The molecular pathways of polyphenols in combating multidrug-resistant microbial superbugs: a comprehensive mechanistic review

1 Introduction

The global crisis of antimicrobial resistance (AMR) is among the most daunting challenges confronting modern medicine and has the potential to reverse decades of treatment advances made through the discovery and clinical use of antibiotics (Ventola, 2015). The World Health Organisation (WHO) has been at the forefront of the alarm that, unless urgent and coordinated action is taken, the world is on the brink of entering a so-called postantibiotic era, a bleak scenario in which routine surgical operations and minor infections once again become life-threatening (World Health Organisation, 2024). In May 2024, a landmark revision of the WHO priority pathogen list reaffirmed that the ESKAPE pathogens Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter retain their critical-priority status, which can be attributed to their extraordinary capacity to acquire and horizontally disseminate diverse resistance mechanisms, rendering multiple classes of antibiotics simultaneously ineffective (World Health Organisation, 2024; Miller and Arias, 2024).

All these pathogens contribute to the fact that by far the majority of healthcare-associated infections are directly caused by these pathogens, especially those in intensive care and surgery, and are directly associated with worsened patient outcomes, extended hospitalisation, and significantly increased healthcare expenses (Miller and Arias, 2024; Sakalauskienė et al., 2025). It is estimated that the economic and public health burden will become catastrophic: the seminal O'Neill Review held an estimate of 10 million deaths annually by 2050 due to MDR infections (O'Neill, 2016). The causes of this crisis are multifactorial and include the overuse and misuse of antibiotics in human and veterinary medicine, insufficient control of infection, and the inherent evolutionary plasticity of bacterial genomes (Ventola, 2015; Kelly et al., 2024). Adding to the biological aspect is a deep and growing lack of innovation in antimicrobial drug discovery. The classic antibiotic development pipeline has effectively stalled: no new antibiotic class has been successfully commercialised to treat Gram-negative infections for more than 30 years, largely because of the high failure rates of clinical trials, regulatory complexities, and unfavourable return-on-investment analyses that have driven major pharmaceutical companies out of the field (Ventola, 2015). Recent approvals, although technically valuable, represent more of a refinement of existing chemical scaffolds than mechanistically novel entities (Miller and Arias, 2024). This stalemate has fundamentally focused scientific interest on alternative approaches, of which the systematic exploration of plant-derived natural products is among the most promising.

In general, polyphenols represent a broad and chemically heterogeneous group of plant secondary metabolites with the defining chemical feature of one or more aromatic rings adorned with hydroxyl substituents. This chemical diversity is unparalleled among natural product families and underlies their remarkable ability to interact with a wide range of biological macromolecules, including proteins, lipids, nucleic acids, and polysaccharides (Manach et al., 2004; Tsao, 2010). The major classes of flavonoids and phenolic acids, tannins, stilbenes and curcuminoids are widely distributed in common food sources, including green tea and berries, turmeric, red wine, and whole grains, and have long been linked to general-effect health benefits (Manach et al., 2004; Daglia, 2012) (Table 1). Mechanistically, polyphenols are characterised by polypharmacology, the concomitant action of multiple, often structurally unrelated cellular targets that creates an overwhelming combinatorial challenge to bacteria attempting to develop resistance (Cueva et al., 2022; Bayode et al., 2024). Initial clinical and epidemiological findings from our group (Bayode et al., 2021) indicated the dissemination of the plasmid-mediated quinolone resistance genes qnrS and qnrB in the nosocomial environment, and the urgency of multitarget intervention strategies is explained. Subsequent studies confirmed the strong anti-MRSA activity of phytophenolic compounds of Jatropha tanjorensis, such as phytol, octadecanoic acid, and n-hexadecanoic acid, against Escherichia coli, S. aureus, and P. aeruginosa (Cushnie and Lamb, 2011; Babayemi et al., 2021; Duda-Madej et al., 2022).

TABLE 1. Primary classes of polyphenols, representative compounds, chemical properties and major sources of polyphenols in the diet (Manach et al., 2004; Tsao, 2010).

Beyond bacterial AMR, new fungal pathogens are equally important components of the AMR crisis. First discovered in 2009, Candida auris has been described to induce resistance by upregulation of the CDR1 efflux gene and thus confer broad azole resistance in most clinical isolates (Cortegiani et al., 2018; Iyer et al., 2020). The mutation of the TR34/L98H in the CYP51A gene is the major panazole resistance determinant in Aspergillus fumigatus (Snelders et al., 2008; Cowen et al., 2015). Positively, polyphenols are also active against these fungal superbugs, extending their therapeutic relevance beyond the bacterial domain (Navarro-Martínez et al., 2006; Fakhim et al., 2025).

The evidence that supports the different mechanisms listed in this review is not consistent or comprehensive. Efflux pump inhibition proves to be the most well supported mechanism: three methods molecular docking, ATPase activity assay, and ethidium bromide efflux experiments yielded convergent results, with each independent research group supporting the others in a manner that is mutually reinforcing. Claims of DNA gyrase inhibition, however, depend primarily on computational docking simulations and older enzymatic studies with no confirmatory enzymatic data from MDR clinical isolates, and thus constitute a materially weaker evidentiary basis. Although mechanistically sound, and sometimes quantitatively impressive, the findings on antibiofilm activity are typically from single-species, static biofilm models that may not completely represent the structural and phenotypic complexity of polymicrobial or clinically generated biofilms. Unlike other reviews on the topic by Alum et al. (2025) and (Rahbardar et al., 2025) which focus on quorum sensing inhibition by natural products and P. aeruginosa resistance mechanisms, respectively, the present review provides a detailed mechanistic integration at the level of the entire ESKAPE group of bacteria and an explicit consideration of the concentration-achievability gap, that is, the difference between in vitro effective concentrations and clinically attainable tissue levels.

