Pulmonary toxicity of brake wear particles: effects of composition and surface reactivity

Abstract

Background

Airborne brake wear particles (BWPs) are increasingly recognized as important non-exhaust contributors to urban particulate pollution; however, their inhalation toxicity and the key determinants of their pulmonary effects remain poorly defined.

Results

Two types of BWPs were generated using a brake dynamometer equipped with non-asbestos organic (NAO) and low-metallic (LM) brake pads on a cast-iron disc. The ≤ 2.5 μm fractions of BWPs generated from the NAO and LM brake pads were designated NAO_2.5 and LM_2.5, respectively. SRM 2975 (diesel exhaust particles) and Fe₂O₃ nanoparticles were included as reference particles. The test particles were characterized for their physicochemical properties, including morphology, size, surface area, crystallinity, colloidal properties, chemical composition, and solubility. The intrinsic oxidative potentials (IOPs) of the test particles, evaluated using a cell-free 2′,7′-dichlorodihydrofluorescein diacetate assay, ranked as NAO_2.5 > SRM 2975 > LM_2.5 > Fe₂O₃ on a mass basis, and as NAO_2.5 > LM_2.5 > SRM 2975 > Fe₂O₃ on a surface area basis. Pulmonary neutrophilic inflammatory responses were evaluated by bronchoalveolar lavage fluid analysis at 24 h after pharyngeal aspiration of 25, 50, and 100 µg/mouse in female BALB/c mice. On a mass basis, the inflammatory response ranked SRM 2975 > NAO_2.5 > Fe₂O₃ > LM_2.5 on a mass basis, whereas on a surface area basis, the ranking was NAO_2.5 > LM_2.5 > SRM 2975 > Fe₂O₃. Lung burden analysis at days 0, 1, and 28 after a single pharyngeal aspiration of 100 µg/mouse showed relatively higher 28-day retention of NAO_2.5 and LM_2.5 (approximately 75% and 67%, respectively) compared to that of SRM 2975 (47%).

Conclusions

These data indicate that NAO_2.5 is more inflammogenic than LM_2.5. Given that BWPs were larger and had substantially lower surface area than the reference particles, their stronger responses under surface area-based comparison suggest greater inflammatory potency per unit surface area. Overall, the BWPs examined in this study exhibited a toxicity profile not fully explained by IOP alone, suggesting an important contribution of particle-specific physicochemical properties.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12989-026-00684-7.

Background

Air pollution remains a critical public health challenge worldwide, with fine particulate matter (PM) exposure strongly associated with premature mortality and cardiopulmonary disease [1, 2]. Despite considerable advances in reducing exhaust emissions through the European Union vehicle emission regulations (Euro 1–6 standards), non-exhaust PM sources, including particles from brake wear, tire wear, and road surface abrasion, have remained outside the main scope of regulatory frameworks and have received limited investigation regarding their impacts on human health [3]. However, the planned introduction of Euro 7 represents a paradigm shift, extending the regulatory focus from exhaust PM emissions to non-exhaust PM emissions for the first time [4]. In particular, brake wear particles (BWPs) are a dominant non-exhaust source of urban PM, accounting for up to ≈ 30% of traffic-related PM_2.5 and ≈ 32–53% of PM₁₀ near major roadways [5].

BWPs originate primarily from high-temperature frictional interactions between brake pads and discs during braking. This process generates a complex mixture of airborne particles across a broad size distribution, encompassing coarse (PM₁₀), fine (PM_2.5), and ultrafine (PM_0.1) fractions [3]. The emission factors of BWPs range from about 4–7 mg/km for passenger cars to 20–40 mg/km for heavy-duty vehicles [6]. In terms of their primary particle size, BWPs are generally smaller than particles from tire or road surface abrasion, and a substantial fraction lies within the respirable (< PM_2.5) range, thereby increasing the likelihood of inhalation and subsequent adverse health effects [7, 8]. The chemical composition of BWPs is complex and variable, comprising metals (e.g., iron, copper, and zinc), carbonaceous phases, phenolic resin binders, inorganic fillers (e.g., alumina and silica), and occasionally organic toxicants such as polycyclic aromatic hydrocarbons (PAHs) [7, 9]. Their relative abundance varies with pad formulations and braking conditions. At the same time, the resulting particles exhibit considerable heterogeneity in morphology, size, and surface chemistry, altogether suggesting the potential for combined adverse effects upon inhalation. In addition, chemicals released from BWPs can migrate into environmental matrices, be taken up by plants, and ultimately pose potential risks to human health via food-chain transfer [10–12]. Although airborne PM has been extensively investigated, most toxicological studies have focused on exhaust-derived particles. By comparison, non-exhaust particles are increasingly recognized as important emerging pollutants in the atmosphere. While their emissions into the air have been documented in several studies, toxicological evidence remains very limited, particularly for BWPs.

Recent in vitro investigations have demonstrated that BWPs can disturb alveolar cellular homeostasis more strongly than diesel exhaust particles (DEPs), mainly due to their high oxidative potential and copper enrichment [13]. A recent in vivo study using a single semi-metallic brake pad, in which particles were artificially generated by grinding and wet milling, showed that repeated intratracheal instillation of brake wear PM_2.5 at a relatively high dose (135 µg/mouse, twice weekly for four weeks) induced pulmonary inflammation, fibrotic changes, and impaired lung function [14]. However, since the particles were not produced under actual braking conditions and the exposure levels were unrealistically high, the toxicological relevance of the findings remains limited. Therefore, this study employed a dynamometer system to generate BWPs from non-asbestos organic (NAO) and low-metallic (LM) brake pads and included Fe₂O₃ nanoparticles and DEP as reference particles for comparative toxicological evaluation. Then, the comparative pulmonary inflammatory potential of these particles was systematically evaluated at realistic exposure concentrations using a mouse aspiration model, providing novel insights into their inhalation toxicity.

