Source differences in the components and cytotoxicity of PM2.5 from automobile exhaust, coal combustion, and biomass burning contributing to urban aerosol toxicity

Abstract

Although air quality guidelines generally use the atmospheric concentration of fine particulate matter (PM${}_{\mathrm{2.5}}$) as a metric for air pollution evaluation and management, the fact cannot be ignored that different particle toxicities are unequal and significantly related to their sources and chemical compositions. Therefore, judging the most harmful source and identifying the toxic component would be helpful for optimizing air quality standards and prioritizing targeted PM${}_{\mathrm{2.5}}$ control strategies to protect public health more effectively. Since the combustions of fuels, including oil, coal, and biomass, are the main anthropogenic sources of environmental PM${}_{\mathrm{2.5}}$, their discrepant contributions to health risks of mixed ambient aerosol pollution dominated by the respective emission intensity and unequal toxicity of chemical components need to be identified. In order to quantify the differences between these combustion primary emissions, 10 types of PM${}_{\mathrm{2.5}}$ from each typical source group, i.e., vehicle exhaust, coal combustion, and plant biomass (domestic biofuel) burning, were collected for comparative study with toxicological mechanisms. In total, 30 types of individual combustion samples were intercompared with representative urban ambient air PM${}_{\mathrm{2.5}}$ samples, whose chemical characteristics and biological effects were investigated by component analysis (carbon, metals, soluble ions) and in vitro toxicity assays (cell viability, oxidative stress, inflammatory response) of human lung adenocarcinoma epithelial cells (A549). Carbonaceous fractions were plenteous in automobile exhaust and biomass burning, while heavy metals were more plentiful in PM${}_{\mathrm{2.5}}$ from coal combustion and automobile exhaust. The overall ranking of mass-normalized cytotoxicity for source-specific PM${}_{\mathrm{2.5}}$ was automobile exhaust $\mathit{>}$ coal combustion $\mathit{>}$ domestic plant biomass burning $\mathit{>}$ ambient urban air, possibly with differential toxicity triggers, and showed that the carbonaceous fractions (organic carbon, OC; elemental carbon, EC) and redox-active transition metals (V, Ni, Cr) assisted by water-soluble ions (Ca${}^{\mathrm{2}+}$, Mg${}^{\mathrm{2}+}$, F${}^{-}$, Cl${}^{-}$) might play important roles in inducing cellular reactive organic species (ROS) production, causing oxidative stress and inflammation, resulting in cell injury and apoptosis, and thus damaging human health. Coupled with the source apportionment results of typical urban ambient air PM${}_{\mathrm{2.5}}$ in eastern China, reducing toxic PM${}_{\mathrm{2.5}}$ from these anthropogenic combustions will be greatly beneficial to public health. In addition to the air pollution control measures that have been implemented, like strengthening the vehicle emission standards, switching energy from coal to gas and electricity, and controlling the open incineration of agricultural straws, further methods could be considered, especially by preferentially reducing the diesel exhaust, lessening the coal combustion by replacement with low-ash clean coals, and depressing the rural crop straw biomass burning emissions.

1 Introduction

As a mixture of multiple sources, ambient particulate matter (PM) arising from anthropogenic activities is continuously deteriorating urban air quality, particularly in developing countries. Among these, fine PM with an aerodynamic diameter of less than 2.5 $\mathrm{\mu }$m (PM${}_{\mathrm{2.5}}$) is recognized as a serious public health concern due to its long persistence in air, carcinogenicity, and acute toxicity to humans (Al-Kindi et al., 2020). There was extensive epidemiological evidence that airborne PM can cause serious negative effects on human health, such as respiratory and cardiovascular diseases, genetic mutations, and developmental disorders (Chowdhury et al., 2022; Clemens et al., 2017; Lelieveld et al., 2021; Smith, 2021). Currently, either the world air quality guidelines or the national air quality standards use the mass concentration of PM${}_{\mathrm{2.5}}$ as a metric for PM${}_{\mathrm{2.5}}$ pollution evaluation and management. However, the particle toxicities are unequal and significantly related to their sources and chemical compositions varying with space and time (Shiraiwa et al., 2017). Therefore, identifying which component(s) and source(s) of ambient PM are most harmful to health will be helpful for evaluating air quality and prioritizing targeted PM control strategies for protecting public health more effectively.

