High-throughput cytokine detection platform for evaluation of chemical induced microglial activation

Abstract

Environmental chemical exposure, such as pesticides and heavy metals, may contribute to neurodegenerative disorders through neuroinflammation. This study aims to identify suitable in vitro microglial models for assessing cytokine responses to potential neurotoxicants, particularly focusing on human induced pluripotent stem cell-derived microglia (hiMG). In this study, we evaluated the cytokine secretion profiles of four microglial cell types—hiMG, HMC3, IM-HM, and BV2—upon stimulation with lipopolysaccharides (LPS) using cytokine arrays. Our findings showed cytokine response patterns in hiMG cells that most closely resemble in vivo conditions, with significant increases in interleukin 6 (IL-6) and tumor necrosis factor-alpha (TNF-α) levels, the latter being uniquely expressed after LPS treatment. Consequently, we developed a homogeneous time-resolved fluorescence (HTRF) assay platform in a 1536-well plate format for high-throughput screening of environmental chemicals using hiMG cells. After LPS treatment, the assay window for secretion of IL-6 and TNF-α increased 3.71-fold and 2.62-fold over the vehicle control group, respectively, with respective EC₅₀ values of approximately 50 ng/mL and 90 ng/mL for IL-6 and TNF-α. We also assessed the response activity of hiMG to other stimuli, including interferon gamma and various catecholamine compounds, and nine environmental chemicals with evidence of cytokine-inducing potential in other in vitro assays. While all nine tested agents stimulated IL-6 and TNF-α production, three compounds (e.g., picoxystrobin) showed significant stimulation of both cytokines. This study establishes a reliable high-throughput platform for detecting inflammatory effects of environmental toxicants in a microglial cell assay, contributing valuable insights into their neuroinflammatory potential and possible implications for neurodegenerative disorders.

1. Introduction

Emerging evidence highlights the critical role of environmental chemicals, including pesticides and metals, in the pathogenesis of neurodegenerative disorders such as Alzheimer’s disease (AD) and Parkinson’s disease (PD) [1]. Notably, exposure to neurotoxic substances has been associated with neuroinflammation and subsequent neurodegeneration via various mechanisms [2,3]. Neuroinflammation is a complex response of the central nervous system (CNS) to a wide range of pathological stimuli, where microglia, the resident immune cells of the brain, play a key role. Depending on their activation state, microglia can mediate protective effects or exacerbate neuronal damage [4]. Upon sensing harmful stimuli, microglia shift from a ramified ‘resting’ state (M0) to a rounded ameboid ‘activated’ state (M1), orchestrating the immune response by releasing pro-inflammatory cytokines, chemokines, and reactive oxygen species [5].

In a healthy brain, microglia perform essential functions, such as synaptic pruning and maintenance of neuronal integrity [6]; however, microglial activation is a hallmark of neuroinflammatory processes observed in a spectrum of neurological disorders [7]. In the context of neurodegenerative diseases like AD and PD, excessive microglial activation may contribute significantly to neuronal damage and cognitive deterioration [8]. The continuous release of pro-inflammatory mediators by activated microglia can invoke a vicious cycle of inflammation, exacerbating neural injury and provoking synaptic dysfunction [9]. Furthermore, emerging research has revealed that microglia are implicated in conditions beyond neurodegeneration, including ischemic stroke, where their activation can facilitate acute neuronal death [10].

Given these facts, there is an important need to develop effective and high-throughput methods to detect inflammation-inducing chemicals in neuro-relevant cell models to understand if they can further inform neurotoxicity assessment. Many commonly used in vitro microglia cell lines could not accurately represent the complex in vivo behavior of these cells, particularly concerning tumor necrosis factor alpha (TNF-α) production in response to lipopolysaccharides (LPS) stimulation [11]. Therefore, selecting an appropriate in vitro microglia model that closely mimics in vivo conditions is critical for developing neuroinflammation detection assays on high-throughput platforms for screening environmental chemicals and other potential neurotoxicants.

In this study, we evaluated the cytokine secretion profiles of four microglia cell lines in response to LPS stimulation. The models assessed included human induced pluripotent stem cell (iPSC)-derived microglia (hiMG) [12], the human embryonic microglial cell line HMC3 [13], the immortalized human microglia (IM-HM) cell line [14], and the murine BV2 microglia cell line [15]. Based on the cytokine expression pattern, we selected hiMG as a model system and identified interleukin 6 (IL-6) and TNF-α as biomarkers, given the pronounced response of hiMG cells to LPS stimulation. IL-6 serves as a general marker for inflammatory states in human microglia and significantly increases in response to LPS [16,17], making it valuable for monitoring microglial activation in response to chemical exposure. Conversely, TNF-α, as a specific pro-inflammatory marker [18], is uniquely expressed in LPS-stimulated hiMG, making it effective for screening potential pro-inflammatory compounds. Using homogeneous time-resolved fluorescence (HTRF) technology [19], we developed IL-6 and TNF-α secretion assays in a 1536-well plate format with hiMG cells. Furthermore, we evaluated the response of this hiMG-based cytokine detection platform to other stimuli, such as interferon γ (IFNγ), cyclosporine A, and environmental chemicals with evidence of cytokine-inducing potential in other in vitro assays (e.g., picoxystrobin). The assay platform developed in this study can be used to detect environmental toxicants that induce neuroinflammation, suggesting its potential application in hazard assessments.

