Potential of isoquinoline-alkaloid-rich Papaveraceae extracts against bacteria causing ocular infections – In vitro and In vivo study

Abstract

Introduction

There is an evident need for novel ophthalmic therapeutics that operate outside the conventional antibiotic and chemically synthetized antiseptic paradigms. Medicinal plants represent a chemically rich and cost-effective source of bioactive compounds, though their use requires careful toxicological assessment.

Methods

In this study, extracts from four Papaveraceae species - Chelidonium majus (roots), Corydalis cheilanthifolia (aerial parts and roots), Glaucium flavum (aerial parts and roots), and Fumaria officinalis (herb) - were investigated across three experimental levels: in silico, in vitro, and in vivo, using the Galleria mellonella larval model.

Results

Chemical profiling identified several isoquinoline alkaloids, including berberine, coptisine, protopine, chelidonine, allocryptopine, glaucine, and tetrahydropalmatine. In silico assessment suggested low systemic toxicity but possible blood-brain barrier permeability for some compounds. In vitro antimicrobial testing against Staphylococcus aureus ATCC 6538 and Pseudomonas aeruginosa ATCC 15442, performed using broth microdilution and bacterial-cellulose-based diffusion assays, demonstrated moderate but reproducible activity of extracts, particularly in Ch. majus roots (MIC equal 1.56 mg/mL for S. aureus and 6.25 mg/mL for P. aeruginosa) and in C. cheilanthifolia, where depending on the type of extract (herb_2, herb_1, root), range of MIC was 0,78–1.56 mg/mL for S. aureus and 3,12–6,25 mg/mL for P. aeruginosa. Cytotoxicity testing on fibroblast cell lines showed no significant toxicity in applied concentrations. In vivo testing showed that all extracts, except C. cheilanthifolia (herb_1) and Ch. majus (root) were virtually non-lethal to G. mellonella larvae, indicating a favorable safety profile in 0.05 mg/mL to 0.002 mg/mL concentrations. To evaluate their suitability for ocular application, wettability tests were performed using commercial tear substitute formulations. The incorporation of the C. cheilanthifolia herb_2 extract at bactericidal concentrations did not alter these physicochemical properties.

Discussion

Overall, the Papaveraceae extracts demonstrated moderate antibacterial activity and low cytotoxicity in applied concentrations, supporting their potential use as adjuvant components in eye drop formulations - enhancing therapeutic efficacy, reducing required antiseptic dosage, and potentially lowering the risk of resistance development.

1. Introduction

Antimicrobial resistance (AMR) has emerged as one of the most pressing global health threats of the 21st century. According to the landmark Lancet report prepared by the Global Burden of Disease AMR Collaborators, drug-resistant infections were directly responsible for an estimated 1.27 million deaths in 2019 and contributed to nearly 5 million more, surpassing the mortality burden of HIV/AIDS or malaria in the same period (Naghavi et al., 2024). This escalating crisis is driven by the remarkable genetic plasticity of bacteria, which enables the rapid acquisition of resistance determinants via horizontal gene transfer, genomic rearrangements, and mutational events. Selective pressure resulting from the excessive and often inappropriate use of antibiotics further accelerates the emergence of multidrug-resistant pathogens. In parallel with the erosion of antibiotic efficacy, antiseptics of synthetic origin have long been considered a robust alternative for topical infection control. Agents such as povidone-iodine, chlorhexidine, octenidine, or polyhexanide are widely used in clinical practice due to their broad activity spectrum, rapid bactericidal action, and initially presumed low propensity for resistance development. However, mounting evidence indicates that microorganisms exposed to repeated or sublethal doses of these agents may gradually acquire reduced susceptibility (Al-Doori et al., 2007; Hardy et al., 2018; Kim et al., 2018). Reports of Staphylococcus aureus or Pseudomonas aeruginosa isolates exhibiting decreased responsiveness to chlorhexidine highlight that selective pressure is not limited to antibiotics alone (Hassan et al., 2013; Cieplik et al., 2019). Although the mechanisms underlying this phenomenon are multifactorial, ranging from efflux pump activation to alterations in cell envelope composition, the clinical consequence remains the same: antiseptics, once considered resistance-proof, are in the process of loosening their effectiveness. This underscores the urgent need to extend antimicrobial discovery efforts beyond conventional chemotherapeutics. These limitations necessitate the development of a “third line of defense” against resistant microorganisms. Beyond antibiotics and classical antiseptics (the first and the second line of defense, respectively) promising candidates include antimicrobial peptides of diverse origins and bioactive plant-derived products. Antimicrobial peptides represent an evolutionary defense strategy found in virtually all kingdoms of life, characterized by rapid, non-specific disruption of microbial membranes and additional immunomodulatory effects (Chakraborty et al., 2022). Likewise, plants produce a vast and chemically diverse arsenal of specialized metabolites, such as alkaloids, terpenoids, polyphenols (e.g., flavonoids, tannins, stilbenoids) and other phytochemical classes - shaped by millions of years of co-evolution with microbial allies (microbiota) and antagonists (pathogens) (Anjali et al., 2023). This coevolution-driven “arms race” yields molecules that frequently act on microbial communication (quorum sensing) and pathogenicity (for example, by blocking acyl-homoserine lactone signaling, down-regulating virulence genes, or disrupting biofilm architecture) rather than solely killing cells, which can lower the selective pressure for resistance. Well-documented examples include ajoene from garlic, which inhibits quorum-sensing regulated genes and biofilm formation, a broad family of phenolic compounds and flavonoids (e.g., quercetin, chlorogenic acid) that interfere with AHL-dependent signaling and biofilm development, and terpene-rich essential oil constituents (such as linalool and related terpenoids) that compromise biofilm integrity and bacterial membrane function (Nakamoto et al., 2020; Deryabin et al., 2019; Aditi et al., 2020; Saini et al., 2022). These anti-virulence and antibiofilm activities, reviewed extensively in the recent literature, make plant metabolites attractive candidates for sustainable, biocompatible adjuvants to conventional antimicrobials. At the same time, translation to human therapeutics requires rigorous toxicology, local pharmacokinetic profiling (absorption, retention and metabolism at the application site), and formulation work to address issues such as stability and poor oral/ocular bioavailability (Ezrari et al., 2025).

In this context, the eye provides a clinically relevant and highly sensitive model for exploring novel antimicrobial strategies. Ocular infections such as conjunctivitis, keratitis, blepharitis, and endophthalmitis are frequently associated with two major opportunistic pathogens: S. aureus and P. aeruginosa. The first of bacteria mentioned - a Gram-positive coccus - is notorious for its dual role as a commensal colonizer and an aggressive pathogen, increasingly implicated in purulent conjunctivitis and postoperative infections. P. aeruginosa, in turn, is a Gram-negative bacillus characterized by extreme metabolic adaptability, rapid development of multidrug resistance, and a strong predilection for colonizing abiotic surfaces such as contact lenses and intraocular implants (Willcox, 2011; Teweldemedhin et al., 2017; O’Callaghan, 2018). Both species exhibit a remarkable ability to form biofilms - structured, extracellular-matrix-embedded microbial communities that confer enhanced tolerance to antibiotics, antiseptics, and host immune defenses (Uruén et al., 2020). Clinically, biofilm formation underlies the persistence and recurrence of ocular infections, complicates eradication efforts, and substantially increases the risk of vision-threatening sequelae. As such, ocular pathogens and their biofilm-associated phenotypes represent a stringent testing ground for evaluating the efficacy, safety, and translational potential of plant-derived antimicrobials.

Any attempt to utilize plant-derived antimicrobials as ophthalmic medication must be accompanied by careful toxicological scrutiny. The ocular surface is composed of highly specialized, delicate epithelial and stromal cells whose integrity is essential for maintaining vision. Even low cytotoxicity can result in irritation, impaired tear film stability, or delayed epithelial regeneration, while more pronounced toxicity may lead to corneal ulceration or permanent visual impairment. Unlike the skin or mucosa, ocular tissues are uniquely vulnerable due to their continuous exposure, high innervation, and limited regenerative capacity. Consequently, compounds with promising antimicrobial activity must be tested in parallel for their compatibility with human corneal and conjunctival epithelial cells, fibroblasts, and keratocytes (Müller and Kramer, 2008). Furthermore, ocular pharmacokinetics differ substantially from systemic or dermal exposure: absorption is modulated by tear turnover, blinking, and the barrier properties of the corneal epithelium. Therefore, safety assessment cannot rely solely on systemic toxicological data, but must specifically address local tolerability and bioavailability in the ophthalmic setting.

Formulation represents a critical step in implementation of plant-derived products into ophthalmic therapy. Topical eye drops remain the most widely used route of administration in ophthalmology, yet they suffer from rapid clearance through tear turnover and nasolacrimal drainage, resulting in short residence times and limited bioavailability. To achieve therapeutic efficacy, formulations must therefore balance physicochemical stability with favorable rheological properties that prolong ocular surface retention without compromising patient comfort or vision clarity. Approaches such as viscosity enhancers, mucoadhesive polymers, and in situ gelling systems have been extensively studied to optimize residence time. In this context, bacterial nanocellulose (BNC) polymer emerges as a promising next-generation carrier: it is biocompatible, highly hydrated, mechanically stable, and can serve as a depot for sustained release of hydrophilic and hydrophobic phytochemicals alike (Campano et al., 2025). Its transparent, non-irritating nature makes it ideally suited for ocular applications, where optical clarity and biocompatibility are essential. Embedding plant extracts into BNC-based formulations could enable controlled release and sustained antimicrobial activity, while paving the way for multifunctional ophthalmic preparations that integrate biofilm inhibition with anti-inflammatory and antioxidant effects.

