Integrative biochemical and computational analysis of leflunomide-induced redox alterations in Nauphoeta cinerea

Abstract

Leflunomide (LF) is an immunomodulator classified as a disease-modifying anti-rheumatic agent. Its mechanisms of action include inhibition of dihydroorotate dehydrogenase in lymphocytes, resulting in immunosuppressive effects, and blocking tyrosine kinase activation, generating anti-inflammatory effects. This study aimed to study biochemical parameters 24 h after leflunomide administration to Nauphoeta cinerea (Cockroach) model, focusing on toxicity assessment. Biochemical analyses were performed, and molecular docking studies were conducted with the proteins 6VCD, 2B3X, and 2B3Y. Leflunomide altered biochemical parameters by increasing markers such as PSH, NPSH, TBARS, and iron levels, suggesting that LF may induce oxidative stress. This was supported by a dose-dependent increase in PSH, indicating an adaptive defense response. Furthermore, a significant increase in NPSH was observed at 128 µg/mL coupled with increased lipid peroxidation and elevated iron levels at 256 µg/mL. Additionally, LF exhibited high molecular activity by interacting with the target proteins, presenting major alkyl-type bonds, in addition to the interaction of fluorine with 2 proteins (6VCD and 2B3Y). In conclusion, leflunomide appears to induce oxidative stress in a dose-dependent manner, triggering adaptive defense responses and altering key biochemical markers.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-33177-2.

Introduction

Immunosuppressants are essential to prevent graft rejection and improve graft and patient survival, with treatment choice depending on individual conditions and including agents such as antiproliferative drugs, calcineurin inhibitors, and corticosteroids^1,2. Leflunomide (HWA 486 or SU101), an orally administered isoxazole-derived prodrug (C12H9F3N2O2; IUPAC: 5-methyl-N-[4-(trifluoromethyl)phenyl]-isoxazole-4-carboxamide; MW 270.2 a.m.u.), is almost completely metabolized in the intestine and liver to its active metabolite, teriflunomide (A771726)^3–5. As a disease-modifying antirheumatic drug (DMARD), leflunomide modulates immune responses, slows disease progression, and improves long-term outcomes^6–8. It is classified as a BCS class II drug with low solubility and high permeability^9–13, and its mechanism involves inhibition of dihydroorotate dehydrogenase in lymphocytes and blockage of tyrosine kinase activation, producing immunosuppressive and anti-inflammatory effects^14–17.

Over the past few years, there has been a growing substantial resurgence within the scientific community regarding the adoption of alternative methods to animal experimentation. This growing interest is mainly concerned about animal welfare and the rising costs associated with sustaining and searching for traditional animal models^18–20.

Among the species under consideration, Nauphoeta cinerea stands out as a promising alternative model for replacing experiments involving mammals, particularly in the evaluation of toxicity and pharmacological effects of xenobiotics and natural toxins. This viewpoint is bolstered by recent studies^20–24. Nauphoeta cinerea shares biophysical principles with the mammalian nervous system, allowing for the analysis of neurotransmitters such as acetylcholine, octopamine, and gamma-aminobutyric acid (GABA), which are also present in mammals. This characteristic makes it a promising model for pharmacological, neurobehavioral, and basic toxicological studies^25,26.

The literature shows a lack of studies involving alternative models for pharmacological analyses of drugs. Moreover, the use of such models enables the evaluation of a large number of individuals, making them a promising strategy for preliminary assays. Therefore, there is a significant gap regarding the application of alternative models.

In this context, the present study aimed to assess the biochemical parameters after 24 h of leflunomide exposure in the Nauphoeta cinerea model, with a focus on toxicological aspects and their impact on biochemical responses. This approach allows comparison with existing literature, highlighting the potential of alternative models as tools for pharmacological studies without the use of mammalian subjects.

Materials and methods

Computational assessment of acute toxicity using QSAR models with stoptox

Acute toxicity of the compound was evaluated using the STopTox platform, (https://stoptox.mml.unc.edu/), which applies Machine Learning (ML) and Quantitative Structure-Activity Relationship (QSAR) models. Predictions were conducted for six toxicity endpoints: oral, dermal, inhalation, skin irritation/corrosion, eye damage, and skin sensitization. The reliability of the predictions was supported by comparison with validated experimental datasets²⁷.

Prediction of toxicity of chemicals

ProTox 3.0 estimates toxicity by combining molecular similarity analyses, fragment-based likelihoods, recurrent structural features, and machine learning methods. To improve predictive reliability, it applies CLUSTER cross-validation, which relies on fragment similarity. The platform incorporates 61 predictive models to evaluate a wide range of toxicity endpoints, such as acute and organ toxicity, toxicological outcomes, molecular initiating events, metabolism, adverse outcome pathways (Tox21), and toxicity-related targets²⁸.

Breeding and dietary formulation of Nauphoeta cinerea matrices

The N. cinerea specimens choose in the present work was sourced from (BIOTOX) at the URCA, Ceará, Brazil. Farming of these species was conducted under strongly controlled environments, confirming a temperature range of 23–25 °C, a relative humidity level of 70%, and a lighting cycle set at 8–16 h. To endure these species, their diet contained of rodent feed accurately articulated to accomplish the specific nutritional demands of N. cinerea employed in this study.

Toxicity assessment in Nauphoeta cinerea

The medication was administered in the third abdominal segment of N. cinerea using a 1 mL needle, with a precise quantity of 10 µL applied to the experimental model. Each treatment consisted of three biological replicates per treatment, each consisting of pooled samples of 10 samples, and three technical replicates per assay. Subsequently, the mortality of individuals was monitored at specific time intervals, including 1, 2, 3, 4, 5, 6, 12, and 24 h after medication application. This detailed experimental approach allows for precise and temporal assessment of the medication’s effects on cockroach survival, providing data on its acute toxicity and potential short-term effects (Table 1).

Table 1.

Experimental design indicating biological groups, sample size, animal distribution, and technical replicates.

