Investigating Metabolically Altered Pathways in Small Cell Lung Cancer: From RNA Sequencing Analysis to Seahorse-Based Functional Validation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This protocol describes an integrated workflow combining RNA sequencing analysis with Seahorse XF metabolic assays to investigate metabolically altered pathways in chemoresistant small cell lung cancer (SCLC). Using Supinoxin (RX-5902) treatment and DDX5 knockdown in NCI-H69AR cells, transcriptomic analysis revealed significant downregulation of genes involved in mitochondrial respiration, particularly oxidative phosphorylation. These findings were functionally validated by Seahorse XF Cell Mito Stress Tests, which demonstrated impaired basal and maximal oxygen consumption and reduced spare respiratory capacity. Overall, the protocol establishes a robust framework to link transcriptomic changes to mitochondrial dysfunction and provides mechanistic insight into DDX5 inhibition as a therapeutic strategy in SCLC. However, while the topic is important and timely in the field of oncology, several sections of the manuscript require further revision and clarification before it can be considered for publication in Methods and Protocols.

1. What is the key added value of this protocol compared with existing RNA-seq and Seahorse pipelines, and which specific “decision points” make it uniquely suited for studying SCLC, DDX5, and Supinoxin?
2. Please provide a clear schematic of the overall workflow (inputs → intermediate steps → outputs), and explicitly indicate at which stages users may stop if they intend to perform only the RNA-sequencing analysis or proceed to Seahorse-based functional validation.
3. For read trimming and alignment, please specify recommended acceptance thresholds (e.g., minimum percentage of mapped reads, acceptable duplication rates, and minimum read length after trimming), and describe the corrective actions that should be taken if these criteria are not met.
4. The pathway analysis employs MSigDB C2 gene sets with a pre-ranked enrichment strategy. Could the authors clarify the rationale for choosing C2 over alternative collections (e.g., Hallmark or KEGG-only), and comment on the robustness of the results to different ranking metrics?
5. The manuscript states that 70 nM Supinoxin is close to the IC₅₀. Please provide the source of this IC₅₀ value (e.g., citation, figure, or prior dataset), and indicate whether additional dose-response or time-course optimization would be required when applying this protocol to other cell lines or laboratories.

Author Response

Reviewer 1

This protocol describes an integrated workflow combining RNA sequencing analysis with Seahorse XF metabolic assays to investigate metabolically altered pathways in chemo-resistant small cell lung cancer (SCLC). Using Supinoxin (RX-5902) treatment and DDX5 knockdown in NCI-H69AR cells, transcriptomic analysis revealed significant downregulation of genes involved in mitochondrial respiration, particularly oxidative phosphorylation. These findings were functionally validated by Seahorse XF Cell Mito Stress Tests, which demonstrated impaired basal and maximal oxygen consumption and reduced spare respiratory capacity. Overall, the protocol establishes a robust framework to link transcriptomic changes to mitochondrial dysfunction and provides mechanistic insight into DDX5 inhibition as a therapeutic strategy in SCLC. However, while the topic is important and timely in the field of oncology, several sections of the manuscript require further revision and clarification before it can be considered for publication in Methods and Protocols.

Response: Thank you for your comments. We have addressed all the comments and revised the manuscript accordingly.

1. What is the key added value of this protocol compared with existing RNA-seq and Seahorse pipelines, and which specific “decision points” make it uniquely suited for studying SCLC, DDX5, and Supinoxin?

 

Response: We thank the reviewer for this insightful comment.  This protocol integrates two separate methodologies: RNA-sequencing and Seahorse-based functional validation. Protocols that have been published before concentrate solely on either RNA-Seq or Seahorse analysis. Our protocol is distinct as it outlines a detailed procedure to elucidate the mechanism of action of Supinoxin in relation to SCLC.  A key decision point in employing this protocol involves DDX5 knockdown and Supinoxin treatment in NCI-H69AR cells to examine metabolic vulnerabilities associated with SCLC biology.  SCLC cells exhibit a high metabolic demand, and DDX5 activity is essential for sustaining mitochondrial respiration. Initially, we utilized RNA-sequencing to thoroughly examine alterations in gene expression, uncovering that oxidative phosphorylation is impacted in both DDX5KD and Supinoxin-treated chemoresistant NCI-H69AR SCLC cells. To independently validate these findings, we conducted Seahorse-based metabolic analysis, which directly assessed mitochondrial respiration after drug treatment.  The depletion of DDX5 or Supinoxin treatment clearly disrupts mitochondrial function. The combination of RNA-Seq and Seahorse pipelines provides deeper insight into the DDX5-dependent metabolism and potentially provides avenues for novel therapeutic approaches for SCLC. 

