Reversal of tetracycline resistance by clove and peony extracts in a multi-drug resistant Escherichia coli

TL;DR

This study found that clove and peony extracts can reverse antibiotic resistance in a multi-drug resistant Escherichia coli strain.

Contribution

The study identifies clove and peony extracts as potential agents to reverse tetracycline resistance in drug-resistant E. coli.

Findings

Clove and peony extracts increased sensitivity to tetracycline and tobramycin in a multi-drug resistant E. coli strain.

Myricetin, a compound from peony, enhanced tetracycline sensitivity without causing membrane permeability or toxicity in mammalian cells.

Abstract

Infections with Gram-negative bacteria that are resistant to multiple antibiotics are increasingly prevalent and challenging to treat. We sought to identify compounds with potential to reverse resistance to tetracycline and tobramycin in a multi-drug resistant isolate of Escherichia coli (NCTC 13400). A screen of 800 extracts of traditional herbs and medicines revealed that polar extracts of cloves (Syzygium aromaticum) and Peony flowers (Paeonia lactiflora) significantly increased sensitivity to both antibiotics in this strain. Fractionation of clove and peony extracts by high performance liquid chromatography revealed activity within the fractions comprising mainly phenolic acids and flavonoids, respectively. Sampling candidate compounds from these fractions revealed that while no tested compound enhanced the activity of tobramycin, myricetin significantly increased the sensitivity of this strain to tetracycline. Myricetin alone inhibited growth, but was not bactericidal and did not induce membrane permeability or inhibit Nile red efflux. In vitro MTT assays using mammalian HEK-293 cells revealed no significant toxicity of myricetin, or most of the other tested compounds, at up to 100 µM. The results suggest that the screening of plant secondary metabolites has potential for the identification of antibiotic resistance reversers with potentially favourable in vitro toxicity profiles.

Figures

Primary screen of 800 extracts of traditional medicines for potential to enhance antibiotic function in a multi-drug resistant *E. coli*. *E. coli* NCTC 13,400 cells were cultured in 0.25x MIC tetracycline (a) or 0.25x MIC tobramycin (b), with 200 µg/ml of each of 800 plant extracts (individual dots), or 2% DMSO as vehicle control. Growth at 8 h is shown as background corrected absorbance normalised to DMSO control

Dose response of top hits from primary screen for antibiotic potentiation. *E. coli* NCTC 13,400 was cultured with tetracycline (a) or tobramycin (b) from 4 µg/ml to 256 µg/ml, together with a fixed concentration of DMSO (2%) or indicated plant extract (200 µg/ml) for 21 h. Background corrected absorbance at 600 nm was normalised to untreated cultures. Means ± SD of 3 independent experiments are shown

Checkerboard assays for potentiation of tetracycline by clove and peony extracts. *E. coli* NCTC 13,400 was cultured with 8 different concentrations of tetracycline (0 µg/ml to 256 µg/ml) and 8 different concentrations (0% to 2%) of DMSO (a), clove extract (b) or peony extract (c). Means of 3 independent experiments are shown

HPLC and activity-guided separation of clove and peony extracts. Clove (a) and peony (b) extracts were separated by reverse-phase HPLC using a linear gradient from water with 0.1% formic acid to acetonitrile with 0.1% formic acid, over 1 h. Eluted fractions (1–19) were collected at 3 min intervals and added to *E. coli* cultures with 16 µg/ml (0.25x MIC) tetracycline and cultured at 37 °C for 21 h. Upper chart indicates UV absorbance of column eluate. Lower chart indicates bacterial growth as measured by background-corrected absorbance at 600 nm. Means ± SD of 3 experiments are shown

Potential mechanisms of antibacterial activity of myricetin. (a) Bacteria were cultured without antibiotic and with 256 µM of indicated phytochemicals. Absorbance at 600 nm was measured hourly to 12 h. (b) Time kill assay of bacteria grown with 2.56% DMSO, 256 µM myricetin, 256 µg/ml tobramycin or 256 µg/ml tetracycline (the latter two as positive controls for cell killing). (c) Bacteria were cultured in 100 µg/ml PI with 256 µg/ml plant extract, 256 µM myricetin, 2.56% DMSO or 0.1% SDS. Increases in fluorescence indicate increase in bacterial cell permeability. (d) Efflux of the dye Nile red