2. Materials and methods

2.1. Reagents

LPS (catalog: L2630, CAS#: 93,572-42-0), dobutamine hydrochlo-ride (catalog: D0676, CAS#: 49,745-95-1), (-)-epinephrine bitartrate (catalog: E4250, CAS#: 51-43-4), (+)-isoproterenol (+)-bitartrate salt (catalog: I8005, CAS#: 14,638-70-1), 5-chloro-N-(2-chloro-4-nitrophenyl)-2-hydroxybenzamide (catalog: N3510, CAS#: 50-65-7), picoxystrobin (catalog: 33,658, CAS#: 117,428-22-5),), 2-(Thiocyanomethylthio)benzothiazole (catalog: ADVH0430B7E6), Benzethonium chloride (catalog: B8879), 1,4-Naphthoquinone (catalog: 152,757), ß-Nitrostyrene (catalog: AMBH324A4993), 3,3,4,4-Tetrachlorotetrahydrothiophene 1,1-dioxide (catalog: ALNH9AA09B39) and tributyltin benzoate (catalog: 423,882, CAS#: 4342-36-3) were obtained from Sigma-Aldrich (Rockville, MD, USA). IFNγ (catalog: 300-021) and IL-1β (catalog: 200-01B) were sourced from PeproTech (Cranbury, NJ, USA). Cyclosporin A (catalog: sc-3503, CAS#: 59,865-13-3) was purchased from Santa Cruz Biotechnology (Dallas, TX, USA). All other chemicals were supplied by the NCATS compound management group. They are 2-(Thiocyanomethylthio)benzothiazole (CAS#: 21,564-17-0), benzalkonium chloride (CAS#: 122-18-9), 1,4-Naphthoquinone (CAS# 130-15-4), beta-Nitrostyrene (CAS# 102-96-5), and 3,3,4,4-Tetrachlorotetrahydrothiophene 1,1-dioxide (CAS# 3737-41-5).

2.2. Cell culture

Human iPSC-derived differentiated microglia (catalog: BX-0900) were acquired from BrainXell (Madison, WI, USA) and cultured in poly-d-lysine (PDL)-coated plates using Advanced DMEM/F-12 medium (Thermo Fisher Scientific, catalog: 12,634,010, Waltham, MA, USA) supplemented with 1 % N-2 Supplement (Thermo Fisher Scientific, catalog: 17,502,048), 2 % B-27 Supplement (Thermo Fisher Scientific, catalog: 17,504,044), 1 % Glutamax Supplement (Thermo Fisher Scientific, catalog: 35,050,061), 1 % MEM Non-Essential Amino Acids Solution (Thermo Fisher Scientific, catalog: 11,140,050), 1X Chemically Defined Lipid Concentrate (Thermo Fisher Scientific, catalog: 11,905,031), 0.2 mM Ascorbic Acid (Sigma-Aldrich, catalog: A8960), 100 ng/mL IL-34 (PeproTech, catalog: 200-34), 20 ng/mL MCSF (PeproTech, catalog: 300-25), 2 ng/mL TGF-β1 (PeproTech, catalog: 100-21), and 15 μg/mL Geltrex^™ LDEV-Free Reduced Growth Factor Basement Membrane Matrix (Thermo Fisher Scientific, catalog: A1413201).

The HMC3 human microglia cells (CRL-3304) were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured with 90 % EMEM (ATCC, catalog: 30-2003) supplemented with 10 % fetal bovine serum (FBS, Cytiva, catalog: SH30070.03, Marlborough, MA, USA), as well as 100 U/mL penicillin and 100 μg/mL streptomycin (Thermo Fisher Scientific, catalog: 15,140,122).

Immortalized human microglia cells (IM-HM) were purchased from Innoprot (Bizkaia, Spain) and cultured using a microglia medium kit (Innoprot, catalog: P60116), which consists of 500 mL microglial basal medium, 5 mL microglial growth supplement, 25 mL FBS, and 5 mL of penicillin/streptomycin solution.

The murine BV2 microglia cells were purchased from the AcceGen (Fairfield, NJ, USA) and maintained at 90 % RPMI-1640 (ATCC, catalog: 30-2001) supplemented with 10 % FBS as well as 100 U/mL penicillin and 100 μg/mL streptomycin.

For the cytokine detection assay, 0.5 million of each type of microglia cell was seeded into each well of a 6-well plate with their respective culture medium and treated with 100 ng/ml LPS or assay buffer as a control for 24 h.