1. Please provide a clear schematic of the overall workflow (inputs → intermediate steps → outputs), and explicitly indicate at which stages users may stop if they intend to perform only the RNA-sequencing analysis or proceed to Seahorse-based functional validation.

 

Response: We thank the reviewer for this suggestion.  We have incorporated Figure 8 into the manuscript to provide a clear schematic of the overall workflow (inputs → intermediate steps → outputs), explicitly indicating stages where users may stop after RNA-sequencing or proceed to Seahorse-based functional validation.

 

1. For read trimming and alignment, please specify recommended acceptance thresholds (e.g., minimum percentage of mapped reads, acceptable duplication rates, and minimum read length after trimming), and describe the corrective actions that should be taken if these criteria are not met.

 

Response: We thank the reviewer for this valuable comment and have addressed this issue from Lines 180-182. 

 

1. The pathway analysis employs MSigDB C2 gene sets with a pre-ranked enrichment strategy. Could the authors clarify the rationale for choosing C2 over alternative collections (e.g., Hallmark or KEGG-only), and comment on the robustness of the results to different ranking metrics?

 

Response:  Thank you for this question. We indeed started with a Hallmark-curated gene set for first-pass GSEA analysis and followed with a C2 geneset to dive deeper into pathway mechanisms and canonical signaling. We have added an additional code section for enrichment with Hallmark and acknowledged that a C2 gene set was then employed. The subsequent analysis with “KEGG_OXIDATIVE_PHOSPHORYLATION” from C2 is as is.  Below are the specific changes in the manuscript (Lines 316-330).

“Pathway analysis was performed using significant DE genes using the R-package ClusterProfiler.

1. Pathway analysis can be performed using various databases such as KEGG, Reactome, Gene Ontology (GO) and others. The choice of databases depends on the research question and the specific pathways of interest. For this analysis, we started with Hallmark gene sets from Molecular Signature Database (MSigDB) via R-package msigdbr. The KEGG OXIDATIVE PHOSPHORYLATION (which is the most relevant pathway aligned with current hypothesis) was top enriched pathway in Hallmark gene set.
2. To dive deeper into pathway mechanisms and canonical signaling, we did second-pass analysis with “C2” curated gene set from MSigDB. Enrichment analysis was performed with pre-ranked genes (Rank = signed fold change * -log10pvalue).

 

Note: A different MSigDB database of choice can be updated in the function above to perform enrichment with different database.”

 

1. The manuscript states that 70 nM Supinoxin is close to the IC₅₀. Please provide the source of this IC₅₀ value (e.g., citation, figure, or prior dataset), and indicate whether additional dose-response or time-course optimization would be required when applying this protocol to other cell lines or laboratories.

 

Response: Thank you for the suggestion. This issue has been addressed in lines 505–512.

“The chosen concentration of 70 nM Supinoxin was determined from our previous dose–response experiments conducted in NCI-H69AR cells (1). The cells were treated with varying concentrations of Supinoxin in 96-well plates for 24 hours, and cell viability was evaluated using the CyQUANT Direct Cell Proliferation Assay. The IC₅₀ value determined from three biological replicates was 69.38 ± 8.89 nM, which is close to 70 nM. When implementing this protocol in different cell lines or laboratory settings, it is recommended to conduct further dose-response and optimization experiments, since drug sensitivity may differ among cell types and experimental conditions.”

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript is good; however, some modifications are recommended.

1. At the start of the introduction, add the prevalence statistics regarding small cell lung cancer (SCLC) based on latest GLOBOCAN.
2. the last paragraph of the introduction should clearly state the aim of the manuscript rather than giving a conclusion!
3. It is not familial to write "Note", it is better to cite the tables or the figures in text related to their explained content.
4. a and b start from line 81, the format is missy and not clear, please uniform the format throughout the manuscript. Sometimes it is just a and not b!
5. The authors must clarify which software was used in working directory.
6. I suggest moving work directory in a supplementary file instead being in the main manuscript.
7. Why the authors did not determine mitochondrial parameters such as mitochondrial membrane potential, STAR, MFN1, MFN2, and DNM1L.
8. Did the author performed any gene ontology regarding pathways affected upon Supinoxin treatment?
9. Did the author determine the confidence score for Gene- 275
   concept network plot for “KEGG Oxidative Phosphorylation” pathway upon Supinoxin treatment?. It is recommended to perform more analysis regarding the network and the possible interactions among the detected molecules.
10. There is no need for Fig 6 and 7. It is common and well-known instrument among this field's researchers. 
11. Experimental setup should be separated in a supplementary file.
12. What is the aim of Fig. 10? is that figure is original? ETC is a well-know biological process? why the authors need for such illustration and mentioning its inhibitors in the figure and in the legend? again, What is the aim?
13. In general, i can find redundancy specially in the methodology section and too much figures, which maybe not needed for the aim of the work
14. The result and discussion sections should be in more details.
15. Add a conclusion.

