Review Reports - Multi-Omics Landscape of Circadian Clock Dysregulation Across the Chronic Liver Disease Spectrum

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

A brief summary

This manuscript reviews circadian clock dysregulation across a broad spectrum of chronic liver diseases, including MAFLD/NAFLD/NASH, ALD, viral hepatitis, HCC, fibrosis, and cholestatic disease. A strength of the manuscript is that it attempts to integrate not only transcriptomics but also proteomics, post-translational modifications, metabolomics, the gut-liver axis, and emerging single-cell and spatial approaches. Tables 1-3 and the translational section make the review potentially useful for readers. However, in its current form the manuscript requires revision: the literature selection strategy is not described, the central claim that four molecular features define hepatic circadian failure needs more explicit evidence-strength grading, and some translational conclusions are stronger than the available temporally resolved human data can support.

General concept comments

For a review with such a broad scope, the manuscript should include at least a brief description of the literature selection strategy. The review covers multiple liver diseases, several omics layers, animal models, human cohorts, cell lines, and computational analyses, but it does not specify which databases were searched, what time period was considered, which search terms were used, or how sources were included or excluded. This is especially important because the evidence base is very uneven across disease categories and omics layers. Please add a concise methodological paragraph and state whether the review is narrative, scoping, or has any systematic elements. This would make the evidence base more transparent and reduce the impression of selective literature inclusion.
The central conclusion that BMAL1 functional downregulation, REV-ERBα output attenuation, NAD+ amplitude reduction, and gut-liver axis desynchronization together constitute a recognizable molecular phenotype of hepatic circadian failure is interesting, but it needs clearer grading of evidence strength. The manuscript itself shows that the depth of evidence differs substantially: HBV evidence is largely transcript-level, ALD relies on a single genome-wide time-resolved multi-omics study, cholestatic disease has limited direct clock-focused evidence, and temporally resolved human tissue data are nearly absent. The authors should distinguish established findings, plausible mechanistic inferences, and hypothesis-generating claims. This could be done in a separate confidence table or by expanding Table 3 with an evidence-strength/confidence column for each disease category and omics layer.
The translational section is useful, but it should be more tightly connected to the limitations of the available human evidence. Section 4 discusses biomarkers, phase inference, REV-ERBα agonism, chronotherapeutic targeting, and NAD+ restoration, while the manuscript also acknowledges that human time-resolved tissue data are nearly absent, REV-ERBα agonism has not been tested in human clinical trials, and NAD+ interventions were not designed around circadian endpoints. Please add a more explicit limitations paragraph and practical requirements for future studies, including standardized sampling time, PBMC versus liver tissue caveats, disease stage and etiology stratification, and pre-specified circadian endpoints in intervention trials. In the current version, the clinical actionability of the review reads somewhat premature.

The main text should more consistently label the source and level of evidence. Several key arguments move between healthy mouse liver, disease models, human liver tissue, PBMC surrogate data, cell lines, microbiota studies, immune deconvolution, and computational reanalysis opportunities. Because this topic depends heavily on translation across species, tissue compartments, and omics platforms, readers need to see immediately where a conclusion is supported by direct human liver data and where it is based on mechanistic extrapolation. Please add such qualifiers in key paragraphs and make the tables more consistent in marking evidence source.

Specific comments

Tables 1 and 3 are useful, but they are too dense in the current layout. Table 1 spans pages 5-6, Table 3 spans pages 21-22, and the narrow columns produce extensive line wrapping. The authors should improve the formatting, add footnotes defining abbreviations, and consider moving long mechanistic explanations from the tables into the text or supplementary material.
Figure 2 and Figure 3 contain small heatmap labels and panel details. Since these figures carry important evidentiary weight, the authors should check resolution, readability after journal scaling, and permission/adaptation statements for reused material.
A final editorial pass is needed. The first page still contains template placeholders, including Academic Editor, Received/Revised/Accepted/Published dates, Citation, and DOI. The manuscript should also clarify or standardize the use of MAFLD, NAFLD, and NASH terminology so readers understand why different nomenclature is used in different places.

Author Response

A brief summary

Response:

We thank Reviewer 1 for a thorough and constructive evaluation of our manuscript. The comments identified genuine limitations in how we presented and graded evidence across disease categories, and we are grateful for the clear and actionable direction they provided.

We have revised the manuscript extensively in response to all four conceptual comments and all three specific comments. First, we added a methodological paragraph in the Introduction describing our literature search strategy and the narrative nature of this review. Second, we restructured how the four recurring molecular features are presented throughout the manuscript, repositioning them as an inferential framework rather than an established pan-disease mechanism, and supplementing this with evidence-source qualifiers embedded in key paragraphs across all disease discussions. Third, we added a dedicated limitations paragraph at the end of the translational discussion, addressing the points raised regarding human tissue data, PBMC caveats, disease stratification, and trial design requirements. Fourth, we simplified Tables 1 and 3 substantially, reducing column counts and moving mechanistic detail into the main text. Fifth, we revised the figure legends for Figures 3 and 4, verified image resolution, and confirmed copyright attribution statements. Sixth, we standardized disease nomenclature throughout the manuscript and removed all template placeholders from the title page.

All revisions are indicated in red text in the revised manuscript. The point-by-point responses below address each comment in turn.

