Peripheral HMGB1 is linked to O₃ pathology of disease‐associated astrocytes and amyloid

Abstract

INTRODUCTION

Ozone (O₃) is an air pollutant associated with Alzheimer's disease (AD) risk. The lung–brain axis is implicated in O₃‐associated glial and amyloid pathobiology; however, the role of disease‐associated astrocytes (DAAs) in this process remains unknown.

METHODS

The O₃‐induced astrocyte phenotype was characterized in 5xFAD mice by spatial transcriptomics and proteomics. Hmgb1 ^fl/fl LysM‐Cre⁺ mice were used to assess the role of peripheral myeloid cell high mobility group box 1 (HMGB1).

RESULTS

O₃ increased astrocyte and plaque numbers, impeded the astrocyte proteomic response to plaque deposition, augmented the DAA transcriptional fingerprint, increased astrocyte–microglia contact, and reduced bronchoalveolar lavage immune cell HMGB1 expression in 5xFAD mice. O₃‐exposed Hmgb1 ^fl/fl LysM‐Cre⁺ mice exhibited dysregulated DAA mRNA markers.

DISCUSSION

Astrocytes and peripheral myeloid cells are critical lung–brain axis interactors. HMGB1 loss in peripheral myeloid cells regulates the O₃‐induced DAA phenotype. These findings demonstrate a mechanism and potential intervention target for air pollution–induced AD pathobiology.

Highlights

• Astrocytes are part of the lung–brain axis, regulating how air pollution affects plaque pathology.

• Ozone (O₃) astrocyte effects are associated with increased plaques and modified by plaque localization.

• O₃ uniquely disrupts the astrocyte transcriptomic and proteomic disease‐associated astrocyte (DAA) phenotype in plaque associated astrocytes (PAA).

• O₃ changes the PAA cell contact with microglia and cell–cell communication gene expression.

• Peripheral myeloid cell high mobility group box 1 regulates O₃‐induced transcriptomic changes in the DAA phenotype.

1. BACKGROUND

Alzheimer's disease (AD) is the most common neurodegenerative disease and the leading cause of dementia; however, its etiology is largely unknown, and there are few disease‐modifying treatments. ¹ , ² Increasing evidence points to a role for environmental factors in AD risk, ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ with several studies implicating several components of urban air pollution in increased AD risk and cognitive deficits. ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ Urban air pollution is a complex mixture comprising several chemical, particulate, and gaseous components, such as ozone (O₃), which is a prevalent and chronic exposure with health effects spanning several organ systems, ¹⁸ , ¹⁹ including the central nervous system (CNS). In the United States, in 2022 alone, > 85 million people were exposed to air pollution levels exceeding US Environmental Protection Agency safety standards, ²⁰ emphasizing the importance of understanding how these inhaled exposures could be affecting the brain, particularly AD pathology.

Ozone, one of the components of urban air pollution and a common ground‐level air pollutant component, ¹⁹ , ²⁰ increases mortality ¹⁹ , ²¹ , ²² , ²³ with a Global Burden of Disease study attributing 12.1% of all male death in 2019 to ambient O₃ pollution along with ambient air pollution. ²³ Increased temperatures catalyze ground‐level O₃ accumulation; ²⁴ thus, climate change is a concern regarding increasing O₃‐associated health effects. ²⁰ Recent epidemiology studies have implicated O₃ as a strong risk factor for cognitive decline ⁹ , ²⁵ , ²⁶ and AD risk, ²⁷ , ²⁸ with one study finding 10.4% increase in cognitive impairment risk with increased O₃. ⁹ Prior rodent studies support that inhalation of O₃ affects the brain, demonstrating increased oxidative stress, mitochondrial dysfunction, neuronal damage, as well memory deficits. ²⁹ , ³⁰ , ³¹ , ³² , ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ Increased amyloid beta (Aβ) positron emission tomography signals in response to ambient air pollution have been observed in humans; ¹³ however, the underlying cellular mechanisms remain unclear. O₃ is a highly reactive gas and cannot transfer to the brain after inhalation; ³⁸ , ³⁹ , ⁴⁰ thus, circulating factors or trafficking peripheral immune calls have been implicated to contribute to O₃‐induced CNS effects in a pathway named “the lung–brain axis,” ²⁹ , ⁴¹ , ⁴² , among other potential pathways. ⁴³ , ⁴⁴ While the majority of the inhaled O₃ reacts with the alveolar lining fluid and O₃ exposure has been extensively linked to myeloid cell infiltration in prior studies, ²⁹ the cell‐specific and peripheral mechanisms underlying the impact of O₃ on amyloid pathology are unclear.

Aβ plaque deposition is a hallmark of AD. ⁴⁵ , ⁴⁶ , ⁴⁷ , ⁴⁸ Astrocytes ⁴⁹ and microglia ⁵⁰ surround plaques to form a protective barrier, restricting plaque toxicity to the surrounding neuropil and facilitating plaque clearance. ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ Disease‐associated astrocytes (DAAs), a recently identified reactive astrocyte subset with a unique transcriptional signature and high glial fibrillary acidic protein (GFAP) expression, surround plaques in human and 5xFAD mouse tissue. ⁵³ We previously demonstrated that O₃ exposure impacts plaque‐associated microglia in 5xFAD mice; ²⁹ however, little is known regarding how air pollution, including O₃, affects astrocytes and their plaque microenvironment localization.

Increasing evidence suggests that peripheral and systemic immune mechanisms contribute to AD pathobiology; ²⁹ , ⁵⁶ for example, peripheral myeloid cells are implicated in AD and amyloid pathology. ²⁹ , ⁵⁷ How these cells are pathologically modified and contribute to neurological disease is unknown, but urban air pollution exposure could play a role. ²⁹ High mobility group box 1 (HMGB1) is a ubiquitous nuclear DNA‐binding chaperone actively secreted by immune and damaged cells, acting as an autocrine and paracrine signal/cytokine. ⁵⁸ Importantly, circulating HMGB1 is elevated in O₃‐exposed 5xFAD mice, but CNS HMGB1 expression level changes are absent. ²⁹ Circulating HMGB1 is elevated in some AD patient populations, ⁵⁹ , ⁶⁰ implicating peripheral circulating HMGB1, but the role of HMGB1 as a transcription factor regulating specific cellular functions in AD is less known. The roles of these peripheral mechanisms in DAA phenotype development and amyloid pathology are unclear.