Author Response

Reviewer 2

The manuscript is good; however, some modifications are recommended.

Response: We thank the reviewer for the feedback and have updated the manuscript accordingly.

1. At the start of the introduction, add the prevalence statistics regarding small cell lung cancer (SCLC) based on latest GLOBOCAN.

 

Response:  Thank you for the suggestion. This issue has been addressed in lines 44–53.

“Lung cancer emerged as the predominant cause of cancer morbidity and mortality in 2022, with nearly 2.5 million new cases and over 1.8 million deaths globally (2). It accounted for approximately 12.4% of all diagnosed cancers and 18.7% of cancer-related deaths, as reported by GLOBOCAN 2022. In 2022, it was estimated that there were 1,572,045 new cases of lung cancer globally among males, with 180,063 (11.5%) classified as small-cell carcinoma (small cell lung cancer, SCLC) (3). Among females, there were 908,630 new cases, with 87,902 (9.7%) identified as small-cell carcinoma worldwide (small cell lung cancer, SCLC). Small cell lung cancer (SCLC) is currently considered a recalcitrant malignancy that accounts for approximately 15% of all lung cancer cases, exhibiting a five-year survival rate of under 7% (4,5)."

 

1. the last paragraph of the introduction should clearly state the aim of the manuscript rather than giving a conclusion!

 

Response: Thank you for the suggestion. This issue has been addressed in lines 104-107.

“This protocol integrates transcriptomic and metabolic analyses, providing a framework for exploring metabolic vulnerabilities in cancer cells, with potential for wider applications in cancer metabolism and therapeutic assessment.”

 

1. It is not familial to write "Note", it is better to cite the tables or the figures in text related to their explained content.

 

Response: We have removed the use of “Note” where applicable and incorporated the information into the main text. The revised text now reads: “The chemicals, software, and deposited data used are listed in Table 1.”

 

1. a and b start from line 81, the format is missy and not clear, please uniform the format throughout the manuscript. Sometimes it is just a and not b!

 

Response: Thank you for pointing this out. The inconsistent labeling (a and b) has been removed and uniform notation has been applied throughout the manuscript.

 

1. The authors must clarify which software was used in working directory.

 

Response: Thank you for bringing this missing link to our attention. Prerequisites for the RNAseq analysis (Hardware, Bioinformatics software and R-packages) are included in supplementary information as ‘Supplement 1’. We have clarified this in manuscript.  The specific text (Lines 113-116) added in the manuscript is as below:

“Create a working directory to perform analysis and load the required bioinformatics packages. Detailed prerequisites (Hardware, Bioinformatics software and R-packages) are included in supplementary information (Supplement 1).”

 

1. I suggest moving work directory in a supplementary file instead being in the main manuscript.

 

Response:  We have now removed the “work directory” i.e. relevant code snippets from manuscript to supplementary information as ‘Supplement 1’.

 

1. Why the authors did not determine mitochondrial parameters such as mitochondrial membrane potential, STAR, MFN1, MFN2, and DNM1L.

 

Response:  This study aimed to establish a workflow for identifying pathways affected by Supinoxin treatment and DDX5 knockdown through RNA-seq analysis, followed by functional validation of mitochondrial respiration using Seahorse extracellular flux analysis, which directly measures cellular respiration.  Evaluation of additional mitochondrial parameters, such as mitochondrial membrane potential and regulators of mitochondrial dynamics and morphology (e.g., MFN1, MFN2, and DNM1L), although informative, was beyond the scope of this protocol-focused study. These parameters primarily relate to mitochondrial fusion–fission balance and structural remodeling, whereas the present protocol focuses on integrating RNA-seq analysis with Seahorse-based functional validation to assess mitochondrial bioenergetics through measurements of oxygen consumption rate.

 

1. Did the author performed any gene ontology regarding pathways affected upon Supinoxin treatment?

 

Response:  Yes, we did perform gene ontology analysis. However, the results are excluded from current analysis because we focus on “oxidative phosphorylation pathway” which is the most relevant pathway aligned with current hypothesis. However, in the manuscript text, we have acknowledged the same by including text below (Lines 318-320).

