Induction of ferroptotic and amyloidogenic signatures linked to Alzheimer's disease by chemically distinct air pollutants

Abstract

Air pollution (AirP) exposure is associated with increased risk of Alzheimer's disease (AD), yet AirP is chemically heterogeneous, complicating identification of shared pathogenic drivers. We compared acute cortical responses to two chemically distinct but metal-rich AirP sources, diesel exhaust particles (DEP) and World Trade Center (WTC) dust, and contrasted them with woodsmoke (WS), a particulate exposure containing substantially lower metal content. Despite major differences in particle composition and size, DEP and WTC elicited highly convergent transcriptomic responses, sharing more than 1200 differentially expressed genes associated with oxidative stress, interferon signaling, ferroptosis, neuronal remodeling, and amyloid processing. These transcriptional changes were accompanied by disrupted glutathione synthesis, altered ferritin-mediated iron storage and heme metabolism, and selective impairment of lipid raft antioxidant defenses, resulting in a 40% increase in lipid raft 4-hydroxynonenal (HNE) with WTC dust with DEP trending similarly, and at least a 65% reduction in phospholipid hydroperoxide detoxification capacity, and 36% increase in the aggregation-prone Aβ42 peptide for both pollution sources. Notably, both exposures produced acute white-matter abnormalities within the corpus callosum despite the absence of bulk brain iron accumulation, indicating that redistribution of bioactive iron rather than total iron burden may be sufficient to promote oxidative injury. In contrast, WS produced a distinct transcriptional profile, lacked coordinated ferroptotic priming, failed to induce lipid peroxidation or Aβ42 accumulation, and showed minimal effects on iron metabolism. Together, these findings identify metal-associated oxidative mechanisms as a convergent pathway linking chemically distinct forms of AirP to ferroptotic vulnerability, amyloidogenic processing, and AD-relevant pathology.

Graphical abstract

Acute AirP exposure converges on ferroptotic priming, amyloidogenic processing, and white-matter vulnerability. Acute exposure to metal-rich AirP, such as DEP or WTC introduces redox-active metals and particulate matter that promote lipid peroxidation, amyloidogenesis, and altered transcriptional regulation in the brain. AirP exposure engages xenobiotic metabolism pathways (AhR/ARNT), activates iron and heme handling through ferritinophagy (NCOA4) and heme oxygenase activity (HMOX1), and blunts lipid peroxide detoxification systems, including glutathione peroxidase 4 (GPx4), ferroptosis suppressor protein 1 (FSP1), and glutathione (GSH) synthesis. These changes promote ferroptotic priming and lipid raft oxidation, facilitating amyloid precursor protein (APP) processing by secretases (ADAM10, BACE1, γ-secretase) and increasing amyloid-β (Aβ) generation. In parallel, transcriptional and cell-state remodeling involving neuronal and oligodendrocyte responses contribute to selective white-matter vulnerability, particularly within the corpus callosum. Together, these pathways provide a mechanistic framework linking acute AirP exposure to convergent oxidative, amyloidogenic, and microstructural changes relevant to Alzheimer's disease pathology. Abbreviations: Aβ, amyloid-β; AhR, aryl hydrocarbon receptor; APP, amyloid precursor protein; ARNT, aryl hydrocarbon receptor nuclear translocator; BACE1, β-site amyloid precursor protein cleaving enzyme 1; DEP, diesel exhaust particles; FSP1, ferroptosis suppressor protein 1; GPx4, glutathione peroxidase 4; GSH, glutathione; HMOX1, heme oxygenase 1; NCOA4, nuclear receptor coactivator 4; WTC, World Trade Center dust.

1. Introduction

AD is a progressive neurodegenerative disorder characterized by memory loss and cognitive decline. While 30 million people worldwide are currently affected, increasing life expectancy projects a five-fold increase in AD by 2050 [1]. Although the molecular mechanisms underlying AD are incompletely understood, aberrant processing and aggregation of β-amyloid (Aβ) peptides are widely implicated in AD progression [2]. Aβ oligomers, fibrils, and plaques disrupt synaptic function, promote oxidative injury, and contribute to neuronal dysfunction and loss [3]. While genetic risk factors and aging are well-established contributors to AD, growing evidence indicates that environmental exposures also play a significant role in modulating disease risk and progression. Among environmental factors, air pollution (AirP) has emerged as a prominent and independent risk factor for dementia and AD [[4], [5], [6], [7]]. Epidemiological studies across diverse populations consistently demonstrate that chronic exposure to particulate matter (PM) and traffic-related air pollution (TRAP) is associated with increased dementia incidence, accelerated cognitive decline, and earlier AD onset [[8], [9], [10], [11]]. These associations persist after adjustment for socioeconomic status, education, and lifestyle factors, underscoring AirP as a modifiable contributor to neurodegenerative disease. Despite this strong epidemiological signal, the biological mechanisms linking AirP exposure to AD pathology remain poorly defined.

A major challenge in understanding AirP-associated neurodegeneration lies in the chemical heterogeneity of AirP itself [12]. AirP encompasses a complex mixture of PM and gases originating from diverse sources, including traffic emissions, industrial activity, structural fires, and biomass combustion. Consequently, it remains unclear whether neurodegenerative outcomes associated with AirP arise from shared biological responses or exposure-specific mechanisms. Notably, many AirP sources contain substantial quantities of metals, including Fe, Cu, Zn, Al, Pb, Cd, Hg, and Mn, which are independently associated with neurotoxicity and oxidative stress [[13], [14], [15]]. Unlike reactive gaseous pollutants such as ozone or NO₂, which primarily exert transient effects in peripheral tissues [16], metals can accumulate within the brain parenchyma, persist for extended periods, and continuously redox-cycle, sustaining oxidative burden [[17], [18], [19]]. Brain metal accumulation can occur through multiple routes, including blood-brain barrier (BBB) dysfunction, active transport mechanisms, and cerebral microhemorrhages (CMBs) [[20], [21], [22]]. CMBs arise decades before clinical AD, with increasing prevalence beginning in early adulthood, and are markedly elevated in AD brains [23,24]. These microvascular events permit the extravasation of heme-bound iron into the brain parenchyma, where iron accumulates within plaques and surrounding tissue. Because CMBs frequently localize to periventricular and subcortical white-matter regions, iron deposition arising from microvascular injury is positioned to disproportionately impact white-matter tracts, including the corpus callosum, which exhibits early vulnerability in AirP-associated neurotoxicity [25]. Iron-laden amyloid plaques are a consistent feature of AD neuropathology and provide a catalytic environment for lipid peroxidation and oxidative damage [[26], [27], [28]]. Together, these observations position persistent brain metal burden as a plausible convergent driver linking chemically distinct AirP exposures to AD-relevant pathology.

We and others have highlighted iron-dependent lipid peroxidation and ferroptosis-associated pathways as important contributors to neurodegeneration, including AD [18,27,28]. Iron readily cycles between ferrous (Fe²⁺) and ferric (Fe³⁺) states, generating highly reactive hydroxyl radicals capable of indiscriminate molecular damage [29]. Because membrane lipids represent a major target of iron-catalyzed oxidation, sustained lipid peroxidation can overwhelm antioxidant defenses and compromise membrane integrity. When lipid peroxide detoxification systems, most notably glutathione peroxidase 4 (GPx4), are impaired, or glutathione (GSH) availability is limited, cells become increasingly vulnerable to ferroptosis [30]. While definitive demonstration of ferroptotic cell death in human AD remains challenging, multiple convergent biochemical and molecular markers strongly support ferroptosis-associated processes as a component of AD pathology [18,27,28,[31], [32], [33]].

Within neuronal membranes, lipid rafts (LRs) represent microdomains enriched in polyunsaturated lipids and cholesterol that are particularly vulnerable to oxidative damage. Importantly, LRs also harbor the secretase complexes responsible for amyloid precursor protein (APP) processing, positioning them as a potential mechanistic nexus between iron-mediated lipid peroxidation and amyloidogenic processing. Analyses of postmortem human AD brain tissue demonstrate diminished GPx4 and GPx1 protein levels and activity within cortical LRs, accompanied by increased lipid peroxidation [27,28]. Decomposition of oxidized lipids generates reactive aldehydes such as 4-hydroxynonenal (HNE), which form irreversible protein adducts that alter enzymatic function [34,35]. HNE modifications have been shown to favor pro-amyloidogenic APP processing, directly linking lipid oxidation within LRs to increased Aβ production [36,37].

