An Acetyl-CoA-Gated Metabolic Checkpoint Links Precursor Supply to Cordycepin Biosynthesis in Cordyceps militaris

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The authors describe interesting work on the regulation of fungal metabolite productivity using cordycepin in Cordyceps militaris strains as an example. The authors discovered a conserved metabolic checkpoint controlled by acetyl-CoA that regulates the tradeoff between growth and production in fungi and identified a new O-methyltransferase whose activity was enhanced by lysine acetylation associated with excess acetyl-CoA. It was shown that excess acetyl-CoA in a highly productive strain is preferentially directed toward the cordycepin biosynthetic pathway due to a switch in post-translational modification. Flux allocation controlled by post-translational modification is proposed as a general principle for rationally reprogramming metabolic priorities in fungal cell factories. The article is well written, and the experiments appear reliable. The work will be of interest to chemists and biochemists studying the metabolism of microorganisms.

Author Response

3. Point-by-point response to Comments and Suggestions for Authors

Comments : The authors describe interesting work on the regulation of fungal metabolite productivity using cordycepin in Cordyceps militaris strains as an example. The authors discovered a conserved metabolic checkpoint controlled by acetyl-CoA that regulates the tradeoff between growth and production in fungi and identified a new O-methyltransferase whose activity was enhanced by lysine acetylation associated with excess acetyl-CoA. It was shown that excess acetyl-CoA in a highly productive strain is preferentially directed toward the cordycepin biosynthetic pathway due to a switch in post-translational modification. Flux allocation controlled by post-translational modification is proposed as a general principle for rationally reprogramming metabolic priorities in fungal cell factories. The article is well written, and the experiments appear reliable. The work will be of interest to chemists and biochemists studying the metabolism of microorganisms.

Response: Thank you for these highly positive and constructive comments. We greatly appreciate your recognition of the scientific significance and rigor of our work. As suggested, we have strengthened the Introduction by:

(1)Adding a concise paragraph that emphasizes acetyl-CoA’s dual role as a central metabolic hub and signaling molecule;

(2)Integrating key recent reviews on PTM-mediated metabolic regulation (e.g., Etier et al., 2022; Xu et al., 2025);

(3)Streamlining the narrative to sharpen the knowledge gap and rationale.

These revisions improve clarity and contextual impact without altering findings. The changes appear on Lines 43–48 in the revised manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript investigates cordycepin overproduction in the high-yield Cordyceps militaris strain GYS60 and the low-yield strain GYS80 through comparative proteomics, PTM mapping, and metabolomics. Zhang and colleagues identified an acetyl-CoA–gated checkpoint centered on the O-methyltransferase CCM_06472, whose activity is activated by Lys123 acetylation and inhibited by Ser34 phosphorylation. Consistent with this new finding, they showed that a single K123Q acetylation-mimetic mutation boosted cordycepin titers by more than fourfold in wild-type strains. Overall, this study provides mechanistic insight into an acetyl-CoA-gated checkpoint and reveals PTM-gated flux allocation as a key regulatory mechanism, suggesting a minimal-intervention strategy for engineering fungal cell factories. The research design is comprehensive and offers interesting and valuable findings.

Additionally, some minor points should be addressed, as outlined below:

 

 

1, For Figure 3A, what do lanes 1, 2, 3, and 4 represent? Do they correspond to the replicates, or to different experimental conditions? Please clarify this either in the main text or in the figure legend.

 

2, For Figure 4G, it would be helpful to zoom in and show the stick representation of the key residues K123 and S34. In addition, a structural comparison of the wild-type predicted structure with the mutants, or the inclusion of molecular dynamics (MD) simulation results, would make the analysis more comprehensive. AlphaFold2 should be cited appropriately. MD simulations are described in the main text, but no corresponding results are shown in the figures; please include representative MD results and provide a more detailed description of the MD methodology in the Methods section. The software used for structural visualization should also be specified. Finally, please clarify whether the different colors of the cartoon ribbon indicate prediction confidence. If so, this should be explicitly labeled in the figure or legend.

 

3, In the legend for Figure 4E (line 286), the text states: “(E) Site-directed mutagenesis validation showed that K123A (mimicking acetylation) activated, whereas S34D (mimicking phosphorylation) inhibited enzyme activity, confirming that Lys123 and Ser34 are critical PTM sites.” However, K123A does not mimic acetylation; K123Q is the acetylation-mimetic mutation. This appears to be a typographical error. Please correct this description accordingly in the legend (and ensure consistency throughout the manuscript).

 

4, For Figure 4F, it would be helpful to include the sequence alignment for the GYS80 strain as well, which would make the comparison more comprehensive. In addition, apart from the highlighted comparison region, are the remaining regions of the sequence identical between the strains? Please clarify this point and discuss any similarities or differences in the text or legend.

 

5, Line 266: “Sequence alignment (Figure 4F) demonstrated that GYS60 retains the Ser34 and Lys123 residues, as well as the conserved SAM-binding motif (GFGTGH), confirming that enhanced enzymatic activity is driven by post-translational modifications (PTMs) rather than structural mutations.” Here, it is unclear what is meant by “structural mutations.” Does this refer only to amino acid substitutions detectable by sequence alignment, or to predicted conformational/structural changes in the protein (e.g., based on modeling or MD simulations)? Please clarify the intended meaning and, if applicable, rephrase this sentence for precision.

 

6, Line 387: For the Western blot (WB) experiments, the antibodies used should be clearly listed in the Methods section, so that the experiments can be reliably reproduced.