Effects of candidate compounds on mammalian cell viability. HEK-293 cells were cultured overnight with 0 to 100 µM of each candidate natural compound, or equivalent content of DMSO as vehicle control. Cell viability was then assessed by MTT assay, with values normalised to cells cultured in the absence of compound or DMSO. Means ± SD of 3 independent experiments are shown. * *p* < 0.05, *p* < 0.01 vs. control condition (0 µM) by ANOVA with Tukey’s post-hoc test

Full text

Introduction

Gram-negative bacteria displaying resistance to multiple classes of antibiotics are an increasing threat to global healthcare systems (Reza et al. 2025). As the discovery of new classes of antibiotic has slowed in recent years, interest is growing in the development of molecules that block specific mechanisms of resistance, with the aim of restoring sensitivity of resistant microbes to existing antibiotics (Elkady et al. 2025). An example of the successful application of this approach is the use of clavulanic acid to reverse resistance to beta lactam antibiotics (Laws et al. 2019). Laboratory studies have also identified promising candidates with potential to target other resistance mechanisms in Gram-positive bacteria (Harikumar et al. 2025; Rampacci et al. 2023). However, comparatively little progress has been made to target non-lactamase resistance mechanisms in Gram-negative bacteria (Norouzalinia et al. 2025).

The screening of natural product libraries has historically been a highly successful approach to the development of new drugs. Until 1990, approximately 80% of new drugs were derived from natural products, with examples including the statins, several key anti-cancer drugs and most existing antibiotics (Li et al. 2009). Plants in particular may have utility in this context as they have succesfully countered the threat of Gram-negative infection for millennia through their production of antibacterial secondary metabolites (Mattingly et al. 2020). Recent work has also shown that the resistance genes prevalent today evolved millions of years ago, within the antibiotic producing organisms and their competitors (Perry et al. 2016). This raises the possibility that plants may have faced evolutionary pressure to express secondary metabolites that block the same resistance mechanisms encountered in the clinic today.

We therefore sought to identify compounds with capacity to reverse resistance of a multi-drug-resistant Escherichia coli (NCTC 13400) to two model antibiotics - tetracycline and tobramycin. As it is prudent to prioritise molecules that are likely to exhibit low toxicity during drug development, a natural product library comprising 800 extracts of traditional herbs or medicines with a history of safe oral use in man was chosen for screening. Candidate compounds from hit extracts showing antibacterial activity were then taken forward for dose response, in vitro toxicity and mechanistic studies.

Methods

Bacterial strains

E. coli NCTC 13,400 (Public Health England) was chosen for study as it is highly resistant to diverse antibiotic classes and is a common cause of urinary tract infections (Woodford et al. 2009). Resistance in this strain is mediated mainly by the pEK-499 plasmid, which carries 10 resistance genes. The control strain, lacking resistance to any of the tested antibiotics, was E. coli DH5α.

Primary screen for plant extracts with capacity to reverse antibiotic resistance

Tetracycline and tobramycin were chosen as the model antibiotics for study, since they are non-lactam antibiotics of different classes, and the pEK-499 plasmid confers resistance to them via two distinct mechanisms: tetA encodes an efflux plump for tetracycline, and aac(6’)-Ib encodes an aminoglycoside acetyltransferase that inactivates tobramycin (Woodford et al. 2009). The natural product library comprised 800 extracts (400 polar and 400 non-polar) of traditional herbs or medicines with a history of oral use in man, at a concentration of 10 mg/ml in DMSO (Phytotitre collection, Caithness Biotechnologies, UK).

For the initial screening assay, E. coli NCTC 13,400 cells were diluted in luria broth (LB) to ~ 10^6^ CFU/ml, and supplemented with 16 µg/ml tetracycline, or 8 µg/ml tobramycin (equivalent to one quarter of the MIC of each antibiotic for this strain). Suspensions were then treated with 200 µg/ml of each extract or 2% DMSO (on 8 wells per plate as vehicle controls). Absorbance was measured at 600 nm using a microplate reader (Tecan) immediately after challenge (t = 0) and after 8 h and 21 h incubation at 37 °C. Background absorbance values (at t = 0) were subtracted from the 8 h and 21 h values to account for interference from plant pigments in some extracts. Growth was then normalised to the mean of the vehicle only controls on the same plate.