Group/Treatment	Biological replicates (n)	Animals per biological replicate	Technical replicates
Control (-)	3	10	3
64 µg/mL	3	10	3
128 µg/mL	3	10	3
256 µg/mL	3	10	3

Sample preparation for biochemical assays

After toxicological assays, the N. cinerea cockroaches from the “control (-)” (distilled water) and “exposed to the Leflunomide (diluting in distilled water) (64, 128 and 256 µg/mL)” groups were prepared on ice (3 min), and heads were cutoff (3 pooled heads), weighed, and homogenized in 0.1 M phosphate buffer, pH 7.4 (1 mg of precipitate: 40 µL of buffer ratio), and centrifuged at 10,000 rpm for 10 min (7826 × g). Each experiment was conducted with three replicates (Table 1).

Measurement of protein thiol (PSH) and non-protein thiol (NPSH) levels

The sulfhydryl (-SH) content of proteins and peptides in the supernatant was evaluated by measuring protein thiol (PSH) and non-protein thiol (NPSH) levels as indicators of oxidative changes. Briefly, 50 µL of supernatant was combined with 150 µL of phosphate buffer (pH 7.4) and 10 µL of 5 mM DTNB. The reaction mixture was incubated for 30 min at 25 °C in the dark and analyzed at 405 nm using a microplate reader, with glutathione (GSH) as the standard. NPSH levels were determined by precipitating the supernatant with 10% TCA and centrifuging for 3 min²⁹.

Accuracy and precision were assessed following ISO 5725-1:2023. Measurements were performed in 6 replicates for each concentration (Control, 64, 128, and 256 µg/mL), and coefficients of variation (CV%) and standard deviations (SD) were calculated to evaluate reproducibility. Intra- and inter-assay variability were determined by repeating the experiments on the same and on independent days, respectively, using calibration curves with 10 concentration points. The calibration curves showed high linearity with correlation coefficients (r²) of 0.998–0.999. The limits of detection (LOD) and quantification (LOQ) were determined as 0.12 µg/mL and 0.36 µg/mL, respectively.

Determination of reactive species to 2-thiobarbituric acid

Acid-reactive substances, specifically 2-thiobarbituric acid reactive substances (TBARS), were measured to determine lipid peroxidation (PL) products as an indicator of oxidative stress, as described by Barbosa Filho et al.. (2014)³⁰. For this assay, the reaction solution having 100 µL of supernatant, 100 µL of 10% TCA, and 100 µL of 0.75% 2-thiobarbituric acid (TBA) was incubated at 95 °C for one hour. After cooling, they were centrifuged at 10,000 rpm for 10 min, and absorbance was noted at 405 nm using 250 µL of the reaction solution. Malondialdehyde (MDA), obtained by the hydrolysis of 1,1,3,3-tetramethoxypropane (TMP), was used for the standard curve. Results were articulated as mol of MDA/g tissue.

Determination of free Fe²⁺ content

To assess changes in the free Fe (II) ion amount in the supernatant of cockroach brains, the free Fe2 + content was find using a modified method by Kamdem et al.. (2013)³¹. In this experiment, a reaction solution having 110 µL of saline solution (0.9%), 60 µL of Tris-HCl (0.1 M, pH 7.4), 20 µL of the supernatant, and 10 µL of 0.25% 1,10-phenanthroline was mixed to the microplate, followed by incubation for 1 h at room temperature. Subsequently, after the reaction period, absorbance was read at 492 nm, and the level of free iron (II) in the supernatant was quantified using ferrous sulfate for the standard curve. The results were expressed in µmol Fe (II)/g tissue.

Statistical analysis

The results of biochemical markers and repellency were expressed as mean ± standard error of the mean (SEM). The generated data were analyzed using One-way ANOVA, followed by Post hoc Bonferroni test. Statistical significance for all groups was considered when P < 0.05. The calculation of IC₅₀ was obtained through linear regression using the GraphPad Prism 8 program.

Absorption, distribution, metabolism, and excretion (ADME) study

The calculation of molecular absorption, distribution, metabolism, and excretion (ADME) parameters can be performed using the SwissADME platform, which utilizes the SMILES notation “CC1 = C(C = NO1)C(= O)NC2 = CC = C(C = C2)C(F)(F)F”. This approach allows the evaluation of key properties of the studied ligands, including physicochemical characteristics, lipophilicity, water solubility, pharmacokinetics, and synthetic accessibility³².

DIGEP-Pred 2.0 analysis

The prediction of drug-induced gene expression profiles was carried out using the DIGEP-Pred 2.0 platform. This tool provides qualitative predictions of differentially expressed genes for a query structure by integrating data from two main sources: (1) literature-derived information on mRNA and protein expression changes from the Comparative Toxicogenomics Database (CTD, https://ctdbase.org), and (2) experimental microarray data from the Connectivity Map (cMAP build02, https://clue.io/data/CMB02) obtained in three human cell lines (MCF7, PC3, and HL60)³³. For this study, the analysis was performed using the CTD-mRNA dataset with a probability cut-off of p.a. > 0.9. The regulation type selected was up-regulation. The input structure was Leflunomide, provided as a SMILES string: CC1 = C(C = NO1)C(= O)NC2 = CC = C(C = C2)C(F)(F)F.

Prediction of small molecule protein targets

The prediction was performed using the SMILES of Leflunomide (PubChem CID: 3899) to identify the most likely macromolecular targets for the compound, applying both 2D and 3D similarity analyses. This strategy relies on a comprehensive database of approximately 37,000 molecules, covering interactions with over 3000 proteins, although in this study the focus was restricted to Homo sapiens proteins. The predictions were based on structural similarity, enabling the recognition of potential biological targets and interaction patterns^34,35.

Recipient treatment

Based on the literature review, a key protein target was selected for molecular docking analysis^36,37. The protein structures 6VCD (Cryo-EM structure of IRP2-FBXL5-SKP1 complex) (Fig. 1), 2B3X (Structure of an orthorhombic crystal form of human cytosolic aconitase (IRP1)) (Fig. 2) and 2B3Y (Structure of a monoclinic crystal form of human cytosolic aconitase (IRP1)) (Fig. 3) were sourced from the Protein Data Bank, which provides detailed information on protein structures, sequences, and related metadata. The site was selected by the ligands co-crystallized with the protein. To prepare the receptor for docking, inhibitors and water molecules were removed using DISCOVERY STUDIO 2021 CLIENT. Receptor preparation included adding hydrogen atoms and minimizing the protein structures to optimize geometry for docking. Hydrogen atoms were added, and the protein structures were minimized to optimize geometry for docking.

Fig. 1.