 

7, In Section 2.7 (Enzyme Activity Assays), the authors use crude extracts to assess the activities of malate dehydrogenase (MDH, CCM_04892) and O-methyltransferase (CCM_06472). However, when working with crude extracts, other endogenous enzymes with similar or overlapping activities may contribute to the measured signals. Please explain in more detail how the assay was designed to ensure that the observed activities can be specifically attributed to MDH (CCM_04892) and O-methyltransferase (CCM_06472). A clearer description of the experimental design and its specificity would strengthen the conclusions drawn from these assays.

Author Response

3. Point-by-point response to Comments and Suggestions for Authors Comments  The manuscript investigates cordycepin overproduction in the high-yield Cordyceps militaris strain GYS60 and the low-yield strain GYS80 through comparative proteomics, PTM mapping, and metabolomics. Zhang and colleagues identified an acetyl-CoA–gated checkpoint centered on the O-methyltransferase CCM_06472, whose activity is activated by Lys123 acetylation and inhibited by Ser34 phosphorylation. Consistent with this new finding, they showed that a single K123Q acetylation-mimetic mutation boosted cordycepin titers by more than fourfold in wild-type strains. Overall, this study provides mechanistic insight into an acetyl-CoA-gated checkpoint and reveals PTM-gated flux allocation as a key regulatory mechanism, suggesting a minimal-intervention strategy for engineering fungal cell factories. The research design is comprehensive and offers interesting and valuable findings. Additionally, some minor points should be addressed, as outlined below. Comments 1: For Figure 3A, what do lanes 1, 2, 3, and 4 represent? Do they correspond to the replicates, or to different experimental conditions? Please clarify this either in the main text or in the figure legend. Response 1: Thank you for this helpful suggestion. We have clarified the representation of lanes in Figure 3A,which is revised as Figure 5A. Specifically, we have added the following sentence to “2.5. Data Integration, Statistical Analysis” for all experiments: " All quantitative experiments were performed with four independent biological replicates unless stated otherwise. Data are shown as mean ± SD, with CV < 15% con-sidered acceptable. Student’s t-test was used for two-group comparisons, and Pear-son’s test for correlations analyses.." This clarification can be found in the revised manuscript on Lines 210-213.

