Rosmarinic Acid Targets AKR1B1 to Ameliorate Atherosclerosis via Vascular Endothelial Cell Energy Metabolism Regulation

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Reviewer comments:

The manuscript investigated the therapeutic potential of rosmarinic acid in atherosclerosis through AKR1B1-mediated endothelial energy metabolism regulation. The topic is scientifically relevant and may provide mechanistic insights into vascular protection. The manuscript is generally well written. However, several issues should be addressed to consider for publication.

Comments

1. The authors should explain the novelty aspest study compared with previous reports on rosmarinic acid, AKR1B1, and endothelial metabolism, like https://doi.org/10.3892/ijmm.2022.5125 and https://doi.org/10.1155/2017/7091904.
2. The author can include the chemical structure of the pharmacological intervention.
3. The author should include the graphical representation of the animal experimental design.
4. What was the logic to decide the rosmarinic acid dose for the animal study?
5. The authors did not include a per se group in which control (non-diseased) animals received rosmarinic acid alone. The authors should justify this omission or consider adding a rosmarinic acid per se group to strengthen the safety of rosmarinic acid.
6. In the Result and Figure legend section, the number of biological replicates for the samples is not indicated.
7. The author should provide all uncoved western blot images included all replicates.
8. The author should cite all the methodologies with appropriate citations.
9. The introduction section can be strengthened by including pharmacodynamic, pharmacokinetic, toxicity, and safety profiles of rosmarinic acid.
10. All abbreviations should be expanded at first appearance in the text.
11. The authors should ensure that all symbols, superscripts, and subscripts are consistently and correctly formatted throughout the manuscript.
12. The manuscript requires careful proofreading to correct grammatical, typographical, and formatting errors. Professional language editing is strongly recommended.
13. High-resolution versions of all figures should be provided..
14. The authors should include a discussion of the study limitations.

Author Response

Reply to reviewers’ comments:

Thanks for your comments and suggestions concerning our manuscript entitled “Rosmarinic Acid Targets AKR1B1 to Ameliorate Atherosclero-sis via Vascular Endothelial Cell Energy Metabolism Regulation” (Manuscript ID: biomolecules-4154401). These comments are all valuable and very helpful for revising and improving our paper. We have corrected carefully following comments and revised portions are marked in red color in the revised manuscript.

 

Reviewer 1#

Reviewer comments:

The manuscript investigated the therapeutic potential of rosmarinic acid in atherosclerosis through AKR1B1-mediated endothelial energy metabolism regulation. The topic is scientifically relevant and may provide mechanistic insights into vascular protection. The manuscript is generally well written. However, several issues should be addressed to consider for publication.

Comments

1.The authors should explain the novelty aspect study compared with previous reports on rosmarinic acid, AKR1B1, and endothelial metabolism, like https://doi.org/10.3892/ijmm.2022.5125 and https://doi.org/10.1155/2017/7091904.

Response: We thank the reviewer for this important question, which provides an excellent opportunity to clarify the novel contributions of our study in the context of the existing literature on rosmarinic acid (RA).

We have carefully considered the valuable studies mentioned. The work by Nyandwi et al. (2022) elegantly demonstrated that RA protects against ox-LDL-induced endothelial dysfunction under high-glucose conditions by suppressing the ROS/p38/FOXO1/TXNIP pathway, thereby reducing monocyte adhesion and diapedesis. This research firmly established RA as a key regulator of endothelial inflammation. Concurrently, the study by (Zhou et al., 2017) provided important evidence that RA could alleviate oxidative stress-induced endothelial dysfunction through the activation of the AMPK/eNOS pathway.

Building upon this critical foundation, our study offers a conceptually distinct and mechanistically deeper understanding of RA's pharmacology. The primary novelty of our work lies in shifting the focus from RA's established anti-inflammatory and antioxidant effects to a more fundamental role as a direct modulator of endothelial energy metabolism. For the first time, we demonstrate that RA's protective function in atherosclerosis is intrinsically linked to its ability to restore cellular energy homeostasis, evidenced by the rescue of both glycolytic function (ECAR) and mitochondrial respiration (OCR) in ox-LDL-injured endothelial cells.

Mechanistically, our study advances the field by identifying a novel, direct molecular target of RA. While previous studies, including the two mentioned, elegantly delineated downstream signaling cascades (such as p38/FOXO1/TXNIP and AMPK/eNOS), they did not identify a direct binding partner for RA that initiates these events. Through an integrated approach combining network pharmacology, molecular docking, molecular dynamics simulations, and, crucially, experimental validation using CETSA, SPR (with a KD of 472 nM), and enzyme activity assays, we provide robust evidence that RA directly binds to and inhibits the energy metabolism gatekeeper, AKR1B1.

