Thyme and cinnamon essential oils inhibit multidrug resistant Escherichia coli and Klebsiella pneumoniae and alter virulence transcripts

Abstract

The increasing incidence of multidrug-resistant (MDR) Gram-negative pathogens, particularly Escherichia coli and Klebsiella pneumoniae, continues to narrow effective treatment options and motivates evaluation of alternative antimicrobial strategies. Here, 33 essential oils (EOs) were screened against six MDR clinical isolates, identifying thyme and cinnamon oils as the most active. Both oils produced large inhibition zones (up to 26 mm) and low minimum inhibitory concentrations (MICs), with cinnamon oil showing a uniform MIC of 0.0488% (v/v) across all strains. For transcriptional analysis, cultures were exposed to cinnamon oil at 0.0244% (v/v) (0.5×MIC). Because this exposure corresponds to a near-MIC/inhibitory condition, observed decreases in virulence-gene transcript levels should be interpreted as inhibitory/stress-associated transcriptional responses rather than definitive sub-MIC anti-virulence effects. GC–MS profiling indicated enrichment of carvacrol (23.4%) and thymol (3.58%) in thyme oil, while cinnamon oil was dominated by cinnamaldehyde (45.8%) and eugenol (10.89%). Cinnamon also showed slightly higher antioxidant capacity than thyme (DPPH RSA 70.57% vs. 67%). Finally, molecular docking (HADDOCK/AutoDock Vina) was used as supportive in-silico screening to explore plausible ligand–target interaction hypotheses. Collectively, thyme and cinnamon EOs demonstrated antibacterial activity and were associated with reduced virulence-gene transcript abundance under near-MIC exposure, supporting further validation using verified sub-MIC conditions and growth controls.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-38791-2.

Introduction

Antibiotic resistance is among the most serious public health threats of the 21 st century, largely driven by extensive and sometimes inappropriate antibiotic use in human medicine and agriculture. This pressure has accelerated the emergence and dissemination of multidrug-resistant (MDR) bacteria, notably E. coli and K. pneumoniae, two Gram-negative pathogens frequently implicated in healthcare-associated infections such as urinary tract infections (UTIs), bloodstream infections, wound infections, and pneumonia¹. MDR K. pneumoniae—often defined as resistance to three or more antimicrobial classes—poses a particular concern because it is associated with severe disease and increased mortality, and it contributes substantially to difficult-to-treat UTIs². Globally, recurrent UTIs affect more than 1.5 million women annually, causing considerable morbidity and reducing quality of life, with heightened risk in hospitalized patients². Uropathogenic E. coli (UPEC) remains the leading etiological agent of UTIs, while K. pneumoniae similarly employs type 1 pili and the mannose-binding adhesin FimH to initiate cystitis; however, rising resistance to major antibiotic classes—including β-lactams, carbapenems, and fluoroquinolones—has increasingly restricted effective treatment options^1,3.

In addition to antimicrobial resistance, both organisms possess virulence determinants that support colonization, invasion, biofilm formation, and evasion of host immunity³. In E. coli, FimH mediates adhesion to epithelial surfaces, OmpA contributes to serum resistance and biofilm formation, Iss is linked to immune evasion, and LuxS participates in quorum sensing and virulence regulation⁴. K. pneumoniae likewise carries key virulence genes, including uge (capsule-associated pathways), mrkA (the major type 3 fimbrial subunit involved in adhesion and biofilm formation), fimH, and rmpA (associated with the hypermucoviscous phenotype), collectively enhancing pathogenicity⁵. Because virulence contributes directly to persistence and severity of infection, strategies that attenuate virulence rather than solely inhibit growth have gained interest as potentially resistance-sparing approaches¹⁵.

Natural products, particularly essential oils—volatile, aromatic plant-derived mixtures—have a long record of traditional use and are increasingly recognized for broad antimicrobial, antifungal, and anti-inflammatory activities⁶. Recent evidence indicates that certain essential oils can inhibit even drug-resistant bacteria and can interfere with quorum sensing, biofilm development, and virulence-gene expression, thereby acting as dual-function agents that suppress growth while reducing pathogenic potential^7–10. Beyond antimicrobial actions, the antioxidant properties of essential oils are also relevant, as their natural antioxidants can delay spoilage in seasoned foods and are widely applied in food preservation; moreover, major oil constituents have been reported to reduce polyunsaturated fatty acid oxidation and mitigate oxidative stress in biological systems^11,12.

At the mechanistic level, E. coli and K. pneumoniae virulence is mediated by adhesins (FimH), outer membrane proteins (OmpA), fimbrial structural proteins (MrkA), and quorum-sensing regulators (LuxS), which collectively promote colonization, immune evasion, and biofilm formation^13,14. Several essential-oil constituents—including cinnamaldehyde, eugenol, thymol, and carvacrol—have demonstrated antimicrobial and antivirulence activities, partly through modulation of gene expression and disruption of key bacterial processes¹⁶. However, the molecular basis of these effects remains incompletely defined. In this context, molecular docking can provide complementary insight by predicting binding modes and interaction sites between phytochemicals and bacterial virulence proteins, supporting interpretation of experimentally observed changes in virulence-gene expression¹⁷.

Therefore, this study aimed to evaluate the antimicrobial activity of the most active essential oils (thyme and cinnamon) against MDR clinical isolates of E. coli and K. pneumoniae, characterize their major constituents and antioxidant capacity, and use supportive molecular docking to explore plausible interactions between key phytochemicals and virulence-associated protein targets. Together, these analyses provide an integrated screening framework and define priorities for follow-up validation under confirmed sub-MIC exposure conditions.

Materials and methods

Bacterial isolates

Fifty clinical bacterial isolates were collected from Al-Qasr Al-Ainy Hospital. After antimicrobial susceptibility testing, six isolates were confirmed as multidrug-resistant (MDR) using the VITEK-2 automated system and were selected for downstream experiments: two K. pneumoniae isolates (K1 from sputum, K3 from urine) and four E. coli isolates (E5 and E6 from urine; E7 and E8 from wound swabs)¹⁸. All procedures were conducted in accordance with the ethical guidelines of the Faculty of Science, Cairo University. The protocol was approved by the Institutional Ethical Committee (Approval No.: FS-CU-2024-11), and written informed consent was obtained prior to sample collection.

Essential oils – source and storage

Thirty-three essential oils (Cinnamon, thyme, clove, black seed, ginger, eucalyptus, camphor, anise, bitter almond, lavender, castor, pumpkin, fennel, sage, turmeric, frankincense, garlic, argan, tea tree, chamomile, oregano and rosemary, lemon, marjoram, parsley, orange, nutmeg, jojoba, cypress, dill, juniper, black pepper, and sweet almond oils) were evaluated. Twenty reference-grade oils were obtained from the National Research Centre (NRC), Cairo, Egypt, and thirteen commercial oils were purchased from a GMP-certified Egyptian manufacturer to represent market-available preparations. Oils were stored in airtight amber glass vials at 4 °C, protected from light until use.

Antimicrobial susceptibility testing of the bacterial isolates

Disk diffusion assay

Bacterial strains were grown on Mueller–Hinton agar (MHA) at 37 °C for 18–24 h. A fresh bacterial suspension was prepared in sterile saline and adjusted to 0.5 McFarland (≈ 1 × 10^8 CFU/mL). The inoculum was lawn-spread onto MHA plates using a sterile swab. Sterile 6-mm filter paper discs were impregnated with 10 µL of each essential oil and placed onto the inoculated agar. Plates were incubated aerobically at 37 °C for 24 h, and inhibition zones (including the disc diameter) were measured in millimeters using a calibrated digital Vernier caliper. Each condition was tested in triplicate, and results are reported as mean ± SD. Discs containing the solvent vehicle (e.g., DMSO at the same final concentration used for EO dilution) served as negative controls. Interpretation followed general CLSI principles (CLSI, 2020)¹⁹. The absence of a standard antibiotic positive control is acknowledged as a study limitation and is addressed in the Discussion. Disk diffusion assays were performed using three independent plates per condition (biological replicates). MIC/MBC assays were conducted in two independent experiments with technical duplicates per condition.

