Network Topology and Interactomic Analysis Reveal the Regulatory Framework of the Humanin Protein Family (MTRNR2Lx Class)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

In this study, the authors build an interactome of 1,033 nodes. They analyze the topology, reliability, and functional implications of the model, and describe the probable function of the small mitochondrial proteins of the Humanin family (MTRNR2Lx). The authors produced a robust model, with approximately 70% of the interactions supported by experimental evidence. Mitochondrial proteins from the Humanin family were located at the periphery of the interactome, functioning as sensors and mediators of stress signals. Mitochondrial humanins prevent acute damage and cooperate with apoptosis-inhibiting mechanisms, and nuclear humanins (MTRNR2Lx) contribute to regulatory responses in neuroprotection and apoptosis via the interaction with GNB1. One important conclusion is that humanins operate as network modulators instead of agents of intrinsic functions.

The manuscript is well written, in good English. I have three minor points:

Line 188: It is mentioned that “Human PPIs are tiny mitochondrial proteins distributed throughout the cell…”, but PPI is an abbreviation for protein-protein interaction, and PPIs must be for protein-protein interactions, regardless of the size of the proteins that participate in this interaction. Certainly, small mitochondrial proteins participate in protein-protein interactions, but in general, protein-protein interactions occur with all kinds of proteins, cytosolic, mitochondrial, nuclear, independently of the size of the protein.

Lines 853, 861, and 862: It seems that the references were missing.

Final point: The figures are very low resolution. In some of them, it is difficult to read the text. Is it possible to increase their resolution?

Author Response

We sincerely thank the Academic Editor and the three Reviewers for their time, careful evaluation, and constructive comments. The manuscript has been substantially revised in response to the points raised. In particular, we corrected terminology, added missing references, improved the quality of the figures, revised selected sections to reduce overinterpretation, and performed minor English editing throughout the manuscript where necessary. Below, we provide a point-by-point response.

Reviewer 1

Comment 1: Line 188: It is mentioned that “Human PPIs are tiny mitochondrial proteins distributed throughout the cell…”, but PPI is an abbreviation for protein-protein interaction, and PPIs must be for protein-protein interactions, regardless of the size of the proteins that participate in this interaction. Certainly, small mitochondrial proteins participate in protein-protein interactions, but in general, protein-protein interactions occur with all kinds of proteins, cytosolic, mitochondrial, nuclear, independently of the size of the protein.

Response 1: We thank the Reviewer for pointing out this terminological error. We agree that “PPI” must be used exclusively to refer to protein–protein interactions. We have corrected the sentence to clarify that the text refers to Humanin-family micropeptides and their nuclear-encoded paralogs, rather than to PPIs.

Changes to the manuscript:

The sentence has been revised as follows:

“Human micropeptides and their nuclear-encoded paralogs are tiny mitochondrial proteins distributed throughout the cell and extracellular space, but the molecular mechanisms underlying their activity are poorly understood.”

Comment 2: Lines 853, 861, and 862: It seems that the references were missing.

Response 2: We apologize for this oversight. The missing reference placeholders have been replaced with the appropriate bibliographic citations in the revised manuscript, and the corresponding references have been added to the reference list.

Comment 3: Final point: The figures are very low resolution. In some of them, it is difficult to read the text. Is it possible to increase their resolution?

Response 3: We agree with the Reviewer and thank them for raising this point. We have improved the quality and readability of the figures throughout the manuscript. The network figures and plots have been re-exported at higher resolution, and vector versions have also been generated where possible to preserve the clarity of node labels, axes, legends, and network structures. The revised high-resolution figure files have been uploaded separately for production.

Reviewer 2 Report

Comments and Suggestions for Authors

I like this manuscript, but I think that the authors need to:

An additional layer of regulation may involve Mitofusin-2 (MFN2), a mitochondrial outer membrane GTPase known to coordinate mitochondrial fusion, mitochondria–ER contact sites (MAMs), calcium signaling, and stress-adaptive mitophagy (Chen et al., 2003; de Brito and Scorrano, 2008; Filadi et al., 2018). Within the proposed interactome framework, MFN2 could function as a structural and metabolic relay linking peripheral Humanin signaling to mitochondrial network remodeling and stress buffering.

