Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This review systematically discusses coupled, multi‑technology pipelines for next‑generation antibiotic discovery to tackle the global antimicrobial‑resistance crisis. It classifies and evaluates major modern approaches including high‑throughput screening, AI‑machine learning, genomics‑metagenomics, fragment‑ and structure‑based design, proteomics, and microfluidics. Representative pre‑clinical case examples such as cilagicin, halicin and abaucin are summarized to demonstrate the benefits of combining complementary tools over siloed single methods. The paper further outlines scientific, economic, regulatory and social barriers limiting antibiotic innovation, and advocates adopting co‑applied or sequentially coupled workflows to replenish the antibacterial lead pipeline. However, several conceptual, structural and presentation limitations weaken the overall manuscript. My concerns are listed bellowed:

1. Schematic figure logical flaws: Figure 1 uses only converging unidirectional arrows, falsely implying linear one‑step outputs; AI/ML is mis‑positioned as an independent parallel branch instead of a cross‑cutting enabling tool.
2. Several logical and textual issues exist in figure 2: (1) Word ambiguity: the verb “compounds” in the central box is misleading; “combines” is more appropriate to describe integrating multiple technical advantages. (2): Connectivity logic: the current lines imply all five guided strategies must be applied together, whereas the actual meaning is selective co-application of complementary approaches. (3) The five discovery outcomes are presented as mandatory outputs, while they represent potential but not simultaneous results. (4) Concept overlap: Activity-guided and Phenotypic-guided approaches are partially overlapping and should be rephrased.
3. Limited critical comparison: It lacks head‑to‑head quantitative comparison between “co‑application” versus “sequential application”; the proposed self‑defined classification is not benchmarked against established literature frameworks.
4. Insufficient translational discussion: While listing many pre‑clinical candidates, it provides shallow analysis on why most coupled‑approach leads still fail to advance into later clinical phases.
5. Minor redundancy: Tables 1‑2 and main text repeat large volumes of candidate‑agent information; limited depth on real‑world practical costs and infrastructure barriers for laboratories implementing these coupled workflows.
6. The reference list exhibits inconsistent formatting; citations should be standardized following the journal’s reference style requirements.

Author Response

We thank the reviewers for their thoughtful and helpful comments. We respond to each point below; their suggestions have undoubtedly improved the paper, for which we are very grateful. We highlighted the new sections in yellow in the revised manuscript.

Comment 1

Schematic figure logical flaws: Figure 1 uses only converging unidirectional arrows, falsely implying linear one‑step outputs; AI/ML is mis‑positioned as an independent parallel branch instead of a cross‑cutting enabling tool.

Response: We thank the reviewer for the valuable comment. We revised Figure 1 to represent antibiotic discovery as an integrated, iterative process. To illustrate the connections between the different approaches, we have modified the figure to incorporate bidirectional connections and feedback loops. In addition, AI/ML now supports data integration, screening, target selection, and prediction, rather than serving as an independent strategy.

Comment 2

Several logical and textual issues exist in figure 2: (1) Word ambiguity: the verb “compounds” in the central box is misleading; “combines” is more appropriate to describe integrating multiple technical advantages. (2): Connectivity logic: the current lines imply all five guided strategies must be applied together, whereas the actual meaning is selective co-application of complementary approaches. (3) The five discovery outcomes are presented as mandatory outputs, while they represent potential but not simultaneous results. (4) Concept overlap: Activity-guided and Phenotypic-guided approaches are partially overlapping and should be rephrased.

Response:  We thank the reviewer for these constructive observations and helpful suggestions. We agree that the original Figure 2 could lead to unintended interpretations of the scheme; therefore, we revised and edited it accordingly. The changes are visible in the submitted version of Figure 2.

First (1), the wording in the central box was changed from “compounds their strengths…etc” to “combines their strengths… etc”, resulting in the following statement: Co-application of complementary technologies combines their individual strengths and helps overcome their individual limitations.”