Determination of minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC)

Following the primary agar diffusion screening, essential oils that exhibited marked antibacterial activity were selected for quantitative determination of the minimum inhibitory concentration (MIC) and minimum bactericidal concentration (MBC) against multidrug-resistant (MDR) gram-negative isolates of K. pneumoniae and E. coli. MICs were determined by the broth microdilution method in sterile, 96-well microtiter plates, using Mueller–Hinton broth (MHB) as the test medium.

Essential oils were first prepared as 50% (v/v) stock emulsions in 2% (v/v) DMSO to facilitate solubilization; stocks were vigorously vortexed and briefly sonicated immediately before dilution to minimize phase separation. The final DMSO concentration in the assay did not exceed a non-inhibitory level²⁰. For MIC testing, a working starting concentration of 25% (v/v) essential oil was prepared in MHB and then subjected to two-fold serial dilutions across the plate to obtain a concentration range of 25–0.012% (v/v) in rows A, B, C, F, G, and H (each row represented one isolate: A-K1, B-E5, C-K3, F-E6, G-E7, and H-E8). A second refined microdilution assay initiated at 0.20% (v/v) confirmed the same MIC endpoints.

Bacterial suspensions were adjusted to a turbidity equivalent to a 0.5 McFarland standard and 50 µL of this inoculum was added to each test well, yielding a final inoculum of approximately 5 × 10⁵ CFU/mL. Control wells included (i) sterility controls containing only MHB (row D1–D6), (ii) growth controls containing MHB plus the bacterial inoculum without essential oil (row E1–E6)²¹. Plates were sealed with parafilm and incubated for 24 h at 37 °C. All assays were performed in at least two independent experiments with technical triplicates per condition (n ≥ 3), and results are reported as mean values.

To determine bacterial viability and visually identify the MIC and subinhibitory concentrations (SICs), 40 µL of a resazurin solution 0.015% (w/v) was added to each well after incubation, followed by a further 2 h incubation at 37 °C²².Resazurin is a non-fluorescent blue dye that is reduced by metabolically active bacteria to the pink, fluorescent compound resorufin; thus, wells remaining blue were interpreted as growth inhibition, whereas wells turning pink indicated bacterial growth²³. The MIC was defined as the lowest concentration at which no visible growth (no color change) was observed. In this study, the concentration used for qRT-PCR corresponded to a near-MIC/inhibitory exposure (½ MIC of 0.0488% v/v); therefore, transcriptional changes are interpreted as responses under inhibitory/stress conditions rather than confirmed anti-virulence effects at sub-MIC exposure²⁴.

For MBC determination, 10 µL aliquots from wells at and above the MIC were aseptically streaked onto Mueller–Hinton agar (MHA) plates prior to resazurin addition. Plates were incubated at 37 °C for 24 h, and the MBC was defined as the lowest essential-oil concentration that yielded no detectable colony growth on MHA²⁵.

Disk-diffusion assay (inhibition-zone diameters). Inhibition zones (mm) were measured after 24 h for each essential oil–isolate combination. Each combination was tested on three independent plates (biological replicates), and values are reported as mean ± SD. For the screening dataset (Top 16 oils), zone diameters were analyzed using a two-way ANOVA with Oil (16 levels) and Isolate (6 levels: E. coli E5, E7, E8, E6; K. pneumoniae K1, K3) as fixed factors, including the Oil × Isolate interaction. Differences were considered statistically significant at p < 0.05.

Primer design and efficiency

Primers targeting virulence genes (fimH, ompA, luxS, iss, rmpA, uge, mrkA) and the 16 S rRNA/ITS regions were designed based on published gene sequences retrieved from GenBank and previous reports on E. coli and K. pneumoniae virulence determinants. Coding sequences were aligned to identify conserved regions, and primers were then designed using the Primer3Plus online tool (https://www.bioinformatics.nl/cgi-bin/primer3plus/primer3plus.cgi). Design parameters were set to generate amplicons of 100–200 bp, with primer lengths of 18–24 nucleotides, GC content of 40–60%, and melting temperatures (Tm) of approximately 58–62 °C, while avoiding significant self-complementarity and primer–dimer formation.

Primer specificity was first evaluated in silico using the PCR amplification tool (http://insilico.ehu.eus/user_seqs/PCR/) against the corresponding bacterial genomes and then confirmed by BLAST analysis to ensure unique target binding and absence of off-target amplification. Amplicon sizes were verified experimentally by agarose gel electrophoresis.

PCR efficiency and linearity (R²) for each primer pair were assessed by standard curve analysis using serial tenfold dilutions of cDNA. Slopes were obtained from linear regression of Ct values plotted against the logarithm of template concentration. Amplification efficiency (%) was calculated using the Eq²⁶:

Efficiency (%) = (10^(–1/slope) – 1) × 100.

Only primer sets with efficiencies between 90 and 110% and R² ≥ 0.99 were accepted for subsequent qRT-PCR assays as shown in Table 1.

Table 1.

QRT-PCR primers list used in this study.

Gene	Accession no.	Primer sequence (5′–3′)	Amplicon size (bp)	Slope	Efficiency (%)	R²	References
luxS	OL457181	F: 5′- CATACCCTGGAGCACCTGTT-3′ R: 5′- TGATCCTGCACTTTCAGCAC-3′	191 bp	−3.32	100.1	0.9998	This study
iss	EU627770	F: 5′- CGGGAATTGGACAAGAGAAA-3′ R: 5′- TCGATGGGCAACTATTGTGA-3′	175 bp	−3.31	100.5	0.9997	This study
ompA	OQ332854	F: 5′- GGGTTACCCAATCACTGACG-3′ R: 5′- TGGTATTCCAGACGGGTAGC-3′	189 bp	−3.27	102.1	0.9998	This study
fimH	KC357736	F: 5′- GGTATTACCTCTCCGGCACA-3′ R: 5′- GTGCGTAATTCGCCGTTAAT-3′	192 bp	−3.3	100.7	0.9993	This study
16 S rRNA	MN900682	F: 5′- CAGCCACACTGGAACTGAGA-3′ R: 5′- GTTAGCCGGTGCTTCTTCTG-3′	204 bp	−3.32	100.1%	0.9996	This study
rmpA	KY403895	F: 5′- GGGGCGGTTTTATCCTAAAG-3′ R: 5′- TTCAGTAGGCATTGCAGCAC-3′	177 bp	−3.3	101.1	0.9994	This study
uge	KY403938	F: 5′- CAGCTCACCAGTGGAACTGA-3′ R: 5′- TCTTCACGCCTTCCTTCACT-3′	182 bp	−3.33	99.8	0.9997	This study
mrkA	M20720	F: 5′- ACACCCATAGCCAGATGGAG-3′ R: 5′- GGCAAACTGGCTGATGTTTT-3′	155 bp	−3.29	101.3	0.9999	This study
16–23 S ITS	MK253052	F: 5′- CGCATAGCTCCACCATCTTT-3′ R: 5′-TGCGAAAATTTGAGAGACTCG-3′	155 bp	−3.31	100.4%	0.9996	This study

Quantitative real-time PCR (qRT-PCR) analysis of virulence gene expression

Six independent multidrug-resistant (MDR) bacterial isolates (four E. coli and two K. pneumoniae) were subjected to qRT-PCR analysis to assess transcriptional responses of selected virulence-associated genes before and after exposure to cinnamon essential oil at 0.0244% (v/v) (0.5×MIC; MIC = 0.0488% v/v). Because 0.5×MIC represents a near-MIC/inhibitory exposure, any reductions in transcript abundance are interpreted as inhibitory/stress-associated transcriptional responses rather than confirmed sub-MIC antivirulence effects²⁷.

Total RNA was extracted using the RNeasy Mini Kit (Qiagen; Cat. no. 74104) according to the manufacturer’s instructions. To preserve RNA integrity immediately after bacterial harvesting, cells were stabilized using RNAprotect Bacteria Reagent. RNA quantity and purity were assessed spectrophotometrically (A260/A280), and only samples with ratios between 1.8 and 2.1 were used for downstream analysis.