This review synthesises the mechanistic findings of peer-reviewed literature published primarily from 2013 to 2025 and retrieved using the PubMed, Scopus, and Web of Science databases with the search terms “polyphenol” and “flavonoid” and “antimicrobial resistance” and “ESKAPE” and “efflux pump inhibition” and “quorum sensing” and “biofilm” and “MDR”. This review is based on molecular and biochemical studies that present quantitative and mechanism-solving information, mainly concerning ESKAPE pathogens and select fungal superbugs. We aim to construct a rigorous mechanistic model of polyphenol activity, critically evaluate translational barriers, and outline a focused program for future studies. Table 1 presents the key classes of polyphenols analysed during the review process.

2 Mechanistic molecular pathways of plant-derived polyphenols in MDR pathogen inhibition

The development and dissemination of MDR bacteria have turned curable infections into therapeutic dead ends. The cellular organisation of resistance; including efflux pumps that suck out antibiotics, enzymatic machinery that breaks down or modifies drugs, target site mutations, and protective biofilm matrices is a daunting, multilayered challenge (Ventola, 2015; World Health Organisation, 2021; De Oliveira et al., 2020). Polyphenols, which are derived from plants, have been presented as scientifically compelling candidates capable of dismantling each of these layers, often simultaneously, in a mosaic of interconnected processes, which are discussed in detail in the following subsections.

2.1 Polyphenols as multitarget agents and a polypharmacology model

The basic weakness of most traditional antibiotics is that they have one specific molecular target. This evolutionary weakness provides a simple route through which bacteria can develop resistance by acquiring point mutations in target genes or by obtaining mechanisms of enzymatic bypass (Cushnie and Lamb, 2011; Miller and Arias, 2024). In contrast, polyphenols typically operate in polypharmacology, the coordinated disruption of multiple, often mechanistically unrelated, cell processes, which imposes a combinatorial burden on bacterial adaptation that is quantitatively distinct from that of single-target antibiotics (Cueva et al., 2022; Xie et al., 2015; Blair et al., 2015). By simultaneously attacking the cell envelope, intracellular enzymes, energy metabolism and communication networks, polyphenols generate a multifront attack that significantly increases the resistance to resistance development (Daglia, 2012; Miller and Waldrop, 2010; Cueva et al., 2022; Bayode et al., 2024). Table 2 presents a consolidated overview of well-characterised polyphenol–target interactions across clinically relevant MDR pathogens, which are used as a mechanistic reference map to the sections that follow.

TABLE 2. Unified modes of action of the chosen polyphenols on MDR pathogens.

2.2 Interruption of bacterial cell envelope

The bacterial cell envelope is the major physical and chemical border between the organism and an aggressive external environment. As it is the first point of contact of any antimicrobial agent, its failure is a second and, in many cases, fatal occurrence. Polyphenols (Figure 1), by virtue of their amphipathic molecular architecture, have hydrophobic aromatic rings that are capable of embedding within lipid bilayers coupled with hydrophilic hydroxyl groups that form stabilising hydrogen bonds with polar lipid headgroups and are structurally predisposed to interact with and destabilise membrane systems (Borges et al., 2012; Cueva et al., 2022). This intercalation of membrane-bound phospholipids disrupts the ordered architecture of the phospholipid bilayer that is bound to the membrane and results in quantifiable increases in membrane permeability, the collapse of transmembrane potential, and the leakage of critical intracellular constituents, including K+ions, ATP, and nucleic acids, which are collectively incompatible with bacterial survival (Wang et al., 2016; Cueva et al., 2022). The exact mechanism of this disruption, however, is not homogeneous; it is deeply conditioned by the specific architectural features of Gram-positive versus Gram-negative cell envelopes.

2.2.1 Gram-positive bacteria

The cell wall of Gram-positive bacteria is a thick peptidoglycan lattice that is directly exposed to extracellular substances, making it a key molecular target. The most common catechin in green tea, epigallocatechin gallate (EGCG), was found to bind directly to the S. aureus peptidoglycan layer. This binding creates structural strain on the polymer network, which results in the formation of characteristic microscale grooves in the cell wall architecture and progressive deterioration culminating in cell lysis (Taylor et al., 2014; Wang et al., 2016; Reygaert, 2022). Similarly, galangin, a flavonol present in honey and the roots of Alpinia officinarum, causes the collapse of membrane potential and triggers leakage of K⁺ ions from S. aureus (Xie et al., 2015), which is a well-established indicator of disastrous membrane failure; galangin exposure further drives S. aureus cells into aggregation, compounding this membrane-level insult (Cushnie et al., 2007). Condensed tannins have a broad Gram-positive spectrum of activity, binding strongly to cell wall-associated proteins and membrane lipids in Bacillus subtilis, S. aureus, and Enterococcus faecalis, where they inhibit adhesion and cause irreversible cell damage (Trentin et al., 2013; Cueva et al., 2022).

2.2.2 Gram-negative bacteria

The Gram-negative cell envelope presents a structurally more complex challenge, with an outer membrane (OM) highly enriched with lipopolysaccharides (LPS), which functions as a selective permeability barrier and effectively excludes many hydrophilic molecules. In this case, polyphenols adopt a mechanistically different approach. Research on E. coli has shown that the major action of EGCG is not direct attack on peptidoglycans but rather indirect oxidative attack. EGCG reacts with LPS and autoxidises to produce reactive oxygen species, predominantly hydrogen peroxide (H₂O₂), in a pH-dependent manner (optimal pH 5.7), which then attacks the cell envelope internally, resulting in the formation of so-called nanoscale perforations that compromise the barrier effect of the OM and allow further penetration of antibiotics (Wang et al., 2016; Xiong et al., 2017; Xu et al., 2023). Resveratrol was shown to accumulate in the cytoplasmic membrane of Corynebacterium glutamicum, leading to significant membrane rigidification, quantified by an increase in the value of a generalised polarisation (GP) of 0.27 + 0.1 to 0.39 + 0.1 over 72 h in Laurdan fluorescence experiments, accompanied by a molecular dynamics-verified slowing of lipid diffusion (Howard et al., 2023; Rahbardar et al., 2025). Rutin increases the permeability of the membrane in both P. aeruginosa and MRSA, increasing their susceptibility to antibiotics (Al-Shabib et al., 2017). A schematic diagram of these mechanisms of membrane disruption is presented in Figure 1.