Second (2), Figure 2 was updated to specifically clarify that these approaches are illustrative examples that could be selectively coupled based on the discovery routes and objectives, rather than all approaches should be applied together. We changed the heading to (Example of Guided Discovery Approaches; illustrative, not exhaustive). Further, we added a longitudinal line next to Sequence and activity-guided approaches to indicate that these two approaches were applied in a worked example of Cilagicin. 

Third (3), we strongly agree that Discovery Outcomes should not appear as mandatory outcomes of applying coupled approaches; thus, we changed the heading to (Potential Discovery outcomes; may lead to one or more of these outcomes) and changed the lines to dashed lines.

Finally (4),  to address the overlap between the two different approaches (Activity and Phenotypic-guided approaches), we removed the Phenotypic approach. We kept the Activity approach, as this approach is consistent with the worked example of Cilagicin.

We added these modifications to the revised Figure 2 in the revised manuscript.

Comment 3

Limited critical comparison: It lacks head‑to‑head quantitative comparison between “co‑application” versus “sequential application”; the proposed self‑defined classification is not benchmarked against established literature frameworks.

Response: We thank the reviewer for this very important observation. We agree that this was a self-proposed/defined classification, and distinguishing the “co-application” approach from the “sequential-application” approach would require substantial evidence and data not currently supported by the literature.  Therefore, to avoid introducing an unclarified and unvalidated classification, we removed the term “sequential application” and refocused on “co-application” as the only approach in antibiotic discovery.

We removed the corresponding definitions and references to the “sequential approach” from the Abstract and replaced them with the co-application approach in Section 2 and the Conclusion. The WHO table no longer presents evidence of “sequential application”; instead, it includes examples of innovative agents in clinical pipelines, followed by the co-application of available technologies in the antibiotic discovery pipeline. We highlighted all edited sentences in the revised manuscript.

Comment 4

Insufficient translational discussion: While listing many pre‑clinical candidates, it provides shallow analysis on why most coupled‑approach leads still fail to advance into later clinical phases.

Response: We appreciate the reviewer's comment. We agree that a discussion of translational challenges and barriers to antibiotic development is needed for a more balanced evaluation of co-application. Therefore, we extended the manuscript to discuss factors limiting antibiotic development, especially for novel antibiotic candidates. The revised manuscript emphasizes barriers such as safety concerns, lengthy and costly timelines, and low discovery-to-market rates in antibiotic development. Most importantly, we highlight that while complementary co-application may improve hit identification, advancing these complementary technologies does not necessarily translate that identification into clinical success, as a substantial attrition rate remains.

Comment 5

Minor redundancy: Tables 1‑2 and main text repeat large volumes of candidate‑agent information; limited depth on real‑world practical costs and infrastructure barriers for laboratories implementing these coupled workflows.

Response: We thank the reviewer for this valuable suggestion. We revised the manuscript to reduce repetition of antibiotic candidates already mentioned in Tables 1 and 2. We also expanded the challenges section to discuss infrastructure and resource limitations in laboratories, from funding barriers to limited computational and data resources to the need for highly skilled practitioners. We also acknowledged that applying these technologies depends on institutional infrastructure and access.

Comment 6

The reference list exhibits inconsistent formatting; citations should be standardized following the journal’s reference style requirements.

Response: We replaced references [20] and [25]; however, we removed references [67], [139], and [164] from the previous manuscript version. In addition, we removed the references from the Conclusion section and moved them to the main text, specifically to Section 3.1, “Historical Pipeline Decline and the Rediscovery Problem.”

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript addresses an important topic, integrating multiple technologies ("coupled approaches") in antibiotic discovery. However, despite its broad coverage, several substantial issues limit its suitability for publication in its current form.

1.The manuscript needs a brief description in a short, separate section of the literature search methodology (the databases searched, search strategy, inclusion and exclusion criteria, time period covered, and the number of studies screened and included).

2.The novelty of the proposed "co-application" and "sequential application" should be clarified, highlighting how these concepts differ from existing antibiotic discovery approaches.

3.The manuscript needs a more critical discussion of the limitations of integrated discovery strategies (e.g. failure rates, translational challenges, unsuccessful examples, and comparative performance data where available).

