Brain organoids in environmental neurotoxicology: applications, mechanisms, and future perspectives

Abstract

The advent of human induced pluripotent stem cell (hiPSC)-derived brain organoids represents a significant advance in environmental neurotoxicology, propelling the discipline toward human-relevant, mechanistic, and predictive in vitro paradigms. This review explores the utility of brain organoids in environmental neurotoxicology, which uniquely address critical limitations of traditional models by recapitulating key aspects of human brain development, including three-dimensional (3D) cytoarchitecture, multilineage cellular heterogeneity, and functional network activity. This review systematically elaborates on their construction principles, unique advantages in neurotoxicological research, and the significant progress made in elucidating mechanisms of toxicity. Notably, brain organoids exhibit enhanced sensitivity in identifying the subtle adverse outcomes of chronic, low-dose exposures to environmental contaminants, often eluding conventional approaches. Their key advantage lies in the greater capacity to deconstruct complex toxicological pathways, enabling precise tracing of adverse outcome pathways (AOPs). Future development requires enhancing model complexity through vascularization, promoting automation and standardization, and integrating artificial intelligence (AI) for data analysis. Concurrently, establishing sustained ethical oversight and standardized frameworks is essential to ultimately advance the field toward more precise and efficient hazard identification and risk characterization.

Graphical abstract

Highlights

1. Human brain organoids bridge species gaps for more human-relevant neurotoxicity assessment.

2. Brain organoids effectively recapitulate chronic, low-dose and mixture environmental risks.

3. Integrating brain organoid data into the AOP frameworks enhances mechanistic understanding.

4. Coupling brain organoids with organ-on-a-chip and AI advances next-generation risk assessment.

Introduction

The neurotoxic impact of environmental pollutants constitutes a major global public health burden. These toxicants exert adverse effects throughout the entire human lifespan, with particularly pronounced threats to the nervous system during developmental and aging stages. Key neurotoxic agents—including air pollutants, pesticides, endocrine-disrupting chemicals, and heavy metals—are closely linked to the pathogenesis of both neurodegenerative and neurodevelopmental disorders (Bakulski et al. 2020; Chin-Chan et al. 2015; Iqubal et al. 2020). Neurotoxicity refers to adverse structural or functional alterations in the central nervous system (CNS) induced by biological, chemical, or physical agents (Aaseth et al. 2020). However, the mechanisms underlying the majority of the estimated 70,000 potential neurotoxic chemicals remain poorly understood (Vargas and Ponce-Canchihuamán, 2017).

Key characteristics of environmental neurotoxicants

(1) Blood–Brain Barrier (BBB) Penetrability, exemplified by per- and polyfluoroalkyl substances (PFAS) (Cao and Ng 2021) and benzene (Carecho et al. 2024), allowing accumulation in brain tissue. (2) Low-dose cumulative toxicity, often lacking a clear safety threshold, with long-term, low-dose exposure amplifying adverse effects. (3) Life-stage susceptibility, where developmental exposure may cause irreversible damage and later-life exposure accelerates neurodegeneration. (4) Subclinical latency, with early molecular or microstructural alterations preceding overt behavioral deficits, delaying intervention until irreversible damage has occurred.

Limitations of traditional models and the emergence of brain organoids

Traditional in vivo and in vitro models still exhibit significant limitations in elucidating the mechanisms and supporting risk assessment of environmental neurotoxicants. Rodent models suffer from cross-species extrapolation uncertainties due to differences in neuroanatomy and metabolism. For example, methylmercury (MeHg) failed to induce forebrain cortical abnormalities in rodents, but disrupted neural progenitor differentiation in human brain organoids (Aaseth et al. 2020). Similarly, zebrafish models, while rapid and cost-effective, lack human-specific higher brain regions and complex endpoints (Hughes and Hessel 2024). Furthermore, traditional animal experiments also involve high costs, extended durations and ethical controversies (Yin and Horzmann 2024). Traditional in vitro models, such as immortalized cell lines and primary cultures, fail to mimic neural tissue heterogeneity, 3D architecture, BBB, or neuron-glia interactions, and are ineffective for chronic low-dose exposures (Pampaloni and Stelzer 2009).

In contrast, human induced pluripotent stem cells (hiPSCs)-derived brain organoids offer a more physiologically relevant alternative. In 2013, Lancaster et al. successfully induced hiPSCs to form brain organoids with regionalized structures by optimizing the 3D culture systems. These models overcome the limitations of animal models and two-dimensional (2D) cell cultures by recapitulating key features of human brain development, including neural progenitor cell proliferation, neuronal differentiation, and cortical lamination (Lancaster et al. 2013). Brain organoids simulate the extracellular matrix (ECM) and biological barriers, alleviating species differences and in vitro-in vivo translation bottlenecks. They have emerged as a promising neurotoxicant screening platform over the past decade. As shown in Table 1, they offer superior species relevance and cellular complexity compared to traditional models, although challenges remain in standardization, cost, among others. Future integration with AI technology holds the promise of establishing a novel environmental toxicology assessment system.

Table 1.