Comments 2:  For Figure 4G, it would be helpful to zoom in and show the stick representation of the key residues K123 and S34. In addition, a structural comparison of the wild-type predicted structure with the mutants, or the inclusion of molecular dynamics (MD) simulation results, would make the analysis more comprehensive. AlphaFold2 should be cited appropriately. MD simulations are described in the main text, but no corresponding results are shown in the figures; please include representative MD results and provide a more detailed description of the MD methodology in the Methods section. The software used for structural visualization should also be specified. Finally, please clarify whether the different colors of the cartoon ribbon indicate prediction confidence. If so, this should be explicitly labeled in the figure or legend. Response 2: Thank you for the constructive suggestions on Figure 4G and the associated methodologies. The original Figure 4D–G has been reorganized as Figure 9A–D to optimize logical flow, with key updates as follows: (1) Enhanced visualization of key residues and structural comparison We updated Figure 8D to include a zoomed-in view of the flexible loop regions (residues 30–38 and 120–128), with stick representations of Lys123 (acetylation site) and Ser34 (phosphorylation site) (atomic coloring: C=gray, N=blue, O=red). The cartoon ribbon colors in Figure 8D represent AlphaFold2’s prediction confidence (pLDDT score): dark blue (pLDDT>90, high confidence), light blue (70–90), yellow (50–70), and orange (<50, low confidence). This color key is added as an inset in Figure 8D and detailed in the legend. Lys123 and Ser34 localize to yellow/orange regions (pLDDT=58–62), confirming their flexible, surface-exposed localization that facilitates PTM by acetyltransferases/kinases. Please see Lines 318-332 and Figure 9D legend (Lines 338-341). (2) Inclusion of representative MD simulation results A supplementary subpanel (Figure 9E) has been added to present root mean square fluctuation (RMSF) plots from 100 ns MD simulations (3 biological replicates). In wild-type (WT) CCM_06472, Ser34 (phosphorylation site) exhibits low flexibility (RMSF = 0.451 Å) and Lys123 (acetylation site) moderate flexibility (RMSF = 0.955 Å)—consistent with functional PTM sites occupying low-to-moderately flexible surface regions (Jumper et al., 2021; Koubassova et al., 2023; Abramson et al.,). Compared to WT, K123Q (acetylation-mimetic) increases Lys123 flexibility (1.806 Å), K123A (acetylation-deficient) reduces it (0.479 Å), and S34D (phosphomimetic) decreases Ser34 flexibility (0.128 Å). These data directly link structural dynamics to PTM susceptibility and the "acetylation-activation/phosphorylation-inhibition" mode, complementing in vitro functional assays and addressing the comment on MD result visualization. Please see Lines 323-337 and Figure 8E legend (Lines 338-341). (3) Citation and methodology supplements A new paragraph (“2.4.3 Molecular Dynamics Simulation”) has been added to the Methods (Lines 192-199): MD simulations of CCM_06472 used the AlphaFold2-predicted structure [23,24] as the initial model. Run in GROMACS 2023.2 with the AMBER99SB-ILDN force field [25], models were solvated (TIP3P, 10 Å buffer, 0.15 M NaCl), minimized, equilibrated (300 K/1 bar), and simulated for 100 ns (n=3). Root Mean Square Fluctuation (RMSF) was calculated via GROMACS; structural visualizations (key residue sticks) via AlphaFold2. Ribbon colors indicate AlphaFold2 pLDDT confidence (dark blue >90, light blue 70–90, yellow 50–70, orange <50). AlphaFold2 is cited as: 23.Jumper, J.; Evans, R.; Pritzel, A.; Green, T.; Figurnov, M.; Ronneberger, O.; Tunyasuvunakool, K.; Bates, R.; Žídek, A.; Potapenko, A.; et al. Highly accurate protein structure prediction with AlphaFold. Nature 2021, 596, 583–589. doi:10.1038/s41586-021-03819-2. 24.Abramson, J.; Adler, J.; Dunger, J.; Evans, R.; Green, T.; Pritzel, A.; Ronneberger, O.; Willmore, L.; Ballard, A.J.; Bambrick, J.; et al. Accurate structure prediction of biomolecular interactions with AlphaFold 3. Nature 2024, 630, 493–500. doi:10.1038/s41586-024-07487-w. 25.Koubassova, N.A.; Tsaturyan, A.K. Molecular dynamics assessment of mechanical properties of the thin filaments in cardiac muscle. Int. J. Mol. Sci. 2023, 24(5), 4792. doi:10.3390/ijms24054792. (4) Clarification of cartoon ribbon color coding The color coding of the cartoon ribbon in Figure 9D explicitly reflects the prediction confidence of AlphaFold2 (pLDDT score), a key metric for structural reliability. To ensure clarity, we have: Added a color key inset in Figure 9D, directly mapping colors to pLDDT ranges: dark blue (pLDDT > 90, high confidence), light blue (70–90, moderate-high confidence), yellow (50–70, low-moderate confidence), and orange (< 50, low confidence); Detailed this color-code correspondence in the Figure 9D legend (Liines 338-341), specifying that Lys123 and Ser34 localize to yellow/orange regions (pLDDT = 58–62). This confirms their flexible, surface-exposed localization, an essential structural feature that facilitates recognition and modification by acetyltransferases/kinases, aligning with our PTM regulatory mechanism   Comments 3: In the legend for Figure 4E (line 286), the text states: “(E) Site-directed mutagenesis validation showed that K123A (mimicking acetylation) activated, whereas S34D (mimicking phosphorylation) inhibited enzyme activity, confirming that Lys123 and Ser34 are critical PTM sites.” However, K123A does not mimic acetylation; K123Q is the acetylation-mimetic mutation. This appears to be a typographical error. Please correct this description accordingly in the legend (and ensure consistency throughout the manuscript). Response 3: We thank the reviewer for this careful and important observation. The reviewer is absolutely correct: K123A is an acetylation-deficient (loss-of-function) mutant, not an acetylation-mimetic mutant. The acetylation-mimetic mutation is K123Q, which substitutes lysine with glutamine to mimic the charge and structural properties of acetylated lysine. The original Figure 4E is changed to revised Figure 9B. This was correctly stated elsewhere in the manuscript (e.g., Section 3.3, Table 4, and Figure 9B). Correction made: The legend for Figure 9B (Lines 335–337) has been revised to: (B) Mutagenesis assays. K123Q enhances enzyme activity 3.7-fold; K123A abolishes ac-tivity, and S34D inhibits enzyme activity by 71%. We have performed a full-text search to ensure no other inconsistencies exist. We sincerely appreciate the reviewer’s meticulous attention to detail, which has enhanced the clarity and accuracy of our manuscript.   Comments 4: For Figure 4F, it would be helpful to include the sequence alignment for the GYS80 strain as well, which would make the comparison more comprehensive. In addition, apart from the highlighted comparison region, are the remaining regions of the sequence identical between the strains? Please clarify this point and discuss any similarities or differences in the text or legend. Response 4: Thank you for the constructive suggestions on the sequence alignment. We have updated Figure 4F (now Figure 9C in the revised manuscript) to include the full-length sequence alignment of CCM_06472 from WT, GYS60 (high-yield), and GYS80 (low-yield) strains, making the inter-strain comparison more comprehensive. Sequence analysis confirms that the entire CCM_06472 coding sequence is 100% identical across the three strains, including both the highlighted functional regions (Ser34 phosphorylation site, Lys123 acetylation site, and SAM-binding motif GFGTGH) and all non-highlighted regions. No insertions, deletions, or missense mutations were detected, which explicitly rules out structural mutations as the driver of differential enzyme activity between strains. This sequence conservation further underscores that post-translational modifications (PTMs) of Ser34 and Lys123, rather than genetic variations, are the core regulatory mechanism governing CCM_06472 activity and cordycepin biosynthetic flux, aligning with our functional validation data (Figure 9A-B). The updated manuscript (Lins 312-317) and corresponding legend (Lines 337-338) clearly illustrate the sequence identity.   Comments 5:  Line 266: “Sequence alignment (Figure 4F) demonstrated that GYS60 retains the Ser34 and Lys123 residues, as well as the conserved SAM-binding motif (GFGTGH), confirming that enhanced enzymatic activity is driven by post-translational modifications (PTMs) rather than structural mutations.” Here, it is unclear what is meant by “structural mutations.” Does this refer only to amino acid substitutions detectable by sequence alignment, or to predicted conformational/structural changes in the protein (e.g., based on modeling or MD simulations)? Please clarify the intended meaning and, if applicable, rephrase this sentence for precision. Response 5: Thank you for pointing out this ambiguity. We apologize for the unclear terminology "structural mutations" in the original statement, specifically refers to genetically encoded amino acid alterations detectable by sequence alignment, not conformational/structural changes predicted by modeling or MD simulations. To clarify, we have revised the sentence to: "Sequence alignment (Figure 9C) demonstrates that CCM_06472 retains the the phos-phorylation site Ser34, acetylation site Lys123, and the conserved SAM-binding motif (GFGTGH) across WT, GYS60, and GYS80 strains. Full-length sequence alignment fur-ther verifies that the amino acid sequence of CCM_06472 is identical across the three strains (Figure 9C), with no genetic mutations detected in the coding region." (Lines 312-317 ).   Comments 6: Line 387: For the Western blot (WB) experiments, the antibodies used should be clearly listed in the Methods section, so that the experiments can be reliably reproduced. Response 6: Thank you for this comment. We have now added full antibody details to the Materials and Methods section to ensure reproducibility. Specifically, in 2.4.2. Western Blot Validation, we now specify: Custom rabbit polyclonal anti‑CCM_06472 (raised against residues 1–30, affinity‑purified to 1.2 mg/mL); HRP‑conjugated goat anti‑rabbit IgG (Cusabio, Cat. CSB‑PA489724) at 1:10,000; Mouse anti‑GAPDH (Cusabio, Cat. CSB‑MA000071M1m) at 1:5,000; This information can be found in Lines 179–191 of the revised manuscript. We thank the reviewer for helping us improve the clarity of our method description.   Comments 7: In Section 2.7 (Enzyme Activity Assays), the authors use crude extracts to assess the activities of malate dehydrogenase (MDH, CCM_04892) and O-methyltransferase (CCM_06472). However, when working with crude extracts, other endogenous enzymes with similar or overlapping activities may contribute to the measured signals. Please explain in more detail how the assay was designed to ensure that the observed activities can be specifically attributed to MDH (CCM_04892) and O-methyltransferase (CCM_06472). A clearer description of the experimental design and its specificity would strengthen the conclusions drawn from these assays.] Response 7: Thank you for this constructive comment. We have revised Section 2.4.1 Enzyme Activity Assays to clearly explain the specificity controls used in crude extract assays. For CCM_06472 (O-methyltransferase), specific activity was calculated as the difference in 3′-O-methyl-AMP formation between assays with and without preincubation with our custom anti-CCM_06472 antibody, which specifically blocks the target enzyme activity. This antibody subtraction method, now detailed in Section 2.4.1. Enzyme Activity Assays (Lines 168–182) , directly addresses the concern of overlapping enzyme activities in crude extracts and ensures that the measured signals are specifically attributable to the target enzymes. Because no malate dehydrogenase (CCM_04892) activity data are presented in the Results, we have removed all corresponding assay descriptions from the methods. Only CCM_06472 activity data and validation are retained and fully supported by experimental evidence. We thank the reviewer for prompting this methodological clarification.