In addition to natural sources, most aerosols come from anthropogenic activities, especially energy consumption including the combustion of fossil fuels causing industrial emissions and automobile exhaust, together with biomass burning (McDuffie et al., 2021; Wu et al., 2022). These diverse sources make the ambient air PM${}_{\mathrm{2.5}}$ a complex mixture with multiple chemical components, such as salts, organic carbon (OC), elemental carbon (EC), minerals, and trace metals (Bari and Kindzierski, 2016). The physiological mechanisms of PM-induced cell toxicity in respiratory systems have been continuously investigated with some progress (T. Li et al., 2022; Kelly and Fussell, 2012, 2020; Shiraiwa et al., 2017; Mack et al., 2019), such as metabolic activation, oxidative stress, inflammatory response, and apoptosis, which are focused on by the current study. Briefly, after inhalation and deposition onto the epithelium, redox-active materials in PM${}_{\mathrm{2.5}}$ can induce the release of reactive organic species (ROS), which causes oxidative stress (an imbalance between ROS and antioxidants, i.e., disequilibrium of the redox state of a cell), followed by inflammation and cell death. The ROS can mediate subsequent signaling pathways, leading to biomolecular damage (e.g., DNA, lipids, and proteins) and cellular injury by mediating inflammatory response, including the release of proinflammatory cytokines like IL-6 and TNF-$\mathit{\alpha }$ by epithelial cells (Ahmed et al., 2020; Landwehr et al., 2021). For instance, oxidative stress could trigger the induction of proinflammatory transcription factors, such as nuclear factor (NF)-$\mathit{\kappa }$B, via the mitogen-activated protein kinase (MAPK) signaling pathway. Components adsorbed on particle surfaces, such as redox-active metals (transition metals, Fe, Ni, V, Cr, Cu), organic compounds (polycyclic aromatic hydrocarbons, PAHs; quinones), or even carbonaceous cores of particles, are responsible for oxidative stress (Ahmed et al., 2020; Cachon et al., 2014). The non-redox-active metals (Zn, Pb, Al) can also influence the toxic effects of transition metals by exacerbating or lessening the production of free radicals. The EC may not be a directly toxic component of PM${}_{\mathrm{2.5}}$ but may rather operate as a universal carrier of combustion-derived chemicals (semivolatile organic fractions, transition metals) of varying toxicities (Kelly and Fussell, 2020). Inorganic soluble sulfates and nitrates are acidic and can interact with and influence the solubility of other compositions like metal bioavailability (Fang et al., 2017; Weber et al., 2016). However, well-known toxic pollutants in the environment like heavy metals and PAHs, whose specific components and particular sources are the most critical factors dominating ambient aerosol health risks, still need be explored.

Past studies performed in various countries have focused on physicochemical characterization or biological effects of ambient air PM${}_{\mathrm{2.5}}$, respectively (Weagle et al., 2018; Jia et al., 2017; Wang et al., 2020). For example, the source analysis of PM${}_{\mathrm{2.5}}$ by photochemical modeling (Bao et al., 2018), the chemical composition of regional PM${}_{\mathrm{2.5}}$ (Chi et al., 2022), and the mechanism of PM${}_{\mathrm{2.5}}$ toxicity were independently reported recently (Jia et al., 2020). Because differences in particle compositions, sources, and toxicities appear in different urban environments (Zhao et al., 2019; Borlaza et al., 2018), source profiles of different emission inventories are needed to elucidate the local aerosol pollution characteristics for control strategies. For instance, it was reported that increased hospital admission risks were significantly associated with sources of vehicle exhaust, coal combustion, and secondary inorganic aerosols. In particular, coal combustion was positively correlated with increases in mortality risks (Du et al., 2021). Coal combustion and vehicle exhaust contributed more significantly to cancer risks of respiratory exposure to atmospheric heavy metals in Tianjin in northern China during the cold seasons (31 % and 11 %) than during the warm seasons (11 % and 4 %) (Tian et al., 2021), while in Nanjing in eastern China, traffic emissions and non-traffic combustion (coal, waste, and biomass) contributed 35 % and 31 % to carcinogenic risks of urban PM${}_{\mathrm{2.5}}$-associated metals, respectively (Xie et al., 2020). Traffic was suggested to play the most crucial role in enhancing the toxicities of fine particles (Park et al., 2018). The particle composition of motor vehicle exhaust was related to automobile types with various fuels, engines, and loads (Lin et al., 2020). Strong catalytic reactivity of metals in PM emitted from diesel vehicles was observed by dithiothreitol (DTT) assay (Jesus et al., 2018). It was found that straw burning during the harvest season is a major trigger of severe air pollution in many regions (Sahu et al., 2021). Aerosols from open biomass burning in the Amazon had a stronger ability to induce ROS than laboratory-generated secondary organic aerosols (Tuet et al., 2019). Although there are emerging studies on particle emissions from single sources, quantitatively comparative studies on multisource pollutants as well as the differential composition and unequal toxicity of various sources are still limited.