Methods

Preparation of BWPs and reference particles

BWPs were generated using a brake dynamometer equipped with NAO and LM brake pads mounted on a cast-iron disc, as previously described [15]. The brake system was enclosed within a 60 cm circular chamber supplied with HEPA-filtered air, and the generated particles were transported through a 15 cm wind tunnel at a flow rate of 5,300 L/min, corresponding to an air velocity of 5 m/s. Particle samples were collected 1.5 m downstream using an isokinetic probe operated at 16.7 L/min, with particles larger than 2.5 μm removed by an upstream 2.5 μm cut-off impactor connected to a PM_2.5 sampler (KMS-4200, Kemik, Seoul, Korea). The classified BWPs were deposited onto 47 mm diameter PTFE membrane filters (pore size: 2 μm) during repeated Worldwide Harmonized Light-Duty Test Cycle (WLTC) driving cycles. Filters were replaced when the collected particle mass exceeded 2 mg. The recovered NAO and LM fractions were designated as NAO_2.5 and LM_2.5, respectively. Detailed methods for particle recovery are provided in the Supporting Information. As reference materials, γ-Fe₂O₃ (maghemite; US-Nano, Houston, TX, USA) and SRM 2975 (DEP reference material; US National Institute of Standards and Technology) were employed to compare the toxicity of BWPs with that of iron oxide (iron being the predominant element in the inorganic fraction of BWPs) and with a representative exhaust PM, respectively.

Preparation of particle suspension for in vivo study

To prepare the suspension of particles for animal experiments, powdered BWPs and reference particles were dispersed in distilled water (DW) without any dispersants, such as serum albumin. Stock suspensions were prepared at a concentration of 2 mg/mL and sonicated using a bath sonicator (Saehan Sonic, Seoul, Korea) for 10 min to disrupt agglomerates. In contrast, SRM 2975 required 80 min of extended sonication, as determined by a previous dispersion test [16].

Physicochemical characterization of test particles

To characterize the physicochemical properties of the test particles associated with toxicity, particle size, shape, surface area, crystallinity, surface charge, inorganic compound levels, PAH levels, and endotoxin levels were assessed.

Morphological, structural, and colloidal characterization of the test particles

The primary particle size, morphology, and chemical composition of particles dispersed in DW were evaluated using transmission electron microscopy coupled with energy-dispersive X-ray spectroscopy (TEM-EDS; JEM-1200EX II; JEOL, Tokyo, Japan). The average primary particle size was calculated by measuring 100 particles using ImageJ software (National Institutes of Health, Bethesda, MD, USA). The surface area of test particles was evaluated using the Brunauer-Emmett-Teller (BET) method with a BELSORP-MAX analyzer (Microtrac BEL, Osaka, Japan). The crystalline structures of particles were characterized by X-ray diffraction (XRD; SmartLab SE, Rigaku, Japan) using Cu Kα radiation (λ = 1.5406 Å) at 40 kV and 30 mA. The hydrodynamic size, polydispersity index (PDI), and zeta potential of particles dispersed in DW were measured using a Zetasizer Nano ZS (Malvern Instruments, Worcestershire, UK). Fourier transform infrared (FTIR) spectra were obtained using an FTIR spectrometer (JASCO, FTIR-6600, Japan) over a range of 400 to 4000 cm^− 1. Raman spectroscopy was performed using a 532 nm laser Raman microscope (WEVE, RAON-Spec, Korea). Dried particle samples were directly used for BET, XRD, FTIR, and Raman analyses. BET, XRD, and FTIR measurements were each performed once, whereas Raman measurements were performed three times. Detailed methods for physicochemical characterization are provided in the Supporting Information.

Inorganic chemical characterization

Inorganic elemental composition was determined by inductively coupled plasma-optical emission spectrometry (ICP-OES) at the UNIST Central Research Facilities (Ulsan, Korea). Detailed sample digestion and instrumental analysis conditions are provided in the Supporting Information.

PAH analysis

For PAH analysis, eight target PAHs, including benzo[a]anthracene, chrysene, benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene, indeno[1,2,3-c, d]pyrene, dibenzo[a, h]anthracene, and benzo[g, h,i]perylene, were analyzed by gas chromatography-mass spectrometry (GC-MS; 7890 GC coupled with 5977 MSD; Agilent Technologies, Santa Clara, CA, USA) using the internal standards.

Endotoxin analysis

To exclude endotoxin-mediated biological responses, the absence of endotoxin contamination in particle samples was confirmed using a Limulus Amebocyte Lysate (LAL) assay kit (Lonza, Walkersville, MD, USA).

Solubility of test particles in simulated body fluids

To evaluate the solubility of NAO_2.5 and LM_2.5 in simulated body fluids, Dulbecco’s phosphate-buffered saline (DPBS), representing a neutral physiological condition, and artificial lysosomal fluid (ALF) were prepared according to our previous study [17]. In addition, the solubility of the test particles was further evaluated under the same DW dispersion conditions used for the in vivo study. Test particles (1 mg/mL) dispersed in DPBS (pH 7.4), ALF (pH 5.5), or DW were incubated at 37 °C for 24 h. After incubation, particle-free supernatants were collected by three successive centrifugation steps (1 h each) at 21,000 × g for NAO_2.5, LM_2.5, and Fe₂O₃, or at 60,000 × g for SRM 2975. The centrifugation conditions were selected based on preliminary tests to ensure complete particle removal. The concentrations of inorganic elements in the supernatants were quantified using ICP-OES at the UNIST Central Research Facilities, using the identical methods described above (ICP-OES analysis for particle samples). A total of 9 elements–zinc (Zn), manganese (Mn), chromium (Cr), copper (Cu), iron (Fe), barium (Ba), molybdenum (Mo), titanium (Ti), and zirconium (Zr)–were selected as target analytes based on our previous X-ray fluorescence (XRF) analysis, which identified these elements as measurable constituents of NAO_2.5 and LM_2.5 [15].

Determination of intrinsic oxidative and nitrosative potential of test particles

To determine the intrinsic oxidative potential (IOP) and intrinsic nitrosative potential of the particles, the levels of reactive oxygen species (ROS) and reactive nitrogen species (RNS) directly generated by the particles were quantified under cell-free conditions. For the ROS assay, the intrinsic ROS-generating potential of test particles was evaluated using the cell-free 2′,7′-dichlorofluorescein diacetate (DCFH-DA; Sigma-Aldrich, St. Louis, MO, USA) assay, as previously described [18]. Test particles were dispersed in DW at concentrations ranging from 6.25 to 50 µg/mL following the same dispersion protocol used for the animal experiments. For the RNS assay, the intrinsic RNS-generating potential of test particles was evaluated by measuring levels of nitrite, a stable end product of nitrogen oxide (NO), using the Griess Reagent System (Promega, Madison, WI, USA). Detailed methods for measuring ROS and RNS are provided in the Supporting Information.

