Antibacterial drug discovery: challenges and preclinical promises from synthetic small molecules

Abstract

Over the past few decades, antibacterial discovery has changed substantially, yet innovation has remained slow despite the growing need for new agents active against multidrug-resistant pathogens. Progress is constrained by scientific hurdles, including target selection, penetration and efflux in bacteria, and the difficulty of translating in vitro activity into in vivo efficacy, as well as persistent economic disincentives. Here, we discuss these challenges and summarise recent advances in the field, with a medicinal chemistry focus on synthetic small molecules that have reached validated-lead or preclinical development and shown in vivo antibacterial efficacy. We also provide an outline of the current clinical pipeline, highlighting areas of innovation and remaining gaps. Our goal is to offer a comprehensive perspective on antibacterial discovery that supports ongoing efforts to strengthen the pipeline in response to antimicrobial resistance.

1. Introduction

Infectious diseases have shaped human health throughout history, with bacterial, fungal, viral, and parasitic pathogens posing recurring challenges.¹ The discovery of antimicrobials in the 20th century was a major shift, as antibiotics markedly reduced illness and death from bacterial infections.^2–4 Before the 1940s, infections such as tuberculosis (TB) and pneumonia were often fatal. With the introduction of antibiotics, mortality from many bacterial diseases fell sharply, contributing to longer life expectancy.⁵ This progress is now being challenged by the growing problem of antimicrobial resistance (AMR).⁶

Resistance develops naturally, but inappropriate antibiotic use in human health, animal husbandry, and agriculture has accelerated its emergence and spread.^7–10 If this trend continues, infections could increasingly complicate routine procedures such as surgery and immunosuppressive treatments, and make the management of common infections more difficult and costly.^11,12 This is reflected in the growing prevalence of multidrug-resistant (MDR) bacteria, including the ‘ESKAPE’ pathogens.¹³ This group includes Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp. The alarming propensity of these pathogens to develop resistance to multiple antibiotics poses significant challenges in healthcare settings.^14,15 ESKAPE pathogens can rapidly develop resistance to antibiotics in development, with mutations emerging within just 60 days, suggesting that new antibiotics may become ineffective as quickly as those already in use.¹⁶

To frame the scale of AMR, a recent analysis of data from 204 countries and territories showed that it is now a leading cause of death worldwide.⁶ It directly caused over 1.4 million deaths, exceeding deaths from HIV and malaria, and projections estimate around 1.9 million deaths in 2050.^17–20 The burden is highest in low-income regions, but remains substantial in high-income settings: in sub-Saharan Africa AMR caused 24 deaths per 100,000 individuals, compared with 13 per 100,000 in high-income countries.⁶ In the US, AMR accounts for over 2.8 million infections and more than 35,000 deaths each year,²¹ while the EU/EEA reports roughly 670,000 infections and around 33,000 associated deaths annually.²² More recent modelling suggests that improved infection control and faster drug development could avert up to 92 million deaths between 2025 and 2050,¹⁷ and the COVID-19 response provides a reminder of how rapidly R&D can move when collaboration, data sharing, clinical access, and regulatory pathways are aligned.^23,24

The burden of AMR continues to rise, yet antibiotic innovation has remained slow.²⁵ Over the past six decades, only five new antibiotic classes have reached clinical approval (Figure 1).^5,26,27 This limited progress is concerning given the growing clinical and economic costs of drug-resistant infections, and it reinforces the need to prioritise the development of new antibiotics, particularly those with new mechanisms of action.^28,29

Figure 1. Timeline showing the introduction of new antibiotic classes to clinical use by decade excluding antitubercular agents.

Mechanisms of action of traditional antibiotics 1. Cell wall synthesis; 2. Protein synthesis; 3. DNA/RNA synthesis and replication; 4. Folic acid synthesis; 5. Cell membrane synthesis.

The current antibacterial discovery and development landscape is increasingly shaped by rising resistance and the limited pace of new drug classes.^14,30 According to the 2023 World Health Organization (WHO) report, only 97 antibacterial agents are currently in the clinical pipeline (57 traditional and 40 non-traditional), with merely four of the traditional agents demonstrating activity against at least one WHO-classified ‘critical priority’ pathogen.³¹ Moreover, many of these candidates are derivatives of existing antibiotic classes, where established resistance mechanisms may limit their differentiation and long-term clinical impact. The limited number of new antibiotics is underscored by the long gap since the discovery of major new classes; for example, the lipopeptide class was identified in the 1980s, and daptomycin was the first to reach the market in 2003.²⁶ In the following years, progress in bringing truly new antibacterials to market has remained limited. Between 2017 and 2023, cefiderocol, a siderophore cephalosporin, was one of the few approvals with activity against WHO priority pathogens.³² Access to cefiderocol has also been uneven, with use concentrated largely in higher-income nations.

The term ‘antibiotic’ was originally used to describe antagonistic interactions between microorganisms and later became associated with natural products from bacteria and fungi that inhibit or kill other microbes.³³ In modern usage, it can also include semi-synthetic derivatives and fully synthetic agents with antibacterial activity. In this review, we use the terms “antibiotic” and “antibacterial” interchangeably to describe synthetic small molecules with direct activity against bacteria, whether bacteriostatic or bactericidal. Where relevant, we also cover non-traditional antibacterial agents, such as anti-virulence compounds and β-lactamase inhibitors, that act as adjuncts by enhancing the activity of direct-acting agents. We treat these as distinct from antibiotics in the strict sense and refer to them as non-traditional antibacterials.

In the complex process of drug development, identifying a validated lead compound during the pre-clinical stage is a critical milestone.^34,35 Numerous compounds exhibiting promising antibacterial activity in vitro are reported in the scientific literature every year. Nevertheless, only a small proportion of these compounds proceed to the next stages, where their efficacy is rigorously evaluated in animal models. This phase is crucial in the development process, as it determines whether the initial hits identified in vitro possess the necessary characteristics to effectively combat infections in vivo and ultimately transition to clinical trials.³⁶ Preclinical studies have two primary objectives: firstly, to accurately replicate the intended biological effects of a compound in animal models (efficacy); and secondly, to systematically identify and characterise all potential toxicities associated with the drug candidate (safety).^37,38

In this review, we focus on a medicinal chemistry view of synthetic antibacterial small molecules with preclinical promise, which we believe has been underrepresented in existing literature. We first summarise key challenges in antibacterial discovery and current strategies for identifying new small-molecule leads. We then provide a brief overview of candidates in clinical development, before highlighting validated-lead and preclinical synthetic compounds that have shown efficacy in relevant in vivo infection models. It is worth noting that compounds discovered for TB are excluded from this review, as they have been extensively reviewed elsewhere.³⁹ Rather than providing an exhaustive account of antibacterial agents, this review focuses on recent synthetic small molecules with demonstrated in vivo efficacy and clear preclinical potential. Where applicable, we direct readers to relevant existing literature for a more comprehensive understanding.

2. Challenges and Risks in Antibiotic Drug Discovery

2.1. The burden and consequences of AMR

By the late 20th century, antibiotics had become embedded in routine care, enabling interventions such as surgery, intensive care, transplantation, and chemotherapy that depend on reliable infection control.^40,41 As antibiotic use expanded, resistant strains emerged and spread, increasingly limiting treatment options and creating wider public health and economic pressures.¹² This has driven coordinated action plans across scientific, medical, and policy communities to slow resistance and rebuild the antibacterial pipeline.^42,43

The consequences of AMR extend far beyond treatment failure. The World Bank estimated that, between 2015 and 2050, AMR could impose major economic burdens through its effects on healthcare systems, trade, and livestock production, with annual healthcare costs rising by around $330 billion even under managed scenarios and approaching $1.2 trillion if resistance worsens.⁴⁴

More recent analyses similarly project substantial losses in life expectancy, rising healthcare expenditures, and major productivity declines in the coming decades, while emphasising that sustained financing will be required to contain AMR and revitalise antibiotic R&D.⁴⁵ Together, these estimates underscore that AMR is not only a clinical and microbiological problem, but also a systemic threat to health systems, economic stability, and the long-term sustainability of modern medicine.⁴⁶

2.2. Market failure and private-sector withdrawal

Developing a new antibiotic can cost over $1.5 billion, yet stewardship and short treatment courses often translate into modest and unpredictable returns.⁴⁷ An analysis estimated a net present value of around $50 million for a new antibiotic, compared with roughly $1 billion for a neuromuscular drug.^12,48 At the same time, resistance can erode the clinical utility of new agents within a few years, creating a tension between conserving new antibiotics through restricted use and generating sufficient returns to sustain investment.

These conditions have contributed to a prolonged weakening of private-sector engagement in antibacterial R&D. Over the past two decades, relatively few truly novel antibiotics have emerged, reflecting persistently weak economic incentives for sustained innovation.^49,50 A 2013 analysis reported that 15 of 18 large pharmaceutical companies had exited antibiotic R&D, and investment has continued to decline.⁴⁸ Consistent with this, the number of infectious disease clinical trials initiated has fallen over the past decade, while overall programme success rates remain lower than in several therapeutic areas.⁵¹ As a result, only a few dozen antibacterial trials have been active in recent years,²⁵ in sharp contrast to fields such as oncology, where thousands of agents are in development.⁵² Together, these dynamics illustrate a structural market failure: antibiotics generate substantial societal value, but limited and uncertain commercial returns, particularly for new agents reserved for MDR infections.⁵³

2.3. Underinvestment and policy responses

Antibacterial R&D has remained underfunded compared with other therapeutic areas, particularly oncology, which has attracted far greater private investment over the past decade. This disparity is evident not only in private financing, including a reported 44-fold funding gap in 2020, but also in public markets, where antibacterial companies accounted for far fewer IPOs than oncology companies between 2011 and 2020.⁵⁴ In response, a range of legislative and policy measures has been introduced to improve the viability of antibiotic development. One prominent example is the 2012 US GAIN Act, which established the QIDP designation and associated incentives such as Fast Track, Priority Review and additional market exclusivity.⁵⁵ More recently, the antibiotic pipeline has shown limited signs of recovery, driven in part by increased direct public funding. Global investment in antibiotics rose by 25% between 2017 and 2022, reaching $2.08 billion, with much of this funding directed toward therapeutic innovation.⁵⁶

2.4. Push and pull incentives to rebuild the pipeline

Sustained progress will require recognising that new antibiotics deliver societal value that is not captured by sales-driven markets, and that this mismatch must be addressed through public investment and new incentive models.⁵⁷ One emerging approach is the use of subscription-style or value-based reimbursement models that delink revenue from prescription volume. The UK is piloting such a system, in which companies are paid annually based on an antibiotic’s assessed value rather than sales.^58,59 In the US, the proposed PASTEUR Act adopts a similar value-based model, with $750 million to $3 billion in payments for qualifying antibiotics in exchange for commitments on availability, resistance monitoring and reporting, stewardship support, international registration, and completion of required post-approval studies over a multi-year contract.⁶⁰

These mechanisms are examples of pull incentives, which aim to make successful antibacterial R&D commercially viable by providing large, predictable rewards after development.⁵³ Their main appeal is that they reduce the dependence of returns on sales volume, thereby better aligning commercial incentives with stewardship goals. Subscription-style models, in particular, have attracted attention because they can provide more stable returns while discouraging overuse.⁶¹ Modelling studies further suggest that strong pull incentives could expand the late-stage pipeline, whereas their absence is likely to reduce advanced funding and shrink the number of candidates reaching the clinic over time.⁶²

By contrast, push incentives reduce the upfront risk of antibiotic R&D through grants, subsidies, and public or philanthropic support. These mechanisms are often implemented through public–private partnerships spanning early discovery to clinical development. CARB-X, BARDA, GARDP, and the AMR Action Fund illustrate different parts of this landscape, supporting programmes from preclinical research to later-stage development.⁶³ Together, these initiatives help sustain entry into the pipeline, while pull incentives are intended to ensure that successful products remain financially viable once developed. A durable rebuilding of the antibacterial pipeline will likely require both approaches in combination.^60,64–66

2.5. Sustaining the antibiotic discovery ecosystem

Beyond financial incentives, sustaining antibiotic discovery requires institutional models that preserve expertise, share risk, and maintain continuity across a long and fragile innovation pathway. Several initiatives are therefore exploring more open models for antibiotic R&D, in which data, ideas, and early-stage outputs are shared more transparently and parts of discovery are conducted in the public domain.^67,68 Such approaches, may complement push and pull incentives by reducing duplication, widening participation, and helping promising assets progress when conventional funding pathways are poorly aligned.⁶⁹ However, open models do not remove the need for substantial long-term funding and dedicated sponsorship, particularly in a field where market incentives remain weak and drug development is inherently costly and complex.

Sustaining the ecosystem also depends on maintaining core expertise in antibacterial R&D and supporting the next generation of researchers.⁴³ A recent AMR Industry Alliance report highlights a growing skills and workforce gap, noting that the AMR R&D community is relatively small and has shrunk over time.⁷⁰ It also identifies declining authorship, reduced research output, and weaker patenting activity relative to areas such as cancer and HIV/AIDS. Together, these trends suggest that the challenge is not only a shortage of compounds in the pipeline, but also a gradual erosion of the scientific capacity and infrastructure needed to sustain antibiotic discovery over the long term.

Academic Drug Discovery Centres (ADDCs) have expanded in response to reduced pharmaceutical involvement in early-stage discovery, including antibiotics.^71,72 These university-based centres aim to translate academic innovation into credible drug candidates until commercial investment becomes feasible. They also offer a way for industry to engage with higher-risk early discovery, but their impact in antibiotics remains to be proven, given long timelines, the difficulty of translation, and the challenge of sustained funding.⁷³ At present, ADDCs are best viewed as promising components of an evolving discovery ecosystem rather than as fully validated solutions.

Ultimately, rebuilding the antibiotic pipeline will require more than correcting individual market failures. It will depend on sustaining the people, institutions, and collaborative structures that allow discovery programmes to emerge, mature, and survive across long development timelines. In this sense, push and pull incentives are necessary, but they are unlikely to be sufficient unless accompanied by continued investment in scientific capacity, translational infrastructure, and durable cross-sector collaboration.

3. Modern Approaches for Uncovering New Antibiotics

3.1. Natural products and improved access to microbial diversity

Natural products from bacteria and fungi have historically been a major source of antibiotics and remain central to the field.⁷⁴ Most clinically important antibiotic classes trace back to natural scaffolds, while a smaller number, such as sulphonamides, fluoroquinolones, and oxazolidinones, are fully synthetic.⁷⁵ Natural products often offer greater structural diversity and complexity,^76,77 which continues to motivate efforts to explore less studied microbial sources for new antibacterial leads.^5,78,79 In parallel, advances in isolation, cultivation, and screening methods have expanded access to previously inaccessible microbial diversity and metabolite space, enabling discoveries such as teixobactin and clovibactin.^80,81

3.2. Peptide-based antibiotics

Peptide-based antibacterials remain an important source of renewed interest in antibiotic discovery because they span multiple structural classes and mechanisms of action. Antimicrobial peptides (AMPs), in particular, have attracted attention because of their broad antibacterial activity and, in some contexts, a lower apparent propensity for resistance emergence.^82–85 Many AMPs are short, often cationic peptides that interact with negatively charged bacterial membranes, although peptide-based antibacterials are not limited to membrane-disrupting mechanisms.⁸⁶ A small number of peptide-derived or peptide-based antibacterials have reached clinical use, and additional candidates remain in development.⁸⁷

Beyond classical AMPs, ribosomally synthesised and post-translationally modified peptides are emerging as a valuable source structurally and mechanistically distinct antibacterial scaffolds.⁸⁸ Lariocidin, a lasso peptide from Paenibacillus sp. M2, illustrates this potential: it binds a previously unexploited site on the bacterial ribosome, disrupts protein synthesis, and shows broad activity, including against MDR A. baumannii, together with low cytotoxicity and in vivo efficacy.⁸⁹ More broadly, its discovery highlights how even accessible environmental sources can continue to yield chemically novel peptide-based antibacterial leads.

3.3. Bacteriophage-based therapies

Phage therapy has a long history and has recently attracted renewed interest.^90,91 Bacteriophages are viruses that infect and kill bacteria, typically with a high degree of specificity.⁹² This selectivity can be advantageous, and phages also offer distinctive practical features, including self-amplification at the sites of infection and adaptability to different routes of administration.^92,93 In some settings, phages can also act synergistically with antibiotics in some settings, improving activity against particular pathogens.⁹² At the same time, the narrow host range of many phages means that effective treatment often depends on accurate strain identification, while regulatory, manufacturing, and quality-control requirements remain major barriers to broader clinical adoption.⁹² Together, these features make phage therapy a promising but operationally complex antibacterial approach.

3.4. Drug repurposing as a source of antibacterial starting points

Drug repurposing is often discussed as a route to identify new antibacterial activities from existing drugs, with the theoretical advantage of leveraging prior pharmacological and safety knowledge.^94,95 In practice, however, repurposing has not yet produced an approved anti-infective, and there is continued scepticism about whether it can reliably deliver clinically useful antibacterials. Many compounds flagged in screens fall short because the exposures needed for antibacterial activity are not achievable or tolerable, or because key properties such as permeability and PK at infection sites are unfavourable. Nevertheless, systematic profiling of approved and investigational drugs can still be highly informative. For example, a large study across genetically diverse E. coli strains identified 186 non-antibiotic drugs with measurable antibacterial effects, providing a useful resource for mechanistic insight, hypothesis generation, and, in selected cases, lead-inspired optimisation.⁹⁶ In this sense, repurposing may be more valuable as a discovery-enabling strategy than as a dependable source of directly translatable antibacterial therapies.

3.5. AI-enabled antibacterial discovery and molecular design

Traditional antibiotic discovery has often relied on screening large compound collections for antibacterial activity, either phenotypically or against defined targets, but both approaches face the practical challenge of exploring an enormous chemical space.⁹⁷ These campaigns can be slow and expensive, with many tested molecules ultimately lacking the right combination of potency and drug-like properties. Recent advances in artificial intelligence (AI) and machine learning offer a way to prioritise smaller, higher-quality sets for synthesis and testing.^98–105

A notable example is halicin, which was discovered using a deep-learning model trained on E. coli growth-inhibition data from 2,335 compounds through a graph neural network approach. The system ranked vast chemical libraries (including ~100 million molecules) for predicted antibacterial activity and prioritised novel chemotypes (i.e., low similarity to known antibiotics). That virtual screening, run in a few days, identified halicin as a broad-spectrum candidate while cutting both time and screening costs.¹⁰⁶

The same group later reported abaucin, a narrow-spectrum antibiotic active against A. baumannii. They first screened approximately 7,500 molecules to generate growth-inhibition data, then trained an ensemble message-passing neural network to score a library of 6,680 compounds. High-scoring, structurally novel candidates were prioritised for testing, leading to the discovery of abaucin, which disrupts LolE-mediated lipoprotein trafficking and effectively suppressed A. baumannii infection in a mouse wound model.¹⁰⁷ More recently, generative AI approaches have been used to propose novel antibacterial candidates from very large make-on-demand spaces, with a subset of synthesised molecules showing activity against A. baumannii and other pathogens.¹⁰⁸

Beyond these AI-driven approaches, quantum and hybrid quantum–classical methods are also being explored for selected problems in molecular modelling and ligand design, although their practical value for antibacterial discovery remains prospective and will depend on further validation and technological maturation.^109–111

3.6. Screening strategies and target deconvolution

Traditionally, antibiotic discovery has relied on two broad screening paradigms: phenotypic screening, which identifies compounds based on their effects on bacterial growth or survival, and target-based screening, which starts from a defined molecular target. These approaches can be applied to the search for either broad-spectrum agents or compounds directed toward specific drug-resistant pathogens. During the Golden Age of antibiotic discovery, researchers relied on in vitro growth inhibition assays to identify new antibiotics, often elucidating their mechanisms of action years later.¹¹² However, with the advent of genomics, bioinformatics, and HTS, the field has shifted towards molecular target-based platforms, aiming to accelerate target identification and validation.^94,113 Contrary to expectations, this shift in approach did not lead to the discovery of new antibiotics.

In practice, this transition did not deliver the anticipated wave of clinically useful antibiotics. For example, between 1995 and 2001 GSK ran 67 HTS campaigns against genomics-derived antibacterial targets, screening hundreds of thousands of compounds, yet none of the resulting leads progressed to clinical trials.¹¹⁴ A major limitation of purely target-based screening is that activity against an isolated target often fails to translate into whole-cell efficacy because of permeability barriers and efflux.^115–117 More broadly, target-based assays contribute a relatively small fraction of approved small-molecule drugs,¹¹⁸ whereas in areas such as oncology, target-based strategies, particularly kinase programmes, have been more productive.¹¹⁹

Phenotypic screening has therefore regained prominence in antibacterial discovery, particularly when combined with strategies that facilitate target deconvolution.^120–122 One practical route is to compare responses in wild-type bacteria versus strains with defined genetic or pathway perturbations to enrich for target or mechanism-linked hits.¹²³ This “target-, pathway-, or mechanism-based” whole-cell screening is increasingly used in antibiotic discovery.¹²⁴ Ideally, target-based whole-cell screening should prioritise less explored targets where there is little or no existing chemical matter.

3.7. Shifting focus to Gram-negative pathogens and target selection and emerging strategies

Over the past decade, antibacterial discovery has increasingly shifted from Gram-positive to Gram-negative pathogens, driven in part by a better understanding of the physicochemical features needed to achieve cell entry and evade efflux.¹²⁵ An influential paper by Piddock and colleagues highlighted a set of less explored targets for Gram-negative drug discovery and outlined practical criteria for target selection, including essentiality, conservation, and the likelihood of achieving selective inhibition over human homologues.¹²⁵ Target location is also critical, since many Gram-negative targets require penetration of two membranes.¹²⁵ The study pointed to opportunities across several pathways, including outer membrane assembly (e.g., lipopolysaccharide (LPS) biosynthesis and lipoprotein maturation and transport), cell wall biosynthesis (e.g., Mur ligase enzymes and FtsZ), core metabolic routes (e.g., isoprenoid and fatty acid biosynthesis), and protein synthesis (e.g., transcription inhibitors).¹²⁵

This renewed focus on Gram-negative pathogens has also encouraged exploration of complementary strategies that extend beyond conventional direct-acting antibiotics.¹²⁶ These include efflux inhibitors,^127–129 antibiotic adjuvants,^130,131 peptidomimetics,^132–134 siderophores,^135–137 and small-molecule degraders, such as BacPROTACs.^138–140

Overall, while AMR continues to drive the need for new treatments, the breadth of approaches now being pursued provides multiple routes to new antibacterial agents. In the next sections, we summarise candidates in clinical trials and selected compounds with promising preclinical efficacy.

4. Overview of New Antibiotics in Clinical Development

Recently, the WHO issued an updated list of 15 families of antibiotic-resistant priority pathogens that pose the greatest threat to human health.¹⁴¹ In the WHO 2023 review of the clinical antibacterial pipeline, 32 drugs were reported as being in development against these pathogens, but only 13 met at least one innovation criterion, namely no known cross-resistance, a new class, a new target, or a new mode of action.³¹ Important gaps remain, particularly for MDR A. baumannii and P. aeruginosa, and there are relatively few candidates addressing the growing problem of metallo-β-lactamases (MBLs).^142,143 Around 40% of the pipeline, 13 of 32 candidates, comprises β-lactam and β-lactamase inhibitor combinations.³¹

Overall, the pipeline remains limited, and much effort is still directed toward modifying existing scaffolds rather than developing innovative agents.¹⁴⁴ In parallel, there has been renewed interest in narrow-spectrum antibacterials that were previously deprioritised. Broad-spectrum agents were often favoured for clinical convenience, especially when rapid diagnostics were limited, but increased emphasis on stewardship and microbiome preservation has shifted attention toward more targeted therapies that can reduce collateral effects on commensal flora and limit bystander selection pressure. In this context, reassessing earlier narrow-spectrum scaffolds is a practical route to new treatments.

Given that several reviews already cover the broader clinical landscape in detail,^145–149 we provide a concise overview here, focusing on candidates that represent new chemical classes and/or novel mechanisms of action, often referred as first-in-class antibiotics (Table 1).

Table 1. Current global pipeline of new antibacterials in clinical development with innovative modes of action.

Name	Structure	Chemical class	Phase	Target
Taniborbactam (1)		Cyclic boronate	III	Cell wall biosynthesis (β-lactamase)
Zoliflodacin (2)		Spiropyrimidenetr ione	III	DNA replication (GyrB)
Gepotidacin (3)		Triazaacenaphthyl ene	III	DNA replication (GyrA and TopoIV)
Afabicin (4)		Benzofuran naphthyridine	IIa	Fatty acid biosynthesis (FabI)
MGB-BP-3 (5)		Polyamide	III	Gene expression
Ibezapolstat (6)		6-anilinouracil	IIb	DNA replication (DNA pol IIIC)
CRS-3123 (7)		Diaryldiamine	II	Protein synthesis (MetRS1)
Fluorothiazinon (8)		1,3,4-thiadiazine- 5-one	II	Anti-virulence (T3SS)
Ledaborbactam etzadroxil (9)		Cyclic boronate	I	Cell wall biosynthesis (β-lactamase)
TXA709 (10)		Thiazolo[5,4- b]pyridine	I	Cell division (FtsZ)
Zosurabalpin (11)		Macrocyclic peptide	I	Lipopolysaccharide transport (LptB2FGC complex)
QPX7728 (12)		Cyclic boronate	I	Cell wall biosynthesis (β-lactamase)
BWC0977 (13)		Oxazolidinone	I	DNA replication (Gyr and TopoIV)

4.1. Taniborbactam

Taniborbactam (1) is a boronate-based β-lactamase inhibitor (BLI) being developed in combination with cefepime for the treatment of severe bacterial infections caused by DR Gram-negative bacteria, particularly carbapenem-resistant Enterobacterales (CRE) and carbapenem-resistant Pseudomonas aeruginosa (CRPA).^150,151 The emergence of β-lactamase-mediated resistance has rendered CRE and CRPA major clinical threats. Existing β-lactam-BLI combinations, which primarily target widespread serine β-lactamases (SBLs), do not provide sufficiently broad coverage across all clinically relevant SBLs, such as KPC and OXA-48, and MBLs, including NDM-1. Taniborbactam is distinguished by its broad inhibition profile, with activity reported against Ambler class A, B, C, and D enzymes, leveraging unique substrate mimicry and distinct mechanisms to combat both SBLs and MBLs.¹⁵² Moreover, its broad-spectrum activity against all β-lactamases makes it an attractive therapeutic option for treating severe infections, including acute pyelonephritis, complicated urinary tract infections, hospital-acquired pneumonia, and ventilator-associated pneumonia. Clinical studies have demonstrated taniborbactam’s safety and pharmacokinetic (PK) profile in individuals with varying degrees of renal impairment, with no serious adverse events reported and no treatment discontinuations due to adverse events.¹⁵³ Positive results from a phase III clinical trial in patients with complicated urinary tract infections, including acute pyelonephritis, provide support for the efficacy and safety of cefepime-taniborbactam.¹⁵⁴ Additionally, it t has shown activity against various strains of K. pneumoniae and P. aeruginosa, including MDR isolates, with confirmed efficacy in both neutropenic mouse lung and ascending urinary tract infection models.¹⁵¹ In February 2024, the FDA issued a Complete Response Letter to Venatorx Pharmaceuticals and Melinta Therapeutics regarding their New Drug Application (NDA) for cefepime-taniborbactam, citing the need for additional data on manufacturing and testing methods, but not expressing concerns about the drug’s clinical safety or efficacy.¹⁵⁵ The FDA’s acceptance of the NDA for cefepime-taniborbactam, developed jointly by Venatorx and the GARDP, marked an important step towards the development of new treatment options for DR Gram-negative infections.¹⁵⁶ With exclusive distribution rights, GARDP aims to ensure equitable global access, particularly in low- and lower-middle-income countries, where the clinical need is substantial.

4.2. Zoliflodacin

Zoliflodacin (2), an orally administered compound, features a novel chemical scaffold known as a spiropyrimidinetrione.¹⁵⁷ Developed initially by AstraZeneca and later by its spin-off Entasis Therapeutics, this agent is currently in phase III clinical trials for the treatment of uncomplicated gonorrhea and has since been approved by the FDA. Zoliflodacin exhibits a unique mode of inhibition targeting GyrB, with a distinct binding site that differs from fluoroquinolones, and no cross-resistance has been detected to date.¹⁵⁸ Although it has demonstrated potent activity against Neisseria gonorrhoeae, including MDR strains,^159,160 its suboptimal PK properties necessitate high doses in clinical trials.¹⁵⁸ Results from Phase III trials, sponsored by GARDP and Entasis, supported further clinical position of zoliflodacin as a treatment option for uncomplicated gonorrhoea.¹⁶¹ The study met its primary endpoints, demonstrating non-inferiority to the current global standard regimen.

Furthermore, zoliflodacin was found to be generally well-tolerated, with no serious adverse events or fatalities reported during the trial.¹⁶¹ Zoliflodacin (Nuzolvence) was approved by the FDA on 12 December 2025 as a first-in-class, single-dose oral antibacterial for uncomplicated urogenital gonorrhoea caused by Neisseria gonorrhoeae.