Cryo-EM structure of IRP2-FBXL5-SKP1 complex (PDB ID: 6VCD).

Fig. 2.

Structure of an orthorhombic crystal form of human cytosolic aconitase (IRP1) (PDB ID: 2B3X).

Fig. 3.

Structure of a monoclinic crystal form of human cytosolic aconitase (IRP1) (PDB ID: 2B3Y).

Ligand treatment

The compound leflunomide was selected for in silico molecular docking assessment. The ligand chosen for the study was designed in 3D using ACD/ChemSketch software, and the 2D model was obtained from ChemSpider (leflunomide ID: 3762). Ligands were prepared with consideration of protonation states and minimized prior to docking. The compounds were subjected to ‘rigid protein-flexible ligand’ docking using the Autodock VINA system in the PyRx software³⁸. Post-docking, the most stable ligand conformations were analyzed in DISCOVERY STUDIO.

Grid calculation and Docking

The grid calculation was processed with 100 conformations in the Autodock VINA system of the PyRx software. For the ligand-protein docking procedure, the grid dimensions in the X, Y, and Z axes were set to 40 × 40 × 44 Å for protein 2B3X, 40 × 40 × 40 Å for proteins 2B3Y and 6VCD, with a grid spacing of 0.375 Å. The center of the grid in the X, Y, and Z axes was defined as 36.974, 0.868, and 33.336 Å for protein 2B3X, 38.873, 32.582, and 6.375 Å for protein 2B3Y, and 187.693, 162.643, and 152.005 Å for protein 6VCD. The docking score (binding energy, kcal/mol) (Table S1) was recorded for each ligand-protein complex. AutoDock VINA allows moderate leniency, enabling flexible ligand conformations while maintaining the protein rigid, offering a balance between computational efficiency and reliable binding predictions. Ligand-protein interactions were visualized and analyzed using DISCOVERY STUDIO.

Results

Toxicological parameters

The in silico toxicity assessment of the compound revealed a generally safe profile across multiple exposure routes. Acute inhalation and dermal toxicity were predicted as non-toxic, with probabilities of 69% and 74% (Fig. 4A and B), respectively. Similarly, the compound showed no potential for skin sensitization (70%) (Fig. 4C) or skin irritation/corrosion (80%) (Fig. 4D). However, oral toxicity and eye irritation were indicated as toxic, with probabilities of 86% and 65%, respectively (Fig. 4E and F).

Fig. 4.

In silico analysis of toxicity parameters: Acute Inhalation Toxicity (A), Acute Oral Toxicity (B), Acute Dermal Toxicity (C), Eye Irritation and Corrosion (D), Skin Sensitization (E), and Skin Irritation and Corrosion (F).

After administering the medication in the third abdominal segment of the N. cinerea species, an evaluation was conducted over a 24-hour period to investigate possible toxic effects. The results revealed no manifestation of toxicity throughout the observation period (Fig. 5). The ProTox 3.0 analysis predicted an LD₅₀ of 235 mg/kg, corresponding to toxicity class 3. The compound was identified as active for multiple endpoints, including hepatotoxicity (0.85), neurotoxicity (0.84), respiratory toxicity (0.57), carcinogenicity (0.65), blood–brain barrier penetration (0.91), ecotoxicity (0.69), and clinical toxicity (0.61). Predicted targets included the aryl hydrocarbon receptor (AhR, 0.99), mitochondrial membrane potential (MMP, 1), ATPase family AAA domain-containing protein 5 (ATAD5, 1), transthyretin (TTR, 0.59), voltage-gated sodium channel (VGSC, 0.59), and cytochrome CYP3A4 (0.75) (Table 2).

Fig. 5.

% survival at different concentrations of leflunomide during a 24-hour observation period in the Nauphoeta cinerea model after leflunomide administration.

Table 2.

Toxicity classification of Leflunomide predicted by ProTox 3.0, showing the identified targets, prediction outcomes, and associated probability values.

Classification	Target	Prediction	Probability
Organ toxicity	Hepatotoxicity	Active	0.85
Organ toxicity	Neurotoxicity	Active	0.84
Organ toxicity	Nephrotoxicity	Inactive	0.56
Organ toxicity	Respiratory toxicity	Active	0.57
Organ toxicity	Cardiotoxicity	Inactive	0.79
Toxicity end points	Carcinogenicity	Active	0.65
Toxicity end points	Immunotoxicity	Inactive	0.99
Toxicity end points	Mutagenicity	Inactive	0.54
Toxicity end points	Cytotoxicity	Inactive	0.80
Toxicity end points	BBB-barrier	Active	0.91
Toxicity end points	Ecotoxicity	Active	0.69
Toxicity end points	Clinical toxicity	Active	0.61
Toxicity end points	Nutritional toxicity	Inactive	0.52
Tox21-Nuclear receptor signalling pathways	Aryl hydrocarbon receptor (AhR)	Active	0.99
Tox21-Nuclear receptor signalling pathways	Androgen receptor (AR)	Inactive	0.98
Tox21-Nuclear receptor signalling pathways	Androgen receptor ligand binding domain (AR-LBD)	Inactive	0.99
Tox21-Nuclear receptor signalling pathways	Aromatase	Inactive	0.79
Tox21-Nuclear receptor signalling pathways	Estrogen receptor alpha (ER)	Inactive	0.92
Tox21-Nuclear receptor signalling pathways	Estrogen receptor ligand binding domain (ER-LBD)	Inactive	0.96
Tox21-Nuclear receptor signalling pathways	Peroxisome proliferator activated receptor gamma (PPAR-Gamma)	Inactive	0.98
Tox21-Stress response pathways	Nuclear factor (erythroid-derived 2)-like 2/antioxidant responsive element (nrf2/ARE)	Inactive	0.95
Tox21-Stress response pathways	Heat shock factor response element (HSE)	Inactive	0.95
Tox21-Stress response pathways	Mitochondrial membrane potential (MMP)	Active	1
Tox21-Stress response pathways	Phosphoprotein (Tumor supressor) p53	Inactive	0.92
Tox21-Stress response pathways	ATPase family AAA domain-containing protein 5 (ATAD5)	Active	1
Molecular initiating events	Thyroid hormone receptor alpha (THRα)	Inactive	0.90
Molecular initiating events	Thyroid hormone receptor beta (THRβ)	Inactive	0.53
Molecular initiating events	Transtyretrin (TTR)	Active	0.59
Molecular initiating events	Ryanodine receptor (RYR)	Inactive	0.94
Molecular initiating events	GABA receptor (GABAR)	Inactive	0.76
Molecular initiating events	Glutamate N-methyl-D-aspartate receptor (NMDAR)	Inactive	0.99
Molecular initiating events	alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionate receptor (AMPAR)	Inactive	0.99
Molecular initiating events	Kainate receptor (KAR)	Inactive	0.99
Molecular initiating events	Achetylcholinesterase (AChE)	Inactive	0.93
Molecular initiating events	Constitutive androstane receptor (CAR)	Inactive	1
Molecular initiating events	Pregnane X receptor (PXR)	Inactive	0.51
Molecular initiating events	NADH-quinone oxidoreductase (NADHOX)	Inactive	0.91
Molecular initiating events	Voltage gated sodium channel (VGSC)	Active	0.59
Molecular initiating events	Na+/I- symporter (NIS)	Inactive	0.93
Metabolism	Cytochrome CYP1A2	Inactive	0.54
Metabolism	Cytochrome CYP2C19	Inactive	0.72
Metabolism	Cytochrome CYP2C9	Inactive	0.55
Metabolism	Cytochrome CYP2D6	Inactive	0.68
Metabolism	Cytochrome CYP3A4	Active	0.75
Metabolism	Cytochrome CYP2E1	Inactive	0.99