Pharyngeal aspiration of test particles

Pharyngeal aspiration was used as a practical method for hazard identification because it enables controlled delivery of a defined pulmonary dose and allows direct comparison of toxicological responses among particles. However, because it is a bolus administration method, it does not fully mimic inhalation exposure and may produce heterogeneous particle deposition in the lung. Six-week-old female BALB/c mice (Hana Bio, Gyeonggi-do, Korea) were acclimated for 1 week prior to the experiment. Animal protocols were reviewed and approved by the Institutional Animal Care and Use Committee of Dong-A University (No. DIACUC-25-14). Mice were housed and maintained in a micro-ventilation cage system (MVCS) under controlled conditions (temperature: 22 ± 1 °C; humidity: 50 ± 10%; 12-h dark/light cycle). Mice were anesthetized with isoflurane (Piramal Critical Care, Bethlehem, PA, USA) delivered via a rodent anesthesia system (VetEquip, Pleasanton, CA, USA), and particle suspensions were introduced into the lung by pharyngeal aspiration, achieved by placing the suspension at the base of the fully extended tongue. Each mouse received 50 µL of suspension, corresponding to doses of 25, 50, and 100 µg/mouse (1.25, 2.5, and 5 mg/kg body weight, respectively), while the vehicle control group received the same volume (50 µL/mouse) of DW. In this study, the tested dose of 100 µg/mouse was contextualized using the Multiple-Path Particle Dosimetry (MPPD) model to estimate the cumulative alveolar deposited dose under selected inhalation exposure scenarios, assuming particles with a mass median aerodynamic diameter of 600 nm. The input parameters used for the MPPD modeling are presented in Table S1 (see Supporting Information). Lung inflammation was evaluated using bronchoalveolar lavage fluid (BALF) and histological analysis at 1 and 7 days after a single pharyngeal aspiration. At each time point, six mice per group were used for BALF and histological analyses, including four for BALF and two for histological evaluation. For lung burden analysis, four additional mice per group were used at each time point.

BALF analysis and histological evaluation

To evaluate particle-induced lung inflammation, BALF was collected using our previously established method [16] and analyzed for cytological, biochemical, and cytokine parameters. Mice were euthanized by exsanguination under deep isoflurane anesthesia. After opening the thoracic cavity, the trachea was intubated with a stainless-steel catheter and lavaged four times using 700 µL of ice-cold DPBS. The first lavage fluid was collected separately and used to measure lactate dehydrogenase (LDH), total protein, and cytokines. LDH activity and total protein concentration were quantified using an LDH assay kit (Roche Diagnostics, Mannheim, Germany) and a bicinchoninic acid (BCA) assay kit (Thermo Fisher Scientific, Waltham, MA, USA), respectively. Pro-inflammatory cytokines, including tumor necrosis factor-α (TNF-α), interleukin-1β (IL-1β), and IL-6, were measured using DuoSet ELISA kits (R&D Systems, Minneapolis, MN, USA). Cells collected from all four lavages were pooled by centrifugation at 375 × g for 5 min and resuspended in 1 mL of DPBS containing 10% fetal bovine serum (FBS). The total number of nucleated cells in BALF was determined using a NucleoCounter (Chemometec, Allerod, Denmark). Cells were cytocentrifuged onto glass slides at 4 × 10⁴ cells per slide using a cytocentrifuge (Hanil, Seoul, Korea) and stained with Diff-Quik (Thermo Fisher Scientific, Waltham, MA, USA). Differential cell counts were performed under a light microscope (Nikon, Tokyo, Japan) by evaluating 300 cells per slide based on cellular morphology. Lung tissues were fixed in 10% neutral-buffered formalin, followed by routine dehydration, paraffin embedding, and sectioning at a thickness of 4 μm at the Neuroscience Translational Research Solution Center (Busan, Korea). Tissue sections were stained with hematoxylin and eosin (H&E) for general histopathological evaluation.

Lung burden analysis to evaluate the lung clearance pattern of test particles

Lung burdens of BWPs and reference materials were determined at day 0 (immediately after aspiration), day 1, and day 28 post-aspiration using our previously described method [19]. Detailed methods for proteinase K (PK)-based recovery of particles deposited in lung tissue and their quantification by UV–Vis spectrophotometry are provided in the Supporting Information. Particle concentrations were quantified by measuring absorbance at 750 nm using a UV-Vis spectrometer (Lambda 365; PerkinElmer, Waltham, MA, USA), based on a standard calibration curve generated from the corresponding original particles dispersed in DW. The standard curves for NAO_2.5, LM_2.5, Fe₂O₃, and SRM 2975 exhibited excellent linearity with R² values (R² = 1.0000 for NAO_2.5 and SRM 2975; R² = 0.9999 for LM_2.5 and Fe₂O₃), indicating reliable quantification over the tested concentration ranges (Figs. S1A-D, see Supporting Information). Lung tissues from non-exposed control mice were subjected to the same PK digestion and recovery procedure as particle-exposed lungs. The resulting pellets were re-dispersed in DW after centrifugation, and the absorbance values were used as background signals attributable to the tissue matrix. These values were subtracted from those of particle-exposed lung samples, and the corrected absorbance values were used to calculate the final particle burden based on the corresponding standard curves. The detection limits estimated from the standard curve were approximately 16, 24, 12, and 1.2 µg/mL for NAO_2.5, LM_2.5, Fe₂O₃, and SRM 2975, respectively. Quantification accuracy ranged from 87% to 94% with minimal variability across particle types (Fig. S1E). Recovery experiments were performed by spiking test particles into 20 mg of dried lung tissue homogenate at final concentrations of 62.5 µg/mL for NAO_2.5, LM_2.5, and Fe₂O₃, and 12.5 µg/mL for SRM 2975. The resulting recovery values were 99% for NAO_2.5, 106% for LM_2.5, 90% for Fe₂O₃, and 104% for SRM 2975 (Fig. S1F). Retention was expressed relative to day 0 (the initial delivered dose) rather than the nominal administration dose.

Evaluation of the PAH-related gene expression in lung tissues

The expression of two cytochrome P450 genes, Cyp1a1 (TaqMan assay ID: Mm00487218) and Cyp1b1 (TaqMan assay ID: Mm00487229), was evaluated in lung tissues 24 h after aspiration at a dose of 100 µg/mouse, as these genes are involved in the aryl hydrocarbon receptor (AhR)-mediated metabolic activation of PAHs [20, 21]. Glyceraldehyde-3-phosphate dehydrogenase (Gapdh; TaqMan assay ID: Mm99999915_g1) was used as the internal control because its raw Ct values showed only limited variation across the experimental groups under the present conditions. Detailed methods are provided in the Supporting Information.