General concept comments

1. For a review with such a broad scope, the manuscript should include at least a brief description of the literature selection strategy. The review covers multiple liver diseases, several omics layers, animal models, human cohorts, cell lines, and computational analyses, but it does not specify which databases were searched, what time period was considered, which search terms were used, or how sources were included or excluded. This is especially important because the evidence base is very uneven across disease categories and omics layers. Please add a concise methodological paragraph and state whether the review is narrative, scoping, or has any systematic elements. This would make the evidence base more transparent and reduce the impression of selective literature inclusion.

Response:

We thank the reviewer for this important observation. We agree that the absence of a methodological statement reduced the transparency of our evidence base, particularly given the breadth of disease categories and omics layers covered.

We have added a dedicated methodological paragraph at the end of the Introduction. The paragraph reads as follows.

"Literature was identified through PubMed and Web of Science searches, with no fixed date restriction, using terms related to circadian clocks, liver disease, and omics approaches. This is a narrative review. The depth of coverage across disease categories reflects the underlying evidence base, as available data differ substantially across disease types and omics layers."

We also note that we have explicitly acknowledged the uneven distribution of evidence throughout the manuscript. In the introductory paragraph of the disease discussion, we state that transcriptomic and epigenomic data are comparatively mature for MAFLD and HCV hepatitis, while time-resolved proteomic and metabolomic data remain absent for HBV-related disease and cholestatic liver disease. This acknowledgment is further reinforced in the revised Table 3, which now includes a dedicated column summarizing the principal evidence gaps for each disease category, and in the Conclusions, which explicitly addresses the uneven distribution of the multi-omics evidence base across disease contexts.

2. The central conclusion that BMAL1 functional downregulation, REV-ERBα output attenuation, NAD+ amplitude reduction, and gut-liver axis desynchronization together constitute a recognizable molecular phenotype of hepatic circadian failure is interesting, but it needs clearer grading of evidence strength. The manuscript itself shows that the depth of evidence differs substantially: HBV evidence is largely transcript-level, ALD relies on a single genome-wide time-resolved multi-omics study, cholestatic disease has limited direct clock-focused evidence, and temporally resolved human tissue data are nearly absent. The authors should distinguish established findings, plausible mechanistic inferences, and hypothesis-generating claims. This could be done in a separate confidence table or by expanding Table 3 with an evidence-strength/confidence column for each disease category and omics layer.

Response:

We thank the reviewer for this substantive comment. We agree that the original manuscript did not adequately distinguish between conclusions of different evidential maturity, and that this weakened the credibility of the central framework.

We considered the reviewer's suggestion of adding an evidence-strength column to Table 3. However, we concluded that a column-based rating system would be difficult to apply accurately at the disease-category level, because individual studies within the same disease category often differ substantially in their evidential quality. Assigning a single rating to an entire row would obscure rather than clarify these within-category differences.

Instead, we have addressed this concern through two complementary approaches. First, we have repositioned the four molecular features throughout the manuscript as an inferential framework rather than an established pan-disease mechanism. The Abstract now states that these features "together form a proposed inferential framework for hepatic circadian failure" and explicitly notes that this "is not an established pan-disease mechanism." The Introduction and Conclusions carry equivalent language. Second, we have embedded evidence-source qualifiers systematically throughout the disease discussion. Conclusions drawn from animal models, cell lines, human chimeric models, PBMC surrogate data, and direct human tissue are now explicitly labelled at the paragraph and sentence level. For example, the HBV discussion notes that all findings derive from transcript-level measurements in small cohorts with no time-resolved data. The ALD discussion notes that the sole genome-wide multi-omics study requires independent replication. The cholestatic disease discussion notes that molecular clock gene expression in human PBC or PSC liver tissue has not been measured.

These within-text qualifiers are reinforced by the revised Table 3, which now includes a "Principal Evidence Gaps" column that summarizes the key evidentiary limitations for each disease category. Together, these changes allow readers to assess the strength of evidence for each disease context without requiring a simplified numerical rating that would misrepresent the heterogeneity within categories.

3. The translational section is useful, but it should be more tightly connected to the limitations of the available human evidence. Section 4 discusses biomarkers, phase inference, REV-ERBα agonism, chronotherapeutic targeting, and NAD+ restoration, while the manuscript also acknowledges that human time-resolved tissue data are nearly absent, REV-ERBα agonism has not been tested in human clinical trials, and NAD+ interventions were not designed around circadian endpoints. Please add a more explicit limitations paragraph and practical requirements for future studies, including standardized sampling time, PBMC versus liver tissue caveats, disease stage and etiology stratification, and pre-specified circadian endpoints in intervention trials. In the current version, the clinical actionability of the review reads somewhat premature.

Response:

We thank the reviewer for this well-founded criticism. We agree that the original translational discussion presented therapeutic and biomarker opportunities without adequately contextualizing them against the current limitations of human evidence, which created an imbalance between mechanistic ambition and clinical readiness.

We have addressed this through two targeted revisions. First, we restructured the biomarker discussion in the revised discussion to include an explicit paragraph on practical design requirements for future studies. That paragraph specifies four concrete requirements. Sampling time should be anchored to habitual wake time rather than absolute clock time. PBMC-derived phase estimates should be treated as systemic surrogates rather than direct proxies for hepatic clock status, given that peripheral and hepatic circadian programs can diverge substantially in liver disease. Studies should stratify participants by disease stage and etiology, as clock dysregulation patterns differ meaningfully across disease categories. Intervention trials targeting circadian pathways should pre-specify circadian endpoints rather than treating them as exploratory outcomes.