Broth microdilution and checkerboard assays

For dose-response assays, E. coli NCTC 13,400 cells were diluted to ~ 10^6^ CFU/ml in LB without antibiotic, and challenged with doubling dilutions of plant extract or antibiotic from 256 µg/ml to 4 µg/ml, in 96-well plates. Absorbance at 600 nm was measured immediately after plating, and again after 21 h culture. For checkerboard assays, 8 concentrations of extract or compound, and 8 concentrations of tetracycline (including zero) were arranged across 64 wells of a 96-well microplate in a total volume of 150 µl. Absorbance at 600 nm was measured at 0 h and 21 h post incubation. For measurement of growth kinetics in the absence of antibiotics, bacteria were plated with 256 µM of each phytochemical and cultured at 37 °C in a BMG Fluostar microplate reader. Absorbance at 600 nm was recorded every hour until 12 h post plating. Baseline absorbance was subtracted from each measurement to account for interference from coloured compounds. Antibiotic concentrations are reported in µg/ml and test compounds in µM, following standard nomenclature for these agents.

Activity-guided fractionation of extracts

Polar extracts of cloves (Baldwins, UK, origin Indonesia) and peony petals (Baldwins, UK, origin Bulgaria), were prepared by grinding 10 g material briefly in 100 ml boiling water, allowing to cool over several hours then steeping at 4 °C overnight. The suspension was filtered, freeze dried and resuspended at 10 mg/ml in DMSO. 40 µl of each extract was then loaded onto a reverse-phase high performance liquid chromatography (HPLC) C18 column (Phenomenex, UK) using a Perkin Elmer series 200 HPLC system. Extracts were then subjected to linear gradient HPLC, transitioning from water with 0.1% formic acid (A) to acetonitrile with 0.1% formic acid (B), over 1 h. Eluted fractions were collected at 3 min intervals then dried in a speedvac before resuspension in 40 µl DMSO. 8 µl of each resuspended fraction was then added to 142 µl E. coli culture containing 16 µg/ml (0.25x MIC) tetracycline, and cultured overnight in 96-well plates. Absorbance at 600 nm was measured after 0 h and 21 h culture.

Well diffusion assays

A 5 mm straw was used to remove plugs from LB agar plates with no antibiotic selection, before streaking the plates with bacteria at ~ 10^6^ CFU/ml. 25 µl of DMSO or plant extract (at 10 mg/ml), with or without 5 µl antibiotic (at 10 mg/ml), were pipetted into respective wells. Zones of clearance around each well were measured after overnight incubation at 37 °C.

Time kill assays

Bacteria were resuspended in phosphate-buffered saline (PBS) to an absorbance of 0.5 at 600 nm, then supplemented with 2.56% DMSO, 256 µM myricetin, 256 µg/ml tobramycin or 256 µg/ml tetracycline (chosen as positive controls for bactericidal activity). Samples of each culture were removed at 0, 4 and 24 h. 100 µl of each sample diluted 1:10,000 or 1:1,000,000 in PBS were then plated onto LB agar plates without antibiotic selection. Colonies were counted after overnight incubation at 37 °C.

Cell permeability and efflux assays

For cell permeability assays, bacteria were resuspended in PBS as for time kill assays and supplemented with 100 µg/ml propidium iodide (PI). Baseline fluorescence (excitation 544 nm, emission − 620 nm) was measured every 20 s for 4 min using a BMG Fluostar instrument. Wells were then treated with 256 µg/ml plant extract, 256 µM myricetin, 2.56% DMSO or 0.1% sodium dodecyl sulphate (as positive control), before recording fluorescence for a further 30 min. Efflux assays were performed as described previously (Jenic et al. 2020). Briefly, bacterial cells were de-energised in PBS containing 1 mM MgCl_2_ and 10 µM carbonyl cyanide m-chlorophenylhydrazone (CCCP), before labeling with 5 µM Nile red for 3 h. Cells were then washed and treated with 2.56% DMSO, 256 µM myricetin or 25.6 µM CCCP (as positive control for efflux inhibition). Fluorescence (excitation 544 nm, emission − 620 nm) was measured every 15 s for 2 min, before supplementation with 50 mM glucose and measurement for a further 20 min.