Comparison of key parameters among different neurotoxicology models

Key Parameters	2D cultures	Rodent cells	Animal models	Brain organoids	Organ-on-a-chips
Model Type	Static	Static	Static	Static	Perfused
Cost	★	★★★	★★★★★	★★★★	★★★★★
Cellular Complexity	★	★★★	★★★★★	★★★★★	★★★★★
Species Relevance	★★	★★★	★★★	★★★★★	★★★★★
Throughput	★★★★★	★★★	★	★★★	★★

★: Low, ★★: Medium–Low, ★★★: Medium, ★★★★: Medium–High, ★★★★★: High

This review aims to explore the applications of brain organoids in environmental toxicology. We will sequentially elaborate on their construction principles and model characteristics, their unique advantages in neurotoxicological research, and the progress made in elucidating neurotoxicity mechanisms. Subsequently, we will systematically summarize the current landscape, critically analyze existing limitations, and discuss the future potential of brain organoids.

Brain organoids: generation and applications in environmental toxicology

Fundamentals of brain organoid technology

Brain organoids are 3D in vitro culture systems, primarily derived from hPSCs, including hiPSCs and embryonic stem cells (ESCs), and can recapitulate key features of human brain development and structure (Chhibber et al. 2020; Xie et al. 2020). The protocol for their generation typically follows a stepwise strategy that mimics in vivo neural development. As illustrated in Fig. 1, the process encompasses four core stages (labeled A–D in the figure): first, hPSCs are cultured and prepared; next, embryoid bodies (EBs) are formed and subjected to neural induction, and these EBs can spontaneously differentiate into the three germ layers; subsequently, EBs are embedded in Matrigel (a basement membrane matrix) to provide a 3D scaffold for cell polarization and self-organization, while neural progenitor cells are expanded and maintained in 3D culture; finally, neuronal differentiation and long-term maturation are promoted via growth factor regulation, and the culture system is transferred to appropriate platforms (e.g., spinning bioreactors) with neurogenic media to support nutrient exchange and tissue maturation (Lou and Leung 2018; Rajan et al. 2020; Zheng et al. 2021).

Fig. 1.

Generation of human brain organoids. The figure provides a brief introduction to the experimental steps involved in culturing and passaging hiPSC to generate human brain organoids. (A) hiPSC initial culture and preparation; (B) Embryoid body (EB) formation and neural induction; (C) Neural progenitor cell proliferation; (D) Neuronal differentiation and long-term maturation

A core characteristic of brain organoids is the self-organization capability, which enables them to recapitulate early in vivo brain developmental processes, including neural tube formation, neuroepithelial differentiation, and regional patterning. The self-organization process relies on the intrinsic developmental potential of embryonic progenitors and the regulatory role of ECM components (e.g., Matrigel) in guiding cell migration and lineage specification (Lancaster and Knoblich 2014; Lou and Leung 2018). This inherent ability to mimic in vivo developmental trajectories makes brain organoids a robust foundational platform for toxicology research.

Generating region-specific neural organoids

Region-specific differentiation and assembloids

As an important tool for neurotoxicology research, brain organoids can recapitulate the similar cell composition, unique structures, and developmental trajectories of the human brain. They feature high culture stability, support high-throughput screening (HTS), and are convenient as well as reproducible, effectively overcoming the limitations of traditional 2D cell models and animal models (Lisa et al. 2024; Renner et al. 2020). Specifically, “high-throughput” in the context of brain organoids refers to the transition from manual, low-scale culture to automated, standardized platforms. These systems facilitate the simultaneous assessment of neurotoxicity across multiple samples using 96- or 384-well formats, enabling the collection of multiple endpoints, such as structural integrity via high-content imaging and functional activity via microelectrode arrays (MEAs) in a single experimental run.

Brain organoids can be directed toward specific brain regions, by modulating key pathways such as SMAD (Shafique 2022; Zheng et al. 2021), WNT (Cao et al. 2023; Shafique 2022), SHH (Hua et al. 2022a, b; Yan et al. 2016) and FGF (Hu et al. 2025; Hua et al. 2022a, b) with small molecules. Currently, successfully constructed region-specific organoids cover key functional regions in the brain, including the forebrain (Hua et al. 2022a, b), cerebellum (Hua et al. 2022a, b), midbrain (Mendes-Pinheiro et al. 2023), cortex (Takata et al. 2023), and hippocampus (Sakaguchi et al. 2015). Each of these models can highly express specific markers corresponding to the target brain region (e.g., TH for the midbrain and FOXG1 for the cortex) (Mendes-Pinheiro et al. 2023; Takata et al. 2023).

A notable trend is the development of assembloids, which fuse or co-culture organoids of different regions, defined as organoids with spatially organized multiple cell types (Pașca et al. 2022), to simulate neural cell migration (Birey et al. 2017; Tang 2021), synaptic connectivity (Wu and Nowakowski 2025) and neuroinflammation interactions (Liu et al. 2025; Kim et al. 2025) Compared with single-region organoids, assembloids provide a more physiologically relevant platform for studying complex neurotoxicity.

Regional specificity in environmental toxicology

Different brain regions exhibit significant differences in sensitivity to toxicants. For example, the midbrain dopaminergic neurons are vulnerable to heavy metals and pesticides, linked to Parkinson’s disease (PD)-like pathological changes (Hongen et al. 2024; Pamies et al. 2022). The cerebral cortex is sensitive to endocrine disruptors (e.g., BPA, DEHP) and air pollutants (e.g., PM2.5) (Cao et al. 2023; Han et al. 2024; Yang et al. 2023), while the hippocampus is particularly affected by PFAS and microplastics, leading to impaired synaptogenesis (Lu et al. 2024; Park et al. 2025).