Evaluation of soluble fraction-mediated inflammogenic potential of test particles

Given that particles deposited in the lung may release organic and inorganic toxic constituents, leachates from particle suspensions were evaluated for their inflammogenic potential as a combined chemical mixture. Inorganic components were confirmed by ICP-OES analysis, and further chemical identification was planned only if significant inflammatory responses were observed. To collect leachates, test particles were suspended in DW at 1 and 4 mg/mL and incubated at 37 °C for 24 h. After incubation, the suspensions were centrifuged three times for 1 h each to remove the particles, and the resulting particle-free supernatants (leachates) were collected for in vivo experiments. Seven-week-old female BALB/c mice were anesthetized with isoflurane (Piramal Critical Care) delivered via a rodent anesthesia system (VetEquip). A volume of 50 µL of the collected leachates was administered to the lungs by pharyngeal aspiration, and lung inflammation was evaluated by BALF analysis 24 h after exposure.

Statistical analysis

All statistical analyses were performed using GraphPad Prism software (ver. 10.6.0; GraphPad Software, La Jolla, CA, USA). Comparisons among multiple groups were conducted using one-way analysis of variance (ANOVA), followed by Tukey’s post hoc test. Data are presented as mean ± standard deviation (SD). Formal normality testing was not performed because of the small sample size in each group. The n values varied depending on the experimental endpoint; for toxicological endpoints derived from animal experiments, n indicates biological replicates (i.e., the number of mice), whereas for physicochemical analyses, n refers to the number of independent measurements. The exact n value for each experiment is provided in the corresponding figure legend or table footnote. A p-value less than 0.05 was considered statistically significant. Spearman’s rank correlation analysis was performed to assess the relationships between physicochemical parameters, including oxidative potential, and toxicity-related endpoints.

Results

Physicochemical properties of test particles

Primary size, surface area, hydrodynamic size, zeta potential, and endotoxin levels

Table 1 summarizes the primary particle size, BET surface area, hydrodynamic size, and zeta potential of the test particles measured in this study. NAO_2.5 and LM_2.5 particles exhibited heterogeneous morphologies, whereas the reference materials displayed relatively homogeneous morphologies (Fig. 1A). The primary particle sizes of NAO_2.5 and LM_2.5 showed broad distributions, with mean diameters of 511.7 ± 195.5 nm and 611.3 ± 277.9 nm, respectively. While Fe₂O₃ and SRM 2975 exhibited narrow size distributions, with mean diameters of 12.3 ± 2.8 nm for Fe₂O₃ and 27.1 ± 5.2 nm for SRM 2975 (Fig. 1B). Consistent with these characteristics, the BET specific surface area of BWPs was relatively low (5–7 m²/g), whereas Fe₂O₃ and SRM 2975 particles showed substantially higher surface areas (67–91 m²/g). The measured BET for SRM 2975 (90.80 m²/g) was very similar to the NIST reference value (91 m²/g). The hydrodynamic sizes of NAO_2.5 and LM_2.5 particles were comparable to their primary particle sizes, indicating stable dispersion in DW. In contrast, Fe₂O₃ and SRM 2975 exhibited larger hydrodynamic diameters (303 and 159 nm, respectively) than their primary sizes, suggesting partial agglomeration. The zeta potentials of all test particles in DW were negative, ranging from −20 to −40 mV. Endotoxin levels in all test samples were below the detection limit (< 0.1 EU/mL).

Table 1.

Physicochemical properties of brake wear particles (NAO_2.5 and LM_2.5) and reference materials (Fe₂O₃ and SRM 2975)

Measurements	NAO2.5	LM2.5	Fe2O3	SRM 2975
Primary size (nm)	511.7 ± 195.5	611.3 ± 277.9	12.3 ± 2.8	27.1 ± 5.2
Specific surface area (m2/g)	5.42	6.62	67.25	90.80
Hydrodynamic size (nm)	511.9 ± 25.2	432.1 ± 17.4	302.9 ± 9.5	159.3 ± 1.8
Polydispersity	0.15 ± 0.07	0.45 ± 0.06	0.38 ± 0.05	0.15 ± 0.01
Zeta potential (mV)	−31.33 ± 2.48	−22.18 ± 1.25	−20.50 ± 0.59	−40.55 ± 0.50
Endotoxin	ND	ND	ND	ND

Data are expressed as mean ± standard deviation (SD); n = 4 for hydrodynamic size, polydispersity index, and zeta potential measured using a Zetasizer Nano ZS. Primary particle size was determined from TEM images using ImageJ based on measurements of 100 individual particles per sample (n = 100). Endotoxin levels were measured using the Pierce Chromogenic Endotoxin Quant Kit and analyzed once for each sample (n = 1). ND, not detected (detection limit: < 0.1 EU/mL)

Fig. 1.

Physicochemical properties of test particles. (A) Transmission electron microscopy (TEM) images (scale bars: 2 μm for NAO_2.5 and LM_2.5, 50 nm for Fe₂O₃, and 200 nm for SRM 2975). (B) Particle size distributions. The size distribution for each sample was determined by measuring 100 individual particles from TEM images using ImageJ. (C) Elemental fraction (%) of each element relative to the total mass in NAO_2.5 and LM_2.5. (D) Concentration of polycyclic aromatic hydrocarbons (PAHs). (E) Solubility of NAO_2.5 and LM_2.5. (F) Fourier transform infrared (FTIR) spectra of NAO_2.5 and LM_2.5 particles. Raman spectra of (G) NAO_2.5 and (H) LM_2.5. Characteristic carbonaceous features were observed at the D band (~ 1312–1315 cm^− 1), G band (~ 1546–1566 cm^− 1), and 2D band (~ 2681 cm^− 1). In addition, a low-wavenumber Raman feature at approximately 240 cm^− 1 was observed exclusively in LM_2.5, which is attributable to metal oxide-related lattice vibrations

Inorganic and organic composition profiles of test particles

TEM–EDS analysis was used to identify the elemental constituents present in selected particle areas, whereas ICP-OES was used for bulk quantitative analysis of inorganic elements in whole particle samples. TEM–EDS analysis revealed that both NAO_2.5 and LM_2.5 particles contained Fe, O, C, Cu, Ti, Mg, Al, Zr, Ca, Si, S, Zn, and K (Table S2, see Supporting Information). Based on ICP-OES analysis, the major inorganic elements accounting for more than 5% of the total inorganic mass were, Fe (57.92%), Ba (12.9%), Ca (8.49%), and Cu (7.05%) in NAO_2.5, whereas those in LM_2.5 were Fe (68.35%), Cu (9.61%), and Zn (6.1%) (Fig. S2 and Table S3, see Supporting Information). When expressed relative to total particle mass, Fe accounted for 21.55% and 24.79% of NAO_2.5 and LM_2.5, respectively (Fig. 1C), and the remaining fraction was presumed to consist mainly of organic/polymeric and carbonaceous components. The purity of the Fe₂O₃ particles was higher than 99.5%, and the elemental composition of SRM 2975 was predominantly black carbon, as reported by the manufacturers. PAH analysis showed that SRM 2975 contained multiple PAHs at appreciable levels, while NAO_2.5 and LM_2.5 shared largely overlapping PAH profiles with substantially lower PAH levels. In particular, chrysene (Chry) and benzo[b]fluoranthene (B[b]F) levels in the BWPs were approximately 10-fold lower than those in SRM 2975, while indeno[1,2,3-cd]pyrene (I[c, d]P) was approximately 6-fold lower. Detailed PAH species and concentrations are summarized in Fig. 1D and Table S4 (see Supporting Information).

