, for easier identification of the authors:

Overall, the number of independent experiments included in the study's figures is low. However, the study retains indicative value as a proof of concept regarding its initial proposed objective.
Page 7, Section 2.6: the authors state that “Data had a normal distribution,” but do not specify the population or the Number of cases for which the result was obtained. This aspect needs further clarification.
Page 8, Figure 3 d, e, f: The magnitude of the Y-axis should be standardized across the three graphs, as the current format creates visual confusion when comparing the effect between the different treatments.
Page 8, caption (Supplementary fig 3-5): Nowhere in the text or figure is it indicated what supplementary fig. 5 represents. This must be resolved.
Page 9, Fig. 4: The magnitude of the Y-axis of the figures reflecting the mRNA expression of the same gene (SMN1 or SMN2) must be standardized. The variability of the scales makes it difficult to accurately perceive the changes as a function of the treatment. Furthermore, why is the volume per well (µL/well) not indicated in figures c and d as it is in figures a and b? This must be addressed in the figure or justified in its legend.
Page 9, Legend Fig. 4: Consider including the µL (3 or 1.5) in figures c and d.
Page 10, caption “(Figure 16)”: There must be an error. This figure does not exist.
Page 11, Fig. 5b: Same as fig. 4. Standardize the magnitude of the Y-axis. In order to maintain a realistic understanding of the changes in SMN protein expression as a function of treatment, the axis can be divided into two sections, with 80% representing the maximum protein value and the remaining 20% visualizing the maximum dispersion of the data.
Page 12, Figure 6: The text indicates that the EC50 value of nusinersen in HEK293 cells is 293 nM. Although this value might be consistent with the figure, its characteristics do not allow for a precise determination of this value. Therefore, it is necessary to confirm that the value appearing in the text (293 nM) is correct and not a typographical error.

Author Response

Comments 1: Overall, the number of independent experiments included in the study's figures is low. However, the study retains indicative value as a proof of concept regarding its initial proposed objective.

Response 1: We agree with the reviewer that the number of independent experiments is limited. This reflects the exploratory nature of the study, which was designed primarily as a proof of concept, addressing the initial proposed objective. Despite the limited sample size, the results are consistent across experiments and support the feasibility of the approach.

Comments 2: Page 7, Section 2.6: the authors state that “Data had a normal distribution,” but do not specify the population or the Number of cases for which the result was obtained. This aspect needs further clarification.

Response 2: The normal distribution for gene expression data was calculated using Shapiro-Wilk test and four values were included in analysis. For statistical analysis 1-way Anova with Dunnett´s multiple comparison test was used for data with normal distribution and Kruskal-Wallis with Dunn´s multiple comparison test for data without normal distribution. P value below 0.05 was considered statistically significant. As suggested, we have now included additional information in Methods section (2.6 Statistical analysis).

Statistical analysis was performed using GraphPad Prism version 10.1.2 for Windows, GraphPad Software, USA. Normal distribution of gene expression data (four values) was determined by Shapiro-Wilk test. For statistical analysis 1-way Anova with Dunnett´s multiple comparison test was used for data with normal distribution and Kruskal-Wallis with Dunn´s multiple comparison test for data without normal distribution. P value below 0.05 was considered statistically significant.

Comments 3: Page 8, Figure 3 d, e, f: The magnitude of the Y-axis should be standardized across the three graphs, as the current format creates visual confusion when comparing the effect between the different treatments.

Response 3: We would like to thank the Reviewer for the comment, and we agree with the suggestions. We have normalized data to control, adjusted the y-axis and provided the new Figure 3.

Comments 4: Page 8, caption (Supplementary fig 3-5): Nowhere in the text or figure is it indicated what supplementary fig. 5 represents. This must be resolved.

Response 4: We thank the reviewer for pointing this out. The description of Supplementary Fig. S5 was inadvertently included as a continuation of Supplementary Fig. S4. This has now been corrected, and a separate, clearly labeled description for Supplementary Fig. S5 (as well as for all supplementary figures) has been provided in the revised manuscript for improved clarity.

Comments 5: Page 9, Fig. 4: The magnitude of the Y-axis of the figures reflecting the mRNA expression of the same gene (SMN1 or SMN2) must be standardized. The variability of the scales makes it difficult to accurately perceive the changes as a function of the treatment. Furthermore, why is the volume per well (µL/well) not indicated in figures c and d as it is in figures a and b? This must be addressed in the figure or justified in its legend.

Response 5: We would like to thank the Reviewer for the comment, and we agree with the suggestion. We have normalized data to the control (untreated) and set the y-axis scale for all samples and for both genes. The only exception is Lipofectamine 2000 (SMN1) where the max of y-axis is 15 and we believe that setting all scales to max 15 would make the observation of results more difficult. We have also added the volumes of reagents to both, Figure 4 and legend.

Comments 6: Page 9, Legend Fig. 4: Consider including the µL (3 or 1.5) in figures c and d.