 

Reviewer 3 Report

Comments and Suggestions for Authors

Summary:

This study reframes the conventional metabolic engineering paradigm by shifting from static pathway amplification to dynamic checkpoint engineering. By examining the metabolic divergence between high- and low-yield Cordyceps militaris strains, Hucheng Zhang and colleagues identify a molecular “valve” centered on the O-methyltransferase CCM_06472. The regulatory mechanism is governed by an acetyl-CoA–gated checkpoint mediated through antagonistic post-translational modifications: enzymatic activity is enhanced by Lys123 acetylation and inhibited by Ser34 phosphorylation. This design leverages acetyl-CoA not only as a biosynthetic precursor but also as a signaling metabolite, establishing a self-reinforcing regulatory circuit that redirects a 4.1-fold metabolic surplus toward cordycepin production. Notably, the authors demonstrate that a minimal intervention, a single K123Q substitution is sufficient to recapitulate the high-yield phenotype, resulting in more than a fourfold increase in cordycepin titers. Overall, this work proposes a broadly applicable design principle for overcoming yield limitations in microbial cell factories by harnessing, rather than overriding, endogenous regulatory logic. I therefore recommend this manuscript for publication pending minor revisions, as detailed below.

Minor comments:

1. For Figure 3, some of the annotations are too small such as the numbers in the Figure 3D, H, I and J. The author is advised to re-layout these figures.
2. What is the difference between one, two, three and four in the Figure 3A? Just samples were subsequently collected four times?
3. It would be more helpful for readers to understand this study if the author can show the biosynthetic pathway of cordycepin.

Author Response

3. Point-by-point response to Comments and Suggestions for Authors

Comments: This study reframes the conventional metabolic engineering paradigm by shifting from static pathway amplification to dynamic checkpoint engineering. By examining the metabolic divergence between high- and low-yield Cordyceps militaris strains, Hucheng Zhang and colleagues identify a molecular “valve” centered on the O-methyltransferase CCM_06472. The regulatory mechanism is governed by an acetyl-CoA–gated checkpoint mediated through antagonistic post-translational modifications: enzymatic activity is enhanced by Lys123 acetylation and inhibited by Ser34 phosphorylation. This design leverages acetyl-CoA not only as a biosynthetic precursor but also as a signaling metabolite, establishing a self-reinforcing regulatory circuit that redirects a 4.1-fold metabolic surplus toward cordycepin production. Notably, the authors demonstrate that a minimal intervention, a single K123Q substitution is sufficient to recapitulate the high-yield phenotype, resulting in more than a fourfold increase in cordycepin titers. Overall, this work proposes a broadly applicable design principle for overcoming yield limitations in microbial cell factories by harnessing, rather than overriding, endogenous regulatory logic. I therefore recommend this manuscript for publication pending minor revisions, as detailed below.

Minor comments:

Comments 1: For Figure 3, some of the annotations are too small such as the numbers in the Figure 3D, H, I and J. The author is advised to re-layout these figures.

Response1: We thank the reviewer for this constructive suggestion. To improve readability, we have reorganized the original Figure 3 into three separate figures.

The original Figure 3D (functional annotation rates) is now presented as Figure 6A.

The original Figure 3H, I, J (protein molecular weight distribution, peptide length distribution, and high‑confidence protein filtering) have been redesigned and enlarged as Figure 7A, 7B, and 7C, respectively.

All annotations have been enlarged for clarity.

The revised figures can be found on Figure 5, Figure 6, and Figure 7.

 

Comments 2: What is the difference between one, two, three and four in the Figure 3A? Just samples were subsequently collected four times?

Response2: Thank you for this helpful suggestion. We have clarified the representation of lanes in Figure 3A (now revised as Figure 5A) in the figure legend.