Second, we added a dedicated limitations paragraph at the end of the therapeutic discussion. That paragraph explicitly states that human time-resolved liver tissue data are nearly absent across all disease categories, identifies the ethical and logistical constraints that account for this gap, confirms that REV-ERBα agonism has not advanced beyond preclinical models in any liver disease indication, and notes that existing NAD⁺ intervention trials were not designed around circadian endpoints and therefore cannot be used to infer circadian efficacy. The paragraph concludes by stating that addressing these gaps requires prospective studies with temporally designed sampling, etiology-stratified cohorts, and pre-specified circadian endpoints.

We believe these additions bring the translational discussion into appropriate alignment with the current state of human evidence and address the reviewer's concern that clinical actionability was overstated in the original version.

4. The main text should more consistently label the source and level of evidence. Several key arguments move between healthy mouse liver, disease models, human liver tissue, PBMC surrogate data, cell lines, microbiota studies, immune deconvolution, and computational reanalysis opportunities. Because this topic depends heavily on translation across species, tissue compartments, and omics platforms, readers need to see immediately where a conclusion is supported by direct human liver data and where it is based on mechanistic extrapolation. Please add such qualifiers in key paragraphs and make the tables more consistent in marking evidence source.

Response:

We thank the reviewer for this comment. We agree that the original manuscript moved between different evidence sources without consistently signalling these transitions to the reader, which made it difficult to assess the translational weight of individual conclusions.

We have conducted a systematic revision of all disease discussion paragraphs to address this concern. Evidence-source qualifiers are now embedded at the sentence level throughout the manuscript. Conclusions derived from healthy mouse liver studies, disease animal models, human liver chimeric mouse models, human biopsy cohorts, PBMC surrogate data, cell line experiments, and non-hepatic model extrapolations are each labelled explicitly at the point where the conclusion is introduced. At the table level, the revised Table 1 includes brief evidence-source notations for disease categories where the evidence base is particularly limited. The revised Table 3 consolidates evidence coverage and principal gaps by disease category in a format that allows readers to assess the depth and source of available evidence at a glance.

Specific comments

1. Tables 1 and 3 are useful, but they are too dense in the current layout. Table 1 spans pages 5-6, Table 3 spans pages 21-22, and the narrow columns produce extensive line wrapping. The authors should improve the formatting, add footnotes defining abbreviations, and consider moving long mechanistic explanations from the tables into the text or supplementary material.

Response:

We thank the reviewer for this feedback. Both tables have been substantially restructured. In Table 1, the "Predominant Silencing Mechanism" column has been removed entirely, with mechanistic detail relocated to the corresponding disease discussion paragraphs. The table now presents only the downregulated and upregulated clock components by disease category, reducing it to a concise reference index. In Table 3, the original five-column structure has been condensed to three columns by merging the "Available Omics Evidence" and "Omics Depth" columns and removing the "Representative Multi-Omics Finding" and "References" columns. A "Principal Evidence Gaps" column is retained.

2. Figure 2 and Figure 3 contain small heatmap labels and panel details. Since these figures carry important evidentiary weight, the authors should check resolution, readability after journal scaling, and permission/adaptation statements for reused material.

Response:

We thank the reviewer for drawing attention to this. These figures now appear as Figures 3 and 4 in the revised manuscript, reflecting the addition of a new schematic figure. We have verified that the original high-resolution source files for both figures meet the journal's minimum resolution requirements. The figure legends have been simplified to improve readability and to ensure that the central message of each figure is immediately clear without requiring readers to parse dense explanatory text. Regarding attribution, Figure 3 is adapted from Duong et al. (2024), *Nature Communications*, under CC BY 4.0, and Figure 4 is reproduced from Mukherji et al. (2024), *Nature Communications*, under CC BY 4.0. Both statements are included in the revised figure legends.

3. A final editorial pass is needed. The first page still contains template placeholders, including Academic Editor, Received/Revised/Accepted/Published dates, Citation, and DOI. The manuscript should also clarify or standardize the use of MAFLD, NAFLD, and NASH terminology so readers understand why different nomenclature is used in different places.

Response:

We thank the reviewer for these editorial observations. All template placeholders on the title page, including the Academic Editor field, date fields, Citation, and DOI, have been removed in the revised manuscript.

Regarding terminology, we have standardized disease nomenclature throughout the revised manuscript. MAFLD and MASH are used as the primary terms, consistent with current nomenclature recommendations. A brief explanatory note is provided in the Introduction to clarify this convention for readers.

Reviewer 2 Report

Comments and Suggestions for Authors

This review synthesizes the disruption of the circadian clock in chronic liver diseases from an integrated multi-omics perspective and provides a cross-sectional discussion of pathologies including MAFLD, ALD, viral hepatitis, HCC, fibrosis, and cholestatic diseases. In particular, it proposes a common phenotype characterized by “decreased BMAL1 activity, attenuated REV-ERBα amplitude, reduced NAD⁺ oscillations, and desynchronization of the enterohepatic axis,” thereby offering a substantial body of information that contributes to an integrated understanding of this field.