Metal release of test particles in simulated body fluid

The elemental release of metals from the test particles was evaluated after 24 h of incubation in DPBS, ALF, and DW at 37 °C. NAO_2.5 and LM_2.5 particles exhibited minimal elemental release in DPBS, with most elements detected at levels below 1% of the total particle mass (Fig. 1E and Table S5, see Supporting Information). In contrast, incubation in ALF resulted in higher Fe release, accounting for 3.5% and 7.8% of the total particle mass in NAO_2.5 and LM_2.5, respectively (Fig. 1E). Elemental release from NAO_2.5 and LM_2.5 particles in DW was also below 1% of the total particle mass (Table S5, see Supporting Information).

Crystalline phase and carbon structure of test particles

XRD analysis revealed that α-Fe₂O₃ was the predominant crystalline phase in NAO_2.5, with diffraction peaks well matched to standard α-Fe₂O₃ reference patterns (JCPDS Nos. 33–0664, 00–024-0072) [22, 23]. In contrast, LM_2.5 exhibited a mixed iron oxide phase, characterized by a diffraction peak corresponding to both Fe₃O₄ and γ-Fe₂O₃ (JCPDS Nos. 85–1436 and 89–5892) [24, 25]. Because Fe₃O₄ and γ-Fe₂O₃ share similar spinel structures and lattice parameters, unambiguous discrimination between the two phases based solely on XRD peak positions is not feasible. Detailed 2θ values and Miller indices are provided in Fig. S3 (see Supporting Information).

FTIR characterization of BWPs

The FTIR spectrum of NAO_2.5 exhibited a broad absorption band at 3336 cm^− 1, corresponding to the hydroxyl (–OH) stretching vibrations (Fig. 1F). Absorption bands at 2928 and 2862 cm^− 1 were attributed to asymmetric C–H stretching vibrations, while peaks at 1647 cm^− 1 (C = O stretching), 1336 cm^− 1 (O–H bending), 1185 cm^− 1 (C–O stretching), and 1101 cm^− 1 (C–O–C stretching) were associated with organic binders such as phenolic resin. A distinct band at 614 cm^− 1 represented Fe–O stretching vibrations, and absorption features near 450 cm^− 1 suggested the presence of additional metal oxides, potentially including barium-, calcium-, and copper-containing compounds. In contrast, the FTIR spectrum of LM_2.5 showed a similar overall pattern but exhibited markedly stronger absorption bands below 700 cm^− 1, particularly within the Fe–O stretching region, consistent with its higher metal oxides content (Fig. 1F). Furthermore, the FTIR results indicate that both NAO_2.5 and LM_2.5 contain organic binders and metallic fillers; however, NAO_2.5 is relatively enriched in organic constituents, whereas LM_2.5 comprises a greater proportion of metallic components.

Raman spectral analysis

The Raman spectra of NAO_2.5 exhibited prominent D (~ 1312 cm^− 1) and G (~ 1566 cm^− 1) bands, indicating the presence of graphitic carbon (Fig. 1G). Compared with LM_2.5, the higher intensities of the D and G bands observed in NAO_2.5 are likely attributable to its greater carbonaceous contribution (Figs. 1G and H). The spectrum of NAO_2.5 was characterized mainly by carbon-related features, while distinct vibrational features attributable to metal oxides were not clearly observed below 700 cm^− 1 under the present measurement conditions (Fig. 1G). Similarly, the Raman spectrum of LM_2.5 displayed D (~ 1315 cm^− 1) and G (~ 1561 cm^− 1) bands, confirming the presence of graphitic carbon (Fig. 1H). In addition, a distinct peak at ~ 240 cm^− 1, Fe–O lattice vibrations, indicated the presence of iron oxide species. This observation is consistent with the high Fe content of LM_2.5 quantified by ICP-OES (Fig. 1C).

Oxidative and nitrosative reactivity (ROS and RNS) of test particles

Intrinsic (particle-generated) ROS measurements showed that all test particles produced significant ROS levels in a concentration-dependent manner (Fig. 2). However, the relative ranking of ROS-generating potency among the particles differed markedly when normalized by mass versus surface area. On a mass basis, NAO_2.5 exhibited the highest ROS-generation potency, followed by SRM 2975, LM_2.5, and Fe₂O₃ (Fig. 2A). Notably, when normalized to surface area, the ranking of ROS-generating potency shifted, with NAO_2.5 remaining the highest, followed by LM_2.5, SRM 2975, and Fe₂O₃ (Fig. 2B). To investigate the effects of BWPs and their leachates on RNS generation, we measured nitrite, a stable breakdown product of NO. Particle suspensions incubated for up to 48 h showed no significant change compared with the baseline control (data not shown). In contrast, leachates prepared from NAO_2.5 and LM_2.5 in RPMI-1640 medium at 250 µg/mL after 48 h of incubation contained approximately 7–10 µM nitrite, which was higher than that of the reference particles (≈ 0.5 µM) (Fig. 2C).

Fig. 2.

