2026 Volume 51 Issue 9 Pages 471-487
Methylisothiazolinone (MIT) is a widely used biocidal preservative in industrial and cosmetic products. However, evidence of its neurotoxic effects is limited, and the mechanistic pathways are poorly defined. This study investigated the neurotoxic potential of MIT and the molecular mechanisms underlying its cellular toxicity in SH-SY5Y cells. After exposing SH-SY5Y cells to MIT various concentrations (30 to 120 μM) for 24 hr, apoptosis and inflammatory responses were quantified alongside the determination of apoptotic-, oxidative stress-, cell survival-, and angiogenic-related biomarkers. To reinforce the experimental findings, a set of in silico chemoinformatics studies, including network toxicology and molecular docking were conducted to pinpoint and evaluate the potential molecular targets of MIT. The IC50 value of MIT was found 115 µM. Total apoptotic cells increasing to 56.5% at 120 µM. ELISA results indicated activation of the intrinsic apoptotic pathway, evidenced by significant upregulation of p53 (9.54-fold), BAX (4.52-fold), and APAF-1 (8.85-fold), along with a disruption in the BAX/BCL-2 (4.52/4.37-fold) balance. Furthermore, an elevation in the COX/COX-2 (8.04/6.93-fold), SRC (7.08-fold), and VEGFR2 (20.96-fold) levels which indicated the concurrent oxidative stress and stress-induced pro-survival signaling at 120 µM. MIT selectively amplified some inflammatory mediators, suggesting inflammasome-driven neuroinflammation. In silico analysis further support MIT as a polypharmacological toxicant with the potential to modulate key molecular targets, including PARP-1, GSK-3β, and iNOS. MIT induced multifaceted neurotoxicity in SH-SY5Y cells through the simultaneous activation of intrinsic apoptotic signaling, oxidative stress, and selective neuroinflammatory pathways. The findings emphasize the potential neurotoxic risks associated with MIT exposure and underscore the necessity for strengthened regulatory evaluation and in vivo validation.
Methylisothiazolinone (MIT), a heterocyclic isothiazolinone derivative, is widely utilized as a preservation agent in cosmetics, healthcare products, household cleaning items, and numerous industrial applications such as adhesives, coatings, and cooling water systems due to its extensive biocidal efficacy and advantageous solubility characteristics. Its antimicrobial efficacy derives from the ability of its nitrogen–sulfur heterocyclic ring to react with thiol-containing residues in proteins and enzymes, leading to the disruption of microbial metabolism and death of cells (He et al., 2006; Burnett et al., 2010; Li et al., 2024). Regulatory surveys indicate its use in hundreds of cosmetic formulations worldwide; for example, 2019 Food and Drug Administration (FDA) Voluntary Cosmetic Registration Program (VCRP) data reported 915 formulations containing MIT, predominantly bath soaps and detergents, with maximum concentrations ranging between 0.02 and 97.5 ppm. Despite these advantages, MIT belongs to a chemical class recognized for its sensitizing potential, raising concerns about dermatological reactions, ecological hazards, and, more recently, possible neurotoxic effects (Du et al., 2002; Burnett et al., 2010; Lundov et al., 2011; Scherrer et al., 2015; Silva et al., 2020).
Household exposure also represents an important route of concern, since MIT and related isothiazolinone compounds are widely incorporated into both personal care items and home cleaning formulations, frequently at quantities reaching 15 ppm, and have also been identified in indoor airflow (Spawn and Aizenman, 2012). Initial safety assessments determined that MIT is tolerable at concentrations as high as 100 ppm (Spawn and Aizenman, 2012); however, subsequent evaluations revealed significant sensitization risks, leading to more stringent regulatory restrictions, including the 2020 The Cosmetic Ingredient Review (CIR) amendment that reaffirmed safety only for rinse-off products at ≤100 ppm and highlighted the absence of a safe threshold for leave-on formulations (Shank et al., 2020). Importantly, while current specifications focus on dermatological safety, the enduring neurological effects of chronic environmental contact with “acceptable” concentrations of MIT remain poorly characterized (Spawn and Aizenman, 2012). Previous in vitro studies have demonstrated that acute exposure to elevated concentrations of MIT induces widespread and selective cell death of neurons, underscoring the need to evaluate neurotoxic effects even within low-dose exposure contexts (He et al., 2006).
Exposure to environmental toxicants has a significant impact on numerous neurodevelopmental and neurodegenerative disorders, with MIT increasingly acknowledged as a potential risk factor (Du et al., 2002; He et al., 2006; Van Huizen et al., 2017; Lee et al., 2025). In vitro research has shown that even brief, acute exposures to MIT at concentrations similar to those found in household items (~15 ppm) can induce widespread neuronal cell death (Spawn and Aizenman, 2012). Previous research has also shown that MIT exerts highly selective neurotoxicity, largely sparing glial populations while causing extensive neuronal loss following acute exposure in cerebrocortical cultures (He et al., 2006). Experimental work in embryonic rat cortical cultures has revealed dose-dependent neuronal death at higher concentrations (100–300 µM) within 24 hr, accompanied by disruptions in glutathione homeostasis and MAPK signaling pathways (Du et al., 2002). Additional studies at lower concentrations (≤3.0 µM) indicated inhibition of neuronal process outgrowth, reductions in focal adhesion kinase (FAK) phosphorylation, and suppression of Src family kinase (SFK) activity, suggesting that even modest, prolonged exposures may interfere with neuronal development (He et al., 2006). More recent evaluations have confirmed that MIT disrupts synaptic growth and neuronal signaling in vitro, underscoring its potential to impair neurodevelopment (Li et al., 2024). However, despite mounting in vitro evidence, the 2020 CIR amended safety assessment reported that no direct in vivo neurotoxicity studies are available, leaving the long-term neurological consequences of MIT exposure unresolved (Shank et al., 2020). Furthermore, prior research has inadequately investigated the interactions among apoptotic, inflammatory, and survival pathways in human neuronal cells, which may be essential for understanding MIT-induced neurotoxicity. In this context, mechanistic investigations in neuronal cell models remain essential to bridge this knowledge gap and to guide future in vivo and regulatory studies.