Below, we outline the issues with this review manuscript section by section.

The Abstract presents four elements as a “common molecular phenotype”; however, direct evidence supporting this across all disease categories remains limited. Given that this represents an important limitation of the manuscript, it should at least be briefly acknowledged in the Abstract.
In the Introduction, the descriptions of individual diseases are somewhat lengthy, which obscures the primary aim of the review. These sections should be streamlined. In addition, the statement that this review “bridges two bodies of work” lacks specificity. It would improve clarity to explicitly state the novelty and originality of this review, for example in bullet-point form.
In Section 2, the discussion of non-transcriptional rhythms—particularly the persistence of rhythmicity in Bmal1-null models—should be clearly framed in terms of whether it reflects established evidence or remains hypothetical. Furthermore, the relationships among the transcriptome, proteome, and metabolome are not clearly articulated; a schematic illustration summarizing the interactions among these omics layers would be helpful. The mention of single-cell and spatial approaches is currently descriptive; a more concrete explanation of what specific insights these approaches can provide is needed.
In Section 3, the presentation of “common patterns” does not sufficiently account for heterogeneity among different diseases. In addition, Table 1 is information-dense and difficult to interpret, particularly the “Predominant Silencing Mechanism” column. Simplification or division of the table would improve readability.
In Section 3.1, the influence of model dependency is not sufficiently discussed, and the relationship to human data requires clarification. In the metabolomics section, the discussion of amino acid metabolism remains largely hypothetical. Moreover, the interpretation of “clock-independent” rhythms is ambiguous and not clearly distinguished from feeding-dependent rhythms. Human data are also limited.

The Gut Microbiota section is particularly hypothesis-driven, relying primarily on animal studies with substantial extrapolation to humans. Because the causal link between microbiota and liver disease progression is not firmly established, it would be more appropriate to describe microbiota as a “potential mediator” rather than implying a direct causal role.
Section 3.2 relies heavily on chimeric mouse models, and treating these findings as equivalent to human data may be somewhat overstated. The discussion of epigenetic memory is also based on limited evidence; therefore, the limitations of these models should be more clearly acknowledged.

In the HCC section, the content is complex and combines transcriptional regulation, post-translational modifications, metabolism, and signaling pathways. A clearer structural organization is needed. As a possible approach, the section could be reorganized into subsections such as “transcriptional dysregulation,” “post-translational rewiring,” and “metabolic reprogramming.” Such restructuring would substantially improve clarity.

The “Immune & Cell Cycle” subsection relies heavily on extrapolation from other cancer types; therefore, these interpretations should be explicitly presented as hypotheses.
In Section 3.3, the specificity of REV-ERB agonists in the context of fibrosis is a concern, and potential off-target effects should be clearly emphasized. In the ALD subsection, mechanisms involving the NAD⁺ pathway and ROS are intertwined and would benefit from clearer organization. In addition, human evidence is limited to PBMC-based studies, which should be acknowledged as a limitation. The discussion of cholestatic disease is relatively sparse and unbalanced; if this topic is considered important, the section should be expanded.
As a cross-cutting issue, there is insufficient distinction between established findings and hypotheses, and the lack of human data is a common limitation across sections. In addition, temporal resolution is often lacking, and there is occasional conflation of causation and correlation. To address these issues, the manuscript should explicitly indicate the level of evidence, clearly label hypotheses, avoid overstating causality, and further refine its structure.

In particular, while the manuscript correctly highlights the “lack of temporally resolved human data,” it would be valuable to elaborate on the extent of this limitation, the reasons underlying it (e.g., ethical constraints, sampling timing, and invasiveness), and potential alternative approaches such as phase inference algorithms or circulating biomarkers.
There is also room for improvement in the figures and tables. Table 1, while comprehensive, is difficult to interpret due to its density; dividing it by disease category or incorporating color coding may improve clarity. Figure 2 includes extensive explanatory text that obscures the central message; simplifying the figure to emphasize a single key point would enhance its impact.

Minor Comments

The terms NAFLD and MAFLD are used somewhat interchangeably; consistent terminology should be adopted throughout the manuscript.
The balance of citations is uneven, with some sections relying heavily on studies from the early 2010s. Inclusion of more recent single-cell and spatial omics studies is recommended.

Author Response

Below, we outline the issues with this review manuscript section by section.

Response:

We thank Reviewer 2 for a detailed and rigorous evaluation of our manuscript. The comments identified important structural and evidential concerns that have substantially improved the clarity and intellectual honesty of the revised manuscript. We are grateful for the effort and specificity with which each issue was articulated.

All revisions are indicated in red text in the revised manuscript. The point-by-point responses below address each comment in turn.

Comments 1：

The Abstract presents four elements as a “common molecular phenotype”; however, direct evidence supporting this across all disease categories remains limited. Given that this represents an important limitation of the manuscript, it should at least be briefly acknowledged in the Abstract.

Response:

We thank the reviewer for this observation. We agree that the original phrasing overstated the evidential basis for the four molecular features, and that the Abstract should reflect the same epistemic caution applied in the main text.

We have revised the relevant sentences in the Abstract to read as follows.