“Pathway analysis can be performed using various databases such as KEGG, Reactome, Gene Ontology (GO) and others. The choice of databases depends on the research question and the specific pathways of interest.”

 

1. Did the author determine the confidence score for Gene- 275
   concept network plot for “KEGG Oxidative Phosphorylation” pathway upon Supinoxin treatment?. It is recommended to perform more analysis regarding the network and the possible interactions among the detected molecules.

 

Response: Yes and we have added a separate section (Lines 357-425) in response. See below:

 

“2.1.11. Confidence Assessment of the KEGG Oxidative Phosphorylation (OXPHOS) Network

• To assess the robustness of gene-concept network within the KEGG Oxidative Phosphorylation (OXPHOS) pathway, we determined if transcriptional enrichment (adjusted pvalue = 3.90E-06, normalized enrichment score = -2.11) correspond to subsequent protein-level organization.

 

• Gene symbols for the OXPHOS pathway were mapped to corresponding STRING IDs. The majority of genes were successfully mapped and interaction querying yielded greater than 2000 pairwise connections among 60 OXPHOS gene symbols indicating extensive inter-protein connectivity and coherent protein interaction network. Visualization of resulting STRING network was performed to demonstrate a densely interconnected structure with highly significant Pvalue.

 

• Each STRING interaction within this network was supported by combined confidence score metric which was computed by combining the probabilities from the different evidence channels (e.g. experimental data, conserved neighborhood, Gene fusions, Phylogenetic co-occurrence, Co-expression, Database imports) [43]. STRING combined confidence score ranges from 0-1000 (some sources may show range from 0-1 which simply divide the scores by 1000). The combined confidence score was interpreted as (0-400 = low-confidence; 400-700 = high-confidence and 900-1000=very high-confidence) [43]. Networks with low-confidence scores generally have more connecting edges, dense networks and may contain indirect evidence with higher chances of false positives while networks with high-confidence scores (>700) have fewer edges with strong experimental support and clean core network with higher biological evidence. Distribution of STRING confidence scores within OXPHOS interaction network was examined. As denoted in Figure S1 >79% interactions within OXPHOS network have very-high-confidence (>900) scores and remaining >20% interactions have high-confidence scores (>700) indicating highly structured and biologically cohesive protein interaction network.

 

 

 

 

• Next, protein–protein interaction (PPI) enrichment analysis was performed. The OXPHOS network contained 1102 observed edges (ppi_edges) compared to 32 expected edges (ppi_lambda). PPI enrichment analysis Pvalue (ppi_Pvalue) was 0, indicating highly significant overrepresentation of interactions than by random chance. To account for machine precision limits (i.e. Pvalue reported as 0), the smallest positive double value was used to compute a conservative −log10-transformed enrichment metric, yielding a robust quantitative estimate of PPI confidence. Further, a composite network confidence score was computed as function of OXPHOS pathway enrichment significance, average of gene-level differential expression strength and PPI enrichment magnitude. High composite network confidence score affirms the significance of OXPHOS pathway network.

 

• Next, topological analysis was performed to quantitatively characterize the structural organization of a protein–protein interaction (PPI) network. Topological analysis revealed the organization of protein interactions and associated key genes (termed as hub genes) that are strong contributors to network integrity. As denoted in respective R-code, first an graph object (g) was built from OXPHOS interactions followed by simplification by removing duplicate edges and self-loops (i.e. removing technically redundant interactions). The resulting graph was comprised of 60 nodes and 1,102 non-redundant edges and corresponding network density of 0.62 (i.e. 62% of all theoretically possible pairwise interactions were retained within simplified graph object). Subsequent calculation of the degree of centrality identified nodes with a high number of connections. Nodes with highest degree of centrality were used to determine top 10 hub genes that form core components within the OXPHOS network. Visualization of core OXPHOS network (i.e. protein-protein interactions associated with top 10 hub genes) was performed.
•  
• “Mapping of the OXPHOS network to STRING interactions indicated extensive inter-protein connectivity and >79% interactions with very high confidence score (>900) denoted biological and functional relevance of protein-protein interactions. Enrichment testing of protein-protein interactions revealed highly significant P value confirming that PPI network connectivity exceeds random expectation. Topological analysis moved beyond simple interaction counts and quantitatively characterized the structural organization of a protein–protein interaction network and revealed the hub genes that contribute most strongly to network integrity. Collectively, these analyses indicated that OXPHOS pathway is not only transcriptionally enriched but also formed a structured, biologically cohesive protein-protein interaction network with hierarchical organization and central hub genes. This reinforced our findings that OXPHOS perturbations are a coordinated pathway level remodeling activity rather than simple gene level variations.”