Applications in environmental toxicology

Model systems

Brain organoids have been widely used to establish models of neurodevelopment, neurodegeneration, and basic neuronal functions associated with environmental toxicant exposure. Neurodevelopmental models can mimic early brain development (gestational weeks 8–20) and are used to study the potential developmental neurotoxicity (DNT) of toxicants such as heavy metals (e.g., lead) (Li et al. 2024) and endocrine disruptors (e.g., paroxetine, BPA) (Cao et al. 2023; Zhong et al. 2020). In neurodegenerative models, organoids can mimic the physiological characteristics of the adult brain (e.g., containing mature neurons, astrocytes, and microglia) and are used to study toxicant-induced neurodegenerative changes, such as PD (Pamies et al. 2022) and Alzheimer’s disease (AD) (Takata et al. 2023). Brain organoids with electrophysiologically active neuronal networks, detected via micro-electrode arrays (MEAs) (Parmentier et al. 2023), can be used to assess toxicant-induced functional impairments, such as in vivo seizure-like activity (Enright et al. 2020) and neuropsychiatric diseases caused by disrupted brain acetylcholine regulatory functions (N. C. Liu et al. 2022a, b).

Studying specific toxicants

Brain organoids have been used to investigate the neurotoxicity of various environmental toxicants (Table 2), including heavy metals, pesticides, air pollutants, endocrine disruptors, pharmaceuticals, among others.

Table 2.

Neurotoxicity research findings of environmental toxicants in brain organoids

Toxicant Categories	Examples	Dose	Time	Model and Generation Protocols	Key Findings
Heavy Metals	Cd (Huang et al. 2021)	10, 40, 80 µmol/L	24 h	Brain organoids from hESC (Lancaster and Knoblich 2014)	Neuronal apoptosis, reduced neural progenitor proliferation
	Pb (Li et al. 2024)	10, 50, 200 µmol/L		Cortical organoids from hESC (Paşca et al. 2015)	Reduced neurogenesis, premature neuronal differentiation
	MeHg (Pamies et al. 2022)	0, 20, 100, 500, 1000, 5000 µmol/L	24 h	Brain organoids from hiPSC (Pamies et al. 2017)	Dopaminergic neuron damage, reduced neuronal viability
Pesticides	CPF (Modafferi et al. 2021)	100 µmol/L	24 h	Brain organoids from hiPSC (Pamies et al. 2017)	Neurodevelopmental abnormalities, neurotransmitter imbalance
	Rotenone (Hongen et al. 2024)	1, 10, 25 µmol/L	72 h	Brain organoids from hiPSC (Lancaster and Knoblich 2014)	Dopaminergic neuron apoptosis, α-synuclein aggregation
	TOCP (Chen et al., 2021)	0, 1, 5, 10 mmol/L	24 h	Commercial brain organoids (Zhejiang Holdbio Co., Ltd.)	Cell apoptosis, reduced organoid viability
Air Pollutants	PM2.5 (Zeng et al. 2021)	0, 25, 50, 100 µg/mL	7, 14, 21 d	Retinal organoids from hESC (Kuwahara et al. 2015)	Cell apoptosis, retinal progenitor proliferation inhibition
	PS-NPs (Chen et al. 2023)	0.025, 0.05, 0.1 mg/mL	14 d	Cortical organoids from hESC (Paşca et al. 2015)	Cortical layer formation disruption, mitochondrial dysfunction
Endocrine Disruptors	BPA (Chesnut et al. 2021)	0.1- 100 µmol/L	7 d	Brain organoids from hiPSC (Pamies et al. 2017)	Neural progenitor depletion, cortical layer patterning disruption
	DEHP (Yang et al. 2023)	0, 10, 50, 100, 200, 400 µmol/L	7, 14 d	Cortical organoids from hESC (Giandomenico et al. 2021; Lancaster and Knoblich 2014)	Neural progenitor migration disruption
	PFOS (Tian et al. 2024)	0, 30, 100, 300 µg/mL	7 d	Midbrain organoids from hESC (Jo et al. 2016)	Astrocyte activation, neuronal apoptosis, and alterations in neurite morphology
Pharmaceuticals	DTG (Caiaffa et al. 2024)	0, 10, 20 µmol/L	24 h	Brain organoids from hESC (Lancaster and Knoblich 2014; Lancaster et al. 2013)	Modifies the gene expression patterns, decreased organoid volumes and internal stiffness
	Cocaine (Kindberg et al. 2014)	3 µmol/L	0.5 h	Neocortical organoids from hESC (Lancaster et al. 2013)	Premature neuronal differentiation and enhanced neurogenesis of various cortical neuronal subtypes
	Ketamine (Du et al. 2023)	20, 100 µmol/L	3, 12 d	Cerebral organoids from hESC (Lancaster and Knoblich 2014)	Abnormal development of cortical organoids
Others	ACR (Bu et al. 2020)	2.5, 5 mmol/L	24 h	Brain organoids from hESC (Lancaster and Knoblich 2014)	Neuronal differentiation inhibition, cell apoptosis
	AgNPs (Y. Huang et al. 2022a, b)	0.1, 0.5 µg/mL	7 d	Brain organoids from hiPSC (Paşca et al. 2015)	Neuronal apoptosis, cell cycle arrest
	Phenylalanine (Kim et al. 2022)	10, 30, 50 mmol/L	5 d	Cerebral organoids from hiPSC (Lancaster et al. 2017)	Microcephaly, abnormal corticalexpansion, and myelination lesions in the developing human brain
	Nicotine (Wang et al. 2018)	1, 10 µmol/L	0, 5, 14 d	Brain organoid-on-a-chip (Lancaster and Knoblich 2014; Lancaster et al. 2013)	Premature neuronal differentiation, disrupted brain regional organization, abnormal cortical development and neuronal outgrowth