First-strand cDNA was synthesized from purified RNA using RevertAid Reverse Transcriptase (Thermo Fisher Scientific) following the manufacturer’s protocol. No-reverse transcriptase controls were prepared in parallel to exclude genomic DNA-derived amplification.

Quantitative PCR was performed using the QuantiTect SYBR Green PCR Kit (Qiagen; Cat. no. 204443) on a Stratagene Mx3005P real-time PCR system (96-well optical plates). Each 25 µL reaction contained 12.5 µL of 2× SYBR Green Master Mix, 0.5 µL of each primer, template cDNA, and RNase-free water. Cycling conditions were: 95 °C for 15 min, followed by 40 cycles of 94 °C for 15 s, gene-specific annealing (50–58 °C) for 30–40 s, and 72 °C for 40 s. A melt-curve analysis was performed at the end of each run to confirm single-product amplification.

The 16 S rRNA gene was used as the internal reference for normalization in E. coli, whereas the 16–23 S rRNA intergenic spacer (ITS) served as the housekeeping control for K. pneumoniae. Relative expression levels were calculated using the comparative 2^-ΔΔCt method and are reported as fold changes versus untreated controls. No-template controls (NTCs) were included in every run to exclude contamination.

DPPH radical scavenging assay

The antioxidant activity of essential oils from Thymus vulgaris (thyme) and Cinnamomum verum (cinnamon) was evaluated using the 2,2′-diphenyl-1-picrylhydrazyl (DPPH) radical scavenging assay. A 0.2 mM DPPH solution was prepared in methanol immediately before use. For each test condition, 0.5 mL of the essential oil sample was mixed with 1.0 mL of the DPPH solution in a glass tube, vortexed briefly, and incubated in the dark at room temperature for 30 min. The decrease in absorbance was then recorded at 517 nm using a JENWAY 6300 UV–Vis spectrophotometer, with methanol serving as the blank. A control reaction was prepared in parallel by replacing the essential oil sample with methanol.

All measurements were performed in triplicate in two independent experiments, and results were reported as mean ± SD. The radical scavenging activity (RSA) was calculated according to the following Eq²⁸.:

Where A_control is the absorbance of the control reaction (DPPH solution without essential oil) and A_sample is the absorbance of the reaction mixture containing the essential oil.

GC-MS spectroscopy

The phytochemical profiles of Thymus vulgaris and Cinnamomum verumessential oils were characterized by gas chromatography–mass spectrometry (GC–MS/MS) using an Agilent 7890B gas chromatograph coupled to an Agilent 7000 C triple quadrupole mass spectrometer (Agilent Technologies, USA). Sample preparation and analytical procedures were conducted in accordance with the standardized protocol described by Huwaimel et al²⁹., with minor modifications implemented to optimize extraction efficiency, enhance analytical sensitivity, and ensure compatibility with the specific experimental conditions and instrumentation used in the present study²⁹. Chromatographic separation was achieved on a TG-5MS capillary column (30 m × 0.25 mm i.d. × 0.25 μm film thickness). The oven temperature was initially set at 40 °C for 2 min, then increased to 120 °C at 3 °C/min, followed by a ramp to 280 °C at 5 °C/min with a final hold of 1 min. Helium served as the carrier gas at a constant flow rate of 1 mL/min. The injector temperature was maintained at 250 °C, the transfer line at 260 °C, and 1 µL of each sample was injected in split mode. Mass spectra were acquired in electron ionization (EI) mode at 70 eV and processed using MassHunter software (version B.08.00). Tentative identification of constituents was based on comparison of their mass spectra and retention behavior with those in the NIST mass spectral library^30,31.

Molecular docking and interaction visualization

Molecular docking was performed to explore potential binding interactions between the major essential-oil constituents and selected virulence-associated targets. Protein–ligand docking was conducted using the HADDOCK web server. Crystal structures were retrieved from the Protein Data Bank: FimH (PDB ID: 7QUO), LuxS (PDB ID: 5V2W), MrkA (PDB ID: 9HW9), and OmpA (PDB ID: 9FZC). Prior to docking, non-receptor chains and non-essential heteroatoms (e.g., crystallographic waters and any co-crystallized partners) were removed, and the receptor structure was prepared using the default HADDOCK preprocessing.

The three-dimensional structures of the phytochemical ligands carvacrol, cinnamaldehyde, eugenol, and thymol were obtained from the PubChem database in SDF format, converted into PDB format, and subjected to geometry optimization and energy minimization using a force-field–based minimization protocol to eliminate steric clashes and ensure conformational stability. Each ligand was docked individually against each protein target under default HADOCK protein–ligand docking parameters, generating up to 100 ranked docking poses per complex.

Docking scores (kcal/mol), confidence scores, and root-mean-square deviation (RMSD) values were extracted for all predicted complexes to assess both binding affinity and cluster convergence/pose variability descriptor. To ensure robustness and address potential concerns regarding docking accuracy, only poses that simultaneously exhibited (i) favorable binding energy, (ii) acceptable RMSD values, and (iii) consistent clustering behavior among the top-ranked conformations were selected for downstream interpretation. Docked poses associated with unstable conformations or unusually high RMSD values were excluded from further analysis to avoid unreliable structural predictions³².

Docking outputs (HADDOCK score, cluster size, and cluster RMSD relative to the overall lowest-energy structure) were recorded. RMSD values were used only as an internal clustering/pose-convergence descriptor (i.e., lower cluster RMSD indicating tighter convergence of solutions) rather than as an experimental measure of pose accuracy. For visualization, the top-ranked cluster(s) were selected based on HADDOCK score/Z-score and convergence (cluster RMSD), and representative poses were analyzed in BIOVIA Discovery Studio.

AutoDock Vina docking was performed for four receptors (9HW9, 7QUO, 5V2W, and 9FZC) using receptor-specific grid boxes while keeping the sampling settings constant across runs. For 9HW9, the grid box was centered at x = − 1, y = − 6, z = − 3 with a box size of 20 × 20 × 20 Å, and exhaustiveness (sampling exhaustively) = 4. For 7QUO, the grid center was x = − 64, y = − 31, z = − 34 with the same 20 × 20 × 20 Å box and exhaustiveness = 4. For 5V2W, the grid center was x = 53, y = 5, z = 18, again using a 20 × 20 × 20 Å box and exhaustiveness = 4. Finally, for 9FZC, the grid center was set to x = 1, y = 5, z = − 43, with a 20 × 20 × 20 Å box and exhaustiveness = 4. These grid definitions ensured consistent search volume while positioning the docking space over the intended binding region of each receptor.

In-silico prediction of ADME and toxicity using PkCSM online tool

The ADME and toxicity properties of four essential-oil constituents (carvacrol, cinnamaldehyde, eugenol, and thymol) were predicted using the pkCSM online platform, https://biosig.lab.uq.edu.au/pkcsm/prediction³³. (Canonical SMILES strings for each compound were retrieved from PubChem and checked for correctness before submission³⁴. The SMILES for each molecule were entered on the pkCSM “Prediction” page either by pasting a single SMILES into the input field or by uploading a SMILES file (up to 100 molecules), following the pkCSM help guidance that recommends canonical SMILES and notes that non-compliant strings may be ignored. On the submission page, the “All” (comprehensive) prediction option was selected to generate pharmacokinetic and toxicity outputs in a single run. The resulting pkCSM output tables were exported manually by recording predicted values for the endpoints reported in Table 7, including absorption (human intestinal absorption, %), distribution (blood–brain barrier permeability, LogBB; CNS permeability, LogPS), metabolism (CYP substrate status for CYP2D6 and CYP3A4; CYP inhibition predictions for CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4), excretion (total clearance; log mL/min/kg), and toxicity (AMES mutagenicity, hepatotoxicity, and skin sensitization).

Table 7.

Predicted ADME & toxicity profile (pkCSM) for four major essential-oil compounds.