Despite these insights, it remains unclear whether chemically distinct AirP sources converge on shared molecular pathways relevant to AD or whether neurodegenerative risk reflects exposure-specific mechanisms. To address this gap, we directly compared three well-described environmentally and chemically distinct AirP exposures: diesel exhaust particles (DEP) [12,25,38,39], a canonical component of TRAP; dust generated by the World Trade Center (WTC) terrorist attacks, which has been associated with an approximately decade-earlier AD trajectory [[40], [41], [42], [43]]; and woodsmoke (WS), a biomass-derived pollutant of increasing relevance due to the growing frequency and severity of wildfires, including recent events in the Los Angeles region [[44], [45], [46]]. By examining whether these exposures converge or diverge in their ability to promote AD-associated molecular, biochemical, and microstructural changes, we tested whether metal-associated oxidative stress represents a common mechanism linking specific forms of AirP exposure to AD-relevant pathology. Acute exposure to the chemically distinct, metal-rich DEP and WTC elicited convergent responses characterized by oxidative stress, impaired lipid peroxide detoxification, ferroptosis-associated priming, altered amyloidogenic processing, and preferential vulnerability of white-matter tracts. In contrast, WS induced a divergent transcriptional and biochemical profile that did not recapitulate these convergent features.

2. Methods

Chemical Composition analysis DEP and WTC chemical composition analysis was performed at the Desert Research Institute's environmental analysis facility. The measurement of OC and EC was conducted using a multiwavelength thermal/optical carbon analyzer (Magee Scientific, Berkeley, CA, USA), adhering to the protocols delineated by the interagency monitoring of protected visual environments [47]. Inorganic ion content was conducted by ion chromatography (IC) by eluting the particles into ultrapure deionized water using sonication, followed by filtration and analysis as previously described [48,49]. Inductively coupled plasma mass spectroscopy (ICP-MS) was employed to determine the concentrations of metals and trace elements, involving a hot block acid digestion process to extract by nitric (HNO₃), hydrochloric (HCl), and hydrofluoric (HF) acid [50,51]. After digestion, the samples were diluted with deionized water, aerosolized, and then introduced into the ICP-MS instrument (Thermo Finnigan Element 2, Thermo Fisher Scientific Inc., Waltham, MA, USA). Woodsmoke metal analysis was also performed by ICP-MS at Michigan State University. Chemical composition and particle sizes for DEP, WTC, and WS have been previously described [42,44,52].

Animal All animal procedures were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals and approved by the University of Southern California Institutional Animal Care and Use Committees (IACUC) under protocol #20842. Male and female mice C57BL/6 (n = 10/group/sex) were housed separately in groups of five, at 22°C/30% humidity and 12 h of light-dark cycles with standard nesting, food, and water ad libitum. 2-month-old mice were exposed to re-aerosolized DEP (NIST SRM 2975), WTC dust, or control filtered air (FA) for 5 h at 100 μg/m³ in whole-body exposure chambers. Briefly, DEP or WTC dust were Milli-Q deionized water-solubilized, mixed, and sonicated for 30 min for re-aerosolized animal exposure [38]. After the 5-h exposure, mice were humanely euthanized and perfused with PBS. One hemisphere of the brain was micro-dissected, frozen, and then stored at −80°C, second hemisphere was fixed in 4% PFA for 24 h and embedded in paraffin for immunofluorescent analysis. Four whole brains per sex in each exposure group (n = 4) were saved for ex vivo MRI.

For the WS exposure, all animal procedures were conducted in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the University of New Mexico IACUC under protocol #23-201419-HSC. Female C57BL/6J mice (n = 10/group) at 2 months of age were exposed to laboratory-generated WS (500 μg/m³) or FA for 4 h in whole-body exposure chambers. WS was derived from cedar chips harvested in the Blue Gap Tachee, Arizona region [45]. Post-exposure, mice were euthanized and perfused with PBS. One hemisphere was micro-dissected and stored at −80°C, the second hemisphere was frozen in −80°C and embedded in OCT for immunofluorescent analysis on frozen sections.

MRI acquisition was performed at the Biological Imaging and Spectroscopy Core at the Zilkha Neurogenetic Institute (University of Southern California). The cryogen-free MR Solutions PET/MRI 7T system was equipped with a bore size of ∼24 cm, a maximum gradient up to 600 mT/m, and a 20 mm internal diameter quadrature bird cage mouse head coil. Ex vivo fixed brains with the skulls intact were prepared following cardiac perfusion and were incubated in 5 mM Gd-DTPA (BioPal, Inc.) for 3 days. The samples were then immersed in perfluorinated polyether fluid (Galden) and positioned with a rubber silicone spacer in a 15 mL centrifuge tube prior to MRI.

A positioning gradient echo sequence was first acquired to prepare the slice stacks for the 3D multi-echo multi-slice (MEMS) spin echo sequence. The MEMS parameters are as follows: echo times (TEs) = 10, 20, 30, 40, 50, 60, 70, 80, 90, and 100 ms; TR = 176 ms; slice thickness = 0.6 mm; field of view (FOV) = 18 mm × 18 mm x 19.2 mm; acquired matrix size (MS) = 256 x 128 x 64 and resized to 192 x 192 x 32; flip angle (FA) = 50°; acquisition bandwidth = 50 kHz; and NA = 1.

2D spin echo diffusion weighted imaging (SEDWI) was performed to calculate diffusion and fractional anisotropy maps and included the following parameters: TE = 23 ms, TR = 4000 ms, 3 b-values (600, 1200, and 1800 s/mm²), 20 directions, FOV = 18 mm × 18 mm, slice thickness = 0.6 mm, 32 slices, MS = 150 x 150, and acquisition bandwidth = 50 kHz.

MRI Analysis T₂ maps were generated from the MEMS images through a pixel-by-pixel exponential fitting of the signal intensities across the different TE times. MATLAB Rocketship v.1.4 module was used to perform the parametric fits [53]. All fits with an r² > 0.6 were included. Using Fiji software, pixels with fits having r² values < 0.6 were set to not-a-number (NaN, i.e., missing data) and were not included in the analysis. Additionally, brain regions were extracted by manually delineating brain outlines on each slice, and outside brain regions were set to NaN. R₂ maps were generated from the T₂ maps, using the relationship T₂ = 1/R₂. Mean R₂ values for each ROI, for each subject, were obtained. Apparent diffusion coefficient (ADC) and fractional anisotropy (FA) maps were generated from the SEDWI scans using DSI studio software [54]. Regions of interest (ROI) were manually delineated using the polygon tool in Fiji.

Western Blot Lipid rafts, nuclear lysates, and whole cell lysates (RIPA) were generated from mouse cortex as previously described and validated [27,28,39,55,56]. Whole cell lysates (20 μg), lipid rafts (5 μg), or nuclear lysates (20 μg) were boiled at 75°C under denatured conditions and resolved on 4-20% gradient gels. Proteins were electroblotted using a Criterion blotter (Bio-Rad Laboratories, Hercules, CA) and transferred onto 0.45 μm polyvinyl difluoride membranes. Membranes were stained using Revert 700 fluorescent protein stain and imaged before blocking with LI-COR Intercept blocking buffer (LI-COR Biosciences, Lincoln, NE), followed by primary antibody incubation overnight at 4°C. Membranes incubated with IRDye 800CW and/or 700CW secondary antibodies and visualized by Odyssey (LI-COR Biosciences). Western blot data were quantified with ImageJ and normalized by total protein per lane and/or loading control protein.

Dot blot RIPA (20 μg) or lipid raft (5 μg) lysates were loaded onto a dot blot apparatus for gravity filtration for 3 h. After filtration, membranes were stained with Revert 700 fluorescent protein stain and imaged before blocking with LI-COR Intercept blocking buffer (LI-COR Biosciences, Lincoln, NE), followed by primary antibody incubation overnight at 4°C.

Immunofluorescence For DEP and WTC exposures, paraffin-embedded brains were sectioned at 5 μm thickness and mounted on slides. Sections were deparaffinized and rehydrated, followed by antigen retrieval using sodium citrate buffer. Slides were then blocked with blocking solution and incubated with primary antibodies overnight at 4°C, followed by incubation with appropriate secondary antibodies for 1 h the following day. Paraffin-embedded sections encompassing the corpus callosum and cortex were stained for markers of lipid peroxidation (HNE) and DNA oxidation (8-OHdG). Paraffin-embedded sections of the corpus callosum were additionally stained for oligodendrocytes (Olig2) and Platelet-Derived Growth Factor Receptor Alpha (PDGFRα).

For WS, frozen OCT-embedded brains were sectioned at 15 μm thickness and mounted on slides. Slides were blocked with donkey serum and incubated with primary and secondary antibodies. Frozen brain sections encompassing the corpus callosum and hippocampus were stained for markers of degraded myelin (dMBP), oxidative damage (HNE, 8-OHdG), complement activation (C5, C5a), and microglial reactivity (Iba1).

