Exploring the Chemical Space of Cephalosporins Across Generations

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript „Exploring the Chemical Space of Cephalosporins Across Generations” by Henrique de Aguiar Mello and Itamar Luís Gonçalves is an interesting short study of design evolution of the cephalosporin family. The manuscript is well suited for the “Drugs and Drug Candidates” journal, as its main aspect is a comparative analysis of the drug-likeness of cephalosporins with ultimate aim of revealing structural and pharmacokinetic patterns influencing the bioavailability and oral administration profile of cephalosporins of different generations. The manuscript is written clearly and the results are presented with care. The reader is easily able to follow the line of reasoning of the authors. The graphical contents is well prepared and designed to underline the main points of discussion. The study is based on simple concepts, which I personally consider its advantage: the relevant features (e.g. logP differences across generations) are visible and not obscured by the use of esoteric descriptors which are hard to connect with physicochemical meaning. Thus, I suggest accepting the manuscript after very minor revisions:

 

The authors do not state (in the Introduction or in the Materials & Methods) how large is the chemical space of cephalosporins in active use. The 38 compounds are described as “representative of different generations of this antibiotic group” (line 300), but are they roughly 50%, 25%, or maybe 10% of the known and used cephalosporins?

 

The submission is accompanied by a supplementary file (Excel spreadsheet with the descriptors and SMILES representations). The manuscript does not contain any information on the Supplementary Files – is it an omission, or was the Excel file intended for review only?

 

The fragment concerning the structure preparation (lines 304-307) should be enlarged. The used descriptors, as far as I see, are in fact 2D descriptors, so the SMILES input would be sufficient. Why were then the molecular structures re-created in ChemDraw? If the 3D structures were generated, were they just standard models (standardized bond lengths etc.), or maybe ChemDraw allowed the authors to carry out molecular mechanics optimization? Please clarify.

 

End of reviewer remarks

Author Response

# Reviewer 1

The manuscript „Exploring the Chemical Space of Cephalosporins Across Generations” by Henrique de Aguiar Mello and Itamar Luís Gonçalves is an interesting short study of design evolution of the cephalosporin family. The manuscript is well suited for the “Drugs and Drug Candidates” journal, as its main aspect is a comparative analysis of the drug-likeness of cephalosporins with ultimate aim of revealing structural and pharmacokinetic patterns influencing the bioavailability and oral administration profile of cephalosporins of different generations. The manuscript is written clearly and the results are presented with care. The reader is easily able to follow the line of reasoning of the authors. The graphical contents is well prepared and designed to underline the main points of discussion. The study is based on simple concepts, which I personally consider its advantage: the relevant features (e.g. logP differences across generations) are visible and not obscured by the use of esoteric descriptors which are hard to connect with physicochemical meaning. Thus, I suggest accepting the manuscript after very minor revisions:

We thank the reviewer for the careful reading of our manuscript and for the very positive evaluation. We greatly appreciate the recognition of the clarity of the writing, the quality of the graphical content, and the relevance of our approach.

 

The authors do not state (in the Introduction or in the Materials & Methods) how large is the chemical space of cephalosporins in active use. The 38 compounds are described as “representative of different generations of this antibiotic group” (line 300), but are they roughly 50%, 25%, or maybe 10% of the known and used cephalosporins?

We thank the reviewer for the positive and encouraging evaluation of our manuscript. Regarding the coverage of the cephalosporin chemical space, we would like to clarify that DrugBank reports approximately 60 cephalosporins and closely related derivatives. Based on this landscape, we constructed a representative dataset comprising 38 compounds, spanning all generations and capturing the major structural and pharmacokinetic trends that have driven the evolutionary design of this antibiotic class. This information has now been added to the Materials and Methods section.

 

The submission is accompanied by a supplementary file (Excel spreadsheet with the descriptors and SMILES representations). The manuscript does not contain any information on the Supplementary Files – is it an omission, or was the Excel file intended for review only?

Thank you for this comment. This was an oversight on our part. We have now clearly indicated throughout the manuscript the existence of Supplementary Materials and explicitly referenced the Excel file containing the molecular descriptors and SMILES representations. In addition, the availability of this supplementary dataset has been properly stated in the Data Availability Statement section. 

 

The fragment concerning the structure preparation (lines 304-307) should be enlarged. The used descriptors, as far as I see, are in fact 2D descriptors, so the SMILES input would be sufficient. Why were then the molecular structures re-created in ChemDraw? If the 3D structures were generated, were they just standard models (standardized bond lengths etc.), or maybe ChemDraw allowed the authors to carry out molecular mechanics optimization? Please clarify.

Thank you for this comment. All molecular descriptors used in this study are 2D, graph-based descriptors. Complexity indices (Zagreb, Bertz, and eccentric connectivity) were calculated solely from molecular connectivity. ChemDraw was used only for structure curation and standardization, verification of SMILES consistency, and figure preparation; no 3D structures or geometry optimizations were generated.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Manuscript Title
Exploring the Chemical Space of Cephalosporins Across Generations

The study is concise and scientifically sound for its scope. This manuscript is clearly written and delivers on its stated aim of using simple physicochemical descriptors and PCA to illustrate how cephalosporins have evolved across generations. The contribution is modest and purely computational, but it is appropriate for the journal and should be useful for readers interested in medicinal chemistry trends and oral bioavailability rules. I have a few minor suggestions to improve clarity and reproducibility.