4.Without quantitative evidence, many conclusions remain largely speculative (e.g. the paper generally assumes that coupled approaches will outperform individual methodologies).

5.Generaly, many topics receive only a brief approach because too many topics are included in a single manuscript. The manuscript is slightly unbalanced in this respect.

6.Several statements require verification: "No new clinical antibiotics have been discovered since HTS was implemented in the 1980s" (Section 4.8). The statement that coupled approaches increase hit rates by "at least 14-fold" is based on a specific study.

7.Some descriptions of development status and clinical phases should be updated (e.g. cefepime-zidebactam, zoliflodacin, which are already approved).

8.I consider that several sections could be condensed substantially:

-the Introduction section contains repeated discussions of antimicrobial resistance, the decline in antibiotic discovery, and the need for innovative strategies

-the technology-specific sections (HTS, genomics, proteomics, metabolomics, and metagenomics) that provide extensive methodological descriptions

-the AI/ML section is particularly overextended, with repeated references (such as halicin and abaucin) and recurring explanations of AI advantages and limitations

-the same case studies (halicin and abaucin) are discussed across multiple sections, leading to unnecessary duplication (these become redundant)

-the section on the benefits of coupled approaches reiterates points already made throughout the manuscript

-the Future Perspectives section should be more focused by providing specific future directions and priorities (less general statements about interdisciplinarity and innovation)

9.Tables are missing legends at the front.

10.The abbreviations are used inconsistently, with some defined multiple times or used before being defined (e.g., AMR throughout the manuscript).

11.Please revise the inconsistent capitalisation of (some examples):

-Artificial Intelligence (AI) and artificial intelligence (AI)

-Machine Learning (ML) and machine learning

-High-Throughput Screening (HTS) and high-throughput screening or High-throughput Screening

-Generative AI and generative AI

-CRISPR/Cas9 and crispr/cas9 or CRISPR-cas9

-Inconsistent capitalisation of section titles (e.g. "Future Perspectives", "Future perspectives", "Future Perspectives and Challenges")

-Drug Discovery Pipeline in one table or figure and drug discovery pipeline in the main text

12.Please revise the variability in compound, bacterial species, and gene/protein nomenclature formatting (e.g., italics used inconsistently) throughout the manuscript.

Author Response

We thank the reviewers for their thoughtful and helpful comments. We respond to each point below; their suggestions have undoubtedly improved the paper, for which we are very grateful. We highlighted the new sections in yellow in the revised manuscript.

1. The manuscript needs a brief description in a short, separate section of the literature search methodology (the databases searched, search strategy, inclusion and exclusion criteria, time period covered, and the number of studies screened and included).

Response: We thank the reviewer for this constructive comment. Following this suggestion, we added a separate paragraph in the Introduction section describing the search methodology for this narrative review. It describes the searched databases, the included and excluded publications, and the timeframe covered. Because this is a narrative review, we did not conduct quantitative counts of selected publications; we also clarified this in the revised manuscript.

2. The novelty of the proposed "co-application" and "sequential application" should be clarified, highlighting how these concepts differ from existing antibiotic discovery approaches.

Response: We thank the reviewer for this very important observation. In the revised version of the manuscript, the term “co-application” is used to describe the simultaneous or coordinated use of technologies to achieve a complementary approach within the early-discovery stage of an antibiotic, where each technology contributes directly to performing a specific function in candidate identification, mechanism of action determination, or candidate or antibiotic target characterization.

We believe this was a self-proposed classification, and distinguishing the “co-application” approach from the “sequential-application” approach would require substantial evidence and data not currently supported by the literature. Therefore, to avoid introducing an unclarified and unvalidated classification, we removed the term “sequential application” and refocused on “co-application” as the only approach in antibiotic discovery.

We removed the corresponding definitions and references to the “sequential approach” from the Abstract and kept them in the co-application approach, Section 2, and the Conclusion. The WHO table no longer presents evidence of “sequential application”; instead, it includes examples of innovative agents in clinical pipelines, followed by the co-application of available technologies in the antibiotic discovery pipeline. We highlighted all edited sentences in the revised manuscript.