In this context, MFN2 may occupy an intermediate position between the MTRNR2Lx–GNB1/FPR2 signaling module and TP53-mediated transcriptional adaptation. Humanin signaling through G proteins and PI3K/AKT pathways could indirectly stabilize or modulate MFN2 activity, thereby influencing mitochondrial connectivity and preserving oxidative phosphorylation during chronic non-acute stress. Because MFN2 is strongly associated with mitochondrial quality control and resistance to metabolic dysfunction (Chen et al., 2007; Sebastián et al., 2016), its participation would fit the observed shift from acute anti-apoptotic defense toward long-term resilience programming.

Mechanistically, a plausible sequence may emerge:

1. Peripheral MTRNR2Lx variants activate FPR2/GNB1 signaling.
2. PI3K/AKT survival cascades reduce excessive mitochondrial fragmentation and favor MFN2-dependent mitochondrial fusion.
3. MFN2 stabilizes mitochondrial bioenergetics and ER–mitochondrial communication, reducing abrupt cytochrome-c release and buffering ROS propagation.
4. The attenuated stress signal reaches TP53 in a regulated manner, favoring senescence-associated adaptation rather than irreversible apoptosis.
5. TP53 subsequently coordinates chromatin remodeling through KMT2A/WDR5 and KAT6A pathways.

In this model, MFN2 would not merely act as a mitochondrial fusion protein, but as a dynamic “stress rheostat” capable of translating Humanin-mediated metabolic signals into structural mitochondrial resilience. This interpretation is consistent with evidence linking MFN2 to neuroprotection, mitochondrial homeostasis, and selective mitophagy under chronic stress conditions (Chen and Chan, 2009; Fang et al., 2019).

Importantly, MFN2 may also provide a mechanistic explanation for the transition between mit-HN and nuclear MTRNR2Lx signaling. Whereas mit-HN directly neutralizes BAX through stoichiometric sequestration, the MTRNR2Lx/FPR2/GNB1 axis could preserve mitochondrial integrity indirectly by maintaining MFN2-mediated network plasticity. In this view, the protective effect is no longer based on consuming anti-apoptotic peptide reserves, but on sustaining mitochondrial adaptability and energetic coherence across the cellular network.

A further speculative possibility is that mitofusin agonists may amplify this adaptive branch of the network. Pharmacological activation of MFN2 could synergize with MTRNR2Lx signaling by enhancing mitochondrial fusion and restoring mitochondria–ER coupling, thereby increasing the threshold required for TP53-driven apoptotic commitment. Small-molecule mitofusin agonists have already shown the capacity to reverse mitochondrial fragmentation and improve neuronal survival in models of Charcot–Marie–Tooth disease and neurodegeneration (Rocha et al., 2018; Dang et al., 2021). Under chronic neurodegenerative or metabolic stress, Humanin signaling and MFN2 agonism may therefore converge on a common systems-level outcome: stabilization of mitochondrial topology, suppression of catastrophic fragmentation, and promotion of a senescence-like survival phenotype instead of cell death.

This hypothesis would also align with the observed systems-biology architecture of the interactome, where peripheral nodes modulate activation thresholds rather than acting as direct executors. In such a framework, MFN2 could represent a structural effector downstream of Humanin signaling, integrating metabolic sensing with mitochondrial morphology and cellular fate decisions.

Author Response

We sincerely thank the Academic Editor and the three Reviewers for their time, insights, and constructive comments. The manuscript has been significantly revised, speculative narratives removed, and our biological discussion expanded in careful response to all points raised. We have also done some English editing here and there, when necessary. Below, we provide detailed answers to each point.