2.3 Repair of antibiotic susceptibility by inhibiting efflux pumps

One of the most clinically common and therapeutically relevant processes that contributes to MDR is the overexpression of efflux pumps. These transmembrane proteins actively export machinery, which constantly expels a wide range of toxic substrates, including structurally diverse antibiotics, out of the bacterial cytoplasm before they can reach their intracellular targets (Du et al., 2018; Duda-Madej et al., 2022). The tripartite AcrAB-TolC system of E. coli (resistance-nodulation-division, RND family) confers resistance to β-lactams, tetracyclines and chloramphenicol, and the tripartite MexAB-OprM system of P. aeruginosa enhances intrinsic resistance by working together with the OM barrier (Duda-Madej et al., 2022; Lawani et al., 2023). These systems are countered by polyphenols through a variety of mechanistically different strategies:

Antibiotic recognition and transport are prevented by direct competitive binding at the substrate binding pocket. Quercetin increases the intracellular accumulation of ciprofloxacin in S. aureus by binding directly to NorA at its efflux site; curcumin destabilises the AcrB pump in P. aeruginosa through a similar competitive mechanism. The phenolic diterpene totarol decreases NorA-mediated ethidium efflux by 50% at subinhibitory concentrations (Smith et al., 2007; Duda-Madej et al., 2022). A second, complementary mode is gene-level suppression: luteolin and naringenin transcriptionally downregulate the norA and acrAB genes, respectively, inhibiting pump production at the source; the gene-suppressive activity of quercetin in MRSA rivals that of the synthetic EPI verapamil (Duda-Madej et al., 2022; Lu et al., 2022; Howard et al., 2023). Indirect sabotage of energy-dependent pumps of the RND and MFS families, which use the transmembrane H⁺ gradient to transport their cargo, is indirect sabotage of the dissipation of the proton motive force by membrane-disrupting polyphenols, such as thymoquinone and carvacrol (Bayode et al., 2024; Lawani et al., 2023).

The most clinically important effect of efflux pump inhibition with polyphenols is the stunning return of susceptibility to otherwise-failed antibiotics, an effect that is effectively converting resistance into susceptibility. Baicalein restores tetracycline and methicillin activity against MRSA; quercetin reduces the MICs of ciprofloxacin, tetracycline, and erythromycin in S. aureus by 3- to 16-fold; and EGCG reduces the MIC of tetracycline against TetK-positive S. aureus by up to 256-fold (Hirai et al., 2010; Duda-Madej et al., 2022). The quantitative data of efflux pump inhibition by NorA are summarised in Table 3 and Figure 2 schematically illustrates the mechanism of NorA inhibition.

TABLE 3. Inhibition of efflux pumps and quantitative data on MIC reduction in MDR pathogens using polyphenols.

2.4 Perturbation of essential intracellular targets

When the cell envelope is compromised or bypassed by the enhanced membrane permeability induced by their own activity, the polyphenols gain access to a rich landscape of intracellular targets. These second, intracellular fronts of antimicrobial activity are the disruption of the following fundamental molecular processes: DNA replication, metabolic homeostasis, and cellular energy production (Cushnie and Lamb, 2011; Kohanski et al., 2010).

2.4.1 DNA replication and transcription inhibition

The bacterial DNA replication machinery interacts with several flavonoids, inhibiting essential enzymes that manage the topology of the chromosome. The most important targets are DNA gyrase and topoisomerase IV. EGCG prevents DNA gyrase activity by occupying the ATP-binding site of the GyrB subunit, inhibiting the energy-dependent supercoiling reaction required to advance replication forks (Gradisar et al., 2007). Quercetin has the same inhibitory effect on GyrB ATPase activity, which results in damage to DNA strands and irreversible arrest of replication (Ohemeng et al., 1993; Plaper et al., 2003; Wu et al., 2013b). A recent molecular dynamics study confirmed the interaction of quercetin with the MurG enzyme in S. aureus, a glycosyltransferase that is essential for peptidoglycan biosynthesis and offers an additional intracellular target that complements its membrane and efflux pump activities (Plaper et al., 2003; Kang et al., 2006; Wu et al., 2013a; D’Andrea, 2015).

2.4.2 Interruption with metabolic homeostasis and energy production

As the main junction of cellular respiration, the tricarboxylic acid (TCA) cycle is a high-value intracellular target. Catechins inhibit TCA cycle enzymes, such as isocitrate dehydrogenase and succinate dehydrogenase, reducing the production of NADH and FADH₂ and thereby starving the respiratory chain of electron carriers needed to produce ATP (Navarro-Martínez et al., 2006). Similarly, cranberry proanthocyanidins downregulate the expression of TCA cycle proteins in P. aeruginosa, disrupting energy metabolism (Ulrey et al., 2014). A parallel susceptibility is the synthesis of fatty acids: EGCG and quercetin are strong inhibitors of the fabI and fabG reductases of the bacterial type II fatty acid synthase (FASII) pathway, which inhibits fatty acid elongation and deprives bacteria of the lipid precursors needed to build and maintain membranes in a biogenic state (Zhang and Rock, 2004). Apigenin acts on FabZ, the acyl-ACP dehydratase (Zhang et al., 2008). At the terminal ATP synthesis level, cranberry proanthocyanidins and tea catechins directly inhibit ATP synthase in P. aeruginosa and S. aureus, respectively, which leads to energy loss and renders essential cellular processes untenable (Stapleton et al., 2004; Ulrey et al., 2014). These intracellular target interactions are summarised in Table 4.

TABLE 4. Classes of polyphenols, intracellular bacterial targets, targeted enzyme pathways, and physiological effects.