In the current study, we began to address these unresolved questions in the field by exploring: (1) how O₃‐induced changes in the periphery may regulate the astrocyte phenotype (transcriptomic, proteomic, cell number, and cell contact) and how this is modified by localization with plaques; (2) the potential role of peripheral myeloid cells in this process; and (3) whether the loss of peripheral myeloid cell HMGB1 is linked to changes in the DAA phenotype.

2. METHODS

2.1. Reagents

All reagents are listed in Tables S1–S5 in supporting information.

2.2. Animals

Male transgenic 5xFAD mice hemizygous for five familial AD mutations (APP K670N/M671L, Swedish; I716V, Florida; V717,I London; PSEN1 M146L; and PSEN1, L286V), ⁶¹ littermate controls on a C57Bl/6J background (B6. Cg‐Tg(APPSwFlLon,PSEN1*M146L*L286V)6799Vas/Mmjax; RRID:MMRRC_034848‐JAX), C57BL/6J (RRID:IMSR_JAX:000664), and LysM‐Cre (B6.129P2‐Lyz2tm1(cre)Ifo/J; RRID:IMSR_JAX:004781) mice were obtained from the Jackson Laboratory. Female 5xFAD mice exhibit markedly exacerbated amyloid pathology that increases rapidly over time, ⁶² , ⁶³ risking a ceiling effect when combined with O₃ exposure; thus, to prevent potentially confounding analyses, only male mice were used in the current study. Homozygous HMGB1 floxed (HMGB1^fl/fl) mice ⁶⁴ were obtained from Riken (B6.129P2‐Hmgb1 < tm1Ttg > ; BRC No. RBRC06240). Hmgb1 ^fl/fl mice were crossed with LysM‐Cre^+/+ mice to generate Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice with deletion of HMGB1 in only peripheral myeloid cells. ²⁹ Hmgb1 ^fl/fl _. LysM‐Cre^+/− mice were bred to produce the experimental Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice and control Hmgb1 ^fl/fl _. LysM‐Cre⁻ littermates. All Hmgb1 ^fl/fl _. LysM‐Cre mice were genotyped to confirm the presence or absence of Cre recombinase and homozygous floxed HMGB1 alleles.

The mice were acclimated to the housing facility for at least 1 week before all studies. All mice were maintained on a 12‐hour light/dark cycle (7:00 am–7:00 pm) in a specific pathogen‐free environment, excluding Helicobacter. The complete list of pathogen exclusions is provided in Table S6 in supporting information. Experimental mice were individually housed in high‐efficiency particulate‐absorbing filtered ventilated polycarbonate cages with food and acidic water (pH 2.2–2.7) provided ad libitum. All experiments were completed in strict accordance with the Indiana University School of Medicine Institutional Animal Care and Use Committee protocols (29002 and 27001) and National Institutes of Health (NIH) guidelines for housing, breeding, and experimental use. All mice were treated humanely to alleviate suffering.

RESEARCH IN CONTEXT

1. Systematic review: We reviewed articles in PubMed with combinations of keywords “astrocytes,” “Alzheimer's disease,” “air pollution,” and “ozone.” We found very little and limited work on the effects of urban air pollution on astrocytes, and how ozone (O₃) exposure affects astrocytic phenotype in Alzheimer's disease (AD) model is mostly unexplored.

2. Interpretation: Our findings indicate that O₃ exposure dysregulates astrocytes, where the specific proteomic and transcriptomic effect is dependent on localization near or away from plaques. We demonstrate that the disease‐associated astrocyte (DAA) phenotype that is important in responding to amyloid plaques is perturbed by the O₃‐induced lung–brain axis response and the consequent interacting communication between astrocytes and peripheral myeloid cells.

3. Future directions: Here, we highlight the important role of astrocytes in the lung–brain axis and the mechanisms by which the peripheral immune response to air pollution modulates this pathway, which is linked to augmented amyloid pathology. The detailed cellular mechanisms require significant additional inquiry. While we revealed O₃ dysregulation of the DAA phenotype associated with increased plaque load features increased cellular contact with microglia in the plaque microenvironment, the mechanistic underpinnings and impact on AD physiology require further exploration. Finally, this work also points to the vasculature as another point of impact in AD pathology that is of significant concern in understanding these complex mechanisms in future inquiry.

2.3. O₃ exposure

Mice were exposed to O₃ in full‐body Hinners inhalation chambers ⁶⁵ as previously described. ²⁹ Briefly, O₃ was produced with an HFL‐10 O₃ generator (Ozonology). The O₃ concentration was continuously monitored using an ultraviolet photometric O₃ analyzer (465L, Teledyne API), and the temperature was maintained at 21 ± 2°C. Rodents have lower sensitivity to O₃ toxicity than primates ⁶⁶ , ⁶⁷ due to their complex nasal turbinates, lung morphological differences, and unique airway surfactant. ⁶⁷ , ⁶⁸ Thus, increasing exposure by a factor of three is traditionally accepted for extrapolating to environmentally relevant human exposures. ⁶⁶ O₃ concentrations of 0.2 to 0.3 ppm are frequent in areas of high air pollution, similar to 1 ppm O₃ exposure in rodents. ⁴¹ , ⁶⁹ Regarding single exposures, 2 ppm was used to compensate for rodent insensitivity, as previously reported. ⁴² Rochester‐style Hinners chambers were used for whole‐body O₃ inhalation exposure. The mice were placed in wire mesh individual housing cages that were transferred to the chambers for exposure. Before the experiment, mice were habituated to the exposure chambers for 5 consecutive days (4 hours/day).

2.4. Exposure‐specific experimental design

For each study, animals were assigned to experimental groups using a randomized block design. Random numbers were generated using http://www.jerrydallal.com/random/randomize.htm.

2.4.1. Subchronic O₃ exposure

For the subchronic O₃ exposure experiments exploring how astrocytes are modified during O₃‐augmented amyloid plaque pathology, 120 male 10‐ to 11‐week‐old 5xFAD mice and littermate controls were exposed in two separate experiments (60 for CNS and 60 for pulmonary measures) to filtered air (FA), 0.3, or 1.0 ppm O₃ for 4 hours/day, 3 consecutive days/week, for 13 weeks (n = 10 per group). The mice were then euthanized, and samples were collected 18 tot 24 hours after the last exposure, as reported previously. ²⁹ Because the purpose of the study was to determine how O₃ modified the astrocyte phenotype during ongoing O₃‐augmented amyloid pathology, the littermate control strain (no plaques) and the mice exposed to 0.3 ppm O₃ (no O₃‐induced change in plaque pathology ²⁹ ) were excluded from processing and analysis.