4.3. Gepotidacin

Gepotidacin (3), formerly known as GSK2140944, is a novel antibiotic that belongs to the triazaacenaphthylene class.¹⁶² Its distinctive mechanism of action involves the simultaneous inhibition of two essential topoisomerase enzymes, namely the GyrA subunit of DNA gyrase and the ParC subunit of topoisomerase IV, thereby interfering with bacterial DNA replication.^163,164 The unique binding mode of gepotidacin requires mutations in both enzymes to confer resistance, which may reduce the likelihood of resistance emerging through a single target-based mutation. This antibiotic demonstrates broad-spectrum bactericidal activity against a range of pathogens, including methicillin-resistant Staphylococcus aureus (MRSA), penicillin-resistant isolates of Streptococcus pneumoniae and E. coli expressing extended spectrum β-lactamase (ESBL). Gepotidacin has demonstrated activity against various strains of levofloxacin-resistant S. aureus, S. pneumoniae, and E. coli.^165,166 Gepotidacin’s efficacy against MRSA was further evidenced in a murine lung infection model under neutropenic conditions.¹⁶⁷ Additionally, clinical trials have demonstrated the antibiotic’s effectiveness, safety, and tolerability in treating patients with suspected or confirmed Gram-positive acute bacterial skin and skin structure infections.¹⁶⁸ Phase III trials have shown gepotidacin to be non-inferior to nitrofurantoin in treating uncomplicated urinary tract infections.¹⁶⁹ Gepotidacin has also yielded positive results in treating urogenital gonorrhea in the EAGLE-1 phase III clinical trial, demonstrating non-inferiority to the standard treatment combining intramuscular ceftriaxone and oral azithromycin.¹⁷⁰ Gepotidacin received US FDA approval on 25 March 2025 for the treatment of uncomplicated urinary tract infections.

4.4. Afabicin

Afabicin (4), a prodrug of afabicin desphosphono, is an antibiotic belonging to the benzofuran naphthyridine class, which selectively targets S. aureus FabI.¹⁷¹ This NADH-dependent enzyme catalyses the final step in the bacterial fatty acid biosynthesis pathway (FAS-II), specifically the reduction of enoyl-acyl carrier protein (enoyl-ACP), a process required for phospholipid synthesis and bacterial membrane integrity.¹⁷² Screening of GSK’s proprietary compound collection led to the identification of potent FabI inhibitors, ultimately resulting in the discovery of afabicin desphosphono through a structure-guided optimisation campaign.¹⁷³ Afabicin is the first FabI inhibitor to enter the clinical pipeline and has shown efficacy against MDR staphylococcal infections.^174,175 Its phosphate prodrug enhances the solubility of afabicin desphosphono, enabling both intravenous and oral formulations. Following administration, it is rapidly converted to its active moiety, which exhibits high serum protein binding and a prolonged elimination half-life, contributing to the favourable pharmacokinetic profile observed in phase I studies.¹⁷¹ Afabicin, currently in phase II clinical development by Debiopharm, has been found to have no significant impact on gut microbiota in both mice and healthy subjects when administered orally, whereas clindamycin, linezolid, and moxifloxacin produced substantial dysbiosis in the murine model.¹⁷⁶ Afabicin is being evaluated for the treatment of staphylococcal infections, including acute bacterial skin and skin structure infections and bone and joint infections caused by both methicillin-sensitive and methicillin-resistant S. aureus.^177,178

4.5. MGB-BP-3

MGB-BP-3 (5), a synthetic polyamide classified as a minor groove binder (MGB), exhibits antibacterial properties against Gram-positive bacteria, with particular interest in the treatment of gastrointestinal infections caused by C. difficile.¹⁷⁹ Developed through a collaborative effort between the University of Strathclyde and MGB Biopharma, MGB-BP-3 acts through a distinct mechanism of action by binding non-covalently to the minor groove of DNA, with a preference for AT-rich regions.^180,181 This interaction interferes with transcriptional processes and distinguishes the compound from fluoroquinolones, although the full mechanism of action of MGB-BP-3, as well as the basis for its limited toxicity toward mammalian cells, remains incompletely understood.¹⁸² MGB-BP-3 exhibits substantial antibacterial activity against a range of Gram-positive bacteria, including the hypervirulent BI/NAP1/027 strain of C. difficile, and serial passage experiments did not detect resistance emergence over more than 80 generations.¹⁸³ Although its activity against Gram-negative bacteria is limited, combination with an efflux pump inhibitor or membrane permeabiliser has been proposed as a possible strategy to overcome intrinsic resistance barriers.¹⁷⁹ In a hamster model of CDI, MGB-BP-3, administered both as a free base suspension and in colonic delivery microparticles, exhibited protective effects against mortality and prolonged post-treatment survival.^148,183 MGB-BP-3 has shown favourable safety, tolerability, and efficacy in Phase II clinical studies, achieving initial and sustained cure in the studied cohort. It has been declared “Phase III ready” by the FDA, having received GAIN status.¹⁴⁸

4.6. Ibezapolstat

Ibezapolstat (6), an orally administered 6-anilinouracil derivative developed by Acurx Pharmaceuticals, exhibits a unique mechanism of action, selectively targeting DNA polymerase IIIC in Gram-positive bacteria. This gut-restricted antibiotic demonstrates efficacy against Clostridioides difficile, an infection localised in the gastrointestinal tract, in both in vitro and in vivo models.^184,185 DNA polymerase inhibitors are widely used in antiviral therapy, comparatively few antibacterial agents targeting bacterial DNA polymerases have advanced in clinical development. Ibezapolstat is one of the first bacteria-specific DNA polymerase IIIC inhibitors to reach late-stage clinical development. As an analogue of deoxyguanosine triphosphate, ibezapolstat forms an inactive ternary complex by pairing with deoxycytidine triphosphate of DNA, effectively binding to DNA polymerase IIIC.¹⁸⁶ The compound’s narrow spectrum is attributed to its selective inhibition of DNA polymerase IIIC, which is not expressed in Gram-negative bacteria. In vitro studies have shown that ibezapolstat’s potency is comparable to that of metronidazole and vancomycin against C. difficile.¹⁸⁷ In a hamster model, a dose of 12.5 mg/kg administered twice daily for 3 days provided complete protection, and the recurrence rates decreased with prolonged treatment. Ibezapolstat demonstrated no toxicity at an oral dose of 1 g/kg, and at 75 mg/kg it exhibited limited absorption (5%), resulting in higher concentrations in the intestine.¹⁸⁵ In a phase IIa clinical trial involving 10 adults with C. difficile infections (CDIs), a 10-day oral treatment regimen of 450 mg ibezapolstat administered twice daily achieved complete eradication of colonic C. difficile within three days, with no recurrences were observed throughout the 28-day study period. The phase IIa data demonstrated not only the efficacy of ibezapolstat against C. difficile but also a significant proliferation of healthy gut microbiota during and after treatment.¹⁸⁸ Acurx completed its phase IIb trial, which assessed the efficacy of ibezapolstat versus vancomycin, and the program is being advanced toward international phase III clinical development.¹⁸⁹

4.7. CRS-3123

Developed by Creston Inc., CRS-3123 (7) (formerly REP-3123) is a novel antibiotic that selectively targets type 1 methionyl-tRNA synthetase (MetRS1) in Gram-positive bacteria, with a particular emphasis on C. difficile.¹⁹⁰ As a vital enzyme, MetRS1 catalyses two consecutive reactions essential for protein synthesis.¹⁹¹ CRS-3123 has showed antibacterial activity against a broad range of Gram-positive bacteria, including C. difficile, S. aureus, and S. pneumoniae.^190,192 CRS-3123 displays limited potency against Gram-negative bacteria.¹⁹⁰ CRS-3123 not only exhibits antibacterial efficacy but also reduces the production of toxins A and B, essential for C. difficile’s pathogenicity.¹⁹³ Moreover, it inhibits spore production, a key challenge in treatment of C. difficile infections and the prevention of recurrences.¹⁹³ The preclinical profile of CRS-3123 is reinforced by its favourable PK profile, characterised by low systemic exposure and peak concentrations in the gut.¹⁹³ In animal models, CRS-3123 significantly improved survival rates by inhibiting spore formation and thereby reducing reinfection.¹⁹³ CRS-3123 has completed two phase I clinical trials, demonstrating safety and tolerability profile. Its low absorption, minimal systemic accumulation, and high faecal concentrations support its further development for CDI.^194,195 A phase II study comparing CRS-3123 with vancomycin in adult with CDI has since been completed, with published results supporting its clinical activity and microbiome-sparing profile.¹⁹⁶

4.8. Fluorothiazinon

Fluorothiazinon (8) is an inhibitor of the type III secretion system (T3SS), a highly conserved target in Gram-negative bacteria that has garnered attention as an anti-virulence strategy.¹⁹⁷ Discovered by the Gamaleya Research Institute of Epidemiology and Microbiology in Moscow, fluorothiazinon has been shown to be efficacious in preventing both acute and chronic systemic infections. It reduces bacterial loads in rodent models infected with P. aeruginosa,¹⁹⁸ E. coli,¹⁹⁹ Salmonella enterica serovar Typhimurium,²⁰⁰ and A. baumannii.²⁰¹ Fluorothiazinon has also been evaluated in a phase II clinical trial in combination with cefepime for the treatment of complicated urinary tract infections caused by P. aeruginosa (NCT03638830), and recent reports have described clinical and microbiological benefits for the combination compared with cefepime/placebo.²⁰²

4.9. Ledaborbactam etzadroxil

Ledaborbactam etzadroxil (9), formerly known as VNRX-7145, is an orally bioavailable BLI developed by Venatorx Pharmaceuticals, which exhibits potent activity against a broad range of serine β-lactamases, particularly Ambler class A, C, and D enzymes.²⁰³ Currently, ledaborbactam etzadroxil is being developed in combination with the oral cephalosporin ceftibuten to restore ceftibuten’s activity against cephalosporin-resistant Enterobacterales. The bicyclic boronate prodrug undergoes rapid and extensive biotransformation to the active inhibitor ledaborbactam and has shown favourable oral bioavailability across multiple preclinical species.^203,204 Ledaborbactam has demonstrated in vivo efficacy in a lethal murine sepsis model, achieving an ED₅₀ of 12.9 mg/kg when administered orally in combination with ceftibuten.²⁰³ Phase I clinical studies published in 2025 showed that ledaborbactam etzadroxil, with or without ceftibuten, was generally safe and well tolerated, with pharmacokinetic properties supporting further clinical development.²⁰⁵ Additional studies, including evaluation in subjects with varying degrees of renal impairment, are being used to inform dose selection for a registrational phase III program of ceftibuten-ledaborbactam etzadroxil.²⁰⁶

4.10. TXA709

TXA709 (10), an antibiotic compound developed by Taxis Pharmaceuticals, targets Filamenting temperature-sensitive mutant Z (FtsZ), a key protein involved in bacterial cell division.²⁰⁷ As the first FtsZ-directed antibacterial to reach clinical development, TXA709 is an oral prodrug formulation of TXA707, designed to combat MRSA infections. FtsZ plays a vital role in bacterial cell division, forming single-stranded filaments through GTP-dependent polymerisation, resulting in the formation of the Z-ring at midcell, a critical step in peptidoglycan synthesis, septum formation, and subsequent cell division induction.²⁰⁸ Treatment with TXA707 caused rapid enlargement of S. aureus cells, leading to lysis within one hour. TXA707 has demonstrated potent bactericidal activity against MDR clinical strains of S. aureus, with an MIC of 1 μg/mL.²⁰⁷ PK evaluation in mice revealed that TXA709 is rapidly converted to TXA707, resulting in 95% bioavailability following oral administration. Both intravenous and oral doses of TXA709 ensured survival in murine peritonitis models infected with MRSA strains.²⁰⁷ In addition, TXA709 and its active form, TXA707, demonstrated linear PK after oral administration in a murine thigh infection model.²⁰⁹ Having completed phase I clinical trials with no reported serious adverse events, TXA709 is being developed as an oral anti-MRSA agent, particularly in combination with other antibiotics such as cefdinir.^210,211

4.11. Zosurabalpin

Zosurabalpin (11) is a structurally distinct macrocyclic peptide with a mode of action that differs significantly from existing antibiotics.^212,213 Developed through a collaborative effort between scientists at Roche and Harvard University, this compound targets carbapenem-resistant A. baumannii (CRAB), a WHO priority 1 pathogen. By targeting the LptB₂FGC protein complex, zosurabalpin blocks LPS transport, a process essential for maintaining the outer-membrane structure of Gram-negative bacteria. This inhibition results in an intracellular accumulation of LPS, ultimately leading to bacterial cell death. Zosurabalpin has shown potent activity against highly drug-resistant CRAB in both in vitro and in vivo models, including sepsis models. Structural analysis via X-ray crystallography further showed that zosurabalpin interacts with LptB₂FGC in an LPS-dependent manner, supporting this mechanism of action.²¹³ The drug’s specificity for A. baumannii has been attributed to the limited evolutionary conservation of Lpt proteins across bacterial species. Zosurabalpin demonstrated activity against more than 100 clinical CRAB isolates and substantially reduced bacterial loads in murine pneumonia models while preventing mortality in murine sepsis models. Phase I clinical studies showed that single intravenous doses ranging from 10 mg to 2000 mg were safe and generally well tolerated,²¹⁴ and the program has since advanced to phase III clinical development.

4.12. QPX7728

QPX7728 (12), also known as xeruborbactam, is a novel ultra-broad-spectrum BLI with activity against both serine and metallo-β-lactamases and potential for both intravenous and oral administration.^215,216 Kinetic studies revealed that QPX7728 exhibits a slow-binding profile toward serine β-lactamases, with K_i values ranging from 0.13 to 9.9 nM, and a fast-on/fast-off mode of action toward metallo-β-lactamases, with K_i values between 4.5 and 270 nM.²¹⁶ In vitro evaluations against eleven β-lactamase-producing strains demonstrated marked reductions in MIC values when QPX7728 was combined with β-lactam antibiotics, in several cases consistent with PK-PD breakpoints for the partner β-lactams. Extensive microbiological assessments have shown that combining QPX7728 with β-lactam antibiotics markedly enhances antibacterial activity, reducing MIC values to as low as 0.015 and 0.03 μg/mL, respectively. In a murine thigh infection model using a carbapenem-resistant K. pneumoniae strain, QPX7728 enhanced meropenem activity, reducing the MIC from >64 μg/mL to 0.25 μg/mL and producing a >1 log₁₀ CFU reduction in bacterial burden. PK studies also showed favourable oral bioavailability, minimal inhibition of mammalian proteases, and an acceptable short-term toxicology profile in rats.²¹⁶ Further in vitro and in vivo studies confirmed the efficacy of QPX7728 in restoring antibiotic potency against clinically relevant MDR strains.^217–221 More recently, phase I clinical studies of xeruborbactam showed that intravenous administration was generally safe and well tolerated, with pharmacokinetic properties supporting further clinical development.²²² Additional phase I studies, including mass-balance (NCT07104162) and fixed-combination evaluations with cefiderocol (NCT07083817), are ongoing or have recently been launched.

4.13. BWC0977

BWC0977 (13) is a broad-spectrum antibiotic discovered by Bugworks Research Inc. and co-developed with GARDP to combat MDR bacterial infections. It selectively targets bacterial DNA replication through inhibition of DNA gyrase and topoisomerase IV.²²³ BWC0977 has demonstrated potent activity against key hospital-acquired pathogens, including WHO critical-priority pathogens such as carbapenem-resistant A. baumannii and K. pneumoniae.²²⁴ Identified through a whole-cell phenotypic screen against E. coli, BWC0977 was subsequently optimised through medicinal chemistry and SAR studies, achieving MIC₉₀ values of 0.03–2 µg/mL against MDR Gram-negative and Gram-positive bacteria. It also showed efficacy in neutropenic mouse thigh and rat lung infection models infected with P. aeruginosa. PK studies in mice, rats, and dogs revealed linear PK, good tissue distribution, and dose-proportional exposure. In a Phase I clinical trial (NCT05088421), single-ascending intravenous doses were generally well tolerated in healthy volunteers, with a favourable safety profile and PK characteristics consistent with preclinical predictions, supporting further clinical development.²²⁴

5. Promising Preclinical Antibiotic Candidates

Innovation in antibiotic discovery continues, and despite advances that have reduced the cost barriers for biologics, small molecules still account for the majority of development programmes.²²⁵ The WHO’s 2023 antibacterial pipeline report provides a snapshot of the preclinical landscape, listing 244 products in development, with most programmes based in Europe (52%) and the United States (34%).³¹ Nearly half are direct-acting small molecules (47%), while 38% fall into non-traditional approaches. These candidates span a range of mechanisms, including cell-wall synthesis (16%), membrane activity (28%), protein synthesis inhibition (7%), immune modulation (5%), and anti-virulence strategies (10%). Notably, 37% are being developed as narrow-spectrum agents, consistent with a shift toward more targeted antibacterial therapies.³¹

While the WHO analysis primarily reflects industry-led programmes, where details can be limited by IP constraints, academic research continues to identify promising compounds that may feed into the development pipeline.²²⁶ Several excellent reviews have summarised preclinical antibacterials from a drug discovery perspective.^{27,38,226–230}

In this section, we provide a medicinal chemistry-focused overview of antibacterial compounds that have reached the “validated lead” or preclinical development stage, with coverage from 2020 to 2025. Our manually curated set, assembled through targeted literature searches and database queries, highlights compounds with demonstrated in vivo efficacy and presents their chemical structures alongside the supporting biological evidence. Although previous reviews have surveyed antibacterial discovery efforts more broadly, relatively few have focused specifically on preclinical candidates within a structure-driven medicinal chemistry framework. The compounds discussed here are synthetic and were identified through approaches including phenotypic and target-based screening, hit-to-lead development, and lead optimisation. For clarity, they are organised according to their reported mechanism of action.

5.1. Compounds that target cell wall biosynthesis

Bacterial cell wall synthesis is essential for bacterial viability and structural integrity, with peptidoglycan being the primary component.²³¹ This protective barrier, significantly thicker in Gram-positive bacteria than in Gram-negative bacteria, maintains the rigidity and integrity of bacterial cells.²³² Disrupting peptidoglycan synthesis compromises bacterial cell wall integrity, leading to cell death.²³³ By targeting bacterial cell wall synthesis, antibiotics can selectively inhibit bacterial growth, as mammalian cells lack a similar structure.²³⁴ In recent years, extensive research has led to the discovery of numerous novel inhibitors, some of which have advanced to preclinical or clinical stages. This section provides an overview of promising compounds targeting distinct pathways related to cell wall and cell envelope biogenesis, including PgcA, DPPS, UppP, LolE, β-lactamase, Mβ-lactamase, and penicillin-binding proteins (PBPs).

α-Phosphoglucomutase (PgcA) is a key enzyme involved in the synthesis of lipoteichoic acid (LTA), a crucial cell envelope-associated polymer found in numerous Gram-positive bacteria.²³⁵ Within the cytoplasm, PgcA facilitates the transfer of a phosphate group from position 1 to position 6 on glucose, thereby contributing to the early stages of LTA synthesis, a vital process that underpins bacterial growth, membrane stability, and biofilm formation.^235,236 Bacterial PgcA shares only ~25% amino acid identity with its mammalian counterpart,²³⁷ making the inhibition of PgcA a promising antimicrobial strategy.

HSGN-94 (13) (Figure 2), an oxadiazole derivative developed by Stintim’s team, effectively inhibits lipoteichoic acid biosynthesis in drug-resistant Gram-positive bacteria.²³⁸ It demonstrates efficacy against various staphylococcal strains, including MRSA and MSSA, with MICs ranging from 0.25 μg/mL to 1 μg/mL. HSGN-94 exhibits activity against other clinically relevant Gram-positive bacteria, such as S. pneumoniae, Streptococcus pyogenes, E. faecalis, E. faecium, and Listeria monocytogenes, with MICs spanning 0.06 μg/mL to 0.25 μg/mL. HSGN-94 displays negligible cytotoxicity against human HaCaT cells at concentrations up to 64 μg/mL and shows a low propensity for inducing MRSA resistance. Mechanistic studies reveal that HSGN-94 selectively targets the PgcA protein, involving direct binding and downregulation of phosphatidylglycerol phosphate synthase in LTA biosynthesis. An in vivo efficacy study using a murine skin infection model of MRSA infection demonstrates a remarkable 93.4% reduction in bacterial load in skin lesions.²³⁸

Figure 2. Antibacterial agents targeting cell wall synthesis.

Cis-prenyltransferases, specifically undecaprenyl diphosphate synthase (UPPS) and decaprenyl diphosphate synthase (DPPS), are crucial in the initial stages of bacterial cell wall synthesis, as they facilitate the elongation of peptidoglycan precursors within the bacterial cytoplasm. These enzymes are absent from the human genome.²³⁹

A promising diphenyl lead compound, redoxal (14) (Figure 2), was identified through a novel screening campaign.²⁴⁰ Initially, a library of 750 compounds, known as the NCI Diversity Set 5, was screened against UPPS in E. coli, followed by cross-screening against DPPS in M. tuberculosis. Among the initial 17 hits, redoxal showed the most potent inhibition of MtDPPS, with an IC₅₀ value of 3.6 μM, and exhibited comparable inhibition of EcUPPS, with an IC₅₀ value of 7.8 ± 3.3 μM, suggesting its potential for multi-targeting. Redoxal had moderate yet selective activity against S. aureus and C. difficile, with MIC values of 11 μg/mL and 4 μg/mL, respectively. It exhibited synergism with aminoglycoside and β-lactam antibiotics against clinical S. aureus strains, resulting in up to a 16-fold reduction in MICs when combined with these antibiotic classes. In a haemolysis assay using human embryonic kidney cells (HEK293T) and mouse embryonic fibroblasts (NIH/3T3), 14 showed no apparent haemolytic toxicity up to concentrations exceeding 300 μM. Additionally, in a C. elegans model, 14 displayed low toxicity (LD₅₀ = 89 μM) and achieved a 70% reduction in S. aureus survival at a concentration of 16 μM.²⁴⁰

Undecaprenyl diphosphate phosphatase (UppP), a crucial membrane protein, plays a vital role in bacterial cell wall formation by catalysing the conversion of undecaprenyl pyrophosphate (C55-PP) to undecaprenyl phosphate (C55-P).²⁴¹ Knockout studies have shown a reduction (75%) in overall C55-PP phosphatase activity in membrane extracts compared to wild-type E. coli, highlighting the enzyme’s importance.²⁴²

Mayhoub’s group has explored arylthiazole derivatives as promising agents for combating MDR Gram-positive strains.^243,244 Through a series of structure-activity relationship optimisations, they developed aminomethylpiperidine-containing biphenylthiazole (15) (Figure 2), an UppP inhibitor with favourable antibacterial activity and PK profile.²⁴⁵ The development of compound 15 involved the replacement of a potentially cytotoxic hydrazine moiety with a pyrimidine group and of the metabolically labile n-butyl group with a phenyl group, thereby enhancing metabolic stability and antibacterial efficacy. Compound 15 has bactericidal properties against drug-resistant Gram-positive pathogens, including MRSA, with a MIC of 2−4 μg/mL and a minimum bactericidal concentration (MBC) of 2−8 μg/mL. Time-kill assays confirmed its rapid bactericidal activity against MRSA, resulting in a 0.7 log₁₀ CFU reduction within 6 hours. Compound 15 showed no cytotoxicity to Caco-2 and HaCaT cells at concentrations up to 64 μg/mL and effectively eliminated intracellular MRSA. Compound 15 demonstrated higher metabolic stability (T_1/2 = 4.28 h) and lower clearance rate (CL = 7.35 L/h) than the parent lead compound. Its moderate inhibition activity against UppP (IC₅₀ = 29 μM) indicated its mode of action in perturbing bacterial cell wall synthesis. In vivo studies using a murine MRSA wound infection model validated that 15 effectively suppressed infection in skin lesions, resulting in a 2 log₁₀ CFU reduction in bacterial load.²⁴⁵

LolE is a transmembrane protein involved in the lipoprotein-releasing system within the lipoprotein pathway of Gram-negative bacteria.²⁴⁶ Specifically, it mediates the transport of essential lipoproteins from the inner membrane to the outer membrane.²⁴⁷ Stokes et al. developed an innovative machine learning approach, utilising a directed message-passing deep neural network, to predict antibiotic structures capable of inhibiting A. baumannii growth.¹⁰⁷ By leveraging a training dataset of 480 active hits from screening 7,500 diverse small molecules against A. baumannii, along with a prediction dataset of 6,700 molecules from the updated Drug Repurposing Hub, they identified nine priority compounds that met specific criteria. Among the compounds evaluated, abaucin (16) (Figure 2) stood out as a promising candidate due to its favourable MIC of approximately 2 µg/mL against A. baumannii. It showed distinct characteristics, including minimal structural and activity overlap with known antibiotics and negligible nonspecific membrane interactions. Abaucin exhibited a restricted spectrum of activity, showing no efficacy against other Gram-negative species, such as CRE and P. aeruginosa, and modest potency against Gram-positive species like S. aureus, C. difficile, Bifidobacterium breve and Bifidobacterium longum. Abaucin retained activity across 41 diverse clinical isolates of A. baumannii despite multiple intrinsic and acquired resistance determinants. Due to the limitations of machine learning in explaining abaucin’s mechanism of action against A. baumannii, the researchers employed whole-genome sequencing to analyse abaucin-resistant A. baumannii mutants. This approach identified mutations in the lolE gene, which was further validated through transcriptomic profiling and CRISPRi-mediated perturbations, revealing a mechanistic link between abaucin and LolE-mediated lipoprotein trafficking. In vivo experiments using a murine wound infection model of A. baumannii infection confirmed abaucin’s efficacy in reducing infection in skin lesions, resulting in a 1.2 log₁₀ CFU decrease in bacterial load.¹⁰⁷

5.1.1. Non-traditional antibacterials: β-lactamase inhibitors

β-Lactamases constitute a class of enzymes that primarily catalysed the hydrolysis and inactivation of β-lactam antibiotics, characterised by a distinctive β-lactam ring. These enzymes are broadly classified into two main categories based on their amino acid sequences. The first category, comprising class A, C, and D enzymes, collectively referred to as serine-β-lactamases, utilise serine residues to facilitate β-lactam hydrolysis. Conversely, class B enzymes, known as metallo-β-lactamases, rely on divalent zinc ions to catalyse substrate hydrolysis.^248,249 This selective ring-opening mechanism contributes to bacterial resistance against β-lactam antibiotics such as penicillins, cephalosporins, and carbapenems.²⁵⁰ β-Lactamase inhibitors are therefore non-traditional antibacterial agents that act as adjuncts, restoring the activity of co-administered β-lactam antibiotics rather than directly inhibiting bacterial growth. Several clinical candidates have been developed to inhibit these enzymes and restore β-lactam activity, including taniborbactam (1), ledaborbactam etzadroxil (9), and QPX7728 (12).

Multiple research teams have developed a range of inhibitors targeting class A, C, and D β-lactamases, showcasing diverse mechanisms of action, and demonstrating efficacy in both in vitro and in vivo settings. In 2021, Fujiu et al. introduced compound 17 (Figure 3), a 2-sulfinyl-diazabicyclooctane derivative, which has proven to be a potent BLI capable of enhancing the antibacterial properties of β-lactam antibiotics against class A, C, and D serine β-lactamases.²⁵¹ Compound 17 was identified as the most active derivative in the series for restoring the antibacterial activity of ceftibuten, exhibiting substantial inhibition against carbapenem-resistant strains, with an IC₅₀ of 0.012−0.028 μM. Although compound 17 displayed moderate total clearance and a short half-life in multiple animal models, including mouse, rat, dog, and monkey, its bioavailability in rats was limited to 1.4%. To address this limitation, the development of compound 18 (Figure 3), a cyclohexyl ester prodrug of compound 17, enhanced bioavailability across multiple species. This improvement in bioavailability translated to enhanced in vivo efficacy, as indicated by the greater reduction of bacterial load when 18 was co-administered with ceftibuten in a murine urinary tract infection model against K. pneumoniae. Compound 18 exhibited a favourable safety profile, with no observed toxicity at a dose of 600 mg/kg/day.²⁵¹

Figure 3. Diazabicyclooctane derivatives as β-lactamase inhibitors.

LCB10-0200 (19) (Figure 4), a siderophore-conjugated cephalosporin antibiotic developed by LegoChem Biosciences and investigated by Kwak’s team, exhibited remarkable potency against MDR P. aeruginosa in vitro and in vivo.²⁵² The incorporation of a siderophore, an iron-chelating compound, represents a promising strategy for enhancing antibiotic delivery into bacterial cells.²⁵³ In 2021, the same team expanded their investigation to combat another MDR pathogen, K. pneumoniae, and found that 19 had in vitro antibacterial activity against this pathogen, including β-lactamase-expressing strains.²⁵⁴ The MIC values ranged from 0.125 to 1-2 mg/L, indicating up to a 32-fold increase in potency compared to established antibiotics such as ceftazidime and ciprofloxacin. In vivo efficacy studies conducted in three different murine infection models revealed promising results, as 19 exhibited effective MIC and ED₅₀ values (MIC ≤ 1 mg/L, ED₅₀ ≤ 2.5 or = 25 mg/Kg) in a systemic model. Treatment with 19 led to a reduction in bacterial load, decreasing it by 3 log₁₀ and 1 log₁₀ CFU units in murine thigh and urinary tract infections, respectively, at a concentration of 50 mg/kg. In addition, 19 showed no observed toxicity at concentrations up to 300 μM, as indicated by previous hERG experiments.²⁵⁴ In the same year, researchers at the US Army Medical Research Institute of Infectious Diseases, developed a novel therapeutic approach involving 19 and a broad-spectrum BLI as a combination treatment targeting a broad spectrum of MDR and ESKAPE pathogens. The combination (1:1 ratio) displayed notable in vitro inhibition, with MICs ≤ 4 μg/mL against 114 MDR bacteria and three biothreat bacteria, including Burkholderia pseudomallei, Yersinia pestis, and Francisella tularensis. This therapeutic combination exhibited in vivo efficacy in murine lung models of Y. pestis infection, achieving a maximum survival rate of 100% at the highest concentration tested.²⁵⁵

Figure 4. Siderophore conjugate (19) and metallo-β-lactamase inhibitors (20 and 21).