Endpoints measured

In brain organoids, a variety of endpoints can be used to evaluate neurotoxicity, covering cellular, molecular, structural, and functional levels. Cell viability/apoptosis is typically detected using CellTiter-Glo (CTG) assay, Terminal Deoxynucleotidyl Transferase-mediated dUTP Nick-End Labeling (TUNEL) assay, and lactate dehydrogenase (LDH) release assay. Molecularly, quantitative real-time polymerase chain reaction (qPCR) (Bu et al. 2020), RNA sequencing (RNA-seq), and single-cell RNA sequencing (scRNA-seq) (Hua et al. 2022a, b), immunofluorescence and Western blotting (Chesnut et al. 2021; Zhong, 2020) are commonly used. Morphologically, high-resolution imaging (confocal microscopy, light-sheet microscopy) reveals structural abnormalities (Hua et al. 2022a, b; Huang et al. 2021). Functionally, MEA and patch-clamp recordings capture neural network activity and single-cell electrophysiological properties (e.g., bursting frequency, action potential amplitude) (Pamies et al. 2022; Parmentier et al. 2023).

Advantages and unique capabilities for environmental toxicology

Physiological sophistication and cell-type specificity

Unlike 2D monocultures, hiPSC-derived brain organoids self-organize into complex architectures, including cortical layers and ventricular zones, comprising neurons, astrocytes, and microglia (Schwartz et al. 2015; Yan et al. 2016; Zheng et al. 2021). This structural heterogeneity enables the capture of cell-type-specific toxic responses, overcoming interspecies extrapolation limitations (Fig. 2). For instance, specific neuronal subtypes like dopaminergic neurons or neural progenitors exhibit differential sensitivity to toxicants such as 6-hydroxydopamine or BPA (Pamies et al. 2022; Yan et al. 2016). By integrating scRNA-seq and spatial transcriptomics, researchers can resolve molecular dysregulation within specific cellular subsets (Monzel et al. 2020), providing a high-resolution map of selective susceptibility.

Fig. 2.

Advantages of brain organoids in environmental toxicology. The figure shows three advantages: (1) Physiological Sophistication and Cell-Type Specificity (2) Chronic Exposure and Dose–Response Realism (3) Modeling Complex Mixtures and Functional Qutcomes

Chronic exposure and dose–response realism

A definitive advantage of brain organoids is their longevity, which enables the simulation of chronic, low-dose exposure paradigms (e.g., months-long treatment) that are technically unfeasible in 2D systems and ethically/financially prohibitive in animals (Fig. 2). Such models have successfully identified subtle neurodevelopmental disruptions from cadmium exposure over 49–77 days—effects that are typically masked in acute, high-dose assays (Huang et al. 2024). Crucially, organoids demonstrate the sensitivity required to model concentrations proximate to human biomonitoring data (nM to low µM range). Recent studies on BPA, DEHP, and AgNPs emphasize the importance of this sensitivity, as these chemicals frequently exhibit non-monotonic dose responses (NMDR) (Cao et al. 2023; Y. Huang et al. 2022a, b; Yang et al. 2023). This underscores the risk of relying on high-dose extrapolation, as low-dose effects may be driven by nuanced endocrine or epigenetic dysregulation rather than the nonspecific cytotoxicity observed at µM–mM levels.

Modeling complex mixtures and functional outcomes

In the real world, human exposure involves complex mixtures of environmental chemicals. While conventional 2D and co-culture models can be employed to test compound mixtures, brain organoids provide a more sophisticated platform to study synergistic toxicity (Fig. 2). Their complex 3D architecture and diverse cell–cell interactions allow for a more nuanced simulation of real-world exposure scenarios, capturing how pollutants interact within a tissue-like environment that simpler models may not fully replicate. This capability was demonstrated in modeling Gulf War Illness, where the combined impact of organophosphates and cortisol revealed multifaceted pathologies, including tau hyperphosphorylation, that single-agent studies could not replicate (Yates et al. 2022).

The biological relevance of these interactions is further validated through advanced functional and structural readouts. Technologies such as confocal imaging, MEAs, and patch-clamp allow for the quantification of disrupted synaptic density, neurite outgrowth, and network synchrony (Amartumur et al. 2024; Park et al. 2022; Chesnut et al. 2021; Fan et al. 2022; Yang et al. 2023). Recent integration of ultra-high-field MRI has even enabled non-invasive tracking of fiber tract integrity (Habib et al. 2025; Versace et al. 2024), directly linking chemical-induced structural damage to functional neural network impairment (Pamies et al. 2022; Yan et al. 2016).

Revealing neurotoxic mechanisms of environmental toxicants

Novel and unique mechanistic insights

Dysregulation of 3D intercellular communication

Human brain development is governed by an intricate 3D intercellular communication network involving neurons and glia cells. Compared to traditional 2D cultures, brain organoids enable the investigation of toxicity-induced neuron-neuron, neuron-glia interactions (Fig. 3). For instance, the assembloid technique was employed to study the interaction of neurons between the cortex and thalamus (Xiang et al. 2019). A brain-on-a-chip (CANDY) enables observation of the interactions of neurons with astrocytes and accurate assessment of responses to penetration of MeHg across the blood–brain barrier. It is confirmed that the promising vulnerability of neurons is identified to MeHg toxicity as shattered neurites and neuroprotection by astrocytes against MeHg-induced toxicity (Choi et al. 2024). Moreover, Cd exposure regulated neural cell proliferation and death, upregulated the astrocytic marker GFAP, induced neuroinflammation, and impaired ciliogenesis, which hinder the normal development of the fetal brain (Huang et al. 2021). Single-cell RNA sequencing (scRNA-seq) has been instrumental in deconstructing these 3D networks. Kanton et al. (2019) utilized scRNA-seq to map the developmental trajectories of human organoids, discovering that specific gene regulatory networks in glial cells are uniquely sensitive to environmental stressors and genetic mutations (Kanton et al. 2019).