To quantify immunofluorescent signal in both frozen and paraffin-embedded tissue, brain sections were imaged at 40x magnification using a BZ-X810 Keyence microscope (Keyence, USA). Image analysis for each brain region was performed by two observers blinded to exposure group using NIH ImageJ software. Image analysis for cell counts was performed by a single observer blinded to exposure group through manual counting of cells expressing Olig2 and Olig2+PDGFRα per area of interest. Olig2 and PDGFRα are shown in representative images as pseudo-colors green and red, respectively. Representative images were selected for display and adjusted uniformly for brightness and contrast across all groups.

Biochemical Assays Total protein was quantified by 660 nm assay for lipid rafts and nuclear lysate, and BCA for whole cell lysates (Thermo Fisher Scientific, Waltham, MA). Total glutathione peroxidase activity was calculated by activity assay (Cayman Chemical, Ann Arbor, MI) following the manufacturer's protocol. For phospholipid hydroperoxidase activity, oxidized phosphatidyl choline (PCOOH) was generated [57] and used in place of the provided cumene hydroperoxide [27,28,58]. The same batch of PCOOH was utilized for all assays. Tissue heme levels were quantified by Quantichrom assay (BioAssay Systems, Hayward, CA). All optical densities were measured using a SpectraMax M2 spectrophotometer equipped with a temperature regulator (Molecular Devices, San Jose, CA).

RNA sequencing 20 mg of cortex was homogenized in TRIzol reagent using a BeadBug Benchtop Homogenizer. Brain tissue was suspended in 1 mL of TRIzol and homogenized for 6 rounds of 10 s homogenization, and 60-s holds between each round. Following homogenization, 300 μL of chloroform was added, and samples were centrifuged with heavy gel phase-lock tubes (VWR, 10847-802) to separate the aqueous phase. The aqueous phase was applied to a standard column-based RNA purification kit (Quantabio, Extracta Plus, Cat# 95214-050) following the manufacturer's protocol. RNA concentration and integrity were assessed with a Qubit 4.0 Fluorometer and Agilent Bioanalyzer prior to sequencing. Library preparation (mRNA library, poly A enrichment) and RNA sequencing (NovoSeq PE150. 6G raw data per sample) was performed by Novogene. Reads were trimmed with trim_galore-0.6.5-1 and mapped to the Mus musculus GRCm39 reference genome with STAR-2.7.3a [59]. Mapped reads were counted to genes using featureCounts (Subread-2.0.3) [60]. Differential expression analysis was performed with DESeq2-1.34.0 [61]. For DEP/WTC exposure datasets, sex was included as a covariate. For the WS dataset, expression data exhibited dominant variance unrelated to exposure status, as assessed by principal component analysis. To improve detection of exposure-associated transcriptional changes, residual technical and latent biological variation was adjusted using RUVSeq (v1.32.0). RUV correction was not applied to DEP or WTC datasets, which showed strong exposure-driven separation without adjustment. Statistical significance for differential expression was defined using Benjamini-Hochberg false discovery rate-adjusted p-values (FDR<0.05). DEGs were subjected to Gene set enrichment analysis (GSEA) analysis for pathway enrichment.

To determine the influence of cellular composition effects, estimated cell-type proportions were derived from bulk RNA-seq using MuSiC v1.1.0 [62] using single-cell reference datasets from Celldex; MouseBrainData, and MouseRNAseqData. Estimated cell proportions were incorporated as covariates in reanalysis by DESeq2 to isolate transcriptomic effects independent of compositional differences among exposure groups. Transcription factor activity was inferred using DoRothEA V1.6.9 by high-confidence regulon and VIPER v1.28.0 for enrichment scoring through full ranked gene lists as inputs. Visualization of RNA data by upset, PCA, or hexbin plots used UpSetR, ggplot2, pheatmap, and RColorBrewer. Raw RNA-seq data are available through Annotare: E-MTAB-16199.

Statistics were performed using SPSS (Version 29.0.20.0; IBM, Chicago, IL). For the majority of biochemical, molecular, and transcriptomic endpoints, means among exposure groups were compared using analysis of covariance (ANCOVA) with sex included as a covariate to adjust for sex-related differences among study groups. Pairwise comparisons were performed using Bonferroni correction with significance set at p < 0.05. Homogeneity of variances was assessed using Levene's test, and non-normally distributed data were log-transformed prior to analysis when appropriate. MEMS and DTI measures were analyzed using two-way ANOVA to assess the effects of exposure and brain region, followed by Tukey's post hoc testing for multiple comparisons. Because regional differences were not the biological focus, post hoc testing and reporting were restricted to within-region exposure comparisons; cross-region comparisons were not performed or reported. For WS analyses, comparisons were performed between WS exposed and FA groups using two-tailed unpaired t-tests. Data are presented as mean ± SEM unless otherwise indicated.

3. Results

3.1. Chemically distinct DEP and WTC particles converge on enrichment of redox-active metals

Comprehensive chemical profiling by ion chromatography, thermal-optical analysis, and inductively coupled plasma mass spectrometry (ICP-MS) revealed that DEP and WTC dust exhibit fundamentally distinct ionic, carbonaceous, and size characteristics, but converge on enrichment of redox-active transition metals. DEP was enriched in NH₄⁺ and SO₄²⁻ and contained high elemental and organic carbon content, consistent with combustion-derived exhaust, whereas WTC dust showed elevated levels of soluble Cl⁻, Na⁺, and K⁺, consistent with pulverized building materials (SFig. 1A and B).

Furthermore, PM, the major component of AirP, exists in three categories: PM₁₀ (coarse, <10 μm), PM_2.5 (fine, <2.5 μm), and PM_0.1 (ultrafine, <1 μm). PM_2.5 and PM_0.1 exhibit heightened cytotoxicity due to their deeper penetrance into the lungs and nasal passageways with PM_0.1 most easily passing through alveolar epithelium [63]. DEP particles are predominantly ultrafine (PM_0.1-PM_2.5), while WTC is largely coarse (PM₁₀) [52,64]. Despite these differences, both DEP and WTC contained abundant transition metals, including Fe, Cu, Zn, Mn, and post-transition metal Pb (SFig. 1C), that are capable of catalyzing lipid peroxidation and implicated in amyloidogenic processing. Thus, while DEP and WTC dust differ fundamentally in their ionic and carbonaceous profiles, their shared metal enrichment provides the most plausible mechanistic link between these distinct exposures and AD pathology. In contrast to DEP and WTC, WS contained at least 1000-fold lower metal content (SFig. 1D). Here, we compared the DEP and WTC exposures to acute WS exposure, which was performed separately and at a higher concentration, exclusively in female mice. Due to this asymmetry, WS results are presented in the supplemental figures.

3.2. Acute DEP and WTC exposure elicits convergent cortical transcriptomic responses in mice despite chemical heterogeneity

Acute exposure to DEP or WTC induced robust transcriptional responses in the cortex, a region moderately afflicted by AD neuropathology. DEP exposure altered over 4000 genes, while WTC exposure altered approximately 1800 genes, with only ∼200 genes differing between the two exposures, suggesting high overlap in their transcriptomic response (Fig. 1A–C). GSEA revealed shared upregulation of oxidative stress, IFNα, and DNA repair pathways, alongside downregulation of protein secretion and UV response pathways compared to filtered air (Fig. 1D and E).

Fig. 1.

Acute 5-h exposure to DEP or WTC elicits highly conserved cortical transcriptomic responses in mice. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³, followed by bulk RNA sequencing of the cerebral cortex. Volcano plots show differential gene expression following A) DEP versus FA, B) WTC versus FA, and C) WTC versus DEP exposure. Gene set enrichment analysis (GSEA) identifies pathways altered following D) DEP versus FA, E) WTC versus FA, and F) WTC versus DEP exposure. G) Rank-rank plot comparing gene-level expression changes following DEP and WTC exposure across the transcriptome. H) Rank-rank hexbin plot showing the density and concordance of gene-level responses between DEP and WTC. I) UpSet plot showing numbers of differentially expressed genes (DEGs) unique to each exposure or shared between DEP and WTC, including concordant and discordant directional responses. Differential expression was determined using DESeq2 with sex included as a covariate and Benjamini-Hochberg false discovery rate correction (FDR<0.05).

DEP exposure preferentially enhanced NF-κB signaling, whereas WTC exposure was associated with increased heme metabolism, suggestive of cerebral microhemorrhages (Fig. 1F). Importantly, Ingenuity Pathway Analysis predicted activation of amyloid precursor protein (APP) and microtubule associated protein tau (MAPT) regulatory networks following both exposures, directly linking these convergent transcriptional responses to AD-relevant pathways.

