Comparative anthelmintic activity of artemisinin, dihydroartemisinin, and artesunate against sheep gastrointestinal nematodes: in vitro evaluation and exploratory β-tubulin molecular modeling

Abstract

Background

Gastrointestinal nematode infections are an important constraint on small-ruminant production, and increasing anthelmintic resistance has intensified the search for alternative therapeutic compounds. Artemisinin and its derivatives have demonstrated activity against several helminth species, although comparative information regarding their effects on gastrointestinal nematodes remains limited. This study compared the in vitro anthelmintic activity of artemisinin (ART), dihydroartemisinin (DHA), and artesunate (ARTE) and explored β-tubulin as a candidate molecular target using complementary computational analyses.

Results

All three compounds showed concentration-dependent activity against gastrointestinal nematode eggs and third-stage larvae. In the egg hatch assay, ARTE showed the lowest EC₅₀ value (63.45 µg/mL), followed by ART (82.02 µg/mL) and DHA (99.06 µg/mL). A similar activity pattern was observed for the predefined 24-h larval non-viability endpoint, with operational LC₅₀ values of 66.08 µg/mL for ARTE, 102.82 µg/mL for ART, and 185.41 µg/mL for DHA. Exploratory molecular docking against β-tubulin showed broadly comparable predicted binding at the colchicine-binding pocket, whereas ARTE exhibited the most favorable predicted interaction at the GTP-associated region. Molecular dynamics simulation supported the conformational stability of the predicted ARTE–β-tubulin complex over 150 ns, including persistent residue contacts and Mg²⁺ coordination. In silico ADMET profiling indicated distinct permeability and distribution characteristics among the three compounds without major structural toxicity alerts in the applied prediction tools.

Conclusions

ARTE showed the highest in vitro anthelmintic activity among the three evaluated artemisinin derivatives, with a consistent activity ranking of ARTE > ART > DHA in both experimental assays. The computational findings provide hypothesis-generating support for a potential interaction of ARTE with the GTP-associated region of β-tubulin but do not establish β-tubulin as its molecular target. Further parasite-specific target studies, in vivo efficacy experiments, pharmacokinetic evaluation, and direct target-engagement studies are required to clarify the biological mechanism and veterinary relevance of these findings.

Introduction

Gastrointestinal nematode (GIN) infections represent a major constraint to small ruminant production worldwide, causing substantial economic losses through reduced weight gain, decreased milk yield, increased mortality, and elevated treatment costs [1, 2]. Among the parasitic nematodes infecting sheep and goats, Haemonchus contortus is among the most pathogenic gastrointestinal nematodes because of its blood-feeding activity, which induces severe anemia, hypoproteinemia, and death, particularly in young animals and periparturient ewes and does [2]. Field studies have quantified the production impact of GIN in small ruminant systems, demonstrating that effective parasite control directly influences farm profitability and animal welfare [1]. However, sustainable control of GIN has become increasingly challenging due to the widespread emergence of anthelmintic resistance across all major drug classes, threatening the long-term viability of conventional chemotherapy-based management strategies.

The current anthelmintic arsenal for small ruminants comprises three main drug classes: benzimidazoles (e.g., albendazole, fenbendazole), macrocyclic lactones (e.g., ivermectin, moxidectin), and imidazothiazoles (e.g., levamisole), with monepantel and derquantel representing more recently introduced compounds [2]. Unfortunately, resistance to benzimidazoles and macrocyclic lactones has been documented globally in GIN populations from sheep and goat herds, with multidrug resistance becoming increasingly prevalent [3–5]. This widespread resistance, coupled with the limited availability of new anthelmintic compounds, has intensified the search for alternative chemotherapeutic agents, including the repurposing of drugs with established safety profiles in other therapeutic areas and the exploration of plant-derived natural products with nematicidal activity.

Artemisinin and its derivatives are sesquiterpene lactones (Table 1) originally isolated from Artemisia annua L. and widely used as antimalarial agents. Recent investigations have reported anthelmintic activity of Artemisia species and artemisinin-related compounds against several parasitic helminths [6–8]. Bio-guided fractionation studies have isolated nematicidal lignans and sesquiterpenes from Artemisia cina, demonstrating potent in vitro larvicidal activity against H. contortus third-stage larvae (L3), with reported LC₅₀ and LC₉₀ values supporting their potential as anthelmintic leads [6, 8]. While artemisinin, dihydroartemisinin, and artesunate share a common endoperoxide bridge critical for bioactivity, they differ in their chemical structures, solubility profiles, and pharmacokinetic properties. These differences may influence their comparative anthelmintic efficacies and suitability for veterinary application [9]. The pharmacological activity of artemisinin derivatives is closely associated with their characteristic endoperoxide bridge. In malaria, activation of this peroxide moiety through iron- and heme-associated processes generates reactive intermediates capable of interacting with multiple cellular targets [10]. Artesunate is a water-soluble semisynthetic derivative that undergoes rapid conversion to dihydroartemisinin (DHA), its principal active metabolite, whereas artemisinin-class compounds generally exhibit relatively short systemic half-lives [9, 11]. Differences in solubility, metabolic conversion, membrane permeability, and systemic exposure among ART, DHA, and ARTE may therefore contribute to differences in their biological activity beyond their established antimalarial effects. Despite these promising findings, systematic comparative studies evaluating the anthelmintic efficacies of artemisinin, dihydroartemisinin, and artesunate against GIN under standardized experimental conditions remain scarce, and the molecular mechanisms underlying their nematicidal actions are incompletely understood.

Table 1.

Molecular structures and chemical properties of artemisinin, dihydroartemisinin, and artesunate

Computational approaches such as molecular docking and molecular dynamics (MD) simulations are widely used to explore plausible ligand–protein interactions and to prioritize candidate targets for subsequent experimental investigation [12–14]. In anthelmintic research, β-tubulin is a biologically relevant protein because microtubule function is essential for multiple cellular processes in nematodes and β-tubulin represents a well-established pharmacological target of benzimidazole anthelmintics. Previous computational studies have also used helminth β-tubulin, including Haemonchus contortus β-tubulin, to investigate and prioritize potential anthelmintic compounds [14, 15]. Accordingly, β-tubulin was selected in the present study as a biologically relevant exploratory candidate target.

