Arsenic Attenuates Responsiveness to Inflammatory Stimuli in the SIM-A9 Microglial Cell Line

Abstract

Arsenic (As) is a major public health threat, with more than 200 million people at risk of consuming drinking water that exceeds World Health Organization safety guidelines. Given that inorganic arsenic (iAs) is linked to various neuropsychiatric and neurodegenerative disorders, a better understanding of its mechanisms of toxicity is warranted. Current evidence suggests that microglia are central to the pathophysiology of As-induced effects in the central nervous system. Microglia are resident immune cells in the brain that play a crucial role in surveillance, clearance of pathogens, and wound healing. They undergo distinct stages of development throughout life, and their behavior is known to be disrupted by environmental insults such as iAs. To characterize the mechanisms by which iAs alters microglial function, we examined the impact of subtoxic exposure to trivalent inorganic arsenic (As(III)) on microglial activity, both in the presence and absence of immune challenges, using a spontaneously immortalized murine cell line derived from the neonatal cerebral cortex (SIM-A9). Results indicate that iAs causes early activation of SIM-A9 cells through upregulation of toll-like receptor 4-mediated NF-κB signaling, followed by slower onset of anti-inflammatory effects mediated through increased Nuclear Factor Erythroid 2-related Factor 2 (Nrf2) activity. This later attenuation of responses to inflammatory stimuli suggests that iAs exposure may impair neonatal microglial function and sensitize individuals to secondary challenges relevant to a range of neurological functions and disorders.

Graphical Abstract

1. Introduction

In 2020, the World Health Organization (WHO) listed arsenic (As) among its top 10 chemicals of major public health concern due to health risks in multiple organ systems, including the brain.^1–3 Inorganic As (iAs) is a heavy metalloid with two primary oxidation states: trivalent (III) and pentavalent (V).⁴ Of these, the trivalent form is more toxic due to its capacity to interact with sulfur-containing proteins.³ Trivalent arsenic is also the predominant valence state found in contaminated groundwater due to anoxic conditions, though the pentavalent form can be found in more oxygenated surface level groundwater.⁵ While inorganic arsenic may sometimes be found in the −3 valence state (arsines/arsenides), such compounds are less common due to thermodynamic instability.⁶ Thus, unless otherwise stated, references to iAs hereafter refer to the trivalent form. Upwards of 200 million people globally are potentially exposed to greater than 10 parts per billion (ppb) iAs in their drinking water⁷: levels that exceed WHO and U.S. Environmental Protection Agency (EPA) safety guidelines.^8,9 The impacts of this exposure are potentially devastating. Indeed, iAs exposure has been linked to neurobehavioral disorders such as anxiety and dementia by numerous groups. For example, Spanish adults living near contaminated sites are disproportionately burdened by neuropsychiatric conditions, and associations are dose-dependently linked to iAs levels in soil.¹⁰ This is recapitulated in rodent studies that demonstrate iAs’s ability to induce anxiety-like behavior.^11–14 With regard to dementia risk, iAs has been shown to induce neuronal damage, hyperphosphorylation of tau protein, and upregulation of amyloid precursor protein expression.¹⁵ Thus, while the clinical manifestations of anxiety and dementia are unique and diverse, current evidence suggests iAs may drive development of these neurobehavioral disorders via converging pathogenic mechanisms. This is supported by the fact that individuals with early-life anxiety have a 2-fold higher risk of dementia in later life.^16,17 Indeed, while lewy bodies, amyloid accumulation, periventricular atrophy, and neurofibrillary tangles are canonically associated with dementia pathology, they are also highly associated with anxiety.^18,19 As such, microglial dysfunction is believed to play a central role in the pathophysiology of both neuropsychiatric and neurocognitive disorders.^20,21

Although the mechanisms by which iAs causes neurobehavioral dysfunction is not fully known, disruptions in microglia appear to be central to the pathophysiology of many As-associated diseases.^22–24 Microglia are resident macrophages of the brain parenchyma that serve a crucial role in immune defense. Their progenitors are derived from the yolk sac during early embryonic development, originating from the same primordial tissue as lung, epidermal, and liver macrophages.²⁵ Migration of microglial precursors to the murine brain occurs around embryonic day (E) 7, but maturation occurs in three transcriptionally distinct stages: early microglia [E10.5-E14], pre-microglia [E14-postnatal (PN) 28], and adult microglia [PN28 or older].^26,27 Intriguingly, environmental exposures and cross-talk with the peripheral immune system can alter microglial development and function.²⁶ For example, microglia can be primed by neonatal infections or aging, resulting in exaggerated responses to subsequent inflammatory challenges.²⁸ In contrast, prenatal exposure to air pollution and maternal stress has been shown to induce social/communication deficits and neurocircuitry changes in male offspring that are thought to arise from impaired microglial pruning of thalamocortical synapses.²⁹ Thus, environmental insults have the capacity to heighten or dampen normal microglial responses.^30,31 This is particularly important with regard to the Developmental Origins of Health and Disease (DOAD) hypothesis, as environmental exposures in early life may significantly impact disease risk in later life.

Inorganic As has well-documented yet seemingly contradictory immunotoxic effects.³² The toxic metalloid has been shown to induce immunosuppression that contributes to elevated infection or cancer risk, as well as pro-inflammatory effects that contribute to atherosclerotic disease, neuropsychiatric illnesses, and dementia.^33–36 Thus, a more in-depth analysis of how iAs modulates microglial activity may shed light on how these seemingly dichotomous immune disruptions arise. To provide such insights, we investigated the effects of subtoxic concentrations of iAs(III) in a spontaneously immortalized murine microglia cell line (SIM-A9) derived from neonatal cerebral cortex, both in the presence and absence of a lipopolysaccharide (LPS) immune challenge.