In silico chemoinformatics approaches have emerged as valuable tools in toxicological research by helping to elucidate the mechanistic basis of observed in vitro effects (Zhang et al., 2025). Among these, network toxicology is a systems-level strategy used to identify key shared targets by integrating chemical–target associations with disease-related pathways, thereby revealing potential regulatory hubs involved in toxic responses (Wang et al., 2025a; Yanli et al., 2025). In addition, molecular docking is widely applied to evaluate the binding potential and interaction profiles of compounds with predicted targets, providing structural support for target engagement (Lan et al., 2025; Wang et al., 2025b). Together, these complementary approaches offer an integrated framework for understanding multi-target toxicological mechanisms and for supporting the interpretation of experimental findings while reducing the need for extensive experimental screening (Bao et al., 2025).
In the present study, the toxicological profile of MIT was investigated through an integrated experimental and computational strategy. A series of in vitro assays were conducted to characterize its cytotoxic, oxidative, inflammatory, and apoptotic effects. The primary objective was to characterize the dose-dependent neurotoxic effects of MIT and to identify the key molecular mediators involved in these processes. Specifically, mechanistic insights were obtained by examining the expression profiles of key apoptotic and survival-related targets, including tumor protein p53 (p53), proto-oncogene tyrosine-protein kinase Src (SRC), apoptotic protease activating factor-1 (APAF1), vascular endothelial growth factor receptor-2 (VEGFR2), cyclooxygenase (COX), cyclooxygenase (COX-2), BCL-2 associated X protein (BAX), and B-cell lymphoma 2 (BCL-2). Furthermore, our research aimed to identify plausible molecular targets and elucidate potential mechanistic pathways by enabling a comprehensive assessment of MIT’s toxicity, spanning phenotypic outcomes to target-level interactions. This approach allowed for a detailed evaluation of both the cytotoxic and molecular responses elicited by MIT in human neuronal cells, providing critical mechanistic data that complement previous in vitro studies and inform future in vivo and regulatory investigations.
Methylisothiazolinone (MIT, purity 98.4%) powder (CAS No: 2682-20-4) from Santa Cruz (USA), MTT (3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyl-tetrazolium bromide) and dimethyl sulfoxide (DMSO) were sourced from Sigma Chemical Co. Ltd (MO, USA), while phosphate-buffered saline (PBS), cell culture medium, and all other supplements were acquired from Multicell Wisent (Quebec, Canada). Additionally, Annexin V FITC Apoptosis Detection Kit with PI and LEGENDPlex™ human inflammation panel was obtained from Biolegend (San Diego, USA), and sterile plastic materials were obtained from Nest (Jiangsu, China).
Cell culture and exposure conditionsHuman neuroblastoma SH-SY5Y cell line was obtained by the American Type Culture Collection (ATCC, CRL-2266). The cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) enriched with 10% fetal bovine serum (FBS) and 1% penicillin-streptomycin. The cells were grown at 37°C in a humidified incubator containing 5% CO2. In the experiments, MIT was solubilized using DMSO to generate standard solutions (Kovalevich et al., 2021). Each experiment was performed in duplicates with three distinct biological replicates.
CytotoxicitySH-SY5Y cells were added to 96-well plates at a count of 5×103 cells per well and then permitted to adhere for one day under usual culture conditions (37°C, 5% CO2). After the cells were seeded, MIT exposure was performed. The cells were subsequently exposed to MIT at doses ranging from 500 to 15.625 µM for 24 hr, with the untreated cells acting as controls. After 24 hr of exposure, 25 µL of MTT (5 mg/mL) was added to every well, and the plate was incubated in the incubator for 3 hr. The supernatant was removed, and the crystals of formazan were dissolved in 100 µL of DMSO. Absorbance was measured at 570 nm using a microplate reader (Epoch, Germany) (Ghasemi et al., 2021). Cell viability was quantified as a percentage relative to the untreated control group.
Determination of apoptosisThe Annexin V-FITC/PI double staining experiment was performed using a flow cytometer to determine apoptosis. SH-SY5Y cells were cultured in 12-well plates in a density of 104 cells/well, and subjected to MIT treatment (60, 90, and 120 µM) for 24 hr, thereafter, collected via trypsinization, and rinsed with a cell staining buffer. After centrifugation, the supernatant was discarded, and pellets were suspended in Annexin V binding buffer and then stained with Annexin V-FITC and PI, following the manufacturer's directions. Following a 20-min incubation in darkness at ambient temperature, the samples were examined utilizing an ACEA flow cytometer (Agilent, USA). The cell viability was measured as percentages of early-stage apoptosis, late-stage apoptosis, and necrosis using NovoExpress software (Agilent, USA) (Naderi et al., 2018).
Evaluation of apoptotic and angiogenic markersSH-SY5Y cells were cultured in 25 cm2 flasks at a density of 5 x 104 cells/flask, and subjected to MIT treatment (60, 90, and 120 µM) for 24 hr. The protein levels of p53, BAX, BCL-2, APAF1, SRC, VEGFR2, COX, and COX-2 were quantified using specific ELISA (Enzyme-linked immunosorbent assay) kits (BT LAB, China) according to the manufacturers’ protocols. The standard curves were generated for each protein using recombinant standards, and sample concentrations were calculated. The results were expressed as ng/mL normalized to mg of protein.
