Copper-enriched automotive brake wear particles perturb human alveolar cellular homeostasis

Abstract

Background: Airborne fine particulate matter with diameter < 2.5 μm (PM2.5), can reach the alveolar regions of the lungs, and is associated with over 4 million premature deaths per year worldwide. However, the source-specific consequences of PM2.5 exposure remain poorly understood. A major, but unregulated source is car brake wear, which exhaust emission reduction measures have not diminished.

Methods: We used an interdisciplinary approach to investigate the consequences of brake-wear PM2.5 exposure upon lung alveolar cellular homeostasis using diesel exhaust PM as a comparator. This involved RNA-Seq to analyse global transcriptomic changes, metabolic analyses to investigate glycolytic reprogramming, mass spectrometry to determine PM composition, and reporter assays to provide mechanistic insight into differential effects.

Results: We identified brake-wear PM from copper-enriched non-asbestos organic, and ceramic brake pads as inducing the greatest oxidative stress, inflammation, and pseudohypoxic HIF activation (a pathway implicated in diseases associated with air pollution exposure, including cancer, and pulmonary fibrosis), as well as perturbation of metabolism, and metal homeostasis compared with brake wear PM from low- or semi-metallic pads, and also, importantly, diesel exhaust PM. Compositional and metal chelator analyses identified that differential effects were driven by copper.

Conclusions: We demonstrate here that brake-wear PM may perturb cellular homeostasis more than diesel exhaust PM. Our findings demonstrate the potential differences in effects, not only for non-exhaust vs exhaust PM, but also amongst different sources of non-exhaust PM. This has implications for our understanding of the potential health effects of road vehicle-associated PM. More broadly, our findings illustrate the importance of PM composition on potential health effects, highlighting the need for targeted legislation to protect public health.

Keywords: Alveolar; Brake-wear PM; Copper; Diesel PM; Epithelium; Hypoxia-inducible factor (HIF); Metallothionein; Non-exhaust emissions; Non-tailpipe emissions; Pseudohypoxic signalling.

Fig. 1

Overview of the study. Brake-wear PM from four different brake pad types, as well as diesel exhaust PM were collected in a laboratory setting using a high-volume cascade impactor onto foam filters. Foam filters were dried under nitrogen gas and PM was extracted using methanol. The composition of these PM were assessed using ICP-MS. Given that PM2.5–0.1 can reach the alveoli, we investigated the differential effects of the vehicle-derived PM types in a submerged culture of alveolar type-II epithelial cells. Designed using BioRender.com

Fig. 2

NAO and Ceramic brake-wear PM induce the greatest oxidative stress. ATII cells were exposed to 8, 16, and 32 µg/cm² of the 5 different vehicle-derived PM types for 24 h. A LDH release was measured, with higher LDH release indicating higher cytotoxicity; LDH release was determined using a CytoTox 96^® Non-Radioactive Cytotoxicity Assay kit, n = 3. B ROS generation was assessed in ATII cells via oxidation of H₂-DCF into the fluorescent DCF after 24 h of exposure to PM; tert-Butyl hydroperoxide (TBHP) was used as a positive control for ROS induction, n = 5. C ATII Haem Oxygenase-1 (HMOX1) expression was determined using RT-qPCR after exposure to 8 µg/cm² of the 5 different PM types, for 2, 6 and 24 h, n = 3 for 2 h exposure. n = 5 for 6 and 24 h exposure. D ATII cell Glutamate-Cysteine Ligase Modifier Subunit (GCLM) expression was determined using RT-qPCR after exposure to 8 µg/cm² of the 5 different PM types, for 2, 6 and 24 h, n = 3. Data was represented as mean + SEM, and a RM one-way ANOVA test was used with a Dunnett’s post-hoc test to determine significance compared to the control. * = p ≤ 0.05, ** = p ≤ 0.01, **** = p ≤ 0.0001