The objective of this study was to comparatively evaluate the anthelmintic activity of ART, DHA, and ARTE against sheep gastrointestinal nematodes using complementary in vitro and computational approaches. Specifically, we aimed to (i) compare their effects using standardized egg hatch and larval motility inhibition assays and (ii) explore β-tubulin as a candidate molecular target by comparing the predicted binding characteristics and dynamic stability of the three derivatives using molecular docking and MD simulations.

Methods

Chemicals and reagents

Artemisinin (BLDPharm, China; CAS 63968-64-9; lot no. 150443; purity 97%), dihydroartemisinin (Tokyo Chemical Industry Co., Ltd., Japan; CAS 71939-50-9; product no. D3793; purity > 98%), and artesunate (BLDPharm, China; catalogue no. BD22980; lot no. CPX557; CAS 88495-63-0; purity > 98%) were used in the present study. Albendazole (analytical standard; purity ≥ 98%; Sigma-Aldrich, St. Louis, MO, USA; product no. A4673; CAS 54965-21-8) was used as the reference positive control. All test compounds were stored refrigerated at 2–8 °C until use. Dimethyl sulfoxide (DMSO) was used as the solvent, and phosphate-buffered saline (PBS, pH 7.2) was used for the preparation of working solutions and control conditions. Stock solutions were prepared at 10 mg/mL in DMSO. Working solutions used in the assays were freshly prepared immediately before each experiment, with the final DMSO concentration maintained at 5% in all treatment and matched vehicle-control wells. Potential conversion of ARTE to DHA during incubation was not experimentally assessed.

Evaluation of anthelmintic activity

The in vitro activity of artemisinin (ART), dihydroartemisinin (DHA), and artesunate (ARTE) was evaluated using the egg hatch assay (EHA) and larval motility inhibition assay (LMIA). The parasitological procedures followed our previously published sheep gastrointestinal nematode protocol [16], with the test concentrations adapted to the compounds examined in the present study.

Parasite material and experimental design

Fecal samples were collected in November 2025 from 78 naturally infected sheep belonging to a single commercially managed flock in Antalya Province, Türkiye. Samples were pooled before laboratory processing. Gastrointestinal nematode eggs and infective third-stage larvae (L3) were confirmed by microscopic morphological examination. Due to the absence of genus- or species-level differentiation and the lack of anthelmintic-resistance characterization, the biological material was analyzed as a mixed, field-derived gastrointestinal nematode population originating from sheep. The relative genus/species composition of the pooled parasite population was therefore not available.

Treatment and control conditions were manually randomized across pre-numbered wells, and replicate wells were distributed across the plate to reduce potential positional effects. Each condition was tested in three technical replicate wells (n = 3) prepared from the same pooled biological material. Within each assay, all treatment groups were prepared from the same pooled egg or L3 suspension and evaluated in a single assay run; no independent biological replicate preparations were performed on separate days. Observer blinding was not recorded as part of the study protocol.

Egg hatch assay (EHA)

The EHA was conducted according to the general procedure of Kozan et al. [16] and Coles et al. [17]. Eggs were recovered from the pooled fecal material. The suspension was centrifuged at 1500 rpm for 5 min using a DLAB DM0412 centrifuge (DLAB Scientific Co., Ltd., Beijing, China), and the recovered egg preparation was used immediately for testing.

Approximately 100 eggs in 200 µL were dispensed into each well of a 48-well microplate. The target inoculum was determined from the estimated concentration of the egg suspension and the dispensed volume; individual wells were not pre-counted before exposure. ART, DHA, and ARTE were tested at final concentrations of 500, 250, 125, and 62.5 µg/mL. The final DMSO concentration was maintained at 5% in all treatment and matched vehicle-control wells. PBS containing 5% DMSO served as the vehicle control, PBS alone served as an additional untreated control, and albendazole was used at 0.25 mg/mL as the reference positive control. Albendazole was included to verify assay responsiveness and was not used as a concentration–response comparator. The plates were incubated at 27 °C for 24 h, after which Lugol’s iodine solution was added to stop further hatching. Unhatched eggs and first-stage larvae (L1) were counted using an Olympus CX21 microscope (Hamburg, Germany).

where IH represents the percentage inhibition of egg hatching [17].

Preparation of infective third-stage larvae (L3)

Infective L3 were produced from approximately 1 kg of the pooled fecal material. The feces were mixed with wood shavings and cultured at 27 °C for 10–11 days in an SPX-250BIII cooling incubator (Huanghua Faithful Instrument Co., Ltd., China) to permit development to the infective stage. The culture was then placed in a Baermann apparatus overnight to recover migrating larvae. The recovered L3 were concentrated by five centrifugation cycles at 1000 rpm for 5 min using the DLAB DM0412 centrifuge, collected with a micropipette, washed three times with PBS, and stored at 8 °C until use, following the previously described procedure [16, 18].

Larval motility inhibition assay (LMIA)

The LMIA was performed using an adapted protocol based on previously described methods [18–20]. Approximately 100 infective L3 were exposed to serial concentrations of ART, DHA, and ARTE (62.5, 125, 250, and 500 µg/mL) prepared in PBS (pH 7.2). The target larval inoculum was standardized from the pooled L3 suspension and dispensed volume; individual wells were not pre-counted before exposure. DMSO was used to facilitate compound solubilization, with a final concentration of 5% maintained in all treatment wells. PBS containing 5% DMSO served as the vehicle control, while albendazole (0.25 mg/mL), prepared in the same PBS–DMSO vehicle, was included as the reference positive control. Larval responses were assessed at 0, 1, 2, 3, 4, 6, 8, and 24 h after exposure. Larvae were classified as non-viable when loss of spontaneous motility was accompanied by marked rigidity or straightening and increased translucency. No washout or recovery-stimulation step was performed; therefore, the endpoint was treated as non-viability rather than independently confirmed irreversible mortality. Each concentration and control was evaluated in three technical replicate wells (n = 3) prepared from the same pooled L3 suspension. The operational LC₅₀ for the 24-h endpoint, defined as the concentration at which 50% of larvae met the predefined non-viability criteria, was estimated by probit analysis using the Quest Graph™ LC₅₀ Calculator (AAT Bioquest, USA).