2.3. Cytokine array detection

To analyze cytokine expression patterns in each cell type, a human cytokine array containing 42 targets (Abcam, catalog: ab133997, Waltham, MA, USA) was used for human cells, while a mouse cytokine array containing 62 targets (Abcam, catalog: ab133995) was used for murine cells. Experimental procedures were performed according to the manufacturer’s instructions. In brief, the supernatant was collected and assayed for secreted inflammatory cytokines. Cytokine detection membranes were blocked with 1X blocking buffer and incubated overnight with cell culture supernatant at 4 °C. Membranes were subsequently washed three times with wash buffer I, followed by incubation with biotinylated antibody for 2 h at room temperature, and further incubation with horseradish peroxidase (HRP)-conjugated streptavidin (1:1000) for an additional 2 h at room temperature. Membranes were developed using Immobilon Forte Western HRP substrate (Millipore, WELUF0500, Rockville, MD, USA). Spots on membranes were quantified using ImageJ software with a protein array analyzer plugin to determine the integrated signal density for each cytokine, normalized against the internal standards provided with the array. Fold-differences in cytokine secretion were reported for treatment groups relative to controls (Supplementary Table 1).

2.4. HTFR-based cytokine detection assay

The experiment was conducted following the manufacturer’s instructions of the HTRF human IL-6 detection kit or HTRF human TNFα detection kit (Revvity, Waltham, MA, USA). Briefly, hiMG cells were dispensed at 1000 cells/4 μL/well into 1536-well white-wall/solid-bottom plates using a Multidrop Combi dispenser (Thermo Fisher Scientific). After a period of incubation (based on the experiment) at 37 °C with a humidified atmosphere of 5 % CO₂ in air, 23 nL of compounds dissolved in DMSO were transferred to the assay plate using Pintool Workstation (Wako Automation, San Diego, CA, USA), resulting in a 175-fold dilution. Following this, cells were incubated for 24 h at 37 °C. Subsequently, 1 μL/well of a pre-mixed detection solution was added to each well, followed by a 2-hour incubation at room temperature. The HTRF signal was read using the PHERAstar (BMG Labtech, Cary, NC, USA), and the HTRF ratio was calculated as follows: Ratio = Signal 665 nm/Signal 620 nm×10,000. The % of activity was calculated as follows: % activity = [(V_compound – V_DMSO)/(V_LPS – V_DMSO)] ×100, where Vcompound represents compound well values. The mean values of the LPS positive control (final concentration: 517.5 ng/ml) and DMSO negative control were represented by V_LPS and V_DMSO, respectively.

2.5. Chemical selection for toxicological screening comparison

A set of compounds with evidence of cytokine activity in other in vitro cell models were identified using the ToxCast database (invitrodb v4.2, https://doi.org/10.23645/epacomptox.6062623.v13) and a subset were selected for testing in the HTRF detection assay. To identify chemicals with relevant cytokine activity, assays targeting IL-6 and TNF-alpha were identified in the ToxCast database (BSK_CASM3C_IL6, BSK_LPS_TNFa, LTEA_HepaRG_IL6, BSK_BT_xIL6, BSK_BT_xTNFa). The assays included: 1) the BioMAP platform (BSK) which uses complex human primary cell systems to measure inflammatory response [20,21] and 2) a high-throughput multiplexed-readout assay using HepaRG human liver cells (LTEA) [22]. Bioactivity was evaluated by calculating the hit rate for each chemical (total number of active hit calls/ total number of endpoints tested). A hit call (hitc) ≥ 0.9 was taken as ‘active’ in multiple-concentration level 5 data in the ToxCast database [23]. For chemicals with more than one sample tested per endpoint, a representative sample was selected based on the tcplSubsetChid function (tcpl v3.2.0) [23], identifying the most active and/or most potent activity based on a series of logic: 1) the most active sample is identified by hit call, 2) in the case of a tie where both samples are active, the most potent sample is identified as the chemical with the lowest concentration at 50 % maximal activity (AC₅₀). Nine chemicals were selected based on a hit rate of >75 % (number of active hit calls/ total number of endpoints tested) and chemical availability for testing (See Supplemental Table 2 for the ToxCast concentration-response modeling results (AC₅₀ values) from the selected chemicals and endpoints). Chemicals were tested at least three endpoints. They are 5-chloro-N-(2-chloro-4-nitrophenyl)-2-hydroxybenzamide (CAS#: 50–65-7), picoxystrobin (CAS#: 117,428–22-5), 2-(Thiocyanomethylthio)benzothiazole (CAS#: 21,564–17-0), cyclosporin A (CAS#: 59,865–13-3), benzalkonium chloride (CAS#: 122–18-9), tributyltin benzoate (CAS#: 4342–36-3), 1, 4-Naphthoquinone (CAS# 130–15-4), beta-Nitrostyrene (CAS# 102–96-5), and 3,3,4,4-Tetrachlorotetrahydrothiophene 1,1-dioxide (CAS# 3737–41-5).

