Unveiling the Brain-Penetrating Material Basis of Dragon’s Blood: Identification of Active Metabolites and Metabolic Pathways for Ischemic Stroke Therapy

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript <Unveiling the Brain-Penetrating Material Basis of Dragon’s Blood: Identification of Active Metabolites and Metabolic Pathways for Ischemic Stroke Therapy> presents a comprehensive metabolomics-based investigation of Dragon’s Blood and its synergistic interaction with Borneol in the context of ischemic stroke. The topic is timely and relevant, particularly considering the increasing interest in natural products and multi-target therapies for central nervous system diseases. The integration of UHPLC-Q-TOF-MS/MS, metabolomics, and computational approaches is a strong aspect of the study. Overall, the manuscript is informative and supported by a substantial experimental framework. Nevertheless, several aspects could be clarified or further strengthened to improve the manuscript’s scientific rigor, interpretative balance, and translational relevance.

Revisions

1. The use of n = 5 and n = 4 (for CSF in the combination group) is limited for multivariate analysis, especially for OPLS-DA models, which are prone to overfitting. For example, in the sections: Materials and Methods – Section 2.3 where the sample size is defined but not justified; Results – Section 3.2, where OPLS-DA results are presented without sufficient validation (e.g., permutation testing); Results – Figure 9 and associated description, where model performance (R², Q²) is reported but without additional validation metrics; Discussion – Limitations (Section 4.5), where this important limitation is only briefly mentioned and could be expanded.

It would be helpful if the authors could provide a short justification for the chosen sample size and include additional validation steps for the OPLS-DA models (for example, permutation testing). Reporting complementary statistical parameters such as confidence intervals or effect sizes would also increase confidence in the results.

2. Some of the proposed transformations (for example, the cyclization of berberrubine to coptisine, or the broader concept of <deep metabolic modulation> by Borneol) seem to be inferred from MS data without direct experimental confirmation. This is particularly noticeable in Results Section 3.3, as well as in the Discussion (Section 4.2) and the Conclusions, where these processes are sometimes described in fairly definitive terms.

I recommend clearly distinguishing between experimentally confirmed findings and proposed or hypothetical pathways. The language should be moderated accordingly (e.g., <suggested>, <proposed>), and, if possible, a schematic summarizing metabolic pathways with confidence levels could be added.

3. The presence of compounds in cerebrospinal fluid is interpreted as evidence of blood–brain barrier penetration; however, it is not entirely clear whether this alone is sufficient to support such a conclusion. This point comes up in Results Section 3.3 (especially 3.3.2 and 3.3.3) and is further developed in Discussion Sections 4.1 and 4.2, as well as in the Conclusions.

It would strengthen the manuscript if the authors could briefly acknowledge the limitations of using CSF as a proxy for BBB permeability and consider alternative explanations. A short comparison with literature data on known BBB-penetrant compounds or similar metabolomics approaches would also be useful here.

4. The molecular docking results are interesting and add a useful dimension to the study; however, in a few places they seem to be interpreted somewhat strongly. For instance, in Results Section 3.5, expressions such as <exceptional spatial fit> or statements implying direct modulation of target proteins may give the impression of experimental validation rather than computational prediction. A similar tendency appears in Discussion Section 4.4 and in the Conclusions.

I recommend softening the language slightly to emphasize that these results are predictive and hypothesis-generating. Briefly mentioning some of the known limitations of docking approaches (e.g., lack of dynamic biological context, potential false positives) would also help to keep the interpretation well balanced.

5. The manuscript proposes that Borneol may modulate metabolic enzymes and possibly influence gut microbiota, which is an interesting idea. However, these mechanisms are not directly investigated in the present study. This is particularly evident in Results Section 3.3.3, Discussion Section 4.3, and the Conclusions, where these effects are sometimes described with a relatively high degree of certainty.

It may be more appropriate to frame these points as hypotheses supported by indirect evidence. Clarifying that enzyme modulation and microbiota involvement were not directly assessed, and suggesting them as directions for future research, would improve the overall balance and credibility of the manuscript.

 

Overall, the manuscript addresses an important and interesting topic and presents a solid experimental approach, and addressing these major revisions would significantly improve the scientific quality and impact of the work.