 

1. There is no need for Fig 6 and 7. It is common and well-known instrument among this field's researchers. 

 

Response: We agree with the reviewer.  We have removed Figures 6, 7, 8 and 10 from the previous submission.  We have added a new figure (Figure 8) indicating the integrated workflow and optional stopping points of the protocol.

 

1. Experimental setup should be separated in a supplementary file.

 

Response: We agree with the reviewer and have moved the experimental setup to the supplementary information (Tables S1–S3).

 

1. What is the aim of Fig. 10? is that figure is original? ETC is a well-know biological process? why the authors need for such illustration and mentioning its inhibitors in the figure and in the legend? again, What is the aim?

 

Response: Figure 10 was generated using BioRender.com to present a schematic representation of the electron transport chain (ETC) and to depict the sites of action of inhibitors used in the Seahorse mitochondrial stress assay. We agree with the reviewer that this figure does not provide additional insights beyond what is already well established. Therefore, to avoid redundancy and improve clarity, the figure has been removed from the revised manuscript.

 

1. In general, i can find redundancy specially in the methodology section and too much figures, which maybe not needed for the aim of the work.

 

Response: We agree with the reviewer.  We have removed Figures 7, 8 and 10 from the previous submission.  We have added a new figure (Figure 8) indicating the integrated workflow and optional stopping points of the protocol.  We have also checked the manuscript thoroughly to avoid redundancy.

 

1. The result and discussion sections should be in more details.

 

Response:  We thank the reviewer for the comment. The Results and Discussion sections have been further elaborated and clarified.

 

1. Add a conclusion.

 

Response:  We agree with the reviewer.  We have addressed a separate “Conclusion” section (Lines 815-832) to address this concern.

 

“The existing protocol integrates two separate methodologies: RNA-sequencing and Seahorse-based functional validation. Protocols that have been published before concentrate solely on either RNA-Seq or Seahorse analysis. Our protocol is innovative as it outlines a detailed procedure to elucidate the mechanism of action of Supinoxin in relation to SCLC.  Initially, we utilized RNA-sequencing to thoroughly examine alterations in gene expression, uncovering that oxidative phosphorylation is impacted in both DDX5KD and Supinoxin-treated NCI-H69AR cells. To validate these findings functionally, we conducted Seahorse-based metabolic analysis, which directly assessed mitochondrial respiration after drug treatment.  SCLC cells demonstrate a high metabolic demand, and DDX5 activity is essential for these cells to sustain mitochondrial respiration. The depletion of DDX5 notably disrupts mitochondrial function, consequently impacting cellular respiration. Studying metabolic activity is essential in SCLC cells following DDX5 knockdown and Supinoxin treatment.  This comprehensive approach demonstrated that Supinoxin disrupts mitochondrial respiration in SCLC cells, corroborating the RNA-Seq findings and uncovering a new mechanism of action.  In conclusion, the combination of RNA-Seq and Seahorse functional validation provides a deeper insight into the DDX5 dependent metabolic activity and subsequent therapeutic approaches targeting metabolic activity in in SCLC.  The current protocol distinctly emphasizes two experimental approaches—RNA-sequencing and Seahorse-based metabolic analysis that converge to a common mechanistic insight.”

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The protocol is well organized and provides a clear, stepwise workflow from RNA‑seq preprocessing to Seahorse‑based metabolic validation. The inclusion of command‑level detail is helpful, but several methodological choices would benefit from additional justification—particularly the STAR alignment parameters, the criteria for QC acceptance, and the strategy used to collapse duplicated TPM entries. Clarifying these points will improve reproducibility for users implementing the workflow in different computational environments. Strengthening the conceptual link between DDX5 inhibition, transcriptomic changes, and mitochondrial dysfunction would also help readers understand the biological rationale behind the protocol.

1. In the Introduction (lines ~17–34), the transition between the general background on SCLC and the mechanistic discussion of DDX5 is abrupt. It would strengthen the narrative if the authors explicitly connected how DDX5‑mediated transcriptional regulation mechanistically intersects with metabolic vulnerabilities in SCLC, thereby justifying why a metabolism‑focused protocol is relevant for studying DDX5 inhibition.