2.4.2. Single O₃ exposure

Two studies were conducted to investigate the effects of O₃ in Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice. For the single 1.0 ppm exposure, 17 male 6‐ to 8‐week‐old Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice and Hmgb1 ^fl/fl _. LysM‐Cre⁻ mice were exposed to 1.0 ppm O₃ or FA once for 4 hours (n = 4–5 mice). Samples were collected 18 to 24 hours after exposure. To obtain clear neuroimmune measurements in this unique strain and genetic background, the exposure was increased to a single 2.0 ppm O₃ exposure, which results in increased neuroimmune responses in the control Hmgb1 ^fl/fl _. LysM‐Cre⁻ mice. ²⁹ As such, 38 male 6‐ to 8‐week‐old Hmgb1 ^fl/fl _. LysM‐Cre ⁺ mice and Hmgb1 ^fl/fl _. LysM‐Cre ⁻ mice were exposed to 2.0 ppm O₃ or FA once for 4 hours (n = 4–5 mice). Tissue samples were collected approximately 24 hours after exposure.

2.4.3. rHMGB1 IV administration

Sixteen male C57Bl/6J mice (6–7 weeks old; n = 8) were injected intravenously by tail vein with 32.5 µg rHMGB1 (Thermo Scientific, 34‐8401‐85) in 200 µL vehicle (20 mM Tris HCl, pH 8.0, 0.2 M NaCl, 1 mM DTT). Three hours after injection, the mice were euthanized, and samples were collected. Samples were excluded from analysis when identified as statistical outliers, resulting in a final sample size of n = 6 to 8.

2.5. Sample collection

2.5.1. Brain tissue

Mice were euthanized with isoflurane. One brain hemisphere was microdissected (cortex, hippocampus, and midbrain), flash‐frozen in liquid nitrogen, and stored at −80°C. The other half of the brain was fixed in 4% paraformaldehyde (Electron Microscopy Sciences, 1921) by immersion for 2 days, followed by cryopreservation in 30% sucrose in phosphate‐buffered saline (PBS) for another 2 days. Then, the entire hemisphere was embedded with optimal cutting temperature compound (4583, Sakura Finetec) in cryomolds (Tissue‐Tek, 4557, Sakura Finetec).

2.5.2. Pulmonary samples

Bronchoalveolar lavage (BAL) fluid was collected from euthanized mice by lavaging the lung twice with 1 mL Hanks Balanced Saline Solution (21‐622CV, Corning) without Ca²⁺ and Mg²⁺. BAL fluid samples were centrifuged at 1500 × g for 10 minutes at 4°C, and the cell pellets were resuspended in 250 µL of PBS. Total cell counts were determined using a TC‐10 automated cell counter (Bio‐Rad), applied to slides using a Shandon Cytospin centrifuge (Thermo Scientific), and stained with Wright–Giemsa (89013, Thermo Scientific). Cell differentials were determined at 40× by a blinded observer counting at least 300 cells per sample.

2.6. Fluorescent immunohistochemistry

Sagittal sections (40 µM) were collected using a freezing stage microtome (Microm HM 450, Thermo Scientific). For all following immunohistochemistry (IHC) endpoints, three evenly spaced sections approximately 0.24 mm apart spanning the motor cortex starting at the sagittal plane ≈ 0.6 mm lateral to the midline ⁷⁰ were stained per brain. All sections were washed for 10 minutes in 0.1% PBS with Tween 20 (PBST) prior to antigen retrieval. Antigen retrieval was performed in 1 M sodium citrate solution at 85°C for 15 minutes with subsequent cooling to room temperature for 20 minutes. Blocking was performed with normal donkey serum for 1 hour, followed by overnight incubation in primary antibody diluted (1: 500) in blocking serum. Rabbit anti‐GFAP (Agilent DAKO, Z0334; RRID AB_10013382) and goat anti‐ionized calcium‐binding adapter molecule 1 (IBA1; Novus Biologicals, NB100‐1028) were used to stain astrocytes and microglia, respectively. Sections were washed three times in 0.1% PBST and incubated with secondary Alexa Fluor antibodies diluted (1:1000) in blocking serum at room temperature for 1 hour, followed by three washes with 0.1% PBST. Donkey anti‐rabbit 647 (Invitrogen A‐31573; RRID AB_2536183) and donkey anti‐goat 488 (Invitrogen A11055; RRID AB_2534102) were used as secondary antibodies for GFAP and IBA1 staining, respectively. Sections were mounted on slides with Prolong Gold (P36930, Life Technologies), cover slipped, and dried overnight in the dark. Slides were stored at −20°C before imaging. To stain for Aβ plaques, sections were washed in 0.1% PBST for 5 minutes and Methoxy x34 (SL 1954, Millipore Sigma) solution (0.04 g X34 in 400 mL 100% ETOH and 600 mL DI H₂O) for 10 minutes. The slides were then sequentially washed five times with DDH₂O and 0.1% PBST for 5 minutes before the addition of Prolong Gold and coverslipping.

2.7. Imaging

2.7.1. Whole‐cortex GFAP expression

The entire hemisphere was scanned at 10× using a Leica Aperio Versa slide scanner (Leica Microsystems Inc.) to determine whole‐cortex astrocyte expression. Images were processed and analyzed using ImageJ (FIJI, version 2.9.0, NIH). The region of interest (ROI) was drawn around the entire cortex and manually thresholded. The Analyze Particles plugin was used to quantify the GFAP‐positive area and cell number. A sample size of 8 to 10 mice per group was analyzed per endpoint. Statistical outliers were excluded.

2.7.2. Plaque‐associated astrocyte quantification

For plaque‐associated astrocyte quantification, 1 µm Z‐stacks were acquired at 40× with oil immersion with a Nikon A1R confocal microscope in the primary motor region of the cortex due to ease of identification of the finite region using confocal microscopy. Images were analyzed in NIS Elements AR (Nikon) using the General Analysis 3 module. Image stacks were thresholded using consistent criteria across the entire analysis. To consistently define the plaque microenvironment, a 50 µm diameter circle was drawn around a plaque‐positive area of this specific size, defining the periplaque ROI. A blinded observer counted GFAP‐positive cell bodies manually in the maximum intensity projection image. Astrocytes were considered periplaque astrocytes if either the cell body or any branches were contained in the ROI. A sample size of 8 to 10 mice per group was analyzed per endpoint. Statistical outliers were excluded.