Statistical analysis

Data are presented as mean ± standard error of the mean (SEM) from three technical replicate wells (n = 3) prepared from the same pooled biological material. Accordingly, the reported variability reflects within-assay technical precision rather than between-animal or between-flock biological variation. Statistical analyses were performed in R (R Foundation for Statistical Computing, Vienna, Austria), except for LC₅₀ estimation, which was carried out using the Quest Graph™ LC₅₀ Calculator (AAT Bioquest, USA) as described above. For the 24-h egg hatch assay (EHA), the effects of compound (ART, DHA, and ARTE), concentration (62.5, 125, 250, and 500 µg/mL), and their interaction were evaluated using two-way analysis of variance (ANOVA). Vehicle and reference-control groups were summarized separately and were not included in the factorial ANOVA. PBS containing 5% DMSO was considered the matched vehicle control, whereas albendazole was included as a single-concentration reference positive control and was not used for concentration–response potency comparisons. For the larval motility inhibition assay (LMIA), responses recorded from 1 to 24 h were summarized descriptively as mean ± SEM because repeated observations were obtained from the same technical replicate wells. Formal inferential analysis of the LMIA was restricted to the 24-h endpoint, for which the effects of compound, concentration, and their interaction were evaluated using two-way ANOVA. EC₅₀ values for the EHA were estimated in R by linear regression of probit-transformed percentage responses on log₁₀-transformed concentrations, whereas LC₅₀ values for the predefined 24-h larval non-viability endpoint were estimated by probit analysis using Quest Graph. Model fit for the probit regressions was summarized using the coefficient of determination (R²) of the fitted probit–log concentration relationship. Corresponding 95% confidence intervals (95% CIs) were obtained by parametric resampling of the concentration-specific mean responses using their SEMs. These intervals characterize within-assay uncertainty and not between-animal or between-flock variability. With only three technical replicates per cell, formal tests of normality and homogeneity of variance were not considered sufficiently informative; ANOVA results were therefore interpreted as within-assay comparisons within the balanced experimental design. Statistical significance was set at p < 0.05.

Molecular docking studies

To investigate the binding affinity and interaction profiles of the selected compounds against β-tubulin, molecular docking analyses were carried out using the Schrödinger Maestro Suite (Schrödinger LLC, New York, NY, USA). The crystal structure of the β-tubulin–colchicine complex (PDB ID: 1SA0) was retrieved from the Protein Data Bank and prepared using the Protein Preparation Wizard implemented in the Maestro Schrödinger Suite [21]. PDB ID 1SA0, which contains mammalian tubulin components derived from Bos taurus, was selected as an experimentally resolved structural model of β-tubulin. This structure has also been used in previous anthelmintic docking studies [22]. Because it is not parasite-derived, it was used here only as an exploratory structural surrogate. Missing hydrogen atoms were added, bond orders were assigned, and the structure was optimised and minimised under default parameters. Two distinct binding regions were selected for docking studies: (i) the co-crystallised ligand binding site (colchicine binding pocket, defined by the coordinates of CN2) and (ii) the GTP binding site. Grid boxes were generated by centring on the respective ligand coordinates for each binding region. The ligands ART, ARTE, DHA, CN2, and guanosine triphosphate (GTP) were prepared using the LigPrep module and the ionisation states generated at physiological pH (7.0 ± 2.0). Docking simulations were conducted using the Glide module in standard precision (SP) mode.

Molecular dynamics simulations

Molecular dynamics (MD) simulations were performed to assess the conformational stability of the top-ranked docking complex and to characterise the dynamic behaviour of ligand binding at the GTP-associated region of β-tubulin. Based on the docking results, the ARTE–β-tubulin complex at the GTP-associated region was selected for MD simulation because it showed the most favorable predicted docking score among the three artemisinin derivatives. All simulations were conducted using the Desmond module. The protein–ligand complex was embedded in an orthorhombic simulation box solvated with TIP3P explicit water molecules, with a minimum buffer distance of 10 Å between the solute and the box boundaries. The system was neutralised by the addition of appropriate counterions (Na⁺ and Cl⁻), and the ionic strength was maintained at 0.15 M NaCl to approximate physiological conditions. The OPLS4 force field was applied to describe all bonded and non-bonded interactions throughout the simulation. MD simulations were carried out for 150 ns under NPT ensemble conditions, maintained at a temperature of 300 K and a pressure of 1.01325 bar using the Nosé–Hoover thermostat and Martyna–Tobias–Klein barostat, respectively. The root mean square deviation (RMSD) of the protein backbone (Cα atoms) and the ligand heavy atoms and root mean square fluctuation (RMSF) analysis of Cα atoms were analysed. Protein–ligand interaction profiling was performed across the entire trajectory to identify persistent contacts [23].

ADMET analysis

Physicochemical, pharmacokinetic, and toxicological properties of ART, DHA, and ARTE were characterised in silico through the QikProp module of Maestro, using the energy-minimised ligand geometries obtained from LigPrep as input structures. Several physicochemical descriptors were calculated for each compound, namely molecular weight, QPlogPo/w, topological polar surface area (TPSA), solvent-accessible surface area (SASA), and hydrogen bond donor and acceptor counts. Oral drug-likeness was examined against Lipinski’s Rule of Five, which defines acceptable boundaries as MW ≤ 500 Da, QPlogPo/w ≤ 5, HBD ≤ 5, and HBA ≤ 10. Lead-likeness was further checked using Jorgensen’s Rule of Three as a complementary filter. For absorption and distribution, intestinal permeability was approximated through predicted Caco-2 cell permeability values (QPPCaco, nm/s), while central nervous system exposure was estimated from the blood–brain barrier penetration coefficient (QPlogBB) and the associated CNS activity score. Membrane permeability was additionally estimated using predicted MDCK cell permeability (QPPMDCK, nm/s), which serves as a widely accepted surrogate for passive transcellular transport across biological membranes. Toxicological properties, mutagenicity, tumorigenicity, irritancy, and reproductive toxicity were estimated with DataWarrior. These analyses were used to compare the predicted physicochemical, pharmacokinetic, and toxicity profiles of the three compounds and to identify potential in silico liabilities relevant to further investigation.