Determination of inflammatory responseThe release of inflammatory cytokines in SH-SY5Y cells following MIT exposure was measured with the LEGENDplex™ Human Inflammation Panel (BioLegend, USA) in accordance with the manufacturer’s guidelines. SH-SY5Y cells were cultured in 25 cm2 flasks at a density of 5 x 104 cells/flask, and subjected to MIT treatment (30, 60, and 90 µM) for 24 hours. Following 24 hr of exposure to MIT, the cell culture supernatants were harvested and examined using flow cytometry (BD Accuri C6, BD Biosciences, USA). The assay is based on a bead-based immunoassay, in which beads of defined sizes are distinguished in the FL3 channel, while cytokine-specific fluorescence intensities are detected in the FL2 channel and are proportional to cytokine concentrations (pg/mL). The concentrations of tumor necrosis factor alpha (TNF-α), interleukin-1 beta (IL-1β), interleukin-6 (IL-6), interleukin-8 (IL-8), interleukin-10 (IL-10), interleukin-12 (IL-12), interleukin-17 (IL-17), interleukin-18 (IL-18), interleukin-23 (IL-23), interleukin-33 (IL-33), monocyte chemoattractant protein-1 (MCP-1), interferon alpha (INF-α), and interferon gamma (IFN-γ) were determined using standard curves generated for each analyte, with the results being normalized to those of the untreated controls.
Statistical analysisData are expressed as the mean ± standard deviation (SD) of three different analyses. Statistical significance was assessed by one-way ANOVA, then by Tukey’s post-hoc test by GraphPad Prism version 9.0.0. A correlation study was conducted with Pearson’s correlation coefficient. The fold changes have been calculated in relation to control values. P-values below 0.05 have been considered statistically significant.
In silico cheminformatics analysis Network toxicology studyThe network toxicology approach is an emerging in silico cheminformatics strategy that offers substantial value in identifying potential molecular targets underlying the toxicological profiles of xenobiotics. It operates by pinpointing the shared nodes between two main datasets: disease-associated targets and chemical-associated targets. To identify the disease-associated potential targets, the process began with a comprehensive analysis of the in vitro outcomes to extract the keywords that most strongly summarize the toxicity profile of the MIT agent. Given that the MIT structure elicited a heterogeneous spectrum of toxicological responses, a set of keywords was selected to precisely encompass and reflect the comprehensive in vitro findings: “intrinsic neuronal apoptosis”, “mitochondrial oxidative stress”, “neuroinflammatory cytokine response” and “MAPK pathway activation”. The pinpointed keywords were queried in the GeneCards® platform (https://www.genecards.org, accessed January 2026), where a comprehensive search was undertaken. The corresponding Homo sapiens protein-coding targets associated with each keyword were retrieved, extracted, and compiled into Excel sheets separately (Yang et al., 2020). On the flip side, the SwissTargetPrediction platform (https://www.swisstargetprediction.ch/, accessed January 2026) was employed to identify the MIT-associated targets using the Simplified Molecular Input Line Entry System (SMILES) notation of MIT, retrieved from the PubChem database (https://pubchem.ncbi.nlm.nih.gov). The species parameter was set to Homo sapiens to ensure the acquisition of human-relevant predicted targets. Similarly to the disease-associated targets, the dataset was gathered and compiled into an Excel sheet (Daina et al., 2019; Kim and Bolton, 2024). Afterward, to identify the intersecting targets across the five compiled datasets and to generate an informative visual plot, the Jvenn tool (https://jvenn.toulouse.inrae.fr/app/example.html) was employed (Jia et al., 2021). At the final stage of the network toxicology assay, the selected targets—identified as the most frequently shared among all compiled datasets—were subjected to a preliminary validation step through protein–protein interaction investigation. The STRING platform (https://string-db.org/) was utilized for this aim. The search was restricted to Homo sapiens, and the resulting interaction network was generated as a bitmap image (Snel et al., 2000; Szklarczyk et al., 2025).
Molecular docking study and molecular mechanics–generalized born surface area (MM-GBSA) calculationsAs a subsequent step to the network toxicology study, a molecular docking analysis was performed. This assay aims to validate the potential targets identified as actionable nodes underlying the multifaceted toxicological behavior reported for the MIT agent, as well as to clarify its structural behavior at the molecular level once complexed with the respective proteins. Through this approach, the capability of MIT to effectively interact with and potentially modulate the selected targets can be more reliably assessed. This assay encompasses three main consecutive steps: ligand preparation, protein preparation and receptor grid generation, followed by molecular docking.
The ligand preparation step commenced by submitting the SMILES notation of the chemical structures of interest into the Maestro Schrödinger interface (version 14.2) using the graphical user module. Subsequently, the LigPrep interface was employed to prepare the ligands for docking by generating all possible 3D conformations and ionization states at pH 7.0 ± 2.0, adjusting bond angles, and performing structural optimization using the OPLS2005 force field, thereby closely mimicking normal physiological conditions (Çapan et al., 2024).
For the protein preparation stage, the process was initiated by retrieving the crystallographic structures representing the potential targets identified by the network toxicology study: Poly [ADP-ribose] polymerase 1 (PARP-1), Glycogen synthase kinase-3 beta (GSK-3β), and inducible nitric oxide synthase (iNOS). The Protein Data Bank (PDB) database (https://www.rcsb.org/, accessed January 2026) was utilized as a source to extract the crystallographic protein motifs as PDB codes (Burley et al., 2024). For the PARP-1 protein, the crystal structure with PDB ID 4GV7 (resolution: 2.89 Å) was selected, while for the GSK-3β and iNOS proteins, the crystal structures with PDB ID 1Q41 (resolution: 2.10 Å) and 4NOS (resolution: 2.25 Å), respectively, were employed. All the selected proteins were in a holo form state (i.e. complexed with their respective reference ligands) and represented the Homo sapiens species. The selection step was followed by subjecting the retrieved structures to the protein preparation wizard module—integrated into the Maestro Schrödinger interface (version 14.2)— to prepare the crude protein structures and render them applicable for molecular docking studies (Sastry et al., 2013). This step seeks to generate optimized versions of the crystallographic proteins that more closely simulate physiological conditions and mimic natural protein behavior. During this stage, the default parameters were applied, including protonation state adjustment at pH 7.0 ± 2.0, addition of missing hydrogen atoms, side chains, and loops, adjustment of bond angles and conformations, removal of water molecules located within 3 Å of heteroatoms, and finally, energy minimization of the protein structures using the OPLS2005 force field (Shivakumar et al., 2012). After that, in pursuit of obtaining reasonable outcomes as well as enhancing the reliability of the predicted binding poses, the receptor grid generation module was employed to create a 3D receptor grid box for every protein, thus the docked ligands were directed precisely into the protein’s binding site. Its size and dimensions were automatically identified based on the coordinates of the crystallographic native ligand; the other parameters were kept as default (Bilgehan et al., 2025). Upon the completion of both ligand and protein preparation steps, the ligand docking module (XP-glide mode) was applied (Friesner et al., 2006).