(A) Quantitative endpoints
Compound	Human intestinal absorption (%)	BBB permeability (LogBB)	CNS permeability (LogPS)	Total clearance (log mL/min/kg)
Carvacrol	90.843	0.407	−1.664	0.207
Cinnamaldehyde	95.015	0.436	−1.582	0.203
Eugenol	92.041	0.374	−2.007	0.282
Thymol	90.843	0.407	−1.664	0.211

(B) Categorical endpoints (metabolism + toxicity)
Compound	CYP2D6 substrate	CYP3A4 substrate	CYP1A2 inhibitor	CYP2C19 inhibitor	CYP2C9 inhibitor	CYP2D6 inhibitor	CYP3A4 inhibitor	AMES	Hepatotoxicity	Skin sensitization
Carvacrol	No	No	Yes	No	No	No	No	No	No	No
Cinnamaldehyde	No	No	Yes	No	No	No	No	No	No	No
Eugenol	No	No	Yes	No	No	No	No	No	No	No
Thymol	No	No	Yes	No	No	No	No	No	No	No

BBB = blood–brain barrier; CNS = central nervous system; CYP = cytochrome P450.

Statistical analysis

All experiments were performed with three independent biological replicates unless otherwise stated. Data are presented as mean ± SD. For inhibition-zone screening (Top 16 oils), zone diameters were analyzed by two-way ANOVA with Oil (16 levels) and Isolate (6 levels) as fixed factors, including the Oil × Isolate interaction; post-hoc comparisons were performed using Tukey’s HSD test (p < 0.05). For qRT-PCR analysis, statistical testing was performed separately for each gene using ΔCt values, with Treatment (control vs. cinnamon oil exposure) and Isolate as fixed factors, followed by Tukey’s HSD where applicable (p < 0.05).

Results

Antibacterial activity of essential oils

The preliminary analysis confirmed that 2%, 1%, and 0.5% DMSO had no growth-inhibitory effects on the tested MDR isolates (Figure S13). Among the 33 essential oils tested, several exhibited measurable antibacterial activity against the six multidrug-resistant (MDR) clinical isolates of E. coli and K. pneumoniae. Figure 1 illustrate the antibacterial activity of the 16 most active essential oils against four E. coli strains (E5, E6, E7, and E8) and two K. pneumoniae strains (K1 and K3). The heatmap present inhibition zone diameters measured in millimeters. Cinnamon, thyme, and clove oils produced the largest inhibition zones, with several values equal to or exceeding 25 mm. K. pneumoniae K3 showed large inhibition zones with cinnamon and thyme oils, while E. coli E7 showed smaller inhibition zones across most tested oils. The hierarchical clustering in the heatmap showed variation in response patterns among bacterial strains. Oils ranked 17–33 exhibited minimal or no inhibition and were excluded from further analysis.

Fig. 1.

Heatmap of antibacterial activity for the 16 most active essential oils against multidrug-resistant (MDR) E. coli (E5–E8) and K. pneumoniae (K1, K3). Values are inhibition-zone diameters (mm) from agar diffusion assays (mean ± SD; n = 3 plates per oil–isolate). Color intensity reflects zone size (larger zones indicate stronger antibacterial activity). Hierarchical clustering (Euclidean distance) was applied to oils and isolates to visualize similarity in response patterns. Oils ranked 17–33 showed minimal/no inhibition and were excluded. Zone diameters were analyzed by two-way ANOVA (Oil × Isolate); p < 0.05 was considered significant.

Minimum inhibitory concentration (MIC) of thyme and cinnamon oils

MIC values for Thymus vulgaris and Cinnamomum verum essential oils were determined against the six MDR isolates (Fig. 2). For thyme oil, the MIC was 0.0975% for K3, E6, and E8 (lanes C, F, and H), and 0.0488% for K1, E5, and E7 (lanes A, B, and G). For cinnamon oil, the MIC was 0.0488% for all tested MDR strains. Lane D served as the negative control, and lane E served as the positive control.

Fig. 2.

Minimum inhibitory concentration (MIC) determination for thyme and cinnamon essential oils against six MDR isolates using broth microdilution with resazurin readout. Each row corresponds to one isolate (A: K1; B: E5; C: K3; F: E6; G: E7; H: E8). D1–D6 are sterility controls (media only; negative controls) and E1–E6 are growth controls (media + inoculum; positive controls). Thyme oil MIC ranged from 0.0488% to 0.0975% (v/v), while cinnamon oil MIC was 0.0488% (v/v) for all isolates.

Effect of cinnamon oil on virulence gene expression in E. coli

The transcript levels of fimH, ompA, luxS, and iss were quantified in four MDR E. coli isolates (E5, E6, E7, and E8) before and after exposure to cinnamon essential oil at ½ MIC (0.0244% v/v; MIC = 0.0488% v/v) (Figures S1–S5), a near-MIC/inhibitory concentration relative to the reported MIC (0.0488%). The post-treatment fold-change values for fimH were 0.3978 (E5), 0.2017 (E6), 0.3078 (E7), and 0.1416 (E8). For ompA, the fold-changes were 0.7220, 0.5105, 0.4931, and 0.7684 in E5–E8, respectively. The luxS fold-changes were 0.6417 (E5), 0.2973 (E6), 0.2238 (E7), and 0.5035 (E8), while iss values were 0.4414, 0.2813, 0.5249, and 0.2698, respectively. Consistent with reduced transcript abundance were higher than those in the corresponding untreated controls (EC1–EC4). Figure 3 shows the relative expression profiles of fimH, ompA, luxS, and iss in the treated isolates, with different letters above bars denoting significant differences (p < 0.05) within each treatment group. In agreement, amplification plots (Figures S1–S5) showed rightward CT shifts for 16 S rRNA, luxS, iss, ompA, and fimH following cinnamon-oil exposure.

Fig. 3.

Relative expression of virulence-associated genes (luxS, iss, fimH, and ompA) in MDR E. coli isolates after exposure to cinnamon essential oil at 0.0244% (v/v) (0.5×MIC; MIC = 0.0488% v/v). Transcript levels were calculated using the 2^-ΔΔCt method relative to untreated controls and normalized to 16 S rRNA. Data are mean ± SD (n = 3 biological replicates). Statistical significance was assessed by two-way ANOVA (Isolate × Gene) followed by Tukey HSD (p < 0.05). Different letters indicate significant differences; bars sharing letters are not significantly different within the comparison set.

Effect of cinnamon oil on virulence gene expression in K. pneumoniae

qRT-PCR analysis of two MDR K. pneumoniae isolates (K1 and K3) before and after cinnamon oil exposure is presented in Fig. 4. For KT1 (K1), fold-change values were 0.1507 for rmpA, 0.3015 for uge, 0.2415 for mrkA, and 0.3842 for fimH. For KT2 (K3), values were 0.2365, 0.5434, 0.3585, and 0.4444, respectively. Figure 4 shows relative expression of rmpA, uge, mrkA, and fimH in KT1 and KT2. Different letters indicate statistically significant differences among genes within each treatment. Amplification plots (Figures S6–S10) displayed delayed amplification for the 16–23 S ITS, rmpA, uge, mrkA, and fimH after treatment. Because the exposure level corresponded to a near-MIC/inhibitory concentration, the observed reductions in transcript levels are reported as transcriptional changes under inhibitory/stress conditions and should not be interpreted as definitive antivirulence effects at validated sub-MIC exposure. For transcriptional assays, the sub-inhibitory concentration (SIC) was defined as a fixed fraction of the MIC (½ MIC; alternatively ¼ MIC) to minimize growth-inhibition confounding. For cinnamon oil (MIC = 0.0488% v/v), the MIC was therefore 0.0244% v/v (½ MIC) (or 0.0122% v/v for ¼ MIC). Growth impact at the selected exposure (0.5×MIC) was monitored by OD600 and CFU enumeration at the sampling time (DATA not published).

Fig. 4.