Parameters of oxidative stress markers

During the analysis of thiol protein parameters (Fig. 6), a significant increase was observed in the presence of leflunomide, suggesting a potential antioxidant effect of the compound. Specifically, it was found that at higher concentrations of the medication, thiol protein levels increased proportionally. This finding indicates the ability of leflunomide, highlighting its potential to mitigate oxidative stress and provide antioxidant benefits in biological systems. It should be stressed that according to ISO 5725-1:2023, the accuracy of thiol content measurements expressed in coefficient of variation (CV%) and standard deviation (SD) was: 8.81% ± 3.95 for Control (-), 10.00% ± 4.32 (64 µg/mL), 1.72% ± 0.94 (128 µg/mL), and 30.81% ± 14.91 for 256 µg/mL.

Fig. 6.

Total thiol levels in homogenates of N. cinerea cockroaches treated with leflunomide. Data are presented as means ± standard error of the mean (SEM) (n = 6). Statistical differences between groups were assessed using one-way ANOVA followed by Bonferroni post-hoc test.

In the analysis of non-protein thiol parameters, it was observed that the concentration of 128 µg/mL of leflunomide exhibited a significant effect compared to the control group (p > 0.0485). Notably, as the concentration of the medication increased, the levels of non-protein thiols also increased proportionally. However, at a concentration of 256 µg/mL, there was a slight decrease in the levels of these thiols, and this effect was not considered statistically significant (Fig. 7). Similarly, the accuracy for non-protein thiol (NPSH) according to ISO 5725-1:2023 was 2.15% ± 0.65 for Control (-), 10.36% ± 3.63 (64 µg/mL), = 6.92% ± 3.01 (128 µg/mL), and 14.94% ± 6.02 for 256 µg/mL. These results suggest a possible dose-dependent relationship between leflunomide and non-protein thiols, although at higher concentrations, there may be a saturation point.

Fig. 7.

Non-protein thiol levels in homogenates of N. cinerea cockroaches treated with leflunomide. Data are presented as means ± standard error of the mean (SEM) (n = 6). Statistical differences between groups were assessed using one-way ANOVA followed by Bonferroni post-hoc test. * Significant effect compared to the control. Significant differences are indicated by asterisks: *p < 0.05, * *p < 0.01, * **p < 0.001, * ** *p < 0.0001.

In the TBARS assays, it was observed that the levels of MDA consistently increased as the concentration of leflunomide increased (Fig. 8). This trend was observed in parallel with other assays, where the higher concentration (256 µg/mL) of the medication resulted in an increase in the levels of biological markers. These findings suggest a possible relationship between the compound concentration and the biochemical processes associated with the production of MDA and other lipid oxidation products. It is important to note that this increase in biomarker levels may indicate either increased oxidative activity or a complex biological response to leflunomide.

Fig. 8.

Malondialdehyde content in N. cinerea homogenate treated with leflunomide. Data are presented as means ± standard error of the mean (SEM) (n = 6). Statistical differences between groups were assessed using one-way ANOVA followed by Bonferroni post-hoc test.

After applying a concentration of 256 µg/mL of leflunomide to the species N. cinerea, a significant increase in iron levels was observed (p > 0.0162). The reduction was also evident when comparing the concentration of 256 µg/mL with that of 64 µg/mL, demonstrating an equally significant effect (p > 0.0161) (Fig. 9). However, as the dose increased, there was an increase in iron levels, presenting a dose-response effect.

Fig. 9.

Free iron levels in homogenates N. cinerea, treated with leflunomide. Data are presented as means ± standard error of the mean (SEM) (n = 6). Statistical differences between groups were assessed using one-way ANOVA followed by Bonferroni post-hoc test. * Significant effect compared to the control; & significant effect compared to concentrations 64 µg/mL and 256 µg/mL. Significant differences are indicated by asterisks: *p < 0.05, * *p < 0.01, * **p < 0.001, * ** *p < 0.0001.

In silico parameters

The in silico analysis of Leflunomide (C₁₂H₉F₃N₂O₂, 270.21 g/mol) revealed favorable pharmacokinetic and drug-likeness properties. The compound has 4 rotatable bonds, 6 hydrogen bond acceptors, 1 donor, a TPSA of 55.13 Ų, and logP consensus of 2.83, suggesting good lipophilicity. Pharmacokinetics indicate high gastrointestinal absorption, blood–brain barrier permeability, and non-substrate status for P-glycoprotein. Leflunomide inhibits CYP1A2 and CYP2C19. Drug-likeness evaluation shows full compliance with Lipinski, Ghose, Veber, Egan, and Muegge rules, a bioavailability score of 0.55, no PAINS or Brenk alerts, positive lead-likeness, and a synthetic accessibility score of 2.45 (Table 3).

Table 3.

Numerical results from SwissADME for Leflunomide - physicochemical properties, lipophilicity, pharmacokinetics, drug similarity, and medicinal chemistry.