Furthermore, to move beyond correlative observations, we established a causal link. We demonstrated that overexpressing AKR1B1 alone was sufficient to recapitulate the metabolic dysfunction seen in disease models, and critically, that RA treatment could rescue this AKR1B1 overexpression-induced injury. This gain-of-function approach confirms that AKR1B1 is not merely a correlate but a functional target through which RA exerts its metabolic protective effects via the downstream SIRT3/PFKFB3 axis. This represents a significant advancement in understanding the precise mechanism of action of RA, positioning the AKR1B1/SIRT3/PFKFB3 axis as a novel therapeutic strategy for atherosclerosis.

Finally, we have now revised the 1. Introduction to better contextualize this knowledge gap, as follows: Previous studies have provided valuable insights into RA's protective mechanisms, suggesting it can suppress endothelial inflammation by inhibiting the ROS/p38/FOXO1/TXNIP pathway under hyperglycemic conditions [19] and alleviate oxidative stress-induced dysfunction via AMPK/eNOS signaling [20]. However, whether RA can directly target a specific protein to regulate the fundamental process of endothelial energy metabolism in atherosclerosis remains unexplored.

2.The author can include the chemical structure of the pharmacological intervention.

Response: Thank you for your valuable suggestion. We agree that including the chemical structure of the pharmacological intervention is important for clarity. We would like to kindly inform you that the chemical structure of rosmarinic acid (RA) is already provided in the manuscript. Specifically, Figure 3D presents both the 2D and 3D structures of RA, which we included to facilitate a comprehensive understanding of the compound used throughout our study. We appreciate your careful review and hope this addresses your concern. If you have any further recommendations, we would be happy to consider them.

3.The author should include the graphical representation of the animal experimental design.

Response: Thank you for your thoughtful suggestion. We agree that a graphical representation of the animal experimental design enhances the clarity and readability of the manuscript. We would like to respectfully inform you that such a diagram is already included in the manuscript. Specifically, Figure 1A provides a schematic overview of the animal experimental protocol, illustrating the grouping, treatment duration, and intervention timeline for the control, model, RA-treated, and atorvastatin groups. We appreciate your careful review and hope this addresses your concern. Please let us know if you have any further suggestions.

4.What was the logic to decide the rosmarinic acid dose for the animal study?

Response: Thank you for your thoughtful question regarding the rationale for the rosmarinic acid (RA) doses used in our animal study. We appreciate the opportunity to clarify this important point.

The doses of RA (10 mg/kg/day for the low-dose group and 20 mg/kg/day for the high-dose group) were selected based on a logical progression from our previous work, combined with published literature and preliminary experimental data. In our recently published study on Prunella vulgaris polyphenols (PVPs) [Sun T, et al. Fitoterapia 2025], we demonstrated that PVPs at doses of 100 mg/kg/day and 200 mg/kg/day (converted from clinical equivalent doses) exerted significant anti-atherosclerotic effects by regulating lipid metabolism via the STAT3 and PI3K-Akt pathways. Notably, rosmarinic acid is the major active component of Prunella vulgaris, accounting for approximately 10% of the total polyphenol content. Therefore, the effective RA doses derived from our PVP study would correspond to approximately 10 mg/kg/day and 20 mg/kg/day, which align precisely with the doses selected for the current investigation. This direct correlation provided a strong foundation for our dose selection. In addition to this rationale, we conducted preliminary dose-ranging experiments in a small cohort of ApoE⁻/⁻ mice (n=3 per group) evaluating RA at 5, 10, 20, and 40 mg/kg/day. Our preliminary results indicated that 10 mg/kg began to significantly improve serum lipid profiles and reduce aortic plaque area, while 20 mg/kg exhibited more pronounced effects without adverse outcomes; 40 mg/kg did not provide additional benefit, suggesting that 20 mg/kg was within the optimal therapeutic window. These findings were further supported by published literature reporting that RA at doses of 10–20 mg/kg exerts protective effects against atherosclerosis in rodent models. Based on this combined evidence—our prior PVP study establishing the RA-equivalent effective dose range, our own preliminary data, and supporting literature—we selected 10 mg/kg and 20 mg/kg as the low and high doses, respectively, for the formal study. We have now revised the 2.1 Materials and Methods to include this rationale more explicitly and have added the relevant references to support our dose selection, as follows: The doses of RA were selected based on our previous study showing that Prunella vulgaris polyphenols (containing approximately 10% RA) at 100 and 200 mg/kg exerted anti-atherosclerotic effects [21], which was further validated by our preliminary dose-ranging experiments and supported by published literature.

