Review papers with raw data transparency

Topline critique

This paper reports that cancer persister cells (PC9, A375, BT474 models) are selectively sensitized to ferroptosis because they downregulate the alternative anti‑ferroptotic enzyme FSP1 (AIFM2), depend on mitochondrial oxidative phosphorylation (OXPHOS) as a source of pro‑ferroptotic ROS, and can be further sensitized by clinically available pan‑HDAC inhibitors (panobinostat, vorinostat) or by combined GPX4 plus FSP1 inhibition; data are supported by scRNAseq, biochemical assays, drug synergy matrices and rescue with multiple antioxidants Main preprint describing FSP1 HDAC persister study
Click Run AI Biology Analysis to expand: reanalyse GSE303411, recreate figures, or design follow up experiments

FSP1 and HDACs suppress cancer persister cell ferroptosis Paper Review

What the paper did (concise)

• Derived persister cells from PC9 (EGFR mutant lung), A375 (BRAF V600E melanoma) and BT474 (HER2 amplified breast) with sustained targeted therapy and profiled ferroptosis sensitivity, antioxidant systems, and transcriptomes Methods and datasets overview
• Used scRNAseq and pseudotime to show surviving persister cells reduce OXPHOS gene expression after GPX4 inhibition and that metformin (complex I inhibitor) partially protects persisters from RSL3, linking mitochondrial ROS to ferroptosis sensitivity scRNAseq OXPHOS finding

Key results and evidence strength

Independent mechanistic context and supporting references

The identification of FSP1 as an alternative ferroptosis suppressor that reduces ubiquinone to ubiquinol and functions in parallel to GPX4 is well documented biochemically and biologically; the UniProt functional annotation and mechanistic chemistry match the biological rationale used by the authors to target FSP1 in persisters SciAdv higher profile version of this work

Histone deacetylase inhibitors broadly alter transcription and can increase ROS and lipid peroxidation in cancer cells—a mechanism consistent with the authors observation that panobinostat increases ROS and creates ferroptosis sensitivity in persisters HDAC inhibition background

Strengths

• Multi model approach across lung, melanoma, and breast persister models demonstrating conserved FSP1 decrease increases generality beyond a single cell line Cross model data
• Integration of scRNAseq to identify OXPHOS links and then pharmacologic validation (metformin) is good causal triangulation
• Data and raw scRNAseq are publicly available (GEO GSE303411) enabling reproduction and reanalysis

Weaknesses and blindspots (critical)

1. Almost all data are in vitro. The authors acknowledge limited or absent in vivo experiments; persister biology in the tumor microenvironment (immune, stromal, hypoxia) can substantially change ferroptosis sensitivity and HDAC inhibitor effects. This is a critical translational gap Paper limitation statement
2. HDAC inhibitors are pleiotropic. While ROS increase is documented, the mechanism linking HDAC inhibition to OXPHOS upregulation and ROS is not pinned to specific gene targets—without knockdown/rescue of candidate genes the link remains correlative and open to alternative explanations (stress response, metabolic shift, altered mitochondrial biogenesis).
3. FSP1 downregulation is consistent but not universal—some persister cells showed variable GPX4/DHODH/VKORC1L1 levels. The conditional dependence on FSP1 (only after GPX4 inhibition) suggests complex redundancy; a deeper genetic epistasis series (CRISPRi/a of FSP1, GPX4, DHODH in persister states) would strengthen claims.
4. Drug concentrations, timing, and potential off-target toxicity of combined GPX4 + HDAC + FSP1 inhibition require careful pharmacologic safety studies—GPX4 inhibitors have poor therapeutic windows in vivo historically.

How convincing is the central causal claim?

Moderately convincing at the cellular level: multiple orthogonal experimental approaches (scRNAseq, pharmacology, ROS readouts, antioxidant rescue, synergy matrices) point to an emergent vulnerability in persister cells caused by an antioxidant deficit that includes low FSP1 and an OXPHOS ROS contribution. But proof that these mechanisms are therapeutically actionable in vivo and in patient tumors is not yet provided, and alternative HDAC downstream effects remain plausible confounders Independent lipidomics supporting OXPHOS role

Practical implications and next steps

1. Immediate follow up: perform genetic epistasis (CRISPR KO/knockdown/overexpression) of FSP1 and GPX4 in persister vs parental contexts and test whether FSP1 overexpression rescues GPX4i sensitivity in persisters and conversely whether FSP1 deletion sensitizes parental cells.
2. In vivo validation: orthotopic xenografts or PDX models with minimal residual disease paradigms to test whether GPX4i + HDACi or GPX4i + FSP1i preferentially eliminate persisters without major systemic toxicity.
3. Mechanistic mapping: identify which HDAC isoforms mediate the ROS/OXPHOS upregulation (class I vs class II) using isoform selective inhibitors or genetic knockdown to reduce pleiotropy concerns and move toward safer translational strategies HDAC isoform strategy review

Minimal reproducible analyses you can run now

• Reanalyse scRNAseq GSE303411 to 1) reproduce OXPHOS/ROS gene set enrichment 2) quantify FSP1 expression distributions across clusters and surviving vs depleted clusters, and 3) infer trajectory-specific transcription factors.
• Recreate Bliss synergy matrices from raw CTG data (Supplementary) and perform Loewe and ZIP metrics as orthogonal checks.

Concluding evaluation

The paper provides a coherent, well-supported in vitro mechanistic story that persister cell ferroptosis sensitivity involves lowered FSP1 and a dependence on OXPHOS derived ROS, and it proposes two actionable combinatorial strategies: (1) GPX4 inhibition plus FSP1 inhibition and (2) GPX4 inhibition plus transient HDAC inhibition that raises ROS. Evidence within cell lines is strong; translation to in vivo/clinical relevance requires prioritized follow ups (genetic epistasis, HDAC isoform mapping, PDX MRD models, toxicity profiling).