Rank-rank correlation analysis demonstrated a strong concordance between DEP- and WTC-induced gene expression changes (r = 0.73; Fig. 1G), indicating that despite stark differences in chemical composition and particle size, both exposures elicit common gene signatures. Hexbin visualization further confirmed extensive overlap among all genes across both exposures (Fig. 1H). Consistent with these findings, upset plot analysis demonstrated that approximately 1200 genes were regulated in the same direction by both exposures (intersection 1-2), whereas only 47 genes showed discordant regulation (intersection 1-3; Fig. 1I). DEG analysis was split by sex in Supplemental Fig. 2. Collectively, these findings demonstrate that chemically distinct, but metal-rich AirP sources converge on a shared cortical stress transcriptome, characterized by oxidative stress, interferon signaling, and altered protein trafficking.

3.3. Acute WS exposure elicits a distinct cortical transcriptomic response in mice compared with DEP and WTC

In contrast to DEP and WTC, WS exposure did not share overlapping top differentially expressed genes (DEGs) with either metal-rich exposure (SFig. 3A). Although WS modestly engaged IFNα signaling, it uniquely enriched IFNγ signaling, fatty acid metabolism, and unfolded protein response pathways (SFig. 3B). Rank-rank correlation and hexbin analyses revealed no meaningful concordance between WS and either DEP or WTC (SFig. 3C–F). Upset plots of the three exposures showed few DEGs with high discordance (Intersection 1-3, SFig. 3G). These data indicate that the convergent transcriptomic response observed following DEP and WTC exposure is not a generic consequence of particulate exposure, but rather reflects a shared biological response associated with metal-rich AirP.

3.4. Acute DEP and WTC exposure induces ferroptosis-associated priming through increased lipid peroxidation and impaired antioxidant defense in mouse cortex

Given the shared oxidative stress signature induced by DEP and WTC, we examined whether these transcriptomic changes translated into ferroptosis-relevant biochemical alterations, particularly in the antioxidant systems responsible for detoxifying lipid peroxides and maintaining glutathione (GSH) homeostasis (Fig. 2A). Transcriptomic analyses revealed coordinated dysregulation of ferroptosis-associated genes, including reduced expression of glutathione peroxidase 4 (GPx4) and glutathione synthetase (GSS), alongside increased expression of pro-ferroptotic mediators such as PTGS2 and ACSL4 (Fig. 2B). In parallel, the iron-responsive surface receptor CD44 and the cystine importer SLC7A11, critical for cystine uptake and GSH synthesis, were upregulated, consistent with a compensatory response to oxidative stress and increased iron presence (Fig. 2B). Consistent with these changes, lipid peroxidation increased following both exposures, with HNE levels rising by at least 30% following DEP and WTC exposure (Fig. 2C). In contrast, protein nitration product, nitrotyrosine (NT) levels, and HNE assessed by immunofluorescence (IF) remained unchanged at this acute time point (SFig. 4A and B), indicating early lipid-selective oxidative injury.

Fig. 2.

Acute DEP and WTC exposure induces ferroptosis-associated changes and impairs antioxidant defense in mouse cortex. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³. A) Schematic representation of glutathione-dependent and glutathione-independent antioxidant pathways regulating lipid peroxide detoxification and ferroptosis-associated vulnerability. B) Heatmap of differentially expressed genes (DEGs) associated with ferroptosis following DEP and WTC exposure in the frontal cortex. Cortical levels of C) HNE, D) GPx4, E) Prdx6, F) PCOOH activity, G) GPx1, H) total GPx activity, I) FSP1, J) SLC7A11, K) GCLM, L) GCLC, M) GSS, N) GSTA4, O) ALDH2, and P) ACSL4 measured in whole-cell cortical lysates by Western blot, dot blot, or enzymatic activity assay as indicated. Statistical analysis was performed using ANCOVA with sex included as a covariate followed by Bonferroni-adjusted pairwise comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

We found GPx4 protein abundance to increase with both DEP and WTC (Fig. 2D), while Prdx6 and GPx1 protein levels remained unchanged (Fig. 2E–G). Despite preserved or increased abundance of these antioxidant proteins, functional assays revealed a 50% reduction in phospholipid hydroperoxide detoxification capacity following both DEP and WTC exposure (Fig. 2F). This was accompanied by reduced total GPx activity (Fig. 2H), indicating deficits in lipid peroxide detoxification and global peroxide handling. To further evaluate ferroptosis-relevant antioxidant systems, we examined ferroptosis suppressor protein 1 (FSP1), which limits lipid peroxidation through quinone reduction. FSP1 protein levels were reduced by 35% with DEP relative to FA and by 25% relative to DEP exposure following WTC exposure (Fig. 2I). These deficits occurred alongside impaired glutathione synthesis, as GSS protein levels declined by greater than 60% despite compensatory increases in cystine transport and upstream GSH pathway components (Fig. 2J–M).

Proteins involved in the detoxification of lipid-derived aldehydes, including glutathione S-transferase A4 (GSTA4) and aldehyde dehydrogenase 2 (ALDH2), were unchanged following DEP and WTC exposure (Fig. 2N and O). Notably, ACSL4 protein levels declined by at least 40% following both exposures (Fig. 2P), suggesting disruption of phospholipid remodeling pathways that influence ferroptotic susceptibility. Collectively, these findings indicate that acute DEP and WTC exposure induces a state of ferroptotic priming characterized by lipid peroxide accumulation, impaired detoxification capacity, and compromised glutathione synthesis, rather than overt ferroptotic cell death.

3.5. Acute WS exposure alters peroxide-handling capacity without inducing coordinated ferroptosis-associated priming in mouse cortex

In contrast to DEP and WTC, WS exposure did not increase levels of HNE or NT (SFig. 4C and D), indicating an absence of overt lipid or protein oxidative damage at this time point. GPx4, Prdx6, and GPx1 protein levels were unchanged with WS exposure (SFig. 4E,F,H). Although phospholipid peroxide detoxification activity was reduced by 80% following WS exposure (SFig. 4G), total GPx activity increased by 30% (SFig. 4I), suggesting compensatory engagement of non-lipid peroxide detoxification pathways. WS exposure did not alter most ferroptosis-related proteins aside from a reduction in ALDH2 (SFig. 4J–P). These findings indicate that WS exposure lacks the coordinated ferroptotic signature observed following metal-rich AirP.

3.6. Acute AirP exposure differentially alters cortical copper-handling pathways in mice

Given emerging links between copper dysregulation, oxidative stress, and neurodegeneration, we examined copper handling pathways following acute AirP exposure. Copper uptake is mediated by the transporter CTR1 and buffered by metallothioneins, while cytosolic chaperones such as copper chaperone for superoxide dismutase (CCS) and ATOX1 deliver copper to antioxidant and mitochondrial targets, including cytochrome c oxidase via COX17 and COA6 (SFig. 4Q). At the transcriptional level, multiple copper-handling genes were downregulated following DEP exposure and showed similar directional trends following WTC exposure, whereas WS exposure produced opposing expression patterns (SFig. 4R). These findings suggest selective remodeling of copper buffering and trafficking pathways by metal-rich AirP.

At the protein level, CCS abundance was reduced by at least 35% following WTC exposure relative to both filtered air and DEP (SFig. 4S). In contrast, SOD1 protein abundance and total superoxide dismutase activity were unchanged following DEP or WTC exposure (SFig. 4T and U), indicating preserved basal copper-dependent antioxidant capacity. WS exposure did not alter SOD1 protein levels but reduced total SOD activity by approximately 50% (SFig. 4V and W), consistent with functional impairment independent of SOD1 abundance. Together, these data indicate that acute metal-rich AirP exposure perturbs copper trafficking and buffering without engaging overt cuproptosis or SOD1-dependent oxidative collapse, whereas WS disrupts copper-dependent antioxidant activity through a distinct mechanism.

3.7. Acute DEP exposure induces cortical DNA oxidation and DNA damage-response pathways

To determine whether oxidative stress induced by metal-rich AirP extended beyond lipid damage, we examined markers of nucleic acid integrity and transcript surveillance. Transcriptomic analysis revealed coordinated regulation of genes involved in DNA damage signaling and repair following DEP exposure, with similar but weaker trends observed following WTC exposure (SFig. 5A). These included regulators of DNA damage sensing (ATM, CHEK1, GADD45A), base excision repair (APEX1, NEIL1, POLB), and nucleotide excision repair (XPA, ERCC6L2), consistent with activation of oxidative DNA damage response pathways.