Relative expression of virulence-associated genes (rmpA, uge, mrkA, and fimH) in MDR K. pneumoniae isolates after exposure to cinnamon essential oil at 0.0244% (v/v) (0.5×MIC; MIC = 0.0488% v/v). Relative expression was calculated by the 2^-ΔΔCt method and is presented as fold change relative to untreated controls. Bars represent means ± SD (n = 3 biological replicates). For each gene, statistical testing was performed on ΔCt values using two-way ANOVA (Treatment × Isolate) followed by Tukey’s HSD post-hoc testing (p < 0.05). Different letters indicate significant differences within each gene.

Antioxidant activity of thyme and cinnamon oils

DPPH radical scavenging activity is presented in Table 2. Cinnamon oil showed an RSA value of 70.57%, while thyme oil showed an RSA value of 67%.

Table 2.

DPPH radical scavenging activity of thyme and cinnamon essential oils.

Sample	Experiment 1 (OD at 517 nm)	Experiment 2 (OD at 517 nm)	Experiment 1 (Mean ± SD)	Experiment 2 (Mean ± SD)	Experiment 1 (RSA %)	Experiment 2 (RSA %)
Control (DPPH + methanol)	1.075, 1.274, 1.158	1.013, 1.216, 1.139	1.169 ± 0.100	1.123 ± 0.102	100	100
Thyme essential oil	0.441, 0.413, 0.438	0.413, 0.412, 0.421	0.431 ± 0.015	0.415 ± 0.005	59.0–67.6	59.2–66.1
Cinnamon essential oil	0.382, 0.395, 0.377	0.380, 0.370, 0.376	0.385 ± 0.009	0.375 ± 0.005	64.5–69.0	62.5–69.6

Differences among groups within each experiment (control vs. thyme vs. cinnamon) were tested using one-way ANOVA. When comparing the effects of both treatment and experiment, a two-way ANOVA was applied to evaluate main effects and interaction. A p value < 0.05 was considered statistically significant. Data are presented as mean ± SD from triplicate measurements for each independent experiment as shown as in Table 3.

Table 3.

Two-way ANOVA (Response = OD at 517 nm; Factors = Treatment × Experiment).

Source	df	F	p value
Treatment	2	319.43	3.93 × 10 -11
Experiment	1	0.725	0.411
Treatment × experiment	2	0.170	0.846
Residual (error)	12	—	—

The two-way ANOVA (response: OD at 517 nm; factors: Treatment and Experiment) indicates a highly significant main effect of Treatment (F^2,12 = 319.43, p = 3.93 × 10⁻¹¹), demonstrating that absorbance differs markedly among the control, thyme oil, and cinnamon oil groups. In contrast, the main effect of Experiment is not significant (F^1,12 = 0.725, p = 0.411), suggesting that overall OD values were comparable between Experiment 1 and Experiment 2. Importantly, the Treatment × Experiment interaction is also not significant (F^2,12 = 0.170, p = 0.846), indicating that the treatment-related differences are consistent across both experiments (i.e., the rank order and magnitude of treatment effects do not depend on the experiment). Overall, these results support robust and reproducible antioxidant-related reductions in OD517 for both essential oils relative to the control, with cinnamon showing the lowest OD values across experiments.

Phytochemical composition of essential oils

GC–MS analysis of thyme oil identified thymol (3.58%) and carvacrol (23.4%) as the major constituents, along with monoterpene hydrocarbons, sesquiterpenes, and fatty acids (Table 4; Figure S11).

Table 4.

Phytochemical composition of thyme essential oil identified by GC-MS analysis.

No	Compound name	Molecular weight (g/mol)	% Area	Compound nature	Retention index (RI)
1	α-Pinene	136.23	2.15	Monoterpene Hydrocarbon	932
2	Camphene	136.23	1.85	Monoterpene Hydrocarbon	946
3	β-Pinene	136.23	2.6	Monoterpene Hydrocarbon	974
4	Myrcene	136.23	3.4	Monoterpene Hydrocarbon	988
5	α-Terpinene	136.23	1.95	Monoterpene Hydrocarbon	1018
6	p-Cymene	134.22	6.8	Aromatic Monoterpene	1026
7	Limonene	136.23	4.5	Monoterpene Hydrocarbon	1032
8	γ-Terpinene	136.23	7.2	Monoterpene Hydrocarbon	1062
9	Linalool	154.25	2.75	Oxygenated Monoterpene	1095
10	Borneol	154.25	3.1	Oxygenated Monoterpene	1165
11	Terpinen-4-ol	154.25	4.2	Oxygenated Monoterpene	1177
12	α-Terpineol	154.25	3.75	Oxygenated Monoterpene	1191
13	Thymol	150.22	3.58	Phenolic Monoterpene	1290
14	Carvacrol	150.22	23.4	Phenolic Monoterpene	1305
15	β-Caryophyllene	204.36	2.25	Sesquiterpene Hydrocarbon	1420
16	α-Humulene	204.36	1.85	Sesquiterpene Hydrocarbon	1452
17	Germacrene D	204.36	2.5	Sesquiterpene Hydrocarbon	1480
18	Caryophyllene oxide	220.35	1.9	Oxygenated Sesquiterpene	1580
19	Spathulenol	220.35	1.65	Oxygenated Sesquiterpene	1595
20	Hexadecanoic acid (Palmitic acid)	256.43	2.1	Fatty acid	1960
21	Octadecanoic acid (Stearic acid)	284.48	1.25	Fatty acid	2150
22	9,12-Octadecadienoic acid (Linoleic acid)	280.45	3.8	Unsaturated fatty acid	2160
23	9-Octadecenoic acid (Oleic acid)	282.47	2.45	Unsaturated fatty acid	2170
24	Eicosane	282.55	1.1	Alkane	2200
25	Heneicosane	296.58	0.95	Alkane	2300
26	Docosane	310.6	0.85	Alkane	2400
27	Tricosane	324.62	2.75	Alkane	2500
28	Tetracosane	338.65	0.7	Alkane	2600
29	Pentacosane	352.68	0.6	Alkane	2700
30	Hexacosane	366.71	0.5	Alkane	2800
31	Heptacosane	380.74	0.4	Alkane	2900
32	Octacosane	394.77	0.35	Alkane	3000
33	Nonacosane	408.8	0.37	Alkane	3100
34	Triacontane	422.82	0.25	Alkane	3200
35	Dotriacontane	450.87	0.2	Alkane	3400

Cinnamon oil analysis showed cinnamaldehyde (45.8%) and eugenol (10.89%) as the principal compounds, in addition to monoterpenes, sesquiterpenes, and fatty acids (Table 5; Figure S12).

Table 5.

Phytochemical composition of cinnamon essential oil identified by GC-MS Analysis.

No	Compound name	Molecular weight (g/mol)	% Area	Compound nature	Retention index (RI)
1	α-Pinene	136.23	1.8	Monoterpene Hydrocarbon	930
2	Camphene	136.23	1.25	Monoterpene Hydrocarbon	945
3	β-Pinene	136.23	1.9	Monoterpene Hydrocarbon	970
4	Myrcene	136.23	2.1	Monoterpene Hydrocarbon	990
5	Limonene	136.23	3.2	Monoterpene Hydrocarbon	1030
6	Eucalyptol (1,8-Cineole)	154.25	2.75	Oxygenated Monoterpene	1042
7	p-Cymene	134.22	2.35	Aromatic Monoterpene	1025
8	γ-Terpinene	136.23	3.85	Monoterpene Hydrocarbon	1060
9	Linalool	154.25	2.6	Oxygenated Monoterpene	1094
10	Terpinen-4-ol	154.25	1.95	Oxygenated Monoterpene	1175
11	Cinnamaldehyde (major)	132.16	45.8	Aromatic Aldehyde	1260
12	Eugenol	164.2	10.89	Phenolic compound	1355
13	α-Copaene	204.36	1.15	Sesquiterpene Hydrocarbon	1375
14	β-Caryophyllene	204.36	3.25	Sesquiterpene Hydrocarbon	1418
15	α-Humulene	204.36	1.95	Sesquiterpene Hydrocarbon	1450
16	Germacrene D	204.36	2.5	Sesquiterpene Hydrocarbon	1480
17	δ-Cadinene	204.36	1.3	Sesquiterpene Hydrocarbon	1525
18	Caryophyllene oxide	220.35	2.1	Oxygenated Sesquiterpene	1580
19	Spathulenol	220.35	1.06	Oxygenated Sesquiterpene	1595
20	Hexadecanoic acid (Palmitic acid)	256.43	2.4	Fatty acid	1960
21	9,12-Octadecadienoic acid (Linoleic acid)	280.45	2.85	Unsaturated fatty acid	2160
22	Octadecanoic acid (Stearic acid)	284.48	1	Fatty acid	2150