Response 6: We have included the volume of reagents to the Figure 4, c and d.

Comments 7: Page 10, caption “(Figure 16)”: There must be an error. This figure does not exist.

Response 7: We thank the reviewer for noting this error. The reference to “Figure 16” resulted from an earlier version in which all figures were numbered consecutively. This has now been corrected.

Comments 8: Page 11, Fig. 5b: Same as fig. 4. Standardize the magnitude of the Y-axis. In order to maintain a realistic understanding of the changes in SMN protein expression as a function of treatment, the axis can be divided into two sections, with 80% representing the maximum protein value and the remaining 20% visualizing the maximum dispersion of the data.

Response 8: We would like to thank the Reviewer for the suggestion. We have modified the y-axis scale.

Comments 9: Page 12, Figure 6: The text indicates that the EC50 value of nusinersen in HEK293 cells is 293 nM. Although this value might be consistent with the figure, its characteristics do not allow for a precise determination of this value. Therefore, it is necessary to confirm that the value appearing in the text (293 nM) is correct and not a typographical error.

Response 9: We would like to thank the Reviewer for the observation. The analysis and calculations were performed by using GraphPad Prism (V. 10.1.2, GraphPad Software, USA). We have checked the EC50 value and it is correct.

Reviewer 2 Report

Comments and Suggestions for Authors

The presented manuscript “Optimizing In Vitro Efficacy Assessment of the Antisense Oli-gonucleotide Nusinersen in Human Cellular Models” is dedicated to evaluation some in vitro methods of oligonucleotide nusinersen efficacy estimation in HEK293 and GM03813 cell lines.

Major concerns

- The main complaint about this work is that the authors did not point anywhere what specific scientific problem they are solving in this work. Without direct indication of this, it is very difficult to draw a conclusion about the completeness and adequacy of the selected methods described in this work. It can be assumed that the authors are claiming some new important element in the selection of tools for in vitro experiments to identify potential oligonucleotide therapeutic candidates for the treatment of SMA. However, the text lacks a description of the current state of the (supposed) problem being solved.

- What is the novelty of the proposed methods/obtained results? In this work the authors use FDA approved oligonucleotide, previously described methods and well-known cell lines. And as a result, the results obtained are in good agreement with the previously published work of other authors.

- The authors do not explain anywhere why different batches of nusinersen were used.

- Sequence 4 and 5 (Negative oligonucleotide controls) are not scrambled sequence, because they do not match in nucleotide composition of nusinersen, it is just two non-specific sequences with same length as nusinersen. By the way, speaking about modifications, the authors did not mention anywhere that nusinersen contains not only backbone modification (MOE+PS) but modified bases as well (5-methylcytosine). What about Sequence 4 and 5?

- The choice of cell line types is justified only by the fact that they have already been examined in other studies. Why for each cell line only one type of oligonucleotide delivery was applied? What can the authors say about the relevance of examining the biological effect of a drug for correcting motor neuron function on selected cell lines? What is the point of investigating proposed delivery systems for a drug that is administered interthecally, without additional transfection agents?

- Authors should carefully check “Genetic background of SMA”: In Figure 1, SMN1 and SMN2 can not lead to the same pre-mRNA (SMN1 pre-mRNA) and product of SMN2 should lead to 20% of full length SMN1 mRNA and 80% SMN1 delta7 mRNA.

- In many cases oligonucleotides written as nucleotides

- Sections 2.2.1 and 2.2.2 should be 2.3.1 and 2.3.2

In its current form, the work lacks a clear objective, any novelty and should be rejected.

Author Response

Comment 1: The main complaint about this work is that the authors did not point anywhere what specific scientific problem they are solving in this work. Without direct indication of this, it is very difficult to draw a conclusion about the completeness and adequacy of the selected methods described in this work. It can be assumed that the authors are claiming some new important element in the selection of tools for in vitro experiments to identify potential oligonucleotide therapeutic candidates for the treatment of SMA. However, the text lacks a description of the current state of the (supposed) problem being solved.

Response 1: We thank the reviewer for this important comment and agree that the scientific problem addressed by this work should be stated more explicitly.

The specific problem addressed in this study is the lack of standardized and systematically optimized in vitro protocols for evaluating the functional activity of antisense oligonucleotides targeting SMN2, including nusinersen. Although nusinersen is an FDA‑approved therapy with a well‑established mechanism of action, in vitro outcomes are highly sensitive to experimental variables such as transfection strategy, reagent choice, dosing, cell type, and analytical methods. These parameters vary widely across published studies, limiting reproducibility and complicating interpretation and comparison of efficacy metrics.

To address this gap, we systematically evaluated commonly used transfection approaches for nusinersen in two widely applied experimental systems—HEK293 cells and patient‑derived GM03813 fibroblasts—and demonstrate that delivery strategy and reagent selection substantially affect cytotoxicity, SMN2 splicing correction, and SMN protein expression. In addition, we examined antibody performance for SMN detection and identified reagents that provide more reliable protein‑level readouts.