2. The description of Supinoxin’s mechanism of action (lines ~52–75) cites both β‑catenin–related effects and mitochondrial dysfunction. However, the manuscript does not clearly reconcile these two mechanistic models. A brief clarification—perhaps one sentence—explaining whether these mechanisms are independent, sequential, or context‑dependent would help readers understand why mitochondrial assays are prioritized in this protocol.

3. In Section 2.1.1, the instructions for downloading reference files rely on specific Ensembl release numbers (e.g., release 114). It would be helpful to note whether the protocol is version‑dependent or whether any recent Ensembl release is acceptable. This clarification is important because genome annotation versioning can significantly affect downstream gene quantification.

4. The FASTQ quality control section (lines ~94–113) provides detailed commands but does not specify recommended thresholds for adapter content, per‑base quality, or duplication levels that would trigger re‑sequencing or more aggressive trimming. Including explicit QC acceptance criteria would make the protocol more reproducible across laboratories.

5. In Section 2.1.4, the STAR alignment parameters are provided, but the rationale for using --outFilterIntronMotifs RemoveNoncanonical and --twopassMode Basic is not explained. A brief justification for these choices—especially since noncanonical introns may be biologically relevant in cancer—would help readers understand the trade-offs involved.

6. The strandedness determination step (Section 2.1.5) includes example outputs but does not specify how many reads are required for a reliable inference. For low‑depth samples, the infer_experiment.py output can be unstable. Adding a recommended minimum mapped‑read threshold (e.g., >1M uniquely mapped reads) would improve robustness.

7. The quantification section (Section 2.1.6) instructs users to extract only column 1 and 7 from featureCounts output, but the manuscript does not explicitly state that column 7 corresponds to the assigned read counts. Since featureCounts output formats can vary with parameter choices, it would be safer to explicitly state “the final column containing assigned counts” rather than referencing column numbers.

8. The TPM normalization step (lines ~164–179) includes a multi-step R script to collapse duplicated gene IDs. However, the logic for selecting the entry with the highest total counts may bias results toward longer transcripts or multi-mapping regions. A short justification for this choice—or a reference supporting this approach—would help readers assess whether this simplification is appropriate.

9. The correlation analysis (Figure 3) shows high replicate concordance, but the manuscript does not describe how outliers or low‑quality replicates should be handled if correlations fall below ENCODE thresholds. Including guidance on when to exclude a replicate or repeat sequencing would make the protocol more actionable for users encountering problematic datasets.

10. The differential expression section (Section 2.1.7) references a GitHub script but does not provide a DOI, commit hash, or version tag. Because GitHub repositories can change over time, it is important to specify a stable version to ensure reproducibility. Adding a Zenodo DOI or a specific commit ID would resolve this issue.

Author Response

Reviewer 3

The protocol is well organized and provides a clear, stepwise workflow from RNA‑seq preprocessing to Seahorse‑based metabolic validation. The inclusion of command‑level detail is helpful, but several methodological choices would benefit from additional justification—particularly the STAR alignment parameters, the criteria for QC acceptance, and the strategy used to collapse duplicated TPM entries. Clarifying these points will improve reproducibility for users implementing the workflow in different computational environments. Strengthening the conceptual link between DDX5 inhibition, transcriptomic changes, and mitochondrial dysfunction would also help readers understand the biological rationale behind the protocol.

Response: Thank you for your comments. We have addressed all the comments and revised the manuscript accordingly.

1. In the Introduction(lines ~17–34), the transition between the general background on SCLC and the mechanistic discussion of DDX5 is abrupt. It would strengthen the narrative if the authors explicitly connected how DDX5‑mediated transcriptional regulation mechanistically intersects with metabolic vulnerabilities in SCLC, thereby justifying why a metabolism‑focused protocol is relevant for studying DDX5 inhibition.

Response: We thank the reviewer for this valuable comment and have addressed this issue from Lines 71-80 and Lines 104-107.

“Previous studies indicate that the RNA helicase DDX5 is essential for the invasive growth of SCLC. Knockdown of DDX5 leads to a significant disruption in mitochondrial respiration (12). This effect is linked to reduced TCA cycle activity, as evidenced by lower intracellular succinate levels, which typically supply electrons to mitochondrial Complex II via succinate oxidation (12). Furthermore, the knockdown of DDX5 also results in the downregulation of nuclear-encoded mitochondrial genes in SCLC cells, thereby reducing the capacity of these cancer cells to produce the necessary energy for vital cellular functions (12).  Moreover, nuclear DNA-encoded mitochondrial genes are essential for maintaining mitochondrial homeostasis by influencing the expression of mitochondria-related genes in cancer cells (16-19).”