The main objective of the current study was to compare the chemical components and corresponding mass-normalized toxicological effects of individual PM${}_{\mathrm{2.5}}$ from various combustion sources and their unequal contributions to ambient aerosol health risks. The aim is to provide experimental evidence supporting the targeted control of specific anthropogenic sources with prominent risks based on their pivotal toxic components. Therefore, we collected both representative ambient PM${}_{\mathrm{2.5}}$ samples ($n=\mathrm{16}$) from urban air and typical source PM${}_{\mathrm{2.5}}$ samples ($n=\mathrm{30}$) from automobile exhaust, coal combustion, and plant biomass burning. Their independent profiles of chemical compositions and in vitro cytotoxicity (cell viability, oxidative stress, and inflammatory response) were investigated and intercompared to assess the differences in source-to-receptor toxicity and to infer the core toxic components and respective harmful contributions. The pivotal toxic components were identified based on the source–sink bidirectional composition-effect results, which were further used to assess the health toxicity contributions of various emission sources to ambient air PM${}_{\mathrm{2.5}}$, supported by its source apportionment through positive matrix factorization (PMF) and chemical mass balance (CMB) models.

2 Materials and methods

2.1

Collection of PM${}_{\mathrm{2.5}}$ samples from primary emissions of 30 typical combustion sources and from representative ambient urban air

In total, 30 types of primary PM${}_{\mathrm{2.5}}$ samples emitted directly from automobile exhaust, coal combustion, and plant biomass (domestic biofuel) burning were respectively collected as follows for both chemical and toxicological analyses.

A total of 10 types of vehicles were chosen for exhaust investigation. They were further categorized into seven subgroups, including small-duty gasoline coaches (SDGCs), small-duty diesel coaches (SDDCs), middle-duty diesel coaches (MDDCs), heavy-duty diesel coaches (HDDCs), light-duty diesel vans (LDDVs), middle-duty diesel vans (MDDVs), and heavy-duty diesel vans (HDDVs). The detailed information on these representative local automobiles is shown in Table S1.

To cover all the coal types consumed in the city, 10 representative types of coal were gathered for investigation. They were further classified into four subgroups, including two types of honeycomb coal (HC), three types of anthracite coal (AC), two types of bituminous coal (BC) mainly for restaurant or household use, and three types of industrial coal (IC) for coal-fired power plants and the steel-smelting industry. The detailed characteristics of the physics and chemistry of these typical coals purchased from local markets are shown in Table S2.

Considering the plant biomass combustion in rural areas surrounding the megacity, 10 representative types of agricultural and forestry solid waste were gathered for investigation. Straws of rice, wheat, corn, soybean, peanut, rape, and sesame together with corncob and branches of peach and pine were selected as plant biomass fuels and further divided into two subgroups, including eight types of crop straw and two types of firewood. The detailed characteristic analysis of these typical plant biomass fuels collected from rural areas around Nanjing is shown in Table S3.

The PM${}_{\mathrm{2.5}}$ samples directly emitted from these combustion sources were collected by the dilution channel sampling method (Fig. S1) using a four-channel particulate matter dilution sampler (HY-805, Hengyuan Technology Development Co., CN). Each sample included three parallel channels of a quartz microfiber filter (Fig. S2) and one channel of a Teflon membrane filter with diameters of 47 mm through a size selector for PM${}_{\mathrm{2.5}}$ with a flow rate of 160 L min${}^{-\mathrm{1}}$ (each channel is 40 L min${}^{-\mathrm{1}}$). Clean air was pumped for 10 min before and after each sample was collected. Before use, the blank quartz filters were incinerated by a muffle furnace at 500 ${}^{\circ }$C for 3 h to remove any possible organic matter, while the Teflon filters were baked at 60 ${}^{\circ }$C for 4 h. After being equilibrated in a constant temperature and humidity chamber for 24 h, the filters were weighed both before and after sampling for gravimetric measurements, and then the mass of the collected PM${}_{\mathrm{2.5}}$ could be calculated. The sampled filters were stored in a refrigerator at $-$20 ${}^{\circ }$C before analysis. The quartz-filter-loaded PM${}_{\mathrm{2.5}}$ samples were used for carbon and ion analysis and for toxicity tests, while the parallel Teflon-filter-loaded samples were used to determine metals.