After that, the resulting docking poses of each ligand–protein complex were collected and analyzed. The complexes with the highest docking scores (i.e., more negative values) were top-ranked and subsequently subjected to the Prime MM-GBSA module, integrated into the Maestro Schrödinger interface (version 14.2). This process aims to estimate the binding free energy (ΔG) of each ligand and provides a more precise and accurate prediction of binding affinity (Huang et al., 2020). Compared to the molecular docking scores, the ΔG values offer a more reliable and refined validation by integrating post-docking energy minimization and accounting for electrostatic interactions, solvation effects, and van der Waals forces. The ΔG binding free energy for each ligand–receptor complex was calculated using the following thermodynamic equation (Equation 1):
Gbind = Gcomplex (Gprotein + Gligand) (Equation 1)
ΔGbind: the binding energy between the ligand and its corresponding receptor. Gcomplex: the minimized energies of the protein-ligand complex. Gprotein: the free protein. Gligand: the free ligand.
As the last stage in the chemoinformatics studies, the finally selected ligand–protein complexes were further investigated using the Protein–Ligand Interaction Profiler (PLIP) server. This valuable tool provides a more comprehensive, detailed, and precise characterization of their interaction profiles (Xia et al., 2024).
The cytotoxic potential of MIT was determined using the MTT assay with 24-hr exposure of the MIT to SH-SY5Y cells, and the IC50 (half-maximal inhibitory concentration) value was found to be 115 µM (Fig. 1), which indicates doses that are sub-lethal to close to the IC50 level (Henslee et al., 2016). For the analysis of the apoptotic pathway, treatment concentrations of 60, 90, and 120 µM were chosen based on the calculated IC50 value (115 µM). For LEGENDplex™ cytokine analysis, lower concentrations of 30, 60, and 90 µM were used, with the highest dose maintained at IC30 to minimize cytotoxicity.
MTT results of SH-SY5Y cells with MIT.
It has been examined the apoptotic effects of MIT on SH-SY5Y neuroblastoma cells using flow cytometric analysis with the Annexin V/PI method. MIT treatment (60, 90, and 120 μM), showed dose-dependent apoptosis induction. The results of the statistical analysis demonstrated that MIT had significant effects, especially for causing early apoptosis (Annexin V+/PI-) (p<0.05). In comparison to the control group, the application of 120 μM MIT increased the early apoptotic cell rate from 2.01 ± 0.48% to 46.23 ± 3.67%, resulting in an approximately 23-fold increase (p<0.05) (Fig. 2). In parallel, the live cell ratio significantly decreased from 96.07 ± 0.20% to 42.60 ± 3.81% (p<0.05). The late apoptotic cell rate also showed a 7-fold increase compared to the control group (1.42 ± 0.31% vs 10.28 ± 0.31%, p<0.05). No statistically significant change was noted in the fraction of necrotic cells (p:0.14), indicating that MIT induced a clean apoptotic death mechanism. The results suggest that MIT has a potent and selective apoptosis-inducing effect on SH-SY5Y cells, and the neurotoxic potential of this biocide compound occurs through apoptotic pathways. Flow cytometry analysis revealed that MIT induced dose-dependent apoptosis in SH-SY5Y neuroblastoma cells. Total apoptosis (early + late apoptotic cells) increased progressively from 3.4% in control cells to 10.0% at 60 μM MIT, 27.4% at 90 μM MIT, and 56.5% at 120 μM MIT, indicating severe neurotoxicity at the highest concentration tested.
Changes in apoptosis observed with MIT exposure on SH-5Y cells 24 hr and the representative flow cytometry analyses (A), the cell amount (%) of necrosis and apoptosis (B). The cells were exposed with 60-120 μM concentrations. Data are given as mean ± SD, *p<0.05 versus the control group (C).
To clarify the molecular foundation of MIT-induced cytotoxicity, ELISA analyses were performed to evaluate the changes in critical apoptotic pathway proteins in SH-SY5Y neuroblastoma cells (Fig. 3). The results demonstrated that MIT activates the intrinsic apoptotic pathway in a dose-dependent manner. At 120 µM, p53 expression increased 9.54-fold (p<0.05), followed by a 4.52-fold increase in pro-apoptotic BAX protein (p<0.05). APAF-1 expression was elevated 8.85-fold at the same concentration (p<0.05), while BCL-2 levels increased to 4.37-fold (p<0.05). The resulting BAX/BCL-2 ratio favored apoptotic progression. COX and COX-2 levels were upregulated 8.04-fold and 6.93-fold, respectively (p<0.05). SRC kinase showed a 7.08-fold increase, and VEGFR2 exhibited the most dramatic elevation at 20.96-fold (p<0.05). These quantitative changes indicate that 120 µM MIT triggers coordinated activation of p53-BAX-APAF-1-mediated intrinsic apoptosis, accompanied by oxidative stress and vascular signaling responses.