Hence, plant natural products emerge as rational candidates for integration into ophthalmic formulations as most of the conventional antibiotics and synthetic antiseptics face escalating limitations. Some medicinal plants of the Papaveraceae family produce considerable amounts of isoquinoline alkaloids (IQA) with documented antimicrobial and anti-inflammatory potential (Zhang et al., 2024). Corydalis cheilanthifolia Hemsl., traditionally used in Asian phytotherapy for pain relief and wound healing, contains protoberberine and aporphine alkaloids such as corydaline and cheilanthifoline. Preliminary studies indicate antibacterial and antioxidant activity, although data on ocular applications remain scarce, whereas the reported toxicity appears generally low, with only sporadic suggestions of hepatotoxicity in other Corydalis species (Engman et al., 2023). Chelidonium majus L (greater celandine), was reported since antiquity to cure blindness and other eye ailments and has been widely used in European folk medicine for the treatment of skin diseases, warts, and liver complaints (Zielińska et al., 2018). Its distinctive alkaloid profile includes chelidonine, sanguinarine, and chelerythrine and several other IQAs. These compounds exhibit strong antimicrobial and antibiofilm properties but the need for careful safety assessment in ocular contexts using credible approach is necessary (Greater Celandine, 2022; Li et al., 2024). Fumaria officinalis L., traditionally used in gastrointestinal and dermatological disorders, is rich in protopine and other IQA typical for the Fumarioideae subfamily and demonstrates spasmolytic, cholagogue, and antimicrobial effects. Toxicity is generally considered low, although hepatobiliary interactions cannot be excluded, and ophthalmic safety data remain limited (Ahmoda et al., 2025). Glaucium flavum Crantz, has been used in Mediterranean folk medicine for its expectorant and antitussive effects. Glaucine and related aporphine alkaloids from this plant display antimicrobial, bronchodilatory, and anti-inflammatory properties. At the same time, glaucine has been reported to cause central nervous system effects at higher systemic doses, and its local tolerability on epithelial cells has not been systematically evaluated (Akaberi et al., 2021). Collectively, these species represent rational candidates for exploratory ophthalmic formulations, combining evolutionary antimicrobial defenses with diverse phytochemical repertoires, yet their established or suspected toxicities emphasize the necessity of rigorous cytotoxicity testing on ocular cell lines prior to translational application.

In light of these considerations, the present study was designed to evaluate the potential of IQA from Papaveraceae plants extracts in vitro and in vivo as defense agents against ocular pathogens, resistant to conventional therapies. In this work, extracts obtained from C. cheilanthifolia, Ch. majus, F. officinalis, and G. flavum were subjected to chromatographic profiling (HPLC) to characterize their chemical composition, followed by in silico toxicological analysis and microbiological testing against S. aureus and P. aeruginosa in both planktonic and biofilm forms. In parallel, cytotoxicity assays in vitro and in vivo were performed to ensure safety and tolerability, while formulation-oriented analyses explored the feasibility of incorporating these compounds into eye drop vehicles, specifically, bacterial nanocellulose-based systems. By integrating chemical, microbiological, toxicological, and formulation perspectives, this study aims to provide an attempt for assessing the translational potential of these traditional medicinal plants in ophthalmology.

2. Materials and methods

2.1. Plant materials

Plants were collected as shown in Table 1, dried at 35 °C (Counter Intelligence, D-CUBE, LD-9013A dehydrator), and stored in airtight containers at room temperature till the extraction.

TABLE 1.

Origin of Papaveraceae plant material used in the present study. ‘BGMP’– Botanical Garden of Medicinal Plants at the Wrocław Medical University, ‘Wilamowa’ – an experimental garden located in Wilamowa, Paczkow community, Nysa County, Poland. Herb–aerial parts containing green stems, leaves and flowers according to the pharmacopoeia guidelines.

No.	Species	Part	Plant	GPS
1	Chelidonium majus	Herb, root	BGMP	51.11737N 17.07483E
2	Corydalis cheilanthifolia	Herb (H1), root	Wilamowa	50.47449N 17.07382E
3	Corydalis cheilanthifolia	Herb (H2)	BGMP	51.11737N 17.07483E
4	Glaucium flavum	Herb, root	BGMP	51.11737N 17.07483E
5	Fumaria officinalis	Herb	Wrocław, Niskie Łąki, spontaneous ruderal habitat	51.09968N 17.06058E

The plant specimens were identified by the Botanical Garden of Medicinal Plants staff botanists–Agata Biernat and Klemens Jakubowski. The botanical binomial nomenclature follows RBG Kew “Plants of The World” database - https://powo.science.kew.org/ (last accessed on July 10th 2026).

The voucher specimens were deposited in the herbarium of the Botanical Garden of Medicinal Plants of the Wroclaw Medical University under the codes:

Ch.m._R_1–10/2023, Ch. m._H_1–10/2019, C. ch._R_1–10/2021, C. ch._H_1–10/2021, C. ch._H_1–10/2022, G. f._R_1–3/2023, G. f._H_1–3/2023, F. o._H_1–3/2023.

As a comparison plant material, a commercial sample of Euphrasiae herba–a traditional herbal drug used in Europe against eye complaints was purchased from Kawon-Hurt SJ (Authorization no. 2247, Gostyń, Poland).

2.2. Extraction and HPLC analysis

Plant material was crushed and powdered using a mortar and pestle and extracted with the mixture of MeOH (Stanlab, Poland) and 0.1% formic acid (Supelco, Germany) in a ratio 4:1 (v/v). Samples were extracted with ten parts of solvent per plant part. Half the volume of solvent was added to the plant material, sonicated (InterSonic, Poland) for 30 min at 25 °C, and filtered through filter paper. The remaining solvent was added to the residual sediment, left for 24 h at 4 °C, sonicated (30 min), and filtered (Chempur, Poland). The extracts were combined and dried on a rotary evaporator (Buchi, Switzerland). The dried residue was dissolved in 10 mL of MeOH, transferred to test tubes and evaporated under the stream of nitrogen/lyophilized, and stored at 4 °C until analysis. Dry extract was dissolved in DMSO (Sigma, Germany) in a concentration 1 mg/mL and further diluted with extraction solvent in a ratio 1:1. The High-Performance Liquid Chromatography with Diode-Array Detection (HPLC-DAD) analysis was performed using Agilent Technologies 1,260 Infinity II chromatographic system (Agilent Technologies, Palo Alto, CA) coupled with Infinity War diode array detector. The separation was carried out using Kinetex C18 column (2.1 mm × 100 mm, 2.6µm, Phenomenex, United States). The mobile phase consisted of A: 10 mM water solution of ammonium acetate (Fischer Chemical, Belgium) adjusted to pH 4 with glacial acetic acid ((Fischer Chemical, Belgium) and B: ACN (Merck, Germany). The elution program was as follows: 0–8 min 15–22%B; 8–18 min 22%B; 18–25 min 22–30%B; 25–28 min 30%B. Flow rate was 0.35 mL/min and the column was maintained in 25 °C (±1 °C). Detection was set at 280 nm. Identification of compounds was based on the comparison of the actual retention time of authentic reference standards (Chelidonine (later abbreviated as “CHD”), sanguinarine (SNG), berberine (BER), chelerithrine (CHL), protopine (PRO) – Extrasynthese, France; Allocryptopine (ACP) – Sigma-Aldrich, Germany; coptisine (COP) - PhytoLab GmbH, Germany). Quantification of the major alkaloids was performed using the external standard method. Calibration curves were prepared for the available reference standards using five concentrations (10–50 μg/mL), and analyzed under the same chromatographic conditions as the plant extracts. The calibration plots were constructed by relating peak area to analyte concentration, and the contents of individual alkaloids in the samples were calculated from the corresponding calibration equations (Supplementary File). The calibration curves showed good linearity within the investigated concentration ranges, with correlation coefficients (R ²) exceeding the acceptable criteria (Supplementary File). The quantitative results presented in Table 2 were calculated based on these calibration curves.

TABLE 2.

Identified alkaloids and their concentrations (µg/mg DW ± SD) in extracts of Papaveraceae species (herb and root fractions). DW -dry weight; SD - standard deviation. H–herb (aerial parts); R-root. H1 – origin Wilamowa, H2 – origin BGMP Wroclaw.

Extract	Alkaloid	Average (µg/mg DW) ±SD; n = 3
C. cheilanthifolia_H1	Protopine	8.24 ± 0.08
	Coptisine	4.14 ± 0.02
	Protopine derivative	8.11 ± 0.32
	Berberine	10.48 ± 0.28
C. cheilanthifolia_H2	Protopine	13.37 ± 0.14
	Protopine derivative	12.34 ± 0.10
	Berberine	7.32 ± 0.15
C. cheilanthifolia_R	Protopine	35.92 ± 0.22
	Coptisine	11.56 ± 0.11
	Protopine derivative	11.18 ± 0.20
Ch. majus_R	Protopine	44.84 ± 0.61
	Allocryptopine	5.62 ± 0.09
	Chelidonine	3.38 ± 0.12
	Coptisine	69.82 ± 0.55
	Berberine	5.45 ± 0.02
F. officinalis_H	Protopine	4.12 ± 0.07
G. flavum_H	Glaucine derivative 1	3.15 ± 0.06
	Glaucine derivative 2	10.03 ± 1.16
	Glaucine	170.04 ± 20.36
G. flavum_R	Glaucine derivative	3.08 ± 0.03
	Protopine	61.16 ± 0.23
	Chelidonine	4.64 ± 0.07
	Tetrahydropalmatine	3.67 ± 0.09

predicted lethal dose (LD50, mg/kg) [ProTox III]
PRO	BER	CHD	GLA	COP	ACP	THP
940	200	460	400	200	940	400

The data were acquired and processed using Agilent OpenLab Chemstation ver. 2.19.20. The references and data used for the identification of the investigated compounds is given in Supplementary Data 1.