2.7.3. Colocalized astrocyte–microglia volume quantification: cell‒cell contact

To assess astrocyte–microglia colocalization, 1 µm spaced Z‐stacks were acquired from the primary motor region of the cortex at 40× with oil immersion with a Nikon A1R confocal microscope. Images were analyzed in NIS Elements AR using the General Analysis 3 Module. Image stacks were thresholded using criteria maintained across the entire analysis. Within a 50 µm diameter ROI, areas double positive for GFAP and IBA1 were quantified per stack. The total colocalized volume was obtained by adding the colocalized area of all the stacks in the image and normalized by the total GFAP volume. A sample size of 8 to 10 mice per group was analyzed per endpoint. Statistical outliers were excluded.

2.8. NanoString GeoMx digital spatial profiling: protein expression

Fixed 10 µm coronal sections were acquired with a cryostat (CM1900, Leica) and stored at −20°C before processing. Slides were processed per the NanoString Slide Prep manual for protein analysis. Briefly, slices were incubated with the NanoString GeoMx Alzheimer's Morphology kit, which contains the Aβ antibody for plaque visualization and Alexa Fluor 647‐conjugated mouse anti‐GFAP antibody (BioLegend, 837512; RRID AB_2734611) for astrocyte identification, along with NanoString panels (Neural Cell Profiling Core, Alzheimer's Disease Panel, Alzheimer's Disease Extended Panel, Glial Subtyping Panel) containing ≈ 60 antibodies with unique photo‐cleavable oligonucleotide tags. Slices were scanned in the NanoString Digital Spatial Profiler, and polygonal ROIs were drawn around plaque‐associated and non–plaque‐associated astrocytes in the cortex. The GFAP‐positive area within each ROI was delineated for photocleavable oligonucleotides collection. The collected oligonucleotides were hybridized with NanoString codeset, per the NanoString user manual, to map the counts to corresponding antibodies and region from where the oligonucleotides were collected. Digital counts were generated on an nCounter Max/Flex system (NanoString Technologies). The differential protein expression analysis was performed using the NanoString GeoMx Digital Spatial Profiling (DSP) Analysis Suite. The digital counts were tested for quality control and normalized to the housekeeping protein (GAPDH and histone H3) counts. Three sections were scanned per brain from a sample size of four animals per exposure group (FA or 1.0 ppm O₃) for the analysis, resulting in 68 to 70 ROIs analyzed per experimental group. ROIs that did not pass the quality control checks were excluded.

2.9. NanoString GeoMx digital spatial profiling: whole transcriptome assay

Fixed frozen 10 µm coronal brain sections were processed with the NanoString Mouse Whole Transcriptome Atlas (WTA) panel per the NanoString user manual for spatial transcriptomic analysis. Plaques and astrocytes were visualized using Alexa Fluor 594‐conjugated mouse anti‐Aβ antibody (BioLegend 803019; RRID AB_2734552) and Alexa Fluor 647‐conjugated mouse anti‐GFAP antibody (BioLegend; 837512 RRID AB_2734611), respectively. Sections were scanned in the NanoString Digital Spatial profiler, and polygonal ROIs were drawn around plaque‐associated astrocytes in the cortex. Photocleavable oligonucleotides were collected only from the GFAP‐positive segments. Collected oligonucleotides were sequenced using the Illumina Next Generation Sequencing Platform (NanoString Technologies). Differential gene expression and enriched pathway analyses were conducted using the NanoString GeoMx DSP Analysis Suite. Normalization was performed using the ubiquitous astrocyte marker Aldh1l1 ⁷¹ , ⁷² because its expression was unchanged in both groups. For the analysis, two to five sections were scanned per brain from a sample size of three animals per group, with 90 to 92 ROIs analyzed per experimental group. ROIs were excluded from the analysis if they did not pass the quality control test.

2.10. RNA isolation and reverse transcription quantitative polymerase chain reaction

Microdissected whole cortex (including corpus collosum) and midbrain tissue were homogenized in Tissue Protein Extraction Reagent with protease and phosphatase inhibitors. An equal volume of TRIzol was added to each homogenate, and RNA was extracted per the manufacturers’ protocol. RNA was treated with DNAse for purification using an Ambion DNA‐free kit (Invitrogen AM1906) and reverse‐transcribed using a Maxima First Strand cDNA synthesis kit (Invitrogen, K1641). Reverse transcription quantitative polymerase chain reaction (RT‒qPCR) was performed with 1 µL cDNA and TaqMan probes and primers (Tables S2 and S3) on a QuantStudio 6 Flex RT‒PCR system (Applied Biosystems). Gapdh was used as the housekeeping control for 2^−ΔΔCT quantification. For a complete list of primers and probes used, see Tables S2 and S3.

2.11. Statistical analysis

Experimenters were blinded to the experimental groups. The sample size was determined according to prior reports, and power analyses were calculated for 80% power. Data were analyzed in GraphPad Prism 8.0 (GraphPad Prism). Outliers were determined using the ROUT method with Q = 1% and removed from all analyses. Normal distribution was tested using the Shapiro‒Wilk test. A Welch t test was performed for data that passed the normality test. A two‐way analysis of variance with Bonferroni post hoc analysis was performed when applicable. A t test was used for GeoMx DSP protein profiling and a linear mixed model with Benjamini‒Hochberg post hoc correction for the whole transcriptome assay. Data are expressed as the mean ± standard error of the mean. A P value < 0.05 was considered to indicate significance.