Chemicals were tested in a concentration range of 8.015 pM-115 pM (15 points with 1/3 dilution) in half-log spacing. Although cytotoxicity was not directly measured in this assay, the HTRF-based cytokine detection assay measured increasing cytokine signal with the assumption that robust cell death in response to chemical exposure after 24 h would result in a decrease in cytokine signal compared to controls [24].

2.6. Statistical analysis

All data were presented as mean ± standard deviation (SD) from at least three independent experiments (biological replicates conducted on different days), unless stated otherwise, and plotted using GraphPad Prism software. The two-tailed unpaired Student’s test of the mean was used for single comparisons of statistical significance, while ANOVA with Tukey’s multiple comparison test was employed for multiple comparisons among experimental groups. Differences were considered statistically significant at p < 0.05.

3. Results

3.1. Identification of optimal microglial models based on cytokine profile

To identify the most suitable in vitro microglial model for cytokine secretion assay development, we evaluated three commonly used human microglial cell models: hiMG, HMC3, and IM-HM. We characterized their baseline cytokine secretion profiles and assessed their responses to LPS stimulation using a human cytokine array comprising 42 targets related to cellular inflammation.

Among these three models, hiMG exhibited the most pronounced basal cytokine secretion, with ten detectable cytokines (e.g., IL-6 and TNF-α) in the array analysis, outperforming both HMC3 and IM-HM (Fig. 1a–g). Following stimulation with 100 ng/ml LPS for 24 h, hiMG demonstrated the highest increase in cytokine levels among the models assessed. Notably, six cytokines—IL-6, IL-10, CCL8 (MCP-2), CXCL5 (ENA-78), CCL5 (RANTES), and TNF-α—showed significant upregulation in hiMG cells (Fig. 1b–c and supplementary Figure 1A). In contrast, HMC3 cells only displayed significant increases in IL-6 and CCL5, while IM-HM cells exhibited upregulation of IL-6, IL-7, CXCL8 (IL-8), CCL5, and CCL2 (MCP-1) (Fig. 1d–g and Supplementary Figure 1B–C). Remarkably, TNF-α was exclusively expressed in hiMG in response to LPS stimulation, while both IL-6 and CCL5 levels increased across all three human microglial cell types.

Fig. 1. Analysis of Cytokine Secretion by Human Microglia Using a Cytokine Array.

(A) A layout map depicting the human cytokine array with 42 specified targets. (B) Comparative cytokine array results from the culture medium of hiMG exposed to 100 ng/ml LPS or water (control) for 24 h. (C) Quantitative imaging analysis displaying cytokine array spots of hiMG, highlighting discernible cytokines shown in B. (D) Cytokine array data collected from HMC3 cell culture medium after treatment of 100 ng/ml LPS or water (control) for 24 h. (E) Image quantification of cytokine array spots in HMC3 samples, focusing on the visible cytokines shown in D. (F) Cytokine array outcomes from IM-HM cell culture medium after treatment with 100 ng/ml LPS or water (control) for 24 h. (G) Image quantification of cytokine array spots of IM-HM samples, representing visible cytokines shown in F. The red and blue boxes denoted the position of TNF-α and IL-6, respectively. All values are represented by the mean ± SD (n = 3 replicates). Statistical significance was determined by one-way ANOVA followed by Tukey’s multiple-comparison test, with p-values presented in each graph.

In contrast to human microglial cells, the murine BV2 microglial cell line exhibited a distinctly different cytokine expression profile following LPS treatment. Stimulation of BV2 cells resulted in a significant increase in eight cytokines: CSF3 (GCSF), CCL12 (MCP-5), CCL3 (MIP-1α), CXCL2 (MIP-2), CCL9 (MIP-γ), CCL5 (RANTES), TNFRSF1A (sTNF RI), and TNFRSF1B (sTNF RII) (Fig. 2a–c and Supplementary Figure 1D). Notably, interleukins and TNF-α were undetectable in BV2 cells, even following LPS stimulation, suggesting that this murine microglial model could not adequately mimic human microglial responses to extracellular stimuli in vitro (Supplementary Table 1).

Fig. 2. Evaluation of Cytokine Secretion in Murine Microglia Using a Cytokine Array.

(A) A layout map depicting the mouse cytokine array with 62 specified targets. (B) Cytokine profiling from BV2 cell culture medium treated with 100 ng/ml LPS or water (control) for 24 h. (C) Imaging quantification of cytokine array spots of BV2 samples, showing visible cytokines shown in B. The red and blue boxes denoted the position of TNF-α and IL-6, respectively. All values are represented by the mean ± SD (n = 3 replicates). Statistical analysis was performed using a one-way ANOVA followed by Tukey’s multiple-comparison test, with p-values presented in each graph.

After evaluation of the four microglia cell types, hiMG emerges as the preferred model for in vitro assay development due to its cytokine expression patterns in response to LPS, which closely align with those observed in vivo, particularly the upregulation of TNF-α and IL-6 [11]. Based on these findings, we selected IL-6 and TNF-α as biomarkers for developing high-throughput screening assays targeting inflammation-inducing chemicals.