2. Materials and Methods

2.1. Reagents

Spontaneously Immortalized Microglia A9 (SIM-A9; Ref CRL-3265; LOT 70031127) cells and Dulbecco’s Modified Eagle’s: F12 medium (DMEM:F12; Ref 30–2006) were purchased from American Type Culture Collection (Manassas, VA). Fetal bovine serum (FBS; Ref 1600–069) was purchased from Gibco and heat inactivated for 30 min at 56°C prior to use. Heat inactivated horse serum (HI HS; Ref 26050–088), lipopolysaccharide [Salmonella enterica serotype typhimurium (LPS; Ref L6143–1MG, LOT 0000144983)], sodium (meta)arsenite (Ref S7400–100G; LOT SLCB4632), and quantitative polymerase chain reaction (qPCR) primers (Table 1) were purchased from Sigma Aldrich (St Louis, MO). Recombinant mouse interleukin-4 (IL-4; Ref ab259406–1002; LOT GR3406966-3) was purchased from Abcam (Waltham, MA). Recombinant mouse interferon-γ (IFN-γ; Ref 485-MI-100; LOT CFP2923101), recombinant mouse tumor necrosis factor-α (TNF-α; Ref 410-MT-010/CF), and anti-iNOS antibody (Ref: MAB9502; LOT CRF0623031) were purchased from R&D Systems (Minneapolis, MN). Anti-mouse Alexa Fluor 790 (Ref: 115-655-174; LOT 163726) was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA). QIAshredder homogenization kits (Ref 79656) were purchased from QIAGEN. E.Z.N.A Total RNA Kits (Ref 6834–03) were purchased from Omega Bio-Tek (Norcross, GA). qScript cDNA Supermix (Ref 101414–106) was purchased from QuantaBio (Beverly, MA). 2X Universal SYBR Green Fast qPCR mix (Ref RK21203) was purchased from Abclonal (Woburn, MA). pHrodo^™ BioParticles^™ E. Coli Conjugate Phagocytosis Kit (Ref P35361), Promega CellTiter 96^™ AQueous One Solution Cell Proliferation Assay (MTS) (Ref PR-G3580), BCA Protein Assay kits (Ref 23225), Cytochalasin D (Ref PHZ1063; LOT 79612134), and DRAQ5 (Ref 62251; LOT 534DR50200) were purchased from Thermo Fisher Scientific (Waltham, MA). Unless otherwise stated, SIM-A9 cells were cultured in DMEM:F12 with 10% HI FBS and 5% HI HS at 37°C in humidified chambers (5% CO₂). All experiments were conducted using cells between Passage 8 and 12, and all LPS challenges were performed at a concentration of 2.5 ng/mL.

Table 1. Quantitative Polymerase Chain Reaction (qPCR) Primers.

Selected set of pro-inflammatory gene markers. F, Forward primer; R, Reverse primer.

GENE	FORWARD (5’ → 3’)	REVERSE (3’ → 5’)	Length (bp)
HPRT1	GGC CAG ACT TTG TTG GAT TTG	CGC TCA TCT TAG GCT TTG TAT TTG	142
iNOS	GAG ACA GGG AAG TCT GAA GCA C	CCA GCA GTA GTT GCT CCT CTT C	129
TNF-α	GGT GCC TAT GTC TCA GCC TCT T	GCC ATA GAA CTG ATG AGA GGG AG	139
IL-1β	TGG ACC TTC CAG GAT GAG GAC A	GTT CAT CTC GGA GCC TGT AGT G	148
IL-6	TAC CAC TTC ACA AGT CGG AGG C	CTG CAA GTG CAT CAT CGT TGT TC	116

2.2. MTS Cell Viability Assay

SIM-A9 cells were seeded into 96-well plates at 1 × 10⁴ cells/well. Following 24h exposure to iAs at a range of concentrations (0, 0.1, 0.25, 0.5, 0.75, 1, 10 μM), cell viability was determined using the MTS assay according to manufacturer’s instructions (Promega). This was followed by trypan blue staining to verify results.

2.3. RNA Extraction, cDNA Preparation, and Quantitative Real-Time PCR

SIM-A9 cells were seeded in 12-well plates (2 × 10⁵ cells/well) 18h prior to treatment, allowing for three technical replicates per treatment group. Cells received one of the following treatments for 24h: 0 μM iAs (sterile distilled water), 2.5 ng/mL LPS (for the final 6h of the incubation period), 1 μM iAs, or 1 μM iAs + LPS challenge (for the final 6h). The same experimental paradigm was utilized for 6h challenges with IFN-γ (50 ng/mL), TNF-α (10 ng/mL), and IL-4 (10 ng/mL) in the presence or absence of 1 μM iAs. Cells were collected using QIAGEN’s QIAShredder and the E.Z.N.A. Extraction Kit. All cDNA was generated using a Bio-Rad T100 Thermal Cycler. Primers for qPCR are shown in Table 1.