2.3. The following bacterial strains were applied in this experiment

S. aureus 6538 and P. aeruginosa 15542 from American Type Culture Collection (ATCC, Manassas, VA, United States), Komagataeibacter xylinus ATCC 53524 (American Type Culture Collection, Manassas, VA, United States).

2.4. In silico ADME profiling of alkaloids detected by HPLC

The pharmacokinetic properties of alkaloids identified in plant extracts were evaluated using the SwissADME web tool (http://www.swissadme.ch Swiss Institute of Bioinformatics, Lausanne, Switzerland). For each compound, the canonical SMILES notation was retrieved from the PubChem database (National Center for Biotechnology Information, Bethesda, MD, United States of America) and introduced as input. SwissADME provided predictions of physicochemical descriptors (molecular weight, number of rotatable bonds, topological polar surface area [TPSA], hydrogen bond donors and acceptors), lipophilicity (consensus logP), water solubility, and drug-likeness filters (Lipinski, Ghose, Veber, Egan, and Muegge rules). Absorption, distribution, metabolism, and excretion (ADME) parameters were evaluated, including gastrointestinal (GI) absorption, blood–brain barrier (BBB) permeability, and substrate/inhibitor status for P-glycoprotein (P-gp) and major cytochrome P450 (CYP) isoforms (CYP1A2, CYP2C19, CYP2C9, CYP2D6, and CYP3A4). In addition, the BOILED-Egg model implemented in SwissADME was applied to visualize passive gastrointestinal absorption and BBB penetration. Predicted pharmacokinetic profiles were exported and summarized for all alkaloids, enabling comparative evaluation of their oral bioavailability and potential for systemic distribution.

2.5. In silico prediction of alkaloids toxicity

The toxicological profiles of selected alkaloids identified in Papaveraceae extracts were assessed using the ProTox-III webserver (https://tox.charite.de/protox3/) Charité–Universitätsmedizin Berlin, Germany). Canonical SMILES notations of each alkaloid were retrieved from the PubChem database (National Center for Biotechnology Information, Bethesda, MD, United States of America) and submitted as input. For each molecule, ProTox-II provided predicted values of acute oral toxicity (LD₅₀ in mg/kg body weight) and toxicity class according to the Globally Harmonized System (GHS) of classification. In addition, endpoints covering hepatotoxicity, cytotoxicity, carcinogenicity, mutagenicity, immunotoxicity, and Tox21 pathway activation (nuclear receptor signaling and stress response pathways) were generated. Probabilities of compound assignment to toxic versus non-toxic categories were reported as confidence scores. The resulting predictions were compiled for all tested alkaloids. This approach allowed a rapid, systematic estimation of the toxicological risk profile of alkaloids prior to further in vitro and in vivo validation.

2.6. Estimation of minimal inhibitory concentration of plant extracts using microdilution method

Dried plant extracts were dissolved in DMSO (analytical grade, Sigma-Aldritch, Germany) to obtain 5 mg/mL stock solutions, which were subsequently diluted twice with sterile deionized water to working intermediates and finally subjected to two-fold serial dilution in cation-adjusted Mueller–Hinton broth (CA-MHB). The final DMSO content in assay wells did not exceed 1% (v/v), and solvent controls were included on every plate. Bacteria were grown overnight on TSA, suspended in saline, and adjusted to 0.5 McFarland using a densitometer (DEN-1B, SIA Biosan, Latvia), followed by dilution into CA-MHB to yield a final inoculum of approximately 5 × 10^5 CFU/mL per well. Minimum inhibitory concentrations (MICs) were determined in 96-well flat-bottom plates (Wuxi Nest Biotechnology, China) by the standard broth microdilution method. Plates were inoculated and incubated at 37 °C for 24 h. Following incubation, 2,3,5-triphenyl-tetrazolium chloride (TTC; 0.05%–0.10% w/v, Chempur, Poland) was added to each well, and plates were further incubated for 30 min. The MIC was defined as the lowest extract concentration at which no red coloration developed, indicating the absence of TTC-formazan formation. The result was verified by spot method. Each assay plate included growth, sterility, and solvent controls. All experiments were carried out in three independent biological replicates on separate days, with 3 technical duplicates included in each run. MIC values were recorded as the modal outcome across repeats; in cases where values differed by one dilution step, the higher MIC was reported.

2.7. Estimation of minimal biofilm eradication concentration of plant extracts using the microdilution method

Overnight cultures grown on tryptic soy agar were suspended in saline and adjusted to 0.5 McFarland (DEN-1B, SIA Biosan, Latvia), then diluted into CA-MHB to obtain ∼5 × 10^5 CFU/mL. Biofilms were established in 96-well flat-bottom polystyrene plates (Wuxi Nest Biotechnology, China) by inoculating 100 µL per well and incubating at 37 °C for 24 h without agitation. After incubation, medium was carefully aspirated, wells were gently rinsed once with sterile PBS to remove non-adherent cells, and 200 µL of fresh CA-MHB containing the extract dilution series (or controls) was added to each well. Plates were incubated a further 24 h at 37 °C. Following treatment,; 0.05%–0.10% TTC w/v) was added to each well and plates were incubated for 30 min. The MBEC was defined as the lowest extract concentration at which no red coloration developed, indicating absence of TTC-formazan formation by the biofilm. In parallel, eradication was corroborated by viable-count “spot” readout following biofilm disruption (PBS rinse, mechanical scraping of the well contents, then spotting on Mueller–Hinton agar). Each plate included growth, sterility, and solvent controls. Experiments were performed as three independent biological replicates on separate days, each with 3 technical duplicates. When replicate MBECs differed by one two-fold dilution, the higher value was reported.

2.8. Preparation of bacterial nano-cellulose carrier containing the plant extracts

Bacterial nano-cellulose (BNC) discs were produced by Komagataeibacter xylinus ATCC 53524 (American Type Culture Collection, Manassas, VA, United States of America) cultured in Hestrin–Schramm (HS) medium under static conditions at 28 °C for 7 days. Following incubation, 11 mm in diameter pellicles were harvested, and purified by immersion in 0.1 M NaOH (Chempur, Poland) at 80 °C until all bacterial cells and debris were removed. The discs were then thoroughly rinsed with sterile water until neutral pH was achieved. Sterilization was performed by autoclaving at 121 °C for 15 min. Prior to use in antimicrobial assays, BNC discs were equilibrated in sterile deionized water and stored at 4 °C under aseptic conditions. For disk diffusion testing, sterile discs were transferred into wells of sterile 24-well plates and soaked with plant extract solutions for 24 h at room temperature, during which each disc absorbed on average 600 µL of extract solution, displacing the water naturally present in the hydrated matrix.

2.9. Evaluation of antibacterial activity of extracts using modified disk diffusion assay

Antimicrobial activity of the extracts was assessed using a modified disk diffusion assay in which BNC discs replaced conventional paper carriers. BNC discs with a diameter matching the wells of 24-well plates (≈11 mm) were first sterilized and then saturated with plant extracts by incubation for 24 h in extract solutions prepared from DMSO stocks diluted in sterile deionized water, which effectively displaced the water naturally present in the hydrated BNC matrix. The introduced concentration corresponded to 2× the MIC value determined in the broth assay. When the MIC was higher than the highest concentration tested in the broth assay, the maximum tested concentration of 12.5 mg/mL was used. After impregnation, discs were gently blotted under sterile conditions to remove surface excess liquid and immediately transferred onto Mueller–Hinton agar plates (Biomaxima, Poland) previously inoculated with standardized bacterial suspensions. The inoculum was prepared from overnight cultures of S. aureus ATCC 6538 and P. aeruginosa ATCC 15442 (American Type Culture Collection, Manassas, VA, United States), adjusted to 0.5 McFarland with a DEN-1B densitometer (SIA Biosan, Latvia), and uniformly spread onto the agar surface. Plates with BNC discs were incubated at 37 °C for 18–20 h. After incubation, the inhibition zones were photographed under standardized lighting conditions and analyzed using ImageJ software (NIH, United States). The measurement scale was calibrated for each image using the “Set Scale” function based on a millimeter reference. The total area of bacterial nano-cellulose (BNC) discs was subtracted from the measured inhibition halo to quantify only the zone of microbial growth inhibition. Final results were expressed as inhibition area (mm²). Each plate included solvent controls (BNC discs soaked in DMSO at the highest applied concentration), sterility controls, and reference antimicrobials. All experiments were performed in two independent biological replicates, each in three technical duplicates, and data were presented as mean inhibition area ± standard deviation.