In response to the reviewer’s comment, we have revised the Introduction to explicitly describe the current state of the field and the methodological gap addressed by this work, and to clearly state that the objective of the study is to define robust and reproducible experimental conditions for in vitro assessment of nusinersen activity, rather than to investigate a new molecular mechanism. We have also revised the beginning of the Results section to directly link the experimental design to this stated objective and to justify the selection of cell models and optimization strategies. We have also supplemented the Discussion section with additional explanations. Together, these revisions clarify the scope, rationale, and methodological contribution of the study and allow a clearer assessment of the adequacy and completeness of the selected methods.

Section: 1. Introduction (p. 2, lines 76-96): Despite extensive clinical and preclinical use of nusinersen, there is currently no standardized or broadly accepted in vitro framework for evaluating the efficacy of anti-sense oligonucleotides targeting SMN2 splicing. Published studies frequently employ different cellular models, cell seeding densities, and transfection strategies, often selected empirically and without systematic comparison. This methodological heterogeneity hampers reproducibility and complicates interpretation and comparison of quantitative efficacy parameters, including EC₅₀ values. Because in vitro efficacy data are commonly used for candidate selection and comparative evaluation of novel antisense oligonucleo-tides, the absence of optimized and clearly defined experimental conditions represents a significant methodological gap.

The aim of this study was therefore to systematically optimize and benchmark in vi-tro experimental conditions for assessing nusinersen activity in two widely used human cellular models: HEK293 cells and the SMA patient derived fibroblast line GM03813. HEK293 cells were employed as a robust and reproducible heterologous system for eva-luating the mechanistic effects of nusinersen on SMN2 splicing under well controlled transfection conditions, whereas GM03813 fibroblasts were used as a disease relevant model to assess nusinersen activity in an endogenous SMA genetic context. By defining robust delivery conditions, identifying key experimental determinants of efficacy, and linking these parameters to transcript and protein level readouts, this work proposes a practical framework applicable to the evaluation of both established and emerging SMN2 targeting antisense oligonucleotide therapeutics.

Section 3. Results (p8, lines 259-273): The experimental strategy was designed to address the variability and lack of standardization in published in vitro assays for evaluating antisense oligonucleotide efficacy in spinal muscular atrophy (SMA) models. Because transfection efficiency, cytotoxicity, and productive intracellular delivery are highly cell‑type dependent, key experimental parameters were optimized separately for each cellular model used in this study.

HEK293 cells were employed as a robust and reproducible heterologous system to evaluate the mechanistic effects of nusinersen on SMN2 splicing under tightly controlled transfection conditions. As nucleofection in HEK293 cells is associated with substantial cytotoxicity, optimization in this model focused on cell seeding density to ensure comparable post‑transfection confluency and reproducible functional readouts.

In contrast, GM03813 patient‑derived fibroblasts, which exhibit slower proliferation rates and increased sensitivity to electroporation, were used as a disease‑relevant cellular model. Optimization in this system therefore focused on identifying lipid‑based delivery strategies that balance transfection efficiency and tolerability, while maintaining constant cell seeding density to reduce experimental variability.

Section 4. Discussion, p 15, 449-455

This work provides a systematically optimized and experimentally justified in vitro framework for evaluating SMN2‑targeting antisense oligonucleotides. By systematically defining optimal experimental conditions in two widely used human cellular models, we demonstrate how parameters such as delivery strategy, cell density, and analytical tools influence apparent efficacy and potency. These findings are directly relevant for the comparative assessment and prioritization of novel antisense oligonucleotide therapeutics in early‑stage development.

Comments 2: What is the novelty of the proposed methods/obtained results? In this work the authors use FDA approved oligonucleotide, previously described methods and well-known cell lines. And as a result, the results obtained are in good agreement with the previously published work of other authors.

Response 2: We acknowledge that this study employs an FDA‑approved antisense oligonucleotide, established experimental methodologies, and well‑characterized cell lines. The novelty of this work does not lie in introducing a new therapeutic molecule or uncovering a previously unknown biological mechanism, but in the systematic and comparative evaluation of experimental parameters that critically determine the outcome of in vitro antisense oligonucleotide efficacy studies.

Although nusinersen, HEK293 cells, and GM03813 fibroblasts have been used previously, published studies typically apply these models under diverse and empirically selected transfection conditions, seeding densities, and delivery strategies, often without direct comparison. As a consequence, reported efficacy metrics such as fold induction and EC₅₀ values are difficult to compare across studies, limiting their value for candidate selection and optimization.

In this work, we provide a side‑by‑side, experimentally controlled assessment of delivery methods, transfection reagents, dosing, cytotoxicity, and analytical tools within the same framework. We demonstrate that methodological choices, such as cell density and delivery strategy, have a major quantitative impact on apparent nusinersen potency at both transcript and protein levels. In addition, we identify antibodies that enable more reliable SMN protein detection.