[21] Sun T, Yuan W, Li X, Xia B, Huang J, Piao M, Chen W, Tuo Q^*. Prunella vulgaris polyphenols alleviate atherosclerosis by regulating lipid metabolism via suppressing STAT3 and activating the PI3K-Akt pathway. Fitoterapia. 2025, 186:106848.

5.The authors did not include a per se group in which control (non-diseased) animals received rosmarinic acid alone. The authors should justify this omission or consider adding a rosmarinic acid per se group to strengthen the safety of rosmarinic acid.

Response: We thank the reviewer for this thoughtful suggestion. We fully agree that including a per se group (healthy animals treated with RA alone) would provide valuable information regarding the safety profile of RA in non-diseased states. In the present study, our primary objective was to evaluate the therapeutic efficacy of RA in an established atherosclerosis model. To support the safety of RA, we relied on extensive published evidence demonstrating the low toxicity of RA in normal rodents [18], as well as our own in vitro CCK-8 assay results showing no cytotoxic effects of RA on HUVECs at concentrations up to 10 μM. However, we acknowledge that these do not substitute for an in vivo per se group. We have now explicitly addressed this as a limitation in the Discussion section and stated that future studies will include a RA-only control group to comprehensively evaluate its safety profile in healthy animals. We thank the reviewer for highlighting this point, which will help strengthen the rigor of our future work.

The discussion section is revised as follows: We also acknowledge that a healthy control group treated with RA alone was not included; while existing literature and our in vitro data support the safety of RA, this represents a limitation that should be addressed in future studies to further confirm its safety profile in non-diseased states.

[18] Hitl, M.; Kladar, N.; Gavaric, N.; Bozin, B. Rosmarinic Acid-Human Pharmacokinetics and Health Benefits. Planta Med 2021, 87, 273-282, doi:10.1055/a-1301-8648.

6.In the Result and Figure legend section, the number of biological replicates for the samples is not indicated.

Response: Thank you for pointing this out. We appreciate the reviewer's careful observation. We would like to clarify that the number of biological replicates for each experiment has already been indicated in the figure legends (e.g., "n=5" or "n=3"), and the individual data points shown in the bar graphs also represent the number of replicates. Nevertheless, to ensure clarity and avoid any ambiguity, we have carefully reviewed all figure legends and the Results section, and have now explicitly stated the number of biological replicates (n) for each experiment in the corresponding figure legends.

7.The author should provide all uncoved western blot images included all replicates.

Response: Thank you for your request. We respectfully clarify that all membranes were cut prior to antibody incubation to detect multiple targets, which is common practice. Therefore, full, uncropped images are not available. However, we will provide all original images of the cropped membranes for each replicate as Supplementary Materials, with relevant bands and molecular weight markers clearly indicated. We hope this adequately supports our data and addresses the reviewer's concern.

8.The author should cite all the methodologies with appropriate citations.

Response: We have thoroughly reviewed the Methods section and added appropriate references for key techniques, including CETSA, SPR analysis, Seahorse assays, and others, as follows:

Autophagy to Inhibit Macrophage Lipid Accumulation. Biomolecules 2024, 14, doi:10.3390/biom14101226.

23. Sun, T.; Quan, W.; Peng, S.; Yang, D.; Liu, J.; He, C.; Chen, Y.; Hu, B.; Tuo, Q. Network Pharmacology-Based Strategy Combined with Molecular Docking and in vitro Validation Study to Explore the Underlying Mechanism of Huo Luo Xiao Ling Dan in Treating Atherosclerosis. Drug Des Devel Ther 2022, 16, 1621-1645, doi:10.2147/DDDT.S357483.
24. Eberhardt, J.; Santos-Martins, D.; Tillack, A.F.; Forli, S. AutoDock Vina 1.2.0: New Docking Methods, Expanded Force Field, and Python Bindings. J Chem Inf Model 2021, 61, 3891-3898, doi:10.1021/acs.jcim.1c00203.
25. Li, X.; Sun, T.; Liu, J.; Wei, S.; Yang, Y.; Liu, J.; Zhang, B.; Li, W. Phloretin alleviates doxorubicin-induced cardiotoxicity through regulating Hif3a transcription via targeting transcription factor Fos. Phytomedicine 2023, 120, 155046, doi:10.1016/j.phymed.2023.155046.
26. Burris, D.M.; Gillespie, S.W.; Campbell, E.J.; Ice, S.N.; Yadav, V.; Picking, W.D.; Lorson, C.L.; Singh, K. Applications of Surface Plasmon Resonance (SPR) to the Study of Diverse Protein-Ligand Interactions. Curr Protoc 2024, 4, e1030, doi:10.1002/cpz1.1030.