Suggestions

❖ First, consider adding logD at pH 7.4 and pKa estimates to the descriptor set, as these are more informative than logP alone for ionizable cephalosporins. It would also help to briefly comment on the typical ionization/zwitterionic character of cephalosporins, porin-mediated uptake in Gram-negative bacteria, and why Ro5 and Veber-type filters are imperfect for antibacterials.

❖ Second, Please add the following reference to connect your structural observations (C3 tetrazoles, pyridinium/quaternary ammonium motifs) to clinically relevant protein-binding data: Kanis E; Parks J; Austin DL. Structural Analysis and Protein Binding of Cephalosporins. ACS Pharmacology & Translational Science. 2023;6:88–91. DOI: 10.1021/acsptsci.2c00168. Suggested placement: Section 2.3 (C3 tetrazoles) and Discussion (albumin binding, charge/polarity). Sample linking sentence: “These C3/C7 motifs have documented associations with albumin binding—tetrazole positively and pyridinium/quaternary amines negatively—across marketed cephalosporins (Kanis et al., 2023)

❖ Third, figure captions would benefit from including the number of compounds per group and, where relevant, the percentage of compounds with oral formulations; a brief indication of the variance explained by the first two principal components and the main contributing descriptors would further help readers interpret the PCA plots.

❖ Clarify the grouping choice that combines fourth and fifth generations (already noted in figure captions); and note variability in generational labels for agents like ceftaroline/ceftobiprole (often termed “advanced” rather than strictly “fifth”).

Conclusion remarks/suggestions

Overall, the study is concise and scientifically sound within its scope. The suggested changes are minor clarifications rather than substantive new work, and implementing them will improve the paper. I recommend acceptance after minor revision.

Author Response

Reviewer 2

The study is concise and scientifically sound for its scope. This manuscript is clearly written and delivers on its stated aim of using simple physicochemical descriptors and PCA to illustrate how cephalosporins have evolved across generations. The contribution is modest and purely computational, but it is appropriate for the journal and should be useful for readers interested in medicinal chemistry trends and oral bioavailability rules. I have a few minor suggestions to improve clarity and reproducibility.

We thank the reviewer for the careful evaluation of our manuscript and for the positive and balanced assessment of its scope and contribution. We appreciate the recognition of the clarity of the writing, the appropriateness of the methodological approach, and the relevance of the study for readers interested in medicinal chemistry trends. The reviewer’s minor suggestions are valuable and have helped us to further improve the clarity and reproducibility of the manuscript.

 

Suggestions

❖ First, consider adding logD at pH 7.4 and pKa estimates to the descriptor set, as these are more informative than logP alone for ionizable cephalosporins. It would also help to briefly comment on the typical ionization/zwitterionic character of cephalosporins, porin-mediated uptake in Gram-negative bacteria, and why Ro5 and Veber-type filters are imperfect for antibacterials.

We thank the reviewer for this valuable suggestion and fully agree that logD at pH 7.4 and pKa-related descriptors are highly relevant for ionizable antibiotics such as cephalosporins, as well as with the limitations of Ro5- and Veber-type filters for antibacterial agents. In this study, we intentionally limited the descriptor set to simple and widely standardized physicochemical parameters to ensure consistency and direct comparability across the dataset. Incorporation of pKa-dependent descriptors would require a complete re-parameterization of the dataset and was therefore considered beyond the scope of the present work. To address this point, we have added a brief discussion on cephalosporin ionization, zwitterionic character, porin-mediated uptake, and the limitations of classical drug-likeness rules.

❖ Second, Please add the following reference to connect your structural observations (C3 tetrazoles, pyridinium/quaternary ammonium motifs) to clinically relevant protein-binding data: Kanis E; Parks J; Austin DL. Structural Analysis and Protein Binding of Cephalosporins. ACS Pharmacology & Translational Science. 2023;6:88–91. DOI: 10.1021/acsptsci.2c00168. Suggested placement: Section 2.3 (C3 tetrazoles) and Discussion (albumin binding, charge/polarity). Sample linking sentence: “These C3/C7 motifs have documented associations with albumin binding—tetrazole positively and pyridinium/quaternary amines negatively—across marketed cephalosporins (Kanis et al., 2023)

We thank the reviewer for this valuable suggestion. The recommended reference by Kanis et al. (2023) has now been cited in the Discussion, where it was used to support and expand the interpretation of our structural observations in relation to clinically relevant plasma protein binding. This addition allowed us to address an important topic that previously represented a gap in the Discussion, particularly regarding the influence of C3 tetrazole and pyridinium/quaternary ammonium motifs on albumin binding and charge–polarity relationships.

❖ Third, figure captions would benefit from including the number of compounds per group and, where relevant, the percentage of compounds with oral formulations; a brief indication of the variance explained by the first two principal components and the main contributing descriptors would further help readers interpret the PCA plots.

