Author response:
The following is the authors’ response to the original reviews.
eLife Assessment
This multi-omics study provides a comprehensive characterization of the context-dependent roles of the JAK-STAT pathway (JSP) across different cellular compartments within the breast cancer microenvironment. The authors present convincing evidence that high JSP activity paradoxically drives anti-tumor cytotoxicity in T cells but promotes malignancy and immunosuppression in tumor epithelial cells, leading to the fundamental discovery that broad JAK-STAT inhibition could be therapeutically counterproductive. Ultimately, the identification of the immune-related JSP score and the STAT4 axis as predictive biomarkers for anti-PD-1 immunotherapy response, particularly in triple-negative breast cancer, offers critical insights for precise patient stratification and targeted therapeutic interventions.
We greatly appreciate the editor’s insightful and comprehensive summary of our study.
Public Reviews:
Reviewer #1 (Public review):
Summary:
In their manuscript, Zhou and colleagues present a detailed look at how the JSP functions differently in the various cells of a breast tumor. The authors have effectively shown that the JSP acts as a double-edged sword, as it helps T cells fight cancer but also allows tumor cells to grow and avoid ferroptosis. These findings are important because they identify a useful biomarker to predict how TNBC patients might respond to PD-1 inhibitors.
We highly appreciate Reviewer #1’s generous comments and thorough understanding of our study.
Strengths:
This work is important because it provides a clear explanation for the conflicting roles of the JSP in the tumor environment. The evidence is solid, as it combines data from thousands of patients with single-cell analysis and lab experiments to confirm the role of STAT4 in cancer progression and immunity.
Weaknesses:
However, there are areas for improvement in the scope of the review, the depth of analysis, and the potential for broader clinical implications. The authors are encouraged to address these issues to enhance the scientific and clinical impact of the study.
We greatly appreciate the positive recognition and insightful comments from the reviewer. We are grateful that you acknowledge our solid evidence and the significance of clarifying the dual roles of JSP and STAT4. We will fully address your suggestions to expand the research scope, deepen the analysis, and strengthen the clinical implications in the revised manuscript.
Major Issues:
(1) The authors demonstrate that STAT4 upregulates SLC47A1, but this is currently supported only by expression correlation and western blot data. To confirm a direct link, the authors are encouraged to perform ChIP-qPCR or luciferase reporter assays to show that STAT4 binds directly to the SLC47A1 promoter.
We highly appreciate this insightful and important comment. Due to time constraints, the first author has left the laboratory for clinical practice, and this manuscript is critical for fulfilling his degree requirements at Sichuan University. We are making every effort to supplement additional mechanistic experiments where feasible. In the meantime, we have performed protein–nucleic acid docking analysis between STAT4 protein and the SLC47A1 promoter region, and the corresponding results have been added to the supplementary figures.
(2) The conclusion that the MIF-CD74 axis drives immunosuppression is based on computational inference. To support this, the authors could consider mining publicly available breast cancer spatial transcriptomics data to show the co-localization of MIF and CD74. Alternatively, performing simple dual-color immunofluorescence staining on a few clinical sections would effectively demonstrate the physical proximity of these cells.
We sincerely appreciate your careful review and valuable suggestions. We fully agree that the conclusion regarding the MIF-CD74 axis driving immunosuppression requires further spatial evidence. Although we plan to collect additional clinical specimens for direct co-localization validation, the related ethical approval is still ongoing and cannot be completed in a short time. Therefore, we have supplemented analyses on publicly available breast cancer spatial transcriptomics datasets, which now provide solid bioinformatic evidence to support the spatial co-localization and interaction of the MIF-CD74 axis in the tumor microenvironment in the revised manuscript.
(3) TNBC is highly heterogeneous and includes subtypes like mesenchymal and immunomodulatory groups. The authors should analyze whether the JSP score or STAT4 levels vary significantly between these subtypes, as this could further refine the selection of patients for JAK1 inhibitors.
Thank you for this insightful suggestion. We have supplemented the expression levels of JSP score and STAT4 in two independent TNBC cohorts to explore their heterogeneity across the four TNBC subtypes (Fig. S5B-C).