“This protocol integrates transcriptomic and metabolic analyses, providing a framework for exploring metabolic vulnerabilities in cancer cells, with potential for wider applications in cancer metabolism and therapeutic assessment.”

2. The description of Supinoxin’s mechanism of action (lines ~52–75) cites both β‑catenin–related effects and mitochondrial dysfunction. However, the manuscript does not clearly reconcile these two mechanistic models. A brief clarification—perhaps one sentence—explaining whether these mechanisms are independent, sequential, or context‑dependent would help readers understand why mitochondrial assays are prioritized in this protocol.

Response: We agree with the reviewer and have addressed this issue from Lines 88-92.

“However, studies conducted in our lab have demonstrated that Supinoxin does not operate via the β-catenin pathway in either SCLC or MDA-MB-231 cells utilized in earlier studies [1].  Rather, treatment of Supinoxin leads to the inhibition of genes linked to oxidative phosphorylation, resulting in compromised mitochondrial function in H69AR SCLC cell lines [1].”

3. In Section 2.1.1, the instructions for downloading reference files rely on specific Ensembl release numbers (e.g., release 114). It would be helpful to note whether the protocol is version‑dependent or whether any recent Ensembl release is acceptable. This clarification is important because genome annotation versioning can significantly affect downstream gene quantification.

Response: We thank reviewer for bringing these small details to our notice and improving the manuscript. We have now included a note as below (Lines 134-138):

“We have included a specific version (release-114) of human genome in our download link above to ensure reproducibility. The genome assembly and annotations are frequently updated, and it is recommended to use the latest version of assembly and annotations. The underlaying biology and pathways should not change but results for individual genes, counts and significance may have changes pertaining to annotation updates.”

4. The FASTQ quality control section (lines ~94–113) provides detailed commands but does not specify recommended thresholds for adapter content, per‑base quality, or duplication levels that would trigger re‑sequencing or more aggressive trimming. Including explicit QC acceptance criteria would make the protocol more reproducible across laboratories.

Response: As per reviewer’s suggestions, we have now included a table (Table 2) (Lines 180-182) denoting the recommended accepted thresholds for important parameters.

Parameter	Recommended threshold	Rationale and Corrective actions
Median PhredPer-base quality	Q30	Phred score Q30 denotes 0.1% error probability. Including bases below Q30 is not catastrophic but it includes noisy data into analysis and may impact alignment scores.
Minimum read length after trimming	50 bp	Dropping the cutoff includes shorter reads into alignment steps and may have negative impact in terms of ambiguous mapping, inflated multi-mapping reads and false positives in gene expression.
Adapter content after trimming	0-5%	Lower adapter content is better, but small leftover fractions are common. Higher adapter contents have negative implications as adapter does not match genome, cause mismatches at 3’ end and lower the overall mapping quality.
Duplication rates	10-30%	RNA-seq naturally has higher duplication due to highly expressed genes. Duplicates should NOT be removed. However, >30% duplication is concerning and there is a trade-off of accepting the data with caveats or may need to redesign the library preparation.
Mapping rate	80-90%	Typically, >80% mapping rate is acceptable, >85% is good and >90% is excellent. For low mapping rates, primary checks should be verifying the appropriate genome and annotations, verifying data quality before/after trimming and finally checking for library contamination using tools like FastQ-screen.

 

5. In Section 2.1.4, the STAR alignment parameters are provided, but the rationale for using --outFilterIntronMotifs RemoveNoncanonical and --twopassMode Basic is not explained. A brief justification for these choices—especially since noncanonical introns may be biologically relevant in cancer—would help readers understand the trade-offs involved.

Response: As per the reviewer’s suggestions, we have included following note in alignment section (Lines 161-172) about these two important parameters.

“--outFilterIntronMotifs RemoveNoncanonical option in STAR aligner is used to remove non-canonical splice junctions from the output. Non-canonical splice junctions were filtered out to reduce alignment artifacts and improve the robustness of gene-level quantification. Rare canonical introns (GC–AG, AT–AC) were retained. While noncanonical splicing events may be biologically relevant in cancer, they are often associated with splicing errors and can introduce noise in the data. Noncanonical splicing events are typically rare and require junction-level validation. This was beyond the scope of this study and filtering was appropriate for our (gene-level) analysis goals.

--twopassMode Basic option in STAR aligner is used to perform a two-pass alignment. In the first pass, STAR identifies splice junctions from the initial alignment. In the second pass, STAR re-aligns the reads to augmented junction set and in turn improves sensitivity for novel splice junctions and enhance alignment accuracy.”