2.5 Pro-oxidant activity, reactive oxygen species, and metal ion chelation

Although polyphenols are generally lauded as potent dietary antioxidants in the context of human physiology, the same structural features that confer antioxidant capacity under one redox condition can drive pro-oxidant, bactericidal activity under the different conditions prevailing within and around the microbial cells. This context-dependent redox duality is a critical mechanistic observation that has significantly broadened the existing knowledge on the antimicrobial action of polyphenols (Perron and Brumaghim, 2009; Prochazkova et al., 2011; Cueva et al., 2022).

At neutral-to-mildly acidic pH values (5.7), phenolic hydroxyl groups undergo autoxidation, and redox cycling and Fenton-like chemistry dramatically amplify this process. The resulting reactive oxygen species, superoxide radicals (O₂•^-), hydrogen peroxide (H₂O₂), and the highly destructive hydroxyl radical (•OH), indiscriminately attack all major classes of cellular biomolecules (Figure 3), they cause single- and double-strand breaks in DNA, initiate lipid peroxidation cascades in membranes, and oxidatively modify proteins through carbonylation and disulfide cross-linking (Perron and Brumaghim, 2009; Prochazkova et al., 2011). At concentrations greater than 100 μM, EGCG produces H₂O₂ levels as high as 127 µmol in phosphate-buffered saline and prevents the growth of E. coli by inducing endogenous oxidative stress, which in turn causes the cellular uptake of EGCG (Xiong et al., 2017; Xu et al., 2023). Another key secondary dimension is polyphenol-mediated metal chelation: iron is an essential cofactor in many bacterial enzymes and is often a growth-limiting nutrient during infection (Rutherford and Bassler, 2012; Gopu et al., 2015). Quercetin, gallic acid and other polyphenols bind Fe²⁺/Fe³⁺ with high affinity, indicating that bacteria access this vital resource but simultaneously catalyse the Fenton-driven amplification of the ROS load, which is a dual strategy of nutrient inhibition and oxidative attack (Perron and Brumaghim, 2009; Lee et al., 2021). Recently, resveratrol has been shown to prevent the formation of S. aureus biofilms by inhibiting the synthesis of polysaccharide intercellular adhesin (PIA) and extracellular DNA release, as well as by reducing ROS production (Lee et al., 2014; Gholami et al., 2020b). Table 5 provides a summary of the key pro-oxidant activities and their implications for bacteria.

TABLE 5. Mechanisms of polyphenol pro-oxidants, ROS types, molecular targets, and MDR pathogens.

2.6 Virulence attenuation via quorum sensing inhibition and anti-biofilm activity

An advanced and more recognised treatment approach is not the actual killing of bacteria but the disarmament of their pathogenic potential. The cell density-dependent communication systems that control many bacterial virulence behaviors, including toxin secretion and the production of proteases, as well as biofilm maturation and antibiotic tolerance, are controlled by cell density-dependent communication systems, which are collectively termed quorum sensing (QS). Polyphenols can effectively neutralise bacterial threats by targeting QS and its downstream effect, biofilm formation, a strategy that has been described as antivirulence therapy (Rutherford and Bassler, 2012; Eleraky et al., 2020; Alum et al., 2025).

2.6.1 Quorum sensing inhibition

Quorum sensing is mediated by diffusible autoinducer (AI) molecules, such as N-acyl homoserine lactones (AHLs) in Gram-negative organisms and small autoinducing peptides (AIPs) in Gram-positive bacteria (Papenfort and Bassler, 2016; Miller and Arias, 2024). Receptor binding at the population level with AI concentrations exceeding threshold levels triggers coordinated, population-level changes in gene expression that activate virulence programs (Papenfort and Bassler, 2016; Eleraky et al., 2020; Prabhu et al., 2019). Polyphenols disrupt this network at various nodes: (i) interference with AI biosynthetic enzymes, reducing signal production; (ii) structural mimicry of AHLs, enabling competitive antagonism of receptor proteins such as LasR and RhlR in P. aeruginosa without productive stimulation; and (iii) promotion of AI degradation (Singh et al., 2021; Abdel Bar et al., 2022). Curcumin, quercetin, and resveratrol all consistently suppress the las and rhl QS systems in P. aeruginosa, decreasing the production of the virulence factors pyocyanin, elastase, and rhamnolipid (Ouyang et al., 2016; Ouyang et al., 2020; Ding et al., 2018; Praditya et al., 2019; Gholami et al., 2020a; Gholami et al., 2020b; Singh et al., 2021; Batista de Andrade Neto et al., 2021). The sdiA and luxS genes are repressed by EGCG and licochalcone A (LAA) against Salmonella Typhimurium (Hosseinzadeh et al., 2020). Mangiferin suppresses QS activity and biofilm formation in Chromobacterium violaceum, Aeromonas hydrophila, Serratia marcescens, and P. aeruginosa at extremely low MICs (1–8 μg/mL) (Husain et al., 2017; Ham et al., 2019; Al-Shabib et al., 2025).

2.6.2 Multistage polyphenol-mediated disruption strategy via bacterial antibiofilm activity

Biofilms are organised, surface-associated, bacterial communities wrapped in self-secreted extracellular polymeric substances (EPSs) and express the most recalcitrant form of bacterial infection clinically (Hall and Mah, 2017; Flemming et al., 2016). Bacteria trapped within biofilms are able to withstand concentrations of antibiotics 100 to 1,000 times above their planktonic minimum inhibitory concentrations (MICs) because of limited diffusion, altered metabolic states, and the protective effects of the matrix (Hall and Mah, 2017; Braem et al., 2023). Polyphenols affect the biofilm life cycle at a variety of stages simultaneously. Rutin significantly decreases the biofilm biomass of P. aeruginosa and MRSA, reducing exopolysaccharide (EPS) and extracellular DNA (eDNA) levels and inhibiting adhesion, as confirmed by crystal violet assays and electron microscopy (Al-Shabib et al., 2017; Ivanov et al., 2022; Kim et al., 2022; Roy et al., 2022; Pei et al., 2023; Ren et al., 2024). EGCG interferes with the biofilm in Stenotrophomonas maltophilia by inhibiting adhesion and QS simultaneously (Vidigal et al., 2014) and by suppressing autoaggregation and swimming motility in Vibrio mimicus at an MIC of 256 μg/mL by promoting ROS, damaging membranes, and causing K⁺ leakage (Li et al., 2020). Pseudomonas aeruginosa biofilms are inhibited by Cranberry A-type proanthocyanidins (PACs), which reduce the motility of P. aeruginosa and modify protein expression profiles (Ulrey et al., 2014; Long et al., 2000). Genistein inhibits the expression of aerolysin genes (aerA) and the QS genes ahyI/ahyR in A. hydrophila (Choi et al., 2018; Spagnuolo et al., 2018; Dong et al., 2021). An important clinical implication of these studies is that antibiofilm and anti-QS effects are often achieved at subinhibitory concentrations; that is, polyphenols can effectively neutralise bacterial pathogenicity without exerting the bactericidal effect required to drive the selection of resistant phenotypes (Slobodníková et al., 2016). The multistage antibiofilm mechanism in the form of a scheme is shown in Figure 4. The quantitative antibiofilm and combination data in Tables 6 are used to present the data.