Metallo-β-lactamases represent a class of β-lactamases that confer resistance to β-lactam antibiotics through substrate hydrolysis involving divalent zinc ions. MBLs are further categorised into three subtypes (B1, B2, B3), distinguished by their structural features, sequence identities, zinc-catalysed mechanisms, and resistance profiles against β-lactams. Mechanistically, B1 and B3 subtypes harbour two zinc ions within the active site, whereas the B2 subtype contains only one.²⁵⁶ The B1 subtype exhibits an unusually broad substrate profile and high resistance to nearly all clinically utilised bicyclic β-lactam antibiotics, with notable representatives including NDM-1, IMP-1, and VIM-2.²⁵⁷ Similarly, the B3 subclass demonstrates a wide substrate spectrum, featuring enzymes such as AIM-1, GOB-18, and CSR-1.²⁵⁸ In contrast, the B2 MBL subtype displays the narrowest substrate selectivity among MBL subtypes, primarily targeting carbapenems, with prominent representatives including CphA, ImiS, and L1.²⁵⁹ Although metallo-β-lactamases are clinically important, no MBL inhibitor has yet reached routine clinical use.

In 2020, researchers at Antabio SAS, led by Davies et al, developed a preclinical candidate, ANT2681 (20) (Figure 4), an NDM-1 inhibitor, which emerged from an extensive lead optimisation campaign.²⁶⁰ Modifications to the original lead (ANT431),^261,262 including the incorporation of a glycinamide group, substitution of pyridine with a difluorophenyl ring, and the addition of a guanidino terminus, resulted in enhanced in vitro activities and showed promising in vivo efficacy, accompanied by acceptable cytotoxicity profiles. In vitro experiments indicated that 20 exhibits inhibitory activity against the NDM-1 enzyme, with an IC₅₀ value of 0.04 μM, and displays efficacy, particularly when combined with meropenem, against a broad range of MBL-producing Enterobacteriaceae, with MIC values ranging from 0.125−8 μM. In a murine thigh infection model, the combination of 20 and meropenem reduced bacterial load, suppressing K. pneumoniae by 1.8 log₁₀ CFU. Compound 20 exhibited a favourable safety profile in mammalian cytotoxicity, selectivity, and genotoxicity assays.²⁶⁰

In 2022, Hirsch’s team identified compound 21 (Figure 4), a N-aryl mercaptopropionamide, as a B1 MBL inhibitor.²⁶³ Compound 21 showed favourable inhibition potencies, with IC₅₀ values ranging from 1-6.4 µM against three B1 subtype MBLs (NDM-1, IMP-7, and VIM-1). Although compound 21 did not exhibit direct antibacterial activities against various B1-producing E. coli and P. aeruginosa isolates (MIC > 128 or 1024 mg/L), it effectively restored the antibacterial potency of imipenem against NDM-1-expressing E. coli, achieving a maximum 256-fold improvement. Compound 21 had a synergistic effect when combined with imipenem in treating 11 clinical isolates producing various MBLs, resulting in improvements in susceptibility, particularly in Serratia marcescens, with a maximum 64-fold improvement in imipenem susceptibility. The activity observed in vitro was corroborated in vivo using a Galleria mellonella model infected with imipenem-resistant NDM-1-expressing K. pneumoniae. Specifically, treatment with the combination of imipenem and 21 for three days resulted in a increase in the survival rate of G. mellonella larvae, reaching 56%.²⁶³

In 2022, Yang et al. introduced compound 22 (Figure 5) as a broad-spectrum MBL inhibitor, specifically targeting both NDM-1 (B1) and ImiS (B2).²⁶⁴ Biochemical assays confirmed the compound’s dual inhibitory activity, with IC₅₀ values of 0.37 µM against ImiS and 0.45 µM against NDM-1. Further characterisation revealed its reversible inhibition mode with ImiS (K_i = 0.29 µM) and irreversible inhibition with NDM-1, as determined by enzyme kinetics and isothermal titration calorimetry assays. Compound 22 demonstrated robust antibacterial activity in vitro against clinical isolates of E. coli and K. pneumoniae expressing NDM-1 and ImiS, with MIC values spanning 8−32 μg/mL. Enhanced activity was observed when 22 was combined with meropenem against ImiS or NDM-1 in clinical isolates of E. coli or K. pneumoniae, resulting in up to a 32-fold improvement in meropenem MIC values. In vivo studies using a murine infection model revealed that 22 enhanced meropenem activity, resulting in a significant reduction of K. pneumoniae burden, with a ~0.7 log₁₀ CFU decrease in the liver and a 0.6 log₁₀ CFU decrease in the spleen.²⁶⁴

Figure 5. Metallo-β-lactamase inhibitors with in vivo antibacterial efficacy.

Schofield et al. successfully developed another broad-spectrum MBL inhibitor, an indole-2-carboxylate derivative (23) (Figure 5), by employing a strategy that mimics the β-lactam-binding mode, specifically by chelating divalent zinc ions.²⁶⁵ Compound 23 exhibited inhibition against B1 subtype MBLs (NDM-1, VIM-1, VIM-2, and IMP-1), with pIC₅₀ values ranging from 7.2 to 9.5. It effectively restored the antibacterial activity of carbapenems, such as meropenem, imipenem, and doripenem, against a broad range of Enterobacterales, with MIC₉₀ values ranging from 1-2 mg/L. Subsequent studies on resistance reported that combining 23 with carbapenems (meropenem and imipenem) resulted in a low frequency of resistance. Additionally, in vivo experiments demonstrated the tolerability and in vivo activity of the combination of 23 and meropenem, which showed efficacy in murine models of peritonitis/sepsis and thigh infection, reducing bacterial load by up to 7 log₁₀ CFU against carbapenem-resistant E. coli and K. pneumoniae strains.²⁶⁵

In 2022, a group at Merck announced the successful development of compound 24 (Figure 5), a pan-MBL inhibitor with clinical potential. Derived from a historical MBL inhibitor and an NDM-1 inhibitor identified through virtual screening, compound 24 displayed enhanced potency in enzymatic activity assays, with IC₅₀ values in the nanomolar range against a panel of three B1 MBLs. It showed promising activity in cell assays, exhibiting favourable MITC₉₅ values against strains producing multiple MBLs. In vivo studies using a murine model of IMP-producing K. pneumoniae infection showed that co-treatment with imipenem and 24 led to a significant reduction of approximately 3 log₁₀ CFU in bacterial load.²⁶⁶

Recently, Li et al. developed a broad-spectrum MBL inhibitor, 25 (Figure 6), which was specifically designed to replicate the key pharmacophore features of biapenem at the active binding site. Compound 25 showed activity against all three MBL subtypes, with IC₅₀ values of less than 2 μM against B1 subtypes, moderate to weak IC₅₀ values ranging from 11 to 25 μM against B3 subtypes, and relatively poor potency against B2 subtypes, with IC₅₀ values of 61 μM or higher. As expected, the combination of 25 and meropenem reinstated antibacterial activity against B1 MBL-producing strains, resulting in a significant reduction in meropenem MIC values, ranging from 4-to 64-fold (MIC < 0.125 μg/mL or 1−8 μg/mL). The combination of 25 and meropenem failed to exhibit notable activity against B2/B3 clinical isolates, which produce CphA/L1-type MBLs. In vivo studies showed that the combination of 25 and meropenem enhances antibacterial activity against a K. pneumoniae NDM-1-producing strain in a murine sepsis model, resulting in a 1-2 log₁₀ CFU reduction in bacterial load in the spleen, liver, and kidney.²⁶⁷

Figure 6.

Broad-spectrum metallo-β-lactamase inhibitor 25, shown alongside the co-crystal structure of 25 (in yellow) bound to P. aeruginosa VIM-2 MBL (in grey), highlighting key interactions and binding sites (PDB code: 8HYD).

Another interesting study introduced 26 (Figure 7), a functionalised cephalosporin featuring a zinc-chelating motif, as an inhibitor targeting MBL-expressing CREs. In vitro studies noted that 25 effectively restored the efficacy of meropenem against seven CRE isolates, achieving meropenem MIC values of ≤ 0.5 mg/L, a 8-fold improvement. Subsequent enzymatic inhibition studies revealed moderate binding affinity of 26 for NDM-1 (K_iapp = 97.4 μM) and VIM-2 (K_iapp = 24.8 μM). Compound 26 exhibited no toxicity against HepG2 cells (IC₅₀ > 1000 mg/L) and did not cross-inhibit mammalian zinc-containing enzymes. Coadministration of 26 and meropenem showed in vivo efficacy against K. pneumoniae NDM in a murine thigh infection model, resulting in a reduction in bacterial burden, exceeding a 3 log₁₀ CFU decrease.²⁶⁸

Figure 7. Broad-spectrum metallo-β-lactamase inhibitor 26 and penicillin-binding protein inhibitors (27 and 28).

5.1.2. Penicillin-Binding Protein inhibitors

Integral to the bacterial cytoplasmic membrane, penicillin-binding proteins play a crucial role in the final steps of peptidoglycan synthesis, specifically facilitating polymerisation and transpeptidation. The mechanism of β-lactams involves the formation of covalent bonds with PBPs, resulting in the formation of stable acyl-enzyme complexes and subsequent inactivation.²⁶⁹ Resistance to β-lactams typically emerges through three primary mechanisms: the production of β-lactamases, mutations in PBPs that reduce susceptibility (including changes in topology and amino acid substitutions), and decreased accessibility to PBPs due to permeability barriers or changes in membrane-associated transport pathways, leading to reduced β-lactam permeability.²⁷⁰ Genetic analysis reveals that resistance is often associated with alterations in high-molecular-weight PBPs compared to susceptible strains.^271,272 In the case of Group A Streptococcus, research has linked reduced β-lactam affinity to the presence of mutated high-molecular-weight PBPs, specifically PBP1a, PBP2a, PBP2b, and PBP2x.^273,274

In 2020, Chang et al. reported compound 27 (Figure 7), a non-β-lactam antibacterial targeting MRSA.²⁷⁵ Derived from a SAR study, 27 exhibited activity against various strains, including S. aureus (MIC = 0.5 μg/mL), MRSA strains (MIC = 0.25 to 0.5 μg/mL), and Staphylococcus epidermidis and Staphylococcus haemolyticus (MICs ≤ 4 μg/mL). Further studies showed bactericidal activity for the combination with piperacillin-tazobactam against MRSA strains, resulting in a reduction of bacterial count exceeding 3 log₁₀ CFU by 12 hours in a time-kill assay. In vivo studies using a murine neutropenic thigh-infection model with MRSA strain NRS70 resulted in a reduction of bacterial load by a maximum of 1.3 log₁₀ CFU. Compound 27 also had favourable PK properties, including a concentration above MIC for 4 hours (>0.5 μg/mL), a clearance rate of 7.7 mL·min^–1·kg^–1, and half-lives of 13.7 hours (intravenous) and 11.2 hours (oral administration), with an absolute oral bioavailability of 65%. Compound 27 exhibited high plasma protein binding (PPB) (95%).²⁷⁵

Compound 28 (Figure 7) is a cephalosporin-3′-diazeniumdiolate prodrug, featuring an aminothiazole oxyimino ether side chain that enhances stability against β-lactamases.²⁷⁶ It demonstrates antibacterial activity against P. aeruginosa cystic fibrosis clinical isolates (MIC = 0.5– 16 μg/mL) and efficacy against other key human pathogens, including E. coli, K. pneumoniae, N. gonorrhoeae, and S. pneumoniae, with most MIC values ≤ 2 μg/mL. In vivo, treatment with 50–120 mg/kg in a murine lung infection model reduced P. aeruginosa burden by 2.3 log₁₀ CFU. Confocal microscopy showed superior antibiofilm activity compared to ceftazidime against P. aeruginosa biofilms PA30 and PA68. The compound exhibited no haemolytic effects on human red blood cells at 32 μg/mL and showed no toxicity, mortality, or weight loss in mice at doses up to 300 mg/kg. Compound 28 contains an O²-linked diazeniumdiolate moiety that enables nitric oxide release, which may contribute to biofilm dispersal and potentially enhance antimicrobial efficacy.²⁷⁶

In 2021, researchers at Entasis Therapeutics reported ETX0462 (29) (Figure 8), a diazabicyclooctane derived compound that simultaneously inhibits PBPs and functions as a BLI.²⁷⁷ Specifically, 29 selectively targets Gram-negative PBPs, namely PBP1a and PBP3. Its profile was optimised to improve porin-mediated uptake, reduce efflux liability, and enhance biochemical potency. As a result, 29 has yielded remarkable efficacy against ESKAPE pathogens. In vitro studies showed 29’s promising activity, with MIC₉₀ values ranging from 1-2 mg/L against clinical isolates of P. aeruginosa and E. coli. The compound also exhibited adequate activity against MDR P. aeruginosa isolates, with an MIC₉₀ of 4 mg/L. Broad-spectrum potency was confirmed with MIC ranges of ≤0.06-16 mg/L against ESKAPE pathogens and 0.25-1 mg/L against biothreat pathogens, including Francisella tularensis, Yersinia pestis, Bacillus anthracis, Burkholderia pseudomallei, and Burkholderia mallei. Studies conducted in murine lung and thigh models of MDR P. aeruginosa infection indicated a 1−2 log₁₀ CFU reduction in bacterial burden at a dose of 50 mg/kg. The efficacy of 29 against biothreat pathogens was investigated in a murine model of Y. pestis infection, resulting in a 100% survival rate in mice treated for 24 hours. Toxicity studies confirmed the favourable safety profile of 29, with a well-tolerated upper dose limit of 2000 mg/kg.²⁷⁷

Figure 8. Penicillin-binding protein inhibitor 29, shown alongside the co-crystal structure of 29 (in yellow) bound to P. aeruginosa PBP3 (in grey), highlighting key interactions and binding sites (PDB code: 7JWL).

Compound 30 (Figure 9), another β-lactam derivative featuring an aminothiazole side chain, has showed remarkable potency against CRE strains, with MIC values spanning 0.063 to 4 μg/mL. Moreover, it has consistently maintained potency (MIC ≤ 1 μg/mL) against a diverse range of clinical CRE isolates. In vivo studies using a murine neutropenic lung infection model infected with carbapenemase-producing K. pneumoniae demonstrated in vivo efficacy, although the reduction in bacterial burden was moderate (from 2.4 log₁₀ to 2 log₁₀ CFU). Nevertheless, the compound still outperformed comparator antibiotics, including ceftazidime, avibactam, and aztreonam.²⁷⁸

Figure 9. Penicillin-binding protein inhibitors with in vivo antibacterial activity.

The cephalosporin WYBQ-4 (31) (Figure 9), characterised by its high affinity for PBPs (PBP1, PBP2, PBP3, PBP4, PBP2a), has exhibited promising inhibitory activity against MRSA. In vitro studies have indicated antibacterial efficacy of 31 against S. aureus, including both MSSA and MRSA clinical strains, with MIC values ranging from 0.5 to 16 μg/mL. Further evidence of 31’s potency comes from in vivo studies using a murine lung infection model, which resulted in a remarkable 4 log₁₀ CFU reduction in bacterial load. In a systemic infection murine model, treatment with 31 at a dose of 150 mg/kg achieved a 100% survival rate among MRSA-infected mice, surpassing the efficacy of cefepime (40%) and moxalactam (50%) by more than two-fold at the same dosage. Compound 31 had a favourable toxicity profile, exhibiting no signs of haemolytic activity, acute toxicity, or human cytotoxicity even at considerably high concentrations.²⁷⁹

5.2. Compounds that disrupt the cell membrane

Bacterial survival is heavily reliant on membranes, which play critical roles in metabolic processes and cellular homeostasis.^280,281 As a result, they present an attractive target for antibiotics. ^282,283 Recently, membrane-disrupting strategies have gained significant attention, particularly those employing cationic amphiphile agents.^284–286 Although the precise mechanisms of action vary across this class of molecules, many are thought to initially interact with the bacterial cell envelope and subsequently perturb membrane lipids and proteins, leading to structural disorganisation, leakage of low-molecular-weight components, and intracellular damage.^{280,283,287,288} While antimicrobial peptides have long exemplified membrane-disrupting mechanisms,²⁸⁹ their limitations in stability and manufacturability have shifted attention toward small-molecule alternatives. This section focuses on such compounds, which often exhibit more favourable drug-like properties. Accordingly, the compounds discussed below encompass a range of membrane-active strategies, including direct membrane disruption, permeabilisation, adjuvant-mediated enhancement of antibiotic uptake, and mixed mechanisms in which membrane perturbation contributes to antibacterial activity.

IITR00693 (32) (Figure 10) is a 2-aminoperimidine antibacterial derivative identified from 30 hits obtained through screening a library of 11,000 small molecules.²⁹⁰ It has shown strong antibacterial activity against S. aureus and E. coli, with a MIC of 8 μg/mL and a favourable safety profile. Compound 32 exhibits antibiotic-potentiating properties, enhancing the spectrum of polymyxin B to include Gram-positive bacteria. This occurs through disruption of the cell surface of S. aureus, which facilitates the binding of polymyxin B to the bacterial cell envelope. The compound’s mechanism of action involves disrupting the proton motive force (PMF). The weakly basic nature of 32 allows it to exist predominantly in the unionised form at physiological pH, enabling passive diffusion into bacterial cells and subsequently leading to a decrease in swimming motility. Combining 32 with polymyxin B creates additional stresses, such as redox imbalance and PMF disruption, damaging the bacterial cell envelope. Time-kill kinetics revealed that the combination exhibited bactericidal activity, achieving a >3 log₁₀ reduction in S. aureus and P. aeruginosa counts. Compound 32’s low tendency to develop adaptive or spontaneous resistance to polymyxin B underscores its potential against MDR strains. In a murine tissue excision infection model, 32 alone reduced the bacterial burden by 1.5 log₁₀ for S. aureus and 0.2 log₁₀ for P. aeruginosa, while the combination therapy resulted in reductions of 6 and 3 log₁₀ CFU, respectively, comparable to the efficacy of polysporin.²⁹⁰

Figure 10. Compounds targeting the bacterial cell membrane.

A series of membrane-permeabilising compounds based on a natural bisindole alkaloid scaffold were developed and synthesised via an optimised one-pot Masuda borylation–Suzuki coupling sequence.²⁹¹ Compounds 33 and 34 (Figure 10) demonstrated antibacterial activity against Gram-positive bacteria with no observed cytotoxicity below concentrations of 12.5 μM and no haemolytic activity up to 100 μM, indicating a good level of antibacterial selectivity. In vivo studies in a mouse wound infection model using the MRSA strain T144 showed that a single topical dose of compounds 33 and 34 (5 mg/kg) achieved efficacy similar to that of vancomycin. In a mouse peritonitis-sepsis model, systemic activity was limited, likely due to poor aqueous solubility. Bacteria exhibited a low tendency to develop resistance to 33, with only a 2- to 4-fold increase in MIC, and there was no cross-resistance observed with other antibiotic classes. While traditional bisindole compounds inhibit pyruvate kinase, these new molecules act by permeabilising bacterial membranes, causing efflux of internal molecules while largely preserving cell wall integrity.²⁹¹ Recent mechanistic studies have confirmed that these bisindole alkaloids also disrupt the cytoplasmic membrane of Gram-positive bacteria by directly interacting with lipid II and membrane phospholipids, which in turn impairs the lipid II biosynthetic cycle.²⁹²

Mason et al. introduced a series of cationic bolaamphiphile compounds designed to disrupt microbial membranes and tested for broad-spectrum antimicrobial activity. These agents showed membrane-permeabilising effects across a wide range of bacterial, mycobacterial, and fungal pathogens.²⁹³ Among these analogues, 35 (Figure 10) with a C12 acyl chain stood out for its rapid bactericidal activity and excellent antibacterial efficacy, with MIC values ranging from 0.125 to 1 μg/mL against Gram-positive bacteria, fungi, and mycobacteria, while showing moderate activity against Gram-negative bacteria. Compound 35 showed high in vitro toxicity (EC₅₀ 3.8 μM in HEK293 and 4.5 μM in HeLa cells), but it also had a selectivity index of 17.4, supporting its further evaluation as a lead against Gram-positive bacteria. It also enhanced the antibacterial activity of aminoglycosides against A. baumannii and P. aeruginosa, suggesting it could reverse resistance to tobramycin and gentamicin in aminoglycoside-resistant strains. In a Galleria mellonella infection model with E. coli ATCC 25922, 35 administered at varying doses (0.25, 1, 4, or 10 mg/kg) in combination with 5 mg/kg tobramycin, did not enhance larval rescue in the experimental setting. In fact, the lowest dose of 35 hindered the therapeutic effect initially observed with tobramycin, although it did not exacerbate the infection further. Interestingly, while 35 improved outcomes in infected G. mellonella larvae, an analogue with a shorter C6 chain length worsened the infection, likely due to differing self-assembly properties.²⁹³

A series of symmetrical lysine-derived compounds featuring aromatic linkers was developed to selectively target MRSA by mimicking membrane-disruptive mechanisms.²⁹⁴ Compound 36 (Figure 11) was the standout candidate, showing strong antibacterial activity against clinical MRSA isolates and achieving an excellent selectivity index of over 2000. Its MIC was 2 μg/mL against 90% of MRSA and Enterococcus faecium strains. It also had effective activity against carbapenemase-producing Gram-negative bacteria, with MIC values ranging from 2 to 8 μg/mL. Time-kill kinetics indicated that compound 36 effectively inhibited bacterial growth within the first 4 hours, maintaining this effect for 24 hours. The proposed mechanism of action for compound 36 involves the interaction with the cell envelope, including peptidoglycan and membrane components, thereby disrupting envelope integrity and interfering with bacterial metabolism, ultimately leading to cell death. The compound’s antibacterial activity is further supported by its ability to induce endogenous reactive oxygen species (ROS). In vivo efficacy was confirmed using a mouse model of MRSA subcutaneous infection, where compound 36 (at 5 and 10 mg/kg) exhibited comparable antibacterial efficacy to vancomycin, achieving a reduction in bacterial burden of 2 and 3 log₁₀ CFU/abscess in the low-dose group and high-dose group, respectively. Treatment with 36 reduced neutrophil counts in skin tissue.²⁹⁴

Figure 11. Cationic amphiphilic compounds displaying in vivo antibacterial efficacy.

Carbazole-based cationic amphiphilic compounds were developed, with compound 37 (Figure 11) identified as a lead candidate due to its strong antibacterial efficacy and minimal haemolytic effects.²⁹⁵ Compound 37 had antibacterial efficacy comparable to vancomycin against Gram-positive bacteria, including MRSA, with MIC ranging from 0.78 to 1.56 µg/mL. It also exhibited minimal haemolytic activity with a high HC₅₀ value of over 200 µg/mL and low mammalian cell toxicity, resulting in a high selectivity index of over 128. The study indicated that 37 rapidly kills bacteria by increasing inner membrane permeability and disrupting membrane integrity, leading to cell death. This membrane-disrupting action was confirmed through scanning electron microscopy and transmission electron microscopy, which detected membrane perforations. The compound also showed a dose-dependent interaction with bacterial surface lipoteichoic acid. Compound 37 had comparable in vivo efficacy to vancomycin in a murine corneal infection model, where topical administration resulted in a 4.1 log₁₀ CFU reduction in bacterial counts, nearly matching vancomycin’s 3.9 log₁₀ CFU reduction.²⁹⁵

Later, the same group presented amphiphilic coumarin derivatives as membrane-active antibacterial agents, with compound 38 (Figure 11) standing out for its broad-spectrum antibacterial activity.²⁹⁶ It displayed a MIC value of 0.31–0.78 µg/mL against Gram-positive bacteria and 6.25 µg/mL against Gram-negative bacteria, along with low haemolytic activity (HC₁₀ = 43.5 µg/mL). Compound 38 exhibited rapid bactericidal activity against MRSA N315 and S. aureus ATCC 29213, reducing bacterial counts by over 3 log₁₀ CFU/mL within 30 minutes at 8x MIC and within 4 hours at 4x MIC, with no regrowth observed over 24 hours. This rapid bactericidal effect was attributed to enhanced membrane permeabilisation of bacterial cells. Compound 38 exhibited a strong affinity for lipoteichoic acid, mediated by electrostatic interactions between the cationic motif and the negatively charged lipoteichoic acid. Resistance development studies indicated a low tendency to develop resistance, with no more than a 4-fold increase in MIC value observed over 20 days of serial passaging. In an in vivo efficacy assessment using a murine corneal infection model, topical administration of 0.5% compound 38 four times a day for three days led to a 5.3 log₁₀ CFU reduction in bacterial burden against S. aureus ATCC 29213 compared to the negative control group.²⁹⁶

In 2022, the same research team developed guanidine derivatives featuring a tetrahydroquinoline scaffold, incorporating cationic and hydrophobic groups.²⁹⁷ The leading compounds, 39 and 40 (Figure 12), had broad-spectrum antimicrobial activity against Gram-positive bacteria (MIC values of 0.78 - 1.56 µg/mL), Gram-negative bacteria (MIC values of 1.56 - 6.25 µg/mL), and C. albicans (MIC values of 0.78 - 1.56 µg/mL). These compounds retained their MICs even after 21-22 passages, indicating a low likelihood of resistance development. These compounds not only exhibited antibacterial activity against planktonic bacteria but also inhibited biofilm formation at low concentrations (1.56 - 3.125 µg/mL) and eradicated existing biofilms at higher concentrations (>3.125 µg/mL). Both compounds showed low haemolytic activity (HC₅₀ >200 µg/mL) and minimal cytotoxicity. Rapid bactericidal and fungicidal activity was attributed to membrane integrity disruption, as confirmed by studies using SYTOX™ Green, SYTO™ 9, propidium iodide, and 1-N-phenylnaphthylamine. In a murine bacterial keratitis model, topical treatment with 39 and 40 (0.5% four times daily for 3 days) resulted in reductions in bacterial load: a 6.6-log and 6.3-log reduction against S. aureus and a 3.2-log and 3.4-log reduction against P. aeruginosa, respectively, relative to the negative control. These results were comparable to those of vancomycin and gatifloxacin, highlighting their potential as antimicrobial agents.²⁹⁷

Figure 12. Guanidine compounds that disrupt bacterial membranes.

The same researchers reported another guanidine derivative based on a phenyl sulfide scaffold chosen because its antibacterial and antibiofilm activity had been observed in earlier studies.²⁹⁸ The team employed a similar strategy by incorporating basic cationic moieties and hydrophobic groups to achieve amphiphilic properties. The lead compound 41 (Figure 12), which contained two arginine residues, demonstrated activity against eight Gram-positive bacteria (MIC 0.39 - 1.56 µg/mL), with low haemolytic activity (HC₅₀ >200 µg/mL), minimal in vitro cytotoxicity, and a low tendency to induce bacterial resistance. The presence of two positive charges from the arginine residues in compound 41 appeared optimal for antibacterial activity. A higher count (more than four positive charges) or a lower count (one arginine group) reduced antibacterial effectiveness. SYTOX™ Green and LIVE/DEAD BacLight assays confirmed the membrane-disrupting activity of compound 41, which was validated by transmission electron microscopy, showing damaged bacterial membranes with noticeable perforations. Compound 41’s efficacy was further tested in a murine corneal infection model induced by S. aureus ATCC 29213. Topical administration of the compound led to a 6.8 log₁₀ CFU reduction in bacterial burden compared to the untreated control group.²⁹⁸

A recent study by the same research group introduced a novel class of phenothiazine-based cationic amphiphilic derivatives incorporating arginine residues, which exhibited promising broad-spectrum antimicrobial activity.²⁹⁹ The synthesis and characterisation of these compounds, particularly compound 42 (Figure 13), featuring a heptyl group and two arginine residues, revealed antimicrobial efficacy against both Gram-positive (MICs = 1.5 μg/mL) and Gram-negative bacteria (MICs = 3.1−6.2 μg/mL), with minimal haemolysis (HC₅₀ = 281 μg/mL) and cytotoxicity (CC₅₀ > 50 μg/mL) towards mammalian cells. In vivo assessments using a murine corneal infection model indicated reductions in viable bacterial colonies for both S. aureus and P. aeruginosa, with compound 42 exhibiting comparable efficacy to vancomycin against S. aureus and tobramycin against P. aeruginosa. Topical application of a 1% solution of compound 42 to infected murine corneas resulted in a 5.28 log reduction in viable S. aureus colonies, comparable to vancomycin’s effect. Against P. aeruginosa, compound 42 and tobramycin achieved reductions of 5.93 log and 4.71 log, respectively. Compound 42 did not induce corneal injury in treated mice, indicating its potential for treating keratitis caused by both Gram-positive and Gram-negative bacteria. Mechanistic studies using SYTOX Green and LIVE/DEAD BacLight assays reported that the compound induced dose-dependent membrane permeabilisation, leading to a marked loss of bacterial viability.²⁹⁹

Figure 13. Cationic amphiphilic compounds with membrane-perturbing effect.