As stated in the Materials and Methods (Section 2.5), all quantitative experiments were performed with four independent biological replicates unless stated otherwise. Therefore, lanes 1–4 represent four independent biological replicates for each strain.

This clarification can be found in the revised manuscript on 2.5. Data Integration, Statistical Analysis, Line 215-218.

 

Comments 3: It would be more helpful for readers to understand this study if the author can show the biosynthetic pathway of cordycepin.

Respons 3: We appreciate this constructive suggestion. We have added Figure 1 to clearly illustrate the classical two-step cordycepin biosynthetic pathway catalyzed by the cns1–cns2 gene cluster (Lines 60-64), as shown below:

 

Figure 1. Simplified biosynthetic pathway of cordycepin in Cordyceps militaris

Note: The pathway proceeds from 3′-AMP via sequential catalysis by Cns2 (phosphohydrolase) and Cns1 (oxidoreductase/dehydrogenase), generating 2′-C-3′-dA as an intermediate before forming cordycepin. Both enzymes are indispensable for cordycepin biosynthesis.

We have also optimized labeling and layout in Figure 12 to enhance clarity, making the classical pathway easier to follow and directly connecting it to the acetyl-CoA checkpoint mechanism.

Reviewer 4 Report

Comments and Suggestions for Authors

The study brings together multiple approaches and presents a potentially valuable engineering outcome. However, some aspects of the mechanistic interpretation appear to be stated more strongly than the data support, and there are a few internal inconsistencies that would benefit from clarification. Strengthening the alignment between the data and the conclusions, providing clearer methodological detail, and ensuring consistency in the presentation of the mutational analysis would further improve the manuscript.

1. In Section 3.4 (AlphaFold and sequence discussion), it is unclear whether Ser34 and Lys123 differ between GYS60 and GYS80 or are conserved. The text suggests that GYS60 “harbors Ser34 and Lys123 mutations,” but also indicates that these residues are conserved and that the observed differences are due to post-translational modifications rather than sequence variation. This creates confusion, and the manuscript would benefit from a clear and consistent statement of whether these positions are identical or different between the two strains.

2. In Section 3.4 (acetyl-CoA activation assay), the manuscript concludes that acetyl-CoA directly activates CCM_06472 and describes the enzyme as an acetyl-CoA “sensor.” However, the assay is performed using crude extracts, and it is not clear whether acetyl-CoA directly modifies the enzyme or acts indirectly through other components present in the extract. As a result, the current experimental setup does not distinguish between direct and indirect effects, and the wording appears to overstate the level of mechanistic resolution.

3. In Section 3.3, the manuscript suggests that acetyl-CoA is preferentially directed toward cordycepin biosynthesis in GYS60 compared to GYS80. However, it is not clear how this redistribution is quantified. Although metabolomics and isotope labeling data are presented, the Methods do not explain how these data were used to estimate pathway-level allocation, making the interpretation difficult to follow.

4. In Sections 3.5 and 3.6, the link between acetyl-CoA levels, CCM_06472 acetylation, and cordycepin production is supported using acetate supplementation, ACS perturbation, and acetylation-modulating interventions. These approaches affect cellular metabolism broadly and do not isolate the specific contribution of Lys123 acetylation. While the results are consistent with the proposed model, they do not uniquely establish causality, and alternative explanations remain possible.

5. In the Discussion and Figure 6, the manuscript proposes a “universal acetyl-CoA checkpoint” for fungal secondary metabolism. This interpretation appears to extend beyond the scope of the data, which are based on a single enzyme in one organism. The model may be more appropriately presented as a system-specific mechanism or as a hypothesis for broader applicability.

6. In Methods Section 2.4 and the corresponding results, enzyme assays are performed using crude protein extracts. This makes it difficult to interpret acetyl-CoA dependence and post-translational modification effects, as other proteins and enzymatic activities in the extract may influence the observed results. This limitation should be taken into account when interpreting these findings.

7. Across Methods Sections 2.2-2.6 and the related figures, the statistical reporting is not always clear. It would help to specify, for each experiment, the number of replicates and the statistical test used. Presenting this information consistently would make the results easier to interpret.

8. In the Data Availability statement, the manuscript indicates that data are included in the article. However, given the use of proteomics, PTM profiling, and metabolomics, it would be important to deposit the underlying datasets in appropriate public repositories and provide accession numbers.

9. In Section 3.4, the structural analysis (AlphaFold modeling and RMSF values) is used to support the accessibility of Ser34 and Lys123. While this is a reasonable supporting observation, it provides indirect evidence and does not directly validate the proposed regulatory mechanism. The current presentation gives this analysis more weight than it likely supports.

Author Response

3. Point-by-point response to Comments and Suggestions for Authors

Comments: The study brings together multiple approaches and presents a potentially valuable engineering outcome. However, some aspects of the mechanistic interpretation appear to be stated more strongly than the data support, and there are a few internal inconsistencies that would benefit from clarification. Strengthening the alignment between the data and the conclusions, providing clearer methodological detail, and ensuring consistency in the presentation of the mutational analysis would further improve the manuscript.

 

Comments 1: In Section 3.4 (AlphaFold and sequence discussion), it is unclear whether Ser34 and Lys123 differ between GYS60 and GYS80 or are conserved. The text suggests that GYS60 “harbors Ser34 and Lys123 mutations,” but also indicates that these residues are conserved and that the observed differences are due to post-translational modifications rather than sequence variation. This creates confusion, and the manuscript would benefit from a clear and consistent statement of whether these positions are identical or different between the two strains.