The fold changes in the protein levels observed with MIT exposure on SH-5Y cells 24 hr. The cells were exposed with 60-120 μM concentrations. Data are given as mean ± SD, *p<0.05 versus the control group (C). APAF-1: apoptotic protease activating factor-1; BAX: BCL-2-associated X protein; p53: tumor protein p53; SRC: proto-oncogene tyrosine-protein kinase Src; BCL-2: B-cell lymphoma 2; VEGFR2: vascular endothelial growth factor receptor-2; COX: cyclooxygenase; COX2: cyclooxygenase-2.
In summary, the quantitative ELISA data reveal a highly coordinated molecular response at 120 µM, where DNA damage sensing (p53, 9.54-fold), mitochondrial commitment (BAX, 4.52-fold; APAF-1, 8.85-fold), oxidative amplification (COX/COX-2, ~8-fold), signal transduction (SRC, 7.08-fold) and vascular adaptation (VEGFR-2, 20.96-fold) converge to execute irreversible apoptosis (Fig. 4).
Coordinated molecular response cascade induced by MIT at 120 µM concentration, showing fold-change increases in key apoptotic, oxidative stress, and vascular signaling proteins. APAF-1: apoptotic protease activating factor-1; BAX: BCL-2-associated X protein; p53: tumor protein p53; SRC: proto-oncogene tyrosine-protein kinase Src: BCL-2: B-cell lymphoma 2; VEGFR2: vascular endothelial growth factor receptor-2; COX: cyclooxygenase; COX2: cyclooxygenase-2.
To comprehensively evaluate the inflammatory response, we analyzed 13 cytokines using the LEGENDplex Human Inflammation Panel 1, including TNF-α, IL-1β, IL-6, IL-8, IL-10, IL-12, IL-17, IL-18, IL-23, IL-33, MCP-1, IFN-α, and IFN-γ. Our findings indicated a dose-dependent elevation in four principal pro-inflammatory cytokines: IL-1β, IL-6, IL-18, and IL-23 (Fig. 5). IL-1β exhibited a substantial elevation at the highest dose (90 μM, p<0.05), while IL-6 demonstrated a robust dose-dependent response with a significant elevation starting at 90 μM (approximately a 3-fold increase, p<0.05). Both IL-18 and IL-23 exhibited significant upregulation at 90 μM compared to control (approximately 6-fold and 4-fold increases, respectively, p<0.05). The remaining cytokines in the panel (TNF-α, IL-8, IL-10, IL-12, IL-17, IL-33, MCP-1, IFN-α, and IFN-γ) did not show statistically significant changes across the tested concentrations, suggesting a selective activation of specific inflammatory pathways rather than a global inflammatory response (Fig. 6).
The fold changes in the inflammatory response observed with MIT exposure on SH-5Y cells 24 hr. The cells were exposed with 60-120 μM concentrations. Data are given as mean ± SD, *p<0.05 versus the control group (C). IL-1β: interleukin-1 beta; IL-6: interleukin-6; IL-18: interleukin-18; IL-23: interleukin-23.
The levels of TNF-α, IL-8, IL-10, IL-12, IL-17, IL-33, MCP-1, IFN-α, and IFN-γ remained relatively unchanged across all treatment concentrations, with no statistically significant differences detected compared to control groups (C). TNF-α: tumor necrosis factor alpha; IL-8: interleukin-8; IL-10: interleukin-10; IL-12: interleukin-12; IL-17: interleukin-17; IL-33: interleukin-33; MCP-1: monocyte chemoattractant protein-1; IFN-α: interferon alpha; IFN-γ: interferon gamma.
With the goal of elucidating the biological targets underlying the toxicological profile reported for MIT, as well as clarifying the involved physiological pathways, a set of chemoinformatics approaches were performed. The analysis aimed to upgrade our understanding of MIT’s mechanisms of action and to identify its effects on interconnected molecular targets and signaling pathways. Accordingly, a network toxicology analysis was first conducted to identify the common targets between two major datasets: disease-associated targets and MIT-associated targets, using advanced and curated databases. Subsequently, molecular docking studies were performed to assess the binding potential of MIT toward the identified most shared targets and to investigate its interaction profiles in terms of binding poses and physical forces exerted.
Network toxicology studyBased on an in-depth analysis of the results observed for MIT following the conducted distinct in vitro assays, including cytotoxicity using neuroblastoma cell lines, apoptosis induction, inflammation response, and altering the expression of a set of critical biomarkers, the disease-associated Homo sapiens targets were recapped in the following four keywords enable a comprehensive search within the GeneCards® database and to accurately reflect the MIT’s toxicological profile: “intrinsic neuronal apoptosis”, “mitochondrial oxidative stress”, “neuroinflammatory cytokine response” and “MAPK pathway activation”. For the precisely defined keywords, 9984, 290, 1147, and 464 targets were collected, respectively (Fig. 7). On the other hand, the chemical-associated Homo sapiens targets were identified against the MIT chemical structure using its SMILES format via screening through the SwissTargetPrediction database, resulting in the collection of 71 targets. All the collected targets were submitted separately into Excel sheets, where the data were checked and duplications were removed. The total of five data sets was then organized and illustrated as a Venn diagram, showing the overlapping targets among the examined data sets. As depicted, three targets were identified as the most common targets across all five data sets. These targets hold the UniProt codes P09874, referring to Poly [ADP-ribose] polymerase 1 (PARP-1); P49841, referring to Glycogen synthase kinase-3 beta (GSK-3β); and P35228, corresponding to inducible iNOS. Aiming to confirm the existence of a good correlation between these finally selected targets, the protein–protein interactions were explored using the STRING server. As shown in the bitmap figure, there are strong protein–protein interactions between the three selected targets, especially PARP-1 and GSK-3β, which have been experimentally confirmed. In light of the depicted network toxicology results (Fig. 7), PARP-1, GSK-3β, and iNOS were selected as potential targets for MIT. Subsequently, the potential of MIT to bind and modulate these selected targets is going to be evaluated through a molecular docking study.