TABLE 6. The selected polyphenols and their MICs against MDR pathogens and mechanisms of biofilm and QS inhibition (compiled with Bayode et al., 2024; Cueva et al., 2022; Al-Shabib et al., 2017; Ivanov et al., 2022; Long et al., 2000; Dong et al., 2021; Al-Shabib et al., 2025; Ouyang et al., 2016; Singh et al., 2021).

2.7 The non-specific mechanisms and false-positive antimicrobial signals of curcumin, EGCG, and quercetin study gaps

One of the problems that has been demonstrated in polyphenol antimicrobial research, despite having been largely overlooked, is that some of the most commonly studied polyphenols, such as curcumin, EGCG and quercetin, produce non-specific signals in standard assay formats. The ability to aggregate in aqueous media, to react with resazurin and other fluorescent or colorimetric reporters used in cell viability assays, and to chelate metal ions present in culture media, can all inhibit microbial growth without any particular interaction with the drug target (Baell and Holloway, 2010; Nelson et al., 2017). These assay supernatant are especially strong when using broth microdilution format at concentrations ≥32 μg/mL, which is a concentration often reported in the literature as the MIC for curcumin against MDR organisms. EGCG has also been reported to have bactericidal activity in many in vitro studies, which may be due to its non-specific pro-oxidant activity, as it undergoes rapid auto-oxidation under aerobic conditions to produce hydrogen peroxide as a byproduct, which could mediate the bactericidal activity in vitro without requiring interaction with a specific bacterial target (Long et al., 2000). Quercetin also undergoes auto-oxidation in cell culture conditions and is deposited in solution when its concentration is above the solubility limit, which can result in the actual concentration in solution being lower than the one tested in the experiment. The researchers in this field are encouraged to confirm antimicrobial activity with the counter-screening of assay interference (such as non-fluorescent cell viability assays, DMSO precipitation test and H2O2 control) and to report if the activity is present in catalase-containing buffer, which would eliminate any peroxide-mediated artifact. Mechanistic claims based on MIC data alone should not be accepted as final without supporting orthogonal biochemical evidence.

2.7.1 Emerging resistance of bacterial pathogens to polyphenols

Among the most often mentioned benefits of polyphenols over traditional antibiotics is that it is believed that they have a multitarget mode of action, which means that resistance is unlikely to be developed in the clinic. This is a sound argument at a mechanistic level, but new evidence indicates that it needs to be taken with a pinch of salt. Staphylococcus aureus and E. coli strains with higher MICs have been selected during repeated exposure to sublethal concentrations of quercetin and resveratrol in laboratory serial passage experiments, due in part to increased expression of the AcrAB-TolC and NorA efflux systems, which are reported to be inhibited by these same polyphenols at higher concentrations (Li et al., 2020; Zhang et al., 2020; Duda-Madej et al., 2022; Li et al., 2022). The apparent paradox that sub-inhibitory concentrations of polyphenols can induce the expression of efflux pumps which pump the polyphenols themselves, suggests an inherent tension in the concept of using these polyphenols as efflux pump inhibitors when they are administered systemically at sub-therapeutic levels. In addition, the genomes of several intestinal Gram-negative rods and Enterococcus species have been identified to contain genes for phenolic-degrading enzymes (tannases, lactonases, polyphenol oxidases) that can be used for polyphenol inactivation without the classical acquisition of resistance genes (Braune and Blaut, 2016). These enzymatic pathways may play a significant role in clinical treatment failure in polyphenol-adjuvant treatment, which has not yet been investigated, but is important. Another challenge is that biofilm-residing cells develop multi-layered tolerance mechanisms (altered metabolic states, matrix-mediated exclusion, persister cell formation) that have so far proven to be resistant to a single compound, despite its breadth of targets, in clinically relevant biofilm models (Hall and Mah, 2017). Such observations do not diminish the therapeutic potential of polyphenols, and they support the notion that their use should be evidence-informed, concentration-limited, and, when feasible, accompanied by strategies that limit the selection pressure that polyphenols may impose.