Fig. 3

Transcriptomic responses to PM are source-dependent, with NAO and Ceramic inducing the greatest number and magnitude of changes. ATII cells were exposed to 8 µg/cm² of the 5 different vehicle-derived PM types for 6 h, after which bulk RNA-Seq was conducted. The differential gene expression package edgeR was used to make pairwise comparisons between each of the five PM types and the control (e.g. LowM vs Ctrl, SemiMxCu vs Ctrl etc.…) to identify genes that were up or downregulated compared to the medium-only control. Genes with a false-discovery rate (FDR) p-value of ≤ 0.05 were considered differentially expressed. A Biplot showing the first two principal components, demonstrating that NAO and ceramic brake-wear PM are separated on PC1. B Heatmap of all genes that were differentially expressed in at least one of the 5 different pairwise comparisons, with hierarchical clustering of genes, generated using the ‘pheatmap’ package. C Number of DEGs for each PM type compared to the medium-only control. D Bubble plot showing Gene Set Variation Analysis (GSVA) scores for the top 10 most upregulated hallmark pathways by NAO brake-wear PM

Fig. 4

NAO- and ceramic-derived brake-wear PM strongly induce an inflammatory response and drive glycolytic reprogramming. ATII cells and primary ATII cells were exposed to 8 µg/cm² of the 6 different PM types for 24 h, after which various markers of inflammation were examined. A GSVA for Hallmark TNFα signalling via NFκB B ATII cell IL-6 protein secretion was determined after exposure to 8 µg/cm² of the 6 different PM types for 24 h. Determined via ELISA. C ATII cell IL-8 protein secretion was determined after exposure to 8 µg/cm² of the 6 different PM types for 24 h. Determined via ELISA. D GSVA for Hallmark Reactive Oxygen Species (Oxidative Stress Score). E GSVA for Hallmark Hypoxia. F GSVA of Panther Glycolysis pathway. G Heatmap visualising the expression of genes within the GSVA of Panther Glycolysis pathway. H ATP Production Rate % from mitochondria, and from glycolysis in the medium control and NAO BWPM. Negative control represents cells that were not exposed to PM. In A: Box contains median, upper, and lower quartiles, with whiskers representing the range. In B, C, and H: Bars represent mean + SEM. In A-F: A RM one-way ANOVA test was used with a Dunnett’s post-hoc test. In H, the ‘*’ represents a significantly increased glycoATP production rate in NAO compared to med ctrl, determined using a two-tailed paired t-test. Statistically significant values are indicated with the star notation on the graphs. * = p ≤ 0.05, ** = p ≤ 0.01, *** = p ≤ 0.001, **** = p ≤ 0.0001

Fig. 5

NAO- and ceramic-derived PM were most potent at perturbing metal ion homeostasis. ATII cells and primary ATII cells were exposed to 8 µg/cm² of the 6 different PM types for 2–24 h, after which various markers of metal ion homeostasis were examined. A GSVA of “Metallothioneins bind metals (Reactome)” pathway. B Heatmap of genes within the Metallothioneins bind metals (Reactome)” pathway. C Normalised counts for MT1G (counts per million, determined with same RNA-Seq dataset as Fig. 3). D Normalised counts for MT2A C. Normalised counts for MT1E. D. Normalised counts for MT1X. In A: Box contains median, upper, and lower quartiles, with whiskers representing the range. A RM one-way ANOVA test was used with a Dunnett’s post-hoc test. In C-F: Bars represent mean + SEM of the normalised counts, significant differences compared to the ctrl were determined using a generalised linear model with edgeR. Statistically significant values are indicated with the star notation on the graphs. ** = p ≤ 0.01, **** = p ≤ 0.0001