Network toxicology framework linking disease-associated pathways with MIT-related molecular targets.
The fundamental objective of a molecular docking study is to investigate not only the feasibility but also the strength of a chemical structure’s ability to bind and interact with a biological target. Following the network toxicology study, the implementation of a molecular docking approach is essential to evaluate the finally selected targets, which are nominated as potential proteins believed to underlie the main toxicological profile observed in vitro. Herein, this study aims to explore the interaction profile of the MIT structure within the binding site of each target separately and compare it with positive control agents. The extent to which the MIT structure reproduces the interaction patterns and binding characteristics of the reference agents indicates its potential to also mimic the physiological actions by modulating that target and its related pathways.
The study commenced by evaluating the prepared crystallographic proteins and validating the docking protocol. This was achieved by isolating and preparing the native crystallographic ligands (reference agents), re-docking each ligand into its corresponding target, and subsequently superposing the docked conformation onto the original crystallographic pose. The root mean square deviation (RMSD) values were then calculated, where an RMSD below 2 Å indicates a reliable and confident docking procedure. As shown in Fig. 8, the calculated RMSD values for the selected potential targets—PARP-1, GSK-3β, and iNOS—were 0.0291, 0.0516, and 0.0862 Å, respectively, confirming the high precision of the docking protocol and the suitability of the prepared crystallographic structures for subsequent docking analyses.
Superposition of crystallographic native ligands for PARP-1 (A), GSK-3ß (B), and iNOS (C) targets with their respective docked poses, demonstrating RMSD values of 0.0291, 0.0516, and 0.0862 Å, respectively.
For the PARP-1 target, docking of the MIT structure within the protein’s binding domain demonstrated an interaction profile totally matching that of the PARP-Ref structure, that acts as a reference ligand (Fig. 9). Both compounds formed hydrogen bonds with SER904 and GLY863 residues, in addition to establishing π-cation and hydrophobic interactions with TYR907. Superposition of MIT over the PARP-Ref ligand showed the high ability of the isothiazolinone ring of MIT to mimic the pyrimidone core of the reference structure. The larger molecular size of the reference agent provides an advantage in occupying the binding pocket more extensively, allowing a more optimal fit, which is reflected in its superior, but still comparable, docking score and binding energy. Docking of the MIT agent within the binding site of GSK-3β revealed the formation of a hydrogen bond with VAL135, along with hydrophobic interactions involving ALA83 and LEU188 residues. The positive control (GSK-Ref) exhibited similar interaction forces to those observed for MIT; however, it additionally formed hydrogen bonds with ASP133 and ILE62, as well as further hydrophobic interactions with VAL70, LEU132, and THR138. Superposition of MIT over the GSK-Ref structure demonstrated the contribution of the carbonyl moiety in partially simulating the structural behavior of the reference ligand, although to a lesser extent compared to the interaction profile observed for the PARP-1 receptor. Docking simulation of MIT within the binding site of the iNOS target demonstrated its ability to establish interactions with both chain A and chain B, similar to the NOS-Ref agent, with shared critical contacts at ARG381 and TRP463. The presence of multiple polar centers (nitrogen and oxygen atoms) enables the reference ligand to engage in a broader range of interactions. Although MIT possesses fewer polar centers, the carbonyl group and the high electron density of the isothiazolinone ring act as effective interaction sites, allowing for meaningful and comparable docking behavior relative to the reference compound. As summarized in Table 1, the notable degree of similarity in interaction patterns, docking scores, and ΔG binding energies between MIT and the respective reference ligands supports the hypothesis that MIT could potentially modulate these targets.
Molecular docking and superposition of MIT structure and the native reference ligands within the binding sites of PARP-1 (PDB ID: 4GV7), GSK-3 (PDB ID: 1UV5), and iNOS (PDB ID: 4NOS), illustrated in 2D and 3D representations.
;%0A%09%09%09newWindow.document.open();%0A%09%09%09newWindow.document.write('<img src=%22./Graphics/51_471-t1.jpg%22>');%0A%09%09%09newWindow.document.close();%0A%09%09)
MIT is one of the commonly used biocides in cosmetics and industrial applications because of its used advantages (Van Huizen et al., 2017). Although neurotoxic effects due to MIT exposure have been reported, studies on this subject are insufficient (Spawn and Aizenman, 2012; Li et al., 2024). For this reason, the present study was conducted to assess the potential harmful effects of MIT on SH-SY5Y cells. The cells are of human origin and are widely used in the evaluation of neurotoxic effects because they provide a strong association with the human nervous system (Kovalevich et al., 2021).
Apoptosis is an extremely complex and energy-dependent process that has an important role in programmed cell death and cell homeostasis (Jin and El-Deiry, 2005). It could be induced by two ways: intrinsic or mitochondrial pathway, and extrinsic or the death receptor pathways (Wen et al., 2019). MIT showed apoptotic and necrotic effects in keratinocytes (NHK) cells, and apoptosis and inflammatory reactions in human bronchial epithelial cells and liver epithelial cells (Ettorre et al., 2003; Park et al., 2018; Park and Seong, 2020). Exposure doses below the IC50 may trigger early apoptotic changes. However, apoptosis can be completely observed at around IC50 dose (Henslee et al., 2016). The 120 µM dose was also added to the exposure concentrations to elucidate the apoptosis pathway. The present study indicated that MIT exposure triggered dose-related cytotoxicity and apoptosis in SH-SY5Y human neuroblastoma cells, with significant changes in apoptotic (p53, BAX, APAF1), survival and angiogenic (BCL-2, SRC, VEGFR2), and inflammatory (COX-2, COX) pathways.