Reviewer 2:

Comments 1: I like this manuscript, but I think that the authors need to: An additional layer of regulation may involve Mitofusin-2 (MFN2), a mitochondrial outer membrane GTPase known to coordinate mitochondrial fusion, mitochondria–ER contact sites (MAMs), calcium signaling, and stress-adaptive mitophagy (Chen et al., 2003; de Brito and Scorrano, 2008; Filadi et al., 2018). Within the proposed interactome framework, MFN2 could function as a structural and metabolic relay linking peripheral Humanin signaling to mitochondrial network remodeling and stress buffering.

In this context, MFN2 may occupy an intermediate position between the MTRNR2Lx–GNB1/FPR2 signaling module and TP53-mediated transcriptional adaptation. Humanin signaling through G proteins and PI3K/AKT pathways could indirectly stabilize or modulate MFN2 activity, thereby influencing mitochondrial connectivity and preserving oxidative phosphorylation during chronic non-acute stress. Because MFN2 is strongly associated with mitochondrial quality control and resistance to metabolic dysfunction (Chen et al., 2007; Sebastián et al., 2016), its participation would fit the observed shift from acute anti-apoptotic defense toward long-term resilience programming.

Mechanistically, a plausible sequence may emerge:

1. Peripheral MTRNR2Lx variants activate FPR2/GNB1 signaling.
2. PI3K/AKT survival cascades reduce excessive mitochondrial fragmentation and favor MFN2-dependent mitochondrial fusion.
3. MFN2 stabilizes mitochondrial bioenergetics and ER–mitochondrial communication, reducing abrupt cytochrome-c release and buffering ROS propagation.
4. The attenuated stress signal reaches TP53 in a regulated manner, favoring senescence-associated adaptation rather than irreversible apoptosis.
5. TP53 subsequently coordinates chromatin remodeling through KMT2A/WDR5 and KAT6A pathways.

In this model, MFN2 would not merely act as a mitochondrial fusion protein, but as a dynamic “stress rheostat” capable of translating Humanin-mediated metabolic signals into structural mitochondrial resilience. This interpretation is consistent with evidence linking MFN2 to neuroprotection, mitochondrial homeostasis, and selective mitophagy under chronic stress conditions (Chen and Chan, 2009; Fang et al., 2019).

Importantly, MFN2 may also provide a mechanistic explanation for the transition between mit-HN and nuclear MTRNR2Lx signaling. Whereas mit-HN directly neutralizes BAX through stoichiometric sequestration, the MTRNR2Lx/FPR2/GNB1 axis could preserve mitochondrial integrity indirectly by maintaining MFN2-mediated network plasticity. In this view, the protective effect is no longer based on consuming anti-apoptotic peptide reserves, but on sustaining mitochondrial adaptability and energetic coherence across the cellular network.

A further speculative possibility is that mitofusin agonists may amplify this adaptive branch of the network. Pharmacological activation of MFN2 could synergize with MTRNR2Lx signaling by enhancing mitochondrial fusion and restoring mitochondria–ER coupling, thereby increasing the threshold required for TP53-driven apoptotic commitment. Small-molecule mitofusin agonists have already shown the capacity to reverse mitochondrial fragmentation and improve neuronal survival in models of Charcot–Marie–Tooth disease and neurodegeneration (Rocha et al., 2018; Dang et al., 2021). Under chronic neurodegenerative or metabolic stress, Humanin signaling and MFN2 agonism may therefore converge on a common systems-level outcome: stabilization of mitochondrial topology, suppression of catastrophic fragmentation, and promotion of a senescence-like survival phenotype instead of cell death.

This hypothesis would also align with the observed systems-biology architecture of the interactome, where peripheral nodes modulate activation thresholds rather than acting as direct executors. In such a framework, MFN2 could represent a structural effector downstream of Humanin signaling, integrating metabolic sensing with mitochondrial morphology and cellular fate decisions.

Response 1:

We sincerely thank the Reviewer for this insightful and constructive suggestion. We agree that MFN2 is highly relevant to the proposed interactomic framework, given its established role in mitochondrial fusion, mitochondria–ER contact sites, MAMs, calcium signaling, mitochondrial quality control, and stress adaptation.