In Silico molecular docking analysis

Docking analysis was performed between cinnamaldehyde, carvacrol, eugenol, and thymol and the target proteins FimH (7QUO), LuxS (5V2W), MrkA (9HW9), and OmpA (9FZC). The Table 6 summarized molecular docking outputs (HADDOCK and AutoDock Vina) for four ligands—carvacrol, thymol, eugenol, and cinnamaldehyde—against four protein targets (7QUO, 5V2W, 9HW9, and 9FZC). For each receptor–ligand complex, it reported the HADDOCK score (mean ± SD), cluster size (n), RMSD (Å), van der Waals, electrostatic, and desolvation energies (kcal/mol), restraints violation energy (kcal/mol), buried surface area (Ų), Z-score, and Vina affinity (kcal/mol). RMSD values ranged from 0.2 to 0.9 Å, cluster sizes ranged from 17 to 199, and buried surface areas ranged from ~ 301.5 to 426.4 Ų. AutoDock Vina affinities spanned approximately − 3.289 to − 5.604 kcal/mol, with the most negative affinities observed for the 9HW9 target (e.g., eugenol − 5.604, carvacrol − 5.564, and thymol − 5.554 kcal/mol). Van der Waals energies remained negative across all complexes (about − 4.2 to − 14.5 kcal/mol), desolvation energies were also negative (about − 1.9 to − 6.8 kcal/mol), and electrostatic energies varied widely (approximately − 33.3 to + 0.8 kcal/mol).

Table 6.

Molecular Docking metrics (HADDOCK scoring components and AutoDock Vina affinities) of major thyme and cinnamon essential-oil constituents (carvacrol, thymol, eugenol, and cinnamaldehyde) against virulence-related target proteins (PDB ids: 7QUO, 5V2W, 9HW9, and 9FZC).

Receptor -ligand	HADDOCK score	Cluster size	RMSD (Å)	Van der Waals energy (kcal/mol)	Electrostatic energy (kcal/mol)	Desolvation energy (kcal/mol)	Restraints violation energy (kcal/mol)	Buried zsurface area (Ų)	Z-Score	AutoDock Vina Affinity (kcal/mol)
7QUO-carvacrol	229.8 ± 1.5	163	0.5 ± 0.1	−6.3 ± 4.1	0.3 ± 1.8	−5.5 ± 0.3	2415.4 ± 50.4	351.3 ± 9.0	−1.3	−4.189
7QUO-cinnamaldehyde	241.4 ± 8.0	132	0.7 ± 0.1	−5.7 ± 1.9	−1.5 ± 2.6	−4.4 ± 0.3	2518.2 ± 88.3	316.0 ± 10.3	−2.0	−4.034
7QUO-eugenol	220.0 ± 2.4	167	0.4 ± 0.3	−6.7 ± 2.2	0.4 ± 1.2	−5.1 ± 0.4	2316.8 ± 28.6	363.5 ± 6.5	−1.8	−4.534
7QUO-thymol	231.0 ± 4.8	139	0.3 ± 0.2	−9.0 ±1.5	−0.4 ± 0.7	−5.1 ± 0.2	2452.0 ± 56.2	342.1 ± 5.8	−1.8	−4.478
5V2W-carvacrol	241.1 ± 15.6	128	0.3 ± 0.2	−8.3 ± 1.9	−0.9 ± 1.2	−5.6 ± 0.6	2551.9 ± 172.8	342.0 ± 6.1	−1.3	−4.660
5V2W-cinnamaldehyde	241.5 ± 15.7	154	0.4 ± 0.2	−8.5 ± 3.5	−4.2 ± 5.1	−4.3 ± 0.6	2552.3 ± 128.3	301.5 ± 10.0	−1.4	−3.543
5V2W-eugenol	253.9 ± 7.6	127	0.9 ± 0.1	−6.5 ± 2.5	−0.0 ± 0.5	−5.1 ± 0.5	2655.7 ± 76.8	349.0 ± 12.1	−1.3	−4.329
5V2W-thymol	250.1 ± 18.2	129	0.4 ± 0.2	−7.3 ± 4.1	−1.0 ± 0.4	−6.8 ± 0.5	2644.3 ± 205.6	341.0 ± 9.8	−1.3	−4.077
9HW9-carvacrol	265.5 ± 11.3	185	0.4 ± 0.2	−6.2 ± 5.6	−0.1 ± 1.1	−5.9 ± 0.4	2777.0 ± 89.9	379.8 ± 12.7	−1.3	−4.315
9HW9-cinnamaldehyde	272.5 ± 11.1	184	0.3 ± 0.2	−3.1 ± 2.0	−10.2 ± 10.1	−5.0 ± 0.7	2826.4 ± 91.5	345.4 ± 9.3	−1.4	−4.279
9HW9-eugenol	263.9 ± 4.7	187	0.2 ± 0.1	−13.9 ± 0.4	−0.8 ± 0.7	−5.9 ± 1.1	2837.5 ± 47.2	395.2 ± 6.5	−1.0	−4.142
9HW9-thymol	264.9 ± 8.3	183	0.5 ± 0.1	−6.9 ± 2.1	0.3 ± 1.2	−6.3 ± 0.9	2781.3 ± 100.2	370.8 ± 7.1	−1.4	−4.113
9FZC-carvacrol	271.7 ± 22.1	69	0.2 ± 0.2	−4.2 ± 4.3	0.3 ± 0.7	−6.6 ± 2.0	2823.7 ± 212.9	323.7 ± 17.5	−1.5	−4.220
9FZC-cinnamaldehyde	249.5 ± 10.7	78	0.4 ± 0.2	−12.2 ± 2.3	−33.3 ± 4.0	−3.3 ± 0.5	2716.2 ± 121.0	331.6 ± 8.8	−2.0	−3.289
9FZC-eugenol	254.8 ± 13.9	17	0.6 ± 0.5	−7.6 ± 6.2	−0.3 ± 0.3	−5.4 ± 1.4	2678.8 ± 167.9	361.5 ± 18.2	−1.1	−3.859
9FZC-thymol	247.0 ± 15.7	53	0.2 ± 0.1	−13.1 ± 2.3	−0.7 ± 1.4	−4.2 ± 0.3	2645.1 ± 164.9	363.4 ± 1.9	−1.6	−3.535

Figures 5, 6, 7 and 8 presented the molecular docking poses of cinnamaldehyde, carvacrol, eugenol, and thymol in four targets, where each figure showed a 3D binding-pocket view (ligands in stick representation within the cavity surface) alongside 2D residue interaction maps generated in BIOVIA Discovery Studio (BIOVIA_DS2025Client). In Fig. 5 (7QUO), the ligands were positioned within the FimH pocket and were surrounded by labeled residues including Trp103, Leu38, Val20/22/36/105, Ile42/126/148, Phe84, Ala127, Leu107, Val128, and annotated contacts involving Tyr48, Gln133, Asn135, and Phe142. In Fig. 6 (5V2W), the binding environments included residues such as Ala61, Met64, Met81, Met89, Ile53/Ile78, Leu38/Leu56, Val19/Val100, Asp37, Glu57, Gly86, Thr85, Tyr88, Arg39, and Phe36/Phe40/Phe60/Phe87. In Fig. 7 (9HW9), across complexes, ligands occupy a similar predominantly hydrophobic/aromatic region, with recurring contacts including Trp78, Val76/Val109/Val149, Leu108/Leu110, Ile122, Tyr147/Tyr148, and polar-edge residues such as Ser111/Thr112 and Lys131/Lys133. In Fig. 8 (9FZC), the ligands were shown within the binding pocket with labeled surrounding residues such as Gln142, Tyr141, Trp143, Phe123, Ala124, Gly125, Gly126, Thr95, Val122, Arg96, Tyr94, and Glu140.