Bazan et al. reported a series of cationic distyrylbenzene-conjugated oligoelectrolytes (COE) with amidine groups as membrane-active compounds.³⁰⁰ Among the synthesised amidine-functionalised COEs, compound 43 (Figure 13), containing a cyclopropyl group and unsubstituted amino and imino nitrogen, emerged as the most promising. It exhibited a MIC of 0.5 µg/mL against S. aureus and 1-4 µg/mL against Gram-negative bacteria including E. coli, K. pneumoniae, and P. aeruginosa. This compound also showed low cytotoxicity toward HepG2 cells (IC₅₀ 1024 µg/mL), yielding a selectivity index of 512, and no haemolysis at 492 µg/mL. Compound 43 retained antibacterial activity across the AMR phenotypes represented in the test panel. It exhibited rapid bactericidal activity against P. aeruginosa PAO1, achieving a 3 log₁₀ CFU reduction at 4× MIC; however, regrowth was observed after 48 hours, which was overcome by using 8× MIC. Compound 43 had a low propensity for bacteria to develop adaptive or spontaneous resistance. Moreover, 43 showed biofilm inhibition, with a minimum biofilm inhibitory concentration (MBIC) that inhibited 50% and 90% of biofilm growth at 0.125× MIC and 0.5× MIC, respectively. It could also eradicate established biofilm at 32× MIC. In vivo efficacy was tested using a wound infection model with PAO1 in mice. Treatment with 43 at concentrations of 0.1%, 0.5%, and 2% w/v on the wound reduced the bacterial burden in a dose-dependent manner.³⁰⁰

Another research group developed a new COE lead compound, 44 (Figure 13), which was identified through an in vitro efficacy/cytotoxicity screening process.³⁰¹ This compound exhibited broad antibacterial activity, with MIC values ranging from 0.5 to 8 µg/mL, against clinically relevant Gram-positive and Gram-negative bacteria, including MRSA and CRE K. pneumoniae. Compound 44 also displayed low cytotoxicity towards Hep-2 and RAW 264.7 cells and showed no in vivo toxicity, as assessed by general tolerability scoring based on appearance, activity, and clinical signs. In murine sepsis models with K. pneumoniae and MRSA, 44 at a dose of 2 mg/kg/day for 3 days increased survival rates in both models, indicating high protective efficacy. The compound was found to affect multiple membrane-associated functions in a dose-dependent manner. At subinhibitory concentrations, it blocked cell septation, reduced ATP levels by 1.7-fold, inhibited motility, and increased cloxacillin sensitivity by up to 4-fold while decreasing cell viability by 3-fold. At 2 × MIC, increased membrane permeability was observed, along with a 2-to 3-fold reduction in ATP levels and respiration, and up to a 1000-fold decrease in bacterial viability. Compound 44 had low potential for resistance development.³⁰¹

Wang et al. tackled the challenge of developing new antibiotics by using an antibiotic adjuvant strategy.³⁰² Through library screening and initial assessment of sensitisation activity and cytotoxicity, they identified 45 (Figure 14) as an adjuvant capable of extending the antimicrobial spectrum of existing antibiotics without itself affecting bacterial growth. Compound 45 exhibited bacterial sensitisation by reducing the MIC of rifampicin against various Gram-negative bacteria, including CRE, colistin-resistant, and other MDR strains, by up to 512-fold. Compound 45 showed a low propensity to induce bacterial resistance and negligible cytotoxicity (IC₅₀ >100 µM). The mechanism of action for 45 involves selective permeabilisation of the outer membrane of Gram-negative bacteria. The compound binds to the negatively charged phosphate groups of lipopolysaccharides on the bacterial outer membrane, disrupting its integrity, and enhancing antibiotic uptake. This was further supported by simulations showing 45’s interaction with ion-bridged phosphates, weakening the interactions between ions and phosphates. In an ex vivo skin infection model, 45 enhanced the efficacy of novobiocin and clindamycin against E. coli and restored polymyxin B sensitivity toward mcr-1-expressing E. coli. The compound’s effectiveness was further validated in a systemic infection model in mice. A single dose of 45 (10 mg/kg) combined with novobiocin (80 mg/kg) increased the survival rate to 93% in cases of WT E. coli (ATCC 25922) infection and to 87% in mice infected with NDM-1-expressing E. coli.³⁰²

Figure 14. Pentamidine derivatives (45 and 46) and a peptidomimetic derivative (47) as antibiotic adjuvants.

Another group introduced a series of pentamidine analogues as antibiotic adjuvants, with compound 46 (Figure 14) identified as a promising candidate.³⁰³ Compound 46, featuring a central carbon-to-nitrogen substitution in the pentamidine linker region, exhibited strong synergistic activity when combined with novobiocin against A. baumannii, with a fractional inhibitory concentration index (FICI) of 0.094, and showed half-maximal activity to HEK293 cells (AC₅₀ 235 µg/mL). The synergistic activity with novobiocin was further evaluated in a murine model of systemic A. baumannii infection. The combination of 46 (10 mg/kg) with novobiocin (50 mg/kg) led to a 3 log₁₀ CFU reduction, demonstrating greater potentiating activity compared to its parent molecule, pentamidine, which achieved a 2 log₁₀ CFU reduction with novobiocin. Compound 46’s ability to bind LPS and its reduced synergy with novobiocin in the presence of exogenous Mg²⁺ suggested a similar mechanism of action to pentamidine by directly interacting with lipid A phosphates to compromise outer membrane integrity. Although it interacts with LPS, 46 maintained synergistic activity with novobiocin in bacteria expressing the polymyxin resistance gene mcr-1, which masks the negative charge of lipid A, indicating its potential to overcome resistance mechanisms. Unlike pentamidine, 46 exhibited reduced impact on hERG channel trafficking, thereby mitigating the risk of QT prolongation.³⁰³ These findings support further evaluation of compound 46 as an outer-membrane-active antibiotic adjuvant.

Song et al. introduced a new series of peptidomimetic derivatives designed to act as antibiotic adjuvants, aimed at restoring the sensitivity of antibiotics against P. aeruginosa.³⁰⁴ Compound 47 (Figure 14) was developed through the structural optimisation of dipeptide PAβN by inserting a methylene group at the α-position of benzene. It exhibited comparable potentiating activity to PAβN against P. aeruginosa strains, enhancing azotreonam (AZT) efficacy by up to 16-fold, without displaying antibacterial activity when used alone. However, it did not exhibit synergistic effects against A. baumannii and only showed additive effects when combined with other classes of antibiotics. While compound 47 had AZT-sensitising activity, it exhibited low cytotoxicity towards HEK293 and LX2 cells, with a CC₅₀ value exceeding 100 μM. The in vivo efficacy of 47 was evaluated using the G. mellonella lethal infection model with PAO1 and 16-2m strains. Treatment with the AZT/47 combination (25 mg/kg AZT and 100 mg/kg 47) improved survival rates, achieving 40% and 80% at 96 hours post-inoculation for PAO1 and 16-2 strains, respectively, whereas AZT alone yielded much lower survival rates of 0% and 30%, respectively. They demonstrated that 47 exerts antibiotic-potentiating activity through inhibition efflux pumps, specifically by binding to the active site of MexB through hydrogen bonding and hydrophobic interactions with amino acid residues, and disruption the outer membrane through electrostatic interactions.³⁰⁴

Recent research has identified benzopyridone cyanoacetates as promising broad-spectrum antibacterial agents, exhibiting low MIC values against both Gram-positive and Gram-negative bacteria.³⁰⁵ Compound 48 (Figure 15) had MIC values ranging from 0.25 to 16 μg/mL against Gram-positive bacteria and 0.25 to 0.5 μg/mL against Gram-negative strains, while displaying minimal haemolytic activity at concentrations up to 128 μg/mL. In a mouse infection model using K. pneumoniae p1-6305014, compound 48 reduced bacterial counts by approximately 1.5 log₁₀ CFU at a 5 mg/kg dose and by about 4 log₁₀ CFU at 15 mg/kg. PK studies in mice indicated that 48 exhibited instability in plasma following intravenous administration, suggesting it may be more suitable for external applications. The half-life was 2.21 hours for oral administration and only 0.2 hours for intravenous administration, indicating the need for further optimisation to improve PK properties. Mechanistic studies suggested that compound 48 exerts multiple effects, including biofilm disruption and membrane damage, leading to bacterial cytoplasm leakage. The compound’s structure may also bind to intracellular DNA, DNA gyrase, and topoisomerase IV through various noncovalent interactions, potentially inhibiting bacterial replication.³⁰⁵

Figure 15. Compounds with membrane-disturbing activity against Gram-positive and Gram-negative bacteria.

Recently, Haldar et al. introduced a small molecule adjuvant designed to repurpose antibiotics for treating Gram-negative bacterial infections.³⁰⁶ By targeting bacterial membranes, compound 49 (Figure 15) shows promise in combating antibiotic resistance. The compound achieves moderate membrane perturbation in bacterial cells without causing mammalian cell toxicity, thereby enhancing antibiotic internalisation. The compound’s unique balance of hydrophilic and hydrophobic groups enables it to interact with the lipid bilayer of Gram-negative bacteria. Although 49 lacks direct antibacterial activity, it potentiates the effects of various antibiotics, including rifampicin, fusidic acid, minocycline, and chloramphenicol, when used in combination. In vivo toxicity studies revealed that 49 exhibits an LD₅₀ of 130 mg/kg for intraperitoneal administration, and subcutaneous administration at 175 mg/kg showed no adverse effects, indicating that 49 is biocompatible for both systemic and topical applications. The in vivo efficacy of a combination therapy consisting of 49 and fusidic acid was assessed in a murine skin infection model. When administered alone, 49 at a dosage of 40 mg/kg achieved a moderate reduction of 0.2 log in bacterial count compared to the untreated group. Fusidic acid alone failed to decrease the bacterial burden. The combination therapy of 49 (40 mg/kg) and fusidic acid (40 mg/kg) resulted in a reduction of approximately 2.5 log₁₀ units in bacterial viability of NDM-1-producing A. baumannii compared to the untreated case. Histopathological analysis observed minimal tissue damage in mice treated with the combination therapy, indicating its potency in combating NDM-1-producing bacterial-induced infections. Fluorescence assays with A. baumannii A168 and K. pneumoniae R3934 revealed that 49 exerts moderate effects on outer membrane permeabilisation and cytoplasmic membrane depolarisation, respectively, thereby confirming its capacity to disrupt bacterial efflux mechanisms. These findings support the use of 49 as a membrane-active adjuvant that enhances intracellular antibiotic accumulation.³⁰⁶

Recent studies have explored various agents that disrupt bacterial membranes. Yi et al. introduced a phenylthiophene pyrimidinediamine derivative, 50 (Figure 15), targeting bacterial membranes in E. coli and S. aureus, with MIC values of 2 and 3 μg/mL, respectively.³⁰⁷ Compound 50 causes membrane potential dissipation and content leakage, resulting in cell death. This disruption is likely due to its high lipophilicity (cLogP 6.83). In mouse bacteraemia models infected by S. aureus, compound 50 reduced viable pathogen counts in the peritoneal cavity at low doses (12 mg/kg), albeit less effectively than polymyxin B. At high doses (24 mg/kg), compound 50 exhibited superior antibacterial efficacy compared to polymyxin B.³⁰⁷ In another study, He et al. investigated a triphenylphosphonium chloramphenicol conjugate as a membrane-targeting agent against MRSA infections.³⁰⁸ The design centred on the mitochondria-targeting vector triphenylphosphonium cation and the broad-spectrum antibiotic chloramphenicol. Conjugate 51 (Figure 15) exhibited activity against MRSA strains (MIC = 1 μg/mL), significantly higher than chloramphenicol alone (MIC > 64 μg/mL). Mechanistic studies showed that compound 51 disrupts bacterial cell membranes, leading to content leakage and bacterial death. In vivo studies in a mouse model infected with MRSA-USA300 confirmed that treatment with 51 (20 mg/kg) reduced bacterial load in infected thighs, comparable or superior to vancomycin. Conversely, chloramphenicol (10 mg/kg) showed no effect against MRSA-USA300.³⁰⁸

Recent advances in the development of membrane-disrupting antibiotics have introduced a range of structurally diverse compounds with novel mechanisms and promising preclinical profiles. One such example is compound 52 (Figure 16), a broad-spectrum antibiotic developed through structure-based drug design and modular synthesis.³⁰⁹ It targets ornithine carbamoyl transferase (arcB), an essential enzyme in bacterial arginine metabolism. By competitively binding to arcB, 52 disrupts bacterial metabolism with downstream effects on membrane permeability and oxidative stress, leading to bacterial cell death. It exhibits activity against both Gram-positive and Gram-negative bacteria, including resistant strains, with MIC values ranging from 4 to 16 µM. In a murine sepsis model caused by S. aureus, a 60 mg/kg dose of compound 52 reduced bacterial loads in liver, spleen, kidneys, and blood, improving survival rates compared to standard therapies. Remarkably, in infections with florfenicol-resistant S. aureus, treatment resulted in a 50% survival rate at 72 hours, whereas untreated controls showed no survival. Prolonged exposure to sub-inhibitory concentrations implicated in a lower tendency for resistance development relative to conventional antibiotics. Preliminary toxicity studies confirmed a favourable safety profile, with no deaths observed at doses up to 1500 mg/kg over one week.³⁰⁹

Figure 16. Compounds with membrane-disrupting activity against Gram-positive pathogens.

Focusing on Gram-positive bacteria, especially MRSA, compound 53 (Figure 16), a dihydropyrrolidone-thiadiazole derivative, displays MIC values between 6.25 and 12.5 μM.³¹⁰ Its antibacterial activity stems from dual inhibition of the YycG/F two-component regulatory system, critical for cell wall homeostasis and virulence, coupled with direct membrane disruption through cardiolipin binding. Transmission electron microscopy reported extensive membrane damage and cytoplasmic leakage in treated bacteria. Compound 53 enhanced the activity of daptomycin, reducing bacterial load and eradicating persister cells that typically evade antibiotics. In vivo, MRSA-infected mice treated with 53 exhibited substantial bacterial load reductions alone and an enhanced effect when combined with daptomycin, achieving a 3.5 log₁₀ CFU reduction. Toxicity studies at 50 mg/kg indicated no liver or kidney dysfunction or histopathological changes, while resistance development was not observed after 20 serial bacterial passages.³¹⁰

From a different mechanistic angle, compound 54 (Figure 16) was identified through virtual screening against bacterial signal peptidase I (SPase I) and optimised into a tetrahydroacridine-9-carboxylic acid derivative.³¹¹ It selectively inhibits SPase I, causing accumulation of unprocessed proteins and impaired virulence, while also disrupting membrane integrity to induce leakage and cell lysis. In a neutropenic murine MRSA skin infection model, topical administration of 1% and 2.5% (w/v) 54 reduced bacterial counts by 1.2 and 2.3 log₁₀ CFU, respectively, outperforming ampicillin though less potent than linezolid. Compound 54 inhibited biofilm formation by 31.4% at 2 μg/mL, addressing a critical challenge in chronic infections. PK profiling confirmed excellent metabolic stability across human, rat, and mouse liver microsomes (half-life >184.78 minutes) and low intrinsic clearance (<7.50 μL/min/mg), suggesting potential for systemic exposure and oral administration. Toxicological assessments in ICR mice showed no acute toxicity at doses up to 100 mg/kg and no significant organ pathology.³¹¹

Targeting bacterial membranes through peptidomimetic design, compound 55 (Figure 16) is an amphiphilic chalcone derivative combining cationic and hydrophobic elements to selectively disrupt bacterial membranes.³¹² It exhibits broad-spectrum activity with MICs ranging from 0.5–1 μg/mL against Gram-positive and 1–32 μg/mL against Gram-negative bacteria, alongside minimal haemolytic toxicity (HC₅₀ = 329.3 μg/mL). Mechanistic studies validated that 55 depolarises bacterial membranes, increases permeability, and promotes intracellular ROS accumulation, culminating in bacterial death. It also strongly inhibits biofilm formation, shows high proteolytic stability, and exhibits a low tendency for resistance development. In vivo, compound 55 administered at 5 and 10 mg/kg reduced bacterial loads in liver (2.5 log₁₀), spleen (1.9 log₁₀), kidney (1.6 log₁₀), and blood (0.7 log₁₀) in a mouse sepsis model infected with S. aureus ATCC 43300 and E. coli ATCC 25922, with efficacy comparable to colistin. Histopathological analysis confirmed reduced inflammation and tissue damage. PK data indicated retention of antibacterial activity in plasma over time, and toxicity testing established a maximum tolerated dose of 50 mg/kg with full survival at lower doses.³¹²

Similarly, the same group reported compound 56 (Figure 16), an amphiphilic benzopyran derivative designed to mimic antimicrobial peptides.³¹³ It exhibited activity against Gram-positive bacteria, with MICs of 1–4 μg/mL, rapid bactericidal kinetics, low frequency of resistance, and strong biofilm inhibition. It disrupts bacterial membranes, increasing intracellular ROS, and causing leakage of proteins and DNA. In a murine sepsis model infected with S. aureus ATCC 43300, compound 56 reduced bacterial loads in blood and organs by over 3 log₁₀ CFU at 5 and 10 mg/kg doses, with efficacy comparable to vancomycin. Histopathology found diminished tissue inflammation, and cytokine profiling showed reduced infection-induced pro-inflammatory markers. The compound also displayed high plasma stability and minimal haemolytic toxicity (HC₅₀ = 131 μg/mL).³¹³

5.3. Compounds that inhibit DNA synthesis

Small molecules may interfere with DNA through several binding mechanisms.³¹⁴ These include intercalation, where the molecule inserts itself between base pairs; alkylation, which involves covalent bonding with specific DNA bases; and binding to the major or minor grooves, where the molecule interacts with the structural spaces in the DNA helix. These mechanisms can alter DNA’s function and structure, thereby influencing transcription, replication, and repair processes. The replication of DNA at the molecular level hinges on the efficient functioning of the replisome, a highly specialised multi-protein complex dedicated to ensuring the accurate duplication of genetic material.³¹⁵ These proteins also represent important targets for antibiotic development.³¹⁶

A study conducted by researchers at Roche described the discovery and assessment of new compounds with antibacterial activity against Gram-negative bacteria by targeting DNA gyrase and topoisomerase IV.³¹⁷ The researchers used a structure-based drug design approach to identify pyrido[2,3-b]indole derivatives as inhibitors of these key bacterial enzymes. Compound 57 (Figure 17) exhibited potency against a range of Gram-negative bacteria, including MDR strains. This broader activity spectrum was associated with increased rigidity through a ring-cyclisation strategy, supported by X-ray crystallography. Compound 57 exhibited a suboptimal PK profile, characterised by a short half-life of 1.6 hours and a high degree of PPB (97%). To assess in vivo activity, the team evaluated compound 57 in a neutropenic mouse thigh infection model using E. coli ATCC 25922. The results showed dose-dependent bactericidal activity, leading to a 1 log₁₀ CFU reduction compared to the control group.³¹⁷

Figure 17. Compound 57 targeting bacterial DNA synthesis.

Fluoroquinolones are broad-spectrum bactericidal antibiotics that target DNA gyrase and topoisomerase IV.³¹⁸ Resistance is increasingly common, driven mainly by target-site mutations and enhanced efflux.³¹⁹

A team at Novartis Institutes for BioMedical Research embarked on a study to discover and optimise inhibitors of DNA gyrase and topoisomerase IV, focusing on compounds effective against fluoroquinolone-resistant Gram-positive bacteria.³²⁰ Employing a scaffold morphing approach, they designed antibacterial agents that incorporated elements from quinazolinedione and fluoroquinolone.

Initially, compound 58 (Figure 18) was identified as a lead candidate due to their strong antibacterial activity (MIC 0.06-1 µg/mL) against both fluoroquinolone-sensitive and resistant Staphylococcus aureus strains, with low cytotoxicity (>100 µM) and minimal inhibition of hERG (IC₅₀ of 109 µM). Further optimisation aimed at improving safety and PK profiles led to the development of compound 59 (Figure 18), with a tricyclic core derived from levofloxacin and retaining the scaffold from compound 58. Compound 59 displayed antibacterial activity (MIC ≤ 0.03-2 µg/mL), reduced hERG inhibition (IC₅₀ > 300 µM), and favourable ADME properties. The X-ray co-crystal structure of compound 59 with K. pneumoniae topoisomerase IV confirmed a unique binding mode that differed from quinolones, lacking direct interactions or water-metal ion bridges with the quinolone resistance-determining region (QRDR). This divergence from traditional quinolone binding explained the sustained activity of compound 59 against QRDR mutations. Notwithstanding its lower in vitro potency compared to moxifloxacin, 59 showed strong in vivo efficacy in a neutropenic murine thigh infection model when administered subcutaneously at lower dosages, which was attributed to its enhanced PK properties, resulting in relevant higher free plasma exposures.³²⁰

Figure 18. Compound 59 targeting topoisomerase IV, shown alongside the co-crystal structure of 59 (in yellow) bound to K. pneumoniae topoisomerase IV (in grey), highlighting key interactions and binding sites (PDB code: 7LHZ).

Bhawar et al. investigated a series of fluoroquinolone-based derivatives designed to combat drug-resistant S. aureus and S. pneumoniae. The enantiomers 60 and 61 (Figure 19), along with their racemic parent, exhibited antimicrobial activity, with MIC values ranging from 0.1 to 0.8 µg/mL against MDR strains. Compound 60 had a favourable safety profile, with a low incidence of chromosomal aberrations (3.8%) in bone marrow cells at 24 hours post-administration in Swiss mice, compared to 61 (12.0%). Compound 60 displayed an oral ED₅₀ of 20 mg/kg in a mouse sepsis model of infection (MDR S. pneumoniae SPN 718), outperforming other fluoroquinolones such as levofloxacin, moxifloxacin, and gemifloxacin, which required ED₅₀ values exceeding 100 mg/kg to achieve comparable efficacy.³²¹

Figure 19. Small-molecule inhibitors targeting bacterial DNA synthesis.

Compound 62 (Figure 20), a bacterial topoisomerase inhibitor, represents a significant advance in the design of antibacterials.³²² Originating from polarity-focused optimisation, the molecule was designed to improve antibacterial activity, metabolic stability, and safety, particularly in reducing hERG channel inhibition. This compound, based on a fluoronaphthyridine scaffold, showed strong antibacterial effects against S. aureus strains, with a MIC of 0.25 µg/mL, and exhibited a low frequency of spontaneous resistance due to its dual inhibition of DNA gyrase and topoisomerase IV (inhibition ratio of 16). Compound 62 had a safe profile with an hERG IC₅₀ of 103 µM and no cytotoxicity towards THP-1, HepG2, and HeLa cell lines (IC₅₀ > 261 µM), in addition to improved metabolic stability. In an in vivo murine model of sepsis induced by MRSA ATCC 33591, treatment with compound 63 (Figure 20), an optimised version of compound 62, at 25 mg/kg and 50 mg/kg doses achieved a 100% survival rate, contrasting with 0% survival in untreated controls. These outcomes underscore the potential of compound 63 as a therapeutic agent for MRSA infections, demonstrating both potent antibacterial activity and a favourable safety profile in preclinical settings.³²²

Figure 20. Compound 62, shown alongside the co-crystal structure of 62 (in yellow) bound to S. aureus DNA gyrase (in grey) (PDB code: 7MVS), along with its optimised analogue, compound 63.

Taisho Pharmaceutical’s study on TP0480066 (64) (Figure 21), a topoisomerase inhibitor, explored its therapeutic potential for treating gonorrhea.^323,324 TP0480066 exhibited substantial antibacterial activity against N. gonorrhoeae, including strains resistant to ciprofloxacin, with MIC values ranging from ≤0.00012 to 0.0005 µg/mL. The compound effectively reduced the number of viable cells in susceptible, ciprofloxacin-resistant, and ceftriaxone-resistant strains by over 3 log₁₀ CFU/mL. The in vivo efficacy was confirmed in a murine infection model, where doses of 30 and 100 mg/kg of 64 decreased mean viable cell counts for both ciprofloxacin-susceptible and ciprofloxacin-resistant strains. Resistance to 64 emerged readily. All TP0480066-resistant N. gonorrhoeae isolates carried the same ParE mutation (Thr169Ile), analogous to Thr165 in E. coli GyrB near the ATP-binding site. There were no observed alterations in the gyrB gene, indicating that 64 may function as a single-target inhibitor, possibly contributing to the observed frequency of resistance. While 64 shows therapeutic potential, additional optimisation is necessary to mitigate the risk of resistance development.³²³

Figure 21. Compounds that interfere with DNA integrity and replication enzymes.

Zhou et al. explored a series of azolylpyrimidinediol derivatives as DNA-groove binders with activity against A. baumannii.³²⁵ Through a systematic SAR study, compound 65 (Figure 21), a triazolylpyrimidinediol, emerged as the lead compound following structure-guided optimisation. It showed antibacterial activity against various Gram-negative bacteria, excluding E. coli and P. aeruginosa ATCC 27853, with MIC values between 0.001 and 0.008 mmol/L. Compound 65 exhibited 32-fold greater potency than sulfadiazine and 3-fold greater potency than norfloxacin. Compound 65 also had favourable PK profiles in mice. Following intravenous administration of 65 at a dose of 10 mg/kg, a peak concentration of 1990 ng/mL was achieved, with a half-life of 2.79 hours. The compound exhibited moderate plasma clearance, at a rate of 506 mL/h/kg. When administered orally, 65 showed improved drug exposure, with an area under the curve (AUC_0-∞) of 8078 h*ng/mL and a prolonged half-life of 4.39 hours. The volume of distribution was determined to be 7890 mL/kg, with a bioavailability of 35.4%. Compound 65 did not exhibit cytotoxicity towards LO2 and 293T cells, nor did it display haemolytic activity. Mechanistic studies indicated that 65 binds to DNA through groove binding without intercalating or cleaving it, leading to disrupted DNA replication, as supported by docking simulations. It also inhibited cellular metabolic activity, suppressed bacterial growth, and eradicated biofilms. In a mouse model of A. baumannii infection, a dose-dependent response was observed: treatment with 10 mg/kg of 65 reduced viable bacterial count by 1.6 logs, while treatment with 30 mg/kg reduced bacterial burden by 3.5 logs compared to negative controls.³²⁵ These findings support further evaluation of azolylpyrimidinediols as antibacterial agents against A. baumannii infections.

Researchers at Redx Anti-Infectives have developed a series of bicyclic oxazolidinone derivatives that inhibit bacterial type II topoisomerase, exhibiting broad-spectrum antibacterial activity.³²⁶ Molecular docking studies noted that these compounds interact with DNA gyrase through hydrogen bonding with the arginine residue R112, specifically via their carbonyl group. The S,S,S enantiomer, compound 66 (Figure 22), which features a cyclohexyl-oxazolidinone moiety, showed antibacterial activity against both Gram-positive and Gram-negative bacteria, with MICs ranging from <0.008 to 4 µg/mL, while displaying low cytotoxicity towards HepG2 cells, with an IC₅₀ value exceeding 64 µg/mL. Compound 66 yielded superior performance against fluoroquinolone-resistant strains (MIC 0.5-4 µg/mL) compared to levofloxacin and gepotidacin, and selectively targeted bacterial topoisomerase II over human topoisomerase II, with IC₅₀ values of 0.055 µM and 0.01 µM against E. coli DNA gyrase and topoisomerase IV, respectively, versus 14.4 µM against human topoisomerase II. It exhibited a low propensity for spontaneous resistance in E. coli, with a frequency of less than 2.4 x 10^-9 at 4 x MIC, outperforming gepotidacin and ciprofloxacin. Through structural modifications, the team successfully identified compounds with reduced hERG activity, while maintaining antibacterial efficacy. In a mouse thigh infection model infected with A. baumannii NCTC13420, subcutaneous administration of compound 66 at 60 and 120 mg/kg reduced bacterial burden, with a decrease of over 1 log₁₀ CFU compared to pre-treatment levels.³²⁶

Figure 22. Compounds targeting bacterial DNA gyrase and topoisomerase.

