Celery (Apium graveolens L.), coriander (Coriandrum sativum L.) and parsley (Petroselinum crispum) were purchased from a commercial market in Alexandria, Egypt, and their aerial parts were extracted separately with methanol. Each methanolic extract was then partitioned into a methylene chloride fraction and an aqueous mother liquor. The biological tests used 20 uropathogenic E. coli isolates obtained from private laboratories in Alexandria. Susceptibility testing showed that every isolate was resistant to amoxicillin-clavulanic acid, doxycycline and nalidixic acid, while all were susceptible to ertapenem, imipenem and meropenem. Twelve isolates were classified as strong biofilm formers, four as intermediate formers and four as non-biofilm formers.

The antibacterial results varied markedly among the herbs and fractions. Coriander’s total methanolic extract and aqueous mother liquor each produced a minimum inhibitory concentration of 3.75 micrograms per millilitre, the lowest value reported for the plant preparations. Its methylene chloride fraction required 100 micrograms per millilitre. Celery was less active: the total methanolic extract had a minimum inhibitory concentration of 50 micrograms per millilitre, while its aqueous mother liquor reached 25 micrograms per millilitre and its methylene chloride fraction showed no detectable activity under the tested conditions. Parsley’s total extract and aqueous mother liquor likewise had no detectable activity, although its methylene chloride fraction inhibited growth at 100 micrograms per millilitre. Ciprofloxacin and norfloxacin had minimum inhibitory concentrations between 0.9 and 1.9 micrograms per millilitre alone. When combined with coriander extract, the effects were described as additive, with no evidence of synergism or antagonism in the tested combinations.

Untargeted metabolomics produced markedly different profiles for the three herbs. Analyses were run in positive and negative electrospray ionization modes and incorporated retention time, accurate mass, predicted molecular formulas and tandem mass-spectrometry fragmentation patterns. Compound detection and processing initially yielded 4,861 features in negative mode and 3,155 in positive mode; after peak detection, deisotoping, chromatographic alignment and filtering, 580 and 512 aligned features, respectively, remained for chemometric analysis. The final annotation included flavonoids and glycosides, isoflavones, phenolic acids, coumarins, amino-acid derivatives, fatty acids and lipids, chlorins and a lignan. All 167 annotations were assigned level 2 confidence, meaning they were tentative identifications supported by library or literature spectral matches rather than confirmed with authentic standards.

Coriander had the richest and most diverse flavonoid and phenolic profile in this analysis. Its extracts contained quercetin, multiple quercetin conjugates, apigenin glucuronide, luteolin-7-O-glucoside, kaempferol-3-O-glucorhamnoside, several caffeoylquinic-acid derivatives, ferulic acid and assorted coumarins. Celery was distinguished by compounds including apiin, luteolin and apigenin conjugates, 4,5-dicaffeoylquinic acid and the coumarin marmesin. Parsley contained numerous furanocoumarins, including cnidilin, 5-hydroxyxanthotoxin, pinnatin, bergaptol and isoimperatorin, as well as rhamnetin and isorhamnetin conjugates. The researchers also detected chlorin-related molecules in all three herbs and the lignan eucommicin A in celery and coriander.

Principal component analysis, hierarchical cluster analysis and partial least squares discriminant analysis separated the three samples according to their metabolite profiles. In negative-ion mode, the first two principal components accounted for 95% of the observed variability, and the samples formed three distinct clusters. Coriander was associated with high levels of quercetin 3-O-rutinoside, eucommicin A and several hydroxylated C18 fatty acids, including three trihydroxy-octadecaenoic-acid features and hydroxy-octadecatrienoic- and hydroxy-octadecadienoic-acid features. Celery was associated with apiin, luteolin 7-O-apiosylglucoside, pyroglutamic acid, lunularic acid and a glycerophosphoinositol. Parsley was distinguished by fructose-phenylalanine and features including bergaptol. Supervised models separating celery from parsley reported R² and Q² values of 0.99 in both ionization modes, although the limited set of triplicate samples means these internal performance values should not be confused with independent validation.