In vitro assay for mammalian cell toxicity

HEK-293 cells (ECACC 85120602) were plated at 4 × 10^4^ cells/well of 96-well plates and cultured overnight before challenge with 0 to 100 µM of each candidate natural compound, equivalent content of DMSO as vehicle control, or 0.1% sodium dodecylsulphate (SDS, a positive control for toxicity). After culture for 24 h, medium was gently replaced with DMEM / 10% FCS containing 2.5 µM 3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) for 4 hours. Formazan crystals were dissolved overnight by supplementing each well with 5% SDS. Absorbance was measured at 600 nm, and cell viability is expressed as a percentage of the absorbance of cells cultured in medium alone.

Statistical analyses

Hits were defined as the top seven extracts ranked by average enhancement of both tobramycin and tetracycline at the 8 h and 21 h timepoints. One-way ANOVA with Tukey’s post-hoc test was used for dose-response analyses, and for other experiments where more than two groups were compared. Statistical significance was defined as p < 0.05.

Results

Results of primary screen for extracts with antibiotic-anhancing properties

None of the tested extracts fully prevented growth of E. coli NCTC 13,400 in the presence of 0.25x MIC tetracycline or tobramycin at the 8–21 h timepoints. However, several extracts partially inhibited growth, particularly at the 8 h timepoint (Fig. 1). A correlation was seen between inhibition of growth in both the tetracycline and tobramycin experiments across all extracts (r = 0.595, P < 0.001), suggesting potential for existence of non-specific resistance reversal or sensitising mechanisms. Interestingly, this was in spite of a much weaker correlation existing between the inhibitory effects of extracts on growth of E. coli NCTC 13,400 in the presence of antibiotics, and the effects of the same extracts on a non-resistant control strain (E. coli DH5α) grown in the absence of antibiotics (r = 0.114, P = 0.022), suggesting that the extracts may be inhibiting growth in the 13,400 strain partly via resistance reversal mechanisms.

Fig. 1. Primary screen of 800 extracts of traditional medicines for potential to enhance antibiotic function in a multi-drug resistant E. coli. E. coli NCTC 13,400 cells were cultured in 0.25x MIC tetracycline (a) or 0.25x MIC tobramycin (b), with 200 µg/ml of each of 800 plant extracts (individual dots), or 2% DMSO as vehicle control. Growth at 8 h is shown as background corrected absorbance normalised to DMSO control

Several extracts stood out in terms of their potential to enhance the inhibtory properties of both tetracycline and tobramycin. When ranking extracts by average inhibition of growth across both antibiotics and timepoints (% inhibition shown in parentheses), the top hits were: Bayberry bark (32%), Rosemary (32%), Peony flower (31%), Rose bud (30%), Rose petal (30%), Witchazel bark (26%) and Willow bark (26%). Clove extract (19%) was also taken forward for further study since both polar and non-polar extracts of the same plant appeared among the runner-up hits, in spite of no correlation existing between inhibitory activities of polar and non-polar extracts more broadly across the whole library (r = 0.032, p = 0.530).

Replication of hit extracts by dose-response assays

The top seven hits were taken forward to test if their effects on MIC of tetracycline and tobramycin could be replicated in dose-response assays. Most of the tested extracts reduced the maximum OD reached by the cultures at 21 h compared to vehicle control (Fig. 2). Rose bud, rose petal and peony flower extracts significantly lowered the MIC of tetracycline from 64 µg/ml (with DMSO vehicle control) to 32 µg/ml (P = 0.036), and clove extract lowered the MIC of tetracycline to 16 µg/ml (P < 0.001, Table 1). The MIC of tobramycin was also significantly lowered from 32 µg/ml to 16 µg/ml by extracts of rosemary and clove (P = 0.032). The IC_50_ of tetracycline was significantly lowered by bayberry, peony and clove extracts, and that of tobramycin by rosemary, bayberry, peony and clove extracts (Table 1). As the most consistent enhancers were clove and peony extracts, particularly with respect to tetracycline, these combinations were taken forward for further analysis. Well diffusion assays showed very modest zones of inhibition for clove and peony extracts alone, but neither extract significantly increased the zone of clearance around the tetracycline or tobramycin wells (Table 2). However, checkerboard assays confirmed dose-dependent effects for both extracts on reversal of resistance to tetracycline in the more sensitive broth microdilution assay (Fig. 3).

