Acquired Resistance to the PRMT5 Inhibitor Confers Collateral Sensitivity to MEK Inhibition in MTAP-Null Non-Small Cell Lung Cancer

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This study (biomolecules-4379386) focuses on acquired resistance to the second-generation MTA-cooperative PRMT5 inhibitor MRTX1719 in MTAP-null NSCLC. The topic addresses a key clinical challenge in precision oncology and carries high translational value. The experimental design is logically rigorous and well-structured: from the construction of MTAP-isogenic cell lines to validate drug selectivity, through the establishment of in vitro long-term induced resistance models, to high-throughput drug screening, RNA-seq transcriptomic analysis, and functional validation of candidate vulnerabilities, the study forms a complete and compelling chain of evidence. The quality of the figures is particularly commendable, data are presented clearly, elegantly, and professionally. The dose-response curves, volcano plots, Venn diagrams, heatmaps, and immunoblots are all produced to a high standard, greatly enhancing the readability and persuasiveness of the manuscript. However,still have any issuses as followed:

1. It is observed that all figures in the manuscript have figure titles or captions such as "Figure 1 XXX" and "Figure 2 XXX" embedded directly within the images. At the same time, independent figure captions are already provided outside the figures, for example below or above each image. This redundant labeling with both embedded and external captions is unnecessary and detracts from the clarity of the images.
2. In Figure 3C was performed at a single10 μM concentration. While the data are convincing, they remain somewhat limited in scope. Why was this particular concentration chosen? Is there any prior rationale or preliminary experimental evidence supporting this selection?
3. The core findings are currently based on two cell lines, H838 and H1437, both of which are KRAS/RAF wild-type. Considering that KRAS mutations account for approximately 30% of NSCLC cases and that synergistic effects of MRTX1719 combined with MEK inhibition have been reported in KRAS-mutant models in the literature, the generalizability of the findings is somewhat limited. In addition, the negative result that A549 cells failed to develop significant resistance may itself contain important mechanistic information.Why did the A549 cells fail to develop significant resistance?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Acquired resistance to the PRMT5 inhibitor confers collateral sensitivity to MEK inhibition in MTAP-null non-small cell lung cancer

The authors generate in vitro models of acquired resistance to the MTA-cooperative PRMT5 inhibitor MRTX1719 in two MTAP-null NSCLC lines (H838, H1437), show that resistance is not fully explained by altered PRMT5 activity or MTA levels, and report a collateral sensitivity to MEK inhibition accompanied by enrichment of MAPK-related transcriptional programs. The resistance-model resource and the drug-screen dataset are useful contributions, and the supplementary data reproduce the in-text counts faithfully. My principal concerns are that the headline mechanistic and therapeutic claims outrun the supporting evidence and that several specific results require correction or clarification.