Yang et al. introduced a series of pyrimido[4,5-b]indole derivatives as inhibitors of bacterial topoisomerase targeting MDR Gram-negative bacteria.³²⁷ The lead compound 67 (Figure 22) demonstrated a broad spectrum of antibacterial activity, with MICs ranging from 0.125 to 4 µg/mL against drug-resistant clinical isolates. While exhibiting cytotoxicity towards L02 and HEK293 cell lines, with CC₅₀ values of 16 and 19 µM, respectively, its cytotoxic effect remained moderate relative to its antibacterial activity, with a 20-40-fold difference compared to its antibacterial activity against A. baumannii. Compound 67 exhibited a favourable safety profile, as evidenced by its reduced hERG inhibition (IC₅₀ > 40 µM). In a neutropenic mouse thigh infection model against MDR A. baumannii, intravenous administration of 67 at doses of 10, 20, and 30 mg/kg resulted in a moderate dose-dependent reduction in CFU compared to the vehicle control, with the highest dose (30 mg/kg) demonstrating bactericidal activity comparable to levofloxacin. Molecular docking suggested that 67 binds to the DNA gyrase B ATP-binding domain through hydrogen bonding with the Asp73 residue and a water molecule, as well as interaction with the Asp106.³²⁷

Hrast et al. identified compound 68 (Figure 22) as an inhibitor of bacterial topoisomerase, with a reduced risk of hERG-related cardiotoxicity.³²⁸ Compound 68, featuring a tetrahydropyran linker that connects a benzamide to a 1,5-naphthylpyridine moiety, exhibited inhibitory activity against DNA gyrase and topoisomerase IV, with IC₅₀ values below 0.365 µM. It demonstrated a high selectivity, with an IC₅₀ exceeding 30 µM for hERG. The compound showed low cytotoxicity, with an IC₅₀ of 11.24 µM towards HepG2 cells, and in a zebrafish model, it did not exhibit cardiotoxic effects. Compound 68 showed an optimal balance between antibacterial potency and hERG inhibition, with hERG IC₅₀/MIC ratios of ≥953 and ≥492 against S. aureus and MRSA strains, respectively, surpassing gepotidacin (≥139 and ≥281). Evaluation of bacterial resistance development found an 8-fold change in MIC over the initial 7 passages, with MIC retention comparable to ciprofloxacin. In an MRSA infection model in neutropenic mice, intraperitoneal administration of compound 68 at 20 and 40 mg/kg four times daily resulted in a reduction in bacterial load, exceeding 3.8 log₁₀ CFU. Both doses resulted in comparable efficacy, likely due to the compound’s limited solubility.³²⁸

Recently, Mašič et al. reported a study introducing a series of bacterial topoisomerase inhibitors, featuring a benzothiazole core.³²⁹ Through hit-to-lead optimisation, compound 69 (Figure 23) emerged as a lead candidate. This pyrrole-benzothiazole derivative, featuring a morpholine moiety, demonstrated antibacterial activity against a broad range of Gram-positive bacteria, with MIC values below 0.03 µg/mL. Its activity against Gram-negative bacteria was moderate to poor, with the notable exception of K. pneumoniae, which had an MIC of 0.5 µg/mL. The development of bacterial resistance was relatively low in comparison to novobiocin. Compound 69 exhibited inhibition of DNA gyrase and TopoIV at concentrations below 10 nM and 44 nM, respectively, while demonstrating high selectivity for bacterial topoisomerases over human DNA topoIIα, with an IC₅₀ of 17.0 µM. Additionally, it displayed a favourable safety profile, exhibiting no cytotoxicity up to 100 µM against MCF-7 and HepG2 cells, and no inhibition of the hERG channel at 50 µM. This compound showed improved solubility (98 µM) and good metabolic stability in both microsomes and hepatocytes, with a half-life of 103 minutes in human liver microsomes. In vivo PK studies indicated a moderate plasma clearance rate of 54.8 mL/min, accompanied by a moderate volume of distribution of 2.3 L/kg, and a half-life of 1 hour. In a neutropenic mouse model of thigh infection induced by S. aureus ATCC 700699, treatment with intravenous doses of 25 and 50 mg/kg resulted in decreases in CFU, with reductions of 1.4 and 2.7 log₁₀, respectively, when compared to control groups. In comparison, a dose of 50 mg/kg of linezolid resulted in a 2.4 log₁₀ CFU reduction.³²⁹

Figure 23. Compound 69, shown alongside the co-crystal structure of 69 (in yellow) bound to P. aeruginosa DNA gyrase B (in grey), highlighting key interactions and binding sites (PDB code: 8BN6).

Compound 70 (Figure 24), a diarginine-based peptidomimetic, acts through an unusual antibacterial mechanism by inducing intracellular aggregation via DNA-triggered phase separation.³³⁰ Upon entry into bacterial cells, compound 70 binds tightly and selectively to double-stranded DNA, catalyzing the formation of large intracellular aggregates. This aggregation disrupts transcriptional regulation, compromises membrane integrity, and ultimately causes cell lysis. Unlike conventional antibiotics, this mechanism bypasses common resistance pathways, such as enzymatic degradation or target-site mutations. In vitro, it displays activity against MDR S. aureus, including MRSA. Cytotoxicity assays confirm minimal toxicity toward mammalian cells, likely due to weak interactions with eukaryotic lipids like phosphatidylcholine and cholesterol. Its in vivo efficacy is supported by multiple infection models: in a murine sepsis model, compound 70 increased survival to 50% (vs. 0% in controls); in a wound infection model, it reduced bacterial load by ~1.4 log₁₀ CFU within 48 hours and accelerated healing in 73% of animals; and in a subcutaneous abscess model, it performed comparably to vancomycin. PK studies in rats revealed a plasma half-life of ~2.5 hours and systemic exposure (AUC_last ~1104 h·ng/mL) following a 2 mg/kg intravenous dose.³³⁰

Figure 24. Peptidomimetic 70 inducing DNA aggregation and DNA gyrase inhibitor 71.

Compound 71 (Figure 24), a 1,4-dicarbonylthiosemicarbazide derivative, was discovered through structure-based design as a selective inhibitor of S. aureus DNA gyrase, with mechanistic studies confirming its activity (IC₅₀ = 1.8 μM).³³¹ The compound forms key hydrogen bonds with ASP81, GLU58, and GLN91, disrupting ATPase activity and inhibiting DNA supercoiling. Compound 71 demonstrated strong antibacterial activity against Gram-positive pathogens. Metabolic stability studies indicated a microsomal half-life of 28.3 minutes and low intrinsic clearance. In a murine MRSA sepsis model, treatment with 20 mg/kg of compound 71 achieved a 50% survival rate, outperforming vancomycin at equivalent doses. PK analysis reported a plasma half-life of over two hours in mice, supporting its further evaluation as a lead candidate for systemic infection treatment.

5.4. Compounds that disrupt energy production and metabolic pathways

Antibiotics primarily interfere with energy-intensive processes, such as protein synthesis and cell wall biosynthesis, thereby disrupting bacterial metabolic balance.³³² Targeting bacterial metabolic and bioenergetic pathways offers opportunity to develop novel antimicrobials that diverge from the conventional targets of antibiotics, including cell-wall biogenesis, DNA replication, RNA and protein synthesis, and folic acid metabolism.³³³ This approach is particularly important in mycobacterial research. A notable breakthrough occurred with the FDA approval of bedaquiline for the treatment of MDR-TB. Bedaquiline exerts its antibacterial effect by binding to the c subunit of ATP synthase, thereby inhibiting ATP synthesis in both actively replicating and dormant mycobacteria.³³⁴ The approval of bedaquiline has expanded the scope of bacterial bioenergetics as a promising therapeutic target. Another notable example is telacebec, a clinical-stage anti-TB drug candidate that targets the QcrB subunit of the cytochrome bcc complex, thereby disrupting the electron transport chain essential for ATP synthesis in mycobacteria.³³⁵

In addition to core bioenergetic and metabolic pathways, the compounds discussed below also include selected agents acting on essential bacterial processes closely linked to growth and physiological homeostasis.

Bacteria generate ATP through substrate-level phosphorylation of fermentable carbon sources or oxidative phosphorylation via the electron transport chain and ATP synthase. Certain bacteria rely on oxidative phosphorylation for growth when substrate-level phosphorylation is insufficient.³³⁶ Lee et al. explored this by reporting on the discovery of thienopyrimidines as inhibitors of Helicobacter pylori, a bacterium that is becoming increasingly resistant to current treatment regimens.³³⁷ The researchers identified two thienopyrimidine compounds through HTS that selectively inhibited H. pylori over commensal gut microbiota. Four series of new thienopyrimidine compounds were developed to explore SAR around the initial hits. Lead compounds were identified, including compound 72 (Figure 25) with an IC₅₀ of 0.007 μM against H. pylori. Although 72 showed good microsomal stability and permeability, it had low solubility (1 μM) and high PPB (99%). In vivo testing reported no efficacy in a murine infection model, prompting investigation via an ex vivo H. pylori infection model. Compound 72 demonstrated high effectiveness in this model, suggesting that optimisation of PK properties may be necessary for in vivo efficacy. Mode of action studies identified the respiratory complex I subunit NuoD as the target for the lead thienopyrimidines. This enzyme complex is essential for H. pylori, supporting further development of this series.³³⁷

Figure 25. Antibacterial agents targeting energy production.

Menaquinone plays a vital role in the oxidative phosphorylation pathway of Gram-positive bacteria. Targeting the menaquinone biosynthesis pathway, particularly the MenA enzyme, holds promise for combating MRSA.³³⁸ A series of new MenA inhibitors was synthesised and tested against MRSA, with bicyclic substituted compound 73 (Figure 25) exhibiting activity (MIC = 3 µg/mL) and effectively blocking menaquinone biosynthesis. Initial testing in a MRSA infection C. elegans model found compound 73 to be ineffective at 5 and 25 µg/mL.³³⁹ Further evaluation in a pulmonary murine model infected with MRSA reported that 73 reduced bacterial lung load in mice without inducing toxicity.³⁴⁰ The variation in activity across experimental models highlights the importance of selecting the most suitable model for in vivo testing. Further elaboration on this topic will be provided in the perspective section of this review.

Beyond core metabolic pathways, additional essential processes linked to bacterial growth and homeostasis have also yielded promising antibacterial candidates. Among these, FtsZ remains one of the most intensively explored targets. FtsZ is a pivotal cytokinesis protein found in the cytoplasm of most bacteria, akin to tubulin in eukaryotes and it regulates bacterial division by polymerising into microtubule-like protofilaments.³⁴¹ Given its wide occurrence and conservation across bacterial strains, FtsZ stands as a promising target for novel antibacterial agents.³⁴² Currently, TXA709 remains the sole FtsZ inhibitor in clinical development. The research group led by Shutao Ma has identified several FtsZ inhibitors, exemplified by the 4,5-dihydroisoxazole derivative 74 (Figure 26).³⁴³ Compound 74 exhibited notable antibacterial activity (MIC, ≤0.125–0.5 μg/mL) against various S. aureus strains, including MRSA and penicillin-resistant variants. Time-kill experiments confirmed its bactericidal mechanism. Evaluation of in vivo efficacy in a mouse model infected with S. aureus revealed that 74 at 10 mg/kg reduced bacterial burden in the blood comparably to linezolid at the same dose. Despite its superior in vitro efficacy and favourable time-kill kinetics, 74 did not surpass linezolid in vivo, likely due to poor PK properties.³⁴³

Figure 26. Antibacterial agents targeting the FtsZ protein.

The research team also developed a series of 10-methylacridinium derivatives designed as FtsZ inhibitors.³⁴⁴ The design strategy involved simplifying the structures of natural alkaloids sanguinarine and berberine, both known for their quaternary pyridinium cores and recognised as FtsZ inhibitors.^345,346 These new derivatives showed broad antibacterial activity, especially against MRSA and vancomycin-resistant Enterococcus (VRE). Compound 75 (Figure 26) emerged as the most effective, with MIC between <0.125 and 4 μg/mL for several Gram-positive bacteria. Its efficacy surpassed or matched that of known antibiotics like linezolid, ciprofloxacin, and vancomycin. For instance, 75 had an MIC of 4 μg/mL against VRE, considerable lower than vancomycin’s MIC (>64 μg/mL). Compound 75 showed rapid bactericidal action, reducing S. aureus ATCC 25923 populations by more than 99.99% within three hours at a concentration of 4 × MIC. In a murine bacteraemia model with S. aureus ATCC 25923, 75 at 25 mg/kg reduced bacterial loads compared to the control group. At 5 mg/kg, 75’s efficacy was comparable to linezolid, but at higher doses, linezolid was more effective. The simplified planar structure of the new derivatives, compared to the more complex four-membered ring in sanguinarine and berberine, was effective in disrupting bacterial cell division.³⁴⁴

Another study introduced a class of sanguinarine derivatives, known as indolyl-5-methylphenanthridium derivatives, also designed as FtsZ inhibitors.³⁴⁷ The design rationale involved modifying sanguinarine by opening its benzene and dioxane rings on both sides to reduce steric hindrance while maintaining its planar hydrophobicity. This modification allowed the compounds to fit into the narrow hydrophobic groove of the FtsZ protein. The quaternary pyridinium centre, a key functional group, was retained to ensure effective interaction with FtsZ. Compound 76 (Figure 26) exhibited effects against various strains of S. aureus, with MIC of 1 μg/mL against susceptible strains and 2 μg/mL against both MRSA and clinical isolates. In time-kill kinetic studies, 76 reduced S. aureus population to below detectable levels within 24 hours. Compound 76 displayed negligible haemolysis in mice red blood cells, indicating a favourable safety profile. The in vivo efficacy of compound 76 was assessed using a mouse model infected with S. aureus ATCC 25923. Mice treated with 76 at 5 mg/kg showed a slight decrease in bacterial load, while those treated with 25 mg/kg experienced a reduction in bacterial counts, although the reduction was less pronounced compared to linezolid at the same dosage.³⁴⁷

Recently, the same group reported a novel scaffold targeting the FtsZ protein, identified through pharmacophore virtual screening.³⁴⁸ Compound 77 (Figure 26) features a 1-methyl-2,5-diphenylpyridin-1-ium core and has demonstrated effective antibacterial activity against both Gram-positive and Gram-negative bacteria, with MIC ranging from 0.06 to 2 μg/mL for Gram-positive strains and 0.5 to 4 μg/mL for Gram-negative strains. A structural hybridisation strategy was used to incorporate a tert-butyl phenyl substituent, which acts as a membrane-disrupting motif. Molecular dynamics simulations indicated that 77 can block the FtsZ protein’s transition from the tense to the relaxed state, thereby disrupting its polymerisation mechanism. Compound 77 permeabilises bacterial inner and outer membranes, facilitating its entry into bacterial cells and causing leakage of bacterial contents, resulting in bacterial death. Compound 77 has low haemolytic toxicity and proved efficacy in vivo. In a mouse infection model with S. aureus ATCC 25923, compound 77 administered at 25 mg/kg reduced bacterial load in the blood, demonstrating efficacy comparable to linezolid. The compound’s dual-targeting mechanism and low resistance potential suggest it could be a promising candidate for new antibacterial agents.³⁴⁸

PC190723 (78) (Figure 27) was among the earliest lead benzamide ether derivatives identified to inhibit bacterial cell division by disrupting FtsZ function.^349–351 PC190723 showed promising antibacterial activity, but clinical development was limited by poor metabolic stability and formulation issues, largely attributed to the 6-chloro substituent on the thiazolopyridine moiety.^352,353 These limitations prompted efforts to enhance its clinical potential, particularly through the development of prodrug derivatives like the clinical candidate TXA709 (10).²⁰⁷ Other medicinal chemistry campaigns have sought to develop PC190723 analogues with enhanced properties, such as the previously mentioned compound 74. DS01750413 (79) (Figure 27) is another analogue that has shown inhibitory activity against the FtsZ protein in S. aureus, with an IC₅₀ value of 2 μg/mL.³⁵⁴ Specific mutations in the FtsZ enzyme, such as G196S and N263S, completely abolished the activity of 79, while mutations like R236I and T309I retained moderate susceptibility. The compound had greater in vitro bactericidal activity than linezolid, with MICs below 2 μg/mL against a range of Gram-positive bacteria. In a neutropenic murine systemic infection model, 79 at a dosage of 200 mg/kg reduced bacterial loads in the spleen, liver, lungs, and kidneys. It also improved survival rates among treated animals compared to the control group. Compound 79 had moderate solubility but showed acceptable PK in mice and no detectable activity against mammalian tubulin.³⁵⁴

Figure 27. PC190723 analogues targeting the FtsZ protein.

Compound 81 is shown alongside the co-crystal structure of 81 (in yellow) bound to S. aureus FtsZ (in grey), highlighting key interactions and binding sites (PDB code: 8HTB).

Compound 80 (Figure 27) is another promising PC190723 analogue. This compound was designed by replacing the thiazolopyridine moiety with a biphenyl scaffold, based on docking studies that identified a large hydrophobic cavity in the FtsZ protein, allowing room for an additional aromatic group. This modification yielded compound 80 with antibacterial activity against Gram-positive bacteria, especially Bacillus subtilis strains, with an MIC range of 0.008 μg/mL to 0.063 μg/mL. It had favourable pharmaceutical properties, including low cytotoxicity (CC₅₀ > 20 μg/mL), high metabolic stability in human liver microsomes (T_1/2 = 111.98 min), and moderate PK (T_1/2 = 2.26 hours, F = 61.2%). In a murine blood infection model using S. aureus ATCC 25923, compound 80 at 25 mg/kg reduced the bacterial burden, demonstrating efficacy comparable to vancomycin.³⁵⁵

Pilch et al. recently reported TXH9179 (81) (Figure 27), a new benzamide-based FtsZ inhibitor with improved anti-staphylococcal activity compared to earlier-generation compounds like PC190723 (78) and TXA707 (10).³⁵⁶ This next-generation inhibitor incorporates a 6-acetylene group on the thiazolopyridine moiety, demonstrating superior bactericidal efficacy against a wide panel of MSSA and MRSA, including strains resistant to vancomycin and linezolid. The MIC of 81 against these isolates ranges from 0.125 to 0.5 μg/mL, an improvement over TXA707. The 6-acetylene group on the thiazolopyridine moiety forms contacts with key amino acids in a straight conformation, consistent with that previously reported for PC190723. It differs from the bent binding mode for TXA707 in complex with FtsZ. Resistance to 81 was associated with specific mutations in the ftsZ gene, confirming FtsZ as the likely target. A carboxamide prodrug version of 81, named TXH1033, was developed to enhance PK properties. The prodrug rapidly converts to 81 in serum via acetylcholinesterases with a conversion half-life of 1.7 minutes. In mouse models of systemic MRSA-USA100 infection, intravenous administration of the prodrug at doses of 15, 30, 45, and 60 mg/kg resulted in survival rates of 0%, 50%, 100%, and 100%, respectively, with similar outcomes observed when administered orally. These findings indicate that TXH1033 offers enhanced in vivo efficacy compared to earlier prodrugs (TXA709), underscoring its promise for further preclinical development.³⁵⁶

Research on new antibacterial targets has explored various biological processes, such as transcription, replication, and translation. G-Quadruplex (G4) DNA structures in pathogenic bacteria are among these targets.³⁵⁷ A recent study examined the effects of a naphthalene diimide ligand library on both Gram-positive and Gram-negative bacteria, finding that compound 82 (Figure 28), had varying modes of action depending on the type of bacteria.³⁵⁸ MIC values for 82 ranged from 8 μM for E. coli to 4 μM for E. faecium and 16 μM for S. aureus. The study suggests that 82’s antimicrobial activity is linked to its ability to bind and stabilise G4 DNA structures. Interestingly, Gram-positive bacteria, with fewer putative G4 domains, showed higher sensitivity to 82 than Gram-negative bacteria. This disparity may be due to the outer membrane of Gram-negative bacteria acting as a barrier, reducing 82’s efficacy. When combined with an outer membrane-perturbing peptide, the antimicrobial activity of 82 in Gram-negative bacteria improved, bringing MIC values in line with those observed for Gram-positive strains (0.5−8 μM). Compound 82 exhibited bactericidal effects on Gram-negative bacteria and bacteriostatic effects on Gram-positive bacteria, suggesting different inhibition mechanisms in processes related to gene transcription and protein translation. Biophysical studies supported the hypothesis that 82’s antibacterial activity is directly related to its G4-binding capability. In vivo testing using the Galleria mellonella infection model reported no apparent acute toxicity for 82. Treatment with 82 at 5 or 10 mg/kg reduced mortality in S. aureus-infected worms, with survival rates increasing from 30% in the untreated group to 70% in the treated group. A similar protective effect was observed in the E. coli-infected worms, with survival increasing from 50% to 90% after treatment. Variations in the mechanisms of action of 82 and the worms’ ability to metabolise and eliminate the drug may account for the differences observed between S. aureus and E. coli in this study.³⁵⁸

Figure 28. Antibacterial agents targeting metabolic pathways.

Compounds targeting dihydrofolate reductase (DHFR) in the folate biosynthesis pathway show promise against MDR Gram-negative bacteria.³⁵⁹ Derived from a Mycobacterium tuberculosis DHFR inhibitor, the 1,3-diamino-7H-pyrrol[3,2-f]quinazoline 83 (Figure 28) has shown strong inhibitory effects against E. coli DHFR (ecDHFR) and antibacterial activity. Its MIC values ranged from 0.002 to 1 μg/mL against various Gram-negative strains. In vitro assays have confirmed its inhibition of ecDHFR, which was validated by co-crystallisation studies. In a mouse model of E. coli infection, intravenous administration of 83 increased survival rates in a dose-dependent manner, reaching 50% at 15 mg/kg. Compound 83 may also inhibit thymidylate synthase, a downstream enzyme in the folate pathway, hinting at dual enzymatic inhibition.³⁶⁰

Irresistin-16 (84) (Figure 28), a broad-spectrum bactericidal antibiotic, has shown efficacy against both Gram-positive and Gram-negative bacteria, with a low propensity for resistance.³⁶¹ This compound had MIC values of less than 4 µg/mL against a wide range of pathogens, achieving its antibacterial effects by disrupting folate metabolism and membrane integrity, specifically by inhibiting E. coli dihydrofolate reductase FolA. PK studies have confirmed that 84 maintains therapeutic concentrations above the MIC in mice for nearly 48 hours after administration. The maximum tolerated dose of 84 was determined to be 15 mg/kg when administered intravenously, resulting in a peak plasma concentration of 1.4 µg/mL with a half-life of 15.8 hours. This stability was further corroborated by mouse liver microsome studies, which confirmed the exceptional in vivo stability of 84. Compound 84 showed in vivo efficacy in a murine gonorrhea model. BALB/c mice infected with N. gonorrhoeae ATCC 700825 and treated with 10 mg/kg of 84 at 2- and 14-hours post-infection exhibited a reduction in vaginal bacterial load compared to vehicle-treated controls, as measured by CFU counts at 26 hours post-infection. Compound 84 showed promise, but its drug-like properties need optimisation; high hydrophobicity limits aqueous solubility and makes oral dosing impractical.³⁶¹

NAD kinases (NADKs) are promising antibacterial targets because of their critical role in bacterial growth and several metabolic pathways.^362,363 A recent study introduced a series of adenine derivatives that function as bacterial NADK inhibitors, exhibiting bactericidal activity in vitro against S. aureus, including MRSA.³⁶⁴ Researchers used fragment-based drug design to develop a focused library of adenine derivatives with various substituents at N9 and C8 positions. A derivative named NKI1 (85) (Figure 28), a diadenosine linked by a propargyl group, was identified with low micromolar inhibition of Gram-positive bacterial NADK. Compound 85 had high metabolic stability in vitro and showed no toxicity to human cells or mice in vivo. In a murine model of acute hematogenous pyelonephritis induced by S. aureus Xen36, daily intraperitoneal injections of 85 (100 mg/kg) for 7 days led to a reduction in kidney bacterial burden, comparable to that observed in the control group treated with levofloxacin. Similar results were observed in an MRSA infection murine model, where treatment at 100 mg/kg daily decreased bacterial counts by 0.5 log₁₀ CFU/kidney and reduced tissular lesions and cellular damages, resulting in recovery from the infection. The study used a CRISPR-Cas9 approach to knock down the ppnK gene encoding NADK in S. aureus, resulting in a lower NADP(H)/NAD(H) ratio, multiple growth defects, and reinforcing the potential of NADK as a therapeutic target for antibacterial development.³⁶⁴

Membrane-targeting ion-transporting molecules offer a novel approach to antibacterial therapy, distinct from traditional methods.³⁶⁵ By disrupting bacterial membranes, these agents, which can transport metal ions such as Zn²⁺, Mg²⁺, and Cu²⁺, have the potential to combat DR bacteria through an unique mechanism that bypasses common resistance pathways.³⁶⁶ A series of 8-aminoquinoline derivatives has been identified with notable transmembrane Zn²⁺ transport properties.³⁶⁷ In the drug design process, it was theorised that the thiourea group would facilitate ion binding to the quinoline core, while arylthiourea modifications would improve penetration through hydrophobic lipid bilayers. Among these derivatives, compound 86 (Figure 29) emerged with remarkable antibacterial efficacy against S. aureus and MRSA, achieving a MIC of 3.1 μM. Although its high cLogP (6.3) may limit solubility, it may also support membrane penetration, which is important for antibacterial activity. Ion-transport studies confirmed that these compounds function as ionophores, with a pronounced affinity for Zn²⁺ ions. In fact, when combined with ZnCl₂, the antibacterial activity of 86 against S. aureus increased by 6.65 times, whereas other metal ions, like Cu²⁺ or Mg²⁺, had minimal impact. Molecular dynamics simulations illustrated that 86 interacts with Zn²⁺ ions in the lipid bilayer, specifically through interactions with the imidazole ring and sulfur atom, suggesting its role in capturing Zn²⁺ from the solvent and facilitating its transport across the membrane. Compound 86 showed bactericidal effects, reducing the likelihood of resistance development. It exhibited no signs of toxicity in haemolytic and cytotoxicity assays. Combining 86 with Zn²⁺ ions not only boosted its antibacterial potency but also restored the effectiveness of certain antibiotics, such as gentamicin and methicillin, in drug-resistant S. aureus strains. In vivo studies using a mouse model infected with biofilm-associated catheter S. aureus confirmed 86’s antibacterial efficacy. Treatment with 86 alone, and especially in combination with ZnCl₂, significantly reduced bacterial load in liver and spleen tissues, also restoring normal tissue histology. These findings support further evaluation of 86 as an ionophore-based antibacterial strategy against drug-resistant infections.³⁶⁷

Figure 29. Antibacterial agents targeting metabolic pathways with in vivo activity.

A series of halogenated phenazines was developed through a modular synthetic strategy designed to expand SAR exploration for antimicrobial purposes. By harnessing interactions with metal(II) cations, these compounds rapidly induce iron starvation in bacterial cells, effectively targeting DR strains and biofilms.³⁶⁸ Compound 87 (Figure 29) exhibited antibacterial activity against several Gram-positive bacteria, including MRSA, VRE, and M. tuberculosis. It was particularly effective against six clinical MRSA isolates, with MIC values between 0.06 and 0.12 μM, outperforming vancomycin. Compound 87 exhibited low cytotoxicity and minimal haemolytic activity at 200 μM. Further in vivo studies using dorsal wound infection models in mice evaluated 87’s efficacy against S. aureus and E. faecalis. The results indicated a reduction in bacterial counts, with a 0.8 log₁₀ CFU reduction in S. aureus UAMS-1 and a 1.7 log₁₀ CFU reduction in E. faecalis OG1RF compared to controls, following three consecutive days of treatment. These findings suggest that 87 may be a promising candidate for further development as a topical treatment for infections caused by Gram-positive bacteria.³⁶⁸

Antimicrobial peptides are limited by high production costs and vulnerability to enzymatic degradation,^369,370 which has driven interest in synthetic mimics that offer enhanced stability and practicality.^295,371 An interesting advancement in this field is the development of N-arylimidazolium antimicrobial peptide mimics.³⁷² Compound 88 (Figure 29) has shown antibacterial activity against various strains, including MRSA, S. aureus, and E. coli. with MIC values ranging from 0.2 to 1 µg/mL, alongside minimal haemolytic activity. These mimics also show moderate efficacy against ESKAPE pathogens and effectively target both planktonic bacteria and biofilms. Mechanistic studies revealed that rather than disrupting bacterial membranes, these biomimetic antimicrobial agents induce a pronounced increase in intracellular ROS, which disrupts key metabolic pathways and leads to bacterial cell death. Specifically, treatment with 88 triggers up to a 20-fold rise in ROS levels, causing oxidative stress that impairs cellular metabolism and viability. This ROS-mediated mode of action, supported by inhibition studies using ROS scavengers like N-acetyl-^L-cysteine, highlights a distinct antibacterial pathway that differs from traditional membrane-targeting antimicrobial peptides. PK studies indicated a favourable in vivo profile for 88, with a half-life of 0.8 hours and high plasma concentrations (12 mg/L) after intravenous administration. In a mouse model of MRSA lung infection, dosing at 4 mg/kg reduced bacterial burden and inflammation, outperforming vancomycin in improving survival and lung tissue recovery.³⁷²

Enterococci, naturally occurring bacteria in the human gastrointestinal tract, can pose a major clinical threat when they develop antibiotic resistance. The most concerning forms are VRE, which comprise two primary species: Enterococcus faecium and Enterococcus faecalis.³⁷³ These resistant strains pose a serious risk in healthcare settings, where treatment options are limited. A recent study investigated the development of acetazolamide derivatives that target bacterial carbonic anhydrases, a class of zinc-metalloenzymes essential for converting carbon dioxide into bicarbonate and protons.^374,375 Among these derivatives, the 1,3,4-thiadiazole compound 89 (Figure 29) had a MIC of 1 µg/mL against E. faecium HM-952. Specifically designed to reduce intestinal permeability, 89 enhanced its effectiveness in gastrointestinal tract decolonisation. Compound 89 achieved a 1.6−1.9 log₁₀ CFU reduction in VRE burden in intestinal tissues, surpassing the current standard, linezolid. The reduced permeability, confirmed through the Caco-2 assay, contributed to this specificity, minimising the risk of systemic side effects. Compound 89 supports further development for gastrointestinal decolonisation of VRE, given its favourable PK profile and efficacy in reducing VRE colonisation in the gastrointestinal tract.³⁷⁶

Lolamicin (90) (Figure 30) is a first-in-class antibiotic that selectively targets Gram-negative pathogens by disrupting the Lol lipoprotein trafficking pathway,³⁷⁷ essential for outer membrane biogenesis.²⁴⁶ Specifically, it inhibits the LolCDE ATP-binding cassette transporter complex, preventing the release and proper localisation of outer membrane lipoproteins. This blockade leads to envelope destabilisation, cell swelling, and lysis. Molecular modelling and structural studies noted that lolamicin binds to two conserved sites (BS1 and BS2) within LolCDE, overlapping with the lipoprotein binding pocket and forming key hydrophobic interactions essential for function. Resistance mapping identified mutations in LolC and LolE at residues that form salt bridges and stabilise the binding groove. These changes reduced susceptibility but imposed a fitness cost, suggesting limited resistance potential in clinical settings. Lolamicin showed bactericidal activity against over 130 MDR Gram-negative isolates, including colistin-resistant strains of E. coli, K. pneumoniae, and Enterobacter cloacae, while sparing commensal anaerobes and Gram-positive bacteria. Whole-cell accumulation assays correlated its activity with selective uptake in susceptible species but not in P. aeruginosa or A. baumannii, explaining its narrow-spectrum profile. In vivo, lolamicin demonstrated robust efficacy in murine models of colistin- and carbapenem-resistant infections. In a sepsis model, oral treatment completely rescued infected mice, improving survival from 13% (untreated) to 100%. In pneumonia models, bacterial burden was reduced by up to 3 log₁₀ CFU. PK evaluation confirmed favourable oral bioavailability (47%) and good tolerability at doses up to 200 mg/kg, with no observed toxicity. Lolamicin preserved gut microbiota composition and prevented Clostridioides difficile overgrowth, a critical advantage over broad-spectrum agents. Structural modifications introducing primary amines impaired binding to LolCDE, underlining the importance of hydrophobic contacts for target engagement.²⁴⁶

Figure 30. Metabolic pathway inhibitors with in vivo efficacy.