3. The manuscript needs a more critical discussion of the limitations of integrated discovery strategies (e.g. failure rates, translational challenges, unsuccessful examples, and comparative performance data where available).

Response: Thank you for this important comment. We agree that the previous version of the manuscript did not sufficiently address the limitations and translational challenges associated with integrating complementary discovery strategies. We now discuss that integrating multiple discovery technologies does not necessarily improve clinical success or reduce failure rates. We added quantitative examples illustrating the substantial failure rates in antibacterial discovery and development, while clarifying that these historical data should not be interpreted as direct failure rates for integrated strategies. We also discuss unsuccessful or discontinued discovery programs and highlight the gap between discovery-stage performance and progression toward clinical candidates. Furthermore, the revised manuscript addresses key translational barriers that may remain despite successful target identification or hit discovery, including bacterial penetration and permeability, pharmacokinetic and pharmacodynamic properties, tissue and infection-site exposure, toxicity, host-pathogen interactions, resistance development, and limitations of preclinical models. We also discuss practical limitations associated with integrating multiple technologies, including data compatibility, reproducibility, computational prediction errors, experimental validation requirements, increased workflow complexity, cost, and the need for specialized expertise.

4. Without quantitative evidence, many conclusions remain largely speculative (e.g. the paper generally assumes that coupled approaches will outperform individual methodologies).

Response: We thank the reviewer for this important comment. We agree that the available literature does not provide sufficient quantitative evidence to demonstrate whether the coupled approach outperforms the single approach in antibiotic discovery. Accordingly, we clarified that the coupled approach is a proposed strategy to improve hit-rate identification and does not guarantee clinical success or outperform the single approach. The literature has not yet demonstrated improvements in hit rates, reduced development time, or lower costs. We also clarified that candidate antibiotics may fail for several attrition reasons, such as safety concerns or unfavorable pharmacokinetics.

5. Generally, many topics receive only a brief approach because too many topics are included in a single manuscript. The manuscript is slightly unbalanced in this respect.

Response: We thank the reviewer for this valuable comment. We agree that the manuscript's scope was not clearly defined, resulting in uneven coverage of some topics. To improve the review's scope and focus, we removed the ‘Traditional Approach in Antibiotic Section’ because these approaches did not contribute to the review's central argument. We also added a scope statement at the end of the Introduction to clarify that the review focuses on co-applied or coupled approaches in next-generation antibiotic discovery and discusses individual technologies in detail when needed. In addition, this review aimed to identify and review the main antibiotic discovery technologies used in integration with other technologies or approaches.

6. Several statements require verification: "No new clinical antibiotics have been discovered since HTS was implemented in the 1980s" (Section 4.8). The statement that coupled approaches increase hit rates by "at least 14-fold" is based on a specific study.

Response: We thank the reviewer for highlighting these two statements. We agree that neither is adequately supported by the available literature; therefore, we removed both from the revised manuscript.

1. We removed the claim that “no new clinical antibiotics have been discovered since HTS was implemented in the 1980s.” We acknowledge that this statement was too absolute, as some agents have reached the clinics. The sentence now focuses only on HTS challenges.
2. We removed the quantitative statement that “Combining more than two technologies has shown a promising approach that increases the hit rate of previous antibiotic drug discovery efforts by at least 14-fold.” The section no longer claims a measurable advantage, consistent with Section 5 and the Conclusion that the measurable advantages of the coupled approaches have not yet been confirmed in the literature.

 

7. Some descriptions of development status and clinical phases should be updated (e.g. cefepime-zidebactam, zoliflodacin, which are already approved).

Response: We thank the reviewer for this helpful observation. We have updated the developmental status of these two agents. The two antibiotics received FDA approval recently: zoliflodacin was approved in December 2025 as Nuzolvence, and cefepime–zidebactam was approved on 29 May 2026 (Zaynich, for complicated urinary tract infections). We revised all statements accordingly, as well as Table 2. The FDA approval of a first-in-class agent in Section 3.3 also reinforced our discussion of innovative agents in antibiotic discovery.