3 Clinical and translational perspectives

3.1 Synergistic potential with conventional antibiotics

The most directly translatable clinical application of polyphenols is not in their use as single-agent antimicrobials, a role often eclipsed by their current pharmacokinetic limitations, but rather in their strategic utilisation as adjuvants in combination therapy with established antibiotic classes. Once the overall effect of two agents is greater than the sum of the individual effects, it establishes a synergy that is traditionally defined by a fractional inhibitory concentration index (FICI) ≤ 0.5 (Kırmusaoğlu, 2019; Cueva et al., 2022). Mechanically complementary actions of polyphenols include outer membrane disruption of polyphenols, which enhances antibiotic permeation through the outer membrane of Gram-negative cells; efflux pump inhibition entraps antibiotics within cells; and direct inhibition of resistance enzymes such as β-lactamases, which removes the enzyme shell that otherwise shields drugs before they can engage with their target (Cueva et al., 2022; Duda-Madej et al., 2022; Bayode et al., 2024). Curcumin suppresses the expression of penicillin-binding protein 2a (PBP2a), the main determinant of methicillin resistance, in MRSA and once again promotes the bactericidal activity of β-lactams (Adamczak et al., 2020; Wang et al., 2021; Xia et al., 2022). EGCG reinstates the action of aztreonam with MDR P. aeruginosa, a pathogen against which treatment options are grossly inadequate (Betts et al., 2019). A mixture of curcumin and EGCG shows synergistic activity against MDR A. baumannii, one of the critical priority pathogens of the WHO, in checkerboard and time-kill assays (Betts et al., 2014). Plant extracts rich in polyphenols have an overall synergistic effect in combination with cefixime against S. aureus, Klebsiella pneumoniae, and E. coli clinical isolates (Khan et al., 2023). Synergy is not restricted to polyphenol–antibiotic pairings, either: combining different classes of polyphenols amplifies inhibitory activity against the foodborne pathogen Listeria monocytogenes beyond that achievable with any single class (Borges et al., 2013). Table 7 summarises the most important synergistic interactions reported in recent literature.

TABLE 7. Synergistic effects of polyphenols in combination with conventional antibiotics against MDR pathogens.

3.2 Mechanistic extension to fungal pathogens

Although the therapeutic focus of polyphenols in the context of antifungal resistance is of secondary importance to the present review, it focuses on the mechanistic aspects of this resistance. Candida auris is a globally emerging and often pandrug-resistant fungal pathogen that drives resistance primarily through upregulation of the CDR1 efflux gene, which leads to broad azole resistance across most clinical isolates (Cortegiani et al., 2018; Iyer et al., 2020). Evidence shows that polyphenolic efflux pump inhibition can restore azole activity against C. auris by inhibiting this very specific efflux mechanism, establishing a proof-of-concept that polyphenolic efflux pump inhibition strategies can be generalised to the fungal domain (Iyer et al., 2020). EGCG interferes with folic acid metabolism and disrupts hyphal growth and biofilm formation, two key virulence determinants (Navarro-Martínez et al., 2006; Lee et al., 2018). The TR34/L98H CYP51A mutation is a panazole resistance determinant with a concerning clinical prevalence in this context (Snelders et al., 2008). Polyphenol combinations can be used to provide complementary membrane-targeting strategies to azole sensitisation in this context. Polyphenol-mediated biofilm disruption strategies similar to those demonstrated to be effective against bacterial biofilms may be effective against Candida biofilms (Ramage et al., 2012; Taff et al., 2013; Rangel-Castañeda et al., 2019; Fakhim et al., 2025; Paiva Macedo and Dias, 2024). The key antifungal resistance mechanisms through which polyphenols act are shown in Figure 5.

3.3 Synergistic potentiation in polyphenols as β-lactam and fluoroquinolone adjuvants

The most clinically tractable use of polyphenols is unlikely to be as antibiotics in their own right; their stand-alone MICs are, for the most part, too high to be reached at an infection site by oral dosing. The more defensible proposition, and the one the experimental record actually supports, is that polyphenols behave as adjuvants: compounds with modest activity on their own that nonetheless hand a failing antibiotic back its potency. This is the logic that first brought clavulanic acid into clinical use alongside amoxicillin, and a recent survey of natural-product β-lactam potentiators places flavonoids, tannins, and phenolic acids within that same tradition, sorting their effects into a small set of recurring routes: efflux pump inhibition, alteration of membrane permeability, biofilm disruption, PBP2a interference, and direct β-lactamase inhibition (Wang, 2026). Those routes are worth taking one at a time, because the evidence behind each differs in quality, and treating them as interchangeable obscures what is doing the work in any given pairing. Figure 6 maps the four routes most relevant to bacterial synergy onto the checkerboard pattern the literature returns most consistently.

The Gram-negative outer membrane is the obvious place to begin, since it disqualifies whole antibiotic classes before pharmacology is even at issue. When epigallocatechin gallate (EGCG) was screened against seventy carbapenem-resistant Acinetobacter baumannii isolates, subinhibitory concentrations pulled the meropenem and carbenicillin MICs back into the susceptible range across the entire collection, with fractional inhibitory concentration indices below 0.5 and more than a two-log kill within 12 hours; tellingly, the potentiation tracked with efflux interference rather than β-lactamase output and held irrespective of each isolate’s resistance profile (Lee et al., 2017). The same access-granting effect surfaces when two polyphenols are paired with one another: curcumin, effectively inert alone against MDR A. baumannii at more than 256 μg/mL, turned bactericidal once EGCG was supplied, its MIC dropping three-to seven-fold to as little as 4 μg/mL with a four-to five-log reduction at 24 h (Betts et al., 2014). What permeabilisation buys, in both cases, is arrival: the drug reaches a target it could not previously approach.

Efflux blockade is the second route, and in gram-positives it has been traced to a named pump. Tannic acid suppresses norA transcription in Staphylococcus aureus through the ArlRS sensor pathway while also engaging the NorA protein directly, a dual action resolved by ethidium-accumulation fluorimetry, checkerboard testing, and in silico docking that converged on the same target (Tintino et al., 2023). Quercetin acts on the same staphylococcal machinery, raising intracellular ciprofloxacin by binding NorA at its substrate site and lowering the working concentration of co-administered fluoroquinolones (Hirai et al., 2010). Shutting the pump down at the level of transcription and obstructing it at the level of protein together restore the drug retention that overexpressed NorA had abolished, which is why efflux inhibition reads, in the clinic, as a return of susceptibility rather than as fresh killing.

Direct β-lactamase inhibition is the route most often invoked and the least firmly established for polyphenols. The premise is simple enough neutralise the hydrolytic enzyme and the antibiotic survives long enough to reach its target but until recently the support was largely computational. That gap has started to close: quercetin, kaempferol, and caffeic acid bind the clinically important carbapenemases KPC-2, NDM-1, and OXA-48 in docking and molecular-dynamics work, and the same compounds measurably curtailed β-lactam hydrolysis and potentiated carbapenems against Klebsiella pneumoniae, Escherichia coli, and Pseudomonas aeruginosa in checkerboard assays (Mourabiti et al., 2025) (Figure 6). A review now catalogue phenolics among candidate enzyme inhibitors alongside efflux blockers (Zai et al., 2025), but for most individual polyphenols the claim still rests on binding predictions rather than steady-state kinetics, and should be weighed accordingly.