3. RESULTS

3.1. Ozone exposure increases GFAP astrocyte density in the cortex

The role of astrocytes in maintaining neuronal health and function has been extensively studied and described. ⁴⁶ , ⁵³ Astrocytes are a heterogeneous and highly complex population of cells with unique phenotypic responses depending on pathological conditions. Some common characteristics of reactive astrocytes have been identified, such as the overexpression of GFAP with enlarged morphology, ⁴⁶ , ⁴⁹ , ⁵³ which is a characteristic of AD. ⁴⁶ , ⁴⁹ However, the instigating events responsible for astrogliosis in AD and how astroglia respond to air pollution are poorly understood. Here, cortical GFAP‐positive astrocytes were found to increase in number in response to O₃ exposure, as evidenced by increased O₃‐elevated GFAP‐positive cell counts and increased total cortical GFAP expression in 5xFAD mice (Figure 1A and B). To explore whether the O₃‐induced increase in cortical astrocyte density is affected by plaque proximity, we quantified the plaque‐associated and plaque‐distant astrocyte numbers in confocal images of the primary motor cortex. Consistent with prior reports, ²⁹ subchronic O₃ exposure increased the plaque number (Figure 1C and D). O₃ also increased the periplaque astrocyte number; however, only a trend toward an O₃‐induced increase was observed in the number of non‐plaque astrocytes (Figure S1A and S1B in supporting information). No significant difference in the number of periplaque astrocytes per plaque was identified (Figure 1E), suggesting that the close proximity of astrocytes to the higher number of plaques may be associated with the higher number of astrocytes. This association is further supported by the significant correlation between periplaque astrocytes and plaque number irrespective of O₃ exposure (Figure 1F). However, the number of non–plaque‐associated astrocytes increased with O₃ exposure with no significant association with plaque number (Figure 1G), suggesting that O₃ affects astrogliosis differently depending on plaque localization.

FIGURE 1.

O₃ exposure increases astrocyte density in the cortex of 5xFAD mice. Male 5xFAD mice (10–11 weeks old) were exposed to either FA or 1.0 ppm O₃ for 3 consecutive days each week for 4 hours/day for 13 weeks. A, Representative 10× images depicting cortical astrocyte density (GFAP, red) in FA‐ and O₃‐exposed mice. Scale bar: 1000 or 100 µm. B, Quantification of the number of GFAP‐positive areas and GFAP‐positive cells in the entire cortex (layers I–VI). C, Representative maximum intensity projection images taken at 40× in the cortex, staining for plaques (Methoxy‐X34, gray) and astrocytes (GFAP, green). Scale bar: 50 µm. D, Quantification of plaque number in the 40× confocal images. E, Quantification of plaque‐associated astrocytes normalized to plaque number. Correlation of the number of plaques with the number of (F) periplaque and (G) non‐plaque astrocytes in the cortex. Astrocytes were considered periplaque if their cell bodies or branches reached within the circular periplaque region of interest drawn at 50 µm diameter around the plaque center; if not, they were considered nonplaque. Data are represented as the mean ± SEM, n = 8–9 mice/exposure group. ^* = P < 0.05; Welch t test. FA, filtered air; GFAP, glial fibrillary acidic protein; O₃, ozone; SEM, standard error of the mean

3.2. Ozone alters astrocyte‐associated protein expression in the periplaque space

To understand how O₃ exposure and plaque localization could interact to affect the astrocyte phenotype and obtain insight into potential functional changes, we next sought to spatially profile cortical astrocyte protein expression according to their proximity to or distance from amyloid plaques. As such, we compared the protein expression pattern in periplaque astrocytes to that in astrocytes distant from the plaque deposition site (non‐plaque) in both O₃‐ and FA‐exposed mice. More specifically, we sought to elucidate how the astrocyte protein response to plaque deposition changes with O₃ exposure. DSP analysis revealed a baseline change in 25 astrocyte‐associated proteins (Table S7 in supporting information) in periplaque astroctyes, consistent with prior findings. ⁴⁹ , ⁵³ Notably, the expression of 16 proteins changed regardless of O₃ exposure (Figure 2B). However, comparing plaque‐associated and plaque‐distant astrocytes, nine proteins (Figure 2B and C) were changed in only FA‐exposed mice, including neprilysin and Ki‐67, indicating a unique baseline change in plaque‐associated astrocyte proteins that occurs without O₃ exposure. Thus, O₃ exposure appears to impede this shift in protein expression. However, myelin basic protein (MBP), CSF1R, and Clec7a protein expression levels were increased in response to O₃ in plaque‐associated astrocytes (Table S7). The upregulation of microglia‐associated proteins in response to O₃ in GFAP‐positive tissue indicated increased cell‐to‐cell physical overlap, suggesting that O₃ may modify astrocyte–microglia communication in the plaque microenvironment.

FIGURE 2.

O₃ altered the astrocytic protein expression pattern, dependent on spatial localization with plaques. Male 5xFAD mice (10–11 weeks old) were exposed to FA or 1.0 ppm O₃ for 3 consecutive days each week for 4 hours/day for 13 weeks. A, Representative images from the NanoString GeoMX DSP platform illustrating periplaque (left) and non‐plaque astrocytes (right), as defined by plaque staining (Aβ, magenta), astrocyte (GFAP, green) staining, and the area from which samples were collected for analysis (blue). Scale bar: 10 µm. B, Venn diagram showing the number of plaque environment‐induced changes in the astrocyte proteomic profile in FA and O₃ groups, indicating protein changes shared between the two groups. Volcano plots representing differentially expressed proteins on periplaque versus plaque‐distant astrocytes in the (C) FA and (D) O₃ groups. n = 68–72 ROIs/region per exposure group (n = 4 mice/exposure group). P < 0.05. Aβ, amyloid beta; DSP, Digital Spatial Profiling; FA, filtered air; GFAP, glial fibrillary acidic protein; O₃, ozone; ROI, region of interest

3.3. Ozone increases astrocyte and microglia colocalization in the plaque microenvironment

The overlapping GFAP‐positive and IBA1‐positive volume in the motor cortex periplaque space was calculated to directly test whether O₃ exposure affects astrocyte–microglia cell‒cell contact around plaques (Figure 3A and Figure S2A in supporting information). The data demonstrated that O₃ exposure increased astrocyte and microglia colocalization in the plaque microenvironment (Figure 3B) but not in non‐plaque regions (Figure 3C and D), suggesting that O₃ modifies glial cell‒cell communication only near amyloid plaques. We have previously seen an O₃‐induced decrease in plaque associated microglia, ²⁹ making this increased colocalization even more intriguing. To verify that this increased colocalization is not simply a result of increased GFAP in the periplaque space, we normalized the data to total GFAP (Figure 3B). The significant difference was retained regardless of the source of GFAP volume.

FIGURE 3.