Fig. 5.

Representative docking poses and 2D interaction maps of cinnamaldehyde (A), carvacrol (B), eugenol (C), and thymol (D) within the FimH adhesin receptor (PDB: 7QUO). Upper panels show ligands as stick models positioned in the pocket surface; lower panels show 2D interaction maps with labeled contact residues. Visualizations were generated using BIOVIA Discovery Studio.

Fig. 6.

Representative docking poses and 2D interaction maps of carvacrol (A), cinnamaldehyde (B), eugenol (C), and thymol (D) within LuxS (PDB: 5V2W). Upper panels show 3D pocket placement; lower panels show residue-level interaction maps. Visualizations were generated using BIOVIA Discovery Studio.

Fig. 7.

Predicted docking poses of four phytochemicals—carvacrol (A), cinnamaldehyde (B), eugenol (C), and thymol (D)—within the MrkA pocket (PDB ID: 9HW9). For each ligand, the upper panel shows the 3D binding pose (protein in cartoon with pocket surface; ligand as sticks), and the lower panel shows the corresponding 2D interaction map highlighting nearby residues. Across complexes, ligands occupy a similar predominantly hydrophobic/aromatic region, with recurring contacts including Trp78, Val76/Val109/Val149, Leu108/Leu110, Ile122, Tyr147/Tyr148, and polar-edge residues such as Ser111/Thr112 and Lys131/Lys133.

Fig. 8.

Representative docking poses and 2D interaction maps of carvacrol (A), cinnamaldehyde (B), eugenol (C), and thymol (D) within OmpA-short (PDB: 9FZC). Upper panels show ligand placement in the predicted pocket; lower panels show residue-level interaction maps. Visualizations were generated using BIOVIA Discovery Studio.

Table 7 reported predicted ADME and toxicity properties for four compounds (carvacrol, cinnamaldehyde, eugenol, and thymol). Human intestinal absorption values were 90.843% (carvacrol), 95.015% (cinnamaldehyde), 92.041% (eugenol), and 90.843% (thymol). Blood–brain barrier permeability (Log BB) values were 0.407, 0.436, 0.374, and 0.407 for carvacrol, cinnamaldehyde, eugenol, and thymol, respectively, while CNS permeability (Log PS) values were − 1.664, − 1.582, − 2.007, and − 1.664, respectively. For metabolism, all four compounds were predicted as non-substrates for CYP2D6 and CYP3A4, and all were predicted as CYP1A2 inhibitors; for inhibition of CYP2C19, CYP2C9, CYP2D6, and CYP3A4, carvacrol, cinnamaldehyde, eugenol, and thymol were “No” across these enzymes. Total clearance (log ml/min/kg) values were 0.207 (carvacrol), 0.203 (cinnamaldehyde), 0.282 (eugenol), and 0.211 (thymol). AMES test predictions were “No” for carvacrol, cinnamaldehyde, eugenol, and thymol. Hepatotoxicity predictions were “No” for carvacrol, cinnamaldehyde, eugenol, and thymol. Skin sensitization was predicted as “No” for all four compounds.

Discussion

This study highlighted the substantial antibacterial activity of several essential oils against multidrug-resistant clinical isolates of E. coli and K. pneumoniae, together with reduced virulence-gene transcript levels observed under near-MIC cinnamon-oil exposure. The antimicrobial effects observed in the disc diffusion and MIC/MBC assays were consistent with the ability of essential oils to penetrate bacterial envelopes and induce structural and functional damage through their hydrophobic nature, leading to membrane disruption, leakage of intracellular contents, and eventual cell lysis^35,36.

The antibacterial activity of the tested oils was not uniform across strains. Clove oil produced inhibition zone against K. pneumoniae stains, while thyme and cinnamon oil were also highly active against selected E. coli and K. pneumoniae isolates. Conversely, some oils such as camphor showed only modest or negligible inhibitory effects. This variability reflects strain-dependent susceptibility patterns that have been described in other essential oil studies and underscores that not all essential oils are equally suitable as antimicrobial candidates^37,38.

The particularly strong inhibitory effects of thyme and cinnamon oils, with inhibition zones up to 26 mm and low MIC values (0.097–0.0488% for thyme and 0.0488% for cinnamon), were consistent with previous reports that identified these oils among the most potent plant-derived antimicrobials against MDR pathogens^39,40. The resazurin-based microtiter assay corroborated the diffusion data and clearly differentiated between growth-permissive and inhibitory concentrations, reinforcing the concentration-dependent nature of the antibacterial response^41,42.

Essential oils extracted from Thymus species have shown inhibition zones ranging from 17 mm⁴³ and 23 mm⁴⁴ to as high as 30–50 mm⁴⁵, with such variation influenced mainly by the region of growth, developmental stage, and the specific bacterial strains tested⁴⁶. The pronounced antibacterial activity of thyme and cinnamon oils can be attributed to their high concentrations of bioactive compounds—particularly thymol, carvacrol, and cinnamaldehyde—which are recognized for their potent antimicrobial and anti-biofilm properties. These compounds act through multiple mechanisms, including disruption of bacterial cell membranes, interference with essential cellular processes, and inhibition of virulence factors. Cinnamaldehyde, the principal active component of cinnamon essential oil, exhibits strong bactericidal effects by altering membrane permeability, compromising membrane integrity, and inducing morphological changes in bacterial cells⁴⁷. Its antibacterial activity is further linked to the depolarization of the cytoplasmic membrane^48,49, suppression of microbial toxin production, and inhibition of biofilm formation. Additionally, cinnamaldehyde has been shown to reduce fimbriae production and impair swarming motility in Uropathogenic E. coli^50–52.

Chemical profiling by GC–MS provided a mechanistic basis for the observed bioactivity of thyme and cinnamon oils⁵³. Thyme oil was dominated by the phenolic monoterpenes thymol (3.58%) and carvacrol (23.4%), both of which are widely recognized for their strong antimicrobial and antioxidant properties. Cinnamon oil, in turn, contained cinnamaldehyde (45.8%) and eugenol (10.89%) as major constituents, compounds known to exert broad antimicrobial, anti-inflammatory, and antioxidant effects^37,54. The presence of additional monoterpenes, sesquiterpenes, and fatty acids in both oils suggests potential interactions among components that may enhance membrane disruption, interfere with energy metabolism, and potentiate antibacterial efficacy^38,55.

At the molecular level, the virulence profiling of E. coli and K. pneumoniae isolates provided important context for interpreting the qRT-PCR findings. In E. coli, the universal detection of fimH, ompA, luxS, and iss confirmed a virulence repertoire centered on adhesion, outer-membrane stability, quorum sensing, and serum survival^56,57. fimH encodes the mannose-binding adhesin of type 1 fimbriae, which is essential for attachment to epithelial surfaces and initiation of biofilm formation^58,59. ompA contributes to outer-membrane integrity, adhesion, and resistance to host defenses, while luxS regulates quorum sensing and the expression of multiple virulence factors via autoinducer-2^60,61. The iss gene confers increased serum survival and has been linked to enhanced persistence in extra-intestinal sites^62,63.

Exposure of MDR isolates to near-MIC cinnamon oil (0.5×MIC) was associated with reduced transcript abundance of fimH, ompA, luxS, and iss in E. coli and of rmpA, uge, mrkA, and fimH in K. pneumoniae, as reported in the qRT-PCR tables/heatmaps. Because 0.5×MIC represents an inhibitory condition, these transcriptional shifts should be interpreted primarily as inhibition/stress-associated responses rather than definitive growth-independent antivirulence effects⁶⁴. Nevertheless, the affected genes are functionally linked to adhesion, quorum sensing, capsule-associated pathways, and biofilm formation, suggesting that future work under verified sub-MIC, growth-matched conditions (with concurrent OD600/CFU controls) may help resolve whether true antivirulence modulation occurs independent of growth inhibition.