(4) While the JSP score works well in the current datasets, the authors should consider validating its predictive accuracy in additional independent immunotherapy cohorts, such as the TONIC trial, to ensure the biomarker is robust across different treatment settings.
We sincerely appreciate this valuable suggestion regarding the validation of the JSP score in independent cohorts. To address your concern about the robustness of our biomarker across different treatment settings, we would like to provide the following clarification and updates:
Status of TONIC-trial Data Access:
We fully recognize the significance of validating the JSP score in the TONIC-trial (Nat Med 2019; https://www.nature.com/articles/s41591-019-0432-4), a seminal study exploring immune induction strategies for PD-1 blockade in metastatic TNBC. We have made persistent efforts to obtain these data. However, our previous application to the Data Access Committee (DAC) of the European Genome-phenome Archive (EGA, Study ID: EGAS00001003535) was declined. The official reason provided was a restriction on data sharing imposed by the US Department of Justice, related to Executive Order 14117, which prohibits the transfer of bulk sensitive personal data to certain countries.
Compensatory Validation in Available Anti-PD-1 cohorts:
Despite the limitation on the TONIC-trial data, we have rigorously evaluated the predictive accuracy of the JSP score in two additional, independent, and publicly available anti-PD-1 treated breast cancer cohorts to thoroughly demonstrate its generalizability (Fig. S5A):
GSE194040 (I-SPY2-990, Pembrolizumab, anti-PD-1): A cohort investigating anti-PD-1 therapy in metastatic breast cancer.
GSE173839 (I-SPY2 trial, Durvalumab, anti-PD-L1): A cohort evaluating neoadjuvant anti-PD-L1 therapy in TNBC.
We believe these additional validations adequately address your comment.
Minor Issue:
The manuscript mentions a U-shaped trajectory of JSP activity during tumor transition. A more detailed biological explanation of why the pathway activity initially drops and then rises would add depth to the discussion.
We greatly appreciate this constructive comment. The JAK–STAT pathway (JSP) is essential for maintaining normal epithelial growth; its expression is higher in normal epithelium than in tumor tissues and increases during normal epithelial differentiation. In datasets containing both normal and tumor cell populations, JSP activity naturally declines during the transition from normal epithelium to early tumor lesions. In the subsequent tumor differentiation stage, JSP activity gradually rises, which may be driven by intrinsic tumor heterogeneity and pathway-dependency among different subtypes. This dynamic trend is consistent with JSP pathway activity score, which is independent of pseudotime cell trajectory analysis. We have added this explanation in the first paragraph of the Discussion.
Reviewer #2 (Public review):
Summary:
The JAK-STAT pathway (JSP) exhibits cell-type-specific functional heterogeneity in breast cancer. This study investigates the JSP in breast cancer and its response to anti-PD‑1 immunotherapy. JSP displays distinct cell‑type heterogeneity: it promotes malignant phenotypes and immunosuppression in tumor cells, while enhancing cytotoxicity and reducing exhaustion in T cells. Elevated JSP expression correlates with improved immunotherapy responses, especially in triple‑negative breast cancer. These findings highlight the paradoxical roles of JSP, indicating that broad inhibition may compromise anti‑tumor immunity.
Strengths:
The major strengths of this study include the comprehensive characterization of JSP heterogeneity across epithelial, tumor, and T cells in breast cancer. The identification of JSP and STAT4 as predictive biomarkers for immunotherapy response, particularly in triple-negative breast cancer, provides clinically relevant insights for patient stratification.
Weaknesses:
The findings rely heavily on public dataset analyses.
We sincerely appreciate the reviewer’s insightful recognition and comprehensive summary of our study, as well as the positive comments on our strengths.
We fully agree that the current findings are mainly based on multi‑omics analyses of public datasets. In response to this comment, we have supplemented additional validation using independent cohorts (e.g., FUSCC‑TNBC and METABRIC) to reinforce the reproducibility of the cell‑type-specific heterogeneity of the JAK–STAT pathway and the predictive value of JSP/STAT4 for immunotherapy response in TNBC.