2.10. Exploratory composition–activity deconvolution of alkaloid-rich extracts

Exploratory composition–activity deconvolution was performed using net inhibition areas obtained in the modified BNC disk-diffusion assay for S. aureus and P. aeruginosa. This analysis was designed as a hypothesis-generating comparison of activity patterns among alkaloid-rich extracts differing in chemical composition. It should not be interpreted as formal pharmacodynamic interaction testing of purified compounds at defined concentrations. For each extract, six independent replicates were photographed and quantified in ImageJ. The area corresponding to the 12 mm cellulose carrier was subtracted from each measurement to obtain the net inhibition area. These net inhibition areas were then used for exploratory comparison between extracts containing different alkaloid patterns. Because the tested extracts naturally differed in their alkaloid composition, they were treated as composition patterns rather than as experimentally prepared mixtures of purified single agents. Putative marker-contribution values were assigned directly when a given extract was dominated by one major alkaloid marker, including protopine from F. officinalis herb, chelidonine from a Ch. majus extract available in the DoTT collection (Supplementary Data), and glaucine from G. flavum herb. When direct marker-dominant profiles were not available, tentative contribution values were inferred by comparing extracts obtained within the same species and experimental run but differing in alkaloid composition. Thus, coptisine contribution was inferred by comparing C. cheilanthifolia H1, containing coptisine, protopine, and berberine, with C. cheilanthifolia H2, containing protopine and berberine. Berberine contribution was inferred by comparing the C. cheilanthifolia H2 profile with the protopine marker profile. Tetrahydropalmatine contribution was inferred by comparing G. flavum root, containing protopine, chelidonine, tetrahydropalmatine, and glaucine, with the available protopine, chelidonine, and glaucine marker profiles. Allocryptopine was treated as an unresolved component because no reliable marker-dominant profile was available.

For descriptive normalization, the highest net inhibition area observed within each bacterial species was used as the reference maximum. Each putative marker-contribution value was then expressed as a fraction of this maximum. Two simple reference expectations were calculated. The first assumed that individual marker contributions combine multiplicatively; in this approach, the expected fractional activity of a given composition pattern was calculated as 1 − ∏(1 − fractional marker contribution), and then converted back to net inhibition area by multiplying by the reference maximum. The second reference expectation assumed that the activity of a composition pattern is driven mainly by the strongest individual marker contribution; in this approach, the expected value was equal to the highest putative marker-contribution value present in the given composition pattern. For each extract composition pattern, the observed net inhibition area was compared with both reference expectations. The difference between the observed and expected values was described using a symmetric tolerance of ±10% of the observed value. Values above this threshold were described as positive deviations from the reference expectation, values within the threshold as no relevant deviation, and values below this threshold as negative deviations.

2.11. Assessment of in vitro cytotoxicity of extract toward L929 cell using neutral MTT test

Dulbecco’s modified Eagle’s medium (DMEM, Biowest, Riverside, MO, United States) supplemented with 1% (v/v) amphotericin B (Biowest, Riverside, MO, United States), 1% (v/v) penicillin with streptomycin (Biowest, Riverside, MO, United States), and 10% (v/v) fetal bovine serum FBS (Biowest, Riverside, MO, United States) was used for cells’ culturing. A suspension of the L929 murine fibroblast cell line (ATCC CCL) at a density of 1 × 10^5/mL was prepared, and 100 µL was added to the wells of 96-well plates (VWR, Radnor, PA, United States). The plates were incubated for 24 h/37 °C, 5% CO2 (Binder C 150 UL CO2 incubator, Germany). Next, the medium was gently removed. The stock solutions of the compounds were geometrically diluted in medium at the concentration ranging from 0.05 mg/mL to 0.002 mg/mL and transferred to the cells at a volume of 100 µL. The following control samples were prepared: growth control and control of DMSO (dimethyl sulfoxide, Sigma Aldrich, Germany) geometrically diluted in medium at concentrations ranging from 1% (v/v) to 0.03% (v/v). The plates were retransferred to the incubator for 24 h. After that, the medium was removed and 100 µL of the solution of 1 mg/mL MTT ((3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide) (Sigma Aldrich, Germany) in DMEM was added for 2h, and the plates were incubated under the same conditions. Subsequently, the dye was removed, and 100 µL DMSO was added for 30 min to dissolve crystals (shaking at 400 rpm (Plate Shaker-Thermostat PST-60HL-4 (Biosan, Latvia)). The absorbance of the solution was then measured at 570 and 630 nm with a spectrophotometer (Multiscan® GO, Thermo Scientific, United States of America). The values at 630 nm were subtracted from the values at 570, and the viability of the treated cells was calculated as follows: Cells Viability [%] = (AbS/Ab C) x 100%; AbS- mean absorbance value of treated samples; AbC- mean values of the growth control. The experiment was performed in three independent experiments, each with 4 repetitions.

2.12. Assessment of in vivo cytotoxicity using Galleria mellonella larval model

The in vivo efficacy and toxicity of the extracts were examined using larvae of G. mellonella reared in the P.U.M.A. laboratory (Wroclaw Medical University, Poland). The larvae originated from a standardized in-house culture maintained under controlled conditions (30 °C, constant humidity, darkness) and fed a uniform artificial diet to ensure physiological consistency across experiments. Groups of six larvae per compound (weighing 250–300 mg each, free of melanization or physical damage) were injected into the last left proleg with 10 µL of extract solution using a Hamilton syringe (Hamilton, Bonaduz, Switzerland). Control groups received either phosphate-buffered saline or DMSO at the corresponding maximal concentration. A 75% ethanol solution was used as a positive (toxicity) control to verify test responsiveness. Larvae were incubated at 37 °C in the dark and monitored for survival every 24 h for up to 96 h.

2.13. Commercial ophthalmic preparations

Three commercially available eye drop formulations were included in the study. Dexoftal® (TEGE Pharma B.V, Breda, Netherlands), hereafter referred to as DX, contains dexpanthenol (2%) polihexanid 0.0001% and hydroxypropylcellulose (HydraFlex™, 0.5%) as viscosity enhancer and mucoadhesive polymer. Thealoz Duo® (Laboratoires Théa, Clermont-Ferrand, France), abbreviated as TD, is composed of trehalose (3%), sodium hyaluronate (0.15%), trometamol, sodium chloride, hydrochloric acid, and water for injection. Hialeye Free® (Adamed Pharma, Czosnow, Poland), denoted as HF, contains sodium hyaluronate (0.2%), sodium phosphate dodecahydrate, sodium dihydrogen phosphate monohydrate, sodium chloride, and water. All products were used within their shelf life and handled under sterile conditions.

2.14. The wettability of ophthalmic preparations

The wettability of the tested ophthalmic formulations (DX, TD, HF) was evaluated by contact angle analysis on surfaces with different physicochemical properties. Droplets of 40 µL were dispensed using a 100 µL microsyringe (Hamilton, Reno, NV, United States) onto three substrates: (i) hydrophobic poly (L-lactic acid) (PLLA) plates (Goodfellow, Cambridge, UK), (ii) hydrophilic glass slides (Thermo Fisher Scientific, Waltham, MA, United States), and (iii) a mucin-containing artificial tear film model. The artificial tear film consisted of sodium chloride (6.8 g/L), potassium chloride (1.4 g/L), calcium chloride dihydrate (0.08 g/L), magnesium chloride hexahydrate (0.1 g/L), sodium bicarbonate (2.2 g/L), glucose (1 g/L), and mucin (0.5% w/v; porcine stomach type II, Sigma-Aldrich, St. Louis, MO, United States) dissolved in phosphate-buffered saline (PBS, pH 7.4). Measurements were performed at 24 °C under ambient laboratory conditions. For each formulation–substrate combination, at least six replicate measurements were performed. Images of sessile drops were captured under standardized conditions, and contact angles were calculated using ImageJ software (NIH, Bethesda, MD, United States of America) with the Contact Angle plug-in developed by Marco Brugnara and based on the PointPicker plug-in.

2.15. Combination of plant extracts with chosen commercial eye drop

To assess the feasibility of incorporating plant extracts into existing ophthalmic vehicles, C. cheilanthifolia H2 extract was combined with DX tear drop. The extract was added to the formulation to achieve a final concentration corresponding to 2 × MIC, as determined in the broth microdilution assay. The resulting mixture was used to saturate sterile BNC discs for 24 h, as described previously. For comparison, parallel BNC discs were prepared with the same extract diluted in 1% DMSO and water; another control was BNC saturated with 1% DMSO and water. Antimicrobial activity was subsequently evaluated using the modified disk diffusion method (Section Material and Methods 8).

2.16. Statistical analysis

Statistical analyses were performed using GraphPad Prism 10 (San Diego, CA, United States of America).

Normality of distribution was verified using Shapiro–Wilk’s test. An Analysis of Variance (ANOVA) was performed to assess statistical significance. For multiple comparisons, Tukey’s or Šídák’s post hoc test was applied. A p-value threshold of less than 0.05 was set for significance in the ANOVA. For the Tukey post hoc analysis, significance levels were further categorized as p < 0.001 and p < 0.0001 for specific pairwise comparisons.

3. Results

At the initial stage of the study, phytochemical profiling of extracts obtained from Papaveraceae species was performed in order to identify and quantify dominant alkaloids. HPLC analysis revealed the presence of composition of alkaloids depending on the plant species and the part of the plant (herb or root). The detected IQA included ACP, BER, GLA, COP, PRO, THP, as well as their derivatives. The concentrations of detected alkaloids are provided in Table 2.

Extracts from C. cheilanthifolia were characterized predominantly by protopine, berberine, and coptisine, with notably higher concentrations of protopine in the roots compared to the aerial parts. The extract of Ch. majus roots was dominated by coptisine and protopine, together with detectable allocryptopine, berberine and chelidonine. Fumaria officinalis extract was chemically simpler, with protopine as the only major alkaloid identified. In contrast, G. flavum displayed the highest total alkaloid content, with glaucine exceeding 170 μg/mg DW in the aerial part, and high protopine content (over 60 μg/mg DW) in the root. Collectively, these results indicate that each Papaveraceae species exhibits a distinct alkaloid fingerprint, with G. flavum herb and Ch. majus root extracts representing the richest sources. Berberine and coptisine were assigned the lowest LD₅₀ values (200 mg/kg), corresponding to GHS (Globally Harmonized System) class 3, while chelidonine (460 mg/kg), glaucine and tetrahydropalmatine (both 400 mg/kg) showed intermediate toxicity (class 3). Protopine and allocryptopine were classified as less toxic, with predicted LD₅₀ values of 940 mg/kg (class 4).