Comments for author File: Comments.pdf

Author Response

Dear Reviewer 1,

 

We would like to express our sincere gratitude for your thorough and constructive evaluation of our manuscript. Your insightful comments regarding sample size justification, the distinction between proposed and confirmed pathways, the limitations of using CSF as a proxy, and the interpretation of molecular docking results have been invaluable in improving the scientific rigor and balance of our paper. We have carefully revised the manuscript according to your suggestions. Below, please find our point-by-point responses.

 

Comment 1:The use of n = 5 and n = 4 (for CSF in the combination group) is limited for multivariate analysis, especially for OPLS-DA models, which are prone to overfitting. For example, in the sections: Materials and Methods – Section 2.3 where the sample size is defined but not justified; Results – Section 3.2, where OPLS-DA results are presented without sufficient validation (e.g., permutation testing); Results – Figure 9 and associated description, where model performance (R², Q²) is reported but without additional validation metrics; Discussion – Limitations (Section 4.5), where this important limitation is only briefly mentioned and could be expanded.

 

Response: We completely agree with your assessment. The sample size (n=5 and n=4) is indeed small for robust multivariate analysis. This limitation primarily arose from the technical difficulty and ethical considerations of collecting sufficient, uncontaminated cerebrospinal fluid (CSF) from small rodents (rats) following stringent systemic perfusion. Following your excellent suggestion, we have added a justification for the sample size in Section 2.3. Furthermore, we have performed a 200-iteration permutation test to validate the OPLS-DA model and rule out overfitting, with the results now added to Section 3.2. Finally, we have explicitly expanded on the statistical limitations of the small sample size in the Limitations section (Section 4.5).

 

Comment 2:Some of the proposed transformations (for example, the cyclization of berberrubine to coptisine, or the broader concept of <deep metabolic modulation> by Borneol) seem to be inferred from MS data without direct experimental confirmation. This is particularly noticeable in Results Section 3.3, as well as in the Discussion (Section 4.2) and the Conclusions, where these processes are sometimes described in fairly definitive terms.

 

Response:We thank the reviewer for pointing this out. We acknowledge that the metabolic transformations discussed (such as cyclization to coptisine) are putative pathways inferred from LC-MS data and lack direct functional or enzymatic confirmation in this study. We have carefully reviewed Sections 3.3, 4.2, and the Conclusions, and significantly softened our language. Words like "emerged", "elucidation", and "demonstrated" have been replaced with "putatively observed", "suggested", and "proposed".

 

Comment 3:The presence of compounds in cerebrospinal fluid is interpreted as evidence of blood–brain barrier penetration; however, it is not entirely clear whether this alone is sufficient to support such a conclusion. This point comes up in Results Section 3.3 (especially 3.3.2 and 3.3.3) and is further developed in Discussion Sections 4.1 and 4.2, as well as in the Conclusions.

 

Response: We thank the reviewer for pointing out this critical distinction. We agree that CSF presence validates CNS exposure but carries inherent ambiguity regarding the precise anatomical route of entry (endothelial BBB vs. choroidal BCSFB). We have now incorporated a discussion of this limitation in Sections 4.1 and 4.5. Additionally, following the reviewer’s recommendation, Section 4.1 now includes a brief comparison with known BBB-penetrant benchmarks (e.g., resveratrol) and a summary of analogous metabolomics methodologies in TCM research to strengthen the comparative context of our findings.

 

Comment 4:The molecular docking results are interesting and add a useful dimension to the study; however, in a few places they seem to be interpreted somewhat strongly. For instance, in Results Section 3.5, expressions such as <exceptional spatial fit> or statements implying direct modulation of target proteins may give the impression of experimental validation rather than computational prediction. A similar tendency appears in Discussion Section 4.4 and in the Conclusions.

 

Response:We appreciate this feedback. Molecular docking is indeed a predictive, hypothesis-generating tool and lacks the dynamic context of true biological systems. We have meticulously revised the text in Sections 3.5, 4.4, and the Conclusions to temper our claims. We have also explicitly stated the limitations of docking in Section 4.5.

 

Comment 5:The manuscript proposes that Borneol may modulate metabolic enzymes and possibly influence gut microbiota, which is an interesting idea. However, these mechanisms are not directly investigated in the present study. This is particularly evident in Results Section 3.3.3, Discussion Section 4.3, and the Conclusions, where these effects are sometimes described with a relatively high degree of certainty.

 

Response:We agree that the mechanisms regarding enzyme and microbiota modulation should be framed more cautiously. We have now clearly stated in Section 4.3 that these were not directly assessed and have reframed them as hypotheses. Additionally, we have identified these areas as key directions for our future research in the Limitations and Conclusions sections.