Moreover, we have clearly discussed this limitation in the Discussion section and explicitly proposed further prospective experimental validation and clinical sample verification in our future work.
We have carefully revised the manuscript in full accordance with all of your valuable suggestions to further improve the quality and rigor of our work.
Reviewer #3 (Public review):
Summary:
This multi-omics study by Zhou et al elucidates the context-dependent roles of the Janus kinase-signal transducer and activator of transcription (JAK-STAT) pathway (JSP) across different cellular compartments in the breast cancer tumor microenvironment. While bulk JSP activity is associated with a favorable prognosis, single-cell analysis reveals a paradoxical landscape: high JSP in T cells drives anti-tumor cytotoxicity and reduces exhaustion, whereas high activity in tumor epithelial cells promotes malignancy and immunosuppression via the MIF-CD74 signaling axis. The JSP score (immune-related) serves as a robust predictive biomarker for response to anti-PD-1 immunotherapy, particularly in triple-negative breast cancer (TNBC). Furthermore, the study identifies the STAT4/SLC47A1 axis as a critical mechanism through which tumor cells resist ferroptosis, facilitating disease progression. These findings suggest that broad JAK-STAT inhibition may be counterproductive in cancer therapeutics; instead, therapeutic success depends on precise modulation and carefully timed interventions to preserve its T-cell-associated functions. This study may inspire future studies to explore specific factors that selectively modulate JAK-STAT activity in immune cells to achieve favorable therapeutic outcomes.
Strengths:
Significant therapeutic implications.
Weaknesses:
Limited molecular mechanisms.
We sincerely appreciate the reviewer’s highly positive recognition and insightful summary of our work. Fully addressing your comment regarding limited molecular mechanisms, we have comprehensively supplemented and enriched the mechanistic elaborations in the revised manuscript—including detailed explanations of the dual cell-type-specific roles of the JSP pathway, the downstream MIF-CD74 axis, and the STAT4/SLC47A1-mediated ferroptosis resistance mechanism. All related revisions have been carefully incorporated into the text to strengthen the molecular depth and robustness of our findings.
Recommendations for the authors:
Reviewer #2 (Recommendations for the authors):
(1) The Graphic Abstract in the current version fails to provide brief information about the submission.
We appreciate your comment on the Graphic Abstract. We have redrawn a new, concise Graphic Abstract that clearly summarizes the key findings, workflow, and core message of our submission. The updated version now provides brief but complete information about the study.
(2) Information regarding the epidemiology of breast cancer and TNBC is recommended to be included in the Introduction section.
In response to your comment, we have supplemented up-to-date epidemiological data for both breast cancer and triple-negative breast cancer (TNBC) in the revised Introduction section.
(3) Attention should be paid to the superscript, particularly for CD8+.
We have revised the plus sign in CD4/8+ to the standard superscript format (CD8⁺) throughout the entire manuscript.
(4) Typos are present, such as the error in "2.1" (please verify and correct accordingly).
We have carefully checked and revised the entire manuscript, especially the section 2.1 Bioinformatical profiling. All typos, grammatical errors, and formatting inconsistencies pointed out in your comment have been fully corrected throughout the text.
(5) Relevant information about MCF-10A cells in the cell culture protocol is missing.
We sincerely apologize for the omission of MCF-10A cell culture details. We have supplemented the complete cell culture protocol for MCF-10A cells in 2.2.1 Cell culture.
(6) For the Western blot experiments, information about the dilution ratios (of primary/secondary antibodies) is required.
We have supplemented the detailed dilution ratios for all primary and secondary antibodies used in the Western blot experiments.
(7) The Ethics Approval Number must be provided.
We have supplemented the official ethics approval number for animal experiments in Section 2.2.6.
(8) For the IHC staining experiments, information about the dilution ratios (of antibodies) is required.
We have supplemented the detailed antibody dilution ratios for all primary antibodies used in the IHC staining experiments in Section 2.2.7 Immunohistochemistry (IHC).
(9) Up-to-date citations are necessary, especially those published in 2026.