Next, the SwissADME BOILED-Egg analysis was perform to check the properties of above-enlisted active compounds in human body and revealed clear differences in the predicted pharmacokinetic profiles of the tested alkaloids (Figure 1). Allocryptopine, chelidonine, and protopine were located within the white region, indicating high probability of gastrointestinal absorption but no blood–brain barrier (BBB) penetration. In contrast, glaucine and tetrahydropalmatine clustered within the yellow region, consistent with favorable BBB penetration and thus potential neuroactive properties. Berberine and coptisine were positioned outside both regions, suggesting poor oral absorption and negligible BBB access. Regarding interaction with P-glycoprotein (P-gp), compounds represented as blue dots are predicted to be P-gp substrates and therefore actively effluxed, whereas red dots denote non-substrates with higher likelihood of intracellular accumulation.

FIGURE 1.

BOILED-Egg plots of Papaveraceae alkaloids generated with SwissADME. The position of each dot indicates the predicted pharmacokinetic behavior: location of dot within the yellow region (“yolk”) corresponds to a high probability of blood–brain barrier (BBB) permeability; within the white region (“egg white”) to a high probability of gastrointestinal absorption; and outside both regions to low probability of either property. Dot color indicates interaction with P-glycoprotein (P-gp): red dots represent compounds predicted not to be P-gp substrates, while blue dots represent compounds predicted to be P-gp substrates, i.e., subject to active efflux and limited central nervous system accumulation. Abbreviations: ACP–allocryptopine, BER–berberine, CHD–chelidonine, GLA–glaucine, COP–coptisine, PRO–protopine, THP–tetrahydropalmatine.

While the BOILED-Egg model provided insights into the pharmacokinetic behavior of the investigated alkaloids, pharmacological feasibility must be balanced with their toxicological safety. Therefore, in parallel with ADME profiling, we employed the ProTox-III platform to predict no only general toxicity (as presented in Table 1) but also organ-specific and systemic toxicities based on structural alerts (Figures 2, 3).

FIGURE 2.

Predicted organ-specific and systemic toxicity profiles of major Papaveraceae alkaloids based on ProTox-III analysis. Heatmap summarizing the predicted toxicological endpoints for alkaloids detected in Papaveraceae extracts. Red fields indicate a positive (toxic) prediction, while blue fields denote absence of toxic activity. The lower part of the diagram links individual compounds to the plant species and plant parts in which they were identified. Abbreviations: PRT - protopine; CPTZ - coptisine; BRBR - berberine; CHLD - chelidonine; ALLC - allocryptopine; THPL - tetrahydropalmatine; GLCN - glaucine; H - herb; R - root. Toxicity endpoints: hepato tox.- hepatotoxicity; neuro tox. – neurotoxicity; nephro tox. - nephrotoxicity; respiratory tox. - respiratory toxicity; cardio tox. - cardiotoxicity; carcinogenicity - carcinogenic potential; immuno tox. - immunotoxicity; mutagenicity - genotoxic potential; cyto tox. - cytotoxicity; BBB barrier - blood–brain barrier permeability; eco tox. - ecological toxicity; clinical tox. - predicted clinical toxicity; nutritional tox. - nutritional imbalance-related toxicity.

FIGURE 3.

Cumulative in silico toxicity profiles of Papaveraceae extracts predicted using the ProTox-III platform. Each column corresponds to an extract obtained from Fumaria officinalis (H–herb), Chelidonium majus (H–herb, R–root), Glaucium flavum (H, R), and Corydalis cheilanthifolia (H1, H2, R). Each horizontal red bar represents one alkaloid identified within a given extract that was predicted to be active for a specific toxicity endpoint. The number of red bars therefore reflects the number of alkaloids exhibiting the given type of toxicity, whereas blue fields indicate that none of the alkaloids present in the extract were predicted as toxic in that category. Evaluated endpoints include neuro tox. – neurotoxicity; resp. tox. – respiratory toxicity; carcino. gen. – carcinogenicity; immuno. tox. – immunotoxicity; mutagenicity–genotoxic potential; cyto. tox. – cytotoxicity; BBB–blood–brain barrier permeability; clinical tox. – predicted systemic clinical toxicity; eco. tox. – ecological toxicity; hepato tox. – hepatotoxicity; nephro tox. – nephrotoxicity; and cardio tox. – cardiotoxicity.

The in silico toxicity assessment demonstrated that Ch. majus (both herb and root) showed the highest cumulative number of predicted toxic endpoints, indicating the broadest toxicological spectrum among the tested extracts. Moderate levels of predicted toxicity were observed for G. flavum (herb and root) and C. cheilanthifolia (H1, H2, R), while F. officinalis exhibited the lowest overall toxicity, with few or no predicted toxic endpoints across the evaluated categories.

Having established the basic pharmacokinetic and toxicological characteristics of the major alkaloids in silico, providing a preliminary but valuable insight into their potential applicability, we subsequently moved to experimental validation in the laboratory. As a first step, the antimicrobial activity of Papaveraceae extracts was assessed by determining their minimal inhibitory concentrations (MICs) and minimal biofilm eradication concentrations (MBEC) against reference ocular pathogens using the microdilution method in 96-well plates.

The tested plant extracts demonstrated variable activity against S. aureus and P. aeruginosa. In general, S. aureus was more susceptible, with MIC values ranging from 0.78 to 6.25 mg/mL and MBECs between 3.12 and 12.5 mg/mL. Extracts from C. cheilanthifolia (herb and root) and G. flavum root exhibited the strongest activity against S. aureus, with MICs of 0.78–1.56 mg/mL and MBECs of 3.12–12.5 mg/mL. In contrast, P. aeruginosa displayed reduced susceptibility, with several extracts (e.g., Ch. majus herb, F. officinalis herb, G. flavum herb and root) failing to achieve inhibition or eradication within the highest concentrations tested (h.c.t.).

Among the active samples, C. cheilanthifolia herb and Ch. majus root were the most effective, with MICs of 3.12–6.25 mg/mL and MBECs of 12.5 mg/mL; moreover Gram-positive S. aureus was generally more sensitive to the plant extracts than Gram-negative P. aeruginosa.

Building on these observations, we next employed a modified disc diffusion method (MDDM), in which bacterial cellulose (BNC) served as a carrier matrix for controlled release of the plant extracts. The resulting inhibition zones were irregular in shape; therefore, their surface areas (mm²) were quantified using ImageJ software instead of relying solely on diameter measurements (Figures 4, 5).

FIGURE 4.

Antimicrobial activity of isochinolone alkaloid-rich extracts against S. aureus. Inhibition zones (mm²) were quantified from six independent replicates using ImageJ software. The area corresponding to the 12-mm cellulose carrier was subtracted from each measurement to obtain the net inhibition zone. Data are expressed as mean ± SD (n = 6). Statistical significance was determined by one-way ANOVA followed by Tukey’s multiple comparison test (p < 0.01 – **, p < 0.0001 – ****; only selected comparisons are selected). The qualitative alkaloid composition of each extract, determined by LC–MS analysis, is indicated by abbreviations: COP–coptisine, PRO–protopine, BER–berberine, CHEL–chelidonine, ACP–allocryptopine, THP–tetrhydropalmatine, GLA–glaucine. The values (mg/mL) in the bottom part of figure indicates the concentrations of extracts applied.

FIGURE 5.

Antimicrobial activity of isoquinolone alkaloid-rich extracts against P. aeruginosa. Inhibition zones (mm²) were quantified from six independent replicates using ImageJ software. The area corresponding to the 12-mm cellulose carrier was subtracted from each measurement to obtain the net inhibition zone. Data are presented as mean ± SD (n = 6). Statistical significance was determined by one-way ANOVA followed by Tukey’s multiple comparison test (p < 0.05 – *, only selected comparisons are indicated). The qualitative alkaloid composition of each extract, determined by LC–MS analysis, is indicated by abbreviations: COP–coptisine, PRO–protopine, BER–berberine, CHEL–chelidonine, ACP–allocryptopine, THP–tetrhydropalmatine, GLA–glaucine. The values (mg/mL) in the bottom part of figure indicates the concentrations of extracts applied.

The antimicrobial assay revealed distinct activity profiles of the tested Papaveraceae extracts against S. aureus and P. aeruginosa. Among all samples, G. flavum_R and C. cheilanthifolia_H1 exhibited the largest inhibition zones for both bacterial species, indicating the strongest antibacterial potential. Corydalis cheilanthifolia_H2 and C. cheilanthifolia_R showed moderate activity, whereas F. officinalis_H produced only small inhibition zones. Overall, the results indicate that root extracts generally exhibited higher activity than herb extracts within the same species, and that C. cheilanthifolia and G. flavum extracts demonstrated the broadest and most potent antibacterial effects across both Gram-positive and Gram-negative strains.

Based on these data, an exploratory composition–activity deconvolution was performed to compare activity patterns among extracts differing in alkaloid composition (Supplementary Data, Supplementary Tables 3, 4).