Oxidative and nitrosative reactivity of test particles. (A) Intrinsic reactive oxygen species (ROS) generation potential of the test particles expressed on a mass basis at concentrations of 6.25, 12.5, 25, and 50 µg/mL. (B) ROS generation potential of the test particles normalized to particle surface area concentration (n = 4). (C) Reactive nitrogen species (RNS) levels in particle leachates following incubation at 250 µg/mL (n = 2). Data are expressed as mean ± standard deviation (SD). Statistical significance was evaluated within each particle group by one-way ANOVA followed by Tukey’s post hoc test. ^*p < 0.05 versus the lowest concentration of each particle. ^#p < 0.05 versus the vehicle control (VEH) group. Statistical analysis was not performed for panel B because the converted surface area doses were different from each other

Comparative acute lung inflammation potential of test particles

At day 1 post-treatment, the total cell counts in BALF significantly increased in the SRM 2975 treatment group in a dose-dependent manner compared to the vehicle control (VEH) group, whereas no significant changes were observed in the other treatment groups (Fig. 3A). All test particles significantly decreased the percentage of macrophages among total BALF cells in a dose-dependent manner compared with VEH (Fig. 3B). The percentage of neutrophils in BALF was markedly increased, in the descending order of SRM 2975 > NAO_2.5 > Fe₂O₃ > LM_2.5, on a mass dose basis (Fig. 3C). When evaluated on a surface area dose basis, however, the order shifted to NAO_2.5 > LM_2.5 > SRM 2975 > Fe₂O₃ (Fig. 3D). Neutrophilic inflammation was significantly greater in the NAO_2.5 group than in the LM_2.5 group. All test particles induced modest, dose-dependent elevations of LDH and total protein levels in BALF relative to VEH (Figs. 3E and F). Notably, LDH levels were significantly higher in the NAO_2.5 high-dose group (100 µg/mouse) compared with LM_2.5. Among tested pro-inflammatory cytokines (IL-1β, IL-6, and TNF-α), only IL-1β levels in BALF were significantly increased in NAO_2.5 and SRM 2975 treatment groups at day 1, whereas others showed no significant changes (Figs. 3G–I). Notably, the percentage of neutrophils showed a strong positive correlation with IL-1β levels in BALF (Fig. 3J). By contrast, nitrite was not detected in BALF from any group, with a detection limit of 1.56 µM (data not shown).

Fig. 3.

Bronchoalveolar lavage fluid (BALF) analysis at day 1 after pharyngeal aspiration of test particles in mice. The treatment doses were 25, 50, and 100 µg/mouse. (A) Number of total cells in BALF. (B) The percentage of macrophages. (C) The percentage of neutrophils. (D) The percentage of neutrophils in BALF on a surface area dose basis. Surface area dose was calculated using the specific surface area (BET) of the test particles. The levels of (E) lactate dehydrogenase (LDH) and (F) total protein in BALF. (G–I) The levels of (G) TNF-α, (H) IL-6, and (I) IL-1β in BALF of the high dose group (100 µg/mouse). (J) Correlation plot of IL-1β levels versus neutrophil percentages in BALF, analyzed using Spearman’s rank correlation. Data are expressed as mean ± standard deviation (SD) and n = 4. ^*p < 0.05 compared with vehicle control (VEH), ^#p < 0.05 compared with NAO_2.5 treatment group. Statistical analysis was performed by one-way ANOVA with Tukey’s post hoc test

Resolution of acute pulmonary inflammatory responses and lung clearance kinetics of test particles

To evaluate the persistence of acute pulmonary inflammatory responses induced by test particles, lung inflammation was assessed 7 days after a single pharyngeal aspiration of 100 µg/mouse. BALF analysis showed that the inflammatory responses observed at day 1 had largely resolved by day 7 in all groups except SRM 2975, where residual neutrophilic inflammation remained, with neutrophils representing approximately 10% of total BALF cells (Figs. 4A–E). The levels of the tested cytokines, including IL-1β, IL-6, and TNF-α, did not show any significant changes (data not shown). All particles remained detectable in the lung with retention above 80% at day 1 post-aspiration of test particles (Fig. 4F). At 28 days post-exposure, the clearance of NAO_2.5 and LM_2.5 was significantly delayed relative to SRM 2975, with 75% and 67% of the deposited mass remaining, respectively, compared with 47% for SRM 2975. Fe₂O₃ retained 63% of particles, although this was not statistically different from SRM 2975. The retention percentages and deposition masses from the lung burden analysis are presented in Table S6 (see Supporting Information).

Fig. 4.

The lung inflammation of test particles at 7 days post-treatment and their pulmonary clearance kinetics of test particles. Lung inflammation was evaluated by bronchoalveolar lavage fluid (BALF) analysis at 7 days after pharyngeal aspiration of test particles at 100 µg/mouse. (A) The number of total cells in BALF. (B, C) The percentage of (B) macrophages and (C) neutrophils in BALF. (D, E) The levels of (D) lactate dehydrogenase (LDH) and (E) total protein in BALF. (F) Lung burden analysis data, expressed as particle retention over time relative to day 0 (immediately after exposure). Retention was measured at days 0, 1, and 28 after a single pharyngeal aspiration in mice (100 µg/mouse). Data are expressed as mean ± standard deviation (SD) and n = 4. ^*p < 0.05 compared with vehicle control (VEH); ^#p < 0.05 compared with SRM 2975 group (one-way ANOVA followed by Tukey’s post hoc test)

Histopathological changes in lungs

Lung tissues collected 24 h after pharyngeal aspiration showed mild infiltration of inflammatory cells in the alveoli, along with particle-laden alveolar macrophages (Fig. 5). In contrast, lung tissues examined 7 days after a single pharyngeal aspiration showed complete resolution of alveolar inflammatory cell infiltration, whereas particle-laden alveolar macrophages persisted (Fig. S4, see Supporting Information).

Fig. 5.

Histopathological changes in lung tissues at 24 h after pharyngeal aspiration of test particles. The tissue sections were stained with hematoxylin and eosin (H&E). (A) Vehicle control (VEH). (B, C) Lung tissue from mice exposed to NAO_2.5, shown at (B) low magnification and (C) higher magnification to indicate particle-laden alveolar macrophages (arrow). (D–F) Lung tissue from mice exposed to (D) LM_2.5, (E) Fe₂O₃, and (F) SRM 2975. Arrows indicate particle-laden alveolar macrophages. No severe histopathological changes were observed in the lung tissues of the high-dose group (100 µg/mouse). Br, bronchiole; Al, alveoli; BV, blood vessel

PAH-related gene expression in lung tissues

The expression levels of Cyp1a1 and Cyp1b1 in lung tissues showed no significant differences between the treatment groups and VEH groups (Fig. S5, see Supporting Information). Although SRM 2975 contained relatively higher levels of PAHs (Fig. 1D), these levels were insufficient to elicit PAH-associated transcriptional responses under the present exposure conditions. In addition, NAO_2.5 and LM_2.5, which contained substantially lower PAH contents and distinct PAH compositions compared with SRM 2975, also did not induce detectable PAH-related metabolic activation in the lung.