The fact that the optimized conditions reproduce previously reported biological effects is not a limitation, but rather supports the validity and relevance of the proposed framework. By defining optimal and suboptimal experimental conditions and providing practical benchmarking data, this study offers reproducible best‑practice guidance for in vitro evaluation of existing and next‑generation SMN2‑targeting antisense oligonucleotide therapeutics. In response to the reviewer’s comment, we have also added a statement defining the methodological novelty of this work to the Introduction.

Section 1, Introduction, p.3 lines 97-100: Importantly, the novelty of this study does not lie in the identification of new biological effects of nusinersen, but in the systematic dissection of experimental variables that determine the robustness, reproducibility, and quantitative outcome of in vitro antisense oligonucleotide efficacy assays.

Comments 3: The authors do not explain anywhere why different batches of nusinersen were used.

Response 3: We thank the reviewer for this comment. Multiple batches of nusinersen were used to assess potential batch‑to‑batch variability and to ensure the robustness and reproducibility of the observed effects. Evaluation of different manufacturing batches is generally recommended in preclinical studies and is consistent with regulatory expectations (e.g., FDA guidelines) to confirm consistency of performance.

In addition, the use of multiple batches provides further confidence that the observed effects are attributable to nusinersen itself rather than to a batch‑specific artifact, thereby serving as an additional experimental control, confirming that the optimized conditions yield consistent functional readouts.

We agree that this rationale was not sufficiently explained in the original version of the manuscript. In the revised manuscript, we have now explicitly stated the reason for using multiple nusinersen batches in the Materials and Methods section. This explanation improves transparency and clarifies the role of batch comparison in experimental design.

Section Materials and Methods; 2.2. Antisense oligonucleotides (p. 4, line 127): “Two independent batches of nusinersen were used to assess the reproducibility of antisense oligonucleotide activity under the optimized experimental conditions and to exclude potential batch‑specific effects.”

Section Results; 3.2. The effect of nusinersen on SMN1 and SMN2 gene expression in HEK293 cells (p. 9, line 311): As expected, no differences in biological activity were observed between the two batches.

Comments 4. Sequence 4 and 5 (Negative oligonucleotide controls) are not scrambled sequence, because they do not match in nucleotide composition of nusinersen, it is just two non-specific sequences with same length as nusinersen. By the way, speaking about modifications, the authors did not mention anywhere that nusinersen contains not only backbone modification (MOE+PS) but modified bases as well (5-methylcytosine). What about Sequence 4 and 5?

Response 4: We thank the reviewer for this important clarification. We agree that negative control Sequences 4 and 5 are not scrambled versions of nusinersen, as they do not match its nucleotide composition. Rather, they are non‑specific sequences of the same length. To avoid any misunderstanding, we have renamed Sequences 4 and 5 accordingly as negative control oligonucleotides (negative controls) throughout the manuscript.

Regarding chemical modifications, we clarify that Sequences 4 and 5 are fully composed of 2′‑O‑methoxyethyl (2′‑MOE) nucleotides with phosphorothioate backbone. All cytidine residues are present as 5‑methylcytidines, consistent with the modifications used in nusinersen. This information has now been explicitly included in the revised manuscript.

Section 1. Introduction, p. 2. line 72: Nusinersen is fully modified with 2‑O´‑methoxyethyl ribose, which increases resistance to nuclease degradation and improves stability and half‑life, 5‑methylcytidine bases, and a full phosphorothioate backbone linkage that further enhances nuclease resistance and promotes interactions with cellular proteins.

Section 2. Materials and Methods; 2.2 Antisense oligonucleotides, p. 4, line 125. Two nusinersen sources (Table 1) and two negative control oligonucleotides with phosphorotioate backbone, 5‑methylcytidines and methoxy ethyl base modifications (Axolabs) were used (Table 2) in the experiments.

Comments 5. The choice of cell line types is justified only by the fact that they have already been examined in other studies. Why for each cell line only one type of oligonucleotide delivery was applied? What can the authors say about the relevance of examining the biological effect of a drug for correcting motor neuron function on selected cell lines? What is the point of investigating proposed delivery systems for a drug that is administered interthecally, without additional transfection agents?

Response 5: We thank the reviewer for these important points and welcome the opportunity to clarify our rationale.

Choice of cell lines.

The selection of HEK293 and GM03813 cell lines was not based solely on precedent, but on their complementary experimental roles in antisense oligonucleotide research. HEK293 cells provide a robust, highly transfectable human cell model that enables controlled assessment of delivery efficiency, toxicity, and molecular readouts. GM03813 fibroblasts are patient‑derived cells carrying the SMA genotype and therefore offer a disease‑relevant context for evaluating SMN2 splicing correction and SMN expression. These two models are widely used as standardized in vitro screening systems, including in regulatory and patent literature, precisely because they allow reproducible comparison of antisense oligonucleotide activity prior to more complex neuronal models.