2.4. RNA-seq and Transcription Factor Enrichment Analysis

SIM-A9 cells were treated and processed for RNA sequencing as stated above. Three to four samples from each treatment group with OD260/280 and OD260/230 ≥ 2 were sent for sample preparation, library construction, sequencing, and analysis by Novogene (Sacramento, CA). Briefly, data quality control measures (error rate distribution, GC-content distribution, data filtering) were assessed before mapping to the reference genome (GRCm39). Gene expression was then quantified for inter-sample correlation analysis and Gene Set Enrichment Analysis. Differential Expression Analysis was performed using DESeq2 with Benjamini-Hochberg correction for multiple hypothesis testing. Finally, gene enrichment analysis was used to perform functional analyses (Gene Ontology Enrichment, KEGG Enrichment, Reactome Enrichment). Differences in transcription factor activity were inferred using univariate linear models as implemented in decoupleR. In brief, differences in gene expression were represented by a composite statistic calculated for all genes by multiplying the log₂ fold change in a gene’s expression by the negative log₁₀ of the adjusted p-value for that difference. These values were then used to fit univariate models inferring gene regulation based on the Dorothea database of interactions between genes and transcription factors. The resulting significance values were then adjusted for multiple hypothesis testing using Bonferroni correction.

2.5. Protein Expression

Inducible nitric oxide synthase (iNOS) protein expression was quantified via in-cell western using a modified protocol from Dave et al.³⁷ Briefly, SIM-A9 cells were seeded at a density of 1.65 × 10⁴ cells per well and cultured for 18h. Cells were treated as described above (0 μM iAs, 2.5 ng/mL LPS, 1 μM iAs, 1 μM iAs+LPS) for either 4 h or 24 h and then fixed in filtered 4% paraformaldehyde for 20 min. Cells were washed three times (10 min/wash), blocked in Licor blocking buffer (1h), incubated with primary antibody overnight (Mouse anti-iNOS; 5 μg/mL), washed three times, incubated with secondary antibody and nuclear stain (Goat anti-mouse 790 1:700, DRAQ5 1:100,000) for 1h, washed three times, and then imaged on a Biorad CLx imager. iNOS intensity was normalized to DRAQ5 nuclear intensity to account for potential differences in cell seeding density.

2.6. Phagocytosis Assay

The pHrodo^™ BioParticles^™ E. Coli Conjugate phagocytosis protocol was conducted according to Invitrogen recommendations with adjustments as below. SIM-A9 cells were seeded at 2.5 × 10⁴/well to achieve 1 × 10⁵ cells/well after 48h. Twenty-four hours after seeding, cells received one of the following treatments for 24h in serum-free, phenol-red free DMEM:F12 media: 0 μM iAs, 2.5 ng/mL LPS (for the final 6h of the 24h incubation period), 1 μM iAs, or 1 μM Ais + LPS (for the final 6h). Cytochalasin D (0.5 μM; for the final 6h) was used as a negative control for phagocytosis due to its ability to inhibit actin polymerization. Five technical replicates were run in each of three independent experiments. At the end of the treatment period, 10 μL of pHrodo^™ BioParticle E. Coli Conjugate (dissolved in serum-free, phenol-red free DMEM:F12 and warmed to 20–25°C) was added to each well for a final particle concentration of 100 μg/mL. The 96-well plate was loaded into a Keyence Fluorescence Microscope (BZ-X810) with a surrounding water bath and a Tokai Hit stage top incubator (5% CO₂, 37°C). Phase contrast (1 per 30s exposure) and fluorescence images (1 per 15s exposure) were acquired every 15 minutes for 2 hours. Using automated Keyence software, fluorescence area (brightness threshold range: 9 to 255) was calculated and normalized to cell density to calculate average fluorescence area per cell per timepoint.

2.7. Statistical Analysis

For Figure 1 (dose response for iAs alone), Supplementary Figure 1 (dose response for iAs+LPS), and Supplementary Figure 2 (time-dependent effects), statistical testing was completed by one-way ANOVA followed by Dunnett’s test for multiple comparison correction. For qPCR experiments in Figures 2–3, comparisons were made by two-sample t-test. Statistical testing for RNA-seq and TFEA are described above (Fig 2 and 3). All other analyses that examined iAs- and immune challenge-induced effects (e.g., LPS, IFN-γ, TNF-α, or IL-4) in parallel utilized a two-way ANOVA to identify main effects and interactions between the two variables. Post-hoc analyses were completed via Uncorrected Fisher’s Least Significant Difference (LSD) with four a priori comparisons based on RNA-seq data findings. For the phagocytosis assay, analyses of the four −/+ iAs X −/+ LPS groups were completed by two-way ANOVA as described above, then followed by a one-way ANOVA to compare arsenic-treated groups to cytochalasin D (negative control for phagocytosis) with Sídák’s correction for multiple hypothesis testing.

Figure 1. Arsenic is subtoxic at doses ranging from 0 to 1 μM in SIM-A9 cells.

A) Experimental design for iAs dose response studies. B) MTS assay for cell viability. SIM-A9 cells were 85–90% viable for doses up to 1 μM. C-E) Dose-response curve (0–1 μM) for inflammatory cytokines (TNF-α, IL-1β) and nitrosative stress marker (iNOS) transcript expression. Data reflects 4 experiments with 2 technical replicates per group. One-way ANOVA + Dunnett’s multiple comparisons test, *p<0.05, ***p<0.001.

Figure 2. Arsenic-induced transcriptional changes suggest modulation of defense response and inflammatory cytokine production.