Physicochemical properties
Formula	C12H9F3N2O2
Molecular weight	270.21 g/mol
Number. Heavy atoms	19
Number. Aroma. Atoms	11
Csp3 fraction	0.17
Number. Revolving bonds	4
Number. H-link acceptors	6
Number. H-link donors	1
Molar refractivity	60.69
TPSA	55.13 Å2
Lipophilicity
Log Po/w (iLOGP)	2.33
Log Po/w (XLOGP3)	2.49
Log Po/w (WLOGP)	4.22
Log Po/w (MLOGP)	2.14
Log Po/w (SILICOS-IT)	2.97
Log logging Po/w	2.83
Pharmacokinetics
GI absorption	High
BBB permeant	Yes
P-gp substrate	No
CYP1A2 inhibitor	Yes
CYP2C19 inhibitor	Yes
CYP2C9 inhibitor	No
CYP2D6 inhibitor	No
CYP2A4 inhibitor	No
Log Kp (skin permeation)	-6.18 cm/s
Similarity with drugs
Lipinski	Yes; 0 violation
Ghose	Yes
Veber	Yes
Egan	Yes
Muegge	Yes
Bioavailability score	0.55
Medicinal chemistry
Pain	0 alert
Brenk	0 alert
Lead resemblance	Yes
Synthetic accessibility	2.45

The analysis identified several genes predicted to be up-regulated by Leflunomide with high confidence. Among them, BTBD8 (Pa = 0.939, IAP = 0.986), FAM86B3P (Pa = 0.949, IAP = 0.973), and ZNF626 (Pa = 0.969, IAP = 0.972) showed strong probabilities. Additional predicted targets included MBL2 (Pa = 0.919, Pi = 0.001, IAP = 0.971), MTBP (Pa = 0.981, IAP = 0.966), ZNF100 (Pa = 0.981, IAP = 0.966), and CERS6 (Pa = 0.930, Pi = 0.002, IAP = 0.956). Overall, the predictions presented high internal accuracy, as reflected by the leave-one-out cross-validation (LOO CV) scores above 0.95 (Table S2).

The predictive analysis indicated that Leflunomide is likely to interact with different molecular targets. The main predicted targets were dihydroorotate dehydrogenase (CHEMBL1966, oxidoreductase), norepinephrine transporter (CHEMBL222, electrochemical transporter), dopamine transporter (CHEMBL238, electrochemical transporter), and cytochrome P450 1A2 (CHEMBL3356, cytochrome P450), all with the highest probability score (Table 4). The overall distribution of predicted targets showed predominance of nuclear receptors (24%), followed by electrochemical transporters (16%), voltage-gated ion channels (14%), and enzymes (10%). Other classes, including oxidoreductases, cytochrome P450, kinases, G protein-coupled receptors, and others, represented smaller fractions ranging from 2% to 4% (Fig. 10).

Table 4.

Redicted macromolecular targets for Leflunomide based similarity analyses. The table lists the target name, corresponding chembl identifier, classification of the target, and the probability score* associated with the prediction.