Response 1: We appreciate the reviewer’s valuable comment. we have revised the manuscript to include a clear, consistent statement confirming sequence identity at these key residues:

Sequence alignment (Figure 9C) demonstrates that CCM_06472 retains the the phosphorylation site Ser34, acetylation site Lys123, and the conserved SAM-binding motif (GFGTGH) across WT, GYS60, and GYS80 strains. Full-length sequence align-ment further verifies that the amino acid sequence of CCM_06472 is identical across the three strains (Figure 8C), with no genetic mutations detected in the coding region. This explicitly rules out genetic variations as the driver of differential enzyme activity, directly highlighting PTMs of Ser34 and Lys123 as the core regulatory mechanism linking acetyl-CoA abundance to cordycepin biosynthetic flux.

Pleas see Lines 318-322.

Comments 2: In Section 3.4 (acetyl-CoA activation assay), the manuscript concludes that acetyl-CoA directly activates CCM_06472 and describes the enzyme as an acetyl-CoA “sensor.” However, the assay is performed using crude extracts, and it is not clear whether acetyl-CoA directly modifies the enzyme or acts indirectly through other components present in the extract. As a result, the current experimental setup does not distinguish between direct and indirect effects, and the wording appears to overstate the level of mechanistic resolution.

Response 2: We sincerely appreciate the reviewer’s rigorous and professional comment.

We agree that conclusions should be carefully aligned with the experimental system.

We have:

(1)Removed “directly activates” and replace “acetyl-CoA sensor” with acetyl-CoA-responsive regulatory node, and stated that acetyl‑CoA activates CCM_06472 in a dose‑dependent manner. As shown in Figure 10A, CCM_06472 activity exhibits a clear sigmoidal response to acetyl‑CoA with an EC₅₀ of 25.8 μM, demonstrating that acetyl‑CoA acts as a specific, concentration‑dependent activator of the enzyme. Specificity was further validated by antibody competition, which abolished >90% of the activity signal.

(2)Replaced overstated wording with moderate, evidence‑based descriptions throughout the manuscript, especially in:

• 4.1. Enzyme Activity Assays (Lines 169–182);
• 4.1 Acetyl-CoA activates CCM_06472 in a dose-dependent manner (Lines 373–398);
• Figure 10A legend: In vitro CCM_06472 activity exhibits a sigmoidal response to acetyl-CoA (EC₅₀ = 25.8 μM). (Line 398).
• 1. Acetyl-CoA Signaling via CCM_06472 Gates Metabolic Flux to Cordycepin

(3)Refined terminology: Removed absolute terms (e.g., “directly activates”) and replaced “acetyl-CoA sensor” with “acetyl-CoA-responsive regulatory node”/“metabolic responder to acetyl-CoA levels”, aligning language with experimental evidence while retaining that acetyl-CoA modulates CCM_06472 activity concentration-dependently.

(4)Highlighted specificity controls: Our assays included anti-CCM_06472 antibody preincubation (1:100 dilution), which eliminated >90% of acetyl-CoA-induced activity, confirming specificity to CCM_06472 and ruling out off-target effects. This control is now emphasized in Section 3.4 (Lines 385-394).

 

Comments 3: In Section 3.3, the manuscript suggests that acetyl-CoA is preferentially directed toward cordycepin biosynthesis in GYS60 compared to GYS80. However, it is not clear how this redistribution is quantified. Although metabolomics and isotope labeling data are presented, the Methods do not explain how these data were used to estimate pathway-level allocation, making the interpretation difficult to follow.

Response 3:  We appreciate the reviewer’s insightful comment on quantifying acetyl-CoA flux redistribution. To address the unclear pathway-level allocation estimation, we revised the manuscript without adding new experiments:

We added Section 2.3.3 “Isotope-Tracer Flux Analysis” (Lines 153-166) to detail the ¹³C-glucose tracing workflow, UPLC-MS/MS quantification of labeled metabolites (acetylcarnitine as acetyl-CoA surrogate, cordycepin, riboflavin), and flux allocation formula (integrating fractional labeling, product concentration, and acetyl-CoA stoichiometry), aligning with standard protocols [Long & Antoniewicz, 2019].  

Long, C.P., Antoniewicz, M.R. High-resolution ¹³C metabolic flux analysis. Nat. Protoc. 2019, 14, 2856–2877. doi:10.1038/s41596-019-0204-0

In Section 3.1, we explicitly link the 62% (GYS60) vs. 18% (GYS80) acetyl-CoA flux to cordycepin to the supplemented method, confirming quantification via ¹³C-labeling data, stoichiometric correction, and background subtraction.

We updated our manuscript and Figure 4B’s legend (Lines: 233-242) to reference Section 2.3.3 for transparency.

 

Comments 4: In Sections 3.5 and 3.6, the link between acetyl-CoA levels, CCM_06472 acetylation, and cordycepin production is supported using acetate supplementation, ACS perturbation, and acetylation-modulating interventions. These approaches affect cellular metabolism broadly and do not isolate the specific contribution of Lys123 acetylation. While the results are consistent with the proposed model, they do not uniquely establish causality, and alternative explanations remain possible.

Response 4: We appreciate the reviewer’s insightful comment on isolating Lys123 acetylation’s specific contribution to cordycepin production. To address concerns about broad metabolic interventions, we emphasize our orthogonal strategy uniquely establishes causality—retaining all original data and references:

(1) Site-specific PTM-mimetic mutations directly target Lys123: Acetylation-mimetic K123Q boosted wild-type yield 4.3-fold (1.65 g/L, 91% of GYS60’s titer), while acetylation-deficient K123A reduced GYS60 yield by 68% (Table 4, Figure 9B). Phosphomimetic S34D suppressed production to 0.22 g/L, confirming Lys123 acetylation (not global metabolism) drives yield.