Fig. 3.

Revealing neurotoxic mechanisms of environmental toxicants. The illustration highlights novel and unique mechanistic insights supported by brain organoids: (1) Dysregulation of 3D intercellular communication; (2) Disruption of human-specific neurodevelopmental programs. The classical pathways include oxidative stress, neuroinflammation, receptor–ligand interactions, disruption of neurotransmitter homeostasis, and epigenetic regulation

Furthermore, the integration of non-neural cell types, has revealed how immune-neuronal interactions are altered during injury. Astrocytes, in conjunction with capillaries, contribute to BBB integrity. Certain xenobiotics disrupt BBB permeability by altering glial morphology, structure, and population dynamics, thereby impeding their differentiation and maturation. Myelin sheaths, formed by oligodendrocytes around CNS axons, are essential for rapid saltatory conduction. A combined strategy integrating myelin quantification, PLP1 total fluorescence, and Western blot analysis revealed that MeHg and BPA exposure dose-dependently inhibit oligodendrocyte proliferation, leading to hypomyelination or demyelination (Chesnut et al. 2021). This disrupts essential neuron-oligodendrocyte signaling and impairs axonal conduction.

Disruption of human-specific neurodevelopmental programs

During embryogenesis, neural proliferation, differentiation, patterning, and apoptosis are tightly orchestrated in a spatiotemporal manner. The integrity of these processes is fundamental to postnatal brain development and adult homeostasis, and their regulation is both dynamic and highly sensitive to environmental changes (Fig. 3). Numerous environmental insults can cause significant deviation from the proper pathway and are detrimental to brain development and homeostasis (Boda et al. 2020). For instance, Cd disrupts organoid development through induction of cell death, accompanied by a compensatory increase in cell proliferation and subsequent depletion of progenitor cells, leading to the collapse of cortical layers (Hu et al. 2025). AgNPs exposure impairs ciliary assembly and elongation at low concentrations, which may disrupt the cell cycle and lead to aberrant brain organoid growth, whereas higher concentrations disrupt neurite outgrowth (Y. Huang et al. 2022a, b).

Additionally, exposure of forebrain cortical spheroids to polystyrene microplastics (PS-MPs) triggers oxidative stress and disrupts cortical layering. Notably, this exposure elicits biphasic, time-dependent effects: short-term exposure enhances cell proliferation and upregulates neural progenitor genes, whereas prolonged treatment reduces viability and suppresses markers of mature neurons and cortical layer VI (Hua et al. 2022a, b). In contrast, human neural retinal organoids exposed to polystyrene nanoplastics (PS-NPs) demonstrate concentration-dependent neurotoxicity, which is synergistically potentiated by co-exposure to Cd (Gao et al. 2025). These divergent outcomes likely arise from differences in particle size and organoid model systems.

The human brain has a massively expanded population of Outer Radial Glia cells (oRGs), which are critical for the folding and expansion of the neocortex. Rodents naturally lack this abundant cell population. Early-life Al(OH)₃ exposure induces premature neuronal differentiation in a time- and dose-dependent manner. Notably, at later developmental stages (day 90), a marked reduction in oRGs—which guide neuronal migration—was observed, alongside promoted differentiation of neural progenitors into astrocytes (Wang et al. 2023). Similarly, DEHP disrupts oRGs development, with dysregulation of “cell–ECM interactions” identified as a core molecular mechanism (Yang et al. 2023). This specific mode of toxicity is missed in 2D cultures due to the lack of the 3D scaffolding required for oRG migration.

Classical neurotoxic mechanism pathways

Oxidative stress and mitochondrial dysfunction

Oxidative stress is defined by a disruption of the pro-oxidant/antioxidant balance, leading to persistent reactive oxygen species (ROS) accumulation and macromolecular damage (Hodgson and Smart 2008). A range of environmental toxicants, including rotenone (Y. Y. Huang et al. 2022a, b), graphene oxide (GO) (X. Liu et al. 2022a, b), sevoflurane polystyrene plastics (Gao et al. 2025; Hua et al. 2022a, b; Tao et al. 2024), zinc (Hua et al. 2022a, b), and Cd (Hu et al. 2025), are known to induce mitochondrial ROS overproduction, triggering a cascade of macromolecular damage and dysfunction. Furthermore, TOCP exposure induces ROS-mediated dysregulation of calcium signaling, disrupting intracellular Ca²⁺ homeostasis and amplifying oxidative stress, thereby impairing neuronal activity (Chen et al. 2021). Simultaneously, as critical activators of MAPK and PI3K/Akt pathways, ROS contribute to PM2.5-induced suppression of proliferation and potentiation of apoptosis during early human neural retina development, as evidenced by transcriptomic analyses (Zeng et al. 2021).