Reviewer 2 Report

Comments and Suggestions for Authors

1. The study identifies several brain-penetrating metabolites (e.g., oxyresveratrol, coptisine) and infers their roles via network pharmacology and molecular docking. However, no direct functional validation is provided. It's necessary to perform in vivo or in vitro validation (e.g., treatment with individual metabolites in ischemic models, or inhibition of their formation) to establish causal relationships.
2. The conclusion that glycosides act as pro-drugs relies primarily on their absence in CSF and the presence of aglycones. This is indirect evidence. Please incorporate time-resolved pharmacokinetics or isotope tracing to demonstrate the conversion pathway from parent compounds to active metabolites.
3. The manuscript suggests that borneol modulates CYP enzymes or gut microbiota, but no experimental data support this. The authors should include mechanistic assays (e.g., CYP activity assays, microbiome perturbation studies) or temper the claims.
4. The use of CSF as a surrogate for brain penetration has limitations, as CSF concentrations do not necessarily correlate with parenchymal levels. Please include brain tissue quantification or discuss this limitation more explicitly.
5. The study relies on qualitative or semi-quantitative data (e.g., “+ / –” detection, heatmaps), which limits pharmacological interpretation. It would be important to perform targeted LC-MS/MS quantification of key metabolites.
6. The identification of targets such as PTGS2 and TNF is based on in silico analysis only. Please validate key targets using biochemical or molecular assays (e.g., Western blot, ELISA).

Author Response

Dear Reviewer 2

 

We are deeply grateful for your meticulous and high-standard review of our work. Your comments regarding functional validation, the need for quantitative data, and the limitations of indirect evidence have provided us with a roadmap to improve the scientific rigor of our manuscript. While some of the suggested large-scale experiments (e.g., isotope tracing and in vivo target knock-down) fall outside the current scope of this material basis identification study, we have addressed your concerns by moderating our claims, incorporating extensive literature validation, and expanding our discussion on limitations.

 

Comment 1 & 6:The study identifies several brain-penetrating metabolites (e.g., oxyresveratrol, coptisine) and infers their roles via network pharmacology and molecular docking. However, no direct functional validation is provided. It's necessary to perform in vivo or in vitro validation (e.g., treatment with individual metabolites in ischemic models, or inhibition of their formation) to establish causal relationships.

The identification of targets such as PTGS2 and TNF is based on in silico analysis only. Please validate key targets using biochemical or molecular assays (e.g., Western blot, ELISA).

 

Response 1&6:We fully agree that direct functional validation (e.g., ELISA/WB for TNF-α or PTGS2) would provide more definitive proof of the proposed mechanisms. However, the primary objective of this study was to systematically identify the brain-penetrating material basis—a necessary prerequisite for subsequent mechanistic studies. To address your concern, we have:

 

1.Expanded the Discussion (Section 4.4) to include references to existing studies where the identified metabolites (e.g., oxyresveratrol, coptisine) have been independently validated as neuroprotective in ischemic models.

 

2.Clarified that our network pharmacology and docking results are predictive and hypothesis-generating.

 

3. Added the lack of biochemical validation to the Limitations section (Section 4.5).

 

Comment 2: The conclusion that glycosides act as pro-drugs relies primarily on their absence in CSF and the presence of aglycones. This is indirect evidence. Please incorporate time-resolved pharmacokinetics or isotope tracing to demonstrate the conversion pathway from parent compounds to active metabolites.

 

Response：You are correct that isotope tracing or time-resolved PK would provide direct evidence. Given the scope of this untargeted metabolomics study, we relied on the well-documented metabolic fate of flavonoids and stilbenes in the literature (i.e., gut microbiota-mediated deglycosylation). We have now cited several key studies [24,25] that demonstrate this specific conversion for Dragon’s Blood components, providing stronger contextual evidence for our "pro-drug" hypothesis in Section 4.2.

 

Comment 3: The manuscript suggests that borneol modulates CYP enzymes or gut microbiota, but no experimental data support this. The authors should include mechanistic assays (e.g., CYP activity assays, microbiome perturbation studies) or temper the claims.

 

Response:We have taken this advice seriously. Following the suggestions of Reviewer 1 and 4, we have significantly tempered these claims. These are now explicitly framed as "hypotheses" rather than conclusions in Section 4.3 and the Conclusions section.