Target	ChEMBL ID	Target class	Probability*
Dihydroorotate dehydrogenase (by homology)	CHEMBL1966	Oxidoreductase	1
Norepinephrine transporter	CHEMBL222	Electrochemical transporter	1
Dopamine transporter	CHEMBL238	Electrochemical transporter	1
Cytochrome P450 1A2	CHEMBL3356	Cytochrome P450	1
Cathepsin S	CHEMBL2954	Protease	0.112041901
Estradiol 17-beta-dehydrogenase 3	CHEMBL4234	Enzyme	0.112041901
Cathepsin K	CHEMBL268	Protease	0.112041901
Leucine-rich repeat serine/threonine-protein kinase 2	CHEMBL1075104	Kinase	0.112041901
Protein-tyrosine phosphatase 1B	CHEMBL335	Phosphatase	0.112041901
Androgen Receptor	CHEMBL1871	Nuclear receptor	0.112041901
Serotonin 6 (5-HT6) receptor	CHEMBL3371	Family A G protein-coupled receptor	0.112041901
Calcium-activated potassium channel subunit alpha-1	CHEMBL4304	Voltage-gated ion channel	0.112041901
c-Jun N-terminal kinase 1	CHEMBL2276	Kinase	0.112041901
Toll-like receptor (TLR7/TLR9)	CHEMBL5804	Toll-like and Il-1 receptors	0.112041901
Microtubule-associated protein 2	CHEMBL2390810	Unclassified protein	0.112041901
Tyrosine-protein kinase SYK	CHEMBL2599	Kinase	0.112041901
Glycogen synthase kinase-3 beta	CHEMBL262	Kinase	0.112041901
Macrophage colony stimulating factor receptor	CHEMBL1844	Kinase	0.112041901
11-beta-hydroxysteroid dehydrogenase 1	CHEMBL4235	Enzyme	0.112041901
Muscarinic acetylcholine receptor M5	CHEMBL2035	Family A G protein-coupled receptor	0.112041901
Muscarinic acetylcholine receptor M1 (by homology)	CHEMBL216	Family A G protein-coupled receptor	0.112041901
Voltage-gated potassium channel. KQT	CHEMBL2476	Voltage-gated ion channel	0.112041901
Voltage-gated potassium channel subunit Kv7.3	CHEMBL2684	Voltage-gated ion channel	0.112041901
Voltage-gated potassium channel subunit Kv7.5	CHEMBL2925	Voltage-gated ion channel	0.112041901
Voltage-gated potassium channel subunit Kv7.4	CHEMBL3576	Voltage-gated ion channel	0.112041901
Serine/threonine-protein kinase PIM1	CHEMBL2147	Kinase	0.112041901
Serine/threonine-protein kinase PIM2	CHEMBL4523	Kinase	0.112041901
Endothelial PAS domain-containing protein 1	CHEMBL1744522	Unclassified protein	0.112041901
Cyclin-dependent kinase 2	CHEMBL301	Kinase	0.112041901
Cyclin-dependent kinase 4	CHEMBL331	Kinase	0.112041901
P2X purinoceptor 7	CHEMBL4805	Ligand-gated ion channel	0.112041901
Serine/threonine-protein kinase Aurora-A	CHEMBL4722	Kinase	0.112041901
Intercellular adhesion molecule-1	CHEMBL3070	Adhesion	0.112041901
Selectin E	CHEMBL3890	Adhesion	0.112041901
Glutaminyl-peptide cyclotransferase	CHEMBL4508	Enzyme	0.112041901
Trace amine-associated receptor 1 (by homology)	CHEMBL5857	Family A G protein-coupled receptor	0.112041901
c-Jun N-terminal kinase 3	CHEMBL2637	Kinase	0.112041901
Arachidonate 15-lipoxygenase	CHEMBL2903	Enzyme	0.112041901
Metabotropic glutamate receptor 5	CHEMBL3227	Family C G protein-coupled receptor	0.112041901
Proto-oncogene c-JUN	CHEMBL4977	Transcription factor	0.112041901
Adenosine A1 receptor (by homology)	CHEMBL226	Family A G protein-coupled receptor	0.112041901
Xanthine dehydrogenase	CHEMBL1929	Oxidoreductase	0.112041901
Progesterone receptor	CHEMBL208	Nuclear receptor	0.112041901
ALK tyrosine kinase receptor	CHEMBL4247	Kinase	0.112041901
Poly [ADP-ribose] polymerase 2 (by homology)	CHEMBL5366	Enzyme	0.112041901
Gamma-secretase	CHEMBL2094135	Protease	0.112041901
Serotonin 2a (5-HT2a) receptor	CHEMBL224	Family A G protein-coupled receptor	0.112041901
Serotonin 2c (5-HT2c) receptor	CHEMBL225	Family A G protein-coupled receptor	0.112041901
PI3-kinase p110-delta subunit	CHEMBL3130	Enzyme	0.112041901
PI3-kinase p110-gamma subunit	CHEMBL3267	Enzyme	0.112041901
PI3-kinase p110-alpha subunit	CHEMBL4005	Enzyme	0.112041901
Monoamine oxidase B	CHEMBL2039	Oxidoreductase	0.112041901
Tyrosyl-DNA phosphodiesterase 2	CHEMBL2169736	Enzyme	0.112041901
Cyclin-dependent kinase 2/cyclin A	CHEMBL2094128	Other cytosolic protein	0.112041901
Arachidonate 5-lipoxygenase	CHEMBL215	Oxidoreductase	0.112041901
Cyclooxygenase-2	CHEMBL230	Oxidoreductase	0.112041901
Phosphodiesterase 3	CHEMBL241	Phosphodiesterase	0.112041901
Phosphodiesterase 4B	CHEMBL275	Phosphodiesterase	0.112041901
c-Jun N-terminal kinase 2	CHEMBL4179	Kinase	0.112041901
Tyrosine-protein kinase JAK1	CHEMBL2835	Kinase	0.112041901
Protein Wnt-3a	CHEMBL1255137	Unclassified protein	0.112041901
Cyclin-dependent kinase 2/cyclin E	CHEMBL2094126	Other cytosolic protein	0.112041901
Tyrosine-protein kinase JAK2	CHEMBL2971	Kinase	0.112041901
Histone deacetylase 8	CHEMBL3192	Eraser	0.112041901
Tyrosine-protein kinase TYK2	CHEMBL3553	Kinase	0.112041901
Elongation of very long chain fatty acids protein 6	CHEMBL5704	Enzyme	0.112041901
Tankyrase-2	CHEMBL6154	Enzyme	0.112041901
Tankyrase-1	CHEMBL6164	Enzyme	0.112041901
Neuronal acetylcholine receptor protein alpha-7 subunit	CHEMBL2492	Ligand-gated ion channel	0.112041901
Estradiol 17-beta-dehydrogenase 2	CHEMBL2789	Enzyme	0.112041901
G-protein coupled receptor kinase 2	CHEMBL4079	Kinase	0.112041901
Cyclin-dependent kinase 5/CDK5 activator 1	CHEMBL1907600	Kinase	0.112041901
Caspase-1	CHEMBL4801	Protease	0.112041901
Egl nine homolog 1	CHEMBL5697	Oxidoreductase	0.112041901
Tyrosine-protein kinase SRC	CHEMBL267	Kinase	0.112041901
Signal transducer and activator of transcription 3	CHEMBL4026	Transcription factor	0.112041901
Sodium/hydrogen exchanger 1	CHEMBL2781	Electrochemical transporter	0.112041901

Fig. 10.

Biological structures in the species Homo sapiens that interact with Leflunomide.

Figure 11A–F illustrates the binding of the leflunomide compound to target proteins: 6VCD in 3D (Fig. 11A) and 2D (Fig. 11B), 2B3X in 3D (Fig. 11C) and 2D (Fig. 11D), and 2B3Y in 3D (Fig. 11E) and 2D (Fig. 11F). The medication leflunomide demonstrated a significant affinity of -7.8 kcal/mol with the target protein 6VCD. This compound revealed a series of essential interactions at the binding sites, including ASP A:881, GLU A:945, HIS A:882, GLY A:885, LEU A:954, LEU A:308, and ALA A:303 (as shown in Fig. 9B). The formation of alkyl and pi-alkyl bonds was observed at the ALA A:303, LEU A:308, and LEU A:954 sites, while the GLY A:885 site showed halogen bonding with fluorine. Additionally, conventional hydrogen bonds were identified at the GLU A:945 site, accompanied by pi-anion bonds. The presence of carbon-hydrogen bonds at the ASP A:881 and HIS A:882 sites was also observed, with the ASP A:881 site demonstrating an additional pi-anion bond (Table 5).

Fig. 11.

Binding of the leflunomide compound to target protein 6VCD in 3D (A), 2D image of the leflunomide compound on target protein 6VCD (B), Binding of the leflunomide compound to target protein 2B3X in 3D (C), 2D image of the leflunomide compound on target protein 2B3X (D), Binding of the leflunomide compound to target protein 2B3Y in 3D (E), and 2D image of the leflunomide compound on target protein 2B3Y (F).

Table 5.

Molecular docking results showing the docking score of Leflunomide with its predicted targets, along with the interacting residues, bond distances, and bond types involved in the ligand–protein interactions.