Consistent with these transcriptional changes, IF revealed increased cortical levels of 8-hydroxy-2′-deoxyguanosine (8-OHdG) following DEP exposure, whereas WTC exposure did not significantly alter 8-OHdG at this acute time point (SFig. 5B). These findings indicate that DEP induces greater nucleic acid oxidation than WTC, paralleling the more pronounced lipid peroxidation and antioxidant impairment observed following DEP exposure. Collectively, these findings demonstrate that acute DEP exposure induces DNA oxidation and engages DNA repair pathways, consistent with secondary nucleic acid stress arising downstream of oxidative injury.

3.8. Acute DEP and WTC exposure alters iron handling and heme metabolism without increasing total brain iron

To determine whether the ferroptosis-relevant oxidative imbalance observed following DEP and WTC exposure was associated with changes in brain iron availability or handling, we examined proteins involved in iron transport, storage, export, and heme metabolism (Fig. 3A). Iron primarily enters the cell bound to transferrin (TF) via the transferrin receptor (TfR) as Fe³⁺ or through divalent metal transporter 1 (DMT1) as Fe²⁺. Intracellular iron is stored within ferritin complexes composed of ferritin heavy (FTH1) and light (FTL) chains, while iron export occurs exclusively through ferroportin (FPN) or through NCOA4-mediated ferritinophagy.

Fig. 3.

Acute DEP and WTC exposure alters iron handling and heme metabolism in mouse cortex. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³. A) Schematic representation of cellular iron uptake, storage, export, and heme metabolism pathways. Cortical protein levels or biochemical measurements of B) TF, C) TfR, D) DMT1, E) FTL, F) FTH1, G) tissue heme, H) HCP1, I) HMOX1, J) FPN, K) IRP1, and L) IRP2 measured by Western blot or biochemical assay as indicated. Statistical analysis was performed using ANCOVA with sex included as a covariate followed by Bonferroni-adjusted pairwise comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Despite evidence of ferroptotic priming, MRI-based measurements revealed no changes in total brain iron content following DEP or WTC exposure (SFig. 6A and B), indicating that acute exposure does not produce bulk iron accumulation. Instead, both exposures altered iron handling pathways. DEP reduced TF and TfR abundance, whereas WTC primarily impaired ferritin-based iron storage, with pronounced reductions in FTH1 and FTL (Fig. 3B–F). DMT1 and FPN1 protein abundance were unchanged following either DEP or WTC exposure (Fig. 3D–G), suggesting selective suppression of transferrin-dependent iron uptake rather than generalized iron import or export. Overall, these changes indicate impaired ferritin-based iron sequestration, particularly following WTC exposure.

Consistent with altered iron storage, tissue heme levels increased by 25% following DEP exposure and trended upward with WTC exposure compared to FA (Fig. 3H). Despite increased heme content, expression of the heme importer HCP1 was unchanged (Fig. 3I). In contrast, heme oxygenase 1 (HMOX1), which catalyzes heme degradation and liberates Fe²⁺, increased two-fold following WTC exposure (Fig. 3J), suggesting enhanced heme turnover and potential microvascular stress. In contrast, canonical iron-sensing pathways remained intact (Fig. 3K and L). These data indicate that acute metal-rich AirP redistributes iron toward bioactive pools without increasing bulk iron accumulation.

3.9. Acute WS exposure produces limited alterations in cortical iron-handling pathways

Consistent with its low metal content, WS exposure did not induce iron redistribution or heme metabolism changes observed with metal-rich AirP. Instead, modest alterations in iron regulatory proteins were observed without changes in iron storage or export (SFig. 6C–M). Collectively, these findings indicate that WS exposure does not induce the iron redistribution and heme-associated responses observed with the metal-rich DEP and WTC.

3.10. Acute DEP and WTC exposure enhances amyloidogenic processing and impairs lipid raft antioxidant defenses in mouse cortex

We previously identified LRs as hotspots for lipid peroxidation [27,28], which may enhance amyloid processing [36,65] (Fig. 4A). Given the sensitivity of lipid rafts (LRs) to lipid peroxidation and their central role in APP processing, we examined amyloid-related pathways. DEP exposure broadly upregulated transcripts involved in APP processing (ADAM10, BACE1, PSEN1, APH1B), endosomal trafficking (VPS35, VPS26B, SNX27, SORT1), and tau kinase signaling (CDK5R1, GSK3β, MARK1, TTBK2), whereas WTC produced a more restricted transcriptional response. Notably, ABCA7 was the only transcript significantly downregulated by both exposures (Fig. 4B). These findings highlight impaired lipid and amyloid handling as a shared vulnerability, while uniquely engaging DEP in broader amyloidogenic and tau-related genes.

Fig. 4.

Acute DEP and WTC exposure enhances amyloidogenic processing and impairs lipid raft antioxidant defenses. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³. A) Schematic representation integrating lipid raft antioxidant defense, oxidative damage, and APP processing. B) Heatmap of differentially expressed genes (DEGs) associated with AD neuropathology and amyloid processing following DEP and WTC exposure. Lipid raft protein abundance, enzymatic activity, and oxidative damage were assessed by Western blot, dot blot, or biochemical assay, with soluble Aβ peptides measured in whole-cell cortical lysates: C) APP, D) ADAM10, E) BACE1, F) PSEN1, G) ApoE, H) LRP1, I) Aβ40, J) Aβ42, K) FSP1, L) GPx1, M) GPx1 activity, N) GPx4, O) PCOOH activity, P) ALDH2, Q) HNE, and R) NT. Statistical analysis was performed using ANCOVA with sex included as a covariate followed by Bonferroni-adjusted pairwise comparisons. *p < 0.05, **p < 0.01, ***p < 0.001.

Consistent with these transcriptional changes, protein-level analysis demonstrated exposure-specific alterations in APP processing components. LR APP protein levels increased by 60% following WTC exposure but were unchanged following DEP exposure (Fig. 4C). The α-secretase ADAM10 increased two-fold following WTC exposure and trended upward with DEP (Fig. 4D). In contrast, the β-secretase BACE1 increased two-fold with DEP exposure and three-fold with WTC exposure (Fig. 4E). The catalytic γ-secretase subunit presenilin-1 (PSEN1) decreased by 40% following DEP exposure but increased by 40% following WTC exposure (Fig. 4F), indicating divergent regulation of secretase components, but converged on increased amyloidogenic output.

Apolipoprotein E (ApoE), which binds amyloid peptides and facilitates their clearance, increased two-fold following both DEP and WTC exposure (Fig. 4G). LR protein of the ApoE receptor, LRP1, increased by 65% following WTC but was unchanged following DEP (Fig. 4H). Measurement of soluble amyloid peptides in whole cell lysates revealed trends of increase in Aβ40 following both exposures, whereas the more aggregation-prone Aβ42 increased by 35% following DEP exposure and by 70% following WTC exposure (Fig. 4I and J), indicating a shift in amyloidogenic production.

To determine whether altered amyloid processing occurred in parallel with compromised LR antioxidant defenses, we examined LR-specific protective mechanisms and damage. FSP1, increased four-fold following DEP and three-fold following WTC exposures (Fig. 4K), consistent with a compensatory response to oxidative stress. Within lipid rafts, GPx1 protein abundance was unchanged, but GPx1 enzymatic activity declined by 40% following both DEP and WTC exposures (Fig. 4L and M). GPx4 protein trended downward in LRs following both exposures, while phospholipid hydroperoxide (PCOOH) reduction capacity decreased by at least 70% (Fig. 4N and O), indicating impairment of lipid peroxide detoxification within the LR. In parallel, LR ALDH2 levels decreased by at least 55% following both DEP and WTC exposure, while LR HNE, but not LR NT, increased by 40% following WTC exposure and trended upward with DEP (Fig. 4P–R). Collectively, these findings demonstrate that acute exposure to metal-rich AirP enhances amyloidogenic processing and promotes accumulation of the aggregate-prone Aβ42 while simultaneously impairing lipid raft antioxidant defenses, recapitulating biochemical features observed in postmortem Alzheimer's disease brain tissue [27,28,55].

3.11. Acute WS exposure alters cortical amyloid peptide profiles without reproducing the DEP/WTC amyloidogenic signature

To determine whether WS exposure influenced amyloid pathways, we examined APP processing and amyloid peptide production in whole cell lysates. In contrast to DEP and WTC, WS exposure reduced APP protein abundance by 40% (SFig. 7A). WS did not alter ADAM10 or BACE1 levels, but reduced PSEN1 levels by 50% (SFig. 7B–D). Despite these changes, WS exposure increased soluble Aβ40 by 85%, without altering Aβ42 levels (SFig. 7E and F), indicating a shift toward production of the less aggregation-prone amyloid species. ApoE levels increased by 40%, while LRP1 abundance was unchanged (SFig. 7G and H). These data demonstrate that WS exposure alters amyloid peptide balance without engaging the amyloidogenic machinery or lipid raft dysfunction observed following metal-rich AirP, further reinforcing mechanistic divergence between WS and DEP/WTC.