 

Comment 4: The use of CSF as a surrogate for brain penetration has limitations, as CSF concentrations do not necessarily correlate with parenchymal levels. Please include brain tissue quantification or discuss this limitation more explicitly.

 

Response: We acknowledge the limitation that CSF levels do not always mirror parenchymal concentrations. However, our use of stringent transcardial perfusion ensured that the CSF detected was not contaminated by systemic blood, providing a reliable snapshot of central exposure. We have added a detailed discussion of this limitation in Section 4.5.

 

Comment 5: The study relies on qualitative or semi-quantitative data (e.g., “+ / –” detection, heatmaps), which limits pharmacological interpretation. It would be important to perform targeted LC-MS/MS quantification of key metabolites.

 

Response: We agree that absolute quantification is the gold standard for pharmacological interpretation. As an initial "unveiling" of the material basis, we utilized untargeted profiling to maximize the discovery of unknown metabolites. We have now acknowledged in the manuscript that targeted quantification of key candidates (e.g., oxyresveratrol) is the necessary next step and is planned for our follow-up research.

 

 

 

 

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript by Yu Zhu and co-workers describes an investigation of Dragon's Blood resin and its synergistic interaction with Borneol for ischemic stroke therapy. The authors used an integrated strategy combining UHPLC-Q-TOF-MS/MS, metabolomics, network pharmacology, and molecular docking. The work is important in understanding which components of this traditional medicine actually reach the CNS and how Borneol facilitates this process. The results will be interesting to the journal readership, although some issues require attention before publication.

Specific comments:

(1) The manuscript reports only 4 valid CSF samples in the Combination group (n=5 initially, with 4 valid samples obtained). The authors do not explain why one sample was excluded. This reduced sample size (n=4) compromises statistical power for the CSF analyses, particularly the OPLS-DA model where Q2 = 0.604 is acceptable but modest.

(2) Sections 2.2 and 2.3:  Why was the 6:1 ratio of Dragon's Blood to Borneol chosen? Provide a citation or justification based on typical use or prior studies.

(3) Table 3 reports Vina scores and cavity volumes, but crucial parameters are not reported: the docking software version (AutoDock Vina is implied but not stated), grid box dimensions, exhaustiveness, binding site locations, and protein preparation steps (e.g., protonation states).

(4) The reported Vina score for Coptisine with TNF-α (-9.2 kcal/mol) is exceptionally strong but lacks comparison with known TNF-α inhibitors or positive controls.

(5) Figures 2-4, 6, 9, 13:  Some symbols in the pictures are too small.

Summarizing, I recommend major revision of the manuscript before acceptance.

Author Response

Dear Reviewer 3

 

We sincerely thank you for your insightful and rigorous evaluation of our study. Your comments regarding the statistical power of CSF samples, the justification for the dosing ratio, the technical details of our docking protocol, and the need for comparative controls have significantly enhanced the clarity and scientific quality of our manuscript. We have addressed each point as follows:

 

Comment 1: The manuscript reports only 4 valid CSF samples in the Combination group (n=5 initially, with 4 valid samples obtained). The authors do not explain why one sample was excluded... particularly the OPLS-DA model where Q2 = 0.604 is acceptable but modest.

 

Response: We apologize for the lack of clarity. During the collection of CSF via cisterna magna puncture, one sample in the Combination (DB+B) group was contaminated with a visible amount of blood due to accidental vascular rupture. To ensure the purity of the CSF and avoid the interference of blood-borne metabolites on the brain-penetration analysis, this sample was excluded. Regarding the statistical power, we have now included a 200-iteration permutation test (as suggested by Reviewer 1) to validate the OPLS-DA model, confirming that the separation is statistically valid and not overfitted despite the small sample size. (Please see the revised Section 2.3 and Section 3.2).

 

Comment 2: Sections 2.2 and 2.3:  Why was the 6:1 ratio of Dragon's Blood to Borneol chosen? Provide a citation or justification based on typical use or prior studies.

 

Response:-This ratio was chosen based on two considerations: 1) Traditional clinical application: In many traditional Chinese medicine (TCM) formulations containing "guide drugs" like Borneol (e.g., Xuesaitong or related classic prescriptions), Borneol is typically added in a proportion of 10%–20% of the main drug. 2) Our previous pilot studies and published work [Ref 9], which demonstrated that the 6:1 ratio (Dragon’s Blood 540 mg/kg and Borneol 90 mg/kg) provides optimal neuroprotective synergy in tMCAO rats. We have added this justification and citations to Section 2.3.