A) Bulk RNA-seq volcano plot (3–4 samples/group) comparing 1 μM iAs-treated SIM-A9 cells to vehicle-treated cells. B) Transcription Factor Enrichment Analysis showing the 30 most statistically significant hits. C) Gene Ontology Analysis for altered pathways following arsenic exposure. D-G) Quantitative real-time PCR for inflammation-associated genes. Data reflects 11 experiments with 3 technical replicates per group. Student’s t-test, **p<0.01, ****p<0.0001

Figure 3. Exposure to 1 μM iAs prior to a lipopolysaccharide challenge attenuates pro-inflammatory activation of SIM-A9.

A) Bulk RNA-seq volcano plot (4 samples/group) showing differentially expressed genes in cells exposed to iAs prior to an LPS challenge versus LPS alone. B) Transcription Factor Enrichment Analysis showing predicted alterations in transcription factor activity following iAs+LPS relative to LPS alone. C) Gene Ontology Analysis for altered pathways following iAs+LPS vs. LPS alone. D) Differentially Expressed Genes (DEGs) induced by iAs in the presence or absence of an LPS challenge. E-H) qPCR data on pro-inflammatory gene expression following iAs+LPS vs. LPS treatment. Data reflects 11 experiments with 3 technical replicates per group. Student’s t-test, *p<0.05, **p<0.01, ****p<0.0001.

3. Results

3.1. SIM-A9 Viability and Dose Response Following Arsenic Exposure

In order to examine subtoxic effects of arsenic exposure on SIM-A9 microglia, cell viability was first assessed via the MTS Assay. SIM-A9 cells were exposed to a range of iAs doses (0, 0.05, 0.1, 0.25, 0.5, 0.75, 1, and 10 μM) for 24h prior to viability assessments (Figure 1A). All concentrations ≤1 μM met our predetermined criteria of 85–90% viability, while cells exposed to 10 μM showed only 18% viability (F_(8,14) = 3.041, p=0.0331; Figure 1B). These data were concordant with trypan blue staining (not shown). Initial analyses for gene expression of inflammatory cytokines (TNF-α, IL-1β) and nitrosative stress (iNOS) showed potential non-monotonic dose-response relationships (Figure 1C–E). Inorganic As concentrations between 0.25 and 1 μM significantly reduced IL-1β expression. iAs had a large and significant effect on TNF-α expression at 0.5 μM (Cohen’s d = 1.6; p=0.0188 by t-test); however, this effect was not significant after correction for multiple hypothesis testing (p = 0.1716). Lastly, although 0.75 and 1 μM doses had a large effect on iNOS expression (Cohen’s d = 1.6 and 1.7, respectively), the effect was only statistically significant at 1 μM after correction for multiple hypothesis testing (p = 0.0188). This two-fold reduction in iNOS activity at 1 μM was particularly interesting, as iAs has previously been shown to reduce iNOS expression in RAW 264.7 macrophages.³⁸ Given the physiologic importance of iNOS for pathogen defense, all subsequent experiments were conducted using 1 μM iAs.

3.2. Arsenic Modulates Pathways Associated with SIM-A9 Immune Activation and Metabolism

Bulk RNA-sequencing comparing 1 μM iAs treatment to vehicle control showed upregulation of several genes known to be induced by arsenic exposure, including Hmox1 and Gclm (Figure 2A). Other notable changes included disruptions in genes related to iron homeostasis, with Slc48a1 upregulated and Cp downregulated. Arsenic also reduced expression of several cellular metabolism and transporter genes, such as Pck2, Asns, Soat2, Aldh1l2, Slc1a4, Gpt2, and Cox6a2. To identify signaling pathways preferentially affected by iAs exposure, we performed transcription factor enrichment analyses (TFEA). The most significantly upregulated transcription factors (TFs) were proto-oncogenes (Erg and Fli1), supporting current data classifying iAs as a Group 1 carcinogen.³⁹ The activity of several general immune modulatory TFs were also altered. Creb1, Batf2, Foxo1/4, Atf2/3/4/5, and Cebpb showed reduced activity, whereas Nfe2l2 (Nrf2), Rela (p65), and Nfkb1 (p105) showed increased activity (Figure 2B). These results appeared somewhat contradictory, as there was simultaneous downregulation of a subset of inflammation-promoting TFs (Batf2, Foxo4, Atf2/3/4/5) and upregulation of others (Rela, Nfkb1). However, with regard to interferon signaling, there was a consistent pattern of downregulation, including reductions in Irf9, Stat2/3, and Sp110 activity. This is reflected in gene ontology pathway analyses that showed augmented responses to viruses and other organisms (Figure 2C). Indeed, despite some variability in iAs-induced effects, confirmatory qPCR analyses showed significant reductions in IL-1β and iNOS expression in exposed cells (Figure 2D–G), suggesting blunted responses to pathogenic signals.