The MRSA β-lactam axis is where the mechanistic record is most instructive, because the early reading was wrong and was later put right a reminder that an observed synergy and an understood mechanism are separate claims. Two decades ago, less than 25 μg/mL EGCG was shown to reverse high-level β-lactam resistance in MRSA, restoring benzylpenicillin, oxacillin, methicillin, ampicillin, and cephalexin with FIC indices between 0.126 and 0.625 across twenty-five isolates; the original authors attributed this to direct peptidoglycan binding and explicitly ruled out an effect on PBP2’ production (Zhao et al., 2001). The decisive follow-up reframed the phenomenon entirely: epicatechin gallate (ECg) intercalates into the cytoplasmic membrane and delocalises PBP2 from the FtsZ-anchored septal division machinery, breaking the PBP2–PBP2a partnership on which MRSA depends to keep building wall after β-lactams have acylated its native PBPs, with no evidence that ECg binds or inhibits PBP2a itself (Bernal et al., 2010). Earlier architectural work by the same group had already recorded ECg thickening the wall and loosening cross-linking without binding peptidoglycan or lowering PBP2a expression, foreshadowing the membrane-centred account (Stapleton et al., 2007). The practical reading of the “PBP2a downregulation” node in Figure 6 is therefore functional displacement of its partner enzyme, not transcriptional silencing a distinction the schematic keeps explicit so as not to overstate the data. Read together, the four routes are complementary rather than redundant: permeabilisation decides whether a Gram-negative drug arrives, efflux blockade decides whether it stays, β-lactamase inhibition decides whether it survives the periplasm, and the membrane-and-divisome effects decide whether a β-lactam can finally act once inside. Depending on which defences a given strain leans on, the reported MIC reductions run from a modest two-fold to as much as 256-fold, the upper figure set by collections such as the carbapenem-resistant A. baumannii isolates in which EGCG collapsed MICs to ≤1 μg/mL (Lee et al., 2017). That breadth is the point: a single adjuvant can convert an untreatable isolate into a treatable one not by being a better antibiotic, but by dismantling, in parallel, the several defences the organism had assembled.

3.4 Limitations in bioavailability, pharmacokinetics, and preclinical data interpretation

The most significant obstacle to the clinical translation of polyphenol-based antimicrobial agents is their notoriously poor systemic bioavailability. After oral intake, the majority of polyphenols do not undergo significant intestinal absorption because of their inherent insolubility in water and chemical instability at physiological pH (Manach et al., 2004; Williamson, 2017). Absorbed metabolites undergo rapid and extensive first-pass conjugation in the intestinal epithelium and liver glucuronidation, sulfation, and methylation, generating polar metabolites that circulate at subtherapeutic levels and are rapidly excreted through the kidneys (Manach et al., 2004; D'Archivio et al., 2007; Williamson and Clifford, 2017).

A major and often overlooked deficiency in this body of literature is the lack of understanding of the difference between the concentrations that are used in vitro and the actual levels that are achievable in human tissues. The minimal inhibitory concentrations (MICs) and synergy data in the studies summarised in this review were largely measured at polyphenol concentrations between 50 and 200 μg/mL. However, peak plasma levels after the typical oral dose of quercetin, curcumin or EGCG in human pharmacokinetic studies have not exceeded 1–3 μM even when administered under enhanced bioavailability conditions (Manach et al., 2004; Williamson, 2017). This is not a minor discrepancy. This translates to the fact that, based on the available data, the concentrations of these compounds that are directly antimicrobial are not likely to be reached in systemic infection sites by oral dosing alone. The mechanistic conclusions do not preclude the fact that polyphenols might be clinically relevant at sub-MIC concentrations as an adjunctive agent or a sensitising agent, as their efflux pump inhibitory, quorum sensing-disruptive, and antibiofilm properties may be relevant to in vivo levels (Betts et al., 2014; Farha and Brown, 2019). Topical, intraluminal and inhalational formulations are potential clinical applications where local concentrations can more closely resemble those shown to be effective in vitro, for gastrointestinal and respiratory pathogens respectively (Negut et al., 2018; Eleraky et al., 2020). Nanoparticle encapsulation and phospholipid-based phytosome delivery systems have shown enhanced bioavailability in animal models, but there are limited and poor quality human pharmacokinetic–pharmacodynamic data as yet, and these must be improved before meaningful clinical translation can be recommended (Cuomo et al., 2011; Rahbardar et al., 2025).

This gap in systemic bioavailability is the most vigorous translational frontier in polyphenol studies. Lipid-based nanoparticles (liposomes, solid lipid nanoparticles) are compelling solutions, as they increase the solubility of compounds and prevent the degradative metabolism of vulnerable phenolic structures; polymeric nanoparticles (PLGA, chitosan) provide sustained and controlled release, and phytosome technology and phospholipid complexation of polyphenols have been shown to increase the bioavailability of curcumin by 29-fold relative to that of unconjugated materials (Cuomo et al., 2011; Deepika et al., 2022; Elkhalifa et al., 2024). High-pressure homogenisation to prepare EGCG-loaded solid lipid nanoparticles results in extended release profiles under both simulated gastric (pH 1.2) and intestinal (pH 6.8) conditions, which suggests the feasibility of release into the gastrointestinal tract (Elkhalifa et al., 2024). Even in the face of these advances, however, the present scene of clinical trials in this field is dominated by a predominantly anti-inflammatory and metabolic focus rather than systemic antimicrobial end points. Directed clinical trials are urgently needed in this area, starting with topical and wound-healing applications where high concentrations of polyphenols can be achieved, followed by proof-of-concept combination adjuvant studies (Brown et al., 2024).