Neuroinflammation and glial cell activation

Glial cells, comprising astrocytes, microglia, and oligodendrocytes, play pivotal roles in mediating immune responses (Choi et al. 2024). Neurotoxicity often involves glial activation and morphological changes, leading to sustained neuroinflammatory cascades, which is recognized as one of the earliest detectable pathological endpoints following low-dose, chronic exposure. For instance, lead acetate specifically induces neuroimmune responses accompanied by microglial morphological changes in organoid models (Yuan et al. 2025). Similarly, a midbrain organoid study demonstrated that PFAS exposure impairs neurodevelopment via promoted neuroinflammation. Notably, microelectrode array measurements revealed a biphasic effect on neuronal activity: enhancement at low concentrations and suppression at high concentrations (Tian et al. 2024).

Change in receptor–ligand interactions

Receptors are membrane-bound or cytoplasmic proteins that recognize specific ligands to initiate biological effects. The WNT signaling pathway, a central network regulating cellular life activities, is initiated by the binding of WNT ligands to Frizzled receptors and LRP5/6 co-receptors. Chronic low-dose exposure PS-NPs (Chen et al. 2023), bisphenol S (BPS) (Abdulla et al. 2025), and lead (Li et al. 2024) has been shown to impede WNT signaling, alter the expression of brain regionalization-related signaling molecules, and consequently suppress neuronal differentiation. Furthermore, PS-NPs interfere with the neuroactive ligand–receptor interaction pathway by downregulating key neuronal markers, with CYSLTR1 and PTH1R preliminarily identified as potential molecular targets (Huang et al. 2025). Another study on face mask–derived microplastics reported that co-exposure to triphenyl 2-phosphate elicited more pronounced neurotoxicity than microplastics alone (Li et al. 2025). Nowadays, computational molecular docking, a technique for predicting xenobiotic toxicity by identifying binding sites and calculating interaction energies, is widely employed in receptor–ligand interaction studies.

Disruption of neurotransmitter homeostasis

Neurotransmitters homeostasis—encompassing synthesis, storage, release, reuptake, and degradation—is fundamental to neural communication (Südhof 2013). In many cases, however, disrupted neurotransmitter homeostasis may not present overt histopathological alterations but instead manifest as neural dysfunction or behavioral changes. For instance, organophosphates, carbamate insecticides, and nerve agents such as soman selectively inhibit acetylcholinesterase activity, preventing acetylcholine inactivation. The resultant accumulation of acetylcholine in the synaptic cleft leads to excessive stimulation of postsynaptic receptors, inducing cholinergic hyperactivation. In contrast, methylmercury enhances the release of catecholamine neurotransmitters, while exposure to the CPF disrupts the balance between excitatory and inhibitory neurotransmission (Modafferi et al. 2021).

Alterations in epigenetic regulation

Epigenetics refers to heritable changes in gene expression that occur without alterations to the underlying DNA sequence. Key epigenetic mechanisms include DNA methylation, histone modification and so on (Tian et al. 2013). Researchers utilize microarray profiling, RNA-seq, Northern blotting and RT-qPCR to map and quantify the specific epigenetic modifications induced by environmental toxicants exposures. For instance, classical polycyclic aromatic hydrocarbons (PAHs), heavy metals and the pesticide rotenone have been shown to alter global DNA methylation patterns, serving as biomarkers for neurotoxicity assessment. Airborne PM2.5 exposure has been linked to increased histone acetyltransferase activity, elevated histone acetylation, and subsequent upregulation of pro-inflammatory cytokine expression (Gu et al. 2017). In nanotoxicology, gold nanoparticles have been reported to reduce levels of the repressive histone mark H3K27me3 (Jiang et al. 2024).

Integrating the AOP framework for neurotoxicity assessment

The AOP is a conceptual framework designed to describe a sequence of causally linked events, starting from a molecular initiating event (MIE), progressing through measurable key events (KEs) at cellular and tissue levels, and culminating in adverse outcomes (AOs) at the organism or population level (Ankley et al. 2010; Pistollato et al. 2020). It offers a powerful structured approach for systematically understanding the toxicological mechanisms of environmental chemicals. The AOP-Wiki, a centralized repository created by the OECD, currently catalogues over 560 established AOPs. As illustrated in the Fig. 4, typical AOPs of four major toxicants are initiated by MIEs such as AChE inhibition (AOP ID 281). These initiating events trigger a cascade of cellular and tissue-level KEs, including mitochondrial damage (AOP ID 564), synaptic loss (AOP ID 450), and glial activation leading to neuroinflammation(AOP ID 544) (Sachana et al. 2018). The progression of these KEs ultimately converges on critical neurotoxicity AOs, manifesting as cognitive deficits, neuro-development delay and impairment of brain function (Dong et al. 2023; Hirano et al. 2023).

Fig. 4.

AOP framework for environmental neurotoxicants. It integrates the OECD documented AOPs of four major environmental toxicants into a consolidated neurotoxicity pathway and explicitly states the value of brain organoid

The emergence of brain organoids has provided a promising platform for constructing and validating human-relevant AOPs in neurodevelopment. Crucially, the figure highlights tissue level KEs that are measurable in 3D organoids, such as Ventricular Zone (VZ) disruption and network burst desynchronization. Although brain organoids are not yet widely applied in AOP construction, they facilitate AOPs by identifying human-specific KEs, verifying mechanisms in 3D context, establishing precise tissue-level dose–response. By systematically integrating high-throughput screening and multi-omics analyses, these advanced models will advance neurotoxicity risk assessment and early warning strategies for environmental pollutants within the AOP framework.