(2) Metabolic interventions include strict controls: Acetate supplementation, ACS perturbation, and acetylation inhibition (ATi/Sirt-OE) did not alter CCM_06472 transcription/translation (Figure 11A,B) and specifically modulated Lys123 acetylation (4.2-fold increase in ACS-HO; 3.8-fold in acetate-supplemented strains) without affecting other CCM_06472 PTMs (Table 2).

(3) Structural dynamics validate specificity: MD simulations showed K123Q increases Lys123 flexibility to 1.806 Å (within the 1.0–2.0 Å activation range [23,24,25]), aligning with activity/yield trends (Figure 9E) and ruling out alternative mechanisms.

These genetic, biochemical, and structural data collectively confirm Lys123 acetylation of CCM_06472 is a causal, rate-limiting factor for cordycepin production, eliminating alternative explanations for the proposed acetyl-CoA-gated PTM switch model.

 

Comments 5: In the Discussion and Figure 6, the manuscript proposes a “universal acetyl-CoA checkpoint” for fungal secondary metabolism. This interpretation appears to extend beyond the scope of the data, which are based on a single enzyme in one organism. The model may be more appropriately presented as a system-specific mechanism or as a hypothesis for broader applicability.

Response 5: We appreciate the reviewer’s valuable comment. To address the overgeneralization concern, we revised the manuscript. Figure 6 is changed to Figure 12.

(1)Removed all “universal” word: throughout the Results and Discussion to avoid overgeneralization.

(2)Narrowed model scope: Renamed Section 3.6 (Line 474) and adjusted core description (Lines 474–495) to explicitly tie the checkpoint to C. militaris cordycepin biosynthesis (CCM_06472-mediated), removing "universal" framing.

(3)Reformulated as a hypothesis: Revised Section 4.2 (lines 548–571) to present broader applicability as a hypothesis supported by parallel fungal metabolic studies [5,31,32,33.34].

(4)Updated Figure 12 and legend (line 483): Clarified specificity to C. militaris cordycepin biosynthesis, adding a note that broader validation is needed.

These changes adhere to the data scope, reinforce rigor by distinguishing conclusions from hypotheses, and preserve the study’s core impact.

 

Comments 6: In Methods Section 2.4 and the corresponding results, enzyme assays are performed using crude protein extracts. This makes it difficult to interpret acetyl-CoA dependence and post-translational modification effects, as other proteins and enzymatic activities in the extract may influence the observed results. This limitation should be taken into account when interpreting these findings.

Response 6: We appreciate the reviewer’s comment. While crude extracts may contain endogenous components, our experimental design ensures reliable interpretation:

(1) Specificity was validated via antibody competition

Target-specific activity was confirmed by subtracting signals from antibody-preincubated reactions, eliminating non-specific interference, in Section 2.4.1, Lines 168-172.

(2) Assay specificity was ensured by product detection

CCM_06472 activity was quantified via UPLC-MS/MS targeting the unique product 3′-O-methyl-AMP, avoiding cross-reactivity, in Section 2.4.1, Lines 172-176.

(3) Findings are cross-validated by orthogonal data

Dose-dependent responses to acetyl-CoA, site-specific PTM-mimetic mutations (K123Q/K123A/S34D), and Western blot results collectively support the regulatory relationship between acetyl-CoA, Lys123 acetylation, and CCM_06472 activity.

We acknowledge crude extract assays have inherent limitations, which are considered in interpretation. The combined controls and cross-validation ensure conclusion robustness, with all original references retained.

 

Comments 7: Across Methods Sections 2.2-2.6 and the related figures, the statistical reporting is not always clear. It would help to specify, for each experiment, the number of replicates and the statistical test used. Presenting this information consistently would make the results easier to interpret.

Response 7:  We appreciate the reviewer’s comment. We have supplemented Section 2.5 with key details while maintaining conciseness :

(1) Defined biological replicates and explicitly stated statistical tests in Sectiion 2.5 for all experiments (Lines 210-213)

All quantitative experiments were performed with four independent biological replicates unless stated otherwise. Data are shown as mean ± SD, with CV < 15% con-sidered acceptable. Student’s t-test was used for two-group comparisons, and Pear-son’s test for correlations analyses.

(2)Added replicate numbers and statistical indicators (n, p values) to all relevant figure legends for full transparency.

These additions ensure consistent, transparent statistical reporting without redundant content.  

 

Comments 8: In the Data Availability statement, the manuscript indicates that data are included in the article. However, given the use of proteomics, PTM profiling, and metabolomics, it would be important to deposit the underlying datasets in appropriate public repositories and provide accession numbers.

Response 8: We sincerely appreciate the reviewer’s professional suggestion on data deposition.

All raw datasets have been prepared and will be deposited into the public repositories iPROX ( https://www.iprox.cn/) and MetaboLights (https://www.ebi.ac.uk/metabolights/) prior to final publication, with accession numbers to be added then.

The Data Availability statement has been revised accordingly to ensure full compliance with the journal’s data-sharing policy.

 

Comments 9: In Section 3.4, the structural analysis (AlphaFold modeling and RMSF values) is used to support the accessibility of Ser34 and Lys123. While this is a reasonable supporting observation, it provides indirect evidence and does not directly validate the proposed regulatory mechanism. The current presentation gives this analysis more weight than it likely supports.

Response 9: We appreciate the reviewer’s comment. We have revised the manuscript to clarify that AlphaFold modeling and RMSF data provide indirect, supportive evidence for the structural Ser34/Lys123 accessibility, consistent with known flexibility criteria for functional PTM sites [23,24,25]).

Causal validation of the acetylation/phosphorylation-gated model relies on orthogonal functional evidence, including site-directed mutagenesis (K123Q/S34D/K123A), acetyl-CoA dose-response assays, ACS genetic manipulation, and acetylation inhibition (ATi/Sirt-OE. We adjusted wording in Sections 3.2.3, 4.1 and Figure 9 legends to reduce overemphasis on structural data, aligning it with its supportive role. Please see Lines 305-341.

