Heterogeneous, population-level drug-tolerant persisters exhibit ion-channel remodeling and ferroptosis susceptibility

Article activity feed

1. 

   eLife

   Oct 23, 2026

   eLife Assessment

   This valuable study uses single-cell profiling, lineage barcoding, and functional assays to examine drug-tolerant persisters in a BRAF-inhibited melanoma cell line. The data provide solid evidence that persisters arise from many lineages and have altered calcium handling and increased sensitivity to ferroptosis, although the lineage result is difficult to interpret without knowing how many cells survived treatment. The broader claim that these cells represent substates within a single persister state is less well supported, as it rests largely on transcriptomic projections, lacks time-resolved data, and has been examined in only one cell line.

   Read the original source
2. 

   eLife

   Oct 23, 2026

   Reviewer #1 (Public review):

   Summary:

   The study focuses on the characterization of drug‑tolerant persisters (DTPs), a subset of tumor cells that survive therapeutic elimination, in an experimental melanoma model (SKMEL5 cell line) under BRAF‑targeted therapy. Using single‑cell RNA sequencing combined with barcode‑based lineage tracing, the authors interrogate DTP heterogeneity and lineage origins. The major conceptual novelty claimed is a shift from an asserted "consensus view" in which DTPs are treated as distinct, uniform cell states, toward the inference that DTPs comprise a heterogeneous population of sub‑states within a single treatment‑defined regime. They also demonstrate, in this model, a lack of strong lineage bias in the emergence of DTPs. Finally, the authors refine and corroborate prior observations of shifts in …

   Reviewer #1 (Public review):

   Summary:

   The study focuses on the characterization of drug‑tolerant persisters (DTPs), a subset of tumor cells that survive therapeutic elimination, in an experimental melanoma model (SKMEL5 cell line) under BRAF‑targeted therapy. Using single‑cell RNA sequencing combined with barcode‑based lineage tracing, the authors interrogate DTP heterogeneity and lineage origins. The major conceptual novelty claimed is a shift from an asserted "consensus view" in which DTPs are treated as distinct, uniform cell states, toward the inference that DTPs comprise a heterogeneous population of sub‑states within a single treatment‑defined regime. They also demonstrate, in this model, a lack of strong lineage bias in the emergence of DTPs. Finally, the authors refine and corroborate prior observations of shifts in calcium/ion‑channel regulation and increased ferroptosis sensitivity in DTPs.

   Strengths:

   The work offers a conceptually interesting reframing of DTPs, emphasizing population‑level heterogeneity and the existence of sub‑states. Even if the "consensus view" it challenges is somewhat overstated, this perspective is likely to stimulate further discussion and more nuanced DTP conceptualization.

   The combination of single‑cell transcriptomics with barcode‑based lineage tracing provides strong evidence, in the SKMEL5/PLX4720 model, that DTP origin is not limited to rare pre‑existing subpopulations. The observation that most lineages contribute to the idling DTP population, with relatively modest changes in abundance, provides a useful contrast to observations in other experimental systems.

   Integrated bulk RNA‑seq and bulk ATAC‑seq analyses showing concordant signatures across transcriptional and chromatin accessibility data provide useful systems‑level insights into how prolonged BRAF inhibition enriches ion‑transport and calcium‑handling pathways in idling DTP populations.

   Direct calcium‑flux experiments demonstrate reduced store‑operated calcium entry in idling DTPs despite similar ER Ca2+ content in untreated and treated cells. This functional assay provides a convincing, mechanistically interpretable readout that supports the proposed ion‑channel remodeling in the idling state.

   The use of a well‑characterized, widely used BRAFV600E melanoma line (SKMEL5) facilitates cross‑comparison with prior work on BRAF inhibitor responses in melanoma.

   Weaknesses:

   The central "state vs sub‑state" conceptual claim has two major issues. First, the "consensus view" that the authors assert (that DTPs are distinct, uniform cell states) is a straw man. The discreteness of phenotypic states depends on the resolution and interpretive framework (discrete categories versus continuum). I do not think there is a clear consensus on the discreteness of the DTP cell state. Much like the "clone vs subclone" dichotomy, "state vs substate" is typically used operationally rather than in reference to strictly defined biological entities. Therefore, the distinction the authors draw between "states" and "sub‑states" is a subtle reframing of the resolution hierarchy rather than a genuinely new challenge to a deeply held field‑wide belief. The shaky grounds of this distinction are reflected by the internal inconsistencies in the authors' own verbiage: in the introduction, the authors state the inference "idling DTP....are still composed of multiple phenotypic states," which directly contradicts the substate distinction. Second, the experimental evidence is insufficient for a meaningful distinction between states and substates. The distinction primarily relies on low‑dimensional PCA/UMAP representations in Figure 1, where therapy-naive cells form 2 clearly distinct clusters, while BRAFi collapses cells into a single cluster with 2 distinct subclusters. It is unclear if the structure will hold if additional dimensions are considered. Further, as the authors themselves note, the data are essentially snapshots rather than trajectories. Without explicit analysis of higher‑dimensional structure or time‑resolved state transitions, it is unclear that a meaningful distinction can be made.

   The findings are derived entirely from a single experimental model: one BRAFV600E melanoma cell line (SKMEL5) treated with a single research‑grade BRAF inhibitor (PLX4720). However, authors frequently use phrasing that implies broader generalizability to DTPs across cancers and therapies. This can be misleading, especially given convincing experimental evidence for distinct alternative scenarios in other systems. For example, DTPs emerging predominantly from specific rare subpopulations have been well documented in multiple experimental models, including melanoma models and BRAF inhibitors. The authors should more carefully delimit the scope of their conclusions to this particular melanoma/BRAFi model system context and more explicitly acknowledge where other systems have been shown to behave differently.