Partial least squares regression then related metabolite abundance to the measured minimum inhibitory concentrations. The negative-ion model had cumulative R² of 0.92 and Q² of 0.87, while the positive-ion model had R² of 0.94 and Q² of 0.83. In both models, coriander’s metabolite pattern was inversely positioned relative to minimum inhibitory concentration, whereas parsley lay closest to the values associated with inactivity. Prominent variables associated with antibacterial activity included quercetin 3-O-rutinoside, trihydroxy-octadecaenoic-acid isomers, hydroxy-octadecatrienoic acid, hydroxy-octadecadienoic acid and eucommicin A. Positive-ion analysis additionally highlighted furanocoumarin hexosides, a methoxy-ethoxycarbonyl coumarin and a chlorin-related compound, 132-hydroxypheophorbide-A methyl ester. These are statistical associations within extract-level data, not demonstrations that any individual metabolite caused the observed inhibition.

All tested plant preparations also inhibited biofilm formation, whether used alone or in combinations with ciprofloxacin or norfloxacin. Biofilm mass was measured after 24 hours by fixing the attached biomass with methanol, staining it with crystal violet, releasing the dye with acetic acid and measuring absorbance at 630 nanometres. The study reported significant reductions in biofilm mass across the extract and antibiotic-combination conditions, including tests with sub-minimum-inhibitory-concentration levels of the plant extracts. The researchers discussed several possible contributors, including quercetin derivatives, phenolic compounds and coumarins that have activity against cell membranes, quorum sensing, extracellular polymeric substances and other biofilm-maintenance processes in earlier studies. Those mechanisms remain hypotheses for the herb extracts tested here, which were not separated into constituent compounds during the assays.

The bacterial collection also displayed several virulence-associated genes, adding context to the laboratory model. The fimH gene was present in 80% of the isolates, afa genes in 90%, papC in 76% and hly genes in 76%; the cytotoxic necrotizing factor gene occurred in 35%. These findings characterize the tested isolates but do not show that the herbal preparations affect the corresponding virulence traits. Moreover, the experiments were performed in vitro with commercial plant material, methanol-derived fractions and an artificial biofilm assay. They did not examine pharmacokinetics, host toxicity, urinary concentrations, treatment outcomes or the possibility of interactions with conventional drugs in patients.

For that reason, the results are best understood as a chemical and activity map for future research. The strongest candidates identified for isolation include quercetin 3-O-rutinoside, trihydroxy-octadecaenoic acid, hydroxy-octadecatrienoic acid, hydroxy-octadecadienoic acid and eucommicin A, together with selected coumarins and chlorin-related metabolites. The authors propose purifying these compounds, confirming their structures with authentic standards and complementary analytical methods, and testing them individually and in combinations against multidrug-resistant uropathogenic E. coli. Mechanistic work, cytotoxicity and selectivity assessments, and testing in appropriate in-vivo urinary tract infection models would still be required before any therapeutic or preventive application could be considered.

Subject of Research: Metabolite profiles associated with antibacterial and antibiofilm activity of celery, coriander and parsley against multidrug-resistant uropathogenic E. coli

Article Title: Integrative metabolomics and chemometric analysis reveal metabolites associated with antibacterial and antibiofilm activities of celery, Coriander, and parsley against multidrug-resistant uropathogenic Escherichia coli

Article References: Abdel-dayem, S. I. A., El Gedaily, R. A., Eltanany, B. M., Seiffein, N. L., Hani, G. A., Afifi, S. M., & Ali, D. E. (2026). Integrative metabolomics and chemometric analysis reveal metabolites associated with antibacterial and antibiofilm activities of celery, Coriander, and parsley against multidrug-resistant uropathogenic Escherichia coli. BMC Complementary Medicine and Therapies, 26(1), Article 313. https://doi.org/10.1186/s12906-026-05610-0

Image Credits: AI Generated

DOI: 10.1186/s12906-026-05610-0

Keywords: Uropathogenic E. coli, Multidrug resistance, Antibiofilm activity, Coriander, Metabolomics, Plant extracts, Quercetin, Fatty acids, Urinary tract infections, Integrative, chemometric, analysis