O₃ increased astrocyte–microglia cell contact in the plaque microenvironment. Male 5xFAD mice (10–11 weeks old) were exposed to either FA or 1.0 ppm O₃ for 3 consecutive days each week for 4 hours/day for 13 weeks. A,= Representative image showing colocalized areas (yellow) of cell‒cell contact, as indicated by white arrows in a single image from a set of confocal Z‐stack images taken at 40×. Scale bar: 10 µm. Quantification of astrocyte–microglia colocalization in the (B) periplaque and (C) non‐plaque (right) space from confocal Z‐stacks taken at 40× in the cortex. Data are represented as the mean ± SEM, n = 8–9 mice/exposure group. ^* = P < 0.05; Welch t test. D, Representative maximum intensity images taken at 60× showing plaques (Methoxy‐X34, gray), astrocytes (GFAP, green), and microglia (IBA1, red) in the O₃ and FA groups. Scale bar: 10 µm. FA, filtered air; GFAP, glial fibrillary acidic protein; IBA1, ionized calcium‐binding adapter molecule 1; O₃, ozone; SEM, standard error of the mean

3.4. Ozone triggers an astrocytic transcriptional shift

Large‐scale astrogliosis is observed in AD, and impaired clearance of Aβ by astrocytes is thought to be deleterious. ⁴⁹ A recently identified DAA population displays a somewhat distinct transcriptional phenotype when localized around plaques; however, this transcriptomic fingerprint is also closely linked with astrocyte subpopulations in the aged brain. ⁵³ Very little is known about how air pollution exposure changes astrocytes, particularly the transcriptional phenotype of plaque‐associated astrocytes. Thus, we used DSP and the NanoString WTA to assess the transcriptional changes in periplaque versus plaque‐distant astrocytes in FA‐ and O₃‐exposed 5xFAD mice. O₃ exposure triggered a unique transcriptional shift in periplaque astrocytes (Figure 4A and B). On closer examination of the differentially expressed genes in these astrocytes (Table S8 in supporting information), an accelerated DAA phenotype was present, highlighted by increased expression of serpina3n, a serine protease inhibitor associated with increased amyloid accumulation; ⁵³ , ⁷³ ctsb and ctsd, lysosomal cysteine proteases associated with amyloid precursor protein processing; ⁵³ , ⁷⁴ and c1qa, a complement factor associated with astrocyte‐mediated synapse elimination. ⁷⁵ However of the protein targets upregulated in the O₃ group, only csf1r, cd9, and ctsd were also significantly transcriptionally upregulated with DSP analysis. Pathway analysis revealed that O₃ upregulated several functional categories of genes in periplaque versus plaque‐distant astrocytes, including cell‒cell communication and gap junction trafficking pathways (Figure 4C and Table S9 and S10 in supporting information). O₃ also resulted in the downregulation of genes in other categories in periplaque but not plaque‐distant astrocytes, such as matrix metalloproteinase (Figure 4D), suggesting extracellular matrix modifications potentially underlie the spatially defined differences. Overall, these results suggest O₃ triggers a potentially pathologically dysregulated astrocyte phenotype that occurs concomitantly with higher Aβ accumulation.

FIGURE 4.

O₃ alters the astrocytic transcriptional profile in the plaque microenvironment. Male 5xFAD mice (10–11 weeks old) were exposed to either FA or 1.0 ppm O₃ for 3 consecutive days each week for 4 hours/day for 13 weeks. A, Representative image showing the digital spatial profiler scan for regions of interest containing astrocytes localized in the periplaque space, as defined by plaque staining (6e10, magenta), astrocyte staining (GFAP, green), and the area from which GFAP‐positive cells were collected for mRNA analysis (blue). Scale bar, 50 µm. B, Volcano plot showing O₃‐induced gene expression in plaque‐associated astrocytes compared to that in the FA group. Highlighted genes represent a subset of significantly changed genes (red dots) after FDR correction (Benjamini‒Hochberg, P < 0.05). Pathway analysis of O₃ versus FA cortical astrocyte gene expression in plaque‐associated astrocytes depicting significantly (C) increased and (D) decreased pathways of interest (Benjamini–Hochberg, P < 0.05. n = 96 ROIs/region per exposure group (n = 3 mice/exposure group). FA, filtered air; FDR, false discovery rate; GFAP, glial fibrillary acidic protein; O₃, ozone; ROI, region of interest

3.5. Peripheral HMGB1 mediates O₃‐induced dysregulation of astrocytes

We previously demonstrated that peripheral HMGB1 regulates the microglial response to O₃ in 5xFAD mice; ²⁹ however, the impact on astrocytes is unknown. Here, HMGB1 mRNA levels were found to be downregulated in O₃‐exposed BAL fluid (Figure 5A), predominantly comprising infiltrating myeloid cells consistent with prior studies in other tissues. ²⁹ , ⁷⁶ Thus, we hypothesized that peripheral myeloid cell HMGB1 is involved in the association between astrocytes and amyloid pathology. To test this hypothesis, we used the previously described Hmgb1fl/fl.LysM‐Cre⁺ mouse strain, ²⁹ with HMGB1 specifically deleted in only peripheral myeloid cells, including BAL fluid cells. A single 4 hour O₃ exposure (1 ppm) reduced neutrophil and lymphocyte infiltration into the BAL fluid in Hmgb1fl/fl.LysM‐Cre⁺ mice (Figure 5B), indicating that myeloid cell HMGB1 plays an important role in peripheral immune cell trafficking and the pulmonary immune response to O₃. Given the reduced HMGB1 mRNA in these cells in 5xFAD mice after a 13 week exposure, we hypothesize that ongoing AD pathology and extended exposure length could have a differential impact on HMGB1 mRNA expression in these cells. Interestingly, serpina3n levels in the midbrain of Hmgb1fl/fl.LysM‐Cre⁺ mice were reduced in response to O₃ exposure (Figure 5C), further suggesting that peripheral myeloid cells and HMGB1 influence the expression of astrocyte genes in the lung–brain axis.

FIGURE 5.