   Interpretation of the lack of an apparent bias in lineage tracing studies strongly depends on the magnitude of the selection bottleneck before the cells enter the idling DTP state. If a substantial fraction of cells do not make it to the DTP, then the finding poses a challenge to generalizations from other experimental systems. If most cells survive, then the lack of selection at the barcode level is trivial.

   The discussion of DTP heterogeneity and rare cell states would benefit from a more precise and balanced engagement with the existing literature. Some cited papers (including several reviews) are only loosely connected to the specific claims, while several highly relevant primary single‑cell studies on rare‑state variability and drug‑induced reprogramming (for example, work from the Shafer and Raj groups) is not explicitly discussed. This weakens the positioning of the proposed conceptual framework within the broader DTP/systems biology fields.

   Impact:

   In my opinion, the evidence presented in this work, in its current form, is insufficient to support the strong conceptual claims about the cell‑state architecture of DTPs. Nonetheless, the study makes a meaningful contribution by bringing a population‑level, dynamic perspective to understanding DTP biology. Even if one disagrees with the authors' interpretation, the work is valuable in pushing the field beyond the entrenched "gene A is a critical regulator of hallmark B in cancer C" mold and encourages more explicit consideration of phenotypic heterogeneity, trajectories, and treatment‑defined cell‑state landscapes when thinking about overcoming persistence and resistance.

   Read the original source
3. 

   eLife

   Oct 23, 2026

   Reviewer #2 (Public review):

   Summary:

   This work by Hayford, Baleami, Stauffer et al. addresses the standard view of drug-tolerant persister (DTP) cells in cancer. The current model treats each DTP as a uniform, single-cell phenotype that survives therapy. Building on the authors' earlier "idling" DTP concept (Paudel et al. 2018), this study instead provides evidence that DTPs are heterogeneous populations of multiple co-existing sub-states whose individual division/death rates balance out to produce near-zero net population growth.

   Using BRAF-mutant melanoma (SKMEL5) cells treated with the BRAF inhibitor PLX4720, the authors combine single-cell RNA-seq, bulk RNA-seq, ATAC-seq, DNA lineage barcoding (ClonMapper), calcium-flux assays, and ferroptosis drug-response assays to show:

   (1) Idling DTP populations show fewer, but still multiple, …

   Reviewer #2 (Public review):

   Summary:

   This work by Hayford, Baleami, Stauffer et al. addresses the standard view of drug-tolerant persister (DTP) cells in cancer. The current model treats each DTP as a uniform, single-cell phenotype that survives therapy. Building on the authors' earlier "idling" DTP concept (Paudel et al. 2018), this study instead provides evidence that DTPs are heterogeneous populations of multiple co-existing sub-states whose individual division/death rates balance out to produce near-zero net population growth.

   Using BRAF-mutant melanoma (SKMEL5) cells treated with the BRAF inhibitor PLX4720, the authors combine single-cell RNA-seq, bulk RNA-seq, ATAC-seq, DNA lineage barcoding (ClonMapper), calcium-flux assays, and ferroptosis drug-response assays to show:

   (1) Idling DTP populations show fewer, but still multiple, distinct transcriptomic sub-states compared to untreated cells (Figure 1).

   (2) They arise from nearly all pre-existing cell lineages rather than from selective expansion of a resistant clone (Figure 2).

   (3) Transcriptomic and epigenomic data converge on altered ion-channel/calcium-handling genes; functional calcium-flux assays confirm reduced store-operated calcium entry (SOCE) in these cells (Figure 3).

   (4) These cells are around 3-fold more sensitive to the ferroptosis inducer RSL3 (GPX4 inhibitor), an effect reversible by the ferroptosis inhibitor ferrostatin-1 (Figure 4).

   The authors propose targeted landscaping, in which drug treatment reshapes a cell population's phenotypic landscape rather than selecting a single resistant subtype, implying that eradicating DTPs will require sequential interventions that progressively narrow the landscape, rather than targeting a presumed uniform persister state.

   Strengths:

   (1) scRNA-seq, bulk RNA-seq, ATAC-seq, lineage barcoding, calcium-flux assays, and drug-response assays all point toward the same conclusions, strengthening confidence in the findings.

   (2) The ClonMapper approach directly tests (and rules out) clonal selection as the origin of DTPs - a key alternative explanation the authors address rigorously.

   (3) The calcium-flux assay and RSL3/ferrostatin-1 ferroptosis experiments provide direct functional confirmation of transcriptomic/epigenomic predictions, moving beyond correlative sequencing data.

   (4) The "bulk" vs. "single-cell-state" and the "targeted landscaping" analogy is useful for framing future treatment strategies.

   (5) Discussion explicitly flags that the data are not time-resolved single-cell trajectories, and that the ferroptosis vulnerability's exact molecular driver is not identified.

   Weaknesses:

   (1) All experiments use one melanoma cell line (SKMEL5 and its subclones), so generalizability to other BRAF-mutant melanomas or other cancer types/drugs is unproven.

   (2) As the authors acknowledge, the study lacks time-lapse/lineage-resolved tracking of individual cells transitioning between sub-states.

   (3) Much of the ion-channel and metabolic pathway evidence relies on gene set enrichment, which is suggestive but not itself mechanistic; the calcium-flux assay has the main functional role.

   (4) The paper demonstrates increased RSL3 sensitivity and rescue by ferrostatin-1, but doesn't identify the specific molecular driver linking ion-channel remodeling to ferroptosis susceptibility.

   Read the original source
4. 

   Version published to 10.1101/2022.02.03.479045 on bioRxiv

   Feb 5, 2022