3.2. Development and optimization of high-throughput screening IL-6 and TNF-α assays in hiMG cells

To develop high-throughput cytokine detection assays, we evaluated the optimal pre-incubation duration of hiMG cells in 1536-well plates prior to compound treatment because maturation and functional responsiveness of hiMG cells are critical for assay optimization. We examined the effects of LPS on hiMG cells after pre-incubation periods of 2-, 3-, or 4-days post-seeding. As illustrated in Fig. 3a–c, a 2-day pre-incubation yielded minimal LPS-induced responses. In contrast, a 3-day pre-incubation resulted in peak induction to 24 h LPS stimulation, as evidenced by robust secretion of IL-6 and significantly increased production of TNF-α (Fig. 3b–c). A 4-day pre-incubation did not enhance the outcomes significantly. While TNF-α levels remained elevated, IL-6 secretion markedly decreased in response to LPS with a 4-day pre-incubation (Fig. 3b–c).

Fig. 3. Parameter Optimization for the High-Throughput Test on hiMG.

(A) Schematic diagram illustrating the experimental framework for assessing hiMG under various pre-incubation durations. (B) IL-6 secretion profiles for different pre-incubation periods of hiMG treated with varying LPS concentrations. (C) TNF-α secretion profiles for different pre-incubation periods in hiMG treated with varying LPS concentrations. (D) Brightfield images illustrating the hiMG morphology upon treatment with 100 ng/ml LPS or water (control) for 24 h. (E) TNF-α secretion profiles at different hiMG seeding densities treated with various concentrations of LPS for 24 h. (F) IL-6 secretion profiles at different hiMG seeding densities treated with various concentrations of LPS for 24 h. (G) IL-6 secretion from two human microglia cell lines, HMC3 and IM-HM, following exposure to a range of LPS concentrations for 24 h. All values are represented by the mean ± SD (n = 3 replicates).

Morphological analysis of hiMG following a 3-day of pre-incubation in the 1536-well plates revealed a significant transition towards amoeboid morphology in the presence of LPS, characterized by rounded cells, hypertrophic cell bodies, and retracted or absent processes. This morphological change contrasts with the ramified cell structure observed under control conditions, where microglia exhibit numerous thin branches extending from a smaller cell body. These findings support the transition of microglia from a resting (homeostatic) state (ramified) to an activated state (ameboid) upon LPS stimulation [25] (Fig. 3d).

Additionally, increasing the seeding density of hiMG improved the dynamic range of the TNF-α assay while lowering EC₅₀ values of LPS (Fig. 3e). For the IL-6 assay, a seeding density of 1000 hiMGs per well provided the largest assay window, despite higher cell densities resulting in lower EC₅₀ values for LPS (Fig. 3f). This reduced signal ratio at higher densities may stem from increased basal signal, more rapid nutrient consumption, or accumulation of metabolic waste, potentially retarding cell function or altering cytokine expression. Consequently, we selected 1000 hiMGs per well as the optimal cell density for subsequent testing, which yielded over a three-fold increase in signal for both cytokine detection assays. Furthermore, these HTRF-based cytokine detection assays confirmed that LPS effectively stimulated IL-6 secretion in two other human microglial cell lines, specifically HMC3 and IM-HM, with EC₅₀ values of 49.30 ng/ml and 14.30 ng/ml, respectively (Fig. 3g). However, LPS did not induce TNF-α expression in these lines (Supplementary Figure 2A).

3.3. Characterization of high-throughput HTRF cytokine detection assays variability

To further validate our assay conditions, we selected a seeding density of 1000 hiMGs per well and a pre-incubation period of 3 days for subsequent evaluations of assay performance in a high-throughput format. DMSO was used as a vehicle control in the assay optimization. In the HTRF-based TNF-α assay, the coefficient of variation (CV) for the wells of DMSO control was 9.09 %, with a signal-to-background ratio (S/B) of 2.62 compared to the data generated by 517.5 ng/ml LPS, resulting in a Z-factor of 0.5 (Supplementary Figure 2B–D). Similarly, in the HTRF-based IL-6 assay, the DMSO vehicle control wells of the plate exhibited a CV of 9.76 %, an S/B of 3.71 with 517.5 ng/ml LPS, and a Z-factor of 0.58 (Supplementary Figure 2E–G). These data indicate excellent assay performance for both the TNF-α and IL-6 assays, validating their suitability for high-throughput screening applications within a 1536-well plate format [26].

3.4. Response of hiMG to other stimuli

To further evaluate whether hiMG can respond to other inflammation inducers, such as IFNγ and IL-1β, we treated hiMGs with varying concentrations of IFNγ or IL-1β, ranging from 0.035 ng/ml to 1150 ng/ml. IFNγ concentration-dependently increased both IL-6 and TNF-α secretion, with EC₅₀ values of 273.2 ng/ml and 232.6 ng/ml, respectively, showing relatively lower efficacy and activity at higher concentrations in inducing cytokine secretion compared to LPS treatment (Fig. 4a). In contrast, IL-1β primarily increased in IL-6 secretion only at the highest concentration (1150 ng/mL), but did not have a significant effect on TNF-α expression in hiMG, thereby supporting the notion that IL-1β preferentially enhances IL-6 production rather than TNF-α (Fig. 4b).