 

Comment 3:Table 3 reports Vina scores and cavity volumes, but crucial parameters are not reported: the docking software version (AutoDock Vina is implied but not stated), grid box dimensions, exhaustiveness, binding site locations, and protein preparation steps (e.g., protonation states).

 

Response:We thank the reviewer for this necessary correction. We have now provided the full technical parameters for the molecular docking in a new sub-section (Section 2.8) in the Materials and Methods.

 

Comment 4:The reported Vina score for Coptisine with TNF-α (-9.2 kcal/mol) is exceptionally strong but lacks comparison with known TNF-α inhibitors or positive controls.

 

Response:We agree that a comparison with a benchmark inhibitor is essential to contextualize the binding affinity. We performed a docking simulation with SPD304 (the co-crystallized ligand of 2AZ5 and a well-known potent TNF-α inhibitor) using the same computational platform . The results showed that SPD304 achieved a Vina score of -9.6 kcal/mol, which is consistent with its role as a high-affinity standard. Our lead compound, Coptisine, yielded a competitive Vina score of -9.2 kcal/mol. This indicates that Coptisine possesses a binding potential highly comparable to that of a specialized TNF-α inhibitor, thereby supporting its role as a key CNS-active effector in Dragon's Blood therapy.

Reviewer 4 Report

Comments and Suggestions for Authors

The manuscript by Zhu et al. presents a metabolomic and systems-level analysis of Dragon’s Blood and its co-administration with Borneol to define CNS-penetrating constituents in ischemic stroke. The integration of UHPLC-Q-TOF-MS/MS, CSF profiling, and network pharmacology is technically sound, and the identification of convergent metabolism toward bioactive metabolites is of interest. However, several conclusions require clearer separation between observation and interpretation, and stronger support for mechanistic claims.

Major comments

1. The proposed “dual synergistic mechanism” of Borneol (BBB permeation and metabolic modulation) is presented as explanatory, but the data are correlative. It remains unclear whether Borneol alters metabolic pathways or increases substrate availability; this distinction should be clarified.
2. Claims regarding “non-intuitive” transformations (e.g., berberrubine to coptisine) and modulation of CYP450 enzymes or gut microbiota are not experimentally supported and should be framed as hypotheses.
3. The small cohort (n=5; CSF n=4) limits robustness for metabolomic analysis. OPLS-DA results require validation (e.g., permutation testing) to exclude overfitting.
4. The experimental context is unclear. It is not specified whether metabolomic profiling was performed in healthy or ischemic animals, which is critical given BBB disruption in stroke.
5. CSF detection is used to infer CNS exposure and, at times, functional relevance. This should be moderated, as CSF presence does not establish parenchymal distribution or target engagement.
6. The increase in detected metabolites (11 to 16) does not establish pharmacological relevance. Quantitative data are needed to assess whether these metabolites reach meaningful levels.
7. The convergence of stilbenes to oxyresveratrol is a key observation, but its designation as a “central effector” lacks functional validation and should be interpreted cautiously.

Minor comments

1. The rationale for the 6:1 ratio of Dragon’s Blood to Borneol should be clarified.
2. The absence of certain compound classes (e.g., steroids, terpenoids) from CSF despite their presence in vitro warrants discussion.

Author Response

Dear Reviewer 4,

We are deeply grateful for your rigorous and insightful review. Your comments regarding the distinction between substrate availability and pathway alteration, the experimental context (healthy vs. ischemic animals), and the pharmacological relevance of CSF metabolites have significantly strengthened the scientific depth of our manuscript. We have addressed all points as follows:

 

Major Comment 

Comment 1&2: The proposed “dual synergistic mechanism” of Borneol (BBB permeation and metabolic modulation) is presented as explanatory, but the data are correlative. It remains unclear whether Borneol alters metabolic pathways or increases substrate availability; this distinction should be clarified.

Claims regarding “non-intuitive” transformations (e.g., berberrubine to coptisine) and modulation of CYP450 enzymes or gut microbiota are not experimentally supported and should be framed as hypotheses.

 

Response 1&2: This is a profound biochemical point. We agree that the increase in detected metabolites could result from Borneol simply increasing the "substrate availability" (higher blood levels of parent drugs) rather than directly altering metabolic enzyme activity. We have now revised the Discussion (Section 4.3) to explicitly distinguish between these two possibilities and have reframed our claims about CYP450 and microbiota modulation as hypotheses rather than definitive conclusions.