Current limitations and challenges

Standardization and reproducibility concerns

Current brain organoids lack unified protocols, leading to inter-laboratory and inter-batch heterogeneity in cellular composition, structural complexity, maturation, and regional specification due to variations in stem cell lines, differentiation protocols, and culture conditions. This variability impedes direct comparison and integration of research findings across studies. Furthermore, the absence of well-defined quantitative endpoints challenges the standardization of data analysis in toxicological assessments. To bridge this gap, the 2018 OECD Good In Vitro Method Practices (GIVIMP) guidance document provided the foundational criteria. In 2024, the Organoid Standards Initiative (OSI) developed the first version of the guidelines for promoting organoid practical use. Additionally, the USEPA is working to populate databases for Guideline DNT studies (USEPA 870.6300). While these efforts are still in nascent stages, they represent a move toward greater consistency and reliability.

Technical complexity and accessibility issues

The construction of brain organoid models involves intricate procedures requiring specialized skills in stem cell manipulation and microfluidic systems, coupled with extended culture durations spanning several months. High costs associated with specialized reagents, consumables, and bioreactor systems further restrict their feasibility for large-scale toxicant screening. Moreover, advanced phenotyping techniques such as high-resolution imaging and single-cell sequencing for 3D organoids entail expensive instrumentation and complex data analysis, posing additional accessibility barriers.

Incomplete maturation and lack of key physiological structures

Brain organoids generally exhibit limited maturity, predominantly reflecting fetal or early postnatal developmental stages, and full maturation to adult-like phenotypes—such as complete myelination and aging signatures—remains a major challenge. Vascularization is often absent or insufficient, impairing nutrient diffusion, waste removal, and modeling of systemic toxicant delivery. The lack of a functional BBB in standard models further limits the study of neurotoxicant penetration. In addition, most current brain organoids lack microglia. Because microglia that migrate from the yolk sac during development are of mesoderm origin, microglia are not formed in patterned brain organoids derived from the neuroectodermal lineage in which mesoderm and endoderm formation are inhibited (Mosser et al. 2017). However, several researchers have tried to incorporate microglia into brain organoids or prepare separately and then coculture (Xu et al. 2021).

Challenges in research focus, dose relevance and environmental validity

The current toxicological landscape in brain organoid research exhibits an imbalance in chemical coverage. As highlighted in Table 2, most studies focus on well-characterized hazards such as heavy metals and endocrine disruptors. In contrast, the neurotoxicological profiles of pharmaceuticals, PFOA/PFOS, and emerging contaminants remain under-explored. Furthermore, a critical gap exists between experimental dosing and human exposure levels. Many foundational studies utilize µM to mM concentrations to ensure observable phenotypes within feasible experimental windows. While useful for initial hazard identification, such supraphysiological doses risk inducing nonspecific cytotoxicity or metabolic overload rather than specific neurotoxic pathways. This potentially leads to false-positive outcomes, overestimating risks compared to real-world exposures (typically in the nM range). Mechanistically, acute high-dose exposure often overwhelms cellular homeostasis, whereas chronic low-dose exposure is more representative of human reality and tends to subtly dysregulate signaling networks and epigenetic programs. Recent studies on BPA and AgNPs emphasize that these chemicals frequently exhibit non-monotonic dose responses (NMDR), which underscores the danger of high-dose extrapolation.

Scalability and throughput limitations

The 3D structural nature of brain organoids complicates their adaptation to conventional high-throughput screening platforms, hindering standardized data acquisition and analytical workflows. While automation offers partial solutions, the substantial investment required and the complexity of optimizing and validating automated processes present significant practical challenges. Collectively, these factors restrict the broad application of brain organoid in high-throughput neurotoxicity risk assessment and mechanistic toxicology.

Ethics and regulations of brain organoids

With the rapid advancement of brain organoids technology, researchers are increasingly constructing assembloids and attempting cross-species transplantation experiments. A Science article co-authored by leading international scientists highlighted that the primary ethical concerns have now expanded to include: (1) the potential for sentience and consciousness within organoids themselves, (2) animal ethics in cross-species transplantation experiments, and (3) informed consent from cell donors (Pașca et al. 2025). However, significant gaps remain in the current regulatory framework, which often relies on intermittent assessments. This necessitates a sustained, transparent, and global oversight system to strike a balance between scientific and ethical imperatives.

Future perspectives and development directions

Enhancing model complexity and physiological relevance

Future work should focus on integrating organ-on-a-chip technology and incorporating functional vascular networks into brain-like organoids. For instance, a brain organoid-on-a-chip model has demonstrated utility in assessing organophosphate (OP)-induced endothelial barrier disruption, confirming that OPs can traverse the BBB in microphysiological systems and rapidly inhibit AChE activity. Additionally, establishing robust protocols for incorporating immune cells is critical. The CANDY chip platform—a human pluripotent stem cell-derived BBB-brain model co-cultured with neurons and astrocytes—has been successfully developed to precisely simulate the brain microenvironment (Choi et al. 2024). Further advancement may involve multi-organ chips (Rajan et al. 2020) that interconnect brain organoids with hepatic (Wu et al. 2025), intestinal, or other organ models to investigate systemic toxicity and metabolite-mediated effects.