Fig. 4. hiMG Response to Additional Inflammatory Stimuli.

(A) IL-6 secretion profiles of hiMG in response to different concentrations of LPS, IFNγ, and IL-1β given separately for 24 h. (B) TNF-α secretion profiles of hiMG in response to different concentrations of LPS, IFNγ, and IL-1β given separately for 24 h. (C) Chemical structures of dobutamine, epinephrine, and isoproterenol. (D) IL-6 secretion profiles of hiMG following treatment with different concentrations of dobutamine, epinephrine, and isoproterenol individually for 24 h. (E) TNF-α secretion profiles of hiMG following treatment with different concentrations of dobutamine, epinephrine, and isoproterenol individually for 24 h. The percentage of activity was calculated based on the maximal effect of LPS measured at the same time. 100 % represents a similar response level to the LPS, and 0 % represents a basal level the same as the DMSO control. All values are represented by the mean ± SD (n = 3 replicates).

Furthermore, we also tested catecholamines on IL-6 and TNF-α secretion because catecholamines are recognized for their potent regulatory effects on immune function, with recent studies implicating activation of adrenergic receptors (ADRs), including alpha-ADR (ADRA) and beta-ADR (ADRB), as a critical mechanism in the modulating of cytokine secretion [27]. To evaluate the impact of catecholamine compounds, we treated hiMGs with three types of ADR agonists: dobutamine, epinephrine, and isoproterenol, using a concentration-dependent approach within our microglial inflammation system (Fig. 4c). As illustrated in Fig. 4d–e, both epinephrine and isoproterenol significantly elevated IL-6 and TNF-α levels, with epinephrine appearing slightly more potent than isoproterenol. Interestingly, dobutamine, which primarily activates ADRB1, did not elicit a similar response, implying that activation of ADRB1 may not be associated with pro-inflammatory cytokine expression in our model since epinephrine can activate both ADRAs and ADRBs, and isoproterenol is a non-selective ADRBs agonist.

3.5. Evaluation of cytokine-inducing compounds predicted from ToxCast data

To explore the potential use of hiMG as a cellular model to screen chemicals for their cytokine response, we tested nine chemicals that were identified from ToxCast in vitro assays measuring cytokine responses in non-neuronal cell models. In our evaluation, the mean (± SD) of DMSO controls across three replicates for the IL-6 assay and TNF-α assay were 7.75 % ± 2.96 % and 5.81 % ± 0.42 %, respectively. Three compounds demonstrated over 20 % activity in both IL-6 and TNF-α assays, 5-chloro-N-(2-chloro-4-nitrophenyl)-2-hydroxybenzamide, 2-(thiocyanomethylthio) benzothiazole, and picoxystrobin (Fig. 5A–C and Supplementary Table 2). Four additional chemicals showed over 20 % activity in the IL-6 assay and slightly increased TNF-α (>10 % activity), including cyclosporine A, benzethonium chloride, tributyltin benzoate, and 1,4-naphthoquinone (Fig. 5d–g and Supplementary Table 2). This suggests that these seven compounds elicit a moderate cytokine response from microglia compared to LPS. Furthermore, beta-nitrostyrene and 3,3,4,4-tetrachlorotetrahydrothiophene 1,1-dioxide also displayed the capability to slightly elevate both IL-6 and TNF-α levels in hiMGs (>10 % activity), indicating a weak cytokine response from microglia compared to LPS (Fig. 5h–i and Supplementary Table 2). Collectively, these findings indicate that these nine compounds may trigger cytokine increases in hiMG cells, suggesting their potential to induce a neuroinflammatory response in microglial cells. Thus, this study provides an assay platform for IL-6- and TNF-α-based high-throughput neuroinflammation detection utilizing hiMG for assessing chemicals that could potentially trigger neuroinflammation.

Fig. 5. Assessment of Cytokine-Inducing Compounds on hiMG.

Concentration-response curves of IL-6 and TNF-α in hiMG after exposure to various compounds: 5-chloro-N-(2-chloro-4-nitrophenyl)-2-hydroxybenzamide (A), picoxystrobin (B), 2-(thiocyanomethylthio) benzothiazole (C), cyclosporine A (D), benzethonium chloride (E), tributyltin benzoate (F), and 1,4-naphthoquinone (G), beta-nitrostyrene (H) or 3,3,4,4-tetrachlorotetrahydrothiophene 1,1-dioxide (I) over a 24-hour period. The percentage of activity was calculated based on the maximal effect of LPS measured at the same time. 100 % represents a similar response level to the LPS, and 0 % represents a basal level the same as the DMSO control. All values are represented by the mean ± SD (n = 3 replicates).