In response to this comment, we expanded the Discussion by adding a dedicated paragraph on MFN2 as a candidate mitochondrial relay within the Humanin-related network. In the revised manuscript, MFN2 is discussed as a possible interface linking the MTRNR2Lx–FPR2/GNB1-associated module with mitochondrial quality-control and apoptosis-associated nodes, including PRKN, VDAC1, and BAX. We also included the concept that MFN2-associated mitochondrial pathways may provide an intermediate network context between peripheral Humanin-related signaling and downstream stress-response modules, including TP53-associated pathways.

Changes to the manuscript:

We have added a dedicated paragraph in the Discussion section (Section 3.11.1, 3.11.1. MFN2 as an indirect stress rheostat. We also added the relevant references suggested by the Reviewer, including studies on MFN2-mediated mitochondrial fusion, mitochondria–ER contact sites/MAMs, mitochondrial quality control, mitophagy, neuroprotection, and mitofusin agonists.

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript presents an interactome-based analysis of Humanin-family peptides (MTRNR2Lx) using different PPI databases, STRING network expansion, and graph-theoretical analyses. Overall, the study is interesting and challenges some prevailing assumptions in the field of Humanin. However, the major limitation of the study is that most biological conclusions are derived from network topology rather than direct experimental evidence. The biological conclusions go beyond the evidence presented, as the manuscript is mostly computational and interpretative, and it lacks direct data to support the proposed biological mechanisms.

1. The title overstates the level of experimental support. The current study is primarily a computational, interactome-based analysis and does not directly establish “evidence-based functional roles”. Please revise the title to better reflect the hypothesis-generating nature of the work. Also, the conclusion appears too strong. Please revise the abstract and discussion sections accordingly to better reflect this bioinformatic work.
2. The manuscript is very verbose. A substantial portion of the content consists of extended interpretations of network topology, conceptual analogies, and speculative biological narratives that are not directly supported by the analyses presented. Many sections repeatedly reinterpret the same network parameters (e.g., diameter, density, eccentricity, centralization, clustering coefficient) using different conceptual frameworks without providing additional evidence or biological insight. Please reduce the manuscript's overall length, particularly the Results and Interpretation sections, by at least 30–40%. A target length of approximately 800 lines or fewer would significantly improve readability, clarity, and scientific focus.
3. Extensive biological conclusions are derived from network topology alone. It is inappropriate to repeatedly interpret network topology as a biological mechanism. For example, the authors described “peripheral nodes” as sensors, regulators, or modulators, but “core nodes” are interpreted as controllers or decision-making centers. While these interpretations are interesting hypotheses, network topology alone cannot be used to establish signaling directions, regulatory hierarchy, causal relationships, or biochemical mechanisms. Protein interaction networks are largely undirected representations of associations. Therefore, many of the mechanistic conclusions presented in the manuscript extend far beyond what the data directly support.
4. Lack of experimental validation. The manuscript proposes several specific mechanistic models, including Humanins acting as peripheral stress sensors, MTRNR2L8 and MTRNR2L12 functioning via G protein signaling, and GNB1 serving as a major functional mediator. However, none of these hypotheses is experimentally tested. The manuscript would be substantially strengthened by including experimental validation. Otherwise, without solid evidence, these models should be presented as hypotheses rather than functional conclusions.
5. For the 42 interactions listed in Table 1, please provide 1) the original supporting references, 2) experimental methods supporting the interaction, and 3) the validation is confirmed in vivo or in vitro. A detailed evidence table should be supplemented.
6. Table 11 and the associated discussion are highly speculative. The analysis moves from enrichment results and network topology to detailed mechanistic models such as TP53 regulation, GNB1 signaling, and nuclear transport machinery. These models are biologically interesting, but the analyses presented do not directly demonstrate these relationships. At multiple points, the authors convert hypothetical interaction networks into functional regulatory pathways without experimental validation.
7. Figure quality and presentation should be significantly improved, particularly Figure 6.