" Four molecular features recur across these contexts. BMAL1 functional downregulation, REV-ERBα oscillatory output attenuation, NAD⁺ oscillatory amplitude reduction, and gut-liver axis circadian desynchronization together constitute an inferential framework for hepatic circadian failure rather than an established pan-disease mechanism."

This revision explicitly repositions the four features as an inferential framework, consistent with how they are presented throughout the revised manuscript, including in the Introduction, the disease discussions, and the Conclusions.

Comments 2：

In the Introduction, the descriptions of individual diseases are somewhat lengthy, which obscures the primary aim of the review. These sections should be streamlined. In addition, the statement that this review “bridges two bodies of work” lacks specificity. It would improve clarity to explicitly state the novelty and originality of this review, for example in bullet-point form.

Response:

We thank the reviewer for this constructive observation. We agree that the original Introduction devoted excessive space to individual disease descriptions, which delayed the reader's understanding of the review's analytical focus.

We have streamlined the disease overview paragraph substantially. Each disease is now described in a single sentence that identifies only the most relevant clock-disease connection, with mechanistic detail reserved for the disease-specific discussions. This compression reduces the paragraph to approximately half its original length while preserving the necessary clinical context.

We have also replaced the phrase "bridges two bodies of work" with an explicit three-point statement of the review's distinctive contribution. The revised Introduction states that this review differs from prior work in three respects. It integrates evidence from transcriptomics, proteomics, post-translational modification profiling, metabolomics, and emerging single-cell and spatial approaches within a disease-organized analytical framework. It proposes four molecular features that recur across disease categories as an inferential framework rather than an established pan-disease mechanism. It explicitly maps where the evidence is mature and where critical gaps remain, with the aim of orienting future human-focused research.

Comments 3：

In Section 2, the discussion of non-transcriptional rhythms—particularly the persistence of rhythmicity in Bmal1-null models—should be clearly framed in terms of whether it reflects established evidence or remains hypothetical. Furthermore, the relationships among the transcriptome, proteome, and metabolome are not clearly articulated; a schematic illustration summarizing the interactions among these omics layers would be helpful. The mention of single-cell and spatial approaches is currently descriptive; a more concrete explanation of what specific insights these approaches can provide is needed.

Response:

We thank the reviewer for these three observations, each of which has been addressed in the revised manuscript.

Regarding non-transcriptional rhythms, we have reframed the discussion of Bmal1-null findings to distinguish clearly between experimental observation and mechanistic interpretation. The revised text states that the persistence of proteome and phosphoproteome rhythmicity in Bmal1-null animals has been contested on methodological grounds, that independent verification in hepatic tissue has not been reported, and that the mechanistic interpretation of existing data should be regarded as a working hypothesis rather than an established feature of liver circadian biology.

Regarding the relationships among the three molecular layers, we have added a new schematic figure, now Figure 2, that illustrates the transcriptome, proteome with post-translational modifications, and metabolome as partially autonomous oscillatory layers with temporal offsets and directional but incomplete dependencies. Key quantitative benchmarks are incorporated into the figure to ground the illustration in published data. The accompanying text has been revised to articulate explicitly that these layers are not sequential outputs of a single upstream oscillator, and that each carries rhythmic information the others cannot fully substitute.

Regarding single-cell and spatial approaches, the revised text now specifies the concrete insights these methods can provide. Single-cell profiling would resolve cell-type-specific clock gene dysregulation across hepatocytes, Kupffer cells, hepatic stellate cells, and sinusoidal endothelial cells, which bulk approaches conflate. Spatial transcriptomics would clarify whether clock gene dysregulation follows the periportal-to-pericentral gradient of hepatic injury or represents a more diffuse reorganization across the lobular architecture. These capabilities are presented as distinct from what current bulk multi-omics approaches can achieve.

Comments 4：

In Section 3, the presentation of “common patterns” does not sufficiently account for heterogeneity among different diseases. In addition, Table 1 is information-dense and difficult to interpret, particularly the “Predominant Silencing Mechanism” column. Simplification or division of the table would improve readability.

Response:

We thank the reviewer for both observations. We agree that the original framing of shared clock dysregulation features did not adequately acknowledge the substantial differences in disease biology and evidence depth that underlie any cross-disease comparison.

We have revised the introductory paragraph of the disease discussion to address this directly. The revised text acknowledges that while BMAL1 attenuation, REV-ERBα dampening, and CLOCK:BMAL1 disruption recur across disease contexts, these features represent the basis for an inferential framework rather than a uniformly supported pan-disease conclusion. The paragraph explicitly states that the depth of supporting evidence differs substantially across disease categories, and that the discussions that follow identify the evidence source and its limitations for each major conclusion. This framing is maintained consistently throughout the disease-specific discussions, where heterogeneity in upstream mechanisms, model systems, and evidence maturity is noted at the paragraph level.

Regarding Table 1, we have removed the "Predominant Silencing Mechanism" column entirely. The mechanistic content previously contained in that column has been relocated to the corresponding disease discussion paragraphs, where it can be presented with appropriate context and evidence-source qualifiers. The table now functions as a concise reference index of clock gene expression alterations by disease category, with a format that substantially reduces line wrapping and improves readability.

Comments 5：

In Section 3.1, the influence of model dependency is not sufficiently discussed, and the relationship to human data requires clarification. In the metabolomics section, the discussion of amino acid metabolism remains largely hypothetical. Moreover, the interpretation of “clock-independent” rhythms is ambiguous and not clearly distinguished from feeding-dependent rhythms. Human data are also limited.