As the actual mixture of various source particles in a real environment, a total of 16 ambient air PM${}_{\mathrm{2.5}}$ samples (each time lasting 23 h) covering a year monthly were collected from December 2019 to October 2020 at an urban site surrounded by traffic, residential and commercial quarters of Nanjing, Yangtze River Delta of eastern China, using a high-volume air sampler (800 L min${}^{-\mathrm{1}}$) with quartz microfiber filters (H. Li et al., 2022).

2.2 Chemical composition analysis

All collected source and ambient PM${}_{\mathrm{2.5}}$ samples were conducted following component analysis (Li et al., 2023). For the concentrations of heavy metals in particulates, samples were digested by concentrated HNO${}_{\mathrm{3}}$–HClO${}_{\mathrm{4}}$ acids with a progressive heating program and determined by inductively coupled plasma optical emission spectrometry (ICP-OES; Optima8000, PerkinElmer, for Cr, Mn, Ni, and Pb), with elements (V, Co, As) at lower concentrations measured by ICP mass spectrometry (ICP-MS; NexIONTM300X, PerkinElmer). Blank filter, reagent blank, replicates, and standard reference material (NIST SRM 1648a, urban dust) were adopted for analytical quality control, with recoveries ranging from 90 % to 110 %. Carbonaceous species (OC and EC) in PM${}_{\mathrm{2.5}}$ were determined using DRI-2001A $\mathrm{OC}/\mathrm{EC}$ (Atmoslytic Inc., Calabasas, CA, USA). For the concentrations of water-soluble ions (WSIs), the main cations (Na${}^{+}$, K${}^{+}$, Mg${}^{\mathrm{2}+}$, Ca${}^{\mathrm{2}+}$, NH${}_{\mathrm{4}}^{+}$) and anions (NO${}_{\mathrm{3}}^{-}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, Cl${}^{-}$, F${}^{-}$) in PM${}_{\mathrm{2.5}}$ were measured by ion chromatography (IC, Thermo Fisher Scientific, USA) using the Metrosep C6-150/4.0 column for cations and the Metrosep A Supp 5 150/4.0 column for anions, respectively.

2.3

Preparing mass-normalized PM${}_{\mathrm{2.5}}$ suspension for cell exposure

In total, 30 source and 16 ambient PM${}_{\mathrm{2.5}}$ samples also performed cytotoxicity tests. In order to elute the particles completely from the quartz membranes, a whole PM${}_{\mathrm{2.5}}$-loaded sample filter was cut into small pieces, immerged in ultrapure water, and extracted six times (30 min for each) in an ultrasonic bath at 0 ${}^{\circ }$C. Although the ultrasonication might impact the ROS (Miljevic et al., 2014), the inevitable systematic error was ignored in this study. The extract was then suction-filtered through a 2.6 $\mathrm{\mu }$m pore-size nylon membrane to remove possible quartz fragments, and the bulk filtrate was freeze-dried back to pure PM${}_{\mathrm{2.5}}$ powder. Ultimately, based on particle mass, the gathered PM${}_{\mathrm{2.5}}$ was dispersed by sterile phosphate-buffered saline (PBS) to a concentration of 400 mg L${}^{-\mathrm{1}}$ and then diluted to PM${}_{\mathrm{2.5}}$ suspension of 80 mg L${}^{-\mathrm{1}}$ with serum-free Dulbecco's modified eagle medium (DMEM) to follow in vitro cell exposure (H. Li et al., 2022).

2.4

Cell culture and cellular toxicity tests by in vitro PM${}_{\mathrm{2.5}}$ exposure