9.The introduction section can be strengthened by including pharmacodynamic, pharmacokinetic, toxicity, and safety profiles of rosmarinic acid.

Response: We thank the reviewer for this suggestion. We have expanded the Introduction to include a concise overview of RA's pharmacokinetic, toxicity, and safety profiles, as follows: Pharmacokinetically, RA exhibits limited oral absorption (~1% of administered dose) and undergoes extensive metabolism by gut microbiota into absorbable phenolic acids. Following absorption, RA is rapidly distributed via binding to serum albumin and undergoes phase II conjugation reactions in the liver and intestine, with saturation ki-netics observed at higher doses. Renal excretion represents the primary elimination pathway, with most metabolites cleared within 24 hours. Importantly, clinical studies have reported no serious adverse effects associated with RA-containing extracts, sup-porting its favorable safety profile for long-term applications [18].

10.All abbreviations should be expanded at first appearance in the text.

Response: We have meticulously checked the entire manuscript and ensured that every abbreviation is spelled out at its first use in each section.

11.The authors should ensure that all symbols, superscripts, and subscripts are consistently and correctly formatted throughout the manuscript.

Response: We have performed a detailed check and corrected any inconsistencies in formatting of symbols (e.g., µM, °C), superscripts (e.g., ApoE⁻/⁻), and subscripts throughout the text, figures, and tables.

12.The manuscript requires careful proofreading to correct grammatical, typographical, and formatting errors. Professional language editing is strongly recommended.

Response: We have had the manuscript professionally edited by a native English-speaking editing service specializing in scientific publications. This has resolved grammatical, typographical, and stylistic issues, improving overall readability and language quality.

13.High-resolution versions of all figures should be provided.

Response: We thank the reviewer for pointing this out. We have verified that all figures were originally generated at the required resolution (300 dpi) and uploaded accordingly. The reduced image clarity in the manuscript file is due to automatic compression by Word and PDF processing software. High-resolution versions of all figures have now been prepared in TIFF format and will be submitted as separate files, as per the journal's guidelines. These files will be available for review during the revision process.

14.The authors should include a discussion of the study limitations.

Response: We thank the reviewer for this valuable suggestion. In response, we have carefully reviewed the manuscript and confirm that a discussion of the study limitations is already included in the final paragraph of the Discussion section. Previously, we acknowledged limitations such as the lack of in-depth investigation into downstream signaling pathways, the absence of validation using AKR1B1 inhibitors or knockout animal models, and the need for more omprehensive experimental approaches. In addition, we have now further expanded this section to also acknowledge that a healthy control group treated with RA alone was not included; while existing literature and our in vitro data support the safety of RA, this represents a limitation that should be addressed in future studies to further confirm its safety profile in non-diseased states. We believe these acknowledgments strengthen the scientific rigor of our study.

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents a comprehensive investigation into the anti-atherosclerotic mechanism of RA, identifying AKR1B1 as a direct target and elucidating the downstream AKR1B1/SIRT3/PFKFB3 signaling axis in restoring endothelial energy metabolism. The study employs a robust multi-method approach combining in vivo models, network pharmacology, molecular docking/dynamics, and extensive in vitro validation. The findings are novel and significant for the field of cardiovascular pharmacology. However, several aspects require clarification and improvement to strengthen the manuscript's impact and readability.

While the RA-AKR1B1 interaction is well-demonstrated (CETSA, SPR, docking), the direct causal link between AKR1B1 inhibition and the activation of the SIRT3/PFKFB3 axis needs further validation. The study shows correlative protein expression changes. To solidify the mechanism, consider: Using a specific AKR1B1 inhibitor (epalrestat) as a control in key experiments (Ox-LDL-induced HUVEC injury, energy metabolism assays) to see if it phenocopies RA's effects.

Performing rescue experiments in RA-treated cells by knocking down SIRT3 or PFKFB3 to see if the protective effects (on viability, ROS, metabolism) are abolished. Conversely, does SIRT3 overexpression rescue the defects caused by AKR1B1 overexpression?

More directly measuring the polyol pathway flux (sorbitol/fructose levels) and the NAD+/NADH ratio in response to RA and AKR1B1 manipulation would strengthen the proposed metabolic link.

Section 2.1 has anomalous sub-numbering ("2.1.2.1.2.1..."). Please correct this formatting error.

For the network pharmacology (Section 2.5), specify the exact search parameters and date of access for each database. The criteria for "EC protection-related targets" should be explicitly defined.