3.3. Arsenic Attenuates LPS-Induced Upregulation of Pro-Inflammatory Genes

To investigate whether arsenic exposure reduces SIM-A9 responsiveness to pathogenic challenges, cells were treated with either 0 or 1 μM iAs for 24h with further exposure to 2.5 ng/mL LPS or vehicle for the final 6h of the incubation period. Bulk RNA-seq again showed upregulation of canonical arsenic-induced genes (Hmox1, Gclm) but also downregulation of an autophagy marker (Fnbp1l) and a number of pro-inflammatory genes (IL-6, Ifi44, Cd69, Rel, Nos2) in the arsenic-exposed group relative to LPS controls (Figure 3A). In contrast to effects of iAs in unchallenged cells (Figure 2B), iAs treatment in LPS-challenged cells attenuated NF-κB signaling [NFkb1 (p105), and Rela (p65)] based upon TFEA (Figure 3B). Similar to iAs treatment in unchallenged cells (Figure 2B), cells treated with 1 μM As+LPS showed reduced interferon activity (Irf9, Sp110) and upregulated Nfe2l2 (Nrf2) activity when compared to cells that received LPS in the absence of iAs (Figure 3B). Gene ontology analyses demonstrated alterations in cytokine activity, viral response activity, and response to other organisms (Figure 3C). Although these results are somewhat similar to those from iAs relative to vehicle (Figure 2C), iAs appears to differentially regulate gene expression in the presence or absence of LPS. Relative to LPS controls, exposure to iAs prior to an LPS challenge results in 585 non-overlapping differentially expressed genes (DEGs). In contrast, only 88 non-overlapping DEGs were noted for iAs versus vehicle. (Figure 3D) Confirmatory qPCR analyses showed a marked reduction in inflammatory cytokine and iNOS expression in SIM-A9 cells treated with 1 μM iAs+LPS relative to SIM-A9 cells that received LPS in the absence of iAs (Figure 3E–F). Relative to LPS controls, iAs+LPS-treated cells exhibited reduced mRNA expression of TNF-α, IL-1β, IL-6, and iNOS by 48.1%, 33.5%, 55.6%, and 51.9%, respectively. Attenuation of LPS-induced inflammatory gene expression was assessed at lower iAs concentrations as well. Arsenic’s effects on IL-1β was large (Cohen’s d = 1.28) at 0.25 μM, but results did not reach statistical significance (p=0.053). Similarly, large effects were noted at 0.25 and 0.5 μM for TNF-α (Cohen’s d = 0.96 and 0.99, respectively), but neither of these comparisons reached statistical significance (Supplementary Figure 1A–C).

3.4. Arsenic has Time-Dependent Effects on Redox Activity and Synergistically Increases Antioxidant Gene Expression in the Presence of LPS

Since RNA-sequencing was only conducted after a 24h iAs treatment, time-course experiments were utilized to examine shorter-term exposures to 1 μM iAs at 2, 4, and 6h. Gene expression data show that 1 μM iAs alone caused a two-fold increase in TLR4 expression at 2h (F_(3,12) = 9.660, p = 0.0016; Supplementary Figure 2A), followed by a two-fold increase in iNOS expression at 4h (F_(3,12) = 4.770, p = 0.0206; Supplementary Figure 2B). To confirm this effect at the protein level, iAs-induced iNOS expression was assessed after 4h and 24h exposure via in-cell western. Additionally, to evaluate iAs-induced effects in the presence of LPS, cells were either co-exposed to 1 μM iAs and LPS for 4h or exposed to 1 μM iAs (24h) with a 6h LPS challenge at the end of the incubation period (Figure 4A–B). At 4h, the effect of LPS on iNOS expression was significant (F_(1,21) = 8.361, p = 0.0087), but the effect of iAs was not (F_(1,21) = 0.6571, p = 0.6571). This was also true for the 24h timepoint (F_(1,20) = 0.0002, p = 0.0002; F_(1,20) = 0.3661, p = 0.5520, respectively). However, significant interaction between iAs and LPS was noted at 4h and 24h (F_(1,21) = 20.21, p = 0.0002; F_(1,20) = 12.41, p = 0.0021, respectively). At 4h, post-hoc testing showed that iNOS protein expression was significantly higher in iAs-exposed cells relative to vehicle controls (Figure 4C; p<0.01), corroborating gene expression data (Supplementary Figure 2B). As expected, LPS treatment increased iNOS expression significantly; however, this LPS-mediated induction of iNOS was significantly blunted by 37.8% in cells co-exposed to iAs (Figure 4C). After treatment for 24 h, there was a trend toward higher iNOS expression in As-treated cells relative to vehicle controls; however, this effect did not reach statistical significance (Figure 4D; p = 0.278). Similar to the 4 h treatment, however, pre-treatment with iAs attenuated LPS-induced iNOS expression by 39.6% (p<0.05). In summary, these experiments detected a significant main effect of LPS, but not iAs. However, these main effects were qualified by an interaction whereby LPS caused an expected increase in iNOS at 0 μM iAs, but prior iAs exposure prevented this effect.

Figure 4. iAs promotes differential regulation of iNOS protein and antioxidant response gene expression in the presence or absence of LPS.

A) For 4h exposures, SIM-A9 cells were exposed to one of 4 conditions: 0 μM As, 2.5 ng/mL LPS, 1 μM As, or 1 μM As + 2.5 ng/mL LPS. For 24h exposures, SIM-A9 cells were exposed to 0 μM As, 2.5 ng/mL LPS (6h), 1 μM As, or 1 μM As for 24h with a 2.5 ng/mL LPS challenge occurring in the last 6h. B) Representative ICW images. C-D) Quantification of ICW fluorescence intensity normalized to nuclear signal (DRAQ5) at 4h and 24h. E-F) qPCR for heme oxygenase-1 and NADPH quinone oxidoreductase-1 at 24h. Data reflects 4–6 experiments with 3 technical replicates per group. Two-way ANOVA + Uncorrected Fisher’s LSD, *p<0.05, **p<0.01, ****p<0.0001