6. The strandedness determination step (Section 2.1.5) includes example outputs but does not specify how many reads are required for a reliable inference. For low‑depth samples, the infer_experiment.py output can be unstable. Adding a recommended minimum mapped‑read threshold (e.g., >1M uniquely mapped reads) would improve robustness.

Response: Thank you for suggesting this important parameter for low-depth samples. We have updated infer_experiment.py command to reflect parameter “-s 1000000” and added explanation as below (Lines 189-197).

“Note: infer_experiment.py performs random sampling of mapped reads and counts the sense cs. antisense strand reads that overlap with annotated genes. The output provides the fraction of reads assigned to each strand, which helps to determine the strandedness of the data. We used uniquely mapped reads (mapping quality score of 255) for this analysis and remove bias towards repetitive regions and multi-mapped reads.  The specific parameter `-s 1000000` was used to sample 1 million reads for the inference. This is a common practice to sampling noise in low-depth sequencing data and avoids unreliable strandedness estimates.”

7. The quantification section (Section 2.1.6) instructs users to extract only column 1 and 7 from featureCounts output, but the manuscript does not explicitly state that column 7 corresponds to the assigned read counts. Since featureCounts output formats can vary with parameter choices, it would be safer to explicitly state “the final column containing assigned counts” rather than referencing column numbers.

Response: As per reviewer’s suggestion, we have added following in the manuscript text (below) (Lines 215-217) and updated corresponding code snippet to reflect this information as comment.

“From featureCounts’ output column1 and column 7 that correspond to “Gene ID” and “assigned read counts”, respectively were extracted using cut command.”

8. The TPM normalization step (lines ~164–179) includes a multi-step R script to collapse duplicated gene IDs. However, the logic for selecting the entry with the highest total counts may bias results toward longer transcripts or multi-mapping regions. A short justification for this choice—or a reference supporting this approach—would help readers assess whether this simplification is appropriate.

Response: Thank you for the kind suggestion that this approach needs further clarification. We have added following to manuscript text (Lines 230-242):

“For the sake of simplicity, we keep the gene location with maximum assigned counts so that we retain maximum assigned counts for a specific gene to retain the dominant (highly expressed) transcript. This approach prioritizes the dominant expressed locus for the specific gene while reducing the noise from low-expression transcript or ambiguity of annotated gene models. This approach avoids artificial inflation of counts that may occur if counts are summed across multiple loci. This strategy is opted in the context of the goal gene-level differential expression analysis. For other goals such as isoform specific analyses, current strategy may not be appropriate, and it is recommended to keep and review the counts for duplicated genes.”

9. The correlation analysis (Figure 3) shows high replicate concordance, but the manuscript does not describe how outliers or low‑quality replicates should be handled if correlations fall below ENCODE thresholds. Including guidance on when to exclude a replicate or repeat sequencing would make the protocol more actionable for users encountering problematic datasets.

Response: We have added following recommendations in the manuscript (Lines 258-273):

“Recommendations for low correlation among replicates:

Correlation thresholds serve as a screening tool and not an automatic exclusion criterion. Low correlation may represent a true biological variation, and it is important to evaluate various technical aspects. For example, one replicate showing low correlation compared to all others (single outlier) may denote sample-specific technical issues and concerned sample should be assessed in terms of sequencing-depth, quality matrices (base-quality, read-length), mapping rate, duplication rate etc. Technical failures like low-depth sequencing, elevated duplication or modest deviation from quality metrices may be addressed by additional sequencing. On the contrary, technical issues such as significantly low mapping rates (<60%), library size < 2x cohort median, duplication rate > 2x cohort median, inconsistent strandedness indicate severe failures and may need assessment for experimental anomalies (e.g. RNA degradation, library preparation artifacts or contamination) and may flag sample for exclusion. If sequencing was performed in batches, then assessment of batch-effect and batch-correction may be necessary as described elsewhere (35). ENCODE recommendation is three replicates for each biological condition (36).”

10. The differential expression section (Section 2.1.7) references a GitHub script but does not provide a DOI, commit hash, or version tag. Because GitHub repositories can change over time, it is important to specify a stable version to ensure reproducibility. Adding a Zenodo DOI or a specific commit ID would resolve this issue.

Response: As per reviewer’s suggestion a stable version of differential expression analysis script is available on Zenodo with DOI: https://doi.org/10.5281/zenodo.18638583.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Accept in the current form. 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors did extensive improvement for the manuscript. All comments have been covered and addressed.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have addressed all the concerns raised.