Major Comments

1. The dose-escalation scheme as described is mathematically inconsistent. In "Generation of MRTX1719-resistant Cell Lines", the authors state that the dose was doubled every four weeks to a 6th dose, but strict doubling from the stated starting concentrations cannot yield the stated final concentrations: H838 (0.02 µM to 0.64 µM at the 6th dose, not 1 µM), A549 (0.05 µM to 1.6 µM, not 1 µM), and H1437 (0.1 µM to 3.2 µM, not 2 µM). The Methods and Figure 2A should be reconciled — please state explicitly whether the dose was capped at a round value, rounded, or escalated on a non-strict schedule, and give the actual concentration at each step.
2. The proposed mechanistic model is not reconciled with the authors' own signaling data. The model (Figure 5D) and narrative attribute collateral MEK sensitivity to remodeling of MAPK signaling, yet Figure 4D shows that basal pMEK/pERK are not elevated in resistant cells and that acute MRTX1719-induced MAPK activation is blunted in them. A blunted MAPK response is difficult to reconcile with increased MEK dependence; if anything it predicts the opposite. The authors note the mechanism "remains unclear," which is appropriate, but Figure 5D presents a confident RAS/RAF/MEK/ERK/DUSP feedback scheme that the data do not establish. The model should be reframed as hypothesis, and the apparent paradox between blunted MAPK induction and heightened MEK sensitivity should be addressed directly rather than smoothed over.
3. The MAPK-program claim lacks genetic validation of causality. The link between the enriched MAPK transcriptional program and the MEK-inhibitor vulnerability is entirely correlative. Without a genetic perturbation — e.g., MEK1/2 or ERK1/2 knockdown/knockout, or a rescue experiment — it cannot be concluded that the MAPK program causes the collateral sensitivity, particularly given that the KEGG "MAPK signaling" term is enriched among both up- and down-regulated genes (a directionally uninformative result for a large signaling gene set). At minimum, the causal language should be softened; ideally, one loss-of-function experiment establishing MEK/ERK dependence in the resistant state would substantially strengthen the central claim.
4. The reduced-MTA mechanism is underweighted relative to its evidentiary support. Reduced MTA is mechanistically coherent for an MTA-cooperative inhibitor (less PRMT5·MTA complex leads to reduced drug engagement), and the authors' own MTA-supplementation experiment partially restored sensitivity — a genuine, if incomplete, rescue. This is stronger, more direct evidence than the correlative MAPK transcriptional signature, yet it is framed as "insufficient to explain" resistance largely to foreground the MEK narrative. I recommend a more balanced interpretation in which reduced MTA and altered MAPK signaling are presented as parallel, potentially co-operating mechanisms, and that the origin of reduced intracellular MTA in an MTAP-null background be at least discussed.
5. The combination synergy supporting the "sequential therapy" rationale is confined to the reversible-resistance model. The MRTX1719 + MEK-inhibitor combination is robustly synergistic only in H838, the reversible resistance model (which loses resistance on drug withdrawal). In H1437, the stable and arguably more clinically relevant model, trametinib synergy is non-significant (synergy mean 2.16, p = 0.229), and selumetinib is marginal. The combination is also antagonistic in parental sensitive cells. Building a "sequential MRTX1719 to MEK inhibition" clinical hypothesis primarily on the less durable model overstates the in vitro support; the proposed strategy should be presented with this discordance made explicit, and the discussion should address why the stable model does not show the effect.
6. The collateral sensitivity is not MEK-specific. In the high-throughput screen, mTOR inhibitors (rapamycin overlaps both resistant lines; everolimus and temsirolimus also score in H838) and a cluster of BET inhibitors (five BET-related compounds in H1437) sensitize comparably to the MEK inhibitors, and the resistant cells were themselves described as enlarged and slow-growing. The selective emphasis on MEK over equally well-supported vulnerabilities (notably mTOR, which overlaps both lines and is consistent with the reported PI3K-AKT downregulation) reads as narrative selection rather than a data-driven conclusion. The authors should either justify the MEK focus on grounds beyond the screen overlap or present the broader vulnerability landscape and frame MEK as one of several candidates.

Minor Comments

1. The background gene universe for the Fisher's exact test is unstated. The reported overlap p-values (upregulated 2.02×10⁻⁹; downregulated 8.39×10⁻²²) are not reproducible without knowing the assumed background. Recomputation against the union of detected coding genes yields similar but not identical values (~6.7×10⁻¹⁰ and ~1.5×10⁻²²). Please specify the universe (N) used so the test can be reproduced; the conclusion of significant overlap is unaffected.
2. Supplementary tables contain typographical and annotation gaps. The Table S2 column header reads "Compund" (Table S3 correctly reads "Compound"), and the Table S1 annotations are incomplete (the MC38 and MCA205 rows lack annotation; the B16-F10 row gives only "RNP delivery"). In addition, the KEGG enrichment results shown in Figure 5B are not provided as a supplementary table and should be, to allow independent inspection of the enriched pathways and statistics.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