Use of one delivery method per cell line.

Each cell line was paired with the delivery method that is known to be most compatible with its biological properties, thereby minimizing confounding variables. HEK293 cells are highly amenable to nucleofection, which allows efficient nuclear delivery and has been extensively validated for this cell type. In contrast, GM03813 fibroblasts exhibit poor survival following electroporation, making lipid‑based delivery the preferred option. Testing all delivery methods across both cell lines would have substantially increased experimental complexity and cytotoxic confounders without improving interpretability. Instead, we focused on optimizing and benchmarking the most relevant delivery strategies within each model, which reflects common practice in antisense oligonucleotide development.

Biological relevance to motor neuron function.

We acknowledge that neither HEK293 cells nor fibroblasts are motor neurons. However, the aim of this study was not to model motor neuron physiology, but to evaluate the molecular pharmacological activity of nusinersen, specifically SMN2 splicing correction and SMN protein expression. These molecular endpoints are cell‑type independent and represent the primary mechanism of action of nusinersen. In vitro non‑neuronal models are routinely used as first‑line systems to optimize delivery, dosing, toxicity, and analytical methods before advancing to more complex neuronal or animal models.

Relevance of delivery systems for an intrathecally administered drug.

Although nusinersen is administered intrathecally in patients without transfection agents, its intracellular uptake in vitro differs substantially from in vivo conditions and relies on endocytic uptake pathways that are often inefficient and variable. Controlled transfection systems are therefore essential in vitro to ensure reliable intracellular delivery, target engagement, and interpretable dose–response relationships. Importantly, the purpose of this study was not to mimic intrathecal delivery, but to establish robust and reproducible in vitro assay conditions for preclinical evaluation. Such assays are critical for early‑stage screening, benchmarking, and comparative assessment of antisense oligonucleotide candidates, which aligns directly with pharmaceutical development workflows.

This work thus addresses a practical and methodological gap by defining experimentally justified in vitro conditions for assessing antisense oligonucleotide activity. The selected cell lines and delivery strategies were chosen as complementary and translationally relevant screening tools rather than as physiological disease models, enabling reliable evaluation of molecular efficacy prior to advancement into more complex neuronal or in vivo systems.

We are grateful to the reviewer for these insightful comments, which have helped improve the clarity and positioning of the study. In response, we have revised the manuscript to more explicitly describe the rationale for the selection of cell line types and delivery methods, and to clarify the scope and intended interpretation of the biological readouts. Additional explanatory text has been added to ensure that the choice of experimental models and delivery strategies is clearly justified and transparent to the reader.

Section 3. Results, p 8. 264-273: HEK293 cells were employed as a robust and reproducible heterologous system to evaluate the mechanistic effects of nusinersen on SMN2 splicing under tightly controlled transfection conditions. As nucleofection in HEK293 cells is associated with substantial cytotoxicity, optimization in this model focused on cell seeding density to ensure comparable post‑transfection confluency and reproducible functional readouts.

Comments 6. Authors should carefully check “Genetic background of SMA”: In Figure 1, SMN1 and SMN2 can not lead to the same pre-mRNA (SMN1 pre-mRNA) and product of SMN2 should lead to 20% of full length SMN1 mRNA and 80% SMN1 delta7 mRNA.

We thank the reviewer for carefully examining Figure 1 and for pointing out this important conceptual inaccuracy. We agree that SMN1 and SMN2 do not give rise to the same pre‑mRNA. In the revised manuscript, Figure 1 has been corrected.

Comments 7. In many cases oligonucleotides written as nucleotides.

Response 7: We thank the reviewer for pointing out this terminology issue. We agree that the term “nucleotides” was incorrectly used in several instances where “oligonucleotides” (or more specifically, “antisense oligonucleotides”) is the appropriate term.

In the revised manuscript, all occurrences referring to nusinersen and control sequences have been carefully reviewed and corrected to use consistent and accurate terminology. This change improves precision and avoids potential confusion between nucleotide monomers and oligonucleotide therapeutics.

page 4, line 145: “1 μL nucleotides (200 μM) were added” corrected to “1 μL oligonucleotides (200 μM) were added”

page 5, line 150: “full medium was added to the cells and the mixture of cells, oligonucleotides and medium”

Section 3.3, p-12, line 362: “no oligonucleotides added or sequence 5”

Comments 8. Sections 2.2.1 and 2.2.2 should be 2.3.1 and 2.3.2

Response 8: We are grateful to the reviewer for this observation. The section numbering has now been corrected, with Sections 2.2.1 and 2.2.2 renumbered as Sections 2.3.1 and 2.3.2.

Reviewer 3 Report

Comments and Suggestions for Authors

Please add titles to supplementary figures in addition to the description in the legend in the text, as many of them look very similar to each other.

For 2.2.2. add a table with the different protocols.