Aerosol pollution can harm lung alveoli and epithelial cells, and the A549 adenocarcinoma epithelial cell has long been used as a suitable epithelial alveolar model (T. Li et al., 2022; Park et al., 2018). The A549 cells were cultured in RMPI-1640 medium (Gibco, USA) supplemented with 10 % fetal bovine serum (FBS, Hyclone, USA) and 1 % antibiotic penicillin-streptomycin (100 U mL${}^{-\mathrm{1}}$) at 37 ${}^{\circ }$C in a 5 % CO${}_{\mathrm{2}}$ incubator. After PM${}_{\mathrm{2.5}}$ exposure, cell viability and the indicators reflecting oxidative damage and inflammatory response, respectively, were determined. While the cell viability assay was helpful in determining PM${}_{\mathrm{2.5}}$ doses to cells, the endogenous ROS measurements revealed the status of cellular oxidative potential after PM${}_{\mathrm{2.5}}$ exposure followed by the relative effects of ROS on various stages of cellular toxicity like inflammatory response (Gali et al., 2019). The cell viability (metabolic activity) was evaluated by mitochondrial activity and determined by the methyl-thiazol-tetrazolium (MTT) assay (Chen et al., 2019). After trypsin action, the density of cells in the logarithmic growth phase was adjusted to $\mathrm{1}\times {\mathrm{10}}^{\mathrm{5}}$ mL${}^{-\mathrm{1}}$. Cell suspensions were inoculated into 96-well plates (Costar, USA) at 100 $\mathrm{\mu }$L per well. The blank control well (without medium and PM${}_{\mathrm{2.5}}$ suspension) and reagent control well (with medium but without PM${}_{\mathrm{2.5}}$ suspension) were set together. After incubation for 24 h and removal of the cellular supernatant, various types of PM${}_{\mathrm{2.5}}$ suspension (concentration of 80 mg L${}^{-\mathrm{1}}$) were added to 96-well plates and incubated for 24 h. Based on pre-experiments, the oxidative stress and inflammation response are sensitive to this dose, while the cell viability can remain sufficient. Fresh medium and MTT reagent (Solarbio, Beijing, CN) were added to each well and the supernatant was discarded, and then 100 $\mathrm{\mu }$L of formazan lysate was added to each well. The optical density (OD) values were measured at 490 nm using a microplate reader (Thermo MULTISKAN FC, USA). Cell viability was (%) $=$ (OD${}_{\mathrm{treatment}}$ $-$ OD${}_{\mathrm{blank}\mathrm{control}}$) / (OD${}_{\mathrm{reagent}\mathrm{control}}$ $-$ OD${}_{\mathrm{blank}\mathrm{control}}$). The levels of cellular ROS production causing oxidative stress in cells, proinflammatory cytokines including tumor necrosis factor alpha (TNF-$\mathit{\alpha }$), and interleukin-6 (IL-6) production for determining the expression of genes related to the inflammatory response in the supernatant were analyzed by enzyme-linked immunosorbent assay (ELISA) kits (Jiangsu Meibiao Biotechnology Co., Ltd., CN), and OD values were measured at 450 nm (Huang et al., 2020; Pang et al., 2020).

2.5 Data analysis

The statistical analysis was performed by IBM SPSS statistics 24 and was plotted by the Origin 2020b software. Spearman correlation coefficients were produced by the correlation analysis. The variance was statistically significant when the statistical test level was $p\mathit{<}\mathrm{0.05}$ and extremely significant when $p\mathit{<}\mathrm{0.01}$. Statistical analyses were performed using the Kruskal–Wallis test (Kruskal and Wallis, 1952).

The source apportionment of PM${}_{\mathrm{2.5}}$ mass in urban ambient air was conducted by the receptor models PMF (EPA PMF version 5.0) and CMB (EPA CMB 8.0). All the measured constituents (OC, EC, Cu, Cr, Co, Ni, As, Pb, Mn, V, Na${}^{+}$, K${}^{+}$, Mg${}^{\mathrm{2}+}$, Ca${}^{\mathrm{2}+}$, NH${}_{\mathrm{4}}^{+}$, Cl${}^{-}$, F${}^{-}$, NO${}_{\mathrm{3}}^{-}$, and SO${}_{\mathrm{4}}^{\mathrm{2}-}$) were selected as PMF model input data, and a four-factor solution was chosen as the optimal solution based on an assessment of the interpretability of the source profiles and the seasonal variability of the source contributions. Due to the high concentrations of sulfate and nitrate in ambient PM${}_{\mathrm{2.5}}$ and the lack of a specific actual source to emit sulfate and nitrate, we added the virtual source profiles of secondary sources in the CMB model (Table S4). The virtual source profiles of secondary sources are represented by the proportion of sulfate, nitrate, and ammonium in pure ammonium sulfate and ammonium nitrate.

3 Results

3.1

Contributions of combustion primary sources to urban ambient air PM${}_{\mathrm{2.5}}$

As shown in Fig. S3, although there have been significant improvements in national air quality in recent years, the estimated annual PM${}_{\mathrm{2.5}}$ concentrations of the representative city Nanjing ($\mathrm{59.1}\pm \mathrm{20.5}$ $\mathrm{\mu }$g m${}^{-\mathrm{3}}$) were 1.7 times higher than the Chinese national standard (35 $\mathrm{\mu }$g m${}^{-\mathrm{3}}$) and 11.8 times higher than the WHO guidelines (5 $\mathrm{\mu }$g m${}^{-\mathrm{3}}$). Urban air PM${}_{\mathrm{2.5}}$ pollution levels in the cold season were higher than the warm season. The similar source apportionment results from the PMF and CMB models are illustrated in Fig. 1. Four major sources of the ambient PM${}_{\mathrm{2.5}}$ were produced by the PMF model (Fig. S4), including secondary aerosols and primary particles of automobile exhaust, coal combustion, and plant biomass burning, which account for 34.0 %, 27.7 %, 25.2 %, and 13.1 %, respectively, of the total PM${}_{\mathrm{2.5}}$ mass concentration. The CMB model source profiles are shown in Table S4, and we normalized the contribution of secondary aerosols (32.4 %), automobile exhaust (32.2 %), coal combustion (25.1 %), and plant biomass burning (10.3 %). Therefore, although the contribution of secondary aerosols cannot be ignored, the main anthropogenic sources of urban air PM${}_{\mathrm{2.5}}$ were primary emissions from the various fuel combustions.