Use of generative artificial intelligence

ChatGPT and SciSpace were used solely for language editing and rephrasing during manuscript preparation. No generative artificial intelligence tools were used for data generation, statistical analysis, interpretation of experimental findings, or generation of scientific conclusions. All AI-assisted text was critically reviewed and edited by the authors.

Results

Anthelmintic activity

In the egg hatch assay, inhibition at 24 h increased with concentration for each compound. ARTE produced inhibition values of 52.67%, 72.33%, 100%, and 100% at 62.5, 125, 250, and 500 µg/mL, respectively. Corresponding values for ART were 47.00%, 66.33%, 85.33%, and 100%, and those for DHA were 39.00%, 55.00%, 72.33%, and 85.67%. Albendazole (0.25 mg/mL) produced 90.00% inhibition, whereas PBS and the 5% DMSO vehicle control showed no detectable inhibition. Two-way ANOVA showed significant effects of compound and concentration (both p < 0.0001) and a significant compound × concentration interaction (p < 0.001). EC₅₀ estimates were 63.45 µg/mL for ARTE (95% CI: 59.38–67.32; R² = 0.862), 82.02 µg/mL for ART (95% CI: 79.46–84.52; R² = 0.883), and 99.06 µg/mL for DHA (95% CI: 93.87–104.25; R² = 0.999) (Table 2).

Table 2.

Egg hatch inhibition (%) at 24 h and corresponding EC₅₀ values with 95% confidence intervals of ARTE, DHA, and ART against field-derived gastrointestinal nematode eggs (mean ± SEM, n = 3)

Test material	Concentration (µg/mL)	Egg hatch inhibition (%) at 24 h, mean ± SEM	EC₅₀ (µg/mL) (95% CI)
ARTE	500	100.00 ± 0.00	63.45 (59.38–67.32)
	250	100.00 ± 0.00
	125	72.33 ± 2.52
	62.5	52.67 ± 2.52
DHA	500	85.67 ± 2.08	99.06 (93.87–104.25)
	250	72.33 ± 1.53
	125	55.00 ± 1.00
	62.5	39.00 ± 1.00
ART	500	100.00 ± 0.00	82.02 (79.46–84.52)
	250	85.33 ± 1.53
	125	66.33 ± 1.53
	62.5	47.00 ± 1.00
Albendazole	0.25 mg/mL	90.00 ± 0.00	—
PBS	0	0.00	—
DMSO	0	0.00	—

Data are presented as mean ± SEM of three technical replicate wells from the same pooled egg preparation. The factorial ANOVA included only ART, DHA, and ARTE across the four tested concentrations; control groups were summarized separately. EC₅₀ confidence intervals were derived from the concentration-specific mean ± SEM data by parametric resampling and describe within-assay uncertainty

At 24 h, the proportions of L3 meeting the predefined non-viability criteria were 44.67%, 68.33%, 78.67%, and 85.00% for ARTE at 62.5, 125, 250, and 500 µg/mL, respectively; 37.67%, 58.33%, 66.33%, and 75.67% for ART; and 33.00%, 43.00%, 52.67%, and 68.00% for DHA. The PBS and 5% DMSO vehicle controls showed no detectable non-viability, whereas albendazole at 0.25 mg/mL produced 75.00% non-viability at 24 h. Two-way ANOVA of the 24-h endpoint showed significant effects of compound and concentration (both p < 0.0001) and a significant compound × concentration interaction (p < 0.01). Operational LC₅₀ estimates were 66.08 µg/mL for ARTE (95% CI: 54.18–77.33; R² = 0.953), 102.82 µg/mL for ART (95% CI: 94.63–111.09; R² = 0.958), and 185.41 µg/mL for DHA (95% CI: 167.77–203.23; R² = 0.986) (Table 3). The 24-h concentration–response data are shown in Fig. 1.

Table 3.

Time-dependent larval non-viability (%) and corresponding operational LC₅₀ estimates with 95% confidence intervals of ARTE, DHA, and ART against third-stage larvae (L₃) of field-derived gastrointestinal nematodes (mean ± SEM of three technical replicate wells, n = 3)