Page 2 at the bottom - correct "a SMA" to "an SMA".

Page 3 - correct "embryotic" to "embryonic"

Page 5 - correct "A260/A260" to "A260/A280".

Page 10 - referral to figure 16 is wrong. There is no such figure.

Author Response

Comments 1. Please add titles to supplementary figures in addition to the description in the legend in the text, as many of them look very similar to each other.

Response 1: We thank the reviewer for this suggestion. Titles have now been added to all supplementary figures, in addition to the existing legends, to improve clarity and facilitate distinction between figures that appear visually similar.

Comments 2. For 2.2.2. add a table with the different protocols.

Response 2: To improve visualization and clarity, tables summarizing the different protocols have been added. As the protocols are already described in detail in the manuscript, the tables were included as supplementary material to avoid redundancy and are now provided as Supplementary Tables S1–S4. In addition, we note that the reference to Section 2.2.2 in the manuscript was made in error and has been corrected to Section 2.3.2.

Comment 3. Page 2 at the bottom - correct "a SMA" to "an SMA".

Response 3: Thank you, this has now been corrected.

Comment 4. Page 3 - correct "embryotic" to "embryonic"

Response 4: Thank you, this has now been corrected. (p. 3, line 107)

Comment 5. Page 5 - correct "A260/A260" to "A260/A280".

Response 5: Thank you, this has now been corrected. (p. 6, line 196)

Comment 6. Page 10 - referral to figure 16 is wrong. There is no such figure.

Response 6: We thank the reviewer for noting this error. The reference to “Figure 16” resulted from an earlier version in which all figures were numbered consecutively. This has now been corrected.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors investigated the effect of various transfection conditions on the effect of nusinersen, an antisense oligonucleotide designed to correct SMN2 pre-mRNA splicing in spinal muscular atrophy (SMA), using cell culture experimental models. Two cell lines were used: HEK293 cells and patient-derived fibroblasts GM. The study addresses an important practical question, as the reliability of data interpretation are critically dependent on transfection conditions. Optimization is essential for the reliable evaluation of antisense drug candidates both in vitro and in the broader context of preclinical development. The findings of this study may contribute to the standardization of cell-based assays used for the assessment of nusinersen and related antisense therapeutics.

I have some comments listed below:

- The authors should clarify how HEK293 cells were rendered relevant as a model system for evaluating nusinersen efficacy. Since standard HEK293 cells carry functional SMN1 alleles and do not recapitulate the pathological SMN2 splicing pattern characteristic of SMA.

- The rationale for studying cell density effects in HEK293 cells and transfection conditions in GM fibroblasts is not sufficiently justified. Is cell seeding density also important for GM fibroblasts? Some explanations on experiment design are briefly mentioned in the Discussion (the middle of paragraph two, and the beginning of paragraphs three and four), however it would be more appropriate to provide this explanation earlier in the text, at the beginning of the Results section, with a more detailed justification of the experimental design choices for each cell line.

- The authors should clarify whether transfection efficiency was monitored across the different conditions tested, as variations in intracellular drug delivery could significantly influence the observed differences in nusinersen activity.

- In section 3.2, the authors refer to the full-length and truncated SMN transcripts as 'SMN1' and 'SMN2' respectively. This terminology is misleading, as both transcripts originate from the SMN2 gene and should be referred to as FL-SMN2 (exon 7-inclusive) and SMN2Δ7 (exon 7-skipped). The authors should address this point earlier in the text, in section 3.1 rather than 3.2.

- In the figures, the Y-axis displays absolute expression values in the thousands and tens of thousands. If this is not of particular significance, normalizing the data to untreated controls (set as 1.0) and presenting results as fold change would improve clarity.

- The error bars in the protein expression data appear quite large, raising concerns about the statistical reliability of the observed differences. The protein expression data are based on only two biological replicates, which is insufficient for reliable statistical analysis. Could the authors comment on this? At least one additional independent experiment may be needed to allow proper statistical evaluation of the reported differences.

- In section 2.2.1, the authors should indicate the final working concentration of nusinersen used per well. Additionally, the authors should provide the concentration used in figure legends for reader convenience.

Author Response

Comments 1. The authors should clarify how HEK293 cells were rendered relevant as a model system for evaluating nusinersen efficacy. Since standard HEK293 cells carry functional SMN1 alleles and do not recapitulate the pathological SMN2 splicing pattern characteristic of SMA.

Response 1. We thank the reviewer for raising this important point. We fully agree that standard HEK293 cells express functional SMN1 and do not intrinsically model the pathological SMN2 splicing defect characteristic of SMA. In this study, HEK293 cells were not intended to serve as a disease model per se, but rather as a controlled, heterologous system to evaluate the molecular activity of nusinersen on SMN2 splicing.