Figure 1

Source contributions (%) to the urban ambient air PM${}_{\mathrm{2.5}}$ (models PMF and CMB).

[Figure omitted. See PDF]

3.2

Chemical compositions of different PM${}_{\mathrm{2.5}}$ from 30 combustion sources and from representative urban ambient air

Typical chemical components, including carbonaceous fractions, heavy metals, and the WSIs of all the PM${}_{\mathrm{2.5}}$ samples from both ambient air and combustion sources, were analyzed and compared with each other.

According to the comparisons of PM${}_{\mathrm{2.5}}$-bound carbonaceous fractions (Fig. 2), automobile- and biomass-sourced PM${}_{\mathrm{2.5}}$ contained a significantly higher total carbon (TC) content than coal combustion and ambient air, while the $\mathrm{OC}/\mathrm{EC}$ ratio trend was ambient air $\mathit{>}$ coal combustion $\mathit{>}$ biomass burning $\mathit{>}$ automobile exhaust sources. This indicated that the carbon content of the ambient PM${}_{\mathrm{2.5}}$ mixture was lower and more dominated by OC than that of combustion primary sources, implying that the OC in ambient air may be aged or cleaned. The OC undergoes various chemical reactions in the atmosphere, such as oxidization by ozone and hydroxyl radicals, resulting in degradation. Figures S4–7 show the detailed carbon fraction characteristics (contents and ratios) of PM${}_{\mathrm{2.5}}$ from each specific source. Carbonaceous fractions in automobile exhaust PM${}_{\mathrm{2.5}}$ were high, but the difference between the OC and EC contents was small. Depending on the diverse automobile fuels, loads, and tailpipe emission standards, the concentrations of carbon fractions in exhaust PM${}_{\mathrm{2.5}}$ varied widely with vehicle categories. The carbonaceous portion of PM${}_{\mathrm{2.5}}$ gradually declines as emission regulations rise, and EC likewise declines dramatically (Fig. S5). However, such differences between coal types were lower, except for the bituminous coal with extremely high OC (Fig. S6). The carbonaceous fraction of PM${}_{\mathrm{2.5}}$ from domestic plant biomass burning differed from raw material species in that tree-branch-sourced PM${}_{\mathrm{2.5}}$ generally contained higher carbon contents than those from crop straws (Fig. S7).

Figure 2

Carbon contents (mg kg${}^{-\mathrm{1}}$) and ratios in PM${}_{\mathrm{2.5}}$ from various specific sources ($n=\mathrm{10}$ for each combustion source and $n=\mathrm{16}$ for urban ambient air).

[Figure omitted. See PDF]

Based on the grouped (Fig. 3) and individual (Figs. S9–12) distributions of the measured heavy metals in various PM${}_{\mathrm{2.5}}$, the V concentrations of combustion sources were generally higher, while Co and Mn were lower than ambient urban air. Coal combustion emissions had the highest levels of Pb and were enriched in Cu and As (Fig. S10), while biomass burning was rich in Cr and Ni (Fig. S11). However, automobile exhausts were enriched in most heavy metals, especially Cu, Cr, Ni, V, and Mn (Fig. S9). Heavy metals from different types of automobile exhausts with the same emission standard vary greatly. Anthracite and industrial coal combustions contain similar heavy metals much more than bituminous coal. Generally, Pb, V, Mn, As, and Cu in branch-sourced PM${}_{\mathrm{2.5}}$ were higher than straws, while Cr, Ni, and Co were dominant and higher in straw-burning emissions. A special discovery was that corncob-burning PM${}_{\mathrm{2.5}}$ had more heavy metals than corn straw and was the biomass with the highest emission levels of heavy metals. Correspondingly, ambient air PM${}_{\mathrm{2.5}}$ was also rich in most metals, especially Mn, Pb, Ni, Cu, and Cr. Therefore, coal combustion sources might contribute the most Pb to urban ambient air and contribute significant Cu and As with automobile exhaust emissions, while plant biomass burning and automobile sources contribute the Cr and Ni. In addition to natural dust, automobile exhaust should be the main anthropogenic source of airborne Mn. Considering the PMF source apportionments of ambient aerosols, automobile exhaust should be the main source of Cr in urban air PM${}_{\mathrm{2.5}}$ and also the source of Cu together with coal combustion.