Test material	Conc. (µg/mL)	1 h	2 h	3 h	4 h	6 h	8 h	24 h	LC₅₀ (µg/mL) (95% CI)
ARTE	500	10.33 ± 0.58	16.00 ± 0.00	24.67 ± 2.08	32.33 ± 2.52	42.67 ± 2.08	56.00 ± 3.00	85.00 ± 2.00	66.08 (54.18–77.33)
	250	8.00 ± 1.00	12.67 ± 2.08	19.67 ± 2.08	28.33 ± 2.52	37.67 ± 3.06	50.67 ± 3.06	78.67 ± 2.08
	125	5.00 ± 0.00	11.00 ± 1.00	16.33 ± 1.53	24.00 ± 0.00	33.00 ± 1.00	44.00 ± 2.00	68.33 ± 1.53
	62.5	4.00 ± 1.00	8.00 ± 0.00	13.00 ± 1.00	19.00 ± 1.00	26.00 ± 0.00	35.00 ± 2.00	44.67 ± 2.52
DHA	500	5.00 ± 0.00	10.00 ± 1.00	16.33 ± 1.53	24.00 ± 0.00	34.00 ± 1.00	46.00 ± 2.00	68.00 ± 1.00	185.41 (167.77–203.23)
	250	3.00 ± 0.00	7.00 ± 1.00	12.33 ± 1.53	19.00 ± 0.00	28.00 ± 1.00	38.00 ± 2.00	52.67 ± 1.53
	125	1.00 ± 0.00	4.33 ± 0.58	8.00 ± 1.00	14.00 ± 0.00	22.00 ± 1.00	31.33 ± 1.53	43.00 ± 1.73
	62.5	0.00 ± 0.00	0.00 ± 0.00	4.00 ± 0.00	9.00 ± 1.00	16.33 ± 1.53	24.00 ± 0.00	33.00 ± 2.00
ART	500	6.00 ± 0.00	12.00 ± 1.00	19.33 ± 1.53	27.00 ± 0.00	39.00 ± 1.00	51.00 ± 2.00	75.67 ± 2.08	102.82 (94.63–111.09)
	250	4.00 ± 0.00	9.00 ± 1.00	15.33 ± 1.53	22.00 ± 0.00	32.00 ± 1.00	43.33 ± 1.53	66.33 ± 1.53
	125	2.00 ± 0.00	5.00 ± 1.00	10.33 ± 1.53	15.00 ± 0.00	23.00 ± 1.00	33.00 ± 2.00	58.33 ± 1.15
	62.5	1.00 ± 0.00	3.00 ± 1.00	6.00 ± 1.00	10.00 ± 0.00	16.00 ± 1.00	24.00 ± 2.00	37.67 ± 1.15
Albendazole	0.25 mg/mL	31.00 ± 0.00	35.00 ± 0.00	46.00 ± 0.00	55.00 ± 0.00	63.00 ± 1.00	75.00 ± 1.00	75.00 ± 1.00	—
PBS	0	0	0	0	0	0	0	0	—
DMSO	0	0	0	0	0	0	0	0	—

Time-course values are presented descriptively as mean ± SEM because the same technical replicate wells were followed repeatedly over time. Formal factorial inference was restricted to the 24-h endpoint. LC₅₀ estimates were derived from the 24-h concentration-response data; the 95% confidence intervals describe within-assay uncertainty and do not represent between-animal or between-flock variability

Fig. 1.

Dose–response activity of artemisinin derivatives against field-derived gastrointestinal nematode eggs and third-stage larvae (L₃). A Egg hatch assay (EHA). Concentration–response curves for ARTE, DHA, and ART at 62.5–500 µg/mL show egg hatch inhibition at 24 h. EC₅₀ estimates (95% CI) were ARTE: 63.45 µg/mL (59.38–67.32), ART: 82.02 µg/mL (79.46–84.52), and DHA: 99.06 µg/mL (93.87–104.25). The horizontal dotted line represents the response of the single-concentration albendazole reference positive control (90.00% at 0.25 mg/mL) and is not a concentration-response comparison. B Larval motility inhibition assay (LMIA). Concentration-response curves show the proportion of L₃ meeting the predefined non-viability criteria at 24 h. Operational LC₅₀ estimates (95% CI) were ARTE: 66.08 µg/mL (54.18–77.33), ART: 102.82 µg/mL (94.63–111.09), and DHA: 185.41 µg/mL (167.77–203.23). The horizontal dotted line shows the response of the single-concentration albendazole reference positive control (75.00% at 0.25 mg/mL). Data are presented as mean ± SEM of three technical replicate wells (n = 3)

Molecular docking studies

The β-tubulin–colchicine complex (PDB ID: 1SA0) was used as the target structure, and docking simulations were carried out at two distinct binding regions: the colchicine-binding pocket, defined by the coordinates of the co-crystallised ligand CN2, and the GTP-binding site. ART, DHA, ARTE, CN2, and guanosine triphosphate were docked into their respective binding regions, and the resulting GlideScores are summarised in Table 4. At the colchicine-binding pocket, all three artemisinin derivatives yielded closely clustered docking scores, − 7.105 kcal/mol for DHA, − 7.027 kcal/mol for ARTE, and − 6.969 kcal/mol for ART, none of which approached the affinity of the co-crystallised reference ligand CN2 (− 8.192 kcal/mol). The narrow spread across the three compounds suggests that structural differences between the derivatives have little influence on binding within this pocket. A notably different picture emerged at the GTP-binding site. Artesunate achieved a docking score of − 9.553 kcal/mol, closely approaching the value obtained for GTP itself (− 9.798 kcal/mol), whereas ART (− 6.371 kcal/mol) and DHA (− 5.868 kcal/mol) bound considerably less favourably. This marked difference suggests a more favorable predicted interaction of ARTE with the GTP-associated region compared with ART and DHA. The 2D interactions of the ligands are given in Fig. 2. Artesunate formed hydrogen bonds with Gly146, Thr145, and Lys254, interacting with residues located near the GTP-binding loop. The compound established a salt bridge with Lys254; glycine interactions with Gly10, Gly143, and Gly144; hydrophobic contacts with Tyr224, Ala180, and Ala12; polar interactions with Ser178, Thr179, and Gln247; and charged interactions with Asp69, Glu71, and Glu183. Mg²⁺ at the GTP-binding site of β-tubulin is an essential cofactor that stabilises the negative charges of the triphosphate moiety of GTP and facilitates the GTPase activity required for tubulin polymerisation and dynamic instability. The docking scores for artesunate are − 9.553 kcal/mol against − 9.798 kcal/mol for GTP at the same site. Although ARTE showed a favorable predicted docking score within the GTP-associated region, docking alone cannot demonstrate competition with GTP or establish a functional effect on β-tubulin. These findings therefore identify this region only as a potential interaction site requiring experimental validation.

Table 4.

Molecular docking scores (kcal/mol) of ARTE, ART, and DHA at the colchicine-binding pocket and GTP-binding site of β-tubulin, compared with the co-crystallized ligand and GTP

Compound	Colchicine-binding pocket (kcal/mol)	GTP-binding site (kcal/mol)
Artesunate	-7.027	-9.553
Artemisinin	-6.969	-6.371
Dihydroartemisinin	-7.105	-5.868
Co-crystal	-8.192	-
GTP	-	-9.798

Fig. 2.