8. I consider that several sections could be condensed substantially:

- the Introduction section contains repeated discussions of antimicrobial resistance, the decline in antibiotic discovery, and the need for innovative strategies

- the technology-specific sections (HTS, genomics, proteomics, metabolomics, and metagenomics) that provide extensive methodological descriptions

- the AI/ML section is particularly overextended, with repeated references (such as halicin and abaucin) and recurring explanations of AI advantages and limitations

- the same case studies (halicin and abaucin) are discussed across multiple sections, leading to unnecessary duplication (these become redundant)

- the section on the benefits of coupled approaches reiterates points already made throughout the manuscript

- the Future Perspectives section should be more focused by providing specific future directions and priorities (less general statements about interdisciplinarity and innovation)

Response: We sincerely thank the reviewer for this constructive suggestion, which helped us improve the manuscript's clarity and focus. We fully agree that several sections contained material not essential to the central argument, and we have substantially condensed the text.

In addition to shortening the Introduction and Section 3.1, we have removed four sections in their entirety:

1. the subsection on traditional approaches in antibiotic discovery (former Section 3.2);
2. the subsection on proteomics for drug target discovery and drug repurposing (former Section 4.2.2);
3. the individual case descriptions of halicin and abaucin (former Sections 4.1.4 and 4.1.5);
4. the subsection on coupling more than one technology (former Section 4.8).

The halicin and abaucin cases are already summarized in Table 1 and discussed in the main text where relevant, so their separate descriptions were duplicative. The remaining sections presented background or peripheral material not needed to support the review’s focus on coupled discovery approaches.

We updated the section headings and numbering accordingly. We believe the revised manuscript is considerably more concise while retaining all content essential to its argument, and we are grateful to the reviewer for prompting these improvements.

9. Tables are missing legends at the front.

Response: We thank the reviewer for this comment. We have now clearly presented the legends at the front of the tables. We have listed abbreviations beneath the table.

10. The abbreviations are used inconsistently, with some defined multiple times or used before being defined (e.g., AMR throughout the manuscript).

Response: We thank the reviewer for this careful observation.  We have systematically reviewed all abbreviations throughout the manuscript and removed duplicate definitions. Each abbreviation now appears only once, at its first use. Specifically, the following abbreviations: AMR, HTS, MoA, and DL.

11. Please revise the inconsistent capitalisation of (some examples):

-Artificial Intelligence (AI) and artificial intelligence (AI)

-Machine Learning (ML) and machine learning

-High-Throughput Screening (HTS) and high-throughput screening or High-throughput Screening

-Generative AI and generative AI

-CRISPR/Cas9 and crispr/cas9 or CRISPR-cas9

-Inconsistent capitalisation of section titles (e.g. "Future Perspectives", "Future perspectives", "Future Perspectives and Challenges")

-Drug Discovery Pipeline in one table or figure and drug discovery pipeline in the main text

Response: We thank the reviewer for these careful readings. We have revised capitalization throughout the manuscript and now apply a single convention. Technical terms such as artificial intelligence (AI), machine learning (ML), deep learning (DL), high-throughput screening (HTS), and generative AI appear in lowercase in the running text and in title case only in section headings when applicable.

12. Please revise the variability in compound, bacterial species, and gene/protein nomenclature formatting (e.g., italics used inconsistently) throughout the manuscript.

Response: We thank the reviewer for identifying this inconsistency. We have standardized nomenclature formatting throughout the manuscript, in the main text and in tables.  All bacterial genus and species names are now italicized at every occurrence, including abbreviated forms (e.g., A. baumannii, E. coli, S. pneumoniae) and higher taxonomic names (Enterobacteriaceae, Enterobacterales).

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have fully addressed all my concerns raised in the first round of review. I have carefully checked the revised manuscript and the point-by-point response letter. All revisions are adequate and appropriate. The manuscript has been substantially improved and meets the journal’s publication requirements. I recommend accept for this manuscript in its current form.

Author Response

Thank you for your valuable comments 

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have appropriately addressed the comments and have made revisions that have enhanced the quality of the manuscript. I appreciate all their efforts.

Author Response

Thank you for your valuable comments