3.12. Acute DEP and WTC exposure engages cortical xenobiotic, redox, and stress-adaptive transcriptional responses

To define regulatory programs engaged by acute air pollution exposure, we examined transcription factors associated with xenobiotic metabolism, oxidative stress, and inflammatory signaling (Fig. 5A). Transcriptomic analysis revealed increased expression of MEF2C, a regulator of neuronal development and synaptic function, following both DEP and WTC exposure, alongside DEP-specific induction of the aryl hydrocarbon receptor (AhR) and its neuronal binding partner ARNT2 (Fig. 5B), which mediate transcriptional responses to polycyclic aromatic hydrocarbons (PAHs) and related xenobiotics. In parallel, transcripts encoding the AhR repressor (AhRR), ARNT1, and ATF4 were reduced, consistent with engagement of xenobiotic-responsive transcriptional circuits (Fig. 5B).

Fig. 5.

Acute DEP and WTC exposure alters nuclear stress-responsive transcriptional programs in mouse cortex. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³. A) Schematic representation of nuclear stress-sensing pathways linking xenobiotic exposure, oxidative stress, and inflammatory signaling. B) Heatmap of differentially expressed genes (DEGs) encoding transcription factors following DEP and WTC exposure. Western blot analysis of cortical nuclear lysates for C) AhR, D) ARNT1, E) ARNT2, F) NF-κB, G) Nrf2, H) ATF4, and I) NCOA4. J) DoRothEA-inferred transcription factor activity derived from RNA-seq gene expression profiles following DEP and WTC exposure. Statistical analysis was performed using ANCOVA with sex included as a covariate followed by Bonferroni-adjusted pairwise comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Assessment of nuclear protein abundance revealed partial divergence from transcript-level changes, consistent with the temporal dynamics of transcription factor activation. Nuclear AhR protein levels were reduced following both DEP and WTC exposure, while ARNT1 abundance was unchanged (Fig. 5C and D). Neuronal ARNT2 levels declined following both exposures, whereas nuclear NCOA4 increased selectively following DEP exposure (Fig. 5E and F), suggesting exposure-specific modulation of xenobiotic and iron transcriptional responses.

To evaluate redox-associated transcriptional control, we assessed BACH1 and Nrf2, which competitively regulate antioxidant and iron metabolism genes. BACH1 protein levels declined by at least 50% following both DEP and WTC exposure (Fig. 5G), consistent with elevated heme levels and relief of BACH1-mediated repression. In contrast, nuclear Nrf2 levels declined by 45% following DEP exposure and 60% following WTC exposure, while ATF4 protein abundance remained unchanged (Fig. 5H and I), suggesting redistribution rather than amplification of canonical antioxidant signaling. Nuclear NF-κB p65 levels were also significantly reduced following DEP and WTC exposure (Fig. 5J), indicating selective remodeling rather than global inflammatory activation.

Because transcription factor activity is not fully captured by steady-state nuclear abundance, we inferred regulatory activity using DoRothEA based on validated target gene expression (Fig. 5K). This analysis revealed increased activity of AhR pathway components, including AhR itself, following DEP exposure, and its obligate partners ARNT and ARNT2 following both DEP and WTC exposure. In parallel, activity of small Maf proteins (MAFF, MAFK, MAFG), which heterodimerize with Nrf2 to regulate antioxidant and detoxification genes, and ATF4 were increased, despite reduced Nrf2 (NFE2L2) activity. AirP exposure also selectively remodeled inflammatory and stress signaling pathways. STAT3 activity increased, whereas RELA (NF-κB p65) and STAT1 activity declined, indicating noncanonical inflammatory regulation. Transcription factors associated with cellular resilience and metabolic adaptation, including FOXO3, HIF1A, and SP1, were reduced following DEP and WTC exposure. Together, these data indicate that acute metal-rich AirP exposure engages coordinated xenobiotic metabolism, redox modulation, and stress-adaptive transcriptional programs without global inflammatory activation, consistent with an early adaptive response to oxidative challenge.

3.13. Acute WS exposure induces a limited and divergent cortical transcription factor response

In contrast to DEP and WTC, WS exposure elicited a more restricted and divergent transcription factor response. Nuclear AhR levels were unchanged, while ARNT1 and ARNT2 abundance declined (SFig. 8A–C). NCOA4 increased 2.5-fold following WS exposure (SFig. 8D), suggesting limited engagement of xenobiotic response signaling. Unlike metal-rich AirP, WS increased nuclear BACH1 levels while reducing Nrf2 (SFig. 8E and F), indicating a shift toward transcriptional repression of antioxidant and iron-handling genes. ATF4 protein abundance increased, while NF-κB p65 levels were unchanged (SFig. 8G and H). Consistent with these findings, WS exposure produced minimal transcription factor activity changes by DoRothEA analysis, further supporting the absence of coordinated xenobiotic or redox regulatory engagement.

3.14. Acute DEP and WTC exposure converges on neuronal and oligodendrocyte-associated transcriptional responses in mouse cortex

To determine whether bulk transcriptomic changes reflected intrinsic transcriptional regulation or shifts in cellular composition, cell-type deconvolution was performed on our RNA-seq dataset. Both DEP and WTC exposures were associated with a reduction in oligodendrocyte proportions and a reciprocal increase in neuronal proportions relative to FA (Fig. 6A and B). Because changes in cellular composition can mask or distort DEGs, we next adjusted the bulk RNA-seq analysis for oligodendrocyte and neuronal proportions. After adjusting for cellular composition, additional ferroptosis-, lipid remodeling-, and vascular-associated genes emerged (Fig. 6C–K) in DEP-exposed brains. In contrast, WTC exposure revealed population-adjusted changes associated with heme metabolism and synaptic modulation (Fig. 6L and M).

Fig. 6.

Acute DEP and WTC exposures converge on neuronal- and oligodendrocyte-associated transcriptional responses in mouse cortex. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³. Cell-type deconvolution of cortical bulk RNA-seq showing estimated A) oligodendrocyte and B) neuronal proportions following acute AirP exposure. Differential expression was subsequently reanalyzed after adjustment for the estimated oligodendrocyte and neuronal proportions, with volcano plots showing C) DEP versus FA adjusted for oligodendrocytes, D) DEP versus FA adjusted for neurons, E) WTC versus FA adjusted for oligodendrocytes, F) WTC versus FA adjusted for neurons, G) WTC versus DEP adjusted for oligodendrocytes, and H) WTC versus DEP adjusted for neurons. Venn diagrams compare DEGs identified in the unadjusted bulk RNA-seq analysis with those identified following oligodendrocyte- and neuron-adjustment for I) DEP versus FA, J) WTC versus FA, and K) WTC versus DEP. Heatmaps show selected population-adjusted DEGs associated with L) ferroptosis, lipid remodeling, and perivascular remodeling following DEP exposure and M) heme metabolism and synaptic modulation following WTC exposure. N) DoRothEA-inferred activity of neuronal- and oligodendrocyte-associated transcription factors. Western blot analysis of cortical nuclear lysates for O) Olig1 and P) NeuroD1. Statistical analysis of estimated cell-type proportions was performed using Kruskal-Wallis testing for A,B, and protein measurements were analyzed using ANCOVA adjusted for sex with Bonferroni's post hoc test for O,P. **p < 0.01.

To further assess whether these changes reflected engagement of lineage-associated transcriptional responses rather than cell loss or gain alone, transcription factor activity was again inferred using DoRothEA. Transcription factor inference revealed increased activity of oligodendrocyte lineage regulators and reduced neuronal transcriptional programs, without evidence of overt cell loss (Fig. 6N). At the protein level, nuclear Olig1 abundance was unchanged following DEP or WTC exposure, whereas NeuroD1 protein levels were reduced by 45% following WTC exposure, mirroring the DoRothEA activity (Fig. 6O and P). Consistent with the absence of overt cell loss, levels of myelin basic protein (MBP) and the neuronal marker NeuN were unchanged following DEP and WTC exposure (SFig. 9A and B). Direct quantification of oligodendrocyte lineage cells (Olig2+) and oligodendrocyte precursor cells (Olig2+PDGFRα+) in the corpus callosum by immunofluorescence similarly revealed no change in cell number following DEP or WTC exposure (SFig. 9C and D), indicating that the deconvolution-based shift in oligodendrocyte transcriptional signature reflects altered transcriptional state rather than a change in cell number.