To further examine the redox consequences of iAs, an environmental toxicant known to promote oxidative stress, antioxidant enzyme expression was assessed after 24 h of iAs exposure (Figure 4E–F). LPS and iAs each had a significant effect on HO-1 expression (F_(1,12) = 30.04, p =0.0001; F_(1,12) = 65.33, p < 0.0001, respectively), but only iAs had a significant effect on NQO1 expression (F_(1,20) = 8.285, p = 0.0093). The effect of LPS on NQO1 trended toward, but did not reach, significance (F_(1,20) = 2.502, p = 0.1294). There was a significant interaction between iAs and LPS on heme oxygenase-1 (HO-1) expression (F_(1,12) = 29.79, p = 0.0001), but not NADPH quinone oxidoreductase-1 (NQO1) expression (F_(1,20) = 5.060, p = 0.2747). Indeed, arsenic alone caused an approximate 2.5-fold increase in both HO-1 (p<0.01) and NQO1 (p<0.01) transcript levels whereas LPS alone did not alter HO-1 expression (p = 0.9872) and only modestly increased NQO1 expression (1.4-fold; p<0.01). However, pre-treatment with 1 μM iAs followed by a 6h LPS challenge synergistically increased HO-1 (8.3-fold) and NQO1 (4.6-fold) transcript levels.

3.5. Arsenic Reduces Phagocytic Capacity

To examine whether iAs caused physiologic impairments in pathogen response mechanisms, phagocytosis assays were performed using pH-sensitive Escherichia coli (E. coli) particles (Figure 5A–B). These particles only become fluorescent once ingested and incorporated into phagosomes (pH~5). Phagocytic capacity was determined by calculating the area-under-the-curve (AUC) for fluorescence over time (Figure 5C). Significant main effects were detected for LPS and iAs (F_(1,8) = 11.52, p = 0.0094; F_(1,8) = 90.02, p <0.0001, respectively), as well as a near significant interaction effect (F_(1,8) = 4.384, p=0.0696). SIM-A9 cells exposed to 2.5 ng/mL LPS showed a 23.3% lower phagocytic capacity than vehicle-treated controls (p<0.01), while exposure to 1 μM iAs for 24h reduced phagocytic capacity by 49.3% (p<0.001). The addition of a 6h LPS challenge to iAs exposure reduced phagocytosis in 1 μM iAs-treated cells by 11.3% (Cohen’s d = 0.78), but this was not statistically significant (p = 0.3234). Lastly, 6h treatment with 0.5 μM Cyt D, an actin inhibitor, significantly reduced phagocytic capacity relative to 1 μM iAs-treated cells (p=0.0048) and 1 μM iAs+LPS-treated cells (p = 0.0243).

Figure 5. Exposure to 1 μM iAs reduces phagocytosis in SIM-A9.

A) Representative images of pHrodo bioparticle fluorescence at 0, 60, and 120-min following 0 μM iAs, 2.5 ng/mL LPS, 1 μM iAs, or 1 μM iAs + 2.5 ng/mL LPS treatment (SB = 50 μm). B) pHrodo fluorescence normalized to cell count over time. C) Area Under the Curve (AUC) calculated from 0 to 120 min for each exposure group. Data reflects 3 experiments with 5 technical replicates per group. Two-way ANOVA + Uncorrected Fisher’s LSD for LPS- and iAs-mediated effects followed by one-way ANOVA + Sídák’s test for multiple comparisons against Cyt D, *p<0.05, **p<0.01, ***p<0.001.

3.6. Arsenic-Induced Attenuation of Inflammatory Signaling is Not Specific to LPS

Similar to effects observed with LPS, stimulation with either IFN-γ or TNF-α for the final 6h of a 24h exposure to 1 μM iAs also attenuated expression of TNF-α, IL-1β, IL-6, and iNOS at the transcript level relative to IFN-γ or TNF-α-treatment alone. For IFN-γ induction, transcript levels of these four pro-inflammatory genes were reduced by 50.8%, 58.0%, 66.2%, and 53.2%, respectively (Figure 6A–D). Interactions between iAs and IFN-γ were significant for TNF-α (F_(1,24) = 16.76, p = 0.0004), IL-1β (F_(1,16) = 11, p = 0.0044), and iNOS (F_(1,16) = 4.903, p = 0.0417), but not IL-6 (F_(1,20) = 0.2131, p = 0.6494). The individual effects of iAs and IFN-γ were significant for all four pro-inflammatory genes. For TNF-α treatment, pro-inflammatory gene expression was reduced by 16.7% for TNF-α, 21.1% for IL-1β, 18.5% for IL-6, and 25.2% for iNOS (Figure 6E–H). Interaction between iAs and TNF-α trended toward significance for TNF-α expression (F(1,36) = 3.201, p = 0.0820), but was not significant for IL-1β, IL-6, or iNOS. Even so, individual effects of iAs and TNF-α were significant for all four pro-inflammatory genes except TNF-α, which showed no significant effect of iAs.