 

Typos and unprofessional words need to be carefully corrected, including but not limited to:

• Please correct the “Excel file_1, Excel file_2, etc.” to “Supplementary Table S1, Supplementary Dataset S1, or Supplementary File S1”.
• Line 188, “Human” to “Humanin”
• Line 345, what is the “interaction cost”

Author Response

We sincerely thank the Academic Editor and the three Reviewers for their time, insights, and constructive comments. The manuscript has been significantly revised, speculative narratives removed, and our biological discussion expanded in careful response to all points raised. We have also done some English editing here and there, when necessary. Below, we provide detailed answers to each point.

Reviewer 3

General comment: This manuscript presents an interactome-based analysis of Humanin-family peptides (MTRNR2Lx) using different PPI databases, STRING network expansion, and graph-theoretical analyses. Overall, the study is interesting and challenges some prevailing assumptions in the field of Humanin. However, the major limitation of the study is that most biological conclusions are derived from network topology rather than direct experimental evidence. The biological conclusions go beyond the evidence presented, as the manuscript is mostly computational and interpretative, and it lacks direct data to support the proposed biological mechanisms.

Response: We thank the reviewer for this conceptual comment. We fully agree that in classical molecular biology, direct laboratory (wet-lab) validation is necessary to demonstrate a linear pathway. This is still true today, but computational interactomics is a well-established independent discipline that maps the deep metabolic space of physically possible interactions, as in our case. We would respectfully like to clarify the nature and robustness of our systems biology model:

1. The model is based on robust experimental data: Interactome-1033 is not a speculative network based on computational test mining or co-expression algorithms. By applying rigorous filtering protocols, 69.23% (approximately 70%) of interactions in our final network are supported by high-confidence evidence from the independent literature curated by BioGRID, MINT, and IntAct. So, our work does not involve new, unverified connections; it contextualizes pre-existing, fragmented experimental evidence within a global framework. Network functional characteristics are unpredictable and "emergent" properties that depend on the physical organization of the network itself. Initial interactors and their interactions are sometimes eliminated as a direct consequence of the ranking calculation performed by the enrichment algorithm (shell expansion and hub competitors) and of configuration parameters, including the exclusion of text mining, which is one of the most critical factors. STRING has a strict rule for network enrichment: if a node in the initial set, after recalculating scores without text mining or due to the introduction of stronger competitor nodes, loses connections to other nodes exceeding the 0.700 threshold, it is demoted or shows no edge. Their direct edge is removed if it does not satisfy the new stringent constraints of the recalculated global topology. Therefore, the network's functional properties faithfully reflect the percentage of reliable interactions actually present in the expanded network, which, in our case, is very high. Finally, biological causality does not always require a very high-affinity physical bond but can also be achieved through brief, momentary interactions (that we are not yet able to define today), or indirectly through regulatory molecules that act as a "shuttle" for information, as in our case (G-proteins).
2. Network topology as a biological constraint. In the science of complex systems, topological parameters (such as centrality, local clustering, and eccentricity) are mathematical reflections of real biological boundaries. The peripheral position of the MTRNR2Lx cluster and its centrality with respect to the GNB1 bottleneck are physical properties of the human interactome that define the possible physical solutions for the interaction paths. So, while we agree that undirected networks cannot establish precise temporal causality, they strictly map and define the spatial and physical possibility of interaction pathways within the cell. Obviously, they need to be validated, but that's a matter of laboratory biochemistry. Our role is to point the soundest way forward. Humanins have been known since 2001, and so far, we are unaware of any laboratory researchers who have advanced comprehensive mechanistic hypotheses about their functions.

Comment 1: The title overstates the level of experimental support. The current study is primarily a computational, interactome-based analysis and does not directly establish “evidence-based functional roles”. Please revise the title to better reflect the hypothesis-generating nature of the work. Also, the conclusion appears too strong. Please revise the abstract and discussion sections accordingly to better reflect this bioinformatic work.