The activation of p53 in response to chemical stress is a key event that controls important cellular processes like apoptosis, cell cycle control, and DNA repair (Chen, 2016). As revealed by our study, exposure to MIT alone triggers a p53-centered apoptotic program and a complex cellular stress response in SH-SY5Y neuronal cells. This finding demonstrates the distinct neurotoxic potential of MIT. Upon examining the molecular basis of the apoptotic process, we observed that the p53 tumor suppressor acts as a primary sensor of MIT-induced cellular damage. The role of p53 in mediating chemically induced oxidative stress has been comprehensively discussed in previous studies (Liu et al., 2020). Our findings revealed that p53 upregulation promoted BAX gene expression, driving the Bax/Bcl-2 ratio in favor of apoptosis despite BCL-2 expression, thereby confirming engagement of the intrinsic (mitochondrial) apoptotic pathway. A significant increase of APAF-1 levels further supports this mechanism, confirming the formation of a robust apoptosome complex following cytochrome c release and the irreversible commitment of the cell to apoptosis.
Oxidative stress is important in the initiation of apoptosis. In our study, the increased levels of COX and COX-2 suggest that MIT exposure causes a pronounced redox imbalance within the cells. These findings corroborate previous reports indicating that MIT triggers apoptosis in human bronchial epithelial cells by compromising intracellular organelle function via oxidative stress (Park and Seong, 2020). The observation of a similar mechanism in a different cell line indicates that oxidative stress may represent a general and cell type–independent feature of MIT toxicity.
A particularly notable finding of this study is the substantial activation detected not only in apoptotic pathways but also in the SRC kinase and angiogenesis-associated VEGFR2 signaling axis. The simultaneous activation of these pathways—typically associated with cell survival and proliferation—alongside an intense apoptotic process indicates that cells mount a complex and multifaceted response to the overwhelming stress induced by MIT. SRC kinase is central to the regulation of angiogenesis and cell survival, serving as a major mediator of VEGF signaling. It has been demonstrated that inhibition of SRC kinase suppresses angiogenesis and tumor growth (Schlessinger, 2000; Ischenko et al., 2007; Manukjan et al., 2025). In our study, it was revealed a pronounced activation of both SRC and VEGFR2, suggesting that the cells initiate a strong “rescue signal” in response to severe stress and mitochondrial dysfunction to enhance their chances of survival. Indeed, given the predominance of pro-apoptotic signaling mediated by p53 and BAX, it is evident that this VEGFR2 and SRC-dependent survival attempt was insufficient to counteract cell death in late apoptosis (p<0.05). This finding suggests that MIT neurotoxicity is not merely a passive process leading to cell demise but rather an active event that simultaneously disrupts the cell’s intrinsic survival mechanisms. Also, these results provide a comprehensive perspective on the molecular mechanisms underlying MIT-induced neurotoxicity and underscores the importance of considering both apoptotic and adaptive stress pathways in evaluating the safety profile of this biocide.
Our findings reveal a coordinated activation of both apoptotic and inflammatory pathways in response to MIT exposure, suggesting a complex neurotoxic mechanism that extends beyond simple cell death. The p53-mediated apoptotic cascade, characterized by BAX upregulation, BCL-2 suppression, and APAF-1 activation, occurred in parallel with a selective pro-inflammatory response marked by significant dose-related rises in IL-1β, IL-6, IL-18, and IL-23. This concurrent activation is particularly noteworthy, as these cytokines are key mediators of neuroinflammation and have been implicated in neurodegenerative processes (Guadagno et al., 2015; Huang et al., 2025). The IL-1 family members IL-1β and IL-18 processed by inflammasomes, suggest potential activation of the NLRP3 inflammasome pathway, which is known to be triggered by mitochondrial dysfunction and oxidative stress—hallmarks of the intrinsic apoptotic pathway we observed. Similarly, the elevation of IL-6 and IL-23 indicates activation of broader inflammatory signaling that may amplify neuronal damage through paracrine effects (Nitsch et al., 2021; Kerkis et al., 2024). In the present study, although significant increases in the inflammatory cytokines like IL-1β, IL-6, IL-18 levels were detected at the 90 µM treatment dose (≥ 3-fold), COX-2 activation was only significant at 120 µM (~8-fold). This result suggests that while inflammatory cytokines exhibit rapid and early release, COX-2 induction requires a higher or more prolonged stimulation threshold because it requires additional steps such as gene expression, mRNA stabilization, and protein synthesis (Neeb et al., 2011). Furthermore, the higher expression of COX compared to COX2 suggests that exposure activates physiological/prostaglandin-mediated protective mechanisms in addition to the inflammatory response (Smith et al., 2000; Aïd and Bosetti, 2011). Similarly, the increase in VEGFR2 expression appears to be significant at the 120 µM treatment dose. IL-6 may have induced VEGFR2 expression, and this may suggest that cells are directed toward angiogenesis or a regenerative pathway as part of the inflammatory response (Hegde et al., 2020). The selective nature of this inflammatory response (with no significant changes in TNF-α, IL-8, IL-10, IL-12, IL-17, IL-33, MCP-1, IFN-α, and IFN-γ) suggests that MIT triggers a specific inflammatory signature rather than generalized immune activation, potentially reflecting a mitochondria-centered danger signal that links apoptotic stress to innate immune responses in neuronal cells.
These results correlate with previous research reporting that isothiazolinones elicit neurotoxic, inflammatory, and oxidative responses in various cellular models (Du et al., 2002; He et al., 2006; Li et al., 2024). For instance, chloromethylisothiazolinone (CMIT)/MIT mixtures have been shown to elicit oxidative stress, mitochondrial dysfunction, and activation of MAPK signaling cascades in SH-SY5Y cells, accompanied by upregulation of proinflammatory cytokines and apoptosis-related genes such as p53 and BAX (Molinari et al., 2025). However, the specific contribution of MIT alone has remained unclear, because those studies investigated combined exposures. The present results clarify that MIT itself is sufficient to induce apoptotic and inflammatory responses, independent of CMIT, highlighting its intrinsic neurotoxic potential.