Compound 91 (Figure 30) has shown remarkable efficacy against MRSA, with its (S)-enantiomer emerging as the primary pharmacologically active form.³⁷⁸ Mechanistic studies revealed that 91 inhibits S. aureus enoyl-acyl carrier protein reductase (FabI), a key enzyme in fatty acid biosynthesis. Enzyme inhibition assays showed an IC₅₀ of 94.0 nM for the (S)-91, compared to 174.2 nM for the racemate and no measurable activity for the (R)-enantiomer up to 10 μM. In vivo, compound 91 exhibited superior efficacy in a murine MRSA skin infection model, reducing lesion size and bacterial burden in a dose-dependent manner. At 10 mg/kg, 91 achieved lesion reductions comparable to the clinical candidate AFN-1252 and surpassed linezolid. Histological analysis confirmed minimal tissue damage in treated animals. The compound also had excellent PK properties, with an oral T_max of 0.5 hours and bioavailability of 86%. Oral C_max values reached 1520.7 µg/L, indicating strong systemic exposure. Intravenous administration yielded a C_max of 3245.3 µg/L, confirming good plasma distribution. Molecular docking studies offered insights into its binding mode, showing that the 2-methyl-3-bromophenyl group engages in hydrophobic interactions with Val201 and Phe204, while the pyrrolidone moiety forms hydrogen bonds with Tyr157 and NADPH. Additionally, π–π stacking with Phe96 further enhances binding affinity.³⁷⁸

Fluorofolin (92) (Figure 30) is a selective DHFR inhibitor that exploits a unique metabolic vulnerability in P. aeruginosa.³⁷⁹ Unlike many bacteria, P. aeruginosa cannot scavenge exogenous thymine, rendering it highly dependent on de novo folate biosynthesis.³⁸⁰ Fluorofolin targets this pathway with exceptional potency and selectivity, showing an IC₅₀ of 0.2 nM against P. aeruginosa DHFR—approximately 100-fold stronger than its activity on E. coli DHFR and with negligible activity against the human enzyme. In vitro, fluorofolin inhibited 89% of clinical P. aeruginosa isolates, including MDR and extensively drug-resistant strains, with an MIC of 0.5 µg/mL. Co-culture assays confirmed selective inhibition of P. aeruginosa over E. coli in the presence of thymine, validating its unique dependency-based mechanism. In vivo, fluorofolin demonstrated efficacy in a murine thigh infection model, reducing bacterial burden by 3 log₁₀ CFU after 72 hours of once-daily dosing (10 mg/kg). PK profiling showed 65% oral bioavailability and no observed acute toxicity, supporting its suitability for systemic use. Resistance studies identified efflux pump overexpression as the primary mechanism of decreased susceptibility. These mutants displayed attenuated virulence, suggesting a low likelihood of clinically problematic resistance.³⁷⁹

5.5. Non-traditional antibacterials: anti-virulence agents

Traditional antibacterials act by directly inhibiting bacterial growth (bacteriostatic) or killing bacteria (bactericidal), and their use can select for resistance through mutation and horizontal gene transfer.³⁸¹ In contrast, non-traditional approaches such as anti-virulence strategies aim to disarm pathogens by targeting factors that drive disease, including proteases, haemolysins, and other enzymes involved in host damage.^382–384 These agents are designed to reduce pathogenicity rather than directly inhibit growth,^385,386 and, because virulence pathways are often more specific to pathogens, they may exert less pressure on commensal flora than broad-spectrum, growth-inhibitory antibiotics.³⁸⁷

LasB elastase, a key virulence factor in P. aeruginosa, is encoded by the highly conserved lasB gene and is secreted by the bacterial cell. This mature enzyme exhibits broad substrate specificity, allowing it to hydrolyse various host and pathogen proteins.³⁸⁸ Its cytotoxic effects include causing tissue damage, impairing tissue repair systems, disrupting host immune responses, and promoting inflammation through its proteolytic activity. Because LasB is secreted extracellularly, it does not present the permeability barriers that often limit the effectiveness of antibacterial agents.³⁸⁹ This unique positioning and broad activity support LasB as an attractive target for understanding and potentially mitigating the pathogenicity of P. aeruginosa.

Everett et al. identified a series of indanyl carboxylate compounds as potential antivirulence agents against P. aeruginosa through virtual screening.³⁹⁰ Chemical optimisation focused on modifications to the benzothiazole core, leading to enhanced potency while retaining selectivity. Among the optimised compounds, 93 and 94 (Figure 31) stood out, with K_i values of 35 and 32 nM against LasB elastase, respectively, and no cytotoxicity in HepG2, BSMC, and SAEC cells (IC₅₀ > 100 µM). X-ray crystallography stablished that compound 94 interacts with Zn²⁺ in the active site of LasB via its carboxylic acid group, with its indane moiety occupying a lipophilic pocket and its quaternary side chain extending into a solvent-exposed region. The study also assessed the impact of these LasB inhibitors on IL-1β activation, a pro-inflammatory cytokine involved in immune response, finding that both compounds reduced IL-1β activation. The PK profiles of both compounds were favourable, marked by substantial systemic exposure (610 and 517 μg·h/mL) and a half-life of approximately 6 hours. In a mouse model of P. aeruginosa lung infection, treatment with these inhibitors (10 and 30 mg/kg) resulted in a reduction in bacterial burden and a 4-fold decrease in IL-1β activation compared to controls. Statistically significant reductions in bacterial burden, albeit small, were observed with treatment 93 at both doses and with treatment 94 at the higher dose.³⁹⁰

Figure 31.

LasB inhibitors 93 and 94, shown alongside the co-crystal structure of 94 (in yellow) bound to P. aeruginosa LasB (in grey), highlighting key interactions and binding sites (PDB code: 7QH1).

Hirsch et al. explored a series of α-benzyl-N-aryl mercaptoacetamide compounds as LasB inhibitors, aiming to reduce virulence in P. aeruginosa.³⁹¹ In their earlier work, they identified compound 95 (Figure 32) as a LasB inhibitor using a fragment-merging strategy, achieving an IC₅₀ of 1.2 µM and demonstrating efficacy in a Galleria mellonella model.³⁹² Subsequent structure-based optimisation led to the development of compound 96 (Figure 32), which replaced the N-aryl ring with a benzothiazolyl ring. This modification improved potency, yielding an IC₅₀ of 0.7 µM. The benzothiazolyl ring introduced additional π-π stacking interactions with a histidine residue and cation-π interactions with an arginine residue, enhancing the inhibitory effect compared to compound 95. Compound 96 also demonstrated inhibitory activity against collagenase H, an enzyme secreted by Clostridium histolyticum that degrades collagen and causes tissue damage, with an IC₅₀ of 0.1 µM. It exhibited improved selectivity toward human matrix metalloproteases, with no observed cytotoxicity in HepG2, HEK293, and A549 cells. Moreover, in vivo testing in zebrafish embryos showed no toxicity at concentrations of ≤30 µM. The compound’s in vivo efficacy was further validated in a P. aeruginosa-infected G. mellonella model, where it improved survival rates by up to 50%.³⁹¹

Figure 32. Benzothiazole LasB inhibitor 96.

UDP-3-O-acyl-N-acetylglucosamine deacetylase (LpxC) is a zinc-dependent enzyme crucial for the biosynthesis of lipid A, a key component of the outer membrane in Gram-negative bacteria.³⁹³ Lipid A functions as an endotoxin, playing a key role in the pathogenicity of Gram-negative bacteria by triggering strong inflammatory responses.³⁹⁴ Therefore, inhibitors targeting LpxC offer a strategy to interfere with lipid A/LPS biosynthesis, with consequences for both bacterial viability and virulence-associated inflammatory responses.^395,396

Furuya et al. at Entasis Therapeutics developed a series of N-hydroxyformamide LpxC inhibitors through lead optimisation, with compound 97 (Figure 33) emerging as a promising candidate.³⁹⁷ It displayed MIC values of 0.13, 4, and 4 µg/mL against E. coli, K. pneumoniae, and P. aeruginosa, respectively, along with potent enzyme binding, indicated by IC₅₀ values of <1 nM for P. aeruginosa and <10 nM for E. coli. Crystallographic analysis showed that compound 97’s thiomethyl group fits into a hydrophobic pocket, while the α-substituent does not engage in polar interactions with the sugar pocket of P. aeruginosa LpxC. In a neutropenic mouse thigh infection model with E. coli ARC6599, compound 97, at 200 mg/kg twice daily, achieved a notable reduction of more than 3.2 log in CFU counts. Compound 97 showed promise, but its physicochemical profile was unfavourable and it inhibited acetylcholinesterase (IC₅₀ 3.5 μM). To mitigate these issues, the morpholine tail of compound 97 was replaced with spiroazetidine-thiatane-oxide, resulting in compound 98 (Figure 33), which showed improved PK properties and reduced AChE inhibition. However, even with these optimisations, compound 98 displayed hypotension-related cardiovascular toxicity, highlighting the need for further refinement in developing these LpxC inhibitors with a more favourable safety profile.³⁹⁷

Figure 33. LpxC inhibitor 97 and its optimised analogue 98.

The co-crystal structure of 97 (in yellow) bound to P. aeruginosa LpxC (in grey) is shown, highlighting key binding interactions (PDB code: 7K99).

A research team at Idorsia Pharmaceuticals identified compound 99 (Figure 34) as a LpxC inhibitor through a scaffold-hopping approach, followed by hit-to-lead optimisation.³⁹⁸ This compound features a methylsulfone hydroxamate head group with a pyrroloimidazolone core and a cyclopropyl ring. Compound 99 exhibited strong activity against a range of Gram-negative bacteria, with MIC values between 0.25 and 1 μg/mL, and favourable solubility at pH 7 (384 μg/mL). It also showed good binding affinity to LpxC in E. coli and P. aeruginosa, with dissociation constants of 9.9 K and 12.5 K, respectively. In a neutropenic murine thigh infection model with E. coli, compound 99 displayed dose-dependent efficacy in reducing bacterial load when administered subcutaneously at 50 or 150 mg/kg, though it did not achieve net static efficacy over a 24-hour period.³⁹⁸ To address these limitations, the team further developed compounds 100 and 101 (Figure 34), which emerged as improved leads.³⁹⁹ Compound 100 contains an azetidine ring, while compound 101 features an additional hydroxymethyl motif. Both compounds retained similar MIC potency against E. coli, P. aeruginosa, and K. pneumoniae (MIC < 1 μg/mL) and showed efficacy against a broader range of drug-resistant Gram-negative bacteria, including fluoroquinolone- and carbapenem-resistant strains. Compounds 100 and 101 showed enhanced solubility (>730 μg/mL) and improved systemic drug exposure (free AUC24h > 900 ng·h/mL). In the neutropenic murine thigh infection model with E. coli, treatment with compounds 100 and 101 achieved a reduction in CFU count below net stasis, indicating their potential as effective LpxC inhibitors with improved PK properties.³⁹⁹

Figure 34. LpxC inhibitor 99 and its optimised analogues 100 and 101.

The co-crystal structure of 99 (in yellow) bound to P. aeruginosa LpxC (in grey) is shown, highlighting key binding interactions (PDB code: 6I4A).

Sugiyama et al. at Taisho Pharmaceutical, discovered TP058653 (102) (Figure 35), a non-hydroxamate LpxC inhibitor, which showed antibacterial activity against drug-resistant Gram-negative bacteria, with MIC values ranging from 1 to 8 µg/mL, including strains resistant to carbapenems.⁴⁰⁰ Compound 102 also reduced LPS levels in E. coli while exhibiting selectivity against human matrix metalloproteinases, with IC₅₀ values that were over 700 times higher than those against E. coli LpxC. In vivo efficacy of 102 was tested in three different murine infection models: systemic infection, urinary tract infection, and neutropenic lung infection. In the systemic infection model with E. coli 5799 or K. pneumoniae 4102, compound 102 achieved a 100% survival rate, on par with meropenem, a standard antibiotic. In the urinary tract infection model and the neutropenic lung infection model, 102 reduced bacterial load. Further studies of 102 in neutropenic mice demonstrated a favourable PK profile, characterised by dose-dependent peak plasma concentrations.⁴⁰⁰

Figure 35. LpxC inhibitor 102.

Quorum sensing (QS) is pivotal in governing the pathogenicity of bacteria like P. aeruginosa, enabling them to synchronise behaviour and gene expression according to population density.⁴⁰¹ In P. aeruginosa, QS regulates biofilm formation and the production of virulence factors, directly impacting bacterial pathogenicity and resistance to conventional antibiotics.⁴⁰²

Lin et al. introduced a series of oxazolidinone compounds as QS inhibitors.⁴⁰³ Among the synthesised compounds, 103 (Figure 36), which features a bromine substituent on the terminal benzene ring, exhibited notable quorum sensing inhibition against Chromobacterium violaceum CV026, with an IC₅₀ of 3.6 µM. To evaluate the compound’s effect on QS-regulated virulence in P. aeruginosa PAO1, four key factors—pyocyanin, elastase, rhamnolipid, and protease—were selected. These represent major virulence mechanisms that drive tissue damage, immune evasion, and infection persistence, making them suitable markers for assessing QS disruption.^404,405 Compound 103 effectively inhibited the production of pyocyanin, rhamnolipid, and proteases, while also suppressing elastase activity, highlighting its potential to attenuate P. aeruginosa virulence through QS interference. The compound also reduced swarming motility and demonstrated a dose-dependent inhibition of biofilm formation, with a reduction of 17% to 42% in P. aeruginosa PAO1 biofilms. Compound 103 was found to increase the susceptibility of PAO1 biofilms to the antibiotic meropenem trihydrate (MEPM). When used in combination with 103, MEPM showed greater eradication of biofilm cells compared to MEPM alone. In vivo efficacy was assessed using a Caenorhabditis elegans N2 infection model with PAO1, where treatment with 103 led to a 16.0% increase in survival rate compared to the negative control group, indicating reduced pathogenicity of PAO1. When 103 was combined with MEPM, the survival rate in C. elegans increased by 18% compared to MEPM alone, demonstrating enhanced activity in combination.⁴⁰³

Figure 36. Quorum sensing inhibitors as antivirulence agents.

Another research group explored 4-fluorophenyl-5-methylene-2(5H)-furanone derivatives as QS inhibitors.⁴⁰⁶ Utilising the ligand-binding domain of P. aeruginosa’s LasR as a template, they employed receptor-based drug design to create and optimise compounds. Compound 104 (Figure 36) emerged as the most promising candidate, inhibiting QS-regulated virulence factors—pyocyanin, rhamnolipid, and protease—while exhibiting minimal cytotoxicity (<2.5% at 512 µg/mL). Like compound 103, it also caused a concentration-dependent reduction in the swarming motility of P. aeruginosa PAO1. Further investigation using GFP reporter strains revealed that compound 104 targets key QS genes including LasB, RhlA, and PqsA, thereby disrupting downstream gene expression. It also inhibited β-galactosidase activity, indicating that it impacts multiple QS systems (LasI/LasR, RhlI/RhlR, and PQS/MvfR) in E. coli. Although compound 104 did not inhibit the growth of Gram-negative bacteria (MIC > 512 µg/mL), it reduced biofilm formation by 50% compared to the control. This suggests it likely exerts less selective pressure for resistance, as it modulates virulence without directly compromising bacterial viability. Compound 104 enhanced the susceptibility of P. aeruginosa to antibiotics. When combined with ciprofloxacin or clarithromycin at a concentration of 256 µg/mL, the inhibition rate of clarithromycin increased from 32% to 82%, and ciprofloxacin from 26% to 77%. An in vivo evaluation using a mouse model infected with PAO1 showed that the combination of compound 104 and ciprofloxacin reduced CFUs compared to ciprofloxacin alone.⁴⁰⁶

Recently, Chen et al. conducted a study in which they explored the 3-hydroxypyridin-4(1H)-one scaffold as inhibitors of QS in P. aeruginosa, with the goal of reducing virulence and antibiotic resistance.⁴⁰⁷ By leveraging click chemistry, the researchers efficiently synthesised these derivatives through the coupling of a 3-hydroxypyridin-4(1H)-one derivative (derived from kojic acid) with 4-aminomethyl-1,2,3-triazole. Compound 105 (Figure 36) emerged as a highly selective inhibitor of the pqs system, a key transcriptional regulator of the QS network specific to Pseudomonas and Burkholderia spp. Compound 105 exhibited inhibitory effects on both the pqs system (IC₅₀ = 3.7 μM) and its associated virulence factor, pyocyanin (IC₅₀ = 2.7 μM), by selectively targeting the pqs-QS system, thereby suppressing pyocyanin production through the inhibition of pqsA promoter expression, which in turn hinders the biosynthesis of Pseudomonas quinolone signalling molecules. Additional experiments confirmed the broad-spectrum efficacy of compound 105 against various strains of bacteria, including clinical isolates, resulting in reductions in pyocyanin production, particularly in strain PA14. Furthermore, at a concentration of 50 μM, compound 105 substantially inhibited biofilm formation in PAO1, which was partly attributed to its disruption of siderophores pyoverdine and pyochelin production, thereby interfering with bacterial iron transport. Compound 105 demonstrated a synergistic enhancement of antibacterial effects when combined with antibiotics such as ciprofloxacin and tobramycin in both in vitro and in vivo C. elegans infection models. In a PAO1-induced C. elegans infection model, administration of compound 105 alone prolonged the lifespan of infected nematodes. When 105 was combined with ciprofloxacin at concentrations of 10 μM and 50 μM, respectively, the survival rates increased considerable, reaching 35% and 87%, respectively. Similarly, when 105 was combined with tobramycin at the same concentrations, the survival rates elevated to 46% and 82%, respectively. These findings support the potential of combining antibiotics with antivirulence agents as a therapeutic strategy.

Gram-positive bacteria utilise sortase enzymes to anchor surface proteins to their cell walls.⁴⁰⁸ Sortase A (SrtA) is a key cysteine transpeptidase that plays a vital role in the virulence of Gram-positive bacteria by facilitating the covalent attachment of surface proteins to peptidoglycan, thereby contributing to the infection process.⁴⁰⁹ As a result, sortases have emerged as attractive targets for drug development in pathogenic bacteria, given the essential role of the proteins that anchor on the microbial surface.⁴¹⁰ Zhang et al. have contributed to the field by developing covalent inhibitors that target Staphylococcus aureus SrtA.⁴¹¹ Through a structure-guided approach, the authors designed inhibitors featuring a barbital motif as the warhead, which facilitates binding to the hydrophilic pocket of SrtA, a diene for covalent interaction with the active site cysteine, and a lipophilic tail that enhances ligand affinity within the pocket. Following iterative optimisation guided by structural insights, compound 106 (Figure 37) emerged as the most promising analogue. While 106 alone did not affect Staphylococcal growth (MIC >128 µg/mL), it exhibited inhibitory activity against SrtA, with an IC₅₀ of 0.6 µM. Crystallographic studies found that compound 106 forms a covalent bond with C208 via Michael addition, while its barbital warhead engages in hydrogen bonding with H142 and H143. Toxicity assessments in G. mellonella larvae confirmed the safety profile of 106. In vivo efficacy evaluations in a G. mellonella model infected with S. aureus Newman or USA300 strains revealed promising anti-infective properties. Compound 106 treatment improved survival rates in both infection models, with a 60% survival rate observed in the Newman-infected group and a 40% survival rate in the USA300-infected group at 120 hours post-treatment.⁴¹¹

Figure 37. Small-molecule inhibitors of sortase A (106) and glucosyltransferase (107) enzymes involved in bacterial pathogenicity.

The co-crystal structure of 107 (in yellow) bound to Streptococcus mutans glucosyl transferase (in grey) is shown, highlighting key binding interactions (PDB code: 8FG8).

Streptococcus mutans is well-established as a contributor to dental caries, primarily due to the activity of its extracellular glucosyl transferase (Gtfs) in forming cariogenic biofilms.⁴¹² A recent study led by Velu et al. focused on refining hydrogel-encapsulated biofilm inhibitors derived from aurones through structural optimisation.⁴¹³ Building on previous research,⁴¹⁴ the team addressed the low solubility of these compounds. Through structural modifications, the researchers developed compound 107 (Figure 37), featuring an aurone ring with a 2,4,5-trihydroxyphenyl moiety. Compound 107 emerged as a biofilm inhibitor, with an IC₅₀ of 6.42 µM and notable improved solubility of 120.09 µg/mL. Moreover, 107 showed robust biofilm inhibition against S. mutans, achieving over 90% inhibition, while exhibiting no adverse effects on bacterial growth at a concentration of 50 µM. Microscopic analysis confirmed that concentrations of glucans and extracellular DNA decreased at 5 µM, with complete eradication at 50 µM. 107 displayed a high degree of selectivity towards pathogenic bacterial biofilms, while exerting minimal inhibitory effects on commensal ones, as evidenced by negligible inhibition of Streptococcus gordonii and no impact on Streptococcus sanguinis. Co-crystal structure analysis confirmed that 107 interacts with Gtfs through specific binding interactions, involving hydroxyl groups at positions 4 and 5 of the phenyl ring and key amino acid residues within the active site, facilitated by coordination with a conserved Ca²⁺ ion. The encapsulation of 107 in pH-sensitive poly(methacrylic acid) resulted in comparable antibiofilm activity, achieving 85-90% inhibition at 25 µM without impeding bacterial growth, regardless of whether the strain was pathogenic or commensal. In vivo experiments, including both topical application and hydrogel-encapsulated delivery of 107, revealed reductions in buccal and sulcal caries scores, with no observed cytotoxicity, as evidenced by the absence of weight loss throughout the study duration. These findings were corroborated in a gnotobiotic rat model infected with S. mutans, where 107 treatment effectively reduced caries scores while maintaining a favourable safety profile.⁴¹³

Lin and colleagues reported the development of bifunctional quinoline derivatives 108 and 109 (Figure 38) as dual-action inhibitors targeting P. aeruginosa biofilm formation and virulence.⁴¹⁵ These compounds act via two synergistic mechanisms: induction of cyclic-di-GMP G-quadruplex formation and inhibition of QS. Both compounds 108 and 109 achieved over 40% inhibition of QS at 10 μM. They also suppressed biofilm formation in P. aeruginosa PAO1, with inhibition levels (~40% at 10 μM). In addition to antibiofilm activity, 108 and 109 attenuated the production of key QS-regulated virulence factors. At 10 μM, 108 reduced elastase secretion by 59% and pyocyanin levels by approximately 40%, while 109 achieved 48% and ~40% reductions, respectively. Both compounds also inhibited hemolysin secretion and impaired swarming motility. Compounds 108 and 109 enhanced the antibacterial activity of tetracycline in vitro, lowering MICs by 4- to 8-fold, and improved host survival in a C. elegans infection model. In combination with tetracycline, 108 and 109 increased survivals to 95% and 86%, respectively. Both compounds had excellent safety profiles: they showed no DNA intercalation (at 50 μM), caused no hemolysis (up to 100 μM), and maintained high Vero cell viability at 12.5 μM.⁴¹⁵ Compound 109 differs from its analogue 108 only by a fluorine substituent at the 5-position of the indole ring.

Figure 38. Small-molecule inhibitors of bacterial quorum sensing.

The same group also reported compound 110 (Figure 38),⁴¹⁶ a hybrid agent combining a 3-hydroxypyridin-4(1H)-one moiety for iron chelation with a long-chain 4-aminoquinoline scaffold designed to inhibit QS in P. aeruginosa. Compound 110 selectively inhibited biofilm formation with an IC₅₀ of 5.8 μM, while exhibiting negligible effects on bacterial growth (MIC ≥512 μM for P. aeruginosa PAO1), supporting its antivirulence specificity. QS inhibition assays confirmed its capacity to reduce virulence factor production, with IC₅₀ values of 16.7 μM for elastase and 6.9 μM for pyocyanin. Mechanistic studies revealed that compound 110, despite its moderate iron-binding affinity (pFe³⁺ = 21), disrupted iron homeostasis by inducing a 16-fold increase in pyoverdine and a 2-fold increase in pyochelin expression at 50 μM, contributing to its antibiofilm activity. Synergistic activity was also observed when 110 was combined with sub-MIC levels of clinically relevant antibiotics. At 50 μM, 110 enhanced the bactericidal activity of tobramycin and polymyxin B, reducing bacterial viability by 60-fold and 100-fold, respectively. These effects translated to improved survival in C. elegans models of P. aeruginosa infection, achieving 81% (with tobramycin) and 91% (with polymyxin B) survival. Safety assessments indicated an excellent biosafety profile: 110 caused <1.5% hemolysis at 200 μM, displayed low cytotoxicity (CC₅₀ >800 μM in Vero cells), and was non-toxic to C. elegans even at 200 μM for 72 hours.⁴¹⁶

5.6. Compounds with miscellaneous or unknown targets

The discovery of new drugs, initially driven by chemical principles in its early stages, has undergone major transformations over the past century, fuelled by breakthroughs in genomics, bioinformatics, molecular biology, and biotechnology.⁴¹⁷ Historically, random screening and serendipity have been instrumental in deepening our understanding of the molecular mechanisms underlying pharmacological and toxicological effects. An estimated 7% to 18% of FDA-approved drugs do not have a clearly defined molecular target, indicating that regulatory approval does not necessarily depend on prior target identification.^417,418

In the context of drug discovery, it is essential to consider compounds with unknown targets, as they often serve as valuable starting points for subsequent optimisation steps. The phenotypic strategy, which does not rely on prior knowledge of the molecular target, has gained prominence in recent years due to the complexities of understanding various diseases at the molecular level.^419,420 By focusing on modulation of a biomarker or disease phenotype rather than targeting a predefined molecular target, this approach offers several advantages. These include the identification of compounds that can interact with multiple targets (polypharmacology) and the discovery of novel chemical entities with unique mechanisms of action, ultimately expanding the ‘druggable’ target space. Translating these findings still depends on continued progress in target deconvolution and hit validation, which remain technically challenging even with recent advances in chemical biology tools for target identification.^421,422 In the pursuit of novel antimicrobial drugs, phenotypic assays are often used to uncover bioactive compounds without prior knowledge of their precise molecular targets. This section examines compounds whose targets remain unknown or only partially elucidated, as well as agents with diverse mechanisms that do not fit within the categories discussed in previous sections.

Building upon prior findings of antimicrobial effects associated with scaffolds containing a pyran subunit against P. aeruginosa biofilm, Li et al. reported a small series of pyran derivatives. Evaluation against S. aureus and P. aeruginosa strains and biofilms reported their limited efficacy, with MIC values exceeding 200 µg/mL for both Gram-positive and Gram-negative bacteria. In vivo experiments confirmed that compound 111 (Figure 39) exhibited a partial reduction, though not complete elimination, of pre-formed biofilms in two infection models: S. aureus tissue cage biofilm infection and P. aeruginosa-mediated urinary tract catheter biofilm infection. Due to the limited number of compounds, a comprehensive SAR analysis was not provided; nevertheless, the authors suggested that the presence of an electron-withdrawing group in the aromatic ring linked to the chromene subunit enhances the activity of the most promising compounds. The authors proposed that the observed antibiofilm activity may be related to the modulation of the quorum sensing system.⁴²³

Figure 39. Chemical structures of miscellaneous antibacterials exhibiting in vivo activity.