2D interactions of artesunate and GTP at the GTP-binding site

Molecular dynamics simulations

To further evaluate the docking-predicted complex and investigate the stability of ligand binding at the GTP-associated region of β-tubulin, molecular dynamics (MD) simulations were performed for the highest-scoring compound, ARTE. A 150 ns MD simulation was conducted using the Desmond module of the Schrödinger. The stability of the protein–ligand complex was assessed using RMSD and RMSF analyses. The protein backbone (Cα) RMSD ranged from 1.2 to 3.4 Å across the simulation, and the ligand RMSD stayed within 1.3–3.9 Å. These values indicated overall conformational stability of the simulated complex (Fig. 3). RMSF analysis showed a similar pattern: residue-level fluctuations were mostly below 3.5 Å throughout the binding region, with a somewhat elevated but still modest flexibility (below 4.5 Å) confined to the 410–420 loop. Interaction profiling over the trajectory revealed that several contacts remained persistent throughout the trajectory (Figs. 4 and 5). Asp98 maintained an ionic interaction with the carboxylate arm of artesunate at close to 100% of frames, with a concurrent water bridge also present at ~ 99% occupancy. Water-mediated contacts with Ala12, Asp69, and Asn101 were similarly well-preserved, while Ala180, Glu183, and Tyr224 appeared intermittently at lower frequencies. A notable feature was the persistent coordination Mg²⁺. Through its carboxylic acid moiety, artesunate maintained direct coordination with the active-site Mg²⁺ ion at essentially full occupancy across the simulation. This persistent contact was consistent with the static docking pose and suggests that Mg²⁺ coordination contributes to stabilization of ARTE within the GTP-associated pocket. Asp98 and Asp69 also showed highly persistent contacts throughout the 150 ns simulation.

Fig. 3.

Molecular dynamics analysis of the β-tubulin–artesunate complex showing protein–ligand RMSD and protein RMSF profiles during the 150 ns simulation. A Protein–ligand RMSD profile. B Protein RMSF profile

Fig. 4.

MD trajectory-derived 2D interaction map of artesunate at the β-tubulin GTP-binding site. The map shows protein–ligand interactions present for more than 30% of the 150 ns simulation, including contacts with Asp98, Asp69, Asn101, Ala12, Ala180, and Glu183, together with water-mediated interactions and Mg²⁺ coordination

Fig. 5.

Dynamic interaction profile of artesunate bound to the β-tubulin GTP-binding site. A Residue-wise protein–ligand contact persistence over the molecular dynamics trajectory. B Interaction fractions and contact-type distribution for residues involved in ligand binding. C Time-dependent variations in ligand RMSD, radius of gyration (rGyr), intramolecular hydrogen bonding, molecular surface area (MolSA), solvent-accessible surface area (SASA), and polar surface area (PSA)

ADMET analysis

ADMET analysis was performed for ART, ARTE, and DHA using QikProp (Maestro 14.5) (Table 5). The compounds satisfied Lipinski’s Rule of Five and Jorgensen’s Rule of Three. Predicted oral absorption ranged from 73.31% to 100% within the applied computational model; however, these values should not be interpreted as direct evidence of oral bioavailability in sheep. QPPCaco values for ART (1997.46 nm/s) and DHA (2661.41 nm/s) were considerably higher than those recorded for ARTE (56.06 nm/s). QPPMDCK permeability values followed the same pattern, with ARTE returning the lowest value (27.94 nm/s). QPPCaco and QPPMDCK predictions indicated lower modeled permeability for ARTE than for ART and DHA. CNS activity scores were + 1 for ART and DHA and − 2 for ARTE. These predictions provide comparative in silico information only and cannot be directly extrapolated to pharmacokinetic behavior in ruminant species, in which gastrointestinal physiology may substantially influence drug availability. Descriptors including QPlogPo/w, PSA, molecular volume, SASA, FOSA, and FISA all fell within QikProp’s accepted ranges for each compound. No mutagenicity, tumorigenicity, or reproductive toxicity alerts were identified by the applied in silico prediction tools.

Table 5.

Predicted ADMET Parameters of ART, ARTE, and DHA Using QikProp

Parameter	ARTE	ART	DHA
Lipinski’s Rule of Five	+	+	+
Jorgensen’s Rule of Three	+	+	+
Oral Absorption (%)	73.31	96.513	100
QPPCaco	56.068	1997.469	2661.412
QPPMDCK	27.944	1045.041	1425.102
CNS	-2	1	1
SASA	645.481	466.076	472.985
QPlogPo/w	2.573	1.792	1.989
TPSA	121.759	65.129	55.413
FISA	174.078	73.333	60.191
FOSA	471.403	392.743	412.795
mol MW	384.425	282.336	284.352

Discussion

This study showed a consistent difference in the in vitro activity of ART, DHA, and ARTE against a pooled, field-derived sheep gastrointestinal nematode population. ARTE exhibited the lowest EC₅₀ estimate in the egg hatch assay (63.45 µg/mL; 95% CI: 59.38–67.32), followed by ART (82.02 µg/mL; 95% CI: 79.46–84.52) and DHA (99.06 µg/mL; 95% CI: 93.87–104.25). A similar ranking was observed for the operational 24-h larval non-viability endpoint, for which ARTE showed the lowest LC₅₀ estimate (66.08 µg/mL; 95% CI: 54.18–77.33), followed by ART (102.82 µg/mL; 95% CI: 94.63–111.09) and DHA (185.41 µg/mL; 95% CI: 167.77–203.23). Two-way ANOVA also identified significant effects of compound and concentration and a significant compound × concentration interaction for both the EHA and the 24-h LMIA endpoint. Thus, the comparative activity pattern ARTE > ART > DHA was consistent across both assays. Because the biological material was pooled from one flock and the n = 3 observations were technical replicate wells, these estimates, confidence intervals, and inferential comparisons describe within-assay differences rather than population-level efficacy across sheep flocks.