Another analytical difficulty relates to the determination of often high MICs of polyphenols in isolation. Although the values of MICs in the hundreds of µg/mL range may not seem favourable compared to those of conventional antibiotics, this measure is of limited interest when polyphenols are used as efflux pump inhibitors or antivirulence agents. In these applications, the relevant efficacy parameter is the EPIC 50 (concentration to achieve 50% inhibition of efflux pumps) or the QSIC 50 (quorum sensing inhibitory concentration 50), which is often frequently achievable at subinhibitory concentrations and in topical application (which generates local drug concentrations orders of magnitude greater than those achievable systemically) (Drusano, 2004; Kırmusaoğlu, 2019; Liapikou et al., 2015; Shah et al., 2024).

3.4.1 Limitations of the evidence base

There are a number of limitations parameters to consider when interpreting the findings of this review. The literature search was limited to three English-language databases (PubMed, Scopus, and Web of Science; 2013–2025), and may have missed relevant studies that were published in other languages and/or grey literature repositories. There was no formal instrument used to grade the strength of the included primary studies and thus, the results from rigorously designed, randomisation-based studies and from single concentration endpoint assays have equal weight in a number of subsections, and the lack of a formal grading of evidence is itself a potential source of selection bias, as is well recognised in natural-product pharmacology (Williamson, 2017; Farha and Brown, 2019). The mechanistic picture presented in this review is overwhelmingly based on in vitro data and, to a lesser extent, in vivo models, and the limited human clinical data has not yet validated the plausibility of many interactions described at tissue-relevant polyphenol concentrations. The polyphenol class is structurally heterogeneous, ranging from simple phenolic acids to oligomeric proanthocyanidins, which further complicates generalisation across the class, and readers are advised to interpret such statements as provisional hypotheses and not definitive statements. These restrictions emphasize the need for pre-registered systematic reviews with explicit grading of quality and the need for primary research to help close the gap between in vitro mechanistic studies and clinical use.

4 Conclusion

Plant-based polyphenols are scientifically compelling, structurally diverse, and mechanistically sophisticated armamentaria against the global crisis of multidrug-resistant superbugs. The evidence synthesised in this review is beyond doubt: these compounds disrupt bacterial and fungal pathogens through a polypharmacological approach that both compromises cell envelope integrity and blocks essential intracellular enzymes and metabolic pathways, generates lethal ROS through pro-oxidant and metal-chelating mechanisms, silences pathogen communication via quorum sensing and blocks biofilm communities at multiple stages of development.

4.1 Prospective study directions

More importantly, the antivirulence and effect of polyphenols are that they act at subinhibitory concentrations to disarm bacteria without killing them, and the antivirulence and effect of polyphenols offer a therapeutic modality that may exert significantly weaker selective pressure on the evolution of resistance than bactericidal drugs do. This time benefit may prove critical in the control of chronic infections and in safeguarding the effectiveness of current antibiotics in the clinical pipeline. To achieve this goal, future research should immediately address a number of interrelated priorities. To fully map the intracellular effects of polyphenol-treated MDR pathogens and to develop new targets, the mechanism needs to be expanded through transcriptomic, proteomic, and metabolomic profiling of polyphenol-treated MDR pathogens. Structural–activity relationship (SAR) investigations and medicinal chemistry optimisation of lead polyphenol scaffolds are needed to increase the intrinsic potency, metabolic stability, and selectivity. The creation of novel advanced nanoformulation delivery systems, such as targeted liposomes, chitosan nanoparticles, and phytosomes, must be expedited and tested in robust animal and organoid infection models that faithfully recapitulate the clinical dynamics of infection. Standardised methodological guidelines for MIC determination, time-kill kinetics, efflux pump inhibition assays, and biofilm quantification are urgently needed to facilitate cross-laboratory comparisons and meta-analyses. Finally, the transition to in vitro evidence to controlled clinical trials needs to occur, starting with topical applications where pharmacokinetic restrictions are minimal and moving to adjuvant combination regimens against resistant nosocomial pathogens. The stakes were not high. In the absence of transformative therapeutic innovation, the global burden of antimicrobial resistance will continue its catastrophic course. In terms of scientific rigour and pharmaceutical sophistication, polyphenols provide a true and renewable source of novel molecular tools to help meet this challenge. The convergence of mechanistic knowledge, nanotechnology, and precision medicine places these ancient plant compounds at the center of 21^st century antimicrobial strategies.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Argyrios Periferakis, Carol Davila University of Medicine and Pharmacy, Romania

Reviewed by: Lamprini Troumpata, Carol Davila University of Medicine and Pharmacy, Romania

Iosif-Aliodor Timofticiuc, Carol Davila University of Medicine and Pharmacy, Romania

Ugonna Cassandra Aniokete, David Umahi Federal University of Health Sciences, Nigeria

Footnotes

AI, Autoinducer; AHL, N-acyl homoserine lactone; AMR, Antimicrobial Resistance; ATP, Adenosine triphosphate; CDR1, Candida Drug Resistance 1 (efflux gene); CSH, Cell surface hydrophobicity; DNA, Deoxyribonucleic acid; eDNA, Extracellular DNA; EGCG, Epigallocatechin gallate; EPI, Efflux pump inhibitor; EPS, Extracellular polymeric substances; ESKAPE, Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter species; FASII, Fatty-acid synthesis pathway II; FICI, Fractional Inhibitory Concentration Index; GP, Generalised polarisation; H₂O₂, hydrogen peroxide; LPS, Lipopolysaccharide; MDR, Multidrug-resistant; MFS, Major Facilitator Superfamily; MIC, Minimum inhibitory concentration; MRSA, Methicillin-resistant Staphylococcus aureus; NADH, Nicotinamide adenine dinucleotide; OM, Outer membrane; PAC, Proanthocyanidin; PBP2a, Penicillin-binding protein 2a; PIA, Polysaccharide intercellular adhesin; PK/PD, Pharmacokinetic/Pharmacodynamic; PMF, Proton motive force; QS, Quorum sensing.