Targeting protein arginine methyltransferase 5 (PRMT5) drives synthetic lethality in  methylthioadenosine phosphorylase-deleted (MTAP-null) cancers. Different generations of PRMT5 inhibitors, in particular the second-generation MTA-cooperative type which preferentially target MTAP-null cells while largely sparing MTAP-wildtype (MTAP-WT) cells. However, even these second-generation PRMT5 inhibitors are only able to deliver modest objective response rates, suggesting therapeutic resistance. In this manuscript, Xu et al., interrogated the underlying mechanisms of resistance to PRMT5 inhibitor using model drug BMS-986504/MRTX1719 and genetically engineered MTAP-null non-small cell lung cancer (NSCLC) cells. Using two in vitro derived resistant cell models, they found that acquired resistance was mainly driven by collateral sensitivity to MEK inhibition and enrichment of MAPK-related transcriptional programs, though alterations in PRMT5 activity or reduced MTA levels may also be involved. Overall, the experiments are well designed and data are solid. With the following concerns being addressed, the review would highly recommend its acceptance.

1. MTAP-null can happen across multiple cancer types, including glioblastoma, mesothelioma, urothelial carcinoma, pancreatic cancer, melanoma, and non-small cell lung cancer (NSCLC). The authors may need to expand their validation with other cancer types.
2. It is unclear why A549 cells were dropped from the resistance study (and thus, all subsequent studies) after only showing a “modest shift” in sensitivity. Is there a cutoff they used to decide whether a cell line counts as resistant or not?
3. Given that the two models (H838 and H1437) display different resistance phenotypes (reversible vs stable), the generalizability of the MEK sensitivity finding is uncertain. The authors should discuss what drives this phenotypic difference and extends their experiments to more NSCLC cell lines or MTAP-null patient samples.  
4. It would be interesting to know how MRTXR lines respond to EPZ015666, the SAM-cooperative PRMT5 inhibitors.
5. The contradicting results of the combination of MRTX1719 and MEK inhibition across the sensitive cell models (antagonistic) versus the resistant models (synergistic) may limit the translational potential given the heterogeneity of the tumors.
6. After generating the MRTXR lines, MRTX1719 was removed and MRTXR cells were cultured for 6 months. The detailed culture method should be given, like how often the cells are sub-cultured.
7. It is hard to compare the expression of SDMA given the various amount of α-tubulin in Figure 1C. So did β-actin in Fig. S1A. The labelling of the molecular weight of SDMA across all the western blotting is inconsistent, like it is 25kDa in Figure 1C, while 50kDa in Figure S1C.
8. In Figure 3C, the author used 10 uM MTA for the MTA supplementation. It is unclear whether this specific dose was meant to restore MTA back to the original levels, or if it is just an arbitrary high dose.   
9. There are many data not shown. The authors are strongly advised to provide them in the supplementary material.
10. The panels of Figure 5 could be re-organized for better flow and ease of understanding.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have addressed all major concerns satisfactorily. The dose-escalation schedule is now reported accurately, the mechanistic model and causal language have been appropriately softened, the new EPZ015666 cross-resistance data strengthen the MTA argument, the sequential-therapy rationale is properly decoupled from the combination data, and the broader vulnerability landscape is now presented with a defensible justification for the MEK focus. The supplementary tables are corrected and complete; I was able to reproduce the DEG counts, the Fisher p-values, and all screen hits from them.

The following minor points should be corrected before publication:

1. Methods 2.9 states N = 15,399; the correct universe is 15,339 (this is the union of coding genes in Tables S4/S5 and the value consistent with the reported p-values).
2. The Introduction now gives 9p21.3 deletion as ~13%, which cannot be lower than the 13.5% stated for CDKN2A within the same locus.
3. Table S6: the "Log10(p-value)" column holds negative log10(p) values while the figure plots −log10(p); please harmonise, and add q-values and gene counts.
4. Minor language: the Abstract sentence "acquired resistance ... gains increased sensitivity" needs rephrasing; Methods 2.4 "per 10-dish" should read "10-cm dish."

Author Response

Please see the attachment.

Author Response File: Author Response.pdf