In the molecular dynamics simulation (Section 2.9), provide key parameters: simulation time, temperature, pressure, integration time step, and software used for analysis (GROMACS, AMBER tools).

In the discussion emphasize the novelty of your finding: RA's primary anti-AS mechanism via endothelial energy metabolism, moving beyond its known antioxidant/anti-inflammatory roles. Discuss the therapeutic relevance of targeting AKR1B1 in atherosclerosis. How does RA's IC50 (~22.5 µM) compare to known AKR1B1 inhibitors, and what are the implications for its potency in vivo?

Briefly discuss limitations. For instance, the HUVEC model, while standard, does not fully recapitulate the complex arterial endothelial environment. The use of male mice only should be acknowledged as a limitation regarding potential sex differences.

Abstract: Replace "profiles" with "ratio" for NO/ET-1 for precision.

Introduction: The transition from general AS to endothelial metabolism to RA could be smoother.

Results 3.1: The sentence "RA significantly reduced aortic root and abdominal aorta plaque formation compared to RA" appears to contain a typo ("compared to RA" likely should be "compared to HFD")

Author Response

Reply to reviewers’ comments:

Thanks for your comments and suggestions concerning our manuscript entitled “Rosmarinic Acid Targets AKR1B1 to Ameliorate Atherosclero-sis via Vascular Endothelial Cell Energy Metabolism Regulation” (Manuscript ID: biomolecules-4154401). These comments are all valuable and very helpful for revising and improving our paper. We have corrected carefully following comments and revised portions are marked in red color in the revised manuscript.

 

Reviewer 2#

This manuscript presents a comprehensive investigation into the anti-atherosclerotic mechanism of RA, identifying AKR1B1 as a direct target and elucidating the downstream AKR1B1/SIRT3/PFKFB3 signaling axis in restoring endothelial energy metabolism. The study employs a robust multi-method approach combining in vivo models, network pharmacology, molecular docking/dynamics, and extensive in vitro validation. The findings are novel and significant for the field of cardiovascular pharmacology. However, several aspects require clarification and improvement to strengthen the manuscript's impact and readability.

1.While the RA-AKR1B1 interaction is well-demonstrated (CETSA, SPR, docking), the direct causal link between AKR1B1 inhibition and the activation of the SIRT3/PFKFB3 axis needs further validation. The study shows correlative protein expression changes. To solidify the mechanism, consider: Using a specific AKR1B1 inhibitor (epalrestat) as a control in key experiments (Ox-LDL-induced HUVEC injury, energy metabolism assays) to see if it phenocopies RA's effects.

Response: Thank you for raising this important point regarding the need to strengthen the causal link within the proposed mechanism. We agree that demonstrating that inhibition of AKR1B1 per se leads to activation of the SIRT3/PFKFB3 axis would further solidify our conclusions. In our revised study, we have addressed this insightful suggestion in the following manner:

Comparative Analysis with Epalrestat: While we did not perform new experiments with epalrestat in the Ox-LDL-induced HUVEC model for the current revision, we carefully considered the underlying rationale. The reviewer's suggestion is excellent for confirming that the observed effects are specifically due to AKR1B1 inhibition. Our decision to rely on the existing multi-layered validation is based on the following reasoning:

Our data already demonstrates a direct and specific interaction between RA and AKR1B1 via CETSA, SPR (KD = 472 nM), and enzymatic inhibition assays (IC₅₀ ~22.5 µM). This establishes RA as a bona fide ligand and inhibitor of AKR1B1 in our experimental systems.

The functional rescue experiments presented in Figure 8 provide strong complementary evidence. We show that RA can counteract the detrimental effects (reduced viability, lowered SIRT3/PFKFB3, decreased ATP) induced specifically by AKR1B1 overexpression. This experiment directly links RA's beneficial effects to the attenuation of excessive AKR1B1 activity, supporting the causal role of this target.

We acknowledge that including epalrestat as a control would be a valuable addition. We have therefore incorporated a discussion of this point in the revised Limitations section, stating: "Future studies employing specific AKR1B1 inhibitors (e.g., epalrestat) as positive controls in parallel with RA would provide further confirmation that the observed metabolic improvements are primarily attributable to AKR1B1 inhibition."