The Gut Microbiota section is particularly hypothesis-driven, relying primarily on animal studies with substantial extrapolation to humans. Because the causal link between microbiota and liver disease progression is not firmly established, it would be more appropriate to describe microbiota as a “potential mediator” rather than implying a direct causal role.

Response:

We thank the reviewer for these detailed observations, each of which has been addressed in the revised manuscript.

Regarding model dependency and the relationship to human data, we have added explicit discussion of this limitation in the MAFLD transcriptomics and proteomics discussion. The divergent findings on BMAL1 and fibrosis severity between hepatocyte-specific knockout models and human NAFLD tissue are now presented as a dedicated interpretive paragraph, which concludes that these discrepancies most plausibly reflect differences in model design rather than fundamental interspecies biological divergence. Throughout the section, conclusions from animal models, cell lines, and human biopsy cohorts are now labelled explicitly and presented in distinct evidential layers rather than merged into unified statements.

Regarding amino acid metabolism, we have revised the relevant paragraphs to present this dimension as a mechanistically plausible but evidentially incomplete framework. The revised text clearly distinguishes between the CBS-CRY1 physical interaction, which provides direct molecular-level experimental evidence, and the broader proposition that MAFLD-associated clock disruption alters sulfur amino acid catabolism and one-carbon metabolism, which is explicitly labelled as a hypothesis awaiting time-resolved metabolomic confirmation in disease-relevant models.

Regarding clock-independent rhythms and their distinction from feeding-dependent rhythms, we have revised the relevant passage to clarify that lipid metabolomic rhythms sustained in Per1/Per2 double-knockout animals reflect feeding-fasting entrainment rather than canonical clock output. The text now explicitly distinguishes these two mechanisms and avoids using "clock-independent" as a general descriptor where "feeding-dependent" is the more precise characterization.

Regarding the gut microbiota discussion, we have reframed the language throughout to describe microbiota as a potential mediator rather than a direct causal driver of hepatic metabolic dysfunction. The fecal transplantation findings are described as suggesting a mediating role rather than establishing causation. The section now opens with an explicit statement that the causal link between microbiota rhythm disruption and liver disease progression has not been established in human studies, and that all mechanistic evidence derives from animal models. The sole available human dataset, the Yang and colleagues cross-sectional study in shift workers, is identified as providing directional support only and is clearly distinguished from the animal model evidence that constitutes the majority of this discussion.

Comments 6：

Section 3.2 relies heavily on chimeric mouse models, and treating these findings as equivalent to human data may be somewhat overstated. The discussion of epigenetic memory is also based on limited evidence; therefore, the limitations of these models should be more clearly acknowledged.

In the HCC section, the content is complex and combines transcriptional regulation, post-translational modifications, metabolism, and signaling pathways. A clearer structural organization is needed. As a possible approach, the section could be reorganized into subsections such as “transcriptional dysregulation,” “post-translational rewiring,” and “metabolic reprogramming.” Such restructuring would substantially improve clarity.

The “Immune & Cell Cycle” subsection relies heavily on extrapolation from other cancer types; therefore, these interpretations should be explicitly presented as hypotheses.

Response:

We thank the reviewer for these observations, all of which have been addressed in the revised manuscript.

Regarding the chimeric mouse model, we have added an explicit framing statement at the point where the Mukherji and colleagues dataset is first introduced. The revised text states that this model provides human hepatocyte-level data within an in vivo experimental environment, but that its findings should not be equated with data derived directly from human liver tissue, given the immunodeficient host environment and the absence of intact human immune-hepatic interactions. This caveat is maintained consistently throughout the HCV discussion. Regarding epigenetic memory, the revised text distinguishes between chimeric model findings and clinical sample observations, and explicitly notes that the direct methylation of clock gene promoter CpG islands has not been demonstrated in the cited studies.

Regarding the structural reorganization of the HCC discussion, we have implemented the reviewer's suggested framework. The HCC material is now organized into three subsections covering transcriptional dysregulation of clock genes, post-translational remodeling of clock proteins, and metabolic reprogramming through clock-associated pathways. Cell cycle checkpoint content, including the PER2-Wee1 axis, the NPAS2-CDC25A axis, and the CRY2-FBXL3-cMYC axis, has been relocated to the post-translational subsection, where it fits more naturally given the protein-level mechanisms involved.

Regarding the immune microenvironment content, this material is now presented as a standalone subsection with explicit hypothesis labeling applied throughout. The subsection opens by stating that the relationship between circadian gene dysregulation and the HCC immune microenvironment is currently supported only by correlational and cross-cancer evidence. Evidence from non-hepatic tumor systems is explicitly identified as biological context rather than direct support for HCC-specific conclusions. The proposition that clock gene downregulation promotes M2 macrophage polarization and immune evasion in HCC is characterized as a hypothesis requiring dedicated experimental investigation, with the two inferential steps underpinning it identified and evaluated separately.

Comments 7：

In Section 3.3, the specificity of REV-ERB agonists in the context of fibrosis is a concern, and potential off-target effects should be clearly emphasized. In the ALD subsection, mechanisms involving the NAD⁺ pathway and ROS are intertwined and would benefit from clearer organization. In addition, human evidence is limited to PBMC-based studies, which should be acknowledged as a limitation. The discussion of cholestatic disease is relatively sparse and unbalanced; if this topic is considered important, the section should be expanded.