	Docking score	Interacting residues	Bond distances	Bond type
6VCD	-7.8 kcal/mol	ASP A:881	3.44 A	Carbon-hydrogen
			4.99 A	Pi-anion
		GLU A:945	3.91 A	Pi-anion
			2.49 A	Conventional hydrogen
		HIS A:882	5.47 A	Pi-alkyl
			3.44 A	Halogen
			3.27 A	Carbon-hydrogen
				Halogen
		GLY A:885	3.46 A	Halogen
			2.93 A
		LEU A:954	4.69 A	Alkyl
		LEU A:308	5.14 A	Pi-alkyl
		ALA A:303	4.03 A	Alkyl
			4.33 A	Pi-alkyl
2B3X	-7.8 kcal/mol	MET A:464	2.22 A	Conventional hydrogen
		ALA A:561	4.77 A	Pi-alkyl
		VAL A:429	4.95 A	Pi-alkyl
		ALA A:432	5.27 A	Pi-alkyl
		VAL A:463	5.26 A	Pi-alkyl
		VAL A:430	3.51 A	Carbon-hydrogen
		PHE A:422	5.30 A	Pi-pi t-shaped
			5.03 A	Pi-alkyl
		VAL A:517	5.38 A	Pi-alkyl
		PHE A:415	4.96 A	Pi-alkyl
2B3Y	-6.6 kcal/mol	PHE A:132	4.82 A	Pi-pi t-shaped
		ARG A:135	4.91 A	Pi-alkyl
		ASN A:133	2.28 A	Conventional hydrogen
			3.35 A	Halogen
			2.33 A	Conventional hydrogen
				Halogen
		ILE A:467	2.77 A	Halogen
		ARG A:134	5.49 A	Pi-alkyl
			2.35 A	Conventional hydrogen
				Halogen
		PRO A:514	4.66 A	Alkyl
		PRO A:465	3.44 A	Halogen
			3.29 A
		TYR A:466	3.10 A	Halogen

When leflunomide bound to the protein 2B3X (as shown in Fig. 11D), a remarkable molecular activity in its structure was observed, evidenced by a significant affinity of -7.8 kcal/mol. This binding was established through interactions at the sites MET A:464, ALA A:561, VAL A:429, ALA A:432, VAL A:463, VAL A:430, PHE A:422, VAL A:517, and PHE A:415. At the MET A:464 sites, the formation of conventional hydrogen bonds was identified, while at the VAL A:430 site, carbon-hydrogen interactions occurred. On the other hand, at the PHE A:422 site, a pi-pi t-shaped bond was observed. The remaining sites showed pi-alkyl interactions (Table 5).

When analyzing leflunomide bound to the target protein 2B3Y (as shown in Fig. 11F), a strong interaction with the protein was evidenced, resulting in a significant affinity of -6.6 kcal/mol. This binding occurred at the sites PHE A:132, ARG A:135, ASN A:133, ILE A:467, ARG A:134, PRO A:514, PRO A:465, and TYR A:466. At the ASN A:133 and ARG A:134 sites, the formation of conventional hydrogen bonds was observed, while at the ILE A:467, PRO A:465, and TYR A:466 sites, halogen bonding with fluorine was identified. Additionally, at the PHE A:132 site, a pi-pi t-shaped bond was observed. At the ARG A:135 and PRO A:514 sites, the interactions were of the alkyl and pi-alkyl types, respectively (Table 5).

Discussion

Leflunomide did not exhibit toxicity after 24 h of exposure; however, the biochemical parameters showed an increase in oxidative stress markers. This suggests that the compound may have induced oxidative stress, as higher doses were associated with elevated levels of lipid peroxidation. In contrast, in-silico analyses indicated that leflunomide presents acute oral toxicity potential and shows strong interactions with kinase-type protein structures.

Leflunomide is an antirheumatic drug capable of reducing the expression of inducible nitric oxide synthase, nitrotyrosine, TNF-α, and IL-1β, as well as decreasing serum MDA levels⁴⁰. After the administration of the medication in the third segment of N. cinerea, no toxicity was observed at the tested concentrations. Furthermore, the medication demonstrated activity in increasing levels of protein and non-protein thiols, TBARS, and iron levels in the tested model.

It is widely recognized that isoxazole derivatives exhibit a diversity of biological activities and are acknowledged for their potential use against a variety of diseases, including infectious and parasitic infections, as well as in oncological therapy. A notable example is leflunomide (Avara), an immunomodulator used to treat symptoms associated with rheumatoid arthritis (RA) and psoriatic arthritis⁴¹. Despite being a low molecular weight isoxazole derivative, leflunomide is considered one of the most potent, albeit associated with severe side effects^42,43.

Leflunomide revealed an increase in the levels of biochemical parameters in the conducted tests, particularly notable at the higher concentration of 256 µg/mL, which resulted in a significant increase in levels of protein thiols, non-protein thiols, and TBARS. Although assays for iron levels showed an increase as the concentration increased. However, in a study conducted by Yildiz et al. (2010)⁴⁴. on the protective effects of leflunomide in intestinal ischemia-reperfusion injury, the combination of leflunomide and ischemia-reperfusion demonstrated notable results: 0.872 ± 0.254 for CAT parameter, 51.64 ± 2.97 for GSH, 9.699 ± 0.681 for SOD, and 17.56 ± 0.87 for GR.

Considering that leflunomide is a DMARD capable of inhibiting TNF-α and thereby suppressing the inflammatory process to prevent the recurrence of inflammation characteristic of rheumatic diseases^45,46, the results of this analysis highlighted an increase in TBARS, which serve as indicators of oxidative stress. This was measured through MDA levels, a byproduct of lipid peroxidation. These findings suggest that the organism exhibited signs of oxidative stress due to lipid peroxidation when exposed to higher concentrations of leflunomide. However, the organism showed an increase in protein thiol groups, possibly defending the organism against the stress caused.

According to Moon et al. (2017)⁴⁰, leflunomide, an anti-rheumatic medication, has demonstrated the ability to decrease the expression of inducible nitric oxide synthase, nitrotyrosine, TNF-α, and IL-1β, as well as reducing serum levels of MDA. These results suggest an explanation for its immunomodulatory and anti-rheumatic effects^47–51. Leflunomide is often used as a second-line therapeutic option, either as monotherapy or in combination with other conventional DMARDs, in patients with resistant disease^52–56. However, a significant limitation in achieving remission with leflunomide is that up to 40% of patients discontinue therapy due to toxicity^57,58.