Fig. 2. Dose response of top hits from primary screen for antibiotic potentiation. E. coli NCTC 13,400 was cultured with tetracycline (a) or tobramycin (b) from 4 µg/ml to 256 µg/ml, together with a fixed concentration of DMSO (2%) or indicated plant extract (200 µg/ml) for 21 h. Background corrected absorbance at 600 nm was normalised to untreated cultures. Means ± SD of 3 independent experiments are shown

Table 1. Replication of hit extracts for sensitisation to tetracycline and tobramycinTetracyclineTobramycinMIC(µg/ml) ^a^p vs. DMSOIC_50_(µg/ml)p vs. DMSOMIC(µg/ml)p vs. DMSOIC(µg/ml)p vs. DMSODMSO64-16-32-7-Rose buds320.036151.000321.000100.695Rose petals320.03680.385321.00081.000Willow510.940151.000321.00080.987Rosemary400.317120.977160.032< 40.035Witchazel510.940140.999321.00061.000Bayberry641.000< 4< 0.001321.000< 4< 0.001Peony320.036< 4< 0.001321.000< 4< 0.001Clove160.001< 4< 0.001160.032< 4< 0.001^a^ Mean MIC of 3 independent experiments is shown

Fig. 3. Checkerboard assays for potentiation of tetracycline by clove and peony extracts. E. coli NCTC 13,400 was cultured with 8 different concentrations of tetracycline (0 µg/ml to 256 µg/ml) and 8 different concentrations (0% to 2%) of DMSO (a), clove extract (b) or peony extract (c). Means of 3 independent experiments are shown

Table 2. Well diffusion assays of clove extract, peony extract, and myricetinTreatment ^a^ClovePeonyMyricetinC/E alone6.7 ± 2.16.0 ± 1.77.0 ± 1.0Tet + DMSO12.3 ± 2.111.7 ± 2.912.7 ± 1.2Tet + C/E12.3 ± 1.213.3 ± 3.513.7 ± 1.5Tob + DMSO23.7 ± 1.522.0 ± 2.025.0 ± 2.6Tob + C/E28.3 ± 2.524.0 ± 5.029.0 ± 2.0^a^ Mean diameter of inhibition in mm from 3 independent experiments is shown, well diameter was 5 mm. Abbreviations: C/E - Respective compound or extract, Tet - Tetracycline, Tob - Tobramycin

Fractionation and polarity assessment of active compounds

We hypothesised that the hit extracts may contain compounds with potential for further development as antibiotic enhancers. Fractionation of clove extract by HPLC revealed early elution of one or more highly polar constituents with inhibitory activity against E. coli NCTC 13,400 grown in the presence of 0.25x MIC tetracycline (Fig. 4A). Peony flower extract separated in the same way potentiated tetracycline in the mid-polarity fractions (Fig. 4B). Previous reports suggest that the dominant compounds in the early eluting fraction from cloves separated in this manner are phenolic acids, particularly gallic acid (Akbar et al. 2012; Tashtoush et al. 2016), and compounds eluting in the mid polarity fractions of peony flower comprise mainly flavonoids and their glycosides, such as kaempferol, myricetin, rutin and luteolin (Čutović et al. 2022).

Fig. 4HPLC and activity-guided separation of clove and peony extracts. Clove (a) and peony (b) extracts were separated by reverse-phase HPLC using a linear gradient from water with 0.1% formic acid to acetonitrile with 0.1% formic acid, over 1 h. Eluted fractions (1–19) were collected at 3 min intervals and added to E. coli cultures with 16 µg/ml (0.25x MIC) tetracycline and cultured at 37 °C for 21 h. Upper chart indicates UV absorbance of column eluate. Lower chart indicates bacterial growth as measured by background-corrected absorbance at 600 nm. Means ± SD of 3 experiments are shown

A series of candidate compounds reported to be present in similarly prepared fractions of clove and peony extracts (Akbar et al. 2012; Čutović et al. 2022; Tashtoush et al. 2016) were then tested for their effects on the MIC and IC_50_ of tetracycline and tobramycin in E. coli NCTC 13,400. None of the compounds had a significant effect on sensitivity to tobramycin (Table 3). However, myricetin significantly lowered the MIC and IC_50_ of tetracycline (Table 3). A checkerboard assay using a higher maximum concentration of myricetin (512 µM) gave a minimum fractional inhibitory concentration index (FICI) value of 0.75, suggesting existence of an additive interaction. This appears to be a class-specific effect as myricetin had no effect on the MIC or IC_50_ of ampicillin or chloramphenicol (data not shown).