Response 1: We have revised the Title, Abstract, Results, and Discussion sections. To avoid ambiguity, we have added two periods, one at the end of the abstract and the other at the end of the conclusions, which unequivocally show the nature of our results and future approaches. In addition, we shifted the tone from absolute conclusions to a narrative that generates hypotheses, replacing definitive verbs with appropriate probabilistic terms ("strongly suggests", "provides a mechanistic model for", "defines the topological framework"). Even the title has been revised to: "Network topology and interactomic analysis reveal the regulatory framework of the Humanins protein family".

Comment 2: The manuscript is very verbose. A substantial portion of the content consists of extended interpretations of network topology, conceptual analogies, and speculative biological narratives that are not directly supported by the analyses presented. Many sections repeatedly reinterpret the same network parameters (e.g., diameter, density, eccentricity, centralization, clustering coefficient) using different conceptual frameworks without providing additional evidence or biological insight. Please reduce the manuscript's overall length, particularly the Results and Interpretation sections, by at least 30–40%. A target length of approximately 800 lines or fewer would significantly improve readability, clarity, and scientific focus.

Response 2: We considered the reviewer's criticisms. We examined the manuscript to remove lengthy conceptual analogies, repetitive interpretations of network parameters, and broad biological narratives. The Results and Discussion sections have been simplified, whenever possible, focusing on synthetic mathematical parameters and biological contexts, thereby reducing the manuscript length and sharpening the scientific focus. The reduced text is in blue.

However, we also respectfully point out that the results are complex and presented in a non-standard manner, requiring careful evaluation, particularly topological, with extensive analyses using a combination of numerous parameters. Hence, the need to be verbose in explaining the meaning of the parameters used, even to non-experts. Despite being in 2026, the biomedical science world still abounds with reductionist researchers with an observational mindset. We strive to make ourselves understood, especially by them.

Comment 3: Extensive biological conclusions are derived from network topology alone. It is inappropriate to repeatedly interpret network topology as a biological mechanism. For example, the authors described “peripheral nodes” as sensors, regulators, or modulators, but “core nodes” are interpreted as controllers or decision-making centers. While these interpretations are interesting hypotheses, network topology alone cannot be used to establish signaling directions, regulatory hierarchy, causal relationships, or biochemical mechanisms. Protein interaction networks are largely undirected representations of associations. Therefore, many of the mechanistic conclusions presented in the manuscript extend far beyond what the data directly support.

Response 3: We have revised the manuscript to avoid interpreting network topology as direct biological mechanism. Terms implying causal hierarchy or functional proof, such as “controllers,” “decision-making centers,” “frontier sensors,” “command center,” and similar expressions, were removed or replaced with more appropriate network-based terminology.

Comment 4: Lack of experimental validation. The manuscript proposes several specific mechanistic models, including Humanins acting as peripheral stress sensors, MTRNR2L8 and MTRNR2L12 functioning via G protein signaling, and GNB1 serving as a major functional mediator. However, none of these hypotheses is experimentally tested. The manuscript would be substantially strengthened by including experimental validation. Otherwise, without solid evidence, these models should be presented as hypotheses rather than functional conclusions.

Response 4: We thank the Reviewer for this important point. We agree that the proposed biological relationships require direct experimental validation. Because the present study is computational and interactome-based, we have revised the manuscript to present the MTRNR2Lx–FPR2/GNB1 module, the involvement of MFN2-associated mitochondrial pathways, and the TP53-associated stress-response modules as testable hypotheses rather than functional conclusions.

We also added statements emphasizing that future experimental studies should test these candidate relationships using appropriate cellular models, receptor perturbation, GNB1 modulation, MFN2/PRKN pathway analysis, mitochondrial readouts, apoptosis assays, and stress-response models.

Comment 5: For the 42 interactions listed in Table 1, please provide 1) the original supporting references, 2) experimental methods supporting the interaction, and 3) the validation is confirmed in vivo or in vitro. A detailed evidence table should be supplemented.

Response 5: We thank the Reviewer for this valuable suggestion and fully understand the request for maximum transparency. Although these interaction data are publicly available and traceable in BioGRID and other curated repositories, we agree that a consolidated summary within the manuscript package improves immediate readability and allows the supporting evidence for the seed interactors to be evaluated more easily.