However, when used in the N. cinerea model, leflunomide did not reveal any signs of toxicity during the 24-hour investigation period. This suggests that even when administered over a short period of time and at low concentrations, leflunomide did not exhibit adverse effects in the N. cinerea model. These results show the possibility that leflunomide can be safely used at explicit doses and time periods without causing harmful effects. However, in the long term, it can cause greater stress, leading to toxicity. In addition, toxicity was seen in acute oral use, possibly due to intestinal stress.

Adverse effects associated with the use of leflunomide for the treatment of rheumatoid arthritis may lead to an increase in hepatic transaminases up to twice the levels considered normal, in addition to skin rashes. These symptoms can be mitigated or eliminated by discontinuing treatment or reducing the administered doses⁵⁹.

Hepatic enzyme alterations leading to hepatotoxicity are among the most concerning adverse reactions associated with drugs used to treat rheumatic diseases^60–62. Although DMARDs exhibit anti-inflammatory activity, they can also induce adverse effects, including hepatotoxicity. Nephrotoxicity refers to kidney damage caused by the toxic effects of substances and medications that impair renal function. Both ROS-induced cell death and hepatic injury frequently involve lipid peroxidation as an underlying mechanism^63,64, which may explain the observed increases in excessive MDA production, along with elevated thiol levels. This is an adaptive response to the stresses caused.

Iron (Fe) regulatory proteins (IRPs) regulate the translation of Fe-related proteins by binding to iron-responsive elements (IREs) on mRNA. In Fe deficiency, they confirm Fe homeostasis in animal cells. In Fe surplus, IRP1 acts as cytosolic aconitase. Regulation of IRP1 is complex and not solely dependent on iron availability. Regulation of cellular Fe homeostasis involves FBXL5-mediated degradation of IRP2, dependent on Fe and oxygen. The possible interaction between FBXL5 and IRP2 is influenced by a [2Fe2S] cluster in the FBXL5 binding domain. The IRP2-FBXL5-SKP1 complex reveals that this cluster organizes the C-terminal loop of FBXL5 for IRP2 recruitment^65–68.

One of the consequences of iron deficiency in the body is anemia, resulting in small red blood cells (microcytic) and red blood cells with hemoglobin deficiency (hypochromic)⁶⁶.

In the study proposed by Pereira et al. (2018)⁶⁹, it was demonstrated that Leuflonomide can attenuate anemia in patients with Rheumatoid Arthritis (RA) indicated by increased hemoglobin levels and decreased Erythrocyte Sedimentation Rate (ESR). These results were promising to demonstrate that LFN can be effective in the treatment of anemia in patients with (RA)⁷⁰.

Leflunomide is a prodrug that undergoes rapid conversion through cleavage of the isoxazole ring, resulting in the active metabolite Teriflunomide (TFL)^71–73. Its main action against Rheumatoid Arthritis (RA) is the inhibition of de novo pyrimidine synthesis^74–76.

Cleavage is the process by which a larger molecule is fragmented into smaller units. This process results in the formation of teriflunomide, which is the active form of the drug. During molecular analyses, we observed certain bonds in leflunomide on its isoxazole ring. In the target protein (6VCD), we identified the bonds APS A:881 and GLU A:945. In protein 2B3X, we observed significant binding activity at sites ALA A:561, VAL A:429, ALA A:432, and VAL A:463, all demonstrating a pi-alkyl type binding. Additionally, in protein 2B3Y, we detected the presence of bonds PHE A:132 and ARG A:135.

A pi-alkyl bond is a type of non-covalent interaction that occurs between functional groups in molecules. In this type of bonding, a p (pi) orbital of an unsaturated carbon atom, such as an aromatic ring, interacts with an alkyl group, which is a saturated carbon chain^77–80. In drug chemistry, pi-alkyl interactions can influence a molecule’s affinity for its biological target, thereby affecting its pharmacological activity^81–83.

It should be stressed that the accuracy of molecular docking depends on several factors, including the quality of the receptor model (protein or enzyme), the ligand generation method, and the choice of docking algorithm. In our study, we used AutoDock Vina 1.2.5, which is widely recognized for its reliable binding affinity predictions. However, docking accuracy is also influenced by ligand and receptor flexibility, as well as system conditions. To validate our results, we analysed predicted interactions against available crystal structures where possible and assessed the consistency across different docking configurations. Additionally, we evaluated binding energy, hydrogen bonding interactions, and comparisons with experimental data to enhance result reliability. Future analyses could further improve accuracy through refined docking methods and additional experimental validation.

Conclusion

Leflunomide altered biochemical parameters in Nauphoeta cinerea, promoting increases in biological markers of oxidative stress according to the dose applied and did not show toxicity at the concentrations tested. Molecular analysis indicated interactions with proteins associated with oxidative stress and metal regulation. In addition, in silico assessment revealed toxicity parameters, affinity with kinases, enzymes, and phosphatases, as well as potential inhibitory effects on CYP1A2 and CYP2C19. The compound also showed high affinity for dihydroorotate dehydrogenase (by homology), and interactions with norepinephrine and dopamine transporters, reinforcing its influence on metabolic and oxidative processes. Further studies on enzymatic parameters and mRNA expression will deepen understanding of LF’s effects in Nauphoeta cinerea. Expanding research to alternative models and integrating experimental validation with molecular docking will help clarify and generalize its mechanisms and interactions.

Manuscript limitations

It should be stressed that this study has certain limitations, particularly the absence of molecular dynamics simulations, which could have provided deeper insight into the stability and temporal behavior of the ligand–protein interactions predicted by the docking analysis. Additionally, although the biochemical assays revealed important oxidative stress parameters, further analyses such as enzymatic activity profiling and gene expression studies were not included and could deepen the understanding of the mechanisms triggered by leflunomide in Nauphoeta cinerea. Future work addressing these aspects will help strengthen and expand the conclusions presented here.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Each author participated sufficiently in taking public responsibility for appropriate portions of the content. Study conception and design: CALS and AAAM, conceived the idea and designed experiments and wrote manuscript. JPK, BRST and AED analyzed the data and performed the experiments; HAA, MRA, AAA and MI analyzed the data and revised the manuscript. All authors reviewed and approved the final version.

Data availability

All data generated or analyzed during this study are included in this published article.

Declarations

Competing interests

The authors declare no competing interests.