3.15. Acute WS exposure induces neuronal and oligodendrocyte-associated transcriptional responses distinct from metal-rich AirP

Deconvolution analysis of WS exposure revealed a contrasting pattern, characterized by increased oligodendrocyte proportions and reduced neuronal proportions (SFig. 9E and F). MBP protein levels were unchanged, while NeuN levels decreased by 40% following WS exposure (SFig. 9G and H). In contrast to DEP and WTC, WS exposure produced distinct transcription factor responses, with increased activity of oligodendrocyte-associated transcription factors including TCF12, PRDM14, SOX11, KLF9, MEIS1, and EBF1, accompanied by reduced activity of the neuronal transcription factor FOXP2 (SFig. 9I). NeuroD1 was not identified in DoRothEA analysis and was unchanged at the protein level following WS exposure (SFig. 9J). Collectively, these findings demonstrate that DEP and WTC exposure converge on shared neuronal and oligodendrocyte transcriptional responses that persist after accounting for cellular composition, whereas WS exposure induces a divergent cell-state response characterized by oligodendrocyte enrichment and reduced neuronal representation.

3.16. Acute DEP and WTC exposure alters mouse white-matter microstructure, with DEP inducing oxidative injury in the corpus callosum

To determine whether acute AirP exposure alters brain microstructure, ex vivo diffusion tensor imaging (DTI) was performed across multiple brain regions. DTI revealed increased fractional anisotropy (Fig. 7A) and reduced diffusivity (Fig. 7B) in the corpus callosum following DEP and WTC exposure, and these findings were confined to white matter. Other brain regions exhibited minimal or no changes in these measurements. These observations are consistent with acute alterations in white-matter microstructure, suggestive of increased directional restriction of water diffusion, potentially reflecting cellular or myelin-associated swelling rather than tissue loss.

Fig. 7.

Acute DEP and WTC exposure alters white-matter microstructure in the corpus callosum. Male and female C57BL/6 mice were exposed to filtered air (FA), diesel exhaust particles (DEP), or World Trade Center (WTC) dust for 5h at 100 μg/m³. Ex vivo diffusion tensor imaging (DTI) measurements of A) fractional anisotropy and B) apparent diffusion coefficient (ADC) across individual brain regions and grouped white- and grey-matter regions. Immunofluorescence analysis of the corpus callosum for C) HNE and D) 8-OHdG. E) Representative 40x immunofluorescence images of HNE and 8-OHdG in the corpus callosum; scale bar, 10 μm. Statistical analysis was performed using two-way ANOVA followed by Tukey's post hoc test for A,B, and ANCOVA with sex included as a covariate followed by Bonferroni-adjusted pairwise comparisons for C,D. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

DEP exposure uniquely increased lipid peroxidation and DNA damage within the corpus callosum (Fig. 7C and D), whereas WTC altered diffusion properties without overt oxidative injury. These findings demonstrate that acute metal-rich AirP rapidly perturbs white-matter integrity, with DEP producing the strongest oxidative injury. Collectively, these data demonstrate that acute AirP exposure rapidly perturbs white-matter microstructure, with the corpus callosum exhibiting the greatest sensitivity. The dissociation between diffusion abnormalities and oxidative damage across exposures suggests that microstructural alterations can arise in the absence of overt lipid or DNA oxidation.

3.17. Acute WS exposure produces minimal oxidative or inflammatory injury in the mouse corpus callosum and hippocampus

To determine whether acute WS exposure elicited oxidative or inflammatory injury within white matter or hippocampal regions, IF was performed in the corpus callosum and across multiple hippocampal subregions. In the corpus callosum, WS exposure did not alter levels of HNE, 8-OHdG, degraded myelin basic protein (dMBP), the complement component C5 or its activation fragment C5a, or the microglial marker Iba1 (SFig. 10A–F). Similarly, WS exposure did not alter HNE, 8-OHdG, dMBP, C5, C5a, or Iba1 immunoreactivity within hippocampal subregions, aside from increased dMBP by 50% and C5 by 60% in the radiatum (SFig. 10G–L). Together, these findings indicate that WS exposure induces a distinct and attenuated neurobiological response characterized by altered amyloid balance and cell-state remodeling without overt oxidative or inflammatory injury.

4. Discussion

AirP exposure is increasingly recognized as a modifiable risk factor for AD, yet the biological mechanisms linking heterogeneous airborne pollutants to neurodegenerative pathology have remained unclear. Here, we demonstrate that acute exposure to two chemically distinct but metal-rich forms of AirP, DEP and WTC dust, elicit a highly conserved molecular response characterized by ferroptotic priming, impaired lipid peroxide detoxification, enhanced amyloidogenic processing, and selective white-matter vulnerability. Despite major differences in particle size, carbon content, and ionic composition, DEP and WTC converge on shared transcriptional, biochemical, and microstructural changes relevant to AD. In contrast, WS, produced by the incomplete combustion of wood with a substantially lower metal burden, induces a divergent and attenuated response that does not recapitulate these convergent features. To capture these primary, early molecular events, we used young, wild-type C57BL/6J mice, providing a clean genetic and physiological background free of the age-associated oxidative stress, white-matter change, ferroptotic vulnerability, and amyloid dysregulation that could otherwise confound interpretation of exposure-driven effects. Together, these findings identify metal-associated oxidative mechanisms as a unifying link between specific forms of AirP and AD-relevant pathology.

A central insight from this study is that chemical heterogeneity does not preclude biological convergence. DEP and WTC dust differ profoundly in origin and composition, with DEP representing PM_0.1, carbon-rich traffic-related pollution, and WTC dust representing PM₁₀, mineral-dominated particulate matter. Yet both exposures robustly activate overlapping transcriptomic responses. Approximately 1200 genes were regulated in the same direction following DEP and WTC exposure, with strong rank-rank correlation across the transcriptome. These shared responses encompassed pathways related to oxidative stress, interferon signaling, DNA repair, and protein trafficking, suggesting engagement of a conserved stress-adaptive response to metal-rich AirP. The absence of comparable overlap with WS indicates that this response is not a generic consequence of particulate inhalation and is consistent with a shared biological response associated with metal content and redox activity.

Biochemically, DEP and WTC exposures produced a pattern consistent with ferroptotic priming rather than overt ferroptotic cell death. Both exposures increased lipid peroxidation while simultaneously impairing detoxification of oxidized phospholipids despite increased GPx4 and unchanged Prdx6 protein. Notably, this increase in whole-cell GPx4 was not reflected within lipid rafts, where GPx4 instead trended downward in parallel with a marked reduction in PCOOH detoxification capacity. This compartment-specific dissociation parallels our previous findings in human AD brain, in which increased whole-cell GPx4 was accompanied by reduced lipid raft GPx4 and reduced phospholipid hydroperoxide-reducing activity [27]. Thus, increased total GPx4 abundance does not necessarily indicate preserved antioxidant capacity within oxidation-sensitive membrane compartments. Additional mechanisms may include post-translational modification or oxidation of the enzymes themselves [66], altered membrane localization, changes in endogenous reducing substrates or cofactors, or other alterations in catalytic function. The present experiments cannot distinguish among these possibilities, and the precise molecular basis of the protein-function dissociation remains unresolved. Together, these findings are consistent with a ferroptosis-vulnerable state and closely parallel alterations observed in postmortem AD brain tissue, including impaired PCOOH reduction and increased lipid peroxidation [27,28,33]. Concurrent reduction in FSP1 with WTC dust and disruption of lipid remodeling pathways further support the interpretation that acute metal-rich AirP exposure sensitizes neural membranes to oxidative injury.

Lipid rafts are a critical nexus linking ferroptotic events to amyloidogenic processing. These membrane microdomains are enriched in polyunsaturated lipids and harbor the enzymatic machinery responsible for APP cleavage [28,67,68]. Acute DEP and WTC exposure impaired lipid raft-specific antioxidant defenses, reduced phospholipid hydroperoxide detoxification capacity, and promoted accumulation of lipid peroxidation products. In parallel, both exposures increased amyloid-β production, with a preferential rise in the aggregation-prone Aβ42 species. These findings support previous cell culture experiments with TRAP [67] and a model in which metal-driven lipid peroxidation within lipid rafts alters membrane composition and secretase activity [36,37], thereby biasing APP processing toward amyloidogenic pathways. Notably, this constellation of lipid raft oxidative damage, impaired oxidized phospholipid reduction, and increased Aβ42 production closely parallels biochemical signatures observed in human AD cortex [27,28].

Iron handling and heme metabolism were selectively altered by DEP and WTC exposure, reinforcing the role of redistributed redox-active iron rather than bulk iron accumulation. Although total brain iron levels were unchanged acutely, both exposures disrupted ferritin storage capacity with increased tissue heme for DEP. Induction of HMOX1, particularly following WTC exposure, suggests enhanced heme turnover and liberation of redox-active iron, consistent with vascular stress and microhemorrhagic processes reported in AirP-exposed and AD populations [25,[69], [70], [71], [72]]. These findings align with neuropathological and imaging evidence linking cerebral microbleeds, iron redistribution, and AD progression [23,27,28,[73], [74], [75]], indicating that even transient perturbations in iron handling may be sufficient to initiate lipid peroxidation cascades.