AI methods, particularly those grounded in machine learning, and more specifically, deep learning techniques, including artificial neural networks, restricted Boltzmann machine networks, and convolutional networks, have found extensive applications as screening methods in the discovery of antimicrobial compounds.^107,424,425 Integrating AI methods with image-based phenotypic drug discovery represents a powerful tool for identifying novel bioactive agents. Several publications have provided detailed insights into the advantages of utilising this technology.^426–428

Stokes et al. employed a deep learning model for predicting antimicrobial compounds that act through the inhibition of E. coli growth.¹⁰⁶ From a selected set of 23 empirically tested predictions, originating from a pool of over 107 million molecules curated from the ZINC15 database, the authors identified eight promising lead compounds. Among them, halicin (112) (Figure 39), which was first mentioned in Section 3, is a c-Jun N-terminal kinase inhibitor that exhibited bactericidal effects with an MIC value of 2 □ g/mL against E. coli BW25113 (CGSC identifier 7636). Even in the presence of E. coli strains carrying plasmid-borne antibiotic-resistance genes (resistance to polymyxins, chloramphenicol, β-lactams, aminoglycosides, and fluoroquinolones), this compound maintained its bactericidal effect. In addition to its effectiveness against E. coli, halicin exhibited activity against various other species, including M. tuberculosis, clinical isolates of CRE, and A. baumannii. While the AI model was unable to predict the mode of action, attempts were made to isolate halicin-resistant mutants, albeit without success. Therefore, the authors performed whole-transcriptome sequencing on E. coli treated with halicin to identify the patterns of gene alterations. They noted a decrease in the expression of genes associated with cell motility and iron homeostasis. Halicin disrupted the proton motive force, affecting the transmembrane ΔpH potential. In vivo experiments using a murine wound model of A. baumannii infection showed that halicin treatment reduced bacterial load by up to 5 log₁₀ CFU/g in the skin lesion of BALB/c mice.¹⁰⁶ This serves as a compelling example of combining AI with phenotypic screening and subsequent target deconvolution to identify compounds with initially unknown targets. While the authors did not ascribe mutagenic or carcinogenic properties to halicin, the presence of the nitro-heterocycle may warrant careful future toxicity assessments. It is noteworthy that halicin bears structural similarity to megazol, an antibacterial and antiparasitic compound previously associated with genotoxic effects.^429,430

Building on the seminal work described above, additional antibacterial compounds were identified, including ZINC000100032716.¹⁰⁶ This compound was later developed by other research groups and renamed STK-35 (113), alongside its simplified analogue SK-66 (114) (Figure 39).⁴³¹ Both compounds exhibited antibacterial activity against Gram-negative bacteria, including P. aeruginosa, E. coli, A. baumannii, and K. pneumoniae, with MIC values ranging from 0.0625 to 8 μg/mL. Time-kill experiments, using 2–4 times the MIC of 113 and 114, resulted in a reduction of more than 3 log₁₀ CFU of the tested strains compared to the initial inoculum, confirming their bactericidal effects. These compounds also had synergistic effects with some available antimicrobials. The combination of chlorhexidine with 113 exhibited synergistic effects against Gram-negative species. The authors proposed a hypothesis suggesting that this observed impact might result from damage to the bacterial membrane caused by chlorhexidine, facilitating the entry of 113 into the inner cell and thereby facilitating entry of 113 into the bacterial cell and increasing its access to the target. Compounds 113 and 114 exhibited bactericidal activity in models involving persister cells of Gram-negative strains. They also reduced the biofilm formation of P. aeruginosa at concentrations ranging from 1 to 16 μg/mL, mitigated haemolysis-induced effects, and demonstrated reduced cytotoxicity in human bronchial epithelial and A549 cell lines. A murine peritonitis infection model of peritonitis showed that treatments with either 113 or 114 protected the animals, resulting in survival rates of 87% and 100%, respectively.⁴³¹

A series of diphenylurea derivatives incorporating an aminoguanidine moiety was synthesised and evaluated, with molecular optimisation focused on identifying new bioactive agents effective against MRSA strains. Initial screening against MRSA NRS384 (MRSA USA300) identified five compounds with MIC values of 2 μg/mL, prompting their selection for further investigation. In subsequent assays against methicillin-sensitive S. aureus (ATCC 6538) and various clinical isolates of methicillin- and vancomycin-resistant S. aureus, compound 115 (Figure 40) exhibited the highest potency in the series, with MIC values ranging from 2 μg/mL to 4 μg/mL, except in vancomycin-resistant S. aureus. At concentrations up to 64 μg/mL, compound 115 had no cytotoxic effects on human keratinocyte (HaCaT) and human colorectal adenocarcinoma (Caco-2) cells. Haemolytic effects induced by compound 115 were dose-response dependent, with 50% haemolysis at 64 μg/mL. Multi-step resistance studies suggested a low likelihood of MRSA developing resistance to compound 115. In a murine model of skin infection, compound 115 exhibited in vivo efficacy, showing antibacterial effects in infected wounds comparable to commercially available fusidic acid ointment.⁴³²

Figure 40. Diphenylurea-based antibacterials demonstrating in vivo activity.

The same research group, employing a similar approach, disclosed the anti-MRSA effects of a series of diphenylurea derivatives featuring an aminoguanidine moiety, substituting the subunit 4-methylpent-1-en-1-yl with alicyclic, aromatic, and open-chain acetylenic subunits.⁴³² The most potent compound, 116 (Figure 40), exhibited anti-staphylococcal effects with MIC values ranging from 0.5 to 2 μg/mL against MSSA, MRSA, and VRSA strains. Compound 116 displayed no cytotoxic effects up to a concentration of 64 μg/mL in HaCaT cells. It showed minimal haemolytic activity, and a low likelihood of developing resistance compared to rifampicin. In a murine model of skin infection, compound 116 reduced the MRSA burden in the skin by up to 0.5 log₁₀ CFU and effectively decreased bacterial burden in the spleens by up to 1 log₁₀ CFU.⁴³³

In the pursuit of drug repurposing, the investigation of etrasimod (Figure 41), a sphingosine-1-phosphate receptor modulator currently undergoing phase 3 clinical trials for ulcerative colitis, revealed efficacy against Gram-positive bacteria, with MIC values ranging from 5 to 10 μM.⁴³⁴ Given S. aureus’ propensity to develop resistance within a 10-day treatment period, a molecular optimisation approach utilised etrasimod as a prototype. Approximately 32 compounds were synthesised and assessed to elucidate the SAR. Initial screening using S. aureus ATCC 25923 identified MIC values ranging from 2 to 19 μg/mL. The effects on S. aureus ATCC 25923 biofilm viability and biomass were evaluated at a single concentration (100 μM), with post-treatment viability ranging from 40% to 95%. None of the examined compounds exhibited activity against Gram-negative bacteria (A. baumannii and P. aeruginosa). Compound 117 (Figure 41), selected for further assays, displayed one of the best activity profiles in the series. For this compound, the MIC values against MRSA and penicillin-resistant S. aureus P2 were 1 μg/mL. Time-kill experiments validated its bactericidal capability, reducing up to 3 log₁₀ CFU at a concentration of 8 x MIC after 1 hour. Resistance development studies indicated a low probability of S. aureus (ATCC 25923) and MRSA developing resistance. In the checkerboard assay, a synergistic effect with gentamicin against E. faecalis, S. aureus, and MRSA was observed. Cytotoxicity evaluations using lung carcinoma epithelial cells (A549) and the human leukaemia monocytic cell line (THP-1) found CC₅₀ > 100 μM and 57.7 μM, respectively. Haemolytic activity was not observed at concentrations up to 62.5 μM. In vivo experiments using the Caenorhabditis elegans infection model indicated that compound 117 reduced the S. aureus ATCC 25923 and MRSA ATCC 43300 burden by 85% and 88%, respectively.⁴³⁵

Figure 41. Etrasimod and linezolid analogues (117 and 118) with in vivo activity against MRSA.

Oxazolidinones are important for treating Gram-positive infections, with linezolid as the first widely used agent in the class.⁴³⁶ Limitations such as myelosuppression, drug interactions, and reduced tolerability in some patient groups highlight the need for new derivatives that better balance efficacy and safety. Addressing this need, Bhawsar et al. synthesised and evaluated a series of oxazolidinones. Initial screening involved sensitive and DR strains of S. aureus, S. pneumoniae, and H. influenza. Compound 118 (Figure 41), identified as the most active against Gram-positive bacteria in this series, exhibited MIC values comparable to linezolid and was selected for in vivo assays. In a mouse sepsis model, compound 118 showed an ED₅₀ value of 2.5 mg/kg, while linezolid ranged from 2.5 to 5 mg/kg. PK studies in Wistar rats and Beagle dogs observed that, for rats at a dose of 15 mg/kg, 118 exhibited a C_max of 2.37 μg/mL and an AUC of 10.85 μg·h/kg, whereas linezolid showed a better PK profile with a C_max of 8.70 μg/mL and an AUC of 27.30 μg·h/kg. In dogs, 118 demonstrated improved PK parameters with a C_max of 9.66 μg/mL and an AUC of 135.25 μg·h/kg, while linezolid had a C_max of 17.82 μg/mL and an AUC of 78.24 μg·h/kg.⁴³⁷

Bacterial sexually transmitted infections clinically implicate three primary species: Chlamydia trachomatis, Treponema pallidum, and Neisseria gonorrhoeae. C. trachomatis, primarily impacting human genital, ocular, and respiratory tract epithelial cells, often remains asymptomatic initially. Prolonged infection poses major risks, including trachoma, potential harm to female reproductive health, and conditions like chronic pelvic pain and infertility.^438,439 Existing antimicrobial therapy faces concerns regarding drug resistance and adverse effects, prompting the search for new active agents.⁴³⁸ Previously, Ouellette et al. identified sulfonylpyridine derivatives as active against C. trachomatis, with MIC values of 50 μg/mL. Antimicrobial effects were associated with the inhibition of the Clp protease system and protein turnover.⁴⁴⁰ Building on this, they synthesised new analogues to explore the SAR.⁴⁴¹ Compound 119 (Figure 42), exhibited the best antimicrobial profile, displaying bactericidal activity at 50 μg/mL. Compound 119 altered the morphology and size of C. trachomatis in a dose-dependent manner, with an IC₅₀ value of 1.76 μg/mL. Selectivity for Chlamydia was confirmed, with no relevant activity against E. coli K12 and S. aureus JE 2 (> 256 μg/mL). The checkerboard assay suggested a synergistic effect with azithromycin. In vivo experiments using a C. muridarum mouse model indicated that a 5-day treatment with compound 119 at 100 mg/kg resulted in reduced chlamydial shedding, supporting the efficacy of compound 119.⁴⁴¹

Figure 42. Compounds 119, iboxamycin (120), and cresomycin (121) as antibacterial agents demonstrating in vivo activity.

Discovered in 1963, lincomycin was among the earliest lincosamide antibiotics used clinically, specifically against anaerobic Gram-positive bacteria. Its successor, clindamycin, introduced in 1970, exhibited an improved spectrum of action and favourable PK properties.⁴⁴² Lincosamides target the bacterial ribosome, and extensive research, including over 500 derivatives by Myers research group, led to the discovery of iboxamycin (120) (Figure 42).⁴⁴³ Compared to clindamycin, iboxamycin had broader efficacy against various resistant strains, including those expressing erm-, cfr-, and isaA resistance genes. Iboxamycin exhibited strong activity against Gram-negative bacteria, defying conventional physicochemical predictors. It exhibited MIC values below 16 μg/mL against carbapenem-resistant strains and below 8 μg/mL against strains resistant to cephalosporin, fluoroquinolone, tetracycline, and aminoglycosides. In vivo studies showed efficacy in murine neutropenic thigh infection models, reducing bacterial burden significantly within 12 hours. In a mouse model of systemic infection (S. pyogenes ATCC 19606), iboxamycin treatment led to the rescue of infected mice. PK studies indicated a 24% bioavailability for orally administered iboxamycin without inducing adverse effects. Spontaneous microbial resistance allowed the identification of mutants, indicating the bacterial ribosome as iboxamycin’s target. X-ray crystallography investigations unveiled its binding mode, providing structural insights into its efficacy against resistant strains, as indicated in bacterial ribosomes containing the erm-resistance gene.⁴⁴³

The strategic constraint of rotatable bonds in iboxamycin (120) has led to the development of cresomycin (121) (Figure 42), a bridged macrobicyclic antibiotic that is pre-configured to bind tightly to the ribosome.⁴⁴⁴ This synthetic antibiotic effectively overcomes the antimicrobial resistance mechanisms that undermine the efficacy of current ribosome-binding antibiotics, such as iboxamycin and clindamycin. Unlike iboxamycin, which assumes varying conformations prior to binding, cresomycin seamlessly locks into the optimal binding conformation, resulting in a strong fit that outcompetes obstructive methyl groups that typically hinder the binding site. This approach defies traditional views on antibiotic resistance, showing that high-affinity binding can effectively counter resistance even without directly targeting the specific regions modified by resistance mechanisms. Cresomycin has shown efficacy against a range of high-priority MDR ESKAPE pathogens, outperforming iboxamycin in inhibiting the growth of MDR Gram-positive bacteria, including staphylococci, streptococci, enterococci, and C. difficile, as well as MDR Gram-negative bacteria, such as E. coli, K. pneumoniae, A. baumannii, and N. gonorrhoeae. In mouse sepsis models, cresomycin yielded a high level of efficacy, rescuing mice from systemic infections caused by cfr-expressing S. aureus, a highly resistant strain. Following a single day of subcutaneous treatment at 25 mg/kg four times a day, all mice survived for seven days, whereas most untreated mice died within two days. Cresomycin demonstrated efficacy in reducing bacterial burden in neutropenic thigh-infection models. Intraperitoneal administration of cresomycin reduced bacterial loads by 4.6, 2.2, 2.6, and 2.7 log₁₀ CFU in cfr-expressing S. aureus, ermA-expressing S. aureus, carbapenem-resistant E. coli, and carbapenem-resistant P. aeruginosa, respectively, compared to untreated controls.⁴⁴⁴ Recent optimisation efforts led to a fluorinated analogue exhibiting improved potency and metabolic stability over iboxamycin and cresomycin. X-ray crystallography reported that the fluorine atom establishes an additional van der Waals contact with a nucleobase in the active site, which is hypothesised to strengthen target affinity.⁴⁴⁵

Afabicin (4), currently in phase II clinical trials for staphylococcal infections, faces limitations in its effect on Gram-negative bacteria. Guided by the “eNTRy rules”,^446,447 Hergenrother et al. utilised a thorough SAR analysis, leading to the discovery of fabimycin (122) (Figure 43).⁴⁴⁸ The eNTRy rules centre on improving accumulation in Gram-negative E. coli. These rules emphasise the importance of an ionisable amine (preferably primary), low globularity (≤0.25), and a restricted number of rotatable bonds (≤5) for effective accumulation.⁴⁴⁷ Initial assessments against 54 strains of MDR A. baumannii, E. coli, and K. pneumoniae showed low MIC values (1−8 μg/mL). Confirmation of FabI as the primary target was supported by spontaneous resistant mutants, and crystal structures revealed favourable interactions with the active compound’s key binding sites. In vivo experiments with neutropenic female BALB/c mice infected with drug-resistant A. baumannii exhibited promising results, with doses up to 30 mg/kg showing a dose-response curve, and a 75 mg/kg dose achieving a 2 log₁₀ reduction in CFU/thigh. In a separate experiment with a carbapenem-resistant strain of E. coli in a murine urinary tract infection model, fabimycin yielded reductions in bacterial load in various organs, including the spleen (3 log₁₀), bladder (2.8 log₁₀), liver (2.9 log₁₀), and kidney (1.9 log₁₀).⁴⁴⁸

Figure 43. Fabimycin (122) shown alongside the co-crystal structure of 122 (in yellow) bound to E. coli FabI (in grey), highlighting key interactions and binding sites (PDB code: 7UMW).

Pyrazole scaffolds are widely used in medicinal chemistry. In 2022, at least three of the 17 small molecules approved by the FDA contained a pyrazole motif.⁴⁴⁹ Inspired by this observation and previous research on fluorophenyl-substituted pyrazole-derived hydrazones,^450,451 Alam et al. embarked on a new investigation.⁴⁵² They synthesised a new series of 71 pyrazole derivatives, assessing their antimicrobial activity against Gram-positive bacteria. Three lead compounds, featuring a trifluoromethyl group and a halogen substituent, exhibited MIC values ranging from 0.5 to 8 μg/mL against susceptible and resistant strains of S. aureus, S. epidermidis, and E. faecalis. Cytotoxicity evaluations at 10 μM showed no cytotoxic effects against HEK-293 human cells, and at 64 μg/mL, haemolysis remained below 10%. In time kill assays, these compounds demonstrated bactericidal activity, reducing bacterial viability by up to 4 log₁₀ after 4 hours. Similar findings were noted in the persister kill assay for MRSA persisters. Using a specialised high-throughput biofilm assay system (Calgary Biofilm Device), the minimum biofilm eradication concentration (MBEC) was determined to be 4 μg/mL, achieving a 2–3 log₁₀ reduction in both S. aureus and E. faecalis. Real-time monitoring of biofilms and electron microscopy confirmed the compounds’ effectiveness in inhibiting biofilm formation. In vivo studies using a Caenorhabditis elegans model of MRSA infection, the compounds showed minimal toxicity up to 20 μg/mL, maintaining 80% viability after 4 days. Treatment with pyrazole 123 (Figure 44) at 4 μg/mL preserved nematode survival at up to 90%, in contrast to the untreated group, in which all animals succumbed to the infection. CRISPR-based genome-wide screening revealed alterations in essential genes, implicating the fatty acid biosynthesis pathway, particularly involving FabG, FabK, and FabZ, as potential targets for the efficacy of the selected compound.⁴⁵²

Figure 44. Compounds targeting fatty acid biosynthesis (123), penicillin-binding proteins (124), and efflux pumps (125).

Disrupting the sporulation process of C. difficile has been proposed as a crucial strategy, as spores produced are dormant and resistant to environmental changes. The sporulation-specific class B penicillin-binding protein, CdSpoVD, is a promising target, and cephamycins have shown effectiveness in interrupting spore formation. Their tendency to induce dysbiosis and contribute to relapse necessitates the search for novel compounds.⁴⁵³ Expanding upon this, Pyne et al. synthesised and evaluated a series of cefotetan analogues as inhibitors of sporulation in C. difficile.⁴⁵⁴ Compound 124 (Figure 44) exhibited a MIC value of 5 μg/mL against C. difficile, with values ≥ 32 μg/mL for other tested microorganisms. In vitro binding assays with CdSpoVD showed an IC₅₀ of 15 nM, and antisporulation assays indicated a 10,000-fold reduction in spore count. Cytotoxicity and haemolysis were observed above 32 μg/mL. Homology studies were conducted to model CdSpoVD, and covalent docking suggested the binding mode. Mass spectrometry analysis confirmed the covalent bond for compound 124. In a murine C. difficile infection model, compound 124 had efficacy comparable to cefotetan. Based on its selective activity against C. difficile, the authors concluded that, although its effects resemble those of cefotetan, the potential to avoid disruption of normal gut flora and reduce relapse risk warrants further investigation.⁴⁵⁴

One widely recognised mechanism of resistance in Gram-negative bacteria involves the overexpression of efflux pumps, such as those belonging to the Resistance-Nodulation-Cell Division (RND) superfamily, like AcrAB-TolC.^455,456 To explore the potential of inhibiting pump efflux to restore bacterial sensitivity to antibiotics, Flipo et al. synthesised a series of pyridylpiperazine derivatives and evaluated in vitro and in vivo experiments involving K. pneumoniae.⁴⁵⁷ Compound 125 (Figure 44) enhanced antibiotic activity with an EC₉₀ of 3.3 μM and exhibited good microsomal stability, with a half-life greater than 40 min and intrinsic clearance of 6.0 μL/min/mg proteins. Compound 125 inhibited AcrB, a trimeric multidrug exporter located in the inner membrane, confirmed by the accumulation of berberine in K. pneumoniae using a fluorescent probe. Cryo-EM identified the binding site in the transmembrane region of K. pneumoniae AcrB. ADME studies showed high water solubility (≥1 mM in PBS), low PPB, and modest hERG potassium channel inhibition (8% at 30 µM). PK studies suggested good bioavailability after oral administration, with C_max and T_max values of 2 μM and 4 hours, respectively. The AUC_(0-24h) supported exposures consistent with sustained efficacy, and 125 also showed potential for lung accumulation. In a hypervirulent K. pneumoniae (ATCC 43816) infection model, a proof-of-concept indicated that while the compound alone did not reduce the lung bacterial load in mice, in combination with levofloxacin, a synergistic effect resulted in a 5-fold enhancement.⁴⁵⁷ These findings support the strategy of combining efflux pump inhibitors with antibiotics as an alternative approach to restore antibacterial activity.

Compound 126 (Figure 45), a peptidomimetic based on a dihydrothiazolo ring-fused 2-pyridone scaffold,⁴⁵⁸ shows bactericidal activity against MDR Gram-positive pathogens, including S. pyogenes, MRSA, VRE, and S. pneumoniae.⁴⁵⁹ In vitro, 126 had MICs in the low micromolar range, effectively eradicating both planktonic and biofilm-associated bacterial cells. Time-kill assays confirmed rapid bactericidal activity, with a >3 log₁₀ CFU reduction within 4 hours of exposure. In a murine S. pyogenes skin infection model, 5 mg/kg of 126 reduced bacterial load by ~1.3 log₁₀ CFU by day 3, lowered inflammatory cytokines (TNF-α, IL-6), and accelerated wound healing. Additionally, it downregulates key virulence factors, such as M protein and SpeB, thereby reducing pathogenicity.⁴⁵⁹

Figure 45. In vivo-active antibacterial agents with diverse mechanisms of action.

Compound 127 (Figure 45) is a next-generation oxazolidinone optimised for enhanced potency and PK properties.⁴⁶⁰ It targets bacterial protein synthesis and exhibits broad-spectrum activity against Gram-positive pathogens, including resistant strains such as MRSA, MRSE, VISA, and VRE, with MICs ranging from 0.06 to 8 μg/mL. Reduced potency was noted against linezolid-resistant strains.

In a murine MRSA systemic infection model, 127 achieved a 4 log₁₀ CFU reduction in bacterial load and 100% survival at doses of 25 and 50 mg/kg. The compound shows a prolonged half-life (11.9 hours) and high oral bioavailability (F% = 122). Cytotoxicity assays indicate a favourable safety profile (IC₅₀ > 64 μg/mL), alongside immunomodulatory effects that mitigate cytokine dysregulation and tissue damage. Additionally, 127 disrupts MRSA biofilms and induces membrane damage, highlighting its potential against resistant infections.⁴⁶⁰

Compound 128 (Figure 46) is a boronate-based inhibitor that restores the efficacy of β-lactam antibiotics against MRSA by targeting the BlaR sensor domain. It forms a covalent bond with the catalytic Ser389 residue of BlaR, blocking its activation and downstream signal transduction. This prevents the proteolytic degradation of the BlaI repressor, thereby maintaining repression of resistance genes, including those encoding β-lactamase and the penicillin-binding protein PBP2a. When tested across 40 MSSA and MRSA strains, including virulent USA300 clones, compound 128 potentiated the activity of oxacillin and meropenem. MICs were reduced by up to 4,096-fold for oxacillin and 2,048-fold for meropenem. Checkerboard assays indicated strong synergy (FIC indices ≤ 0.25), and time-kill experiments confirmed bactericidal activity, with combinations of compound 128 and β-lactams leading to >6 log₁₀ CFU reductions. PK studies in mice showed that compound 128 achieves therapeutic concentrations at the site of infection. Following a single subcutaneous dose (25 mg/kg), peak levels reached 28 mg/kg in thigh tissue and 8.5 mg/L in plasma at 4 hours, with sustained exposure up to 8 hours, supporting a thrice-daily dosing schedule. In a neutropenic thigh infection model, co-administration of 128 with either oxacillin or meropenem resulted in a 1.7–1.8 log₁₀ CFU reduction in bacterial burden compared to vehicle, whereas monotherapy had no significant effect. Compound 128 was also well tolerated, showing no cytotoxicity in HepG2 cells at concentrations up to 64 µg/mL.⁴⁶¹

Figure 46. Compounds 128 and 129. The co-crystal structure of 128 (in yellow) bound to S. aureus BlaR1 (in grey) is shown, highlighting key binding interactions (PDB code: 8C0P).

Compound 129 (Figure 46), a 2-aminothiazole derivative, enhances the activity of polymyxin E against both Gram-negative and Gram-positive bacteria.⁴⁶² Although it shows minimal intrinsic activity against A. baumannii, P. aeruginosa, and K. pneumoniae (MIC > 128 μg/mL), 129 strongly synergises with polymyxin E, lowering MICs by 4- to 32-fold (FICI ≤ 0.5). It also exhibits direct antibacterial activity against MRSA, VRE, and B. subtilis (MIC = 1 μg/mL). Time-kill assays confirm durable suppression of A. baumannii when combined with polymyxin E. In vivo, 129 enhances polymyxin E efficacy in a murine infection model, achieving 100% survival at 64 mg/kg plus 0.5 mg/kg polymyxin E. Mechanistic studies reveal that 129 increases bacterial outer membrane permeability, supported by NPN uptake and TEM analyses.⁴⁶²

This section highlighted compounds with a variety of mechanisms of action that fall outside the earlier categories, alongside compounds with unknown molecular targets. For the latter, this underscores the critical role of phenotypic assays in antibacterial discovery as indispensable tools for identifying novel bioactive compounds with unique or yet-to-be-elucidated mechanisms.

6. Discussion and Perspective

The period from 1940 to 1960 is often described as the golden era of antibiotic discovery, when many of today’s antibiotics emerged from whole-cell screening. Subsequent efforts shifted toward medicinal chemistry optimisation of existing scaffolds and later to target-based discovery, but translating target engagement into whole-cell antibacterial activity has proved difficult.⁴⁶³ The current arsenal still relies heavily on natural products, yet finding new agents from natural sources has become slower and lower-yield over time.⁴⁶⁴ This contributed to a move in the late 1990s toward high-throughput screening of synthetic libraries for target-specific inhibitors, an approach that ultimately produced few antibacterial candidates suitable for development.⁴⁶⁵

A major early-stage challenge in antibacterial discovery is finding new compounds and targets that avoid cross-resistance, with increasing emphasis on Gram-negative pathogens.⁴⁶⁶ Efforts to define practical design criteria have highlighted “bacterial bioavailability” as a useful framework, capturing uptake, intracellular distribution, metabolism, and efflux.⁴⁶⁷ This is particularly important for Gram-negative bacteria, where the outer membrane and active efflux substantially limit compound accumulation compared with Gram-positive organisms.^468–470

General physicochemical guidelines, such as the eNTRy rules, have been proposed to improve compound accumulation in Gram-negative bacteria,⁴⁴⁷ but major barriers persist, particularly in poorly permeable pathogens such as P. aeruginosa.⁴⁴⁶ Achieving sufficient cytoplasmic exposure remains essential for intracellular targets,⁴⁷¹ which makes better methods to measure and predict cytoplasmic accumulation and permeability a priority for rational design.^472–474 In parallel, expanding beyond traditional “rule of five” chemical space can open new opportunities, as illustrated by zosurabalpin (11), and may be especially relevant when mining structurally diverse natural product collections for new scaffolds.

Identifying a validated lead compound in the preclinical stage is an important milestone in antibiotic development.^34,35 Many compounds show in vitro activity, but only a small fraction progress to in vivo studies where efficacy and safety can be assessed.^37,38 Animal models remain essential, yet modelling human infection, particularly for ESKAPE pathogens, is challenging because of differences in immunity, disease progression, and virulence.^475,476 Transgenic, humanised, and immunocompromised models may provide more predictive platforms and help improve translation.^477–479

The clinical antibacterial pipeline remains limited, with relatively few candidates representing new chemical classes or novel mechanisms of action (Table 1). This underlines the importance of agents with differentiated modes of action, which may be less affected by existing resistance. Preclinical efficacy in animal models is an important step, but clinical translation still requires extensive trials and careful PK assessment to identify liabilities early, even when in vitro activity is strong.

Evaluating the PK properties is essential for progressing antibacterial leads, as it determines whether in vitro potency can translate into effective exposure at the site of infection in vivo.⁴⁸⁰ Many compounds with strong in vitro activity fail at the preclinical stage because exposures are insufficient or inconsistent, so linking in vitro results with in vivo efficacy and PK early helps guide more efficient optimisation.^481–483 Clinical development failures due to poor drug-like properties have declined, reflecting the value of assessing developability early.^484–486 Despite many antibacterial hits reported each year, only a small number show efficacy in preclinical in vivo models, which remain a key filter for candidates with realistic clinical potential.⁴⁸² A thorough assessment of drug metabolism and PK also helps identify liabilities such as low bioavailability, drug–drug interactions, and reactive or toxic metabolites that can prevent adequate delivery to the target.^487,488

In parallel, improvements in antibacterial development depend on more predictive preclinical validation. Emerging approaches, including humanised model of drug metabolism, organ-on-chip platforms,^489,490 ex vivo organ culture systems,⁴⁹¹ and 3D bioprinting of tissues,⁴⁹² can complement traditional animal models and strengthen decisions on which candidates are most likely to translate into clinical benefit.

Across the preclinical candidates discussed in this review, contributions are concentrated in a small number of regions. In particular, compounds from US and Chinese research institutions represent around 25% and 39% of the total, respectively. European contributions are relatively modest, at 17%. It is essential to acknowledge that our review is not exhaustive and, therefore, these numbers should not be regarded as definitive references or an accurate reflection of the total research output in antibacterial discovery. Nonetheless, they unequivocally highlight the dominant position of the US and China in scientific output in this area.⁴⁹³

Our analysis reveals that most compounds are being developed to combat Gram-negative bacteria (~58% of studies), pointing to a recent shift in focus from Gram-positive to Gram-negative pathogens.¹²⁵ In terms of the infection models employed, there is alignment with the WHO’s priority pathogens list.^494,495 MRSA and S. aureus are prevalent in Gram-positive bacteria studies, while A. baumannii, P. aeruginosa, K. pneumoniae, and E. coli are primary targets in Gram-negative bacteria research.