 

Comment 3:The small cohort (n=5; CSF n=4) limits robustness for metabolomic analysis. OPLS-DA results require validation (e.g., permutation testing) to exclude overfitting.

 

Response:As suggested, we have now included a 200-iteration permutation test for the OPLS-DA model. The results (R2  = 0.321, Q2 = -0.062) confirm that the model is robust and not overfitted despite the small sample size (n=4-5). (Please see revised Section 3.2).

 

Comment 4:The experimental context is unclear. It is not specified whether metabolomic profiling was performed in healthy or ischemic animals, which is critical given BBB disruption in stroke

 

Response:We appreciate this critical clarification. The metabolomic profiling in this study was performed on healthy animals. The rationale was to establish the "basal" brain-penetrating material basis of Dragon’s Blood without the confounding variable of stroke-induced BBB disruption. However, we fully acknowledge that in a stroke pathology, the BBB becomes "leaky," which would likely increase the influx of both parent drugs and metabolites. We have clarified this in Section 2.3 and added a detailed discussion in Section 4.1 and 4.5 regarding how our findings might translate to an ischemic state.

 

Comment 5 & 7:CSF detection is used to infer CNS exposure and, at times, functional relevance. This should be moderated, as CSF presence does not establish parenchymal distribution or target engagement.

The convergence of stilbenes to oxyresveratrol is a key observation, but its designation as a “central effector” lacks functional validation and should be interpreted cautiously.

 

Response 5&7:We have moderated our language throughout the manuscript. We now state that CSF detection "suggests CNS exposure" rather than "establishes target engagement." The designation of oxyresveratrol as a "central effector" has been reframed as a "proposed key candidate based on its convergent metabolic profile and known antioxidant properties," pending further functional validation.

 

Comment 6:The increase in detected metabolites (11 to 16) does not establish pharmacological relevance. Quantitative data are needed to assess whether these metabolites reach meaningful levels.

 

Response : We agree that metabolite count (11 to 16) does not automatically equate to efficacy. While this study provides semi-quantitative profiling, we have added a acknowledgement in the Limitations section (Section 4.5) that absolute quantification is required to determine if these metabolites reach therapeutic thresholds.

 

Minor Comment 

Comment 1:The rationale for the 6:1 ratio of Dragon’s Blood to Borneol should be clarified

 

Response : This ratio was chosen based on two considerations: 1) Traditional clinical application: In many traditional Chinese medicine (TCM) formulations containing "guide drugs" like Borneol (e.g., Xuesaitong or related classic prescriptions), Borneol is typically added in a proportion of 10%–20% of the main drug. 2) Our previous pilot studies and published work [Ref 9], which demonstrated that the 6:1 ratio (Dragon’s Blood 540 mg/kg and Borneol 90 mg/kg) provides optimal neuroprotective synergy in tMCAO rats. We have added this justification and citations to Section 2.3.

 

Comment 2:The absence of certain compound classes (e.g., steroids, terpenoids) from CSF despite their presence in vitro warrants discussion.

 

Response: We have added a discussion in Section 4.1. The absence of these classes in the CSF, despite being abundant in the resin, is likely due to their high molecular weight, high lipophilicity (leading to sequestration in blood lipids), or their role as substrates for efflux transporters like P-glycoprotein (P-gp), which actively prevents their BBB crossing.

 

 

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have addressed the reviewers’ comments and revised the manuscript accordingly. In my opinion, the manuscript is now suitable for publication.

Reviewer 2 Report

Comments and Suggestions for Authors

I suggest accepting the revision in its present form.

Reviewer 3 Report

Comments and Suggestions for Authors

The revised version of the manuscript was significantly improved by the authors. The comments were addressed properly. I recommend acceptance of the manuscript for publication. Please ensure that high-resolution versions of figures 2-4, 11, 13 are provided according to the journal’s guidelines before publication.

Reviewer 4 Report

Comments and Suggestions for Authors

The authors have addressed the reviewers’ concerns carefully and improved the clarity and rigor of the manuscript. The revised discussion now appropriately distinguishes increased substrate availability from direct metabolic modulation by Borneol and frames CYP450 and microbiota-related effects more cautiously. The addition of permutation testing strengthens the metabolomic analysis despite the small cohort size. The authors also clarified the use of healthy animals for metabolomic profiling and moderated claims regarding CNS exposure and the role of oxyresveratrol. Overall, the revisions substantially improve the interpretation, balance, and scientific clarity of the study.