All original references and core conclusions are retained, with the regulatory mechanism’s robustness confirmed by functional and genetic experiments.

 

Round 2

Reviewer 4 Report

Comments and Suggestions for Authors

The manuscript has improved considerably, and most of the major concerns have been addressed. The added clarification on flux analysis, the resolution of inconsistencies between sequence and PTM regulation, and the clearer methodological descriptions all strengthen the work. The study is now close to being suitable for publication, but a few minor points should still be addressed.

• Although the authors have toned down some of the earlier claims, parts of the manuscript still read as if the mechanism is fully established. In particular, statements implying that acetyl-CoA regulates CCM_06472 activity through Lys123 acetylation in a definitive causal way should be slightly softened. Since the enzyme assays were carried out using crude extracts, it would be more appropriate to present these conclusions as well-supported rather than directly demonstrated.

• The mutation data (K123Q/K123A/S34D) provide convincing functional support for the model. However, these substitutions do not directly reflect endogenous PTM dynamics. Claims that these experiments establish causality or rule out alternative explanations should therefore be moderated. It would be more accurate to state that the results are consistent with a causal role of Lys123 acetylation.

• The removal of the “universal checkpoint” claim is appropriate. That said, some parts of the Discussion still imply broader applicability beyond the system studied here. It would help to clearly frame this as a hypothesis and keep the conclusions focused on C. militaris.
• The data availability section needs to be completed. The manuscript mentions that datasets have been deposited, but accession numbers are not provided. These should be included, and the statement made fully consistent.

Recommendation: minor revision.

Author Response

Comments: The manuscript has improved considerably, and most of the major concerns have been addressed. The added clarification on flux analysis, the resolution of inconsistencies between sequence and PTM regulation, and the clearer methodological descriptions all strengthen the work. The study is now close to being suitable for publication, but a few minor points should still be addressed. Recommendation: minor revision.

Comment 1: Although the authors have toned down some of the earlier claims, parts of the manuscript still read as if the mechanism is fully established. In particular, statements implying that acetyl-CoA regulates CCM_06472 activity through Lys123 acetylation in a definitive causal way should be slightly softened. Since the enzyme assays were carried out using crude extracts, it would be more appropriate to present these conclusions as well-supported rather than directly demonstrated.

Response 1: We appreciate the reviewer’s constructive comments.

We have toned down all definitive causal statements throughout the manuscript and framed conclusions as being well-supported by multiple lines of evidence, consistent with data generated from crude extract-based enzyme assays.

In Section 3.2.3 (Lines 305–309), we changed ‘established PTM as the dominant regulatory layer’ to ‘supports PTM as the dominant regulatory layer’.

In Section 3.4.1 (Line 430), we replaced ‘confirmed acetyl-CoA as a direct activator’ with ‘supports acetyl-CoA as an activator’.

Similar adjustments were made in Section 4.1 (Lines 515–517) to align conclusions with the evidence from crude extract assays.

Please see the details in manuscript (tracked change round 2).

 

Comment 2: The mutation data (K123Q/K123A/S34D) provide convincing functional support for the model. However, these substitutions do not directly reflect endogenous PTM dynamics. Claims that these experiments establish causality or rule out alternative explanations should therefore be moderated. It would be more accurate to state that the results are consistent with a causal role of Lys123 acetylation.

Response 2: We appreciate the reviewer’s insightful comment.

We have revised the manuscript to explicitly distinguish between supportive evidence (mutagenesis) and definitive causality (which would require endogenous PTM manipulation). For example:

In Section 3.2.3 (Lines 310–311), we changed ‘validated the PTM switch’ to ‘is consistent with a causal role of Lys123 acetylation’.

In Section 3.3 (Lines 361–363), we replaced ‘established the causal roles’ with ‘is consistent with a causal role’.

Please see the details in manuscript (tracked change round 2).

 

Comment 3: The removal of the “universal checkpoint” claim is appropriate. That said, some parts of the Discussion still imply broader applicability beyond the system studied here. It would help to clearly frame this as a hypothesis and keep the conclusions focused on C. militaris.

Response 3: We appreciate the reviewer’s constructive comment.

We have revised the Discussion to strictly focus conclusions on C. militaris and explicitly frame all broader implications as testable hypotheses within the studied system. Specifically:

In Section 4.2 (Lines 554–555), we now state: ‘The acetyl-CoA-mediated checkpoint we describe in C. militaris may represent a testable hypothesis for conserved regulatory paradigms in fungal specialized metabolism’ and explicitly list three testable hypotheses.

In Section 4.5 (Lines 617–619), we replaced ‘generalizable principle’ with ‘system-specific regulatory principle’ and added: ‘As a testable hypothesis for future work, this acetyl-CoA-gated logic may be explored by engineering analogous PTM-sensing modules in other fungal systems.’

The Figure 12 legend (Line 486) now specifies that the model is proposed for C. militaris cordycepin biosynthesis, with broader applicability remaining to be tested.

Please see the details in the manuscript.

 

Comment 4: The data availability section needs to be completed. The manuscript mentions that datasets have been deposited, but accession numbers are not provided. These should be included, and the statement made fully consistent.

Response 4: We appreciate the reviewer’s constructive comment.

We have completed the data availability section with full repository URLs and valid accession numbers.

The raw datasets have been deposited in iPROX (accession: IPX0016387000) and MetaboLights (accession: REQ20260326218169). These datasets are currently under curation; as per repository policy, they will be made publicly available upon manuscript acceptance, at which point the accession numbers will become active. All other data are available from the corresponding author upon reasonable request.

This statement is now fully consistent with journal data policy and is included in the revised manuscript (Lines 661–664).