Table 3. Effects of candidate natural compounds on sensitisation to tetracycline and tobramycinTetracyclineTobramycinMIC(µg/ml)p vs. DMSOIC50(µg/ml)p vs. DMSOMIC(µg/ml)p vs. DMSOIC50(µg/ml)p vs. DMSODMSO ^a^64.0-24.2-32.0-< 4-Caffeic acid64.01.00021.20.73126.70.9724.50.998Chlorogenic acid64.01.00021.30.76232.01.0004.70.981Gallic acid53.30.94020.50.49826.70.972< 41.000Kaempferol85.30.94025.11.00032.01.000< 41.000Luteolin64.01.00022.70.99032.01.0005.90.798Myricetin32.00.03614.30.00121.30.4875.60.752Quercetin64.01.00021.70.87732.01.000< 41.000Rutin42.70.31721.20.73124.00.8724.60.974^a^ Natural compounds present at 256 µM, DMSO control at 2.56%

Potential mechanisms of anti-bacterial activity of myricetin

Myricetin significantly slowed the rate of growth of E. coli NCTC 13,400 over 12 h when cultured in the absence of antibiotic (P < 0.01, Fig. 5A). Significant growth inhibition was also seen, albeit to a lesser extent, with kaempferol, luteolin and rutin (P < 0.05). Time kill assays revealed that myricetin did not lower the number of viable CFU/ml within 24 h (Fig. 5B), and it also did not alter bacterial membrane permeability, as shown by PI uptake assays (Fig. 5C). Finally, efflux assays suggest that myricetin is not an inhibitor of efflux pumps that utilise Nile red as a substrate (Fig. 5D).

Fig. 5. Potential mechanisms of antibacterial activity of myricetin. (a) Bacteria were cultured without antibiotic and with 256 µM of indicated phytochemicals. Absorbance at 600 nm was measured hourly to 12 h. (b) Time kill assay of bacteria grown with 2.56% DMSO, 256 µM myricetin, 256 µg/ml tobramycin or 256 µg/ml tetracycline (the latter two as positive controls for cell killing). (c) Bacteria were cultured in 100 µg/ml PI with 256 µg/ml plant extract, 256 µM myricetin, 2.56% DMSO or 0.1% SDS. Increases in fluorescence indicate increase in bacterial cell permeability. (d) Efflux of the dye Nile red post-injection of glucose (arrow) was measured in the presence of 2.56% DMSO, 256 µM myricetin or 25.6 µM CCCP. Reduction in rate of decline of fluorescence indicates reduced efflux activity. Means of at least 3 experiments are shown

Mammalian cell toxicity

MTT assays using the human HEK-293 cell-line were used to gain preliminary insight into the in vitro toxicity of the test compounds. While most of the compounds (including myricetin) did not significantly lower the viability of HEK-293 cells by 24 h as measured by this assay, there were significant reductions in viability in response to 100 µM luteolin and rutin (Fig. 6).

Fig. 6. Effects of candidate compounds on mammalian cell viability. HEK-293 cells were cultured overnight with 0 to 100 µM of each candidate natural compound, or equivalent content of DMSO as vehicle control. Cell viability was then assessed by MTT assay, with values normalised to cells cultured in the absence of compound or DMSO. Means ± SD of 3 independent experiments are shown. * p < 0.05, ** p < 0.01 vs. control condition (0 µM) by ANOVA with Tukey’s post-hoc test