To strengthen the mechanistic narrative and establish a more direct link between AKR1B1 inhibition and downstream signaling, the Discussion section has been enhanced as follows:

First, the established metabolic logic is now explicitly outlined: activation of AKR1B1 consumes NAD⁺ through the polyol pathway, resulting in a reduced NAD⁺/NADH ratio. Given that SIRT3 is a key NAD⁺-dependent deacetylase, its activity is inherently sensitive to changes in NAD⁺ availability. This connection is supported by previous literature (e.g., He et al., J Cell Physiol 2021) demonstrating that SIRT3 directly regulates PFKFB3 expression and activity in endothelial cells, thereby controlling glycolytic flux. This mechanistic link is further reinforced by new data showing that RA restores the NAD⁺/NADH ratio in Ox-LDL-treated cells (Figure 5G), providing a direct metabolic readout that bridges AKR1B1 inhibition with the subsequent activation of the NAD⁺-sensitive SIRT3/PFKFB3 axis.

In summary, while we agree that an epalrestat control experiment is an excellent idea, we believe the combination of (a) direct target engagement evidence, (b) functional rescue from AKR1B1 overexpression, and (c) demonstration of the key metabolic change (NAD⁺/NADH restoration) provides a robust and compelling chain of evidence supporting our proposed axis. We have revised the manuscript to present this logic more clearly and have acknowledged the potential value of the suggested experiment for future work. We hope this explanation and the accompanying textual revisions are satisfactory.

2.Performing rescue experiments in RA-treated cells by knocking down SIRT3 or PFKFB3 to see if the protective effects (on viability, ROS, metabolism) are abolished. Conversely, does SIRT3 overexpression rescue the defects caused by AKR1B1 overexpression?

Response: We thank the reviewer for this insightful suggestion regarding SIRT3/PFKFB3 rescue experiments. We fully agree that such genetic approaches would provide definitive evidence for the hierarchical relationship within the AKR1B1/SIRT3/PFKFB3 axis.

While we have not yet performed these specific experiments, our current data collectively establish a coherent mechanistic framework: (1) RA directly binds to and inhibits AKR1B1 (CETSA, SPR, enzyme assays); (2) RA modulates AKR1B1/SIRT3/PFKFB3 expression in both in vivo and in vitro models; (3) RA restores the NAD⁺/NADH ratio, providing a metabolic link between AKR1B1 inhibition and activation of the NAD⁺-dependent SIRT3; (4) AKR1B1 overexpression alone phenocopies Ox-LDL-induced metabolic dysfunction; and (5) RA rescues the defects caused by AKR1B1 overexpression. These findings collectively support AKR1B1 as an upstream regulator of the SIRT3/PFKFB3 axis.

It is important to note that the present study focuses on establishing RA's anti-atherosclerotic effect and identifying its primary target, with the AKR1B1/SIRT3/PFKFB3 axis being elucidated through a combination of literature evidence and experimental validation. As this represents an initial mechanistic exploration, deeper genetic validation—including SIRT3/PFKFB3 knockdown or overexpression experiments—constitutes a logical next step in our ongoing research. We have now acknowledged this in the revised Discussion and proposed these experiments as a key direction for future investigations.

We have supplemented this in Discussion section, as follows: While our data collectively support a model wherein RA inhibits AKR1B1 to restore NAD⁺ homeostasis and subsequently activate the SIRT3/PFKFB3 axis, we acknowledge that direct genetic rescue experiments would provide further definitive evidence for this hierarchical relationship. Specifically, future studies employing SIRT3 or PFKFB3 knockdown in RA-treated cells, or SIRT3 overexpression in the context of AKR1B1 upregulation, would confirm whether these molecules are essential downstream ef-fectors of RA's protective effects. Such experiments represent an important direction for further elucidating the precise molecular architecture of this signaling pathway.

3.More directly measuring the polyol pathway flux (sorbitol/fructose levels) and the NAD⁺/NADH ratio in response to RA and AKR1B1 manipulation would strengthen the proposed metabolic link.

Response: We thank the reviewer for this valuable suggestion. While we did not directly quantify sorbitol and fructose levels, our study employed a comprehensive approach to validate the metabolic link between RA and AKR1B1. First, we measured the NAD⁺/NADH ratio (Figure 5G), which represents the key functional endpoint linking polyol pathway activity to downstream NAD⁺-dependent signaling. Second, we performed targeted energy metabolomics (Figure 6J-K), which comprehensively mapped the metabolic consequences of RA treatment. This analysis revealed that RA significantly restored multiple glycolytic and TCA cycle intermediates (including G-6-P, F-6-P, ATP, and citric acid) that were dysregulated by Ox-LDL, providing direct evidence that RA modulates global cellular energy metabolism downstream of AKR1B1. Together with our CETSA, SPR, and enzyme activity data demonstrating direct RA-AKR1B1 binding and inhibition, these findings establish a clear mechanistic link without requiring direct sorbitol/fructose quantification. We acknowledge that future metabolomic studies quantifying specific pathway intermediates would be a useful extension.