Figure 6. Exposure to 1 μM iAs attenuates IFN-γ and TNF-α responses but promotes IL-4 response.

qPCR was run for (A) TNF-α, (B) IL-1β, (C) IL-6, and (D) iNOS following exposure to 0 μM iAs (24h), IFN-γ (6h), 1 μM iAs (24h), or 1 μM iAs (24h) + IFN-γ (6h). The same experimental paradigm was followed for TNF-α stimulations (E-H) and IL-4 stimulations (I-J). Data reflects 5–11 experiments with 3 technical replicates per group. Two-way ANOVA + Uncorrected Fisher’s LSD, *p<0.05, **p<0.01, ***p<0.001, ****p<0.0001

In contrast, 24h exposure to 1 μM iAs with a 6h IL-4 challenge (Figure 6I) showed a significant interaction effect (F_(1,34) = 7.642, p = 0.0091) on expression of an anti-inflammatory gene marker [i.e., arginase1 (Arg1)]. The individual effect of IL-4 on Arg1 expression was significant (F_(1,34) = 1250, p <0.0001), but the effect of iAs was not (F_(1,34) = 0.07456, p = 0.7865). Relative to IL-4 controls, exposure to iAs prior to an IL-4 challenge increased expression of Arg1 by 28.3%; although this did not reach statistical significance (p=0.09051), the effect size was large (Cohen’s d = 0.87). Interestingly, no significant interactions or individual effects were noted for iAs or IL-4 with regard to iNOS expression (Figure 6J). However, IL-4 exposure significantly reduced iNOS expression at 0 μM iAs (p = 0.0198), and 1 μM iAs significantly reduced iNOS expression in the absence of IL-4 (p = 0.0402).

4. Discussion

4.1. Time-dependent effects of arsenic exposure

Arsenic is a common environmental toxicant known to increase the risk of various neurological and psychiatric disorders, many of which are associated with microglial dysfunction. In this study, we utilized targeted and untargeted approaches to examine the impact of iAs exposure on SIM-A9 microglial biology, both in the presence and absence of an immune challenge by LPS. In the absence of LPS, we showed that exposure to 1 μM iAs upregulated iNOS transcript and protein expression at 4h, an effect likely mediated through upstream activation of TLR4 signaling. In contrast, after 24h of iAs exposure, iNOS protein expression was not significantly altered; however, iNOS and IL-1β transcript levels were reduced. This suggests that SIM-A9 cells were in the process of downregulating iNOS activity at 24h, as demonstrated by reductions in iNOS protein levels. Thus, it is possible that arsenic induces time-dependent effects on microglia, with an early activation phase followed by a later state of inhibition.

Time-dependent shifts in activity may arise from compensation for oxidative stress, as it is well-established that arsenic generates reactive oxygen species (ROS) through inhibition of complexes I, II, and III of the electron transport chain.³ This is supported by our data showing upregulation of HO-1 and NQO1 transcript expression after 24h of exposure as well as by TFEA revealing activation of the Nrf2 antioxidant pathway. Interestingly, the current literature suggests that ROS exert time- and concentration-dependent effects on NF-κB activity.^40,41 Oxidative stress tends to induce early phase activation of NF-κB, followed by later phase inhibition.⁴⁰ Similar to our findings, ROS-induced inhibition of NF-κB has previously been documented in cadmium-exposed Chinese hamster ovary cells and rat renal tubular cells.^42,43 Although TFEA showed that iAs alone mildly upregulated NF-κB activity (Rela [p65]), a majority of the negative regulators in the gene set were upregulated (7 of 13) and a majority of the positive regulators were downregulated (50 of 91). Thus, the overall pattern mirrors the shift toward reduced NF-κB activity seen with iNOS expression. At 24h, exposure to iAs had the greatest suppressive effect on the following NF-κB inhibitors: Fn1, Trib3, Pck2, Ephb2, and Timp1. The most highly upregulated NF-κB activators included Agt, Klk1b27, Serpine1, and Ccl3. Dynamic temporal changes resulting from ROS-mediated feedback inhibition may explain why Rela [p65] and Nfkb1 [p105] showed slightly increased activity while a majority of other immunoregulatory transcription factors showed moderate to largely downregulated activity (Batf2, Foxo4, Atf2, Irf9, Sp110, Stat2/3). Chromatin immunoprecipitation (CHIP)-seq or proteomic analyses may shed more light on how iAs dynamically augments the gene expression profiles of microglia.

4.2. Context-dependent effects of arsenic exposure

Pre-exposure to 1 μM iAs severely blunted LPS-mediated induction of pro-inflammatory transcript expression and iNOS protein expression, suggesting that arsenic-exposed microglia were less responsive to pathogen-specific molecular patterns (PAMPs) than unexposed microglia. This aligned with TFEA data, which showed marked downregulation of NF-κB activity (RelA [p65], NFkb1[p105]) in the presence of iAs+LPS. Indeed, multiple publications have demonstrated downregulation of NF-κB signaling in peripheral macrophages exposed to iAs for ≥24h.^44–46 Additionally, LPS augmented iAs-mediated transcriptional induction of the antioxidant enzymes HO-1 and NQO1, suggesting their combined effect induced more oxidative stress than the sum of their individual effects. This may explain why the immunosuppressive effects of iAs were more pronounced at 24h in the context of an immune challenge than with iAs treatment alone.

SIM-A9 cells exposed to iAs prior to an IFN-γ challenge produced half as many pro-inflammatory transcripts as cells treated with IFN-γ alone, suggesting a reduced capacity to respond to viral infection. This is supported by TFEA showing reduced activity of Interferon Response Factor 9 (Irf9), Signal Transducer and Activator of Transcription 2/3 (Stat2/3), and Speckled Protein 110 kDa (Sp110). GO functional analyses also showed augmentation of responses to viruses and other pathogens. In contrast, iAs caused only mild attenuation of TNF-α-mediated effects on pro-inflammatory gene expression, suggesting iAs may differentially modulate responses to specific immune challenges. Even so, taking into context the neonatal period from which SIM-A9 cell lines are derived, these studies indicate that iAs exposure may disrupt microglia function in a manner that contributes to adverse neurodevelopmental outcomes.^47,48