We have thoroughly updated the reference list according to your suggestion in epidemiology of breast cancer.
(10) Proofreading the language is recommended in order to enhance the fluency and readability of the manuscript.
We have carefully polished the full manuscript with the help of a native English speaker to improve linguistic fluency, readability, and academic expression. All revisions have been completed strictly following your suggestions, and we deeply appreciate your efforts to help optimize this work.
Reviewer #3 (Recommendations for the authors):
Major points for the authors:
(1) Please provide an overview figure of the datasets and approaches used in this study, as Figure 1.
We sincerely appreciate your valuable suggestion. We have supplemented an overview figure (designated as Figure 1A) that systematically summarizes all datasets and experimental approaches used in this study, including the detailed workflow of bioinformatic profiling, pseudotime analysis, and functional validation.
(2) The authors need to improve the organization of figure panels, as they appear cluttered in some regions, which impedes understanding of the figures.
We sincerely appreciate your constructive comment. To address the cluttered figure panels that impeded understanding, we have redrawn Figures 2, 3, 5, and 6, and fine-tuned the image size, layout, and spacing of the panels.
(3) The experimental section utilizes female mice for the MDA-MB-231 xenograft models. Given that a central finding of the paper is the pathway's role in T-cell-mediated anti-tumor immunity, the authors should discuss how the absence of a functional T-cell compartment in nude mice affects the interpretation of tumor growth data, or, ideally, provide data from immunocompetent syngeneic models.
We thank the reviewer’s valuable comment. The MDA-MB-231 xenograft model in nude mice only supports our conclusion that STAT4 promotes tumor growth, given the deficient T-cell immune compartment in this model.
We are currently constructing an orthotopic breast cancer model with stable STAT4 overexpression in 4T1 cells using immunocompetent mice, which possesses a complete immune microenvironment to further validate our immune-related findings. In addition, we plan to establish conditional STAT4 overexpression via the Cre/LoxP system in the MMTV-PyMT transgenic breast cancer mouse model. However, these elaborate in vivo validations cannot be completed within a short time frame due to experimental duration and technical limitations.
This manuscript is critically important for the first author to complete their doctoral degree at Sichuan University. We sincerely appreciate the reviewer’s understanding and generous support for accepting our current data and future follow-up validation plans.
(4) While the study links STAT4 to SLC47A1 upregulation, adding direct mechanistic evidence - such as ChIP-seq or luciferase reporter assays - would confirm that STAT4 directly binds the SLC47A1 promoter rather than acting through intermediary signaling.
We highly appreciate this insightful and important comment. Due to time constraints, the first author has left the laboratory for clinical practice, and this manuscript is critical for fulfilling his degree requirements at Sichuan University. We are making every effort to supplement additional mechanistic experiments where feasible. In the meantime, we have performed protein–nucleic acid docking analysis between STAT4 protein and the SLC47A1 promoter region, and the corresponding results have been added to the supplementary figures.
(5) Are there any potential upstream selective regulators of STAT4 in immune cells?
IL‑12 acts as the upstream activator of STAT4 in immune cells. This cytokine binds to IL12R‑β1/β2, triggering Tyk2/Jak2 signaling to induce STAT4 phosphorylation, dimerization and nuclear translocation, thereby upregulating IFN‑γ transcription and enhancing T cell‑ and NK cell‑mediated antitumor immunity. We have added these details in the Discussion.
(6) Recent studies have identified CD74+ lipid-associated macrophages (LA-MAMs) as a conserved niche in multi-organ metastasis of breast cancer. Linking the tumor-derived MIF-CD74 axis results to this broader metastatic framework could emphasize the clinical relevance of the findings.
Recent study defines a conserved MIF-CD74 LA-MAM axis driving T-cell exhaustion and multi-organ metastasis in breast cancer, predicting poor patient survival. Our work further reveals that tumor-intrinsic JAK-STAT signaling reinforces this immunosuppressive cascade, while T-cell STAT4 activation reverses immune suppression. Combining MIF-CD74 blockade with precise STAT4-targeted strategies may synergize to remodel the metastatic niche and enhance immunotherapy efficacy in TNBC. We have supplemented the relevant mechanistic details and literature discussion in the revised Discussion section.