Specifically, HEK293 cells were used because they allow robust, reproducible transfection and clear assessment of splicing modulation under well‑controlled conditions. The relevance of the system derives from the introduction of an SMN2 splicing context, enabling direct evaluation of exon 7 inclusion in response to nusinersen. This approach is commonly used for initial proof‑of‑concept studies of splice‑modifying antisense oligonucleotides.

We have now clarified this rationale in the revised manuscript and explicitly state that the HEK293 model was used to assess mechanistic splicing effects rather than to recapitulate SMA pathology. We also emphasize that these results should be interpreted as indicative and complementary to studies performed in disease‑relevant cellular models.

Please see: Section 3. Results, p8, lines 264-268: “HEK293 cells were employed as a robust and reproducible heterologous system to evaluate the mechanistic effects of nusinersen on SMN2 splicing under tightly controlled transfection conditions. As nucleofection in HEK293 cells is associated with substantial cytotoxicity, optimization in this model focused on cell seeding density to ensure comparable post‑transfection confluency and reproducible functional readouts.”

Section 4. Discussion: p. 16, line 469-472“Accordingly, HEK293 cells were used as a controlled mechanistic model to study delivery‑dependent effects on SMN2 splicing, whereas GM03813 fibroblasts served as a disease‑relevant system to assess nusinersen activity in an endogenous SMA genetic back-ground.

Comments 2. The rationale for studying cell density effects in HEK293 cells and transfection conditions in GM fibroblasts is not sufficiently justified. Is cell seeding density also important for GM fibroblasts? Some explanations on experiment design are briefly mentioned in the Discussion (the middle of paragraph two, and the beginning of paragraphs three and four), however it would be more appropriate to provide this explanation earlier in the text, at the beginning of the Results section, with a more detailed justification of the experimental design choices for each cell line.

Response 2: We thank the reviewer for highlighting the need for a clearer justification of our experimental design.

The optimization strategies differed intentionally between HEK293 and GM03813 cells due to fundamental differences in cell type, growth characteristics, and susceptibility to delivery methods. HEK293 cells are fast‑growing and highly amenable to electroporation‑based methods such as nucleofection, which is efficient but associated with substantial cell loss. Therefore, cell seeding density was systematically evaluated in HEK293 cells to compensate for nucleofection‑related cytotoxicity and to achieve reproducible post‑transfection confluency and readouts.

In contrast, GM03813 patient‑derived fibroblasts grow more slowly and are particularly sensitive to electroporation. For this reason, lipofection‑based approaches were explored instead of nucleofection. Cell seeding density was kept constant in GM03813 cells to minimize experimental variability and because preliminary experiments indicated that transfection reagent choice and formulation, rather than seeding density, was the dominant factor influencing delivery efficiency and cell viability in this model.

In response to the reviewer’s suggestion, we have presented the rationale for these experimental design choices at the beginning of the Results section, where it now explicitly explains why cell density optimization was performed in HEK293 cells but not in GM03813 fibroblasts.

Section 3. Results: p. 8, line 264 – 273: HEK293 cells were employed as a robust and reproducible heterologous system to evaluate the mechanistic effects of nusinersen on SMN2 splicing under tightly controlled transfection conditions. As nucleofection in HEK293 cells is associated with substantial cytotoxicity, optimization in this model focused on cell seeding density to ensure comparable post‑transfection confluency and reproducible functional readouts.

Comments 3. The authors should clarify whether transfection efficiency was monitored across the different conditions tested, as variations in intracellular drug delivery could significantly influence the observed differences in nusinersen activity.

Response 3: As presented in the study, transfection efficiency was not quantified directly using fluorescently labeled oligonucleotides or reporter constructs, but was evaluated indirectly through functional outcomes, including the magnitude and consistency of nusinersen‑induced changes in SMN2 splicing, SMN protein expression, as well as assessments of cell morphology and viability. These readouts reflect productive intracellular delivery, which is more relevant for antisense oligonucleotide activity than total cellular uptake, as a proportion of internalized oligonucleotides may remain trapped in non‑productive pathways.

Importantly, differences observed between conditions are therefore interpreted as condition‑dependent effects on productive delivery and biological response, rather than as intrinsic differences in nusinersen itself. In HEK293 cells, variations in nucleofection efficiency were manifested primarily as differences in cytotoxicity and post‑transfection confluency across seeding densities, with higher seeding density compensating for nucleofection‑associated cell loss and yielding more robust and reproducible nusinersen activity. In GM03813 fibroblasts, marked differences in nusinersen efficacy and cytotoxicity across transfection reagents were observed, enabling identification of delivery conditions that best support productive uptake under the tested experimental setup.

We have now explicitly clarified in the revised manuscript that transfection efficiency was assessed functionally, within the context of the tested experimental conditions, rather than by direct quantitative measurement.

Section 3. Results: p 8, 274-281: Because intracellular delivery efficiency is inherently influenced by experimental conditions, differences observed between tested conditions are interpreted as reflecting condition‑dependent effects on productive oligonucleotide delivery and downstream biological response. In this context, delivery efficiency was evaluated functionally through SMN2 splicing correction, SMN protein expression, and assessments of cell morphology and viability. The Results therefore illustrate how parameters such as cell seeding density and transfection strategy modulate functionally relevant SMN readouts under controlled in vitro conditions.