The exploratory composition–activity deconvolution revealed that extracts with more complex alkaloid patterns did not always behave as a simple sum of their apparent marker-associated activities. For S. aureus, positive deviations from the reference expectations were observed mainly for C. cheilanthifolia H1 and G. flavum R, suggesting that these extracts may contain alkaloid patterns worth further investigation using purified compounds. C. cheilanthifolia H2, C. cheilanthifolia R, F. officinalis H, and G. flavum H showed either no relevant deviation or less pronounced deviations, depending on the reference expectation applied. In the case of Ch. majus R, the two reference approaches provided divergent results, indicating that its activity pattern should be interpreted with particular caution. A comparable tendency was observed for P. aeruginosa, although the overall activity pattern was more restricted and should be interpreted in the context of the lower susceptibility of this species to the tested extracts. C. cheilanthifolia H1 and G. flavum R again showed the most notable positive deviations from the reference expectations, whereas the remaining extracts showed no relevant or less consistent deviations. These observations are exploratory and hypothesis-generating only; they do not demonstrate formal synergy, additivity, or antagonism between purified alkaloids. Building on these antibacterial and composition–activity observations, we next assessed the preliminary biocompatibility of the extracts in vitro. Since complex alkaloid-rich extracts may differ not only in antimicrobial activity but also in cytotoxic potential, it was necessary to evaluate their effects on mammalian cells before considering further ocular-specific testing. Therefore, the MTT assay was employed on fibroblasts as a first-line screening to determine the effects of individual and combined plant preparations on cell viability (Figure 5). This step provided an initial estimate of tolerable concentration ranges and helped distinguish selective antimicrobial activity from nonspecific cytotoxicity prior to further in vivo evaluation.

The cytotoxicity assay performed on L929 fibroblasts showed that, under the tested low-exposure conditions, most extracts maintained cell viability above the 80% threshold commonly used in ISO 10993-5-based cytotoxicity assessment. This indicates that the majority of extracts did not show relevant cytotoxicity in this preliminary screening setup. Only selected root extracts enriched in protoberberine alkaloids caused a moderate reduction in viability. The solvent control did not affect cell survival, confirming that the observed effects were associated with the extracts rather than the vehicle. To further compare cytotoxicity patterns among extracts differing in alkaloid composition, an exploratory composition–cytotoxicity deconvolution was performed (Supplementary Table 5).

The exploratory composition–cytotoxicity deconvolution indicated that cytotoxicity patterns differed among extracts depending on their alkaloid composition. Extracts containing COP, PRO, and BER showed measurable cytotoxic effects, but these effects were generally low to moderate under the tested conditions. In C. cheilanthifolia, comparison of H2 (PRO + BER) with H1 (COP + PRO + BER) suggested that the presence of COP may be associated with a higher cytotoxicity pattern, although this observation should be treated as exploratory. In contrast, the G. flavum extracts, containing GLA and/or THP together with other alkaloids, did not show a proportional increase in cytotoxicity, suggesting that more complex alkaloid patterns do not necessarily translate into stronger cytotoxic effects. Overall, these observations suggest that the cytotoxicity of alkaloid-rich extracts cannot be inferred solely from the number of detected alkaloids, but depends on the specific composition pattern. This analysis was used only as a hypothesis-generating comparison to support preliminary safety ranking of the extracts. Given these results, in vivo validation was performed using the G. mellonella larval model to further assess systemic tolerance and potential toxicity under physiological conditions. This model provides a rapid and ethically compliant bridge between in vitro cytotoxicity and vertebrate safety testing, allowing integrated evaluation of both dose-dependent survival and visible tissue responses.

After injection, no mortality was observed in the control group of larvae (PBS + 1% DMSO) throughout 120 h conversely, ethanol (75%) caused complete mortality (100%) within 24 h. Among the plant extracts, F. officinalis_H, G. flavum_H, and C. cheilanthifolia_H2 caused no mortality (0%) during the entire observation period, while C. cheilanthifolia_R and G. flavum_R induced lethality at the level of 16.6%. In turn, C. cheilanthifolia_H1 and Ch. majus_R caused higher lethality, reaching 33.2%. Next, before incorporating the extracts into the commercial formulation of artificial eye drops, their wettability was evaluated. Since the eye drops were intended to be subsequently embedded into BNC carriers, the primary property examined was their wettability, which essentially indicates how well the drop spreads across different surfaces (Figure 6).

FIGURE 6.

Wettability of commercial ophthalmic formulations on surfaces with different physicochemical properties. Representative images show sessile droplets DX, TD, and HF deposited on hydrophobic poly (L-lactic acid) (PLLA), hydrophilic glass, and an amphiphilic artificial tear film. Contact angles were determined using ImageJ. Lower contact angles correspond to higher wettability, indicating better spreading behavior of the tested eye drops on the given surface.

The results presented in Figure 6 indicated that DX exhibited the most favorable spreading behavior, characterized by the lowest contact angles across all tested surfaces, suggesting optimal wettability and suitability for further incorporation into BNC-based ocular carriers. Based on the integrated biological and physicochemical outcomes, the C. cheilanthifolia_H2 extract was selected for incorporation into the DX formulation. The resulting composite eye drops were subsequently embedded into BNC membranes, and the outcomes of this stage are presented in Figure 7.

FIGURE 7.

Antibacterial activity of Corydalis cheilanthifolia_H2 extract introduced to the TX eye drops containing compared to set of controls. The modified disk diffusion assay was performed using BNC discs impregnated with DX + C. cheilanthifolia_H2 or TD alone or C. cheilanthifolia_H2. Discs were placed on Mueller–Hinton agar plates inoculated with S. aureus ATCC 6538 or P. aeruginosa ATCC 15442. The inhibition zones were quantified in ImageJ, with the BNC carrier area subtracted. Results represent the mean inhibition area ± SD from 12 repeats. Anova, followed with Sidak’s multiple comparison test; p < 0.001. The values (mg/mL) in brackets indicate extracts’ concentrations applied.

The modified BNC-based diffusion assay demonstrated that the antibacterial activity was driven by the C. cheilanthifolia_H2 extract, while the DX vehicle containing 0.001% polyhexamethylene biguanide (PHMB) alone produced no inhibition zones against either S. aureus or P. aeruginosa. When the extract was incorporated into the DX formulation, the resulting inhibition zones were significantly larger than those generated by the extract alone (p < 0.05, Sidak’s multiple comparison test), for both tested strains. This finding indicates that the ophthalmic vehicle enhanced the bioavailability or diffusion of the active compounds without contributing intrinsic antimicrobial activity. The results confirm that the antibacterial potential of the C. cheilanthifolia_H2 extract was not only preserved but further potentiated following its incorporation into the composite BNC-based carrier.

4. Discussion

The progressive loss of antibiotic efficacy due to bacterial resistance and the limited chemical diversity among available antiseptics justify the search for complementary antimicrobial strategies (Chinemerem Nwobodo et al., 2022; Muteeb et al., 2023). In our opinion, the studied plant-derived agents should not be regarded as substitutes for classical antimicrobials, but after thorough investigation-as tools capable of temporarily relieving selective pressure while supporting infection control. The aim is not to replace, but to sustain and to provide locally active, short-contact compounds that reduce bacterial load without accelerating resistance. Within this framework, phytochemicals serve as a biological buffer zone between antibiotics and chemically-derived antiseptics interventions, extending the usability of both by preserving their clinical efficacy. In this work, the species from the Papaveraceae family were selected due to their unique and complementing phytochemical profiles, long ethnopharmacological history, and our group’s and others prior experience with its alkaloid constituents (Zielińska et al., 2021; Zielińska et al., 2019; Zhang et al., 2020; 2024; Bhambhani et al., 2021). These plants produce a diverse spectrum of isoquinoline alkaloids - structurally varied yet biosynthetically related - offering a unique model for studying additive, synergistic, and antagonistic interactions within natural mixtures (Krzyżek and Zielińska, 2021). Ch. majus is rich in benzophenanthridine, protoberberine, and protopine derivatives, particularly in its roots. Their concentrations may vary depending on the harvest time, habitat, and weather conditions. Numerous scientific reports indicate that the highest content and most diverse composition of isoquinoline alkaloids have been found in the roots of this species collected in autumn (Zielinska et al., 2019; 2021; Tome and Colombo, 1995). C. cheilathifolia as a species is relatively poorly described in the scientific literature. The few available studies indicate that, similarly to Ch. majus, it is a plant fairly rich in alkaloids, and its extracts exhibit strong antibacterial activity (Krzyżek and Zielińska, 2021; Zielińska et al., 2021). The richness of these compounds in both aerial and roots provides strong justification for including this species in studies aimed at developing a model set of plant metabolites with optimal antimicrobial activity. G. flavum and F. officinalis are characterized by a somewhat different alkaloid profile, which is associated with distinct biological properties (Wagner and Bladt, 1996; Zhang et al., 2020; Akaberi et al., 2021; Kang et al., 2015). In both species, a high structural diversity of isoquinoline alkaloids has not been observed, unlike in Ch. majus or C. cheilathifolia, which makes them convenient model systems for comparative studies.

In turn, the ocular system was chosen as the testing ground because it imposes inherently strict pharmacodynamic conditions: continuous dilution (limited retention time) by tear fluid and exposure to opportunistic pathogens such as S. aureus and P. aeruginosa. Ocular infections remain a persistent clinical problem, particularly in postoperative and contact lens-related settings, where rapid microbial proliferation can compromise vision within dozen of hours (Shah and Wozniak, 2023). Therapeutic options in this context are restricted to a spectrum of antibiotics and antiseptics, both of which face aforementioned limitations related to tolerance, bioavailability, and resistance development. Thus, the clinical need is therefore not for stronger agents, but for safer, transiently active compounds that can reduce microbial burden without selecting for resistant strains or act together with antibiotics and antiseptics.