Minor points for the authors:
(1) The use of "spokesperson" to describe STAT4's role as a representative of the JAK-STAT pathway is somewhat informal for a scientific manuscript. Adopting more standard academic phrasing, such as "primary mediator" or "key transcriptional orchestrator," would enhance the professional tone.
Thank you for your valuable comment. We have revised the manuscript accordingly by replacing the informal term "spokesperson" with the standard academic phrase "key transcriptional orchestrator".
(2) The JSP score achieved a predictive AUC of 0.70-0.76. The authors could improve the work by testing whether combining the JSP score with existing clinical biomarkers, such as PD-L1 IHC or Tumor Mutational Burden (TMB), significantly enhances predictive accuracy.
We have made every effort to collect publicly available breast cancer immunotherapy datasets for further validation. Unfortunately, none of these datasets provided immunohistochemistry (IHC) data for PD-L1/PD-1 expression. To address your valuable suggestion, we instead integrated mRNA expression levels of PD-L1/PD-1 with the JSP score to predict immunotherapy response.
In cohorts GSE194040 and GSE173839 (Fig. S5A), this combined model exhibited improved predictive performance with an AUC exceeding 0.8, which is superior to using the JSP score alone. The corresponding results have been added and presented in the supplementary figures.
(3) There is a potential contradiction in which bulk JSP scores correlate with better survival, whereas tumor-intrinsic JSP scores correlate with poor survival. A clearer discussion or a specific figure reconciling how the dominant immune signal overrides the pro-tumor signal in bulk analysis would be beneficial.
In survival profiling, higher T-cells- and normal epithelial-specific JSP scores correlate with favorable patient survival, whereas elevated tumor-intrinsic JSP scores are associated with poor prognosis. This can be attributed to the predominant expression of JSP in T cells, which enhances T cell mediated anti-tumor immunity and counterbalances its pro tumor effects within cancer cells. We have added detailed clarification of this dual regulatory mechanism in the Discussion section.
(4) The authors cite recent publications regarding the benefits of late-stage or intermittent JAK inhibition. Providing a more detailed proposed dosing schedule or "therapeutic window" based on their differentiation data could offer more actionable insights for clinical trial design.
Based on the above clinical evidence and our findings, administering JAK–STAT inhibitors before or concurrently with immunotherapy may impair T‑cell cytotoxicity and disrupt normal epithelial differentiation in breast cancer patients. Instead, sequential delivery of JAK inhibition following immunotherapy represents a promising immune‑sensitizing strategy, particularly for the TNBC subtype. We have added corresponding descriptions in the third paragraph of the discussion section.
(5) The authors note that they are unable to refine the analysis for TNBC subtypes, such as mesenchymal-like (MES), due to data limitations. If possible, using the METABRIC cohort (which was already accessed) to perform a secondary validation of JSP activity across these specific molecular subtypes would add significant depth.
We appreciate this constructive suggestion. To address the subtype heterogeneity of JSP activity in TNBC, we have collected two TNBC datasets (FUSCC-TNBC and 2024_Nat.Comm.) and conducted further validation and analysis across different TNBC molecular subtypes in Fig. S5B-C.
(6) The discussion evaluates both broad JAK inhibitors (Ruxolitinib) and STAT3-selective inhibitors (TTI-101). Explicitly comparing the potential biological impact of selective STAT3 inhibition versus selective STAT4 activation could clarify the most promising therapeutic direction.
We greatly appreciate this valuable suggestion. We have supplemented the Discussion (in the penultimate paragraph) by proposing a translational strategy utilizing the specific cytokine IL‑12 to activate STAT4 for immune sensitization, while explicitly comparing the distinct biological effects and therapeutic directions between selective STAT3 inhibition and targeted STAT4 activation.
In summary, we sincerely thank the editors and reviewers for their constructive comments and valuable suggestions. We have carefully addressed all the comments and revised the manuscript accordingly.