Discussion

The discovery of resistance reversing agents with low toxicity and the ability to target non-lactam antibiotics has been challenging to date, particularly in the context of treating Gram-negative bacterial infections (Elkady et al. 2025; Norouzalinia et al. 2025). A natural extract screening approach was chosen to address these difficulties for several reasons. First, we hypothesise that plants are likely to have evolved secondary metabolites as counter-measures against these bacterial defences. Second, the screening of natural product collections typically yields a higher hit rate in phenotypic assays than synthetic compound collections of equivalent size (some estimate up to 300x higher, Li et al. 2009). Third, a natural extract library offers a much greater chemical diversity than a similarly sized pure compound collection (Wilson et al. 2020). Finally, by screening a collection comprising mainly of traditional herbs and medicines with a history of oral use in man, the aim was to begin with molecules that are likely to exhibit favourable toxicity profiles, to help facillitate future steps towards translation.

It should be acknowledged that screening an extract collection presents some limitations in comparison to pure compound screens. In particular, plant pigments and related compounds commonly present in extracts can interfere with absorbance- and fluorescence-based assays (so-called ‘Pan-Assay Interference compounds’, PAINs) (Wilson et al. 2020). These effects were countered in the present study by using background subtraction of all absorbance measurements, and by seeking confirmation of bioactivity in dose-response and checkerboard assays. The identification of active compounds from hit extracts also requires activity-guided separation and structural studies, particularly if the hit plant is not well studied. Fortunately, the chemical composition of polar extracts of both clove and peony have been reported previously (Akbar et al. 2012; Čutović et al. 2022; Tashtoush et al. 2016), facillitating prompt progression to exploration of candidate bioactive constituents.

As none of the tested compounds enhanced the activity of tobramycin, the agents responsible for this activity in clove and peony extracts remain to be discovered. However, myricetin significantly potentiated the activity of tetracycline in E. coli NCTC 13,400. Some previous studies have reported other anti-microbial properties of this molecule. For example, myricetin showed synergistic interactions with numerous antibiotics against Helicobacter pylori (Krzyżek et al. 2021), Pseudomonas aeruginosa (Zeng et al. 2025) and Klebsiella pneumoniae (Lin et al. 2005). It was also reported to be synergistic with levofloxacin, but not other antibiotics, when screening a range of non-resistant Gram-positive and Gram-negative bacteria (Hanci et al. 2023). However, the present study appears to be the first to report myricetin’s ability to reverse resistance to tetracycline in a highly resistant E. coli.

The mechanisms of myricetin’s anti-microbial properties are little studied. One study used an in silico modelling approach to suggest that myricetin may bind to the FimX protein, a receptor for c-di-GMP that promotes biofilm formation in P. aeruginosa (Zeng et al. 2025). We found that myricetin significantly suppressed growth of E. coli NCTC 13,400, even in the absence of antibiotic, but it does not kill cells directly, nor alter cellular permeability or inhibit efflux pumps that utilise Nile Red as a substrate. Together, these observations suggest that myricetin displays growth inhibition rather than direct bactericidal activity. However, further work will be necessary to establish the mechanism by which myricetin increases the sensitivity of this strain to tetracycline.

Myricetin is abundant in many fruits and vegetables, particularly cranberries and spinach, and dietary intake of the molecule is estimated to be as much as 2.2 mg per day in European adults (Taheri et al. 2020). To our knowledge there have been no clinical trials of myricetin alone. However, one study using a combination pill containing multiple nutraceuticals, delivered 150 mg myricetin per day to 42 subjects and found no signs of toxicity over a 4 week study period (Abidov et al. 2006). These observations, together with the in vitro MTT assay data shown in the present study, suggest that myricetin is likely to be of low toxicity.

Other recent screens of natural product collections support their utility for antimicrobial discovery. For example, a recent screen of 419 pure natural compounds found several that reverse resistance to colistin and tobramycin in Gram-negative bacteria (Mattingly et al. 2020). Likewise, a larger screen of > 300,000 fractions of natural extracts found a ~ 1% hit rate with respect to direct antimicrobial activity in non-resistant Staphylococcus aureus and E. coli (Martínez-Fructuoso et al. 2023).

Overall, the present study further highlights the potential utility of natural extract screening for the discovery of compounds with potential to enhance the effectiveness of existing antibiotics. Myricetin could serve as a useful scaffold for the development of tetracycline resistance reversing molecules with potentially favourable safety profiles.