Peripheral HMGB1 modulates O₃‐induced astrocytic dysregulation. A, Subchronic (13‐week) O₃ (1 ppm) exposure reduced HMGB1 mRNA expression in the BAL cells of 5xFAD mice. Data are represented as the mean ± SEM, n = 6 mice/exposure group. Welch t test. B, Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice have Hmgb1 genetically ablated in peripheral myeloid cells (comprising a substantial component of BAL fluid cells) but not microglia. Hmgb1 ^fl/fl _. LysM‐Cre⁻ and Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice were exposed to O₃ (1.0 ppm) or FA once for 4 hours. Cell counts of eosinophils, neutrophils, and lymphocytes infiltrating the BAL are shown. Data are represented as the mean ± SEM, n = 3–5 mice/group. ^* = P < 0.05, ^**  = P < 0.01, ^***  = P < 0.001; Welch t test. C, Hmgb1 ^fl/fl _. LysM‐Cre⁻ and Hmgb1 ^fl/fl _. LysM‐Cre⁺ mice were exposed to O₃ (2.0 ppm) or FA once for 4 hours. Serpina3n mRNA levels in the midbrain after a single 2 ppm O₃ exposure are shown. Data are represented as the mean ± SEM, n = 9–10 mice/exposure group. ^* = P < 0.05, ^†† = P < 0.01. D, Gfap, C3, and Aqpn4 mRNA levels were assessed 3 hours after a tail vein injection of rHMGB1 (32.5 µg). Data are represented as the mean ± SEM, n = 6–8 mice/exposure group. ^* = P < 0.05; Welch t test. BAL, bronchoalveolar lavage; FA, filtered air; HMGB1, high mobility group box 1; O₃, ozone; SEM, standard error of the mean

Circulating HMGB1 has previously been shown to be upregulated in 5xFAD mice in response to O₃. ²⁹ Thus, we treated C57 mice with 32.5 µg recombinant HMGB1 protein intravenously, as previously reported, ²⁹ , ⁷⁷ to determine whether circulating HMGB1 can regulate the DAA phenotype. Circulating HMGB1 increased Gfap mRNA expression in the midbrain, and there was a trend toward an increase in c3, and aqpn4 mRNA expression (Figure 5D). Most of these DAA markers were demonstrated above to be modified by O₃ in the cortex in 5xFAD mice (Table S8). Overall, these data suggest that the peripheral myeloid cell HMGB1 is modified by O₃ exposure and may play an important role in immune cell trafficking and modulating the astrocytic response to O₃ exposure.

4. DISCUSSION

O₃ is a major component of urban air pollution and abundant at levels deleterious to human health across the United States and worldwide. ²⁴ High O₃ levels have recently been strongly associated with increased AD risk. ²⁷ In addition, O₃ is associated with chronic obstructive pulmonary disease (COPD) and is a well‐known asthma irritant, ⁷⁸ , ⁷⁹ and recent studies have linked both asthma ⁸⁰ and COPD ⁸¹ to increased dementia risk, further emphasizing the importance of understanding the underlying disease mechanisms. In the current study, using 5xFAD mice, we explored the mechanistic underpinnings of how an inhaled gas incapable of translocating to the brain (O₃) could impact the brain and cellular pathology in the CNS parenchyma in association with peripheral immune responses. We define periplaque DAAs and demonstrate the involvement of peripheral myeloid cells in their regulation to delineate an overlooked component of the lung–brain axis that potentially influences how urban air pollution promotes AD pathobiology (Figure 6).

FIGURE 6.

Plaque‐associated astrocytes and peripheral myeloid cells interact in the O₃‐dysregulated lung–brain axis: Implications for Alzheimer's disease. O₃, a reactive gas component of urban air pollution that cannot reach the brain, increased astrocyte density in the 5xFAD mouse cortex, concomitant with decreased bronchoalveolar lavage fluid cell (predominantly myeloid) HMGB1 expression and an exacerbated plaque burden. O₃‐induced astrocyte effects (transcriptomic and proteomic) were dependent on the localization of astrocytes relative to plaques, indicating that this air pollution exposure selectively and qualitatively changes astrocytes in the plaque microenvironment, accelerates the astrocyte transcriptomic shift to a disease‐associated astrocyte phenotype, and increases astrocyte contact with microglia but not plaques. Mechanistically, O₃‐exposed mice with HMGB1 deleted from the peripheral myeloid cells but not microglia exhibited a perturbed pulmonary immune response to O₃ and disrupted DAA markers in the brain, indicating that peripheral myeloid cells and HMGB1 regulate the astrocyte DAA response to O₃. These findings provide much‐needed insight into how urban air pollution may dysregulate the lung–brain axis, disrupt astrocytic function, and increase the amyloid burden. DAA, disease‐associated astrocytes; HMGB1, high mobility group box 1; O₃, ozone

In the current study, we demonstrated that GFAP astrocyte density in the cortex of 5xFAD mice increases in response to O₃ exposure (Figure 1A and B). The number of plaque‐associated astrocytes increased with O₃ exposure; however, the number of astrocytes surrounding each plaque did not increase. Thus, the increase in cortical astrocytes was likely due to the increase in the number of plaques in response to O₃ exposure (Figure 1E). Importantly, the astrocytes surrounding plaques, which are important for plaque clearance and reducing toxicity, ⁵⁴ , ⁵⁵ were found to be qualitatively different after O₃ exposure (Figures 2, 3, 4). Periplaque astrocytes from 5xFAD mice exposed to O₃ exhibited a localization‐dependent altered astrocytic proteomic profile (Figure 2). On closer examination, many of the proteomic changes occurring only with O₃ exposure in plaque‐distant versus periplaque astrocytes involved the loss of several key proteins (Figure 2). Some unexpected proteins, such as MBP, were upregulated in plaque‐associated astrocytes in response to O₃ exposure (Table S7). This finding suggests that astrocytes may take up MBP during neurite breakdown, consistent with our prior work showing that O₃ enhances dystrophic neurites in the periplaque space. ²⁹

Importantly, the upregulation of microglia‐specific proteins in astrocytes (Table S7) likely reflects an increase in cell‒cell interactions and communication between astrocytes and microglia in the plaque microenvironment. We directly tested increased astrocyte–microglia interactions by IHC (Figure 3) and demonstrated that O₃ increases astrocyte–microglia contact only in the periplaque space. Additionally, these periplaque cells differ transcriptionally from plaque‐distant astrocytes, demonstrating altered cell‒cell communication and gap junction trafficking pathway genes in the plaque microenvironment (Figure 4). It has been previously reported that microglia and astrocyte cell‒cell contact may be important for amyloid clearance. ⁸² We hypothesize that increased astrocyte–microglia communication may signify a coordinated effort between these cell types in compensating for the O₃ exposure–induced dysregulation of protective functions in the periplaque space.