These results are consistent with previous studies indicating anthelmintic efficacy of artemisinin-class drugs; nevertheless, direct comparative data among these three particular derivatives against gastrointestinal nematodes are still scarce in the literature. Cala and colleagues reported measurable inhibition of egg hatch and larval motility by Artemisia annua extracts, supporting the genus-level anthelmintic potential of artemisinin-containing preparations [24]. Similarly, in our previous study we documented dose-dependent effects of multiple Artemisia species extracts on egg hatch and larval motility in vitro, reinforcing the biological activity of this phytochemical class against nematodes [25]. However, these investigations utilised crude extracts instead of refined derivatives, which prevents direct potency comparisons with the current data. Keiser and colleagues established the in vivo efficacy of artesunate against Fasciola hepatica in naturally infected ruminants, demonstrating that artemisinin derivatives can exhibit anthelmintic actions in vivo against trematodes [26]. Shaohong et al. additionally documented a decrease in parasite burden subsequent to artesunate treatment in experimental infection models, indicating derived efficacy beyond antimalarial applications [27]. Unlike previous research, this study systematically compares ART, DHA, and ARTE side-by-side. Our findings from standardized in vitro assays demonstrate a consistent comparative activity pattern within this derivative series against gastrointestinal nematodes; however, three compounds alone are insufficient to define a formal structure–activity relationship.

The observed activity ranking (ARTE > ART > DHA) is noteworthy because DHA is the principal active metabolite of artesunate in antimalarial therapy [9]. In the treatment of malaria, the efficacy of artemisinin derivatives predominantly relies on the endoperoxide bridge, which is activated via iron- and heme-dependent mechanisms [10]. In contrast, DHA showed the lowest activity among the three compounds in the present assays. This disparity indicates that the molecular processes necessary for anthelmintic activity may differ from those that regulate antimalarial effects [28, 29]. Structural and physicochemical differences among the three derivatives may contribute to the observed differences in activity [9].

Because the biological material was pooled from a single flock and was not differentiated to genus or species level, the observed responses represent the aggregate behaviour of a mixed field-derived gastrointestinal nematode population rather than taxon-specific susceptibility. The source flock was also not phenotypically or molecularly characterized for anthelmintic resistance; consequently, activity against resistant isolates cannot be inferred.

Albendazole was included at a single reference concentration as a positive assay control rather than as a pharmacological comparator. Consequently, the higher percentage response observed for ARTE at some tested concentrations should not be interpreted as evidence that ARTE is more potent or more efficacious than albendazole. A direct comparison would require concentration–response testing of the reference drug under the same assay conditions. The final DMSO concentration was 5% in all treatment and matched vehicle-control wells, and the matched vehicle produced no detectable effect under the present assay conditions. Nevertheless, this observation is assay-specific and does not establish that 5% DMSO is without biological effects under all nematode assay conditions. A formal comparative solubility assessment of ART, DHA, and ARTE was not performed; therefore, equivalent solubility of the three compounds under the assay conditions cannot be assumed.

Molecular docking was performed to explore β-tubulin as a candidate molecular target and to determine whether the three artemisinin derivatives differed in their predicted interaction profiles. Docking to the β-tubulin colchicine-binding pocket showed that all three compounds had similar binding affinities (ARTE = − 7.027 kcal/mol, ART = − 6.969 kcal/mol, DHA = − 7.105 kcal/mol). The relatively similar docking scores at the colchicine-binding pocket did not parallel the differences observed in the in vitro activity of the three compounds. On the other hand, docking to the GTP-binding region of β-tubulin showed very different affinities. ARTE had the strongest predicted binding (− 9.553 kcal/mol), followed by ART (− 6.371 kcal/mol) and DHA (− 5.868 kcal/mol). The more favorable predicted interaction of ARTE within the GTP-associated region paralleled its higher in vitro activity; however, this correspondence does not establish a causal relationship.

The present docking results suggest a possible alternative interaction of artemisinin derivatives within the GTP-associated region of β-tubulin, which is distinct from the benzimidazole mechanism. This hypothesis is supported by prior in silico studies that have employed helminth β-tubulin models to prioritize small molecules with anthelmintic potential. Ranjan et al. explored interaction zones within the β-tubulin colchicine-binding pocket using structure-based analysis, demonstrating that ligand-contact mapping can guide anthelmintic hit identification [15]. Several discovery studies have integrated docking to β-tubulin alongside other helminth targets to rank candidate compounds before experimental validation [30–32]. The present study extends this approach by identifying the GTP-associated region as a potential interaction site for ARTE that warrants further investigation.

The GTP-binding region of β-tubulin is central to microtubule dynamics because the nucleotide state of β-tubulin directly affects tubulin assembly and microtubule stability. After β-tubulin-bound GTP is incorporated into the microtubule lattice, it is hydrolyzed, and loss of the GTP cap is associated with microtubule catastrophe, whereas GTP binding supports the assembly-competent state of tubulin dimers [33, 34]. On this basis, interference with nucleotide binding, or with the conformational changes linked to it, could disrupt microtubule assembly and impair essential cellular functions in nematodes. In the present study, the favorable predicted interaction of ARTE with the GTP-associated region, together with its greater in vitro activity, provides a hypothesis-generating observation but does not establish that β-tubulin mediates the observed anthelmintic effect. Other molecular targets or mechanisms may therefore contribute to the observed activity of ARTE and the other artemisinin derivatives. If this interaction is subsequently confirmed experimentally, it could represent a mechanism distinct from the classical benzimidazole-binding mode. However, the present study did not test phenotypically or molecularly characterized benzimidazole-resistant isolates. Therefore, no conclusion can be drawn regarding cross-resistance, preserved activity in resistant nematode populations, or utility for resistance management. These questions require direct testing in characterized resistant and susceptible gastrointestinal nematode isolates together with target-engagement and resistance-genetics studies [35–38].