Thank you for this valuable comment. We have revised the figure captions to include the number of compounds in each group, as suggested. In addition, we now report the percentage of data variance explained by PC1 and PC2, which facilitates the interpretation of the PCA results. These changes are highlighted in the current version of the manuscript.

 

❖ Clarify the grouping choice that combines fourth and fifth generations (already noted in figure captions); and note variability in generational labels for agents like ceftaroline/ceftobiprole (often termed “advanced” rather than strictly “fifth”).

We thank the reviewer for this important comment. The rationale for grouping fourth- and fifth-generation cephalosporins has now been more clearly justified in the Methodology section. Briefly, the number of agents classified strictly as fifth generation is limited, and maintaining a separate group would compromise the statistical robustness of comparative analyses across generations. For this reason, and considering the variability in generational labeling, the combined grouping was retained. The reviewer is kindly referred to Section 4.2 of the revised manuscript for the detailed explanation.

 

Conclusion remarks/suggestions

Overall, the study is concise and scientifically sound within its scope. The suggested changes are minor clarifications rather than substantive new work, and implementing them will improve the paper. I recommend acceptance after minor revision.

We sincerely thank the reviewer for this positive and encouraging evaluation. We greatly appreciate the recognition that the study is scientifically sound and that the suggested changes are limited to minor clarifications. This assessment is very important to us, and we believe that addressing these points has further strengthened the manuscript.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The authors of ddc-4064588 studied five generations of popular cephalosporins—which is one of the most important classes of β-lactam antibiotics. A total of 38 compounds were analyzed by molecular descriptors (MW, logP, TPSA, HBA, HBD, rotatable bonds and global complexity indices), then with statistical methods such as ANOVA and principal component analysis (PCA). The molecular descriptors were analyzed systematically according to generations of drugs, and the conclusions were made in association with popular Lipinski’s and Veber’s rules.

The methodologies are well chosen, which are available for general researchers. Interesting and convincing findings were provided. In addition, necessary background is given. The paper is also well written. I recommend the paper be accepted for publication in DDC with minor revisions.

Some comments on concepts and research directions:

1. Since different cephalosporines display distinct affinities for individual penicillin-binding proteins (PBPs), could the authors comment on this direction in connection with their findings?
2. The authors identified important substituents such as tetrazole groups at position C3, Oximes and aminothiazole group at position C7, would the author make a generalization for possible new drugs?
3. Could we use the descriptors to establish a machine learning strategy?    

Author Response

# Reviewer 3

The authors of ddc-4064588 studied five generations of popular cephalosporins—which is one of the most important classes of β-lactam antibiotics. A total of 38 compounds were analyzed by molecular descriptors (MW, logP, TPSA, HBA, HBD, rotatable bonds and global complexity indices), then with statistical methods such as ANOVA and principal component analysis (PCA). The molecular descriptors were analyzed systematically according to generations of drugs, and the conclusions were made in association with popular Lipinski’s and Veber’s rules.

The methodologies are well chosen, which are available for general researchers. Interesting and convincing findings were provided. In addition, necessary background is given. The paper is also well written. I recommend the paper be accepted for publication in DDC with minor revisions.

We thank the reviewer for the careful evaluation and the positive assessment of our manuscript. We appreciate the recognition of the methodological approach, clarity of presentation, and relevance of the findings

Some comments on concepts and research directions:

Since different cephalosporines display distinct affinities for individual penicillin-binding proteins (PBPs), could the authors comment on this direction in connection with their findings?

We thank the reviewer for this insightful comment. We agree that differential affinity for specific PBPs is a key determinant of cephalosporin activity and spectrum. While the present study focuses on global structural and physicochemical trends across generations, future studies will integrate pharmacodynamic data, including PBP selectivity, with the descriptors and framework established here.

 

The authors identified important substituents such as tetrazole groups at position C3, Oximes and aminothiazole group at position C7, would the author make a generalization for possible new drugs?

We thank the reviewer for this insightful comment. While our analysis highlights recurring substituents at the C3 and C7 positions that are associated with specific generational trends, we believe that generalizations for the design of new cephalosporins should be made with caution. The identified motifs reflect historical optimization rather than universal design rules. Therefore, in the present study, these substituents are discussed as trend markers in the evolution of cephalosporins, rather than as predictive features for the development of new drug candidates.

 

Could we use the descriptors to establish a machine learning strategy?  

We thank the reviewer for this insightful and forward-looking comment. The primary objective of the present study was to provide an overview—a snapshot of the physicochemical landscape of marketed cephalosporins across generations—using interpretable, unsupervised methods. In this context, PCA was intentionally employed as an exploratory tool to reveal global patterns and relationships within the chemical space, rather than to build predictive models. We fully agree that this is a very well-raised topic, and that the curated dataset presented here could certainly inspire future studies employing machine learning approaches, including expanded versions of this dataset. However, the implementation of supervised or more complex ML strategies would require the definition of specific endpoints, training and validation schemes, and performance metrics, which fall outside the scope of this concise, hypothesis-generating analysis. We have now highlighted this point in the Discussion as a perspective for future work.

Author Response File: Author Response.pdf