Promoting technological innovation

Development of fully automated culture systems and bioreactors (Abdulla et al. 2025) will enable consistent organoid generation and toxicant exposure assays. The Knight group demonstrated that utilizing micropatterning technologies to constrain cell attachment areas effectively guides organoids to form single, morphologically uniform neural tube structures, known as neural rosettes (Knight et al. 2018). Building upon this foundation, the team introduced the RosetteArray® platform in 2024. This platform enables highly standardized neural tube culture and screening within 96- and 384-well plates, facilitating systematic quantitative assessment of thousands of samples through AI-driven automated image analysis. By integrating robotic liquid handling with real-time biosensing (Kilic et al. 2016), these platforms bridge the gap between complex 3D biology and the rigorous demands of industrial-scale toxicological risk assessment.

Data integration and analysis

AI and machine learning (ML) (Monzel et al. 2020) present powerful tools for automated organoid morphological analysis, cell segmentation, and interpretation of complex electrophysiological data from MEAs. For example, RNA-seq has been employed to construct predictive models assessing mixed exposures to 39 toxins in 3D neural structures derived from human embryonic stem cells, enabling prediction of neurotoxic interactions for novel chemicals (Schwartz et al. 2015). Multi-omics integration—spanning transcriptomics, proteomics, metabolomics, and epigenomics—will be crucial for systematic elucidation of toxicity mechanisms.

Toward regulatory acceptance

Achieving regulatory acceptance for brain organoid models in neurotoxicity assessment requires the establishment of standardized experimental protocols, phenotypic assessment benchmarks, data reporting frameworks, and ethical review processes. Rigorous validation of model reliability, sensitivity, and specificity in predicting neurotoxic outcomes is essential. Case studies demonstrating their application as complementary or alternative approaches to animal testing will help align these models with chemical safety assessment requirements. Regulatory bodies have been adapting to these scientific advancements, with leading health authorities, such as the U.S. Food and Drug Administration (FDA) in the United States and the European Medicines Agency, adopting New Approach Methodologies (NAMs), which range from cell-based assays and microphysiological systems (organ chips) to bioprinting and in silico models, moving away from traditional animal testing. This transition is reflected in legislative changes, such as the FDA Modernization 2.0 Act, which emphasizes nonclinical over conventional testing.

Conclusions

The emergence of brain organoids marks a notable shift in environmental neurotoxicology, moving the field toward more human-relevant, mechanistic, and predictive modeling. Crucially, brain organoids are not merely another in vitro model but a uniquely capable platform that recapitulates key aspects of human brain development and function, including 3D cytoarchitecture, cellular diversity, and emerging network activity. Their application has already shown promising sensitivity in identifying the subtle but adverse outcomes of chronic, low-dose exposures to ubiquitous environmental contaminants, which can be challenging to detect using conventional toxicological approaches. Furthermore, the significant potential of these models lies in elucidating human-specific, mechanistic, and pathway-based neurotoxicity, enabling the precise deconstruction of AOPs from MIEs to AOs. This capability fundamentally bridges a critical gap between traditional toxicology and human health risk assessment. However, challenges remain, such as limited organoid maturation, absence of key physiological structures, difficulties in multi-organ integration, lack of cellular standardization, and the need for further validation of model predictivity. Future work should prioritize the development of more physiologically representative models through vascularization and BBB formation, optimized chip designs for multi-organ coordination, and establishment of standardized frameworks to improve reproducibility and predictive accuracy—ultimately advancing environmental neurotoxicology toward more precise and efficient risk assessment.

Abbreviations

AD

Alzheimer's Disease

AOP

Adverse Outcome Pathway

ACR

Acrylamide

AgNPs

Silver Nanoparticles

BBB

Blood-Brain Barrier

BPA

Bisphenol A

Cd

Cadmium

CNS

Central Nervous System

CPF

Chlorpyrifos

DEHP

Di-(2-ethylhexyl) phthalate

DNT

Developmental Neurotoxicity

ECM

Extracellular Matrix

FGF

Fibroblast Growth Factor

hiPSCs

Human induced Pluripotent Stem Cells

LDH

Lactate Dehydrogenase

LPS

Lipopolysaccharide

MEAs

Micro-Electrode Arrays

MeHg

Methylmercury

oRGs

Outer Radial Glia-like Cells

PD

Parkinson's Disease

PFAS

Per- and Polyfluoroalkyl Substances

PS-MPs

Polystyrene Microplastics

PS-NPs

Polystyrene Nanoplastics

qPCR

Quantitative real-time Polymerase Chain Reaction

RNA-seq

RNA Sequencing

ROS

Reactive Oxygen Species

scRNA-seq

Single-cell RNA Sequencing

TOCP

Tri-ortho-cresyl Phosphate

WNT

Wingless-related Integration Site

Authors contribution

Jiawen Liu and Yanling Xie are co-first authors. Jiawen Liu: Data curation, Conceptualization, Investigation, Writing – original draft. Yanling Xie: Data curation, Conceptualization, Investigation, Writing – original draft. Meihui Zhu: Writing – original draft, Visualization. Zhiqiu Wang: Writing – review & editing, Conceptualization, Methodology. Yan Huang: Writing – review & editing, Conceptualization. Xiaobo Cen: Conceptualization, Supervision, Funding acquisition. Qian Bu: Writing – review & editing, Conceptualization, Supervision, Funding acquisition, Project administration.

Funding

This work was supported by Sichuan Science and Technology Program (24NSFSC0707), National Science Foundation of China (T2350007), 2024 Sichuan Provincial Special Program for Leading Scientists in Basic Research (2024JDKXJ0006) and 135 Project for Disciplines of Excellence of West China Hospital, Sichuan University (ZYGD23011).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Clinical trial number

Not applicable.

Competing interest

The authors declare no competing interests.