We have therefore added a new Supplementary Dataset S1, reporting the supporting evidence for the 42 Humanin/Humanin-like seed interactors listed in Table 1. This dataset includes, when available, the original supporting references, PMIDs, database source, detection method or evidence type, and validation context, including whether the evidence derives from in vivo, cell-based, or in vitro studies.

Comment 6: Table 11 and the associated discussion are highly speculative. The analysis moves from enrichment results and network topology to detailed mechanistic models such as TP53 regulation, GNB1 signaling, and nuclear transport machinery. These models are biologically interesting, but the analyses presented do not directly demonstrate these relationships. At multiple points, the authors convert hypothetical interaction networks into functional regulatory pathways without experimental validation.

Response 6: We have revised the discussion associated with Table 11 to reduce mechanistic overinterpretation. In the revised version, TP53-, GNB1-, MFN2-, and chromatin-associated modules are no longer presented as experimentally demonstrated regulatory pathways. Instead, they are described as candidate network contexts emerging from enrichment and topology analyses.

Comment 7: Figure quality and presentation should be significantly improved, particularly Figure 6.

Repsonse 7: We have improved the quality and readability of the figures throughout the manuscript, with particular attention to Figure 6. The network figures and centrality plots were re-exported at higher resolution, and vector files were generated where possible to preserve the clarity of node labels, axes, legends, and network structures. We also revised the corresponding figure legends to improve clarity.

Other comments: Typos and unprofessional words need to be carefully corrected, including but not limited to:

• Please correct the “Excel file_1, Excel file_2, etc.” to “Supplementary Table S1, Supplementary Dataset S1, or Supplementary File S1”.
• Line 188, “Human” to “Humanin”
• Line 345, what is the “interaction cost”

Response: We thank the Reviewer for these corrections. We carefully revised the manuscript to correct typographical errors, non-standard terminology, and unclear wording.

All previous references to “Excel file_1,” “Excel file_2,” and related terms were replaced with standard supplementary-file nomenclature, including Supplementary Dataset S1–S4, Supplementary Table S1–S5, and Supplementary Figure S1–S15, as appropriate.

The terminology error involving “Human”/“Humanin” was corrected

Regarding the term “interaction cost” (former line 345), we thank the Reviewer for pointing out that this expression could be unclear.. We realize that in traditional molecular biology, the word "cost" evokes consumption of ATP, metabolic energy, or protein synthesis, whereas here it is used in a topological and structural sense, according to graph theory and systems biology. In network topology, the "cost of interaction" refers to the structural and evolutionary constraints associated with network density. A highly dense network implies that a large number of distinct protein-protein interactions must be maintained, physically and temporally, within the crowded cellular environment, thereby requiring high molecular coordination. In short, saying that the interactome has a low "interaction cost" means that the cell optimizes resources through very fast, globally connected (small-world) communication, keeping the network very sparse (low density) without forcing each protein to interact with the rest of the proteome. Interactome-1033 shows that it has minimized this topological cost (density of 0.056) while maximizing communication efficiency (path length of 2.91). It is evident that these values are not imposed but derive from the intrinsic organization of the interactome, which is grounded in its physical foundations. However, to avoid confusion with "thermodynamic" costs, we have revised the sentence:

"Thus, network density reflects structural constraints and interaction coordination requirements (topological cost), while distance reflects communication capability."

Action taken in the manuscript: Below are the sentences added at the end of the Abstract:

"Overall, the methodological approach, results, and proposed model provide new insights into the systems-level organization of Humanin biology and identify prioritized molecular candidates for future in vitro and in vivo validation in the context of neurodegeneration, apoptosis, and cellular stress."

And at the end of the Conclusions:

"Operating via the FPR2/GNB1/MFN2 axis and catalytic PI3K/Akt signaling cascades, this network property sustains active resilience during chronic or neurodegenerative stress, establishing clear, testable molecular hypotheses for future in vitro and in vivo validation."