4. Discussion

Recent studies increasingly highlight the association between exposure to neurotoxic substances and the onset or exacerbation of neuroinflammation, which can progress to neurodegeneration [2,3]. Nevertheless, a significant gap persists regarding effective and efficient platforms for detecting inflammatory response in a brain model, thereby enabling high-throughput assessment of the potential neuroinflammatory effects of environmental chemicals. The findings of this study reaffirm the importance of selecting appropriate human microglial models for investigating neuroinflammation. Among the models evaluated, hiMG emerged as the best choice, demonstrating robust cytokine secretion and responsiveness to LPS stimulation comparable to in vivo microglial behavior [28,29]. This finding indicates that hiMG can effectively replicate in vitro the inflammatory responses observed from LPS treatment in vivo, thus serving as a valuable tool for assessing chemical-induced neuroinflammation [30].

Most microglia cell lines are derived from primary cultures obtained from the brain or the spinal cord, subsequently immortalized through viral transduction with oncogenes [11]. The advantages of using cell lines include their ease of maintenance and abundant availability, stemming from their limitless proliferative potential. However, recent research has indicated that microglia cell lines exhibited significant genetic and functional discrepancies compared to primary microglia and ex vivo microglia [31,32]. Furthermore, microglial cell lines originating from neonatal or embryonic CNS sources are unlikely to accurately represent the phenotype of adult microglia, as exemplified by HMC3, which is derived from the CHME-5 cell line. Notably, a recent investigation suggests that CHME-5 cells are actually of rat origin rather than human, raising questions about the validity of HMC3 [33]. However, another study has affirmed the human origin of HMC3 cells (ATCC^®CRL-3304) through extensive analysis, confirming the absence of cross-contaminating cells from other species [13].

In our evaluation, 100 ng/ml LPS was chosen to stimulate microglia cells because it is a frequently used concentration in microglial studies to mimic neuroinflammation, demonstrating its ability to activate microglia cells and trigger cytokine secretion without significant toxicity to microglial and macrophage cells [25,34–38]. As a result, both human microglia cell lines showed a lack of TNF-α expression even after LPS stimulation. In contrast, hiMG exhibited substantial increases in the expression of various pro-inflammatory cytokines, including TNF-α, highlighting its superiority as an in vitro model that accurately reflects in vivo adult human microglial function. The differing reactions to LPS across various microglia models may be attributed to their developmental stage, the unique genetic backgrounds from the donor, and how they were engineered for indefinite growth, all of which contribute to distinct patterns of cytokine secretion.

Conversely, the immortalized murine BV2 microglial model exhibited a distinct cytokine expression profile, characterized by a marked increase in several cytokines while failing to produce detectable levels of interleukins or TNF-α following LPS stimulation. This disparity complicates the extrapolation of this murine system to concordant human microglial responses, raising concerns regarding its utility. Despite BV2’s long-standing use in studies of neuroinflammation and neurodegenerative disorders, including AD and PD [39,40], our findings resonate with the growing consensus that human-derived microglial models, such as hiMG, provide a more reliable platform for studying human-specific inflammatory responses compared to conventional murine models. Furthermore, stem cell technology provides researchers with a readily available source of microglia, making hiMG the most suitable model for the high-throughput assay development and screening applications.

To facilitate high-throughput screening of pro-inflammatory agents, we optimized culture conditions for hiMG. Our results suggest that a pre-incubation of 3 days allows for optimal cell maturation and responsiveness, as reflected in heightened sensitivity and cytokine production upon LPS exposure. This dynamic change in microglial morphology from a ramified to an amoeboid form is consistent with microglia activation and cytokine secretion in both in vivo and in vitro studies, highlighting the adaptability of microglia in response to inflammatory stimuli [30,41]. Moreover, this optimized condition showcases excellent performance in HTRF-based cytokine detection assay development, specifically for IL-6 and TNF-α.

Beyond assessing response to LPS, our study also evaluated the responsiveness of hiMG cells to other inflammatory mediators, specifically IFNγ and IL-1β. The observed concentration-dependent increase in both IL-6 and TNF-α secretion upon IFNγ stimulation confirms that hiMG cells express the necessary receptors and possess the functional signaling pathways required to respond to this crucial cytokine, reflecting their role in orchestrating immune responses involving interferons [42]. Similarly, the increase in IL-6 secretion following IL-1β treatment, alongside its more limited effect on TNF-α, further corroborates the functional presence of IL-1β receptors and associated signaling cascades in hiMG, consistent with known IL-1β biology [43]. This broader responsiveness of hiMG cells—extending beyond simple reactivity to a single Toll-like receptor (TLR) agonist—is profoundly significant. It demonstrates that hiMG cells possess a multifaceted inflammatory repertoire, enabling them to react to diverse pro-inflammatory signals characteristic of the complex neuroinflammatory milieu observed in vivo [44]. The capacity of hiMG cells to integrate and respond to multiple inflammatory stimuli, including interferon and TLR-agonist stimulation, allows for a more nuanced investigation of synergistic or antagonistic pathway interactions that drive neuroinflammatory pathogenesis. This sophisticated responsiveness makes the hiMG model an invaluable tool for accurately identifying and characterizing environmental toxicants or therapeutic compounds that modulate neuroinflammation, better reflecting the complex and polymodal nature of inflammation within the CNS during disease states [44].