Contribution of oxidative potential to the toxicity of test particles

Intrinsic ROS generation by each particle showed strong positive correlations with multiple toxicity endpoints, including LDH and total protein levels in BALF (Fig. 6). In contrast, inflammatory cell–based endpoints showed weaker associations; although the neutrophil percentage correlated moderately with intrinsic ROS generation (r = 0.55), this relationship did not reach statistical significance (p = 0.07). To further examine this relationship, toxicity endpoints were replotted against ROS-normalized doses, expressed as ROS-normalized mass dose (µg × ROS potential at each dose level) and ROS-normalized surface area dose (cm² × ROS potential at each dose level). A simple plot of surface area dose versus neutrophil percentages showed that each particle exhibited a distinct dose-response (Fig. 6E). In contrast, normalization by ROS-generating capacity markedly reduced inter-particle variation, resulting in a much narrower separation among particles; under these conditions, ROS-normalized surface area dose showed a strong positive correlation with neutrophil percentage for Fe₂O₃ and SRM 2975 (r = 0.94, p < 0.05; Fig. 6F). Similarly, compared with mass-based dose alone (Fig. 6G), normalization by ROS-generating capacity reduced inter-particle variability, resulting in a strong positive correlation between ROS-normalized mass dose and neutrophil percentage for Fe₂O₃ and SRM 2975 (r = 0.89, p < 0.05; Fig. 6H). In contrast, no significant correlations were detected for either ROS-normalized metrics in NAO_2.5 and LM_2.5. (Figs. 6F and H). These results indicate that particle-specific factors (e.g., soluble organic toxicants) associated with NAO_2.5 and LM_2.5 contribute to their toxicity responses even after accounting for ROS-generating capacity.

Fig. 6.

Evaluation of the contribution of oxidative potential to the toxicity of test particles. Intrinsic oxidative potential (IOP) was plotted against toxicity endpoints in bronchoalveolar lavage fluid (BALF). (A–D) Correlation plots between IOP of each particle and (A) total cell counts, (B) lactate dehydrogenase (LDH) levels, (C) total protein levels, and (D) neutrophil percentages. (E–H) Neutrophil percentages in BALF were further plotted against (E) surface area (SA)-based dose, (F) ROS-normalized SA-based dose (cm² × IOP at each dose level), (G) mass-based dose, and (H) ROS-normalized mass-based dose (µg × IOP at each dose level). Notably, reference particles (REF; Fe₂O₃ and SRM 2975) showed significant positive correlations between ROS-normalized SA- or mass-based doses and neutrophil percentages, whereas no significant correlations were observed for either ROS-normalized metrics in brake wear particles (BWPs; NAO_2.5 and LM_2.5). Data are expressed as mean ± standard deviation (SD) and n = 4. Correlations were assessed using Spearman’s rank test

Contribution of particle leachates to the lung inflammatory potential of test particles

Based on these observations, we next examined whether soluble components derived from particles contribute to the lung inflammatory potential of BWPs. To evaluate the contributory role of particle leachates as potential additional toxicity-determining factors, leachates from NAO_2.5 and LM_2.5 were generated in DW, and their effects on lung inflammation were assessed using the mouse pharyngeal aspiration model. The tested leachate concentrations were 1 and 4 mg/mL in DW, corresponding to 0.5- and 2-fold of the particle concentration (2 mg/mL) used in the animal experiments. These concentrations were selected to evaluate the potential effects of particle-derived leachates under conditions relevant to the in vivo exposure. The results showed that leachates from both BWPs did not significantly alter the measured BALF inflammatory endpoints compared with the VEH group (Figs. 7A–D).

Fig. 7.

Pulmonary inflammatory responses to leachates derived from brake wear particles (BWPs). Leachates were prepared by incubating NAO_2.5 and LM_2.5 particles in distilled water (DW) at 1 and 4 mg/mL at 37 °C for 24 h, followed by centrifugation to remove particles. The resulting particle-free leachates were administered to mice via pharyngeal aspiration. Pulmonary inflammation was assessed by bronchoalveolar lavage fluid (BALF) analysis. (A–D) Representative BALF parameters, including (A) total cell counts, (B) neutrophil percentages, (C) lactate dehydrogenase (LDH) levels, and (D) total protein levels. Data are presented as mean ± standard deviation (SD) (n = 4 for 1 mg/mL and n = 3 for 4 mg/mL)

Discussion

Non-exhaust PM has emerged as a critical public health concern, particularly as exhaust emissions have been substantially reduced under stringent regulations such as Euro 6, and the forthcoming Euro 7 standard extends regulatory attention to non-exhaust sources [4]. In this context, the present study provides comparative in vivo evidence that inhalable BWPs, generated under realistic braking conditions, induce acute lung inflammation with particle type–dependent potency in a mouse pharyngeal aspiration model used for comparative toxicological interpretation. Using BWPs generated under controlled dynamometer braking, we demonstrate that both NAO_2.5 and LM_2.5 elicit inflammatory responses in the lung. These responses were partly associated with their IOP, but the overall response pattern also suggested contributions from additional particle-specific properties beyond ROS generation alone. These findings offer new insights into the toxicological relevance of non-exhaust PM and highlight the need to consider physicochemical heterogeneity when assessing health risks associated with BWPs.

The present study demonstrated that the relative inflammogenic potency of the tested particles was highly dependent on the applied dose metric, with normalization by surface area revealing a markedly different toxicity pattern than that observed on a mass basis. This shift suggests that particle surface area may serve as an additional biologically relevant dose metric for comparative interpretation [26–28], given that interactions at the bio–nano interface largely govern particle–cell interactions and downstream biological responses [29]. From this perspective, the greater inflammatory potency of BWPs than that of the reference particles under surface area-normalized conditions may indicate that BWPs possess a higher intrinsic toxic potential per unit surface area than Fe₂O₃ nanoparticles or DEP (SRM 2975). However, this interpretation should be made cautiously, as in vivo responses may also be influenced by other factors, including agglomeration state, particle number, and protein corona formation in the lung. Further studies are needed to clarify the relative contribution of these factors to the observed in vivo responses.

Characterizing the key physicochemical parameters that govern particle toxicity is essential for understanding underlying toxic mechanisms and for developing preventive or therapeutic strategies to mitigate exposure-related health risks [30, 31]. In the present study, SRM 2975 was included as a diesel exhaust reference particle, and its acute neutrophilic effect and dose-response were generally consistent with a previous mouse study [32]. The present dataset indicates that intrinsic oxidative potential was associated with several lung injury-related endpoints [33, 34], but did not fully explain the overall inflammatory potency of BWPs. ROS-normalized analyses should be interpreted cautiously, as they indicate association rather than causation and do not establish ROS generation as the sole mechanistic driver of the observed in vivo responses. These findings suggest that additional particle-specific factors, including fragmented morphology and the organic/polymeric component, may have contributed to BWP toxicity beyond that predicted by oxidative potential alone [18].