Data from IL-4 stimulation indicates differential effects of iAs in the presence or absence of an anti-inflammatory cytokine as well. Exposure to 1 μM iAs alone for 24h significantly attenuated Arg1 expression, which corresponded with TFEA data indicating mildly elevated NF-kB activity. In contrast, exposure to 1 μM iAs prior to an IL-4 stimulation increased Arg1 transcript levels relative to IL-4 controls, suggesting a greater shift toward anti-inflammatory signaling. Notably, while the activities of Arg1 and iNOS are often thought to be intertwined due to competition for arginine substrate, dissociation of Arg1- and iNOS-related activity has been reported in several studies, likely due to arginine-independent mechanisms.^49–51

4.3. Arsenic-induced effects on phagocytosis

Importantly, functional assessment of microglial phagocytic capacity following arsenic treatment support suppositions drawn from gene and protein expression data. Specifically, SIM-A9 cells treated with LPS showed reduced capacity to phagocytize Escherichia coli particles, an effect that matches data reported by other groups.^52,53 Indeed, while it is well-established that TLR4 activation increases phagocytosis in macrophages,^54,55 the opposite is likely true for microglia.⁵⁶ Importantly, microglia exposed to arsenic showed severely attenuated phagocytic capacity relative to vehicle or LPS controls. Furthermore, there was a trend toward even greater inhibition of phagocytosis in iAs-treated cells exposed to an LPS challenge; however, this augmentation requires further confirmation. While the current literature suggests reductions in phagocytic capacity may stem from arsenic’s ability to inhibit actin dynamics and induce cytoskeletal reorganization,^57,58 more research is needed to elucidate the precise molecular mechanisms by which arsenic inhibits phagocytosis under basal and stimulated conditions.

4.4. Limitations

Arsenic exposure induced time-dependent effects on microglial biology, with prolonged exposure attenuating cellular responses to several important inflammatory stimuli, likely due to Nrf2 activation and NF-κB inhibition. Despite these important findings, several limitations of our studies should be noted. First, while SIM-A9 are derived from female neonatal cerebral cortex like BV2 cells,⁵⁹ arsenic-induced effects on inflammatory gene expression and phagocytosis may not be consistent across cell lines. For example, BV2 cells exposed to 1 μM iAs(III) have previously been shown to increase phagocytosis of neuronal debris,⁶⁰ while in another study, exposure to 0.5 and 2 μM iAs(III) increased expression of several pro-inflammatory cytokines.⁶¹ This difference may be explained by several variables: (1) cell line-specific effects, (2) subpopulation-specific phenotypes, or (3) differences in dosage. SIM-A9 cells are spontaneously immortalized microglia, which may behave differently from virally transformed BV2 cells. However, differences in phagocytic behavior between BV2 and SIM-A9 cells might also suggest that unique subpopulations of cortical microglia exist,^62,63 and that they respond differently to arsenic. This is supported by single cell data demonstrating that microglial transcriptomes are highly heterogeneous during the early postnatal period.⁶⁴ Importantly, pentavalent arsenic (As(V)) viability curves from previous work in BV2 cells also show non-monotonic dose-response relationships.⁶⁵ Thus, further work is required to clarify dose-response relationships as well as the underlying mechanisms that distinguish arsenic responses in BV2 and SIM-A9; importantly, effects in primary microglia need to be explored. Additionally, given that BV2 and SIM-A9 cells are both derived from the cerebral cortex of female neonates,^66,67 more work is required to examine male microglia to determine if iAs has sex-specific effects.

4.5. Summary

Using the SIM-A9 microglia model, we showed that exposure to iAs induced time-dependent effects on inflammatory signaling. Supporting this data, we note that anti-inflammatory phenotypes have been reported in several peripheral macrophage subtypes following exposure to iAs for ≥24h.^68–70 In the literature, there is also increasing evidence suggesting arsenic trioxide, an FDA-approved treatment for acute promyelocytic leukemia,⁷¹ may be a beneficial therapy for autoimmune encephalomyelitis due to its ability to reduce microglial activation and inflammatory cytokine expression.⁷² Conversely, arsenic’s ability to attenuate normal responses to inflammatory stimuli such as LPS, IFN-γ, or TNF-α raises concerns that environmental exposure to this common toxicant may increase infection risk and delay inflammation resolution, leading to prolonged infections and increased tissue damage. Indeed, blunted cytokine signaling and reduced phagocytic capacity indicate that iAs-exposed microglia may be slower to recruit other immune cells and clear infections. This can ultimately progress into chronic inflammation, a hallmark feature of numerous neurobehavioral disorders. Hence, further characterization of arsenic effects across cell lines, primary microglia, and microglia subpopulations as well as levels and durations of exposure is needed to better understand how arsenic alters neuroimmune activity. Elucidating these mechanisms will be crucial for illuminating of how iAs contributes to the development of neuropsychiatric/neurocognitive disorders.

5. Conclusions

In the literature, iAs shows contradictory effects on the immune system, having been linked to both immunosuppressive and inflammatory signaling. In this study, we demonstrated that exposure to iAs causes early activation of microglia, followed by later inhibition of NF-κB activity in favor of Nrf2 signaling. This effect is potentially mediated through oxidative stress-induced feedback inhibition, and is more pronounced in the presence of inflammatory challenges like LPS, IFN-γ, and TNF-α. These disruptions in microglia biology may underlie arsenic’s contribution to common neurobehavioral disorders.