4.Section 2.1 has anomalous sub-numbering ("2.1.2.1.2.1..."). Please correct this formatting error.

Response: We apologize for this formatting error, which occurred during manuscript preparation. The numbering in Section 2 has been corrected to a logical, hierarchical structure.

5.For the network pharmacology (Section 2.5), specify the exact search parameters and date of access for each database. The criteria for "EC protection-related targets" should be explicitly defined.

Response: Thank you for your valuable suggestions. We have carefully revised the Table 1,as follows:

Table 1. The involved database URLs in Bioinformatics analysis

Order	Database name	URL	Search parameters	Date of access
1	TCMSP	https://old.tcmsp-e.com/tcmsp.php	a score cutoff of 20 and altered P-value of 0.05	From November 1, 2023 to May 30, 2024
2	BATMAN-TCM	http://bionet.ncpsb.org.cn/batman-tcm/#/home	a score cutoff of 20 and altered P-value of 0.05
3	SwissTargetPrediction	http://www.swisstargetprediction.ch/	probability > 0
4	UniProt	https://www.uniprot.org/	set the organism classification to “Homo sapiens”
5	GeneCards	https://www.genecards.org/	/
6	PharmGKB	https://www.pharmgkb.org/	/
7	TTD	http://db.idrblab.net/ttd/	/
8	DrugBank	https://go.drugbank.com/
9	Venn platform	https://bioinfogp.cnb.csic.es/tools/venny/index.html	/
10	STRING	https://cn.string-db.org/	a minimum interaction score of 0.90

 

6.In the molecular dynamics simulation (Section 2.9), provide key parameters: simulation time, temperature, pressure, integration time step, and software used for analysis (GROMACS, AMBER tools).

Response: We thank the reviewer for the valuable comment regarding the molecular dynamics (MD) simulation parameters. In response, we have now provided the detailed key parameters used in the MD simulation, which were also described in our previously published work (reference [23]) and have been cited in the current manuscript.

The key parameters and reversed version are summarized as follows:

The system was modeled using the AmberFF99SB force field for the protein and the GAFF force field for RA. The complex was solvated in a cubic water box with a 1 nm buffer, and Na⁺ ions were added to neutralize the system.

The simulation workflow included: (1) Energy minimization: A two-step mini-mization was carried out. First, the protein was restrained while water molecules were minimized (1500 steps of steepest descent, total 5000 cycles). Then, the entire system was minimized without restraints (2000 steps of steepest descent, total 5000 cycles). (2) Equilibration: The system was gradually heated from 0 to 310 K over 100 ps using the Langevin thermostat, followed by 100 ps of pressure equilibration at 1 bar using the isotropic Berendsen coupling method. (3) Production run: A 100 ns MD simulation was conducted at a constant temperature of 310 K (human body temperature) and a pres-sure of 1 bar, with an integration time step of 2 fs. The cutoff distance for van der Waals and short-range electrostatic interactions was set to 10 Å, and long-range electrostatics were treated using the Particle‑Mesh‑Ewald (PME) method.

All simulations were performed using GROMACS (version 2018 or later). Trajec-tory analysis, including root‑mean‑square deviation (RMSD) and root‑mean‑square fluctuation (RMSF), was performed using GROMACS built‑in tools, and binding modes were visualized with PyMOL. The stability of the AKR1B1-RA complex was assessed throughout the 100 ns trajectory, and the representative structure was ex-tracted for further analysis.

7.In the discussion emphasize the novelty of your finding: RA's primary anti-AS mechanism via endothelial energy metabolism, moving beyond its known antioxidant/anti-inflammatory roles. Discuss the therapeutic relevance of targeting AKR1B1 in atherosclerosis. How does RA's IC50 (~22.5 µM) compare to known AKR1B1 inhibitors, and what are the implications for its potency in vivo?

Response: Thank you for your valuable suggestion. We provide the following explanation regarding RA's potency as an AKR1B1 inhibitor:

In evaluating RA's potency as an AKR1B1 inhibitor, our enzyme kinetics assays revealed an IC₅₀ of approximately 22.5 µM, a value notably higher than that of epalrestat (EP, IC₅₀ = 1.07 µM). This difference in inhibitory potency may be attributed, at least in part, to distinct binding modes between RA and EP within the AKR1B1 active site. Molecular docking analysis (Figure 3E) reveals that RA forms hydrogen bonds with key residues including W20, H110, W111, L301, and C298. In contrast, EP—a structurally distinct carboxylic acid derivative—has been reported to interact with a different set of residues within the active site, which may contribute to its higher binding affinity and more potent enzymatic inhibition. These structural differences provide a plausible explanation for the observed IC₅₀ disparity.