Figure 3

Heavy-metal contents (mg kg${}^{-\mathrm{1}}$) in PM${}_{\mathrm{2.5}}$ from various specific sources ($n=\mathrm{10}$ for each combustion source and $n=\mathrm{16}$ for urban ambient air).

[Figure omitted. See PDF]

According to the comparisons of water-soluble cation and anion concentrations in various PM${}_{\mathrm{2.5}}$ (Fig. 4), coal combustions contained the highest SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NH${}_{\mathrm{4}}^{+}$, automobile exhausts had the highest contents of NO${}_{\mathrm{3}}^{-}$, Na${}^{+}$ and Ca${}^{\mathrm{2}+}$, plant biomass burning sources contained the highest K${}^{+}$ and Cl${}^{-}$, but Mg${}^{\mathrm{2}+}$ was the lowest for all the sources. However, the urban ambient air PM${}_{\mathrm{2.5}}$ contained the highest NO${}_{\mathrm{3}}^{-}$ and was also dominated by SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NH${}_{\mathrm{4}}^{+}$, for which NO${}_{\mathrm{3}}^{-}$ should be mainly contributed by secondary aerosols and automobile primary sources and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NH${}_{\mathrm{4}}^{+}$ should be from coal combustion to a significant degree. In addition to NO${}_{\mathrm{3}}^{-}$, Na${}^{+}$ and Ca${}^{\mathrm{2}+}$, automobile source PM${}_{\mathrm{2.5}}$ also had higher F${}^{-}$ and Mg${}^{\mathrm{2}+}$ concentrations than other sources. The detailed concentration distributions of WSIs in PM${}_{\mathrm{2.5}}$ from each specific source are provided in Figs. S12–14. The WSI levels vary widely with specific source categories. PM${}_{\mathrm{2.5}}$ from LDDVs-2 had the lowest number of WSIs compared to the other automobile exhausts (Fig. S13). Similarly to the metal composition, bituminous coal also had the lowest number of WSIs among all the coals (Fig. S14). Compared to branches, PM${}_{\mathrm{2.5}}$ from burning crop straws had much higher levels of K${}^{+}$, Cl${}^{-}$ and SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and lower levels of F${}^{-}$ and NO${}_{\mathrm{3}}^{-}$ (Fig. S15).

Figure 4

Water-soluble ion (WSI) contents (mg kg${}^{-\mathrm{1}}$) in PM${}_{\mathrm{2.5}}$ from various specific sources ($n=\mathrm{10}$ for each combustion source and $n=\mathrm{16}$ for urban ambient air).

[Figure omitted. See PDF]

To summarize, the overall concentrations of measured TC, cumulated heavy metals and WSIs in PM${}_{\mathrm{2.5}}$ from each source type are shown in Fig. 5. Among all the source emission and environmental receptor samples, the cumulated heavy metals from coal combustion were highest, and automobile exhaust was higher than ambient PM${}_{\mathrm{2.5}}$, the overall carbon contents from automobile exhaust and biomass burning were both higher than ambient PM${}_{\mathrm{2.5}}$, and only the cumulated soluble ions in PM${}_{\mathrm{2.5}}$ from primary sources of coal combustion were equivalent to the ambient aerosols. In a word, chemical compositions of PM${}_{\mathrm{2.5}}$ were distributed very diversely and varied significantly with the specific source types of combustion emissions.

Figure 5

Cumulated typical measured components (mg kg${}^{-\mathrm{1}}$) in PM${}_{\mathrm{2.5}}$ from various specific sources ($n=\mathrm{10}$ for each combustion source and $n=\mathrm{16}$ for urban ambient air).