Section 4. Discussions. P. 16, lines 473-476: A limitation of the present study is the absence of direct quantitative measurement of oligonucleotide uptake. Instead, delivery efficiency was assessed through functional readouts reflecting productive intracellular delivery, which represents the biologically relevant determinant of antisense oligonucleotide activity.

Comments 4. In section 3.2, the authors refer to the full-length and truncated SMN transcripts as 'SMN1' and 'SMN2' respectively. This terminology is misleading, as both transcripts originate from the SMN2 gene and should be referred to as FL-SMN2 (exon 7-inclusive) and SMN2Δ7 (exon 7-skipped). The authors should address this point earlier in the text, in section 3.1 rather than 3.2.

Response 4: We thank the reviewer for pointing out this important issue regarding terminology. In the revised manuscript, following the reviewer’s recommendation, we have introduced and clarified this nomenclature earlier in the text, specifically in Section 3.1, to ensure that the distinction is clear before presenting the experimental results.

Section 3.1 Cell line characterization, p. 8, line 284-287. Because GM03813 cells lack functional SMN1 and rely exclusively on SMN2 for SMN protein production, two SMN2-derived transcripts were quantified throughout this study: the exon 7–inclusive, full-length SMN2 transcript (FL‑SMN2) and the exon 7–skipped transcript (SMN2Δ7).

Comments 5. In the figures, the Y-axis displays absolute expression values in the thousands and tens of thousands. If this is not of particular significance, normalizing the data to untreated controls (set as 1.0) and presenting results as fold change would improve clarity.

Response 5: We would like to thank the Reviewer for the suggestion. We have normalized the data on Figures 3 and 4.

Comments 6. The error bars in the protein expression data appear quite large, raising concerns about the statistical reliability of the observed differences. The protein expression data are based on only two biological replicates, which is insufficient for reliable statistical analysis. Could the authors comment on this? At least one additional independent experiment may be needed to allow proper statistical evaluation of the reported differences.

Response 6: We thank the reviewer for raising this important point. We agree that the protein expression data are based on a limited number of biological replicates and that this restricts the statistical robustness of the analysis. The relatively large error bars reflect this variability and were intentionally retained to transparently represent the experimental dispersion.

In this study, protein expression analysis by Western blot was performed as a complementary readout following the gene expression experiments, with the primary objective of qualitatively confirming the direction of the effects observed at the mRNA level. As Western blotting is inherently semi‑quantitative and strongly dependent on antibody performance, extensive antibody screening was first conducted (including antibodies from four different suppliers) to ensure specificity and reliability of detection.

We acknowledge that Western blotting is not well suited for precise EC50 determination and that two biological replicates are insufficient for rigorous statistical evaluation of small effect sizes. We therefore interpret the protein expression data as supportive and indicative rather than definitive. This has now been clarified in the revised manuscript.

While an additional independent experiment would indeed strengthen the statistical power, the presented protein data are intended to corroborate the primary RNA‑based findings, which were prioritized for quantitative evaluation.

Section Results 3.5: p. 15, lines 426-430: Protein expression analysis by Western blot was performed to independently verify nusinersen‑induced effects observed at the transcript level. Experiments were conducted using two independent biological replicates. Due to the semi‑quantitative nature of Western blotting and its dependence on antibody performance, protein data were interpreted with appropriate caution and were not intended for precise quantitative comparisons.

Section Discussion: p. 17, lines 527-533: Protein‑level analysis was included to verify that nusinersen‑induced transcript changes translate into corresponding alterations in SMN protein expression. Although the number of biological replicates was limited and Western blotting is inherently semi‑quantitative, protein expression patterns were consistent across experiments and aligned with transcript‑level data. These results confirm the biological effect of nusinersen while acknowledging that the experimental design does not permit rigorous statistical evaluation of subtle differences or precise potency estimation at the protein level.

Comments 7. In section 2.2.1, the authors should indicate the final working concentration of nusinersen used per well. Additionally, the authors should provide the concentration used in figure legends for reader convenience.

Response 7: We would like to thank the Reviewer for the comment and agree with the suggestion. We have added the information about the nusinersen concentration to the name of the Figures 3 and 4.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I appreciate the authors' efforts to improve their manuscript, as well as their point-by-point responses. Although I still recognize the same limitation in terms of originality as in the first version of the manuscript, I understand the importance of collecting and revealing more data on the technical aspects of conducting biological activity tests of therapeutic oligonucleotides.

Reviewer 4 Report

Comments and Suggestions for Authors

I have no further comments and consider the manuscript suitable for publication.
A technical note: several references to figures appear as "Error! Reference source not found." The authors are advised to check the cross-references in the manuscript file before final submission.