Importantly, our Western blot and immunofluorescence analyses demonstrated that RA significantly downregulates AKR1B1 protein expression in both Ox-LDL-induced HUVECs and the aorta of ApoE⁻/⁻ mice. This dual regulatory mechanism, which combines direct enzymatic inhibition with suppression of protein expression, may enhance RA's overall functional impact beyond what is reflected by its in vitro IC₅₀.

Regarding in vivo potency, several factors contextualize RA's efficacy despite its moderate IC₅₀. First, the effective concentration at the target site in vivo can differ substantially from in vitro IC₅₀ values due to tissue distribution, protein binding, and potential accumulation in vascular tissues. Second, RA's well-characterized antioxidant properties may synergistically suppress the oxidative stress that activates AKR1B1, indirectly contributing to reduced enzyme activity. Third, chronic administration at 20 mg/kg/day in our ApoE⁻/⁻ mouse model produced robust anti-atherosclerotic effects (Figure 1), providing functional evidence of RA's in vivo efficacy. Moreover, RA's moderate inhibitory potency may offer pharmacological advantages, potentially resulting in fewer off-target effects compared to highly potent synthetic inhibitors, making it an attractive candidate for chronic preventive strategies in early-stage AS.

We have supplemented this point in Discussion section as follows: Enzyme kinetics assays revealed that RA inhibits AKR1B1 with an IC₅₀ of 22.5 µM, notably higher than epalrestat (1.07 µM). This difference is likely due to distinct binding modes: molecular docking shows RA interacts with residues W20, H110, W111, L301, and C298, whereas EP binds to a different set. Importantly, RA also downregu-lates AKR1B1 protein expression in vitro and in vivo, indicating a dual regulatory mechanism that may enhance its functional impact beyond the IC₅₀ value.

8.Briefly discuss limitations. For instance, the HUVEC model, while standard, does not fully recapitulate the complex arterial endothelial environment. The use of male mice only should be acknowledged as a limitation regarding potential sex differences.

Response: Thank you for your thoughtful suggestion. In accordance with your comment, we have now briefly discussed the limitations of our study in the Discussion section. The revised text is as follows: First, although HUVECs are a well-established and widely used in vitro model for investigating endothelial cell function, they do not fully recapitulate the complex hemodynamic environment and cellular heterogeneity of the arterial endothelium in vivo. Future studies employing more advanced models, such as microfluidic chips or primary arterial endothelial cells, may provide additional insights. Second, our in vivo experiments were conducted exclusively using male ApoE⁻/⁻ mice. Given the known sex differences in the incidence and progression of atherosclerosis, the inclusion of female animals in future studies will be important to assess potential sex-specific effects of RA.

9.Abstract: Replace "profiles" with "ratio" for NO/ET-1 for precision.

Response: Thank you for your careful reading and valuable suggestion. In accordance with your comment, we have revised the abstract to replace the word "profiles" with a more precise term. The original phrase "improved serum NO/ET-1 profiles" has been changed to "improved serum NO and ET-1 levels" to more accurately reflect the specific parameters measured in our study. We believe this modification enhances the clarity and precision of the abstract. Thank you again for your insightful feedback. 10.Introduction: The transition from general AS to endothelial metabolism to RA could be smoother.

Response: Thank you for your suggestion. We have rewritten the latter part of the Introduction to create a smoother logical flow: atherosclerosis → endothelial dysfunction → central role of energy metabolism → current therapeutic gaps → RA as a promising candidate with unknown mechanism in this metabolic context.

11.Results 3.1: The sentence "RA significantly reduced aortic root and abdominal aorta plaque formation compared to RA" appears to contain a typo ("compared to RA" likely should be "compared to HFD")

Response: Thank you for your careful review and valuable comment regarding Section 3.1. We appreciate the opportunity to clarify this point. In the original manuscript, the sentence reads: "RA_H significantly reduced aortic root and abdominal aorta plaque formation compared to RA_L." Here, "RA_H" refers to the high-dose rosmarinic acid group (20 mg/kg/day), and "RA_L" refers to the low-dose rosmarinic acid group (10 mg/kg/day). The comparison is therefore between the two RA treatment groups, not between RA and the HFD model group. This distinction is important because it demonstrates a dose-dependent effect of RA.

To avoid any potential confusion, we have revised the sentence in the manuscript to explicitly state: "High-dose RA (RA_H) significantly reduced aortic root and abdominal aorta plaque formation compared to low-dose RA (RA_L)." This modification ensures clarity for all readers.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript has been drastically improved following revision, and I think it's accepted in the current form.

 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have adequately addressed my comments and concerns.