Response:

We thank the reviewer for these observations, each of which has been addressed in the revised manuscript.

Regarding REV-ERB agonist specificity, we have moved the warning about SR9009 off-target effects to the opening of the pharmacological discussion rather than retaining it as a closing caveat. The revised text now states at the outset that SR9009 has documented REV-ERB-independent effects on cell viability and gene expression, as demonstrated in REV-ERBα/β double-knockout cells, and that mechanistic conclusions from SR9009 treatment studies should not be attributed solely to REV-ERBα activation. This front-loaded placement ensures that readers interpret the subsequent pharmacological findings with appropriate caution.

Regarding the NAD⁺ and ROS mechanisms in the ALD discussion, we have reorganized the relevant paragraph to present these two axes as parallel and independent contributors to clock suppression rather than as intertwined or sequential steps. Each axis is now described in a discrete evidential unit, with an explicit closing statement clarifying that the two mechanisms act in concert but should not be understood as a linear mechanistic cascade.

Regarding PBMC-based human evidence, the revised ALD discussion now explicitly labels the Huang and colleagues dataset as deriving from peripheral blood mononuclear cells rather than liver tissue, and states that PBMC measurements should be understood as an accessible surrogate rather than a substitute for direct hepatic clock gene data. This caveat is applied consistently wherever PBMC data are cited in the ALD context.

Regarding the cholestatic disease discussion, we acknowledge the reviewer's observation that this section is less extensive than others. This reflects the genuine scarcity of direct clock-focused evidence in PBC and PSC rather than a selective omission. We have nonetheless strengthened the section in two respects. We have clarified the mechanistic basis of bile acid-driven clock disruption and cytokine-mediated clock suppression as two independent pathways rather than presenting them as a single unified mechanism. We have also added a paragraph identifying three recently published single-cell and spatial transcriptomic datasets in PSC and PBC liver that offer a near-term opportunity for computational reanalysis of clock gene expression, representing the most feasible path toward generating preliminary data in this undercharacterized disease context.

Comments 8：

As a cross-cutting issue, there is insufficient distinction between established findings and hypotheses, and the lack of human data is a common limitation across sections. In addition, temporal resolution is often lacking, and there is occasional conflation of causation and correlation. To address these issues, the manuscript should explicitly indicate the level of evidence, clearly label hypotheses, avoid overstating causality, and further refine its structure.

In particular, while the manuscript correctly highlights the “lack of temporally resolved human data,” it would be valuable to elaborate on the extent of this limitation, the reasons underlying it (e.g., ethical constraints, sampling timing, and invasiveness), and potential alternative approaches such as phase inference algorithms or circulating biomarkers.

Response:

We thank the reviewer for this cross-cutting observation, which reflects a concern that runs through several of the preceding specific comments and that we have treated as a priority in the revision.

Regarding the distinction between established findings and hypotheses, we have embedded evidence-level qualifiers systematically throughout all disease discussion paragraphs. Conclusions supported by multiple independent studies are presented with direct affirmative language. Conclusions derived from single studies, animal models, or cell lines are introduced with phrases such as "animal model data suggest," "mechanistic studies in cell lines indicate," or "this finding derives from a single study and requires independent replication." Hypothesis-generating propositions, particularly in the gut microbiota, immune microenvironment, and cholestatic disease discussions, are now explicitly labeled as such at the point where they are introduced rather than at the end of the relevant passage.

Regarding the conflation of causation and correlation, we have revised passages where associative data were previously described using causal language. Correlational findings from transcriptomic cohort analyses and immune deconvolution studies are now described using relational rather than mechanistic terms.

Regarding the lack of temporally resolved human data, we have expanded the discussion of this limitation in the translational discussion. The revised text explains that this gap reflects a combination of ethical constraints on serial liver biopsies, the logistical demands of nocturnal and multi-timepoint sampling in clinical populations, and the historical absence of pre-specified circadian endpoints in liver disease trial designs. The text then identifies two methodological alternatives that do not require new tissue collection. Transcriptomic phase inference algorithms, including TimeSignature and TimeMachine, can estimate internal circadian phase from one or two peripheral blood draws and could be applied to existing biobank collections. Circulating biomarkers including plasma bile acids and PBMC clock gene expression profiles offer additional accessible readouts, provided that sampling time is standardized relative to habitual wake time. These alternatives are presented as complementary rather than equivalent to direct temporally resolved tissue data, with their respective limitations acknowledged.

Comments 9：

There is also room for improvement in the figures and tables. Table 1, while comprehensive, is difficult to interpret due to its density; dividing it by disease category or incorporating color coding may improve clarity. Figure 2 includes extensive explanatory text that obscures the central message; simplifying the figure to emphasize a single key point would enhance its impact.

Response:

We thank the reviewer for these observations. Both have been addressed in the revised manuscript.

Regarding Table 1, we have substantially reduced its density by removing the "Predominant Silencing Mechanism" column entirely, as noted in our response to Comment 4. The table now presents only the downregulated and upregulated clock components by disease category, functioning as a concise reference index rather than a mechanistic summary. This restructuring eliminates the line-wrapping problem and makes the table considerably easier to interpret at a glance.