Nevertheless, it has to be noted that discourse surrounding local use of medical plants, including IQA-containing Papaveraceae, has long been polarized between two extremes: overestimated toxicity and exaggerated antimicrobial potency. Reports of hepatotoxicity, neurotoxicity, or mutagenicity are often generalized from extracts or isolated compounds administered systemically, without considering concentration, formulation, or route of exposure. Conversely, claims that these alkaloids kill Gram-negative and Gram-positive bacteria, as well as fungi, although true in general sense (such an activity is observed), often lack quantitative validation and also disregard critical variables such as concentration, exposure time or formulation, all of which profoundly influence the observed antimicrobial outcome. We ourselves - like many other research teams - have not entirely avoided this trap (Zielińska et al., 2020). This study sought to clarify these misconceptions by combining alkaloid profiling with biological analyses and to define the potential translational range of the extractible mixtures. Based on our previous studies, in which we used mixtures containing a limited number of components, namely, two or three alkaloids in various combinations, to evaluate their antimicrobial effectiveness (Zielinska et al., 2019), we undertook investigations to develop a model mixture based on the correlation between the composition and the direction of bioactivity. Indeed, quantitative analysis of alkaloid content demonstrates that the total yield per 100 g of dry plant material was several orders of magnitude below doses associated with systemic toxicity in animal models (Table 1). Even under the conservative assumption of additive effects, the concentrations achievable in ocular formulations remain far below any toxic threshold. Moreover, given the small applied volume of eye drops (typically ≤ 5 mL) and the short exposure time before dilution (>1 min, typically) and clearance by tear flow, the probability of reaching harmful tissue levels is negligible. Thus, while Papaveraceae alkaloids possess measurable bioactivity, their use in topically applied systems such as ophthalmic formulations should operate safely within a non-toxic window. In silico ADME modeling predicted that several major alkaloids, particularly berberine, chelidonine, allocryptopine, and protopine, may cross the blood–brain barrier and undergo hepatic metabolism (Figure 2). However, again, these predictions must be interpreted in the context of the extremely low concentrations involved in topical ocular application. With dosing volumes of only a few milliliters and rapid dilution by tear turnover, systemic absorption is (if occurs at all) minimal. The possibility of local accumulation in ocular compartments such as the corneal epithelium or vitreous humor cannot be fully excluded (Lorrai et al., 2024), but given the rapid drainage kinetics of tear film, such effects would likely be transient.

One should bear in mind that applied in this work SwissADME and ProTox analyses should be interpreted only as supportive tools for early toxicological contextualization. Predictions of gastrointestinal absorption, oral bioavailability, and oral LD values do not establish ocular tolerability and cannot replace ocular-specific safety models. However, they help place the potential toxicity of alkaloid-rich extracts in the context of expected exposure, concentration range, and plausible systemic burden. Thus, these analyses were used as one component of a broader exploratory screening strategy, aimed at prioritizing candidates for subsequent ocular-specific validation rather than confirming the safety of a finished ophthalmic product. Nevertheless, from a basic science perspective, the in silico toxicity modeling revealed an important relationship between alkaloid diversity within an extract and its predicted toxicological load: mixtures containing a greater number of distinct alkaloids exhibited higher cumulative toxicity scores (Figure 3). The practical implication of this phenomenon once again indicates that while chemical complexity of extract increases theoretical toxicity, the administered doses and short exposure times in ophthalmic formulations should remain within the physiological safety margin.

Also, antimicrobial testing performed using both microdilution and modified disk diffusion assays (Figures 4,5,8) revealed consistent activity patterns, showing that extracts containing more alkaloids generally exhibited stronger antibacterial effects. The microdilution method yielded quantitative MIC and MBEC values, providing a screening environment with efficient diffusion of the extract in liquid medium and prolonged contact with bacterial cells, thus ensuring reliable but initial detection of baseline antimicrobial activity. To address this limitation, a diffusion-based assay was employed, quantifying inhibition zones in mm². This method provided additional information, capturing not only antibacterial potency but also the compound’s ability to spread across soft and agar surfaces. When implemented with BNC as a carrier - which is also considered a potential dressing material for infected eyes (Malani et al., 2024) - the assay more accurately reproduced topical release conditions, enabling diffusion of alkaloids through the hydrated agar matrix. In our previous work, we have already demonstrated the favorable physicochemical and structural properties of a three-dimensional agar scaffold, particularly its capacity to mimic the characteristics of soft tissue (Ciecholewska-Juśko et al., 2022). Although chemically synthesized antiseptics are generally active at lower concentrations than crude plant extracts (Kamaruzzaman et al., 2016; Papa et al., 2022; Franch et al., 2025) this principle should not be directly extrapolated to DX in the present BNC-diffusion assay, because PHMB was present in this formulation only at a preservative-level concentration and was further diluted after incorporation into the hydrated BNC matrix. In this light, the activity of Papaveraceae isoquinoline-alkaloid-rich extracts should be regarded as moderate but meaningful, especially considering their natural origin and multifactorial mechanisms of action.

FIGURE 8.

Antimicrobial activity of plant extracts against S. aureus and P. aeruginosa. The table shows minimum inhibitory concentration (MIC) and minimum biofilm eradication concentration (MBEC) values expressed in mg/mL. Values denoted as > “h.c.t.” indicate that MIC or MBEC exceeded the highest concentration tested. Representative assays for C. cheilanthifolia are shown below: “plankton” - planktonic growth, “biofilm” – biofilm growth, SC - sterility control, GC - growth control, solvent control - DMSO (bottom row).

In line with this, the formulation context plays a decisive role in shaping the apparent efficacy of Papaveraceae isoquinoline-alkaloid-rich extracts. The exploratory composition–activity deconvolution indicated that Papaveraceae isoquinoline-alkaloid-rich extracts display diverse activity patterns depending on their chemical composition. The most notable positive deviations from reference expectations were observed for C. cheilanthifolia H1 and G. flavum R, suggesting that ternary or quaternary alkaloid patterns may contribute to enhanced antibacterial activity. Other extracts, such as C. cheilanthifolia H2, showed a more balanced profile, combining reproducible antimicrobial activity with a relatively favorable preliminary cytotoxicity pattern. Thus, the natural diversity of alkaloid composition across the tested extracts provided useful insight into how specific phytochemical patterns may influence both antibacterial performance and safety-related parameters. These observations support further targeted studies using purified alkaloids and defined mixtures to verify which components or component patterns are most responsible for the observed effects. These findings suggest that future (ours and of other teams) studies should adopt a reductionist strategy - systematically de-constructing the mixtures to identify which individual alkaloid contributes most to antimicrobial activity. Such knowledge would then enable a re-construction approach, where this principal compound could be formulated together with selected co-alkaloids providing cytoprotective or toxicity-buffering effects. This dual process of deconstruction and reconstruction may ultimately yield (if combined with proper surfactants/emulgators) optimized, safer, and mechanistically transparent phytopharmaceutical formulations.

The ranking of antimicrobial potency among the tested extracts aligns with existing reports on the biological activity of individual Papaveraceae alkaloids (Su et al., 2011; Zhang et al., 2023; Yang et al., 2025). The extracts richer in glaucine and tetrahydropalmatine, showed weaker effects, mirroring literature data indicating their limited bacteriostatic capacity (Scholar, 2007; Du et al., 2022). The moderate but stable activity of C. cheilanthifolia_H2 fits within this framework, suggesting that extracts containing fewer, non-interfering alkaloids may achieve a clinically relevant antimicrobial profile.