Understanding the biological impact of MIT is essential to contextualize the observed in vitro findings. To this end, a combined network toxicology and molecular docking approach was applied to identify potential molecular targets underlying its toxic effects. The experimental results indicate that MIT induces a multifaceted toxic response, including neuronal cytotoxicity, intrinsic apoptosis, oxidative stress, inflammatory responses, and disruption of key signaling pathways, suggesting a multi-pathway mechanism rather than a single-target effect.
Network toxicology analysis identified PARP-1, GSK-3β, and iNOS as shared key targets associated with MIT-induced toxicity. PARP-1 is closely linked to DNA damage responses, NAD+/ATP depletion, mitochondrial dysfunction, and intrinsic apoptosis, consistent with apoptotic markers observed in MIT-treated cells (Wang et al., 2019). GSK-3β plays a central role in oxidative stress, mitochondrial dysfunction, tau phosphorylation, and neuronal survival, supporting its involvement in MAPK-related toxicity pathways (Gupta et al., 2024; Sayas and Ávila, 2021; Yang et al., 2017). Meanwhile, iNOS contributes to nitrosative and oxidative stress and is associated with inflammatory signaling, which may amplify neuroinflammatory and apoptotic responses (Wang et al., 2021; Üremiş and Üremiş 2025). Collectively, these findings support a multi-target mechanism underlying MIT toxicity and highlight the usefulness of network-based approaches in complementing experimental toxicological data.
The results obtained from the network toxicology analysis were subsequently validated by the molecular docking study. The observed comparability in docking scores and ΔG binding energies, alongside the highly conserved interaction profiles relative to native crystallographic reference ligands, reinforces the hypothesis that MIT is capable of structurally modulating the identified targets. The unique structural features of MIT’s isothiazolinone ring, including its high electron density and the existence of critical and distinct heteroatoms (O, N, and S), appear to underlie its functional leverage to engage and modulate interactions within the binding pockets of the targets of interest.
In unison, the strong coherence between the in vitro phenotypes and the in silico cheminformatics outcomes—encompassing both the network-derived target correlations and the docking-supported interaction feasibility—provides substantial evidence that the multifaceted toxicological profile reported for MIT is likely to stem from the synchronous modulation of a set of central targets: PARP-1, GSK-3β, and iNOS. Conceptualizing the MIT agent as a polypharmacological toxicant (a term used here in a toxicological context to describe a xenobiotic capable of interacting with multiple molecular targets, rather than implying any therapeutic or pharmacological activity), rather than a single-pathway disruptor, provides a more coherent explanation for the concurrent apoptotic cues, oxidative imbalance, inflammatory signaling, and metabolic stress that were observed. This perspective reinforces a multi-target interference model as the mechanistic framework underlying MIT’s toxicological fingerprint. This holistic view not only broadens our understanding of the toxicodynamics of MIT but also highlights the value of network-based computational approaches in forecasting the functionally relevant toxicological nodes for environmental and industrial chemicals.
This study examined the neurotoxic effects of MIT on human SH-SY5Y neuronal cells. MIT exposure resulted in dose-dependent cytotoxicity, as demonstrated by the MTT assay, with an IC50 value of 115 µM. Apoptosis analysis revealed significant induction of programmed cell death at concentrations of 60, 90, and 120 µM. Moreover, ELISA analyses revealed altered expressions of key apoptotic and survival-related genes, including p53, BAX, BCL-2, APAF1, SRC, VEGFR2, COX-2, and COX, indicating that MIT disrupts multiple molecular pathways in human neuronal cells. These results underscore the possible neurotoxic risk of MIT, even at levels pertinent to environmental and domestic exposure. Importantly, we reveal that MIT simultaneously triggers a selective pro-inflammatory response characterized by substantial increases in IL-1β, IL-18, IL-6, and IL-23, suggesting inflammasome activation and neuroinflammatory signaling that may amplify neuronal damage beyond direct cytotoxic effects. The concurrent engagement of both cell death and inflammatory pathways highlights MIT's multifaceted neurotoxic mechanism and raises concerns about its widespread use in consumer products, particularly given potential chronic low-level exposures. Aiming to reinforce the experimental findings, a set of in silico chemoinformatics approaches was performed, starting with a network toxicology analysis and subsequently complemented by a molecular docking study. Network toxicology identified PARP-1, GSK-3β, and iNOS as potential targets, revealing convergent molecular nodes that plausibly govern MIT-induced toxicity. In parallel, molecular docking confirmed the structural compatibility and modulation potential of MIT toward these network-derived hubs. Collectively, these computational outcomes reinforce MIT’s multifaceted toxicological profile and provide a mechanistic scaffold to interpret its diverse toxic effects observed in vitro.
These findings drive home the importance of stricter regulatory oversight of MIT-containing products and warrant further investigation into its long-term neurotoxic potential in vivo, particularly regarding cumulative effects on the developing and aging nervous system. Our results reinforce the growing evidence that necessitates reassessment of isothiazolinone biocides in formulations where human exposure is possible. Overall, this study provides theoretical explanations for MIT-induced neuronal toxicity and highlights the importance of evaluating both cytotoxic and molecular responses to inform future in vivo and regulatory safety assessments.
FundingThis research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Conflict of interestThe authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data availabilityData will be made available on request.
Author contributionsConceptualization: Mohammed T. Qaoud, Merve Arici
Investigation: Gül Küçükkahraman
Methodology: Gül Küçükkahraman, Özge Sultan Zengin, Mohammed T. Qaoud
Formal Analysis: Gül Küçükkahraman, Özge Sultan Zengin, Mohammed T. Qaoud
Writing – original draft: Gül Küçükkahraman, Özge Sultan Zengin, Mohammed T. Qaoud
Writing – review and editing: Merve Arici, Gül Özhan
Supervision: Gül Özhan
Resources: Gül Özhan
Ethical approval and consent to participateNot applicable.
Patient consent for publicationNot applicable.