Additionally, the interaction of catecholamines with microglial cells adds complexity to our understanding of neuroinflammation. Previous studies suggest that catecholamines can modulate microglial activation and cytokine production [27,45–52]. For example, peripheral treatment with ADR antagonists (e.g., prazosin, propranolol) has been shown to block cytokine response in various structures of the CNS, and ADRB activation correlates with IL-1β mRNA expression, with evidence positing microglia as a primary source of IL-1β [51,53]. However, in vitro studies indicate that stimulation of microglial adrenergic receptor ligands can exert anti-inflammatory effects in LPS-induced microglia cells [54]. This is attributed to a reduction in the production of pro-inflammatory cytokines, superoxide anions, and nitric oxide by microglia [54]. However, the specific effects of catecholamines on cytokine expression in microglia cells remain inadequately characterized. In our study, in the absence of LPS, treatment with catecholamines such as epinephrine and isoproterenol resulted in significant increases in IL-6 and TNF-α secretion in hiMG cells, highlighting the potential for these neurotransmitters to amplify the neuroinflammatory response. This context-dependent nature of catecholamine effects indicates their dual role as both mediators of stress-related neuroinflammation and potential influencers of neuroprotective pathways.

While our findings present a platform for assessing neuroinflammation-inducing agents, it is essential to acknowledge certain limitations. The applicability of results obtained from hiMG to complex in vivo environments remains to be fully established. The brain’s microenvironment is intricately influenced by a multitude of factors, including other cell types, extracellular matrix components, and systemic factors that are not fully replicated in vitro. Thus, caution must be exercised when extrapolating our results to the broader biological context of human neuroinflammation. Building a more complex physiological environment that incorporates various brain cells may be necessary to effectively mimic in vivo conditions.

Moreover, potential donor-to-donor variability among human samples could impact the generalizability of the findings. Differences in genetic background, health status, and environmental exposures among individuals may lead to variability in microglial responses. Future studies should aim to incorporate a range of human iPSC lines to account for this variability and enhance the robustness of the findings.

In addition to characterizing the effects of environmental chemicals, our study establishes a robust platform for assessing potential neuroinflammation-inducing agents, which have broader implications in toxicological screening and drug discovery. Future research could leverage this hiMG-based cytokine secretion assay platform to explore the neurotoxic potential of a more extensive array of environmental chemicals or to evaluate the assay results with orthogonal information (e.g., complex neurobiological models comprising microglia), and other potential toxicants, ultimately enabling researchers to prioritize chemicals for safety evaluations and regulatory decisions. Future work is also needed to disentangle potential cytotoxic factors driving cytokine release in response to chemical exposure, as prolonged overactivation of microglia in culture can lead to cell death [55,56]. Integrating this platform within a larger toxicological framework could assist in identifying key neuroinflammatory pathways activated by various compounds, providing insights for risk assessment in public health.

The application of this assay platform also extends to drug discovery and development for neurodegenerative diseases. By utilizing the hiMG model to screen potential therapeutic agents, researchers can evaluate the anti-inflammatory properties of novel compounds, potentially leading to the identification of drugs that modulate microglial responses and attenuate neuroinflammatory processes. Additionally, the platform can facilitate phenotypic drug screening [57], allowing for the examination of compounds that not only reduce pro-inflammatory cytokine secretion under stimulation but also promote microglial homeostasis and neuroprotection. Furthermore, for our broader neuroinflammation screening efforts in the future, we plan to incorporate a cell viability assay as a counter screen to accurately rule out cytotoxic compounds and distinguish true inflammatory responses from cell death-mediated signal decreases.

In conclusion, this study supports the continued application of human-derived microglial models, particularly hiMG, in neuroinflammation research. By refining our methodologies, we pave the way for more comprehensive assessments of the neurotoxic potential of environmental agents in a high-throughput context. Furthermore, this approach has the potential to not only screen for environmental chemicals that may induce neuroinflammation but also to contribute to advancements in neuropharmacology and therapeutic strategies. Future studies may expand to additional cytokines and cover other immune response pathways that contribute to neuroinflammation and neurodegenerative disorders.

5. Disclaimer

The United States Environmental Protection Agency (U.S. EPA) through its Office of Research and Development has subjected this article to Agency administrative review and approved it for publication. Mention of trade names or commercial products does not constitute endorsement for use. The views expressed in this article are those of the authors and do not necessarily represent the views or policies of the US EPA, DTT/NIEHS, and NCATS.

Supplementary Material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.slast.2025.100347.