The in silico docking analysis complemented the gene-expression data by suggesting plausible binding interactions between major essential oil constituents (cinnamaldehyde, eugenol, thymol, and carvacrol) and selected virulence-associated proteins. Docking against FimH (7QUO), LuxS (5V2W), MrkA (9HW9), and OmpA (9FZC) indicated moderate to high binding affinities and stable poses at functionally relevant sites, including residues involved in adhesion, fimbrial assembly, membrane stability, and quorum-sensing regulation^65,66. The docking summary in Table 6 (HADDOCK cluster statistics and energy terms, plus AutoDock Vina affinities) was interpreted using established molecular-docking frameworks, where the reported pose clusters and scoring components reflected data-driven docking (HADDOCK) and empirical scoring (Vina)⁶⁷. Across complexes, the consistently low RMSD values and sizeable clusters supported pose convergence within each target, while the negative van der Waals and desolvation terms indicated that nonpolar packing and solvent-exclusion contributions dominated many solutions, with electrostatic terms varying by ligand–target pair. Because docking scores and rankings remained method- and scoring-function–dependent, the numerical trends were treated as comparative and hypothesis-supporting rather than definitive measures of biological inhibition⁶⁸.

Figures 5–8 visualize the predicted binding poses and interactions of key essential-oil constituents (cinnamaldehyde, eugenol, thymol, and carvacrol) against selected virulence-associated targets, including 7QUO (FimH), 5V2W (LuxS), 9HW9 (type 3 fimbrial subunit MrkA), and 9FZC (OmpA), based on available structural data. These docking outputs are presented as hypothesis-generating only and were used to explore plausible ligand–target interaction hypotheses reported in prior literature^69, 70.

Across targets, predicted docking scores were generally in the low-to-moderate range (approximately -3 to -6 kcal/mol). These values do not confirm binding or biological inhibition; rather, they provide qualitative, hypothesis-generating patterns. Notably, several constituents (eugenol and cinnamaldehyde) showed recurrent predicted contacts across LuxS/FimH/MrkA-related structures, which may be explored in future functional assays. Any alignment between docking patterns and reduced transcript abundance observed under near-MIC exposure should be interpreted cautiously and validated under confirmed sub-MIC, growth-matched conditions.

In addition to their antimicrobial and anti-virulence effects, thyme and cinnamon oils exhibited notable antioxidant activity in the DPPH assay, with cinnamon showing slightly higher radical-scavenging capacity. This antioxidant potential aligns with previous reports for these and related essential oils and may add value in food preservation or therapeutic formulations where oxidative stress control is desirable^15,66. The combination of antimicrobial, anti-virulence, and antioxidant properties therefore positions thyme and cinnamon oils as promising candidates for development as natural adjuvants or complementary agents in strategies targeting MDR Gram-negative infections^71,72.

The pkCSM-derived ADME outputs in Table 7 (quantitative and categorical panels) indicate consistently high predicted human intestinal absorption for carvacrol, cinnamaldehyde, eugenol, and thymol (90.843–95.015%), supporting favorable modeled oral uptake for these small, lipophilic essential-oil constituents. Because these values are structure-based in silico predictions, they should be treated as computational screening estimates rather than experimental absorption measurements^33,73,74. Distribution descriptors were broadly concordant across compounds, with positive LogBB values (0.374–0.436) alongside negative LogPS values (−1.582 to −2.007), suggesting potential for BBB-associated penetration signals while indicating comparatively lower modeled CNS permeability on the LogPS scale^33,74.

For metabolism, all four compounds were predicted as non-substrates for CYP2D6 and CYP3A4, while all were flagged as CYP1A2 inhibitors; in contrast, no inhibition signals were returned for CYP2C19, CYP2C9, CYP2D6, or CYP3A4 in this dataset. Given the role of CYP1A2 (and, more broadly, major CYP isoforms) in the clearance of many therapeutic agents, a CYP1A2 inhibition flag is typically interpreted as an early indicator of potential metabolic interaction liability that warrants confirmatory in vitro enzyme inhibition testing when translational development is intended^75,76. Predicted total clearance values were clustered within a narrow interval (0.203–0.282 log mL/min/kg), with eugenol showing the highest modeled clearance among the four compounds, consistent with compound-specific differences in predicted disposition behavior^33,73.

Toxicity predictions in Table 7 did not raise AMES mutagenicity, hepatotoxicity, or skin-sensitization alerts for any of the four molecules (all “No”). Nonetheless, because first-tier mutagenicity (e.g., bacterial reverse-mutation assays) and liver-safety assessment rely on standardized experimental frameworks and a weight-of-evidence approach, these negative in silico calls should still be viewed as preliminary and best used to prioritize downstream validation rather than as definitive safety conclusions^77–80.

This study has several limitations. First, the disk-diffusion screening did not include a standard antibiotic positive control, which limits direct benchmarking of inhibition-zone sizes across assays. Second, qRT-PCR was performed after exposure to cinnamon oil at 0.5×MIC, which represents a near-MIC/inhibitory condition; therefore, reduced virulence-gene transcript levels may reflect general inhibitory stress responses rather than growth-independent anti-virulence effects. Future work should validate transcriptional effects under confirmed sub-MIC conditions (e.g., 0.25×MIC) with quantitative growth controls (OD600/CFU) and should extend testing to additional isolates and time points. Finally, molecular docking and ADME predictions were used only as computational screening tools and require experimental validation to confirm binding and biological relevance.

Despite these constraints, the integrated phenotypic, molecular, and in silico data support the view that thyme and cinnamon essential oils, rich in thymol, carvacrol, cinnamaldehyde, and eugenol, may provide a multifaceted approach to controlling MDR E. coli and K. pneumoniae by simultaneously inhibiting growth, attenuating virulence, and offering antioxidant benefits. Further in vivo investigations, formulation studies, and combination trials with conventional antibiotics are warranted to translate these findings into clinically relevant applications.

Conclusion

This study provided evidence that essential oils from Thymus vulgaris and Cinnamomum verum showed multitargeted activity against multidrug-resistant (MDR) E. coli and K. pneumoniae. Antibacterial activity was demonstrated in phenotypic assays, with cinnamon oil showing consistent inhibitory performance across the tested strains. In addition, virulence-associated gene expression was reduced, as confirmed through qRT-PCR analysis, and both oils exhibited measurable antioxidant activity. GC–MS profiling identified the major phytochemicals likely contributing to these effects, and the proposed protein–ligand interactions were further supported by in silico docking analyses against selected virulence-related targets. Given the scientific, methodological, and structural limitations highlighted during review, these findings were interpreted cautiously and were presented as a foundation for a more rigorously designed follow-up work, including expanded isolate panels, stronger benchmarking controls, and in vivo/formulation validation, to clarify the potential role of these oils as complementary agents in addressing the persistent challenge of antibiotic resistance. Overall, thyme and cinnamon oils showed measurable antibacterial activity and were associated with reduced virulence-gene transcript abundance under inhibitory exposure, supporting follow-up validation under confirmed sub-MIC conditions and expanded isolate panels.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

N.M.M. and D.T.E. conceived and designed the study. N.M.M., M.M.A.M. and M.E.A.D. conducted the experimental work and performed the phenotypic and molecular assays. M.M.A.M. and K.A.E.G. carried out the GC-MS analysis and antioxidant activity assessment. K.A.E.G. and M.M.A.M. performed the molecular docking and in silico analysis. N.M.M. and M.E.A.D. wrote the main manuscript text. D.T.E., and M.M.A.M. and K.A.E.G. contributed to data interpretation and critical manuscript revisions. All authors reviewed and approved the final version of the manuscript.

Funding

Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB).

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participate

approved by the Ethical Committee of the Faculty of Science, Cairo University, Egypt. Written informed consent was obtained.