Regarding Figure 2, this figure now appears as Figure 3 in the revised manuscript, reflecting the addition of a new schematic figure earlier in the text. The legend has been substantially shortened to focus on a single central message, namely that post-transcriptional and post-translational mechanisms dominate hepatic circadian proteome organization under both standard and environmentally disrupted conditions. Detailed technical descriptions of normalization methods, gene classification categories, and overlap statistics have been removed from the legend, leaving only the information necessary to interpret the figure directly.

Minor Comments

Comments 1：The terms NAFLD and MAFLD are used somewhat interchangeably; consistent terminology should be adopted throughout the manuscript.

Response:

We thank the reviewer for this observation. We have standardized terminology throughout the revised manuscript, adopting MAFLD and MASH as the primary terms in accordance with current nomenclature recommendations.

Comments 2：The balance of citations is uneven, with some sections relying heavily on studies from the early 2010s. Inclusion of more recent single-cell and spatial omics studies is recommended.

Response:

We thank the reviewer for this observation. We have reviewed the citation balance across all sections and removed a number of older studies where more recent references adequately represent the same evidence. The total citation count has been reduced accordingly.

Regarding single-cell and spatial omics studies specifically addressing circadian clock biology in liver disease, we conducted an additional targeted literature search but were unable to identify published studies that directly match the scope of this review. This absence is itself noted in the manuscript as a critical research gap. No published study has yet applied single-cell or spatial transcriptomic profiling to clock gene expression in any liver disease context, and we have highlighted this as a high-priority direction for future investigation. We therefore respectfully maintain the current citation structure for this topic, as adding tangentially related references would not accurately represent the state of the field.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

All the comments I indicated have been taken into account, additions have been made, and the shortcomings have been corrected.

Author Response

Thank you for your careful re-evaluation and positive feedback. We are pleased that the revisions have addressed your concerns satisfactorily, and we appreciate the time and effort you devoted to reviewing our manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

This revised manuscript addresses the issues raised in the previous peer review with great care and sincerity. Structural organization and readability have been improved through the restructuring of the HCC section, simplification of Table 1, and the addition of a multi-omics conceptual framework figure. In particular, redefining the “common molecular phenotype” as an “inferential framework” rather than an established common pathological entity represents an appropriate and intellectually balanced revision in light of the current level of evidence.

The following are minor comments:

In the Conclusion section, it would be helpful to state more explicitly that these features represent “partially overlapping recurring motifs” across heterogeneous disease contexts.
Although the discussion of microbiota-derived rhythmic signaling has been substantially improved, the abundance of mechanistic detail may still lead some readers to overestimate the maturity of the evidence. It would therefore be helpful to briefly reiterate at the beginning of the relevant section that the current evidence base is derived primarily from animal models and preclinical studies.
Overall, many sentences remain quite long and contain multiple mechanistic clauses. Further streamlining of sentence structure and length would improve readability.

Author Response

Response：

We thank the reviewer for the positive assessment of our revised manuscript and for the additional constructive comments. We are pleased that the restructuring of the HCC discussion, the simplification of Table 1, the addition of the multi-omics conceptual framework figure, and the repositioning of the four molecular features as an inferential framework were well received.

We have addressed all three minor comments in the current revision. All changes are indicated in blue text in the revised manuscript. The point-by-point responses below address each comment in turn.

Comment 1：

In the Conclusion section, it would be helpful to state more explicitly that these features represent "partially overlapping recurring motifs" across heterogeneous disease contexts.

Response：

We thank the reviewer for this precise and constructive suggestion. We agree that the phrase "inferential framework" alone does not fully convey the heterogeneous and partially overlapping nature of these features across disease contexts.

We have revised the relevant sentences in the Conclusions to read as follows.

"BMAL1 functional downregulation, REV-ERBα output attenuation, NAD⁺ amplitude reduction, and gut-liver axis desynchronization represent partially overlapping recurring motifs rather than a uniform mechanistic signature. Together, they form a proposed inferential framework for hepatic circadian failure."

Comment 2：

Although the discussion of microbiota-derived rhythmic signaling has been substantially improved, the abundance of mechanistic detail may still lead some readers to overestimate the maturity of the evidence. It would be helpful to briefly reiterate at the beginning of the relevant section that the current evidence base is derived primarily from animal models and preclinical studies.

Response：

We thank the reviewer for this observation. We agree that front-loading the evidential caveat more explicitly would better orient readers before they encounter the mechanistic detail that follows.

We have added two sentences at the opening of the gut microbiota discussion to address this directly. The revised section now begins as follows.

"The evidence reviewed in this section derives almost entirely from animal models and preclinical studies. The causal link between microbiota rhythm disruption and liver disease progression has not been established in human studies, and findings should be interpreted within that constraint."

Comment 3：

Overall, many sentences remain quite long and contain multiple mechanistic clauses. Further streamlining of sentence structure and length would improve readability.

Response：

We thank the reviewer for this observation. We have carefully reviewed the entire manuscript with sentence length as a specific editing criterion. All sentences exceeding 30 words have been identified and revised. In each case, compound sentences containing multiple mechanistic clauses were divided into shorter, discrete statements. This approach preserves the mechanistic content and logical progression of each argument while improving overall readability.