The structural interpretation of these findings should remain exploratory because PDB ID 1SA0 is mammalian rather than parasite-specific. Although 1SA0 has previously been used in anthelmintic docking studies [22], sequence and structural differences between mammalian and nematode β-tubulin may influence predicted ligand interactions. Therefore, the present docking results should not be interpreted as evidence that β-tubulin is the molecular target of ARTE in the tested nematode population. To further evaluate the docking predictions and examine the temporal stability of the ARTE–β-tubulin complex, MD simulations were performed over 150 ns. Across the simulation trajectory, ARTE retained a stable binding pose within the GTP-binding site, as indicated by low RMSD and RMSF values for the ligand and neighboring binding-site residues. Persistent hydrogen-bonding and hydrophobic contacts were also maintained throughout the simulation. Notably, coordination with Mg²⁺, a functionally important cofactor in tubulin guanine-nucleotide binding, was maintained throughout the simulation [21]. Because the nucleotide state of β-tubulin governs tubulin assembly behavior and microtubule dynamic instability, interference with this region could plausibly perturb the conformational transitions required for microtubule polymerization and depolymerization [39, 40]. The 150 ns MD simulation supports the conformational stability of the predicted ARTE–β-tubulin complex within the computational model but does not establish that this interaction occurs under biological conditions. MD simulations are commonly used to evaluate and refine such docking predictions [41, 42]. These findings are also in line with previous structure-based studies that combined docking and MD to prioritize β-tubulin-directed ligands before experimental testing [32, 42].

ADMET prediction revealed differences in the physicochemical and permeability profiles of the three compounds. ART and DHA showed higher predicted Caco-2 and MDCK permeability than ARTE, whereas ARTE showed lower predicted CNS exposure. All three compounds satisfied the evaluated drug-likeness criteria, and no major toxicity alerts were identified by the applied prediction tools. However, these findings represent computational estimates and should not be interpreted as evidence of in vivo pharmacokinetic behavior or safety. Pharmacokinetic, tissue-distribution, and safety studies in target animal species will therefore be required to determine the translational relevance of these predictions.

Limitations

This study has several limitations. The parasite material was pooled from a single flock and was not differentiated at the genus or species level; therefore, the findings represent a mixed gastrointestinal nematode population and cannot be attributed to a specific taxon. The assays were based on technical rather than independent biological replicates, which limits generalizability. The biological experiments were exclusively in vitro, and no in vivo efficacy, pharmacokinetic, or safety studies were performed. The source population was not characterized for anthelmintic resistance. In addition, β-tubulin was evaluated only as a single exploratory candidate target using a non-parasite structure. Accordingly, the docking and MD findings should be regarded as hypothesis-generating and do not establish biological target engagement or causation. Finally, the ADMET results are in silico predictions and may not directly reflect pharmacokinetic behavior in sheep, particularly given species-specific differences and ruminant gastrointestinal physiology. Further parasite-specific, biologically replicated, target-validation, and in vivo studies are warranted.

Conclusions

This study provides a comparative in vitro and computational evaluation of artemisinin (ART), dihydroartemisinin (DHA), and artesunate (ARTE) against a pooled, field-derived gastrointestinal nematode population obtained from naturally infected sheep. ARTE showed the highest in vitro activity among the three compounds in both the egg hatch assay and the operational larval motility/non-viability endpoint, with the lowest EC₅₀ and LC₅₀ point estimates and a consistent activity ranking of ARTE > ART > DHA. Because the biological material originated from one pooled flock sample and the assays used technical triplicates, these findings should be interpreted as preliminary within-assay comparative evidence. They do not establish species-specific efficacy, activity against anthelmintic-resistant isolates, superiority to albendazole, or in vivo effectiveness.

The computational analyses provided complementary, hypothesis-generating evidence but do not establish β-tubulin as the mechanism underlying the observed in vitro activity. Further studies using parasite-specific structural models, direct target-engagement approaches, resistant isolates, and in vivo efficacy, pharmacokinetic, and safety evaluations are required to determine the mechanism and veterinary relevance of artesunate against gastrointestinal nematodes.

Abbreviations

ADMET

Absorption, Distribution, Metabolism, Excretion, and Toxicity

ANOVA

Analysis of Variance

ART

Artemisinin

ARTE

Artesunate

CNS

Central Nervous System

DHA

Dihydroartemisinin

DMSO

Dimethyl Sulfoxide

EC₅₀

Half Maximal Effective Concentration

EHA

Egg Hatch Assay

FISA

Hydrophilic Component of Solvent-Accessible Surface Area

FOSA

Hydrophobic Component of Solvent-Accessible Surface Area

GIN/GINs

Gastrointestinal Nematode(s)

GTP

Guanosine Triphosphate

IH

Inhibition of Hatching

L1

First-Stage Larva

L3

Third-Stage Larva

LC₅₀

Operational concentration associated with 50% larval non-viability

LMIA

Larval Motility Inhibition Assay

QPlogPo/w

Octanol/Water Partition Coefficient

MD

Molecular Dynamics

PBS

Phosphate-Buffered Saline

PSA

Polar Surface Area

QPlogBB

Predicted Brain/Blood Partition Coefficient

QPPCaco

Predicted Caco-2 Cell Permeability

RMSD

Root Mean Square Deviation

RMSF

Root Mean Square Fluctuation

SASA

Solvent-Accessible Surface Area

SEM

Standard Error of the Mean

SMILES

Simplified Molecular Input Line Entry System

SP

Standard Precision

Authors' contributions

SROD contributed to the conceptualization, investigation, methodology, data curation, formal analysis, visualization, project administration, and preparation and revision of the manuscript. MAA contributed to the methodology, computational analyses, data curation, formal analysis, software, validation, visualization, and revision of the manuscript. MSE contributed to the methodology, validation, and revision of the manuscript. EK contributed to the methodology, resources, supervision, and revision of the manuscript. SA contributed to funding acquisition and revision of the manuscript. IITC contributed to supervision and revision of the manuscript. All authors read and approved the final manuscript.

Funding

This study was supported by the Scientific Research Projects Unit (BAP) of Inonu University under grant number TSA-2025-4366. The support covered access to the Schrödinger software program and the procurement of the active compounds used in the study. The authors also acknowledge Ross University School of Veterinary Medicine, Department of Research, for support with open access publication.

Data availability

Data generated in this study are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

Ethical approval was not required for this study. Fecal samples were obtained from naturally infected sheep, and no animals were experimentally infected or treated for the purposes of this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.