In vitro cytotoxicity assessment on L929 fibroblasts (Figure 9) showed that most extracts maintained viability above the ISO 10993–5 safety threshold of 80%, confirming their overall biocompatibility. A moderate reduction in viability was observed only for F. officinalis_H, which was dominated by protopine. Although protopine has been reported by other teams to exert cytotoxic effects in certain cell types, in our opinion the observation presented in Figure 9 should be interpreted cautiously, as 2D cell culture systems are highly sensitive to mechanical detachment caused by viscous or surface-active solutions (Weiskirchen et al., 2023). The apparent loss of viability may therefore reflect partial cell detachment rather than genuine cytotoxicity. Importantly, this effect was absent in multi-alkaloid extracts such as G. flavum_R, suggesting also that the presence of other constituents may mitigate the response and support the overall safety of balanced plant extracts. Extracts characterized by lower phytochemical complexity, such as F. officinalis_H, showed higher cytotoxic indices, while those combining multiple structural classes - particularly C. cheilanthifolia_H2 and G. flavum_R - maintained cell survival within the non-cytotoxic range (Supplementary Table 4). To verify whether these in vitro trends translate into a living organism, the in vivo toxicity test using G. mellonella larvae was subsequently performed and confirmed the overall safety of the evaluated extracts (Figure 10). Complete survival (100%) was observed for F. officinalis_H, G. flavum_H, and C. cheilanthifolia_H2, while partial lethality (16.6%–33.2%) occurred only for Ch. majus_R and C. cheilanthifolia_H1. It is noteworthy that these results do not fully parallel the in vitro data (Figure 10 vs. Figure 9) - fibroblast mortality was observed after exposure to F. officinalis, whereas no larval death occurred following injection of the same extract. This may further support our earlier assumption that cells in 2D cultures are more fragile and susceptible to mechanical or chemical stress than those embedded within an extracellular matrix or organized tissues of a living organism. Additionally, since several Papaveraceae alkaloids possess intrinsic anti-insecticidal properties (including activity against G. mellonella, specifically), the observed larval mortality (or its absence) may also reflect species-specific susceptibility rather than general cytotoxicity (Chelav and Khashaveh, 2013; Reyes-Luna et al., 2024). Nevertheless, the larval model, was in majority consistent with fibroblast cytotoxicity assays, and supported the favorable safety profile of C. cheilanthifolia_H2 as a translational candidate for ocular application. Integration of the selected extract into an ophthalmic formulation demonstrated that biological activity could be preserved and even enhanced under applied conditions. When C. cheilanthifolia_H2 was incorporated into the eye drop and further embedded within BNC carriers, the resulting system exhibited superior spreading behavior and produced significantly larger inhibition zones against both S. aureus and P. aeruginosa compared to the extract or vehicle alone (Figures 6, 7). It should be emphasized that PHMB in the DX formulation was present at a preservative-level concentration of 0.0001%, rather than at a therapeutic anti-infective concentration. After incorporation of DX into the hydrated BNC matrix, the effective PHMB concentration was further reduced to approximately 0.00005%, which is below the lower boundary of published broth MIC ranges reported for polyhexanide against S. aureus ATCC 6538 (Schug et al., 2022). Therefore, the absence of a measurable inhibition zone for DX alone in the BNC-diffusion assay was not unexpected and should not be interpreted as evidence that PHMB lacks intrinsic antimicrobial activity. Rather, it reflects the low preservative-level concentration, additional dilution in the BNC matrix, and the limitations of diffusion-based assays for poorly diffusing or formulation-bound compounds. In this study, DX was used as an ophthalmic vehicle to assess whether the selected Corydalis extract retained release-associated antibacterial activity after incorporation into a commercial eye drop formulation, not as a comparator for therapeutic PHMB activity. These findings validate preliminary the compatibility of C. cheilanthifolia_H2 with modern ophthalmic excipients and highlight its potential as a bioactive component of next-generation, surface-adherent therapeutic eye drops.

FIGURE 9.

Cytotoxicity of plant preparations toward fibroblasts. Cells were exposed for 24 h to the indicated concentrations (0,03%–1% v/v) of extracts; DMSO (red curve) is the matched solvent control at the same concentrations. Data are mean ± SD; the horizontal line marks 80% viability. Concentrations of 1% of the plant preparations corresponded to the concentration of 0.05 mg/mL. Boxes under each plot list alkaloids detected in the corresponding extract: COP- coptisine; PRO - protopine; BER - berberine; CHEL - chelidonine; ACP - allocryptopine; GLA - glaucine; THP - tetrahydropalmatine.

FIGURE 10.

In vivo toxicity assessment of plant extracts in the G. mellonella larval model. Larval survival was monitored for 120 h. PBS + 1% DMSO served as a negative control, while 75% ethanol acted as a positive cytotoxic reference. Representative images show a viable (upper image) and a melanized, non-viable (lower image) larvae.

This study was designed as an exploratory translational framework, and several limitations should be acknowledged. Only two bacterial strains - S. aureus ATCC 6538 and P. aeruginosa ATCC 15442 - were examined. While they represent clinically relevant ocular pathogens, expanding the strain panel to include additional clinical isolates will be necessary to confirm broader applicability (Brożyna et al., 2025). The current results should therefore be interpreted as a conceptual reference point rather than a final antimicrobial profile. Moreover, although alkaloids were identified as the principal bioactive fraction, Papaveraceae extracts also contain flavonoids (quercetin, kaempferol, rutin, isorhamnetin) and phenolic acids (caffeic, ferulic, and p-coumaric acids), which may contribute to biological activity through antioxidant and membrane-modulating effects (Croft, 1998; Nazari et al., 2025). Generally, a higher content of polyphenolic compounds is characteristic of aerial parts, primarily the leaves, whereas underground plant organs, particularly in Papaveraceae species, contain substantially smaller amounts of these substances. In contrast, their alkaloid content is proportionally higher (Zielińska et al., 2021). Thus the focus on alkaloids was intentional, as they are the best-characterized pharmacophores of this family, allowing for clearer structure-activity interpretation. A further limitation is that cytotoxicity was assessed using L929 murine fibroblasts, which served only as a first-line screening model for preliminary extract selection. Moreover, because this was a two-dimensional cell culture assay, the observed reduction in cell viability may partly reflect mechanical detachment of cells exposed to viscous or surface-active solutions, rather than direct cytotoxicity alone.

To, at least partially, mitigate this bias, complementary testing was conducted in G. mellonella larvae, though the insect model has its own limitations because certain Papaveraceae alkaloids possess intrinsic anti-insecticidal properties. Moreover, haemocoel injection in G. mellonella should not be interpreted as a model of topical corneal exposure, epithelial irritation, wound healing, tear-film effects, or local ocular pharmacokinetics. In the present study, this model was used as an organism-level toxicity screen in a living system, providing an additional safety-oriented layer beyond two-dimensional cell culture. Extracts showing acceptable profiles in both two-dimensional cell culture and the G. mellonella model may be considered better candidates for subsequent ocular-specific testing, whereas extracts showing toxicity in either system should be deprioritized or reformulated before further development. Thus, although these models do not predict ocular tolerability directly, they help anticipate which extracts are most likely to proceed successfully into more relevant corneal and conjunctival models. Another limitation concerns the interaction modeling based on inhibition zones obtained in the BNC-based diffusion assay. These calculations were performed using activity patterns observed for whole alkaloid-rich extracts and should therefore be interpreted as exploratory and hypothesis-generating. The predicted interaction patterns were not experimentally verified using defined mixtures of purified reference alkaloids. Such verification requires a separate methodological framework, including controlled alkaloid ratios, concentration ranges, solubility controls, and validation of diffusion/release behavior from the BNC carrier. We are currently developing this approach, and future studies will focus on testing purified alkaloids and their defined mixtures in optimized in vitro systems to confirm or refute the interaction patterns suggested here. The study also included a formulation containing polyhexanide (0.001%) as a preservative, yet a systematic analysis of synergy or additivity between the plant extract and antiseptics remains to be performed. Such work will be critical for designing combination therapies that enhance antimicrobial efficacy while minimizing the risk of resistance. Drug release from BNC was not examined, since the release kinetics of unmodified BNC- rapid uptake and rapid release of small molecules -are well documented, including in our previous studies (Junka et al., 2019). Future stages will focus on implementing modified cellulose matrices, such as citric acid–crosslinked systems developed previously for wound applications, to achieve controlled release (Ciecholewska-Juśko et al., 2021). The physicochemical stability of the extracts within the DX formulation were also not assessed. These factors may influence diffusion and bioavailability and will be addressed in subsequent optimization studies.

An additional formulation-related aspect that should be considered in future studies is the intrinsic coloration of several isoquinoline alkaloids. In the present work, the extracts were used after dilution to working concentrations, where the color intensity was visibly reduced compared with concentrated extracts. However, reduced visible coloration does not exclude potential optical or material-related effects after ocular application. Therefore, future development of alkaloid-rich ophthalmic formulations should include quantitative assessment of formulation transparency, washout behavior under tear-film-like conditions, and potential staining or retention on ocular-surface models and contact lens materials.

All these limitations are presented deliberately and explicitly, as each defines a clear direction that must be pursued before full translational implementation can be achieved. They also collectively illustrate why not all aspects could be addressed within a single study - no research team possesses the experimental capacity to comprehensively cover such a multidimensional framework at once.

Overall, this work constitutes a solid in silico, in vitro, and in vivo foundation that, once expanded to address the above limitations, will support targeted preclinical testing in vertebrate models and eventual translation into human ophthalmic applications.

Funding Statement

The author(s) declared that financial support was received for this work and/or its publication. The in vitro analyses on bacteria and fibroblast cells and analyses on G. mellonella as well as in silico modelling was supported by the Wroclaw Medical University (grant no. SUBZ.D300.26.018 to AJ). Cultivation, identification, collection and preparation of plant material is supported by Polish Ministry of Science and Higher Education grant for special research facility in the Botanical Garden of Medicinal Plants at the WMU - decision number 28/598769/SPUB/SP/2024 to AM. The isolation and analysis of the plant extracts was supported by the Polish National Science Center (NCN) grant - SONATA 2019/35/D/NZ7/00266 to SZ.

Data availability statement

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics statement

The manuscript presents research on animals that do not require ethical approval for their study.

Author contributions

MD: Writing – review and editing, Project administration, Visualization, Writing – original draft, Resources, Data curation, Investigation, Software, Conceptualization, Funding acquisition, Methodology. SZ: Supervision, Writing – review and editing, Investigation. TK: Writing – original draft, Investigation. MQ: Investigation, Writing – original draft. BD: Investigation, Writing – original draft. MlB: Writing – original draft, Investigation. WK: Investigation, Writing – original draft. AM: Validation, Supervision, Writing – review and editing. MrB: Writing – review and editing, Supervision, Resources. YS: Resources, Conceptualization, Validation, Investigation, Supervision, Writing – review and editing. AJ: Methodology, Validation, Conceptualization, Supervision, Formal Analysis, Writing – review and editing, Resources, Funding acquisition, Project administration.

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The authors AJ, AM declared that they were an editorial board member of Frontiers at the time of submission. This had no impact on the peer review process and the final decision.

Generative AI statement

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

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

Publisher’s note

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

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fphar.2026.1786083/full#supplementary-material