The type of animal models used is another important trend. Approximately 84% of the compounds discussed in this review were evaluated in mouse infection models, while the remainder were tested in alternative models, such as Galleria mellonella and Caenorhabditis elegans. These non-mammalian models accelerate in vivo efficacy and toxicity testing while providing ethical benefits by reducing reliance on mammalian subjects.^496–498 Notably, C. elegans serves as a valuable model organism for infection research, owing to its shared innate immune pathways with humans.^499,500 The FDA’s decision to no longer require animal testing for new drugs is expected to pave the way for increased adoption of alternative models, thereby facilitating rapid preclinical assessment.⁵⁰¹ These insights provide valuable context for understanding current trends in antibiotic development, particularly the critical focus on combating high-priority pathogens.

Conventional antibacterials, despite their structural diversity, act through a relatively limited set of mechanisms.⁵⁰² By contrast, the compounds highlighted in this review span a broad range of scaffolds and motifs and include candidates that engage less explored targets and pathways, expanding the antibacterial strategy space and helping to address AMR by opening up new target opportunities.¹²⁵

Within this broader landscape, a key aspect of antibiotic discovery remains the exploration of novel mechanisms of action and targets.¹²⁵ The analysis of compounds in this review reveals that a considerable portion of new compounds targets traditional pathways, with 16% acting as inhibitors of cell wall biosynthesis and 12% inhibiting DNA synthesis. A surprising 22% of compounds that disrupt the cell membrane, and 17% act through non-classical or antivirulence-associated mechanisms, including targets such as LpxC, quorum sensing pathways, and LasB elastase. The relatively small number of compounds that target DNA replication, compared to those that inhibit other cellular processes, is notable given the numerous proteins involved in the bacterial replisome, which presents a multitude of potential drug targets.³¹⁶ One major obstacle is the incomplete characterisation of DNA replication machinery in key pathogens, limiting our understanding of where to focus drug development efforts. Identifying replication inhibitors has proven challenging, often relying on in vitro assays, such as enzymatic activity or DNA replication, which may not accurately predict in vivo efficacy.^315,503

Other innovative mechanisms include protein FtsZ inhibition, crucial for bacterial cell division. Of note, the antibiotic candidate TXA709 (10), currently in clinical phase I, targets FtsZ. Other promising approaches focus on bioenergetic metabolism, exemplified by compounds 72 and 73.

Remarkably, only three compounds, iboxamycin (120), cresomycin (121), and compound 127 were identified as protein synthesis inhibitors, which is notable since protein synthesis is critical for bacterial growth and replication, accounting for approximately 50% of the total energy consumption in growing bacteria.^504,505 Targeting protein synthesis thus remains a promising strategy, including against other pathogens. Likewise, only two compounds, 47 and 125, have been identified as efflux pump inhibitors, specifically targeting the multidrug exporter AcrB and MexB, respectively. This approach holds potential for developing antibiotic adjuvants, as it can synergise with existing antibiotics, thereby boosting their effectiveness.^506,507 The scarcity of efflux pump inhibitors may stem from our incomplete understanding of the intricate efflux systems employed by bacteria, making it a challenging area for drug development.¹²⁹

A key question emerging from the data compiled in this review is the likelihood of these agents advancing to clinical trials. To provide clarity, we have summarised the major physicochemical, pharmacokinetic, and biological properties of the compounds outlined in Section 5, as presented in Table 2. This summary aims to give an overview of the attributes that could influence the transition of these compounds into clinical development.

Table 2.

Summary of some key properties exhibited by the compounds discussed in Section 5. In our colour-coded system, green denotes compounds in advanced preclinical development that have undergone key PK profiling. Pale orange indicates data that falls outside the recommended lead criteria for oral antibiotics, while a darker orange signifies data that raises more relevant concerns for an oral lead candidate. Yellow indicates the ten most commonly used antibiotics worldwide.

Compound	MW	ClogPa	Aq. Sol. (µM)b	MICc (µg/m L)	HC50 (µM)d	hERG IC50 (µM)	PPBe (%) M/H	in vivo T1/2 (h)f M (R)	in vivo F (%)g M (R)	Survival rate (%)	Log10 CFU Reduction
13	508.5	4.81	-	0.25	-	-	-	-	-	-	-h
14	484.5	8.09	-	4-11	-	-	-	-	-	-	-h
15	441.6	6.31	-	1-4	-	-	-	(4.3)	-	-	~1
16	390.4	3.24	-	2	-	-	-	-	-	-	~1
18	280.2	-0.3	-	32	-	-	26/19	0.4	80	-	>0.1
19	865.3	-4.75	-	<2	-	-	-	-	-	-	~1.5 - 3.8
20	435.3	-0.11	10000	-	-	>100	76/84	4.19	-	-	1.8
21	225.2	1.6	-	-	-	-	-	-	-	56	-
22	321.7	4.3	-	-	-	-	-	-	-	-	1 - 2
23	529.5	-1.36	-	-	-	-	6/35	>7.5k	-	-	~7i
24	332.3	-0.44	-	-	-	>60	57/77	-	-	-	~3
25	254.3	1.76	-	-	-	-	-	-	-	50	-
26	605.6	-6.86	-	-	-	-	-	-	-	-	>3
27	411.3	4.75	-	<0.5	-	-	-/95	11.2	65	-	0.78
28	655.7	-1.4	-	0.5-16	>32	-	-	-	-	-	~3
29	346.3	-0.89	-	0.5-2	-	-	-	-	-	100	1
30	517.4	-2.76	-	<0.25	-	-	-	-	-	-	2
31	546.6	1.99	-	0.5-8	-	-	-	-	-	100	-
32	183.2	0.87	-	8	-	-	-	-	-	-	3 – 6i
33	379.2	5.55	-	0.15	>100	-	-	-	-	-	2
34	378.2	6.68	-	0.08	>100	-	-	-	-	-	~1.5
35	592.9	3.43	-	0.25-16	-	-	-	-	-	64	-
36	729.1	7.06	-	0.5-2	-	-	-	-	-	-	2.75
37	366.4	3.49	-	<1.5	-	-	-	-	-	56	4.17
38	413.5	4.08	-	<1.5	-	-	-	-	-	-	5.3
39	388.6	6.08	-	<1.5	-	-	-	-	-	-	6.6
40	386.5	5.67	-	<3.1	-	-	-	-	-	-	6.3
41	899.2	6.31	-	<1.56	-	-	-	-	-	-	6.8
42	682.8	2.18	-	1.5-6.2	-	-	-	-	-	-	5.2 – 5.9
43	843.1	10.28	-	2	-	-	-	-	-	-	2.8
44	1093.8	0.09	-	0.5-8	-	-	-	-	-	>90	-
45	442.4	3.78	-	-	-	-	-	>4.5k	-	93	3.1
46	417.5	3.23	-	-	-	-	-	-	-	100	-
47	418.5	2.48	-	-	-	-	-	-	-	40	-
48	480.5	3.99	57	<0.5	>32	-	-	2.21	26	-	4
49	600.9	1.15	-	-	-	-	-	-	-	-	2.5
50	417.3	6.83	-	2-3	-	-	-	-	-	-	~2
51	738.6	10.72	-	1	-	-	-	-	-	-	~1
52	497.4	2.34	-	2-16	>256	-	-	-	-	50	~3
53	576.3	4.54	-	1-32	>200	-	-	-	-	62	2
54	513.9	9.21	-	1-2	>25	-	-	>3.0k	-	-	2.3
55	680.9	3.25	-	0.5-32	>480	-	-	-	-	-	~2.1
56	512.7	9.13	-	1.32	>255	-	-	-	-	-	~2.8
57	500.4	3.9	30	0.31	-	-	97/-	1.6	-	-	1
59	387.4	-1.1	3270	0.25-4	-	>300	59/82	(4.64)	(92)	-	-h
60	403.4	-0.21	-	<3.1	-	-	-	-	-	-	-h
61	403.4	-0.33	-	<1.5	-	-	-	-	-	-	-h
63	489.5	1.4	69	0.25	-	103	88/85	0.6	49	100	-
64	371.3	2.65	-	<0.0005	-	-	-	1.75	-	-	~3.5
65	250.2	-1.82	98	<0.008	>100	>50	88/-	1.0	-	-	3.5
66	506.5	0.55	76	<4	-	5	15/14	-	-	-	1.5
67	489.55	3.34	-	0.12-4	-	>40	-	(8.5)	-	-	~1
68	508.3	5.14	-	<0.031	-	>30	-	-	-	-	2
69	455.3	3.65	-	<8	>256	-	-	4.39	35.4	-	2.7
70	953.2	4.37	-	4-8	>130	-	-	2.5	-	50	~1.4
71	323.4	2.19	-	0.7-8	-	-	-	-	-	50	-
72	377.4	4.22	1.0	ND	-	-	99/-	2.79	-	-	-h
73	518.6	7.49	-	3	-	-	-	-	-	-	2
74	388.4	3.8	-	<0.5	-	-	-	-	-	-	-h
75	352.5	3.08	-	<4	-	-	-	-	-	-	~0.5
76	415.5	7.47	-	1-8	-	-	-	-	-	-	~0.5
77	460.6	5.23	-	<4	-	-	-	-	-	-	~2
79	488.2	2.36	-	1-2	-	-	-	-	-	-	~1 – 2
80	387.8	4.82	-	<0.25	-	-	-	2.26	61	-	-h
81	345.3	1.15	-	<0.5	-	-	-	-	-	100	-
82	542.6	5.15	-	2.1-8.6	-	-	-	-	-	70-90	-
83	307.3	3.23	-	0.002-1	-	-	-	-	-	50	-
84	365.4	4.98	-	<4	-	-	-	15.8	-	-	~0.7
85	570.5	-4.77	-	-	-	-	-	-	-	-	0.5
86	531.4	6.38	-	1.6	-	-	-	-	-	-	Undetected
87	402.4	5.61	-	<0.12	-	-	-	-	-	-	1.73
88	480.7	6.3	-	0.25-1	1600	-	-	0.85	-	70	ND
89	307.3	-0.84	127	1	-	-	-	-	-	-	1.62 – 1.92
90	380.4	4.50	-	0.008-8	>100	-	-	-	47	70-100	2-3
91	384.2	3.83	>1000	0.25-1	>660	-	-	(8.7)	86	-	-h
92	308.3	1.82	-	0.01-31	>200	-	71/-	12	-	-	~2h
93	512.6	-0.73	34	-	-	-	76/-	6.2	-	-	~0.5
94	608.6	1.26	25	-	-	-	82/-	6.5	-	-	~0.5
96	314.4	3.74	-	-	-	-	-	-	-	~50	ND
97	451.5	2.73	-	0.13-4	-	-	91/-	-	-	-	3.2
98	495.6	1.72	-	0.25-8	-	-	91/-	-	-	-	-
99	433.4	-1.15	886	0.25-1	-	-	91/-	-	-	-	-h
100	432.5	-1.1	>1680	0.5	-	-	63/-	-	-	-	0.8
101	463.5	-1.97	>1680	0.5-1	-	-	97/-	-	-	-	0.6
102	460.5	-1.28	-	1-8	-	-	75/-	-	-	100	-
103	343.1	2.56	-	-	-	-	-	-	-	16	-
104	341.3	3.68	-	-	-	-	-	-	-	-	~1i
105	417.4	1.26	-	-	-	-	-	-	-	87.1	-
106	366.3	3.22	-	-	-	-	-	-	-	100	-
107	270.2	2.11	444	-	-	-	-	-	-	-	-h
108	285.3	-0.13	-	-	-	-	-	-	-	95	-
109	303.3	0.17	-	-	-	-	-	-	-	86	-
110	470.9	4.77	-	-	-	-	-	-	-	81	-
111	452.3	5.56	-	>200	-	-	-	-	-	-	~4
112	261.2	0.8	-	2.0	-	-	-	-	-	-	5
113	541.5	2.3	-	0.06-8	-	-	-	-	-	100	-
114	358.3	-0.96	-	0.06-8	-	-	-	-	-	87.5	-
115	392.5	4.94	-	2.0	-	-	-	-	-	-	~1
116	410.4	4.87	-	0.5-2	-	-	-	-	-	-	0.5
117	417.4	7.4	-	1	-	-	-	-	-	-	~0.6
118	429.4	0.58	-	2	-	-	-	(7.10)	-	-	-
119	430.9	4.33	-	-	-	-	-	-	-	-	~1
120	495	3.22	-	0.03-8	>50	-	95/92	-	27	-	~2
121	498.6	3.75	-	0.06-4	-	-	-	-	-	-	2.2 – 4.6
122	405.4	2.38	-	1-8	-	-	-	-	-	-	2
123	489.8	7.1	-	1-2	-	-	-	-	-	90	-
124	543.5	0.76	-	5	-	-	-	-	-	-	~4
125	330.8	3.01	>1000	1-8	-	>30	46/-	>10	-	-	~4
126	555.6	8.55	-	0.0004	-	-	-	-	-	-	~1
127	381.3	2.49	-	0.03-8	-	-	76/-	11.9	122	100	~6
128	459.1	4.0	-	-	-	-	-	-	-	-	~1.7
129	395.3	4.61	-	0.5-4	-	-	-	-	-	100	-
AMX	365.4	0.87	-	-	-	-	17l	~1l	60l	-	-
AZM	748.9	3.03	-	-	-	-	<50l	68l	38l	-	-
CIP	331.3	0.28	-	-	-	-	30l	3.5l	70l	-	-
DOX	444.4	0.63	-	-	-	-	~8l	10-22l	~100l	-	-
CLN	424.9	2.16	-	-	-	-	95l	~3l	90l	-	-
MTZ	171.1	-0.02	-	-	-	-	20l	8l	90l	-	-
SMX	253.2	0.89	-	-	-	-	70l	10l	~90l	-	-
TMP	290.3	0.91	-	-	-	-	44l	8-10l	~90l	-	-
PEN	350.3	2.09	-	-	-	-	~80l	~1l	~60l	-	-
CLR	747.9	3.16	-	-	-	-	~70l	~3l	50l	-	-

^aCalculated logP values from ChemDraw v23.1.

^bKinetic solubility.

^cMIC range against the bacteria tested.

^dHaemolytic activity red blood cells.

^ePercentage binding to mouse (M) or human (H) plasma proteins.

^fHalf-life in mice (M) or rats (R).

^gOral bioavailability in mice (M) or rats (R).

^hDifferent methods were employed to evaluate the in vivo efficacy.

ⁱAdministered in combination with an antibiotic.

^jData for rabbits.

^kData from in vitro evaluation.

^lClinical data retrieved from the DrugBank database. (-): not determined/available. AMX, amoxicillin; AZM, azithromycin; CIP, ciprofloxacin; DOX, doxycycline; CLN, clindamycin; MTZ, metronidazole; SMX, sulfamethoxazole; TMP, trimethoprim; PEN, phenoxymethylpenicillin; CLR, clarithromycin.

The development of several compounds listed in Table 2 faces significant challenges due to issues such as high lipophilicity, inadequate solubility, cytotoxicity, unfavourable hERG inhibition, and suboptimal PK. In the context of infectious disease drug development, lead selection criteria typically encompass a logP value of less than 5 and ideally less than 3, solubility of over 10 μM at pH 7.4, an oral half-life longer than 2 hours, and oral bioavailability exceeding 25% in rodents.⁵⁰⁸ While in vitro potency is crucial, it should not come at the expense of physicochemical properties or drug metabolism and PK characteristics. To ensure optimal oral bioavailability, guidelines such as Lipinski’s Rule of Five and Veber’s guidelines recommend limiting the number of rotatable bonds to fewer than 10 and the polar surface area to less than 140 Ų.^509,510 These criteria are best regarded as useful guiding principles rather than strict exclusion rules in antibacterial discovery.

Lead compounds that follow Lipinski’s Rule of Five are often prioritised for oral drug development, but strict adherence to these parameters can restrict antibiotic discovery, since many effective antibacterials, particularly natural products, fall outside these guidelines.⁵¹¹ In practice, many widely used antibiotics still fit within Lipinski’s criteria; for example, among the ten most commonly used antibiotics worldwide, only azithromycin and clarithromycin deviate from these rules (Table 2).⁵¹² This trend does not hold for anti-tuberculosis drugs, which more frequently violate Lipinski’s parameters (e.g., bedaquiline and delamanid).⁵¹³

Unlike many drugs developed for areas such as the nervous or cardiovascular systems, antibacterials often fall outside conventional drug-like space, tending to have higher molecular weights, greater hydrophilicity, and lower lipophilicity. They also tend to contain more oxygen and nitrogen atoms, hydrogen bond acceptors, and ring structures than the average drug.⁵¹⁴ Nonetheless, flexibility in applying these guidelines does not lessen the importance of careful developability assessment. These unique properties likely arise from their need to be effective in non-human systems and to penetrate the cell walls of bacteria. However, it is important to note that these extreme physicochemical properties and molecular complexities can increase the predicted risks to developability, including challenges related to bioavailability, permeability, solubility, synthesis, and formulation, which can ultimately compromise their chances of successful development.⁵¹⁵

Many antibiotics have reached clinical use despite these challenges. Aminoglycosides, for example, often violate traditional drug-likeness rules yet remain clinically effective, showing that useful antibacterials can emerge even when they do not fit conventional criteria. They are not isolated examples; numerous other drug candidates in the clinical pipeline and several preclinical compounds discussed herein similarly fail to meet these criteria.

Oral dosing is commercially attractive, but intravenous delivery remains a widely accepted and effective option for treating life-threatening, highly drug-resistant infections.⁵¹⁶ More broadly, strict reliance on conventional drug-like property guidelines can sometimes limit the identification of novel antibacterial candidates.⁵¹¹ While many leads in Table 2 lack complete profiling, particularly in areas like solubility, hERG inhibition, plasma protein binding, and bioavailability, there are still a few compounds (20, 27, 48, 59, 69, 80, 91, 125, and 127) that appear comparatively more promising for further development based on the data currently available. Nonetheless, moving from preclinical success to a clinically viable therapy remains a long and unpredictable process.

Many compounds discussed in this review have not yet met the criteria expected of drug candidates and will require further optimisation before they can progress through preclinical development and into clinical trials. As with drug discovery more broadly, a proportion will not advance, even when they show encouraging efficacy in animal models. This attrition is an inherent part of development in both academic and industrial settings.³⁶

The large number of discontinued antibacterial programmes highlights how difficult it remains to translate hits from HTS and target-based campaigns into robust leads and drug candidates, in part because permeability and efflux rules differ across species and are still not fully understood.^517–519 An example is the pyrazolopyrimidines series developed by AstraZeneca, showcasing nanomolar potency against the ligase MurC, though efficacy is restricted to strains lacking key efflux pumps.⁵²⁰

Looking ahead, there is growing emphasis on narrow-spectrum agents and on expanding into less explored targets, although progress is constrained by gaps in bacterial biology and target understanding.^125,521 Advances in AI-enabled methods, including improved structure and interaction prediction tools such as AlphaFold 3, may help accelerate target prioritisation and design by complementing conventional screening and modelling approaches.⁵²²

AI holds substantial potential to transform antibiotic discovery, although its full impact has yet to be demonstrated.^523,524 A recent analysis by Meier et al. from the Boston Consulting Group suggests AI is likely to improve clinical success rates for drug candidates, boosting pharmaceutical R&D productivity.⁵²⁵ While these projections are cautiously optimistic and AI has yet to produce a fully approved drug, its impact on chemistry is already clear. Rather than being a future development, AI is now becoming embedded across the discovery workflow, from data analysis and virtual screening to hit prioritisation and molecular design, and its influence on how drug discovery is practised will continue to grow as tools and datasets mature.

Recent work in antibacterial discovery illustrates this shift, including the identification of halicin (125),¹⁰⁶ abaucin (16),¹⁰⁷ and compounds 130 and 131 (Figure 47).⁵²⁶ These advancements, made possible by the application of deep learning methods, showcase AI’s capability to identify potential antibiotics within vast chemical libraries.⁵²⁷ These innovative compounds yielded in vivo efficacy in animal models. These AI-discovered molecules can be easily optimised through medicinal chemistry, enabling the design of analogues with improved efficacy and more suitable PK profiles.

Figure 47. Antibacterial small molecules identified via AI-based approaches.

Ultimately, this review highlights examples where antibacterial discovery efforts have translated ideas and compounds into viable preclinical leads. Progress in this area depends on sustained collaboration across disciplines. The medicinal chemistry case studies highlighted in this review illustrate that sustained interdisciplinary efforts can translate promising concepts into potential therapeutic opportunities, despite the high attrition that characterises antibacterial discovery. We hope they help guide and motivate continued work to strengthen the pipeline, since the potential rewards to society are so significant.

7. Final Remarks

We have seen a great many success stories of antibiotic drug discovery, though it is a challenge to convey how much detailed, complicated, innovative work has been carried out by so many researchers. The diversity of chemical structures and biological targets examined in this review is breathtaking, requiring creative, frontier science from our peers vs “venerable” targets such as DNA through to an “industry favourite” of BLIs, and structures from beta-lactam derivatives through to cationic amphiphiles and spiroazetidine-thiatane-oxides. There is an excellent online resource tracking the preclinical pipeline.⁵²⁸ Naturally we need many more promising compounds vs more targets if we are to overcome the growing threat of AMR. We hope that there are many more successes coming through from the earlier stages of discovery. That would require a growth of funding for early stage discovery from government research councils and foundations; several notable funding calls have been run recently (e.g., Pathways to Antimicrobial Clinical Efficacy (PACE) and the Gram-Negative Antibiotic Discovery Innovator (Gr-ADI) run by a coalition of funders) and we hope this continues as a way of mobilising the community around key challenges and providing new starting points that might also be attractive for industrial development.

Success in early-stage discovery would bring with it a challenge to those engaged in translation (whether from industry, governments or NGOs such as GARDP) to ensure that this high-quality science is taken through to the clinic, through the development of mechanisms by which drug development can be funded at scale. This remains a crucial and frequently debated aspect of the development of antibacterials quite broadly, with several bold push- and pull-initiatives and fund-raising mechanisms currently being trialled, such as the UK’s Subscription Model; a survey of this important area is beyond the scope of this review, but excellent overviews are available⁴³ and John Rex continues to update the community on new developments via the AMR Solutions newsletter.⁵²⁹

We have highlighted many successes. It is of course always difficult to be clear on the number of failed projects, since they are not all published, particularly those in the private sector. Industry-sharing “accelerator”-type groups disseminate valuable learnings within that group, but typically not in detail to the wider community.⁶⁴ Failed approaches and “negative” data are valuable to others, even if they are more difficult to progress or publish; such stories help others to learn and help prevent needless duplication of effort. GARDP runs the Revive initiative to try to reduce the loss of “institutional knowledge” when projects or teams come to an end, as has happened all too often in recent years. Ensuring all aspects of projects are in the public domain was a founding principle of The Open Source Antibiotics consortium that carries out all its work openly online⁵³⁰; this approach is far from common, largely because of concerns (justified or not) over the translatability of assets with no protected intellectual property.

In putting together this review, we have had the opportunity to highlight a diversity of projects, but a review naturally raises a question: what is missing? For example, what do we need as a community to accelerate the science? A few years ago, the answer might have been “artificial intelligence” but as described above, this field has already contributed and will only become more powerful as an aide to researchers. AlphaFold has developed into an invaluable tool for target-based drug discovery, built through machine learning as applied to high quality, curated data deposited by humans in the Protein Data Bank⁵²²; if we were to build a similar resource for protein-small molecule interactions, we might see similarly dramatic productivity enhancements for predictive hit finding.⁵³¹ The CO-ADD initiative was notable as a well-funded, community resource aimed at generating phenotypic data from diverse compounds⁵³²; that there was a way of having grassroots compounds evaluated with consistent methods was very useful to us all. (Indeed, there are relatively few such resources in medicinal chemistry quite broadly, to assess critical parameters such as solubility and intrinsic clearance). A related example is the measurement of compound accumulation: a crucial area, helping to inform the direction of antibacterial projects, and yet this remains something without a well-funded evaluation centre for contributed compounds. We would argue this is needed. We would also argue for the need to continue support for other government-supported services that could help accelerate antibacterial development,⁵³³ for example the Chemistry Center for Combating Antibiotic-Resistant Bacteria (CC4CARB), which can be used to accelerate hit to lead chemistry for promising series.

What of biological targets that are not being pursued? We wish to pursue novel targets because these can inevitably help in fighting the development of resistance, including as part of combination therapies. It is often a challenge to pursue new targets because they are less fashionable, less easy to evaluate and are more risky,⁵³⁴ but several are clearly attractive because of their essentiality, level of homology vs human orthologues and their conserved nature.¹²⁵ Some examples might be: 1) the WD40 repeat proteins,⁵³⁵ 2) the group of seven Lpt proteins involved in transport of lipopolysaccharide to the outer membrane, and 3) several of the ca. 20 aminoacyl tRNA synthetase enzymes that have never been drugged; readers will inevitably have their own favourite unexplored corner of the bacterial proteome – there are plenty to choose from, and even if such research does not result in medicines, the development of chemical probes (potent, in-cell active, selective protein binders) can help the research community properly understand microbial protein function. There are tantalising possibilities for multi-targeting compounds, where one molecule could target more than one biological target, such as Gepotidacin or an as-yet unidentified molecule that could target the Mur ligases. Additional strategies to delay the onset of resistance might be to exert less pressure on the bacterium through combating mechanisms such as biofilm formation or virulence pathways. We will almost certainly need a greatly improved understanding of how we might exploit vulnerabilities of dormant persisters, given that they appear to be not entirely dormant.⁵³⁶

What new chemotypes might be promising? Of course, Nature may continue to deliver new molecules for us if we are able to innovate in how we harvest “uncultivatable” sources,⁵³⁷ and in how we mine under-explored microbial genomes.⁵³⁸ Some antimicrobial peptides remain attractive because they are able to function through a physical mechanism of membrane perturbation that might provide advantages in combating the emergence of resistance; similar comments apply to ionophores and other simple structures we have seen in this review (e.g., Figure 10). Perhaps metal complexes will undergo a renaissance.^539,540 There will be a continual need to innovate in the chemistry allowing for the construction of the sometimes-complicated structures involved in BacProtacs and guanidinium/pyridinium containing compounds aimed at improving accumulation.

What is certain is that some combination of the approaches described in this review, the tenacity of the research community across disciplines (which includes our AI-based co-workers) and the sustained financial support of our community to overcome the peculiar funding challenges of antibiotic drug development will provide society’s best chances of developing novel and effective antibiotics that will be of enormous future value to society, worldwide.

Abbreviations

AChE

acetylcholinesterase enzyme

ADDCs

Academic Drug Discovery Centres

AI

artificial intelligence

AMP

antimicrobial peptide

AMR

antimicrobial resistance

AUC

area under the curve

AZT

azotreonam

BARDA

Biomedical Advanced Research and Development Authority

BLI

β-lactamase inhibitor

CARB-X

Combating Antibiotic-Resistant Bacteria Biopharmaceutical Accelerator

CC4CARB

Chemistry Center for Combating Antibiotic-Resistant Bacteria

CDI

Clostridioides difficile infections

CFU

Colony Forming Unity

COE

conjugated oligoelectrolytes

CRAB

carbapenem-resistant A. baumannii

CRE

carbapenem-resistant Enterobacterales

CRPA

carbapenem-resistant Pseudomonas aeruginosa

C55-PP

undecaprenyl pyrophosphate

C55-P

undecaprenyl phosphate

DHFR

dihydrofolate reductase

DNDi

Drugs for Neglected Diseases initiative

DPPS

decaprenyl diphosphate synthase

DR

drug-resistant

enoyl-ACP

enoyl-acyl carrier protein

ESBL

E. coli expressing extended spectrum β-lactamase

FAS-II

fatty acid biosynthesis pathway

FDA

U.S. Food & Drug Administration

FICI

fractional inhibitory concentration index

FtsZ

Filamenting temperature-sensitive mutant Z

G4

G-Quadruplex

GARDP

Global Antibiotic Research and Development Partnership

Gtfs

glucosyl transferase

GyrB

Gyrase subunit B

HTS

high-throughput screening

LpxC

UDP-3-O-acyl-N-acetylglucosamine deacetylase

LPS

lipopolysaccharide

MBC

minimum bactericidal concentration

MBEC

minimum biofilm eradication concentration

MBIC

minimum biofilm inhibitory concentration

MBLs

metallo-β-lactamases

MDR

multidrug-resistant

MEPM

meropenem

MetRS1

type 1 methionyl-tRNA synthetase

MRSA

methicillin-resistant Staphylococcus aureus

MSSA

methicillin-sensitive Staphylococcus aureus

NADKs

NAD kinases

NDA

New Drug Application

PBP

penicillin-binding protein

PgcA

α-Phosphoglucomutase

PMF

proton motive force

PPB

plasma protein binding

PK

pharmacokinetic

QIDP

Qualified Infectious Disease Product

QRDR

quinolone resistance-determining region

QS

Quorum sensing

RND

Resistance-Nodulation-Cell Division

ROS

reactive oxygen species

SBL

serine β-lactamase

SrtA

Sortase A

T3SS

type III secretion system

TB

tuberculosis

UPPS

undecaprenyl diphosphate synthase

UppP

undecaprenyl diphosphate phosphatase

VRE

vancomycin-resistant Enterococcus

WHO

World Health Organisation

Data availability statement

No primary research results, software or code have been included and no new data were generated or analysed as part of this review.