AP-2 Transcription Factors as Regulators of Ferroptosis: A Family-Wide Profiling in Diverse Cancer Contexts

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this manuscript entitled “AP-2 Transcription Factors as Regulators of Ferroptosis: A Family-Wide Profiling in Diverse Cancer Contexts” by Kołat et al., the authors present a comprehensive profiling of transcription factors involved in the regulation of ferroptosis across a wide range of cancer types. This study is timely and of clear interest, providing valuable insights into potential ferroptosis-related scores derived from multi-cancer cohort analyses. I believe this work has merit and is suitable for publication in principle. However, to support a favourable recommendation, the authors should fully address the comments outlined below:

1) The authors should properly document and cite the current state-of-the-art of ferroptosis research. Some work should be discussed, as it is pivotal for the current understanding of what ferroptosis entails, where it is initiated and how it propagates. The original definition of ferroptosis: PMID: 22632970. FSP1 and phase separation: PMID: 37380771. Recent advances showing that ferroptosis is initiated in lysosomes: PMID: 40335696. Work showing that FSP1 drives metastasis in the lymph node environment: PMID: 41193799. It is strange that the authors have ignored this literature. This work also shows that lysosome biology is crucial for ferroptosis. Related genes should be included in the network of ferroptosis genes (Lamp1, Lamp2, Rab5 etc.).

2) The authors mention ‘non-TF-bound iron’ which is misleading, albeit widely used in the literature. This wording should be avoided. Work has emerged showing that CD44 drives iron uptake in cancer settings (PMID : 32747755  and PMID: 41593071). What happens to CD44 in the analysis by these authors and its effect on ferroptosis? CD44 must feature in Figure 6, considering this crucial literature. While canonical iron-handling genes (e.g., TFRC, FTH1/FTL, SLC40A1) are referenced, the analysis places limited emphasis on broader iron regulatory networks such as ferritinophagy regulators, iron chaperones, or heme metabolism pathways. Given that AP-2 factors are implicated in differentiation and metabolic control, a more systematic evaluation of iron homeostasis genes, and how they integrate with AP-2 activity, would strengthen the ferroptosis framework.

3) The ferroptosis score is used as a central readout throughout the manuscript, yet it reflects transcriptional potential rather than ferroptotic execution. Higher AP-2 expression associates with a higher ferroptosis score (FPS) and worse prognosis, which could indicate compensatory antioxidant programs rather than increased ferroptosis. This conceptual ambiguity should be discussed, and the limitations of FPS as a surrogate for ferroptosis should be emphasized.

4) Analyses are based on bulk RNA-seq data, which obscures cell-type-specific effects. Given that ferroptosis sensitivity and AP-2 activity may differ between tumour cells, stromal cells and immune infiltrates, some associations, especially those involving immune-related genes (e.g., PTPRC, CXCL8), may reflect compositional differences rather than tumour-intrinsic regulation. This limitation should be more thoroughly addressed.

5) Several highlighted genes (e.g., CD36, EPHA2, SNAI2, TP63) are closely linked to stemness and EMT programs, which are known to confer ferroptosis resistance through altered iron handling and lipid metabolism. The manuscript soes not explicitly connect AP-2-driven stem-like states to ferroptosis evasion. Expanding this discussion, especially considering CD44, would better integrate the findings with the current ferroptosis literature.

Author Response

Dear Reviewer, thank you so much for all your comments, please see the attachment for our responses. With kind regards, Damian Kołat

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this manuscript, the authors examined a comprehensive family-wide map that links AP-2 transcription factors to expression programs related to ferroptosis and clinical outcomes. They emphasized cohort-specific AP-2/FPS phenotypes and identified AP-2ε along with AP-2α and AP-2γ. The study's concept is solid, the study's aim is clear and novel, and the overall design is sufficient. The references are up to date. However, the discussion could benefit from slight enhancements. Here are some suggestions for improving the study:

In the introduction:

• Some abbreviations need a full name, such as SLC7A11 and NCOA4...etc, if applicable.

In the methodology:

• Did the authors remove genes with zero or very low expression across samples?
• TCGA contains both primary tumors, metastatic samples, and occasionally "solid tissue normal" samples. Did the authors filter for primary tumors (01A) only?
• FerrDb categorizes genes into drivers, suppressors, and markers. Did the authors include all of them, or focus on a specific subset?
• Why did the authors choose Monocle3 for bulk data?
• Since the authors have 20 different tumor cohorts, how did the authors choose which log₂FC values to display on the Path view map?
• The methodology provides an excellent list of the tools used; however, adding a brief explanation of the underlying logic for key functions—such as how find_gene_modules() relates to the ferroptotic clusters—would bridge the gap between the raw code and the biological interpretation, making the paper more accessible to a broader audience.

In the results:

• The authors report AP-2α/γ predominance and AP-2ε enrichment, but do not clearly state why this matters biologically at this stage.

In the discussion:

• The discussion is valuable, but at times it repeats results instead of analyzing them. It should also be compared to previous studies.
• Limitations section is well written.
• Please unify the writing style of abbreviations. For example, NRF2 in the introduction section becomes Nrf2 in the discussion.

Author Response

Dear Reviewer, thank you so much for all your comments, please see the attachment for our responses. With kind regards, Damian Kołat

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This work contributes a valuable resource for ferroptosis research in cancer. It merits publication in the International Journal of Molecular Sciences after minor revisions.
1. The authors used RNA-seq data from TCGA to infer relationships between AP-2 expression, ferroptosis scores, survival, and gene modules and integrated these results with targetomes. But the authors do not perform direct binding assays or functional validation in this study. The authors indicated that AP-2-centered regulatory networks suggest a directed, mechanistic framework. The overinterpretation should be avoided. 
2. The authors evaluated five key TCGA cohorts: CESC, GBM, OV, PAAD, and THCA. To validate or compare findings from TCGA, external datasets are essential. Could you validate the key findings in independent datasets, such as CPTAC or GEO cohorts, and add a section or figure comparing TCGA results?
3. The authors selected 20 cohorts based on the Ferroptosis Database or the literature. However, some cohorts show minimal links between AP-2 and ferroptosis. Could you use a table or flowchart to show cohort filtering criteria and justify why the focus narrows to five cohorts? 
4. According to the TCGA-CESC (https://portal.gdc.cancer.gov/projects/TCGA-CESC), a total of 307 cases are listed. There are 1098 cases on https://portal.gdc.cancer.gov/projects/TCGA-BRCA. The listed sample counts appear slightly outdated. It is better to address the date the authors extracted TCGA data in the Materials and Methods section. Furthermore, could you expand to include the number of tumor-normal pairs and the number of survival events? For reproducibility in bioinformatics, could you share the full R script, parameters, and pipeline via a GitHub or Zenodo link? 
5. Please use consistent y-axis scaling in Figure 2. It is better to include sample sizes per group. 
6. In Figure 4, “Gene ontology” should be capitalized as “Gene Ontology.”
7. The TCGA cohort abbreviations are defined in Table 2. Could you define the abbreviations on first use? Otherwise, readers are confused when they read these abbreviations for the first time. 

Author Response

Dear Reviewer, thank you so much for all your comments, please see the attachment for our responses. With kind regards, Damian Kołat

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors addressed the concerns iterated int he previous round of reviews and I recommend publishing in its current form.

Reviewer 3 Report

Comments and Suggestions for Authors

Thank you for your responses. The revised manuscript is acceptable for publication in the International Journal of Molecular Sciences.