The structural consequences of these molecular changes were evident in selective white-matter vulnerability, most prominently within the corpus callosum. Both DEP and WTC exposure altered DTI MRI metrics, characterized by increased fractional anisotropy and reduced mean diffusivity. This pattern is consistent with acute microstructural alterations such as myelin-associated swelling, altered axonal packing, or changes in extracellular water diffusion rather than overt tissue loss [76,77]. DEP uniquely induced lipid and DNA oxidation within the corpus callosum, whereas WTC altered diffusion properties in the absence of overt oxidative damage, suggesting that microstructural disruption may precede detectable biochemical injury. These findings are consistent with emerging human and experimental evidence identifying white matter as a sensitive target of AirP exposure. Recent human imaging studies have associated ambient AirP exposure with altered white-matter microstructure [69,70], while longitudinal exposure to PM2.5 in older adults was associated with progressive increases in white-matter hyperintensity burden [78]. Experimental studies similarly demonstrate that particulate AirP exposure can promote white-matter demyelination and exacerbate diffusion abnormalities and vascular injury in susceptible mice [79]. Together, these findings support white-matter vulnerability as a potentially important link between AirP exposure and later neurodegenerative risk [80].

In contrast to DEP and WTC, WS exposure produced a fundamentally different biological response. WS contained substantially lower metal content and did not induce coordinated ferroptotic priming, lipid peroxidation, or amyloidogenic bias. Although WS impaired phospholipid peroxide detoxification activity, this occurred in the absence of lipid peroxidation or ferroptosis-associated transcriptional responses. WS instead favored increased Aβ40 production without elevating Aβ42, consistent with a less pathogenic amyloid profile. These findings underscore that not all particulate exposures carry equivalent neurodegenerative responses and are consistent with metal burden contributing to the distinct pathogenic potential of different AirP sources.

Several limitations should be acknowledged. First, this study focused on acute exposures to capture early mechanistic responses rather than cumulative pathology. Behavioral assessments were intentionally not performed, as the acute exposure paradigm was designed to preserve post-exposure molecular, biochemical, and imaging endpoints that would be confounded by behavioral testing. Importantly, the acute exposure paradigm used here does not fully recapitulate the chronic, heterogeneous, and cumulative exposures experienced by humans, and the extent to which these early molecular responses persist or contribute to human air-pollution-associated neurodegeneration remains to be determined. While acute priming does not equate to disease, experimental evidence from independent exposure paradigms indicates that increasing duration or repetition of metal-rich air pollution exposure is associated with amplified lipid peroxidation, inflammatory activation, white-matter injury, and amyloid-related pathology. Together with the present findings, these data support a model in which repeated exposure to metal-rich AirP reinforces ferroptotic vulnerability, lipid raft oxidation, and amyloidogenic bias, thereby accelerating AD-associated pathology. Second, this study used young, wild-type mice; whether these molecular responses are amplified or more persistent in aged or transgenic AD-relevant models remains untested and represents an important future direction. Third, WS exposure was conducted at higher concentrations and in a single sex, reflecting real-world wildfire conditions but limiting direct quantitative comparison across exposures.

In summary, this study identifies metal-associated oxidative vulnerability as a convergent biological mechanism linking chemically distinct forms of AirP to AD-relevant pathology. Acute exposure to DEP and WTC dust primes the brain for ferroptotic stress, disrupts lipid raft antioxidant defenses, enhances amyloidogenic processing, and selectively alters white-matter microstructure, all of which are pathological features of AD. These effects were not reproduced by the independently conducted, metal-poor WS exposure, supporting the concept that particulate exposures can produce distinct neurobiological responses depending on their chemical composition. By delineating a shared, mechanistically coherent pathway connecting AirP to AD, these findings provide a framework for understanding environmental contributions to neurodegeneration and highlight metal handling and lipid peroxide detoxification as potential targets for intervention.

CRediT authorship contribution statement

Kristina Shkirkova: Writing – review & editing, Writing – original draft, Visualization, Supervision, Investigation, Formal analysis, Data curation, Conceptualization. Naomi S. Sta Maria: Writing – review & editing, Writing – original draft, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation. Herbert Anson: Writing – review & editing, Investigation, Formal analysis. Yashar Aghaei: Writing – review & editing, Investigation, Formal analysis. Mohammad Mahdi Badami: Writing – review & editing, Investigation, Formal analysis. Ararat Chakhoyan: Writing – review & editing, Investigation, Formal analysis. Jose A. Godoy-Lugo: Writing – review & editing, Investigation. Claire Chung: Writing – review & editing, Investigation, Formal analysis. Salma Durra: Writing – review & editing, Investigation, Formal analysis. Angela Tang-Tan: Writing – review & editing, Visualization, Investigation, Formal analysis. Lifu Zhao: Writing – review & editing, Investigation, Formal analysis. Alexandra Demetriou: Writing – review & editing, Investigation, Formal analysis. Manuel Morales: Writing – review & editing, Investigation, Formal analysis. Sindhu Daggupati: Writing – review & editing, Investigation, Formal analysis. Isabella Bent: Writing – review & editing, Investigation, Formal analysis. Hyoungjin Park: Writing – review & editing, Investigation, Formal analysis. Caleb Franklin: Writing – review & editing, Investigation, Formal analysis. Selena Chen: Writing – review & editing, Investigation, Formal analysis. Giovanni Chahine: Writing – review & editing, Investigation, Formal analysis. Skyye Dodds-Lewis: Writing – review & editing, Investigation, Formal analysis. Masako Morishita: Writing – review & editing, Investigation, Formal analysis. Russell E. Jacobs: Writing – review & editing, Supervision. Jean-François Gout: Writing – review & editing, Formal analysis. Wendy J. Mack: Writing – review & editing, Formal analysis. Henry Jay Forman: Writing – review & editing, Conceptualization. Bérénice A. Benayoun: Writing – review & editing, Formal analysis. Marc Vermulst: Writing – review & editing, Conceptualization. Mitchell D. Cohen: Writing – review & editing, Resources. Matthew Campen: Writing – review & editing, Resources, Methodology, Data curation. Constantinos Sioutas: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Conceptualization. William J. Mack: Writing – review & editing, Writing – original draft, Supervision, Resources, Project administration, Methodology, Conceptualization. Caleb E. Finch: Writing – review & editing, Writing – original draft, Supervision, Methodology, Funding acquisition, Conceptualization. Max A. Thorwald: Writing – review & editing, Writing – original draft, Visualization, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization.

Ethics declaration

This study was conducted in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. This study was approved by the University of Southern California Institutional Animal Care and Use Committee (USC IACUC); University of New Mexico Institutional Animal Care and Use Committee (UNM IACUC). (Approval No. USC IACUC #20842; UNM IACUC #23-201419-HSC),

Ethics declarations

WJM reports the following: consultant: Viseon, Imperative Care, Q'Apel, Medtronic, Stryker, Stream Biomedical, Spartan Micro; Egret. Investor: Cerebrotech, Q′ Apel, Endostream, Viseon, Rebound, Stream Biomedical, Spartan Micro, Radical Catheters, Vastrax, Borvo.

Funding

Lab studies were supported by NIH grants to CEF (R01-AG051521, P01-AG055367) and the Cure Alzheimer's Fund.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:Caleb E. Finch reports financial support was provided by National Institutes of Health. William J. Mack reports a relationship with Viseon that includes: consulting or advisory and equity or stocks. William J. Mack reports a relationship with Imperative Care that includes: consulting or advisory. William J. Mack reports a relationship with QApel that includes: consulting or advisory and equity or stocks. William J. Mack reports a relationship with Medtronic that includes: consulting or advisory. William J. Mack reports a relationship with Stryker that includes: consulting or advisory. William J. Mack reports a relationship with Stream Biomedical that includes: consulting or advisory and equity or stocks. William J. Mack reports a relationship with Spartan Micro that includes: consulting or advisory and equity or stocks. William J. Mack reports a relationship with Egret that includes: consulting or advisory. William J. Mack reports a relationship with Cerebrotech that includes: equity or stocks. William J. Mack reports a relationship with Endostream that includes: equity or stocks. William J. Mack reports a relationship with Rebound that includes: equity or stocks. William J. Mack reports a relationship with Radical Catheters that includes: equity or stocks. William J. Mack reports a relationship with Vastrax that includes: equity or stocks. William J. Mack reports a relationship with Borvo that includes: equity or stocks. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.