While O₃ mainly induced a loss of protein expression in periplaque astrocytes, a larger transcriptional inquiry revealed the upregulation of genes such as serpina3n, c1qa, c1qb, and ctsb, indicating that O₃ exposure shifted the astrocyte transcriptional signature toward a more enriched DAA phenotype. Among the DAA genes upregulated by O₃ only in the periplaque space, serpina3n is associated with increased Aβ deposition and could contribute to increased plaque burden. ⁵³ , ⁷³ Notably, DAA astrocytes precede plaque deposition in 5xFAD mice and accumulate over time. ⁵³ Here, exposure to O₃ appears to accelerate the DAA phenotype in 5xFAD mice. However, while this O₃‐accelerated DAA phenotype correlated with an O₃‐induced increase in plaque burden in the current study, the beneficial or deleterious consequences of the O₃‐modified astrocyte genes in the periplaque space remain unclear and require substantial further investigation.

The transcriptional data emphasize that O₃ exposure qualitatively changes periplaque astrocytes, revealing multiple important potential targets for future mechanistic inquiry. For example, O₃ exposure has a well‐established impact on stroke and vascular pathology, ³⁰ , ³⁷ , ⁸³ and the astrocyte gene expression pathway analysis in the current study denotes changes in pathways involving vascular wall communication, vegfa–vegr2 communication (which occurs in vascular endothelial cells), and aquaporin communication, all of which point to O₃‐induced phenotypic changes in plaque‐associated astrocytes that may be linked to neurovascular dysfunction or pathology. Interestingly, gene expression changes in neurotransmitter pathways were also altered, which is unsurprising because astrocytes are known to maintain neurotransmitter homeostasis, the disruption of which is neurotoxic. This finding supports our previous data showing that O₃ exposure augments neuritic dystrophy. ²⁹

We next sought to better understand the mechanisms underlying these changes in the periplaque astrocyte phenotype. Peripheral immune cells traffic to the lung upon O₃ inhalation, ⁸⁴ and these cells and their associated circulating factors are key components of the lung–brain axis. ⁴¹ Here, we discovered that with subchronic O₃ exposure in 5xFAD mice, when the plaque load was increased and the proteomic and transcriptomic DAA phenotype was disrupted, HMGB1 gene expression was lowered in the immune (predominantly myeloid) cells that trafficked to the lung (Figure 5). While this decrease in gene expression does not demonstrate a direct impact on Aβ plaque load it implies a potential relation to be directly tested in future studies. Furthermore, peripheral myeloid cell‐specific HMGB1 deletion reduced immune cell infiltration into the lung and midbrain serpina3n expression (Figure 5), implicating the lung–brain axis in modulating the astrocytic transition to the DAA phenotype. Our data demonstrating that circulating HMGB1 upregulates DAA marker expression further indicate that peripheral HMGB1 influences the periplaque astrocyte phenotype.

Overall, our findings demonstrate that O₃ triggers a qualitative change in periplaque astrocytes, dysregulates cell‒cell communication and cellular function in the plaque microenvironment, and accelerates the astrocyte transcriptomic shift toward the DAA phenotype, which is associated with an increased plaque number. These findings identify astrocytes, particularly periplaque astrocytes, as a key mechanistic component of the lung–brain axis associated with the effect of air pollution on AD pathology. Finally, we demonstrate the important role of peripheral HMGB1 in this process, highlighting a critical need to investigate this highly complex mechanistic pathway further.

One limitation of this study is the use of the 5xFAD mouse model. While this model is popular for investigating amyloid pathology, the AD phenotype is aggressive, particularly in female mice. This prevented us from assessing female mice in this study, and sex‐based differences may exist. Additionally, the focus of this study was astrocytic dysregulation in the context of amyloid pathology. To fully elucidate the role of the lung–brain axis in astrocytic dysregulation in AD, its effects on tau pathology must be investigated in the future.

While O₃ is unable to directly interact with the brain parenchyma after inhalation, it is important to note that in addition to the pulmonary epithelium, O₃ also reacts with nasal epithelium potentially causing some lesions and generating cytokines among other factors that could also be an indirect pathway signaling O₃‐induced CNS effects, which is a point of future study, to fully elucidate how O₃ affects the brain.

This article introduces a complex mechanism implicating astrocytes in the lung–brain axis that will require extensive multidisciplinary studies to fully elucidate. The following prospects for future inquiry were identified according to our results: (1) The identification and localization of the culpable immune cell cascade, tracing the first immune cell response in the periphery to the transfer of the cascade to the brain parenchyma and subsequent follow‐up of chronic consequences. (2) An exploration of the CNS cellular contribution and the role of cell–cell contact, including the CNS vasculature unit and border‐associated macrophages, in how O₃ communicates with the brain and impacts astrocyte and amyloid pathology. (3) Rescue studies investigating whether peripheral circulating factors can ameliorate the O₃‐augmented astrocyte and amyloid pathology. (4) Investigations into the O₃ lung–brain axis exposome and its role in this process, comparing identifiable markers of pathological peripheral immune changes, deleterious circulating factors, ongoing astrocyte disruption, and amyloid pathology in AD animal models exposed to O₃ with those in AD, asthma, and COPD patient data.

This study is the first to identify astrocytes as part of the lung–brain axis and explore how trafficking immune cells can modify astrocytes and, potentially, amyloid plaque pathology in response to air pollution. Together, these findings provide much‐needed insight into the underlying mechanisms driving how exposure to high levels of air pollutants, such as O₃, increases AD risk, highlighting the need to investigate this complex mechanistic pathway further to deepen our understanding of AD etiology, identify targets for the prevention and treatment of this disease, and guide the development of policies regulating air pollution.

CONFLICT OF INTEREST STATEMENT

The authors have no competing interests to declare. Author disclosures are available in the supporting information.

CONSENT STATEMENT

No human subjects were used in this study and consent was not required.

Supporting information

Supporting Information

Supporting Information

Ahmed C, Greve HJ, Garza‐Lombo C, et al. Peripheral HMGB1 is linked to O₃ pathology of disease‐associated astrocytes and amyloid. Alzheimer's Dement. 2024;20:3551–3566. 10.1002/alz.13825