Fig. 6

Different vehicle-derived PM types have distinct elemental characteristics, with NAO and ceramic BWPM differentiated by high copper content. 5 different vehicle-derived PM types were collected using a high-volume cascade impactor, after which, the whole particles were digested and analysed via ICP-MS. A Heatmap showing element concentrations with hierarchical clustering, to group elements enriched in each bulk PM type. B A biplot comparing the first two principal components for the bulk PM. NAO and Ceramic are separated along PC2 (circled in red). C A loadings plot showing the top 15% of elements that are contributing to the direction of the PCs for the bulk PM. For example, high iron concentrations are shifting PC1 values higher (as in the case of SemiMxCu), or high copper concentrations are shifting PC2 values higher (as in the case of NAO and ceramic). D Copper concentration in bulk PM demonstrating higher copper concentrations in NAO and ceramic brake-wear PM

Fig. 7

Copper drives the observed effects of NAO brake-wear PM. ATII cells were exposed to 6 concentrations of NAO BWPM (0.5, 1, 2, 4, 8, and 16 µg/cm²) for 24 h, after which the supernatant was harvested and the cells were washed 3 times before lysis with Chelex-100 treated MilliQ H2O. The lysates were then analysed for their elemental composition using ICP-MS, to determine the concentrations of the metals in cells after NAO BWPM exposure. A Intracellular copper concentration following NAO BWPM exposure. n = 3. Following this, ATII cells were exposed to either metal chelators alone (TEPA 50 µM, DFX 12.5 µM, or TPEN 1 µM), NAO BWPM (8 µg/cm²) alone, or the NAO BWPM and metal chelators together (NAO + TEPA, NAO + DFX, or NAO + TPEN) for 24 h. Following this, HMOX1 gene expression (B), MT1G gene expression (C), and IL-6 protein secretion (D) were examined. n = 4. In all cases, bars represent mean + SEM, and a RM one-way ANOVA test was used with a Dunnett’s post-hoc test to test for significance. * = p ≤ 0.05. ** = p ≤ 0.01, *** = p ≤ 0.001. **** = p ≤ 0.0001

Fig. 8

NAO BWPM induces pseudohypoxic HIF pathway activation in a copper-dependent manner. GSVA was used to interrogate a 15-gene signature indicative of HIF transcription factor signalling to generate a HIF score A GSVA of a 15-gene set HIF score. B Heatmap of genes within the HIF Score. C Correlation between the HIF score and Oxidative Stress Score. D Correlation between [Cu] and Oxidative Stress Score. E–G: ATII cells were transfected with either a hypoxia response element (HRE) reporter firefly luciferase plasmid, or hypoxia-inducible factor α C-activation domain (HIFα-CAD) firefly plasmid, and both with a Renilla luciferase control plasmid using lipofectamine LTX. Cells were then exposed to 8 µg/cm² of NAO BWPM for 24 h and luciferase activity was determined using a Promega Dual-Luciferase^® Reporter Assay System. E. The mechanism of the HIF-CAD reporter assay. F. FIH inhibition following NAO BWPM exposure was determined using a HIFα-CAD reporter. n = 3. G. HRE binding following NAO BWPM exposure was determined using an HRE reporter. The plasmid contained an HRE binding region, and increased HIF transcription factor complex binding to this HRE binding site causes increased luciferase activity. PHD inhibitor DMOG was used as a positive control. n = 3. H and I: ATII cells were exposed to NAO brake-wear PM (8 µg/cm²), TEPA (50 µM), NAC (10 mM), or Ascorbate (5 mM) alone for 24 h. The cells were also exposed to NAO and TEPA together for 24 h, with no pre-treatment, NAO + NAC and NAO + Ascorbate refer to 24 h NAO exposure after a one-hour pre-treatment with respective agents. H. Representative HIF1α western blot showing the impact of NAO BWPM on HIF1α stabilisation, as well as the impact of copper chelator TEPA. I. Densitometric analysis of the western blots. n = 3. In F, G, and I data was represented as mean + SEM. In A: Box contains median, upper, and lower quartiles, with whiskers representing the range. In A and I, a RM one-way ANOVA test was used with a Dunnett’s post-hoc test. In C and D, a Pearson’s correlation was used. In F and G, and two-tailed paired t-test was used