Based on the present findings, several physicochemical factors did not appear to be major contributors to the toxic potential of BWPs under the present experimental conditions, including RNS, PAHs, and soluble fraction. Although nitrite concentrations showed an increasing trend in the BWP groups, the observed level (approximately 10 µM) was not considered indicative of a major nitrosative contribution to toxicity under the present experimental conditions [35, 36]. Although PAHs are well-established drivers of oxidative stress and inflammation [37], the present data suggest that the measured PAH alone was insufficient to explain the early inflammatory response pattern observed here. The lack of Cyp1a1 and Cyp1b1 induction further supports the conclusion that PAH levels in both BWPs and SRM 2975 were unlikely to be primary drivers of the observed inflammatory responses. Nevertheless, BWPs generated under harsh braking conditions—such as high-temperature, high-load, or prolonged braking scenarios—may contain elevated levels of PAHs and other organic toxicants, and under such circumstances, PAHs could contribute more substantially to particle-induced inflammatory responses. Therefore, while the present findings indicate that PAHs are unlikely to be the dominant drivers of lung inflammation under representative braking conditions, their contribution cannot be entirely ruled out and should be interpreted in the context of braking intensity and real-world exposure scenarios.

Only minimal metal release from BWPs was detected in physiological saline, whereas approximately 4–8% per day of iron dissolution was observed in ALF. This finding suggests that metal ion release from BWPs is largely restricted to the acidic phagosomal environment, where released iron ions could, in principle, contribute to oxidative stress via Fenton-type reactions [38]. However, the IOP measured in this study did not correspond to the extent of iron release. Specifically, NAO_2.5 exhibited a higher IOP than LM_2.5 despite showing lower iron dissolution, and Fe₂O₃ nanoparticles—despite being a direct source of iron—displayed the lowest IOP among all tested particles. These observations indicate that metal ion release from BWPs is unlikely to be a dominant determinant of their oxidative potential or toxicity. Moreover, iron is an essential element in biological systems and is tightly regulated through well-established homeostatic mechanisms, suggesting that the limited amount of iron released from BWPs under the tested conditions is unlikely to overwhelm cellular defense systems [39, 40]. In this context, the high IOP of BWPs may reflect contributions from multiple particle attributes, including their fragmented morphology [18], crystallinity [41], and the presence of reactive organic/polymeric components [42].

The significant correlation between IL-1β levels and neutrophil percentages in BALF observed in this study suggests that inflammasome-mediated signaling may represent a key pathway underlying the inflammatory responses induced by the test particles [43]. Inflammasome activation is typically triggered when phagocytosed particles destabilize the phagolysosomal membrane, resulting in the leakage of cathepsins and other lysosomal contents into the cytosol, which in turn stimulates NOD-like receptor family, pyrin domain containing 3 inflammasome assembly and the release of IL-1β [44]. ROS are also recognized as critical secondary signals that amplify inflammasome activation, either by directly modifying inflammasome components or by promoting mitochondrial dysfunction [45]. Taken together, particle-specific properties that influence lysosomal destabilization, such as surface reactivity, morphology, and organic constituents, likely play significant roles in determining the extent of inflammasome activation and subsequent neutrophilic inflammation [45, 46].

The MPPD modeling shown in this study indicated that this bolus dose corresponds to approximately 3 days of alveolar deposition following occupational exposure (8 h/day) at the recommended exposure limit (REL) of 5 mg/m³ for iron oxide as established by the National Institute for Occupational Safety and Health (NIOSH) [47]. Furthermore, modeling indicated that, assuming a PM_2.5 concentration of approximately 100 µg/m³ in a subway station of London, UK, and an exposure duration of 8 h/day, the administered dose corresponds to alveolar deposition achieved after approximately 166 days of exposure [48]. These MPPD-based comparisons were intended only to provide an approximate dosimetric context for the tested dose, rather than to imply direct equivalence between single bolus administration and real-world chronic inhalation exposure. Likewise, the resolution of lung inflammatory responses by 7 days after treatment for all tested particles suggests that the recovery pattern observed in the present study may not fully reflect that under inhalation exposure conditions, where particle deposition occurs through continuous or repeated low-dose exposures. However, although single bolus administration has inherent limitations, including likely heterogeneous particle deposition, this approach is useful as a simplified experimental model within the scope of the present study, which aimed to compare the relative inflammatory potential of BWPs and elucidate the underlying mechanisms.

Conclusions

Inhalable BWPs generated under realistic braking conditions exhibited substantial pulmonary inflammogenicity in vivo. On a mass basis, NAO_2.5 showed slightly lower inflammogenicity than SRM 2975, while LM_2.5 exhibited the lowest inflammogenicity among the tested particles. However, given that BWPs were larger and had substantially lower surface area than the reference particles, their greater toxicity under surface area-normalized conditions suggests higher intrinsic toxic reactivity per unit surface area. In addition, the BWPs examined in this study exhibited a toxicity profile that was not fully explained by intrinsic oxidative potential alone, indicating that particle-specific physicochemical properties contribute importantly to BWP toxicity. Although oxidative potential was associated with several toxicity endpoints, additional factors such as fragmented morphology, crystallinity, and reactive organic or polymeric components may also contribute, whereas the contributions of RNS, PAHs under representative braking conditions, and the soluble fraction appeared limited. However, because the present findings are limited to the specific NAO and LM samples tested here, further studies using a wider range of BWP samples are needed to better identify physicochemical properties related to their toxicity responses. Future studies should also consider compositional variability arising from diverse brake pad formulations and evaluate how environmental aging processes may modify the physicochemical properties and toxicity of BWPs.

Author contributions

**Gyuri Kim: ** Writing – original manuscript, Conceptualization, Investigation, Formal analysis, Data curation. **Soyeon Jeon: ** Investigation, Data curation. **Tae Hwan Shin, Yonghyeon Park, and Kiyun Kim: ** Investigation. **Sang-Hee Woo and Yun Suk Huh: ** Resources, Methodology. **Joon-Goo Lee, Seokhwan Lee, and Wan-Seob Cho: ** Writing – review & editing, Project administration, Supervision, Conceptualization.

Funding

This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (RS-2024-00336383 and RS-2025-00518101).

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

All animal experiments for this study were approved by the Institutional Animal Care and Use Committee of Dong-A University (approval number: DIACUC-25-14).

Competing interests

The authors declare no competing interests.