[Figure omitted. See PDF]

3.3

Cell viability, oxidative stress, and inflammation levels exposed to various mass-normalized PM${}_{\mathrm{2.5}}$

Multiple toxicological endpoints (cell viability, oxidative stress, and inflammation) that facilitate identification of the specific particle-triggering ROS and inflammatory response resulting in cell death were evaluated for source-specific PM${}_{\mathrm{2.5}}$. After 24 h exposure to the same dose of different PM${}_{\mathrm{2.5}}$ obtained from specific emission sources, the A549 lung cells also showed varying toxicological responses (Fig. 6). The survival rate of cells exposed to automobile exhaust PM${}_{\mathrm{2.5}}$ was much lower than ambient air PM${}_{\mathrm{2.5}}$ by 16.6 % (Fig. 6a). Automobile exhaust PM${}_{\mathrm{2.5}}$ induced higher ROS production in cells than biomass burning, which was 26.4 % and 14.8 % higher than ambient PM${}_{\mathrm{2.5}}$ (Fig. 6b). Coal combustion induced the highest cellular IL-6 production, followed by automobile exhaust, which was 13.1 % and 4.48 % higher than ambient air PM${}_{\mathrm{2.5}}$, while the PM${}_{\mathrm{2.5}}$ from automobile exhaust and biomass burning induced a similarly (10.4 %) higher cellular production of TNF-$\mathit{\alpha }$ than ambient PM${}_{\mathrm{2.5}}$ (Fig. 6c, d). These results suggested that combustion primary emission PM${}_{\mathrm{2.5}}$ had a stronger ability to induce oxidative stress and inflammatory injury in lung cells than ambient air PM${}_{\mathrm{2.5}}$, thus resulting in the higher probability of apoptosis induction (Victor and Gottlieb, 2002; Wang et al., 2013). Generally, the mass-normalized PM${}_{\mathrm{2.5}}$ from primary sources of automobile exhaust had the strongest overall toxicity. Therefore, to protect public health by controlling PM${}_{\mathrm{2.5}}$ pollution, these anthropogenic combustions were key target sources, and the most toxic automobile PM${}_{\mathrm{2.5}}$ especially should be reduced preferentially.

Figure 6

Cell viability, oxidative stress and inflammation levels of human alveolar epithelial cell lines (A549) exposed to PM${}_{\mathrm{2.5}}$ suspension (80 mg L${}^{-\mathrm{1}}$) from various specific sources ($n=\mathrm{10}$ for each combustion source and $n=\mathrm{16}$ for urban ambient air).

[Figure omitted. See PDF]

3.4

Correlations between various PM${}_{\mathrm{2.5}}$ components and toxicity endpoints

Spearman correlation coefficients between chemical compositions and cellular toxicological response indicators were applied to screen the key components of all PM${}_{\mathrm{2.5}}$ involved in cell injury (Fig. 7). It was found that the degrees of correlations varied with the toxicological mechanisms of different airborne chemicals. Based on the overall PM${}_{\mathrm{2.5}}$ samples from various sources, the proinflammatory cytokine IL-6 showed significantly strong positive correlations with some heavy metals (As, Pb, V, Cu), while TNF-$\mathit{\alpha }$ and oxidative stress (ROS) had similar significantly positive correlations with aerosol components of carbon fractions (EC, OC) and transition metals (V, Cr, Ni). The TNF-$\mathit{\alpha }$ also showed a positive correlation with water-soluble Cl${}^{-}$ and K${}^{+}$, and ROS correlated with F${}^{-}$, Ca${}^{\mathrm{2}+}$, and Mg${}^{\mathrm{2}+}$.

Figure 7

Overall correlations between typical cellular toxicological responses and chemical compositions of PM${}_{\mathrm{2.5}}$ from various sources (${}^{*}$ $p\mathit{<}\mathrm{0.05}$, ${}^{\mathit{\#}}$ $p\mathit{<}\mathrm{0.01}$; $n=\mathrm{46}$).

[Figure omitted. See PDF]

4 Discussion

4.1

Chemical markers for source apportionments of ambient air PM${}_{\mathrm{2.5}}$

Combustion emissions are key anthropogenic sources contributing to urban air PM${}_{\mathrm{2.5}}$ through both primary and secondary aerosols, which were 66 % and 34 % as estimated by the PMF model and 67.6 % and 32.4 % as estimated by the CMB model, respectively (Fig. 1). Compared to the PMF results, the proportions of coal combustion and secondary sources in the CMB model results show minimal changes, while biomass contributions are slightly underestimated, and there is a slight increase in the proportion attributed to vehicular emissions. The high concentrations of chemical markers are usually used in source analysis, such as ammonium sulfate and nitrate for secondary aerosols which originated mainly from gaseous precursors (e.g., NH${}_{\mathrm{3}}$, SO${}_{\mathrm{2}}$, and NO${}_X$) (Mahilang et al., 2021), the EC, Cu, Mn, and Ni for vehicle exhaust (Srivastava et al., 2021), the As, Pb, OC, EC, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and relatively low ${\mathrm{NO}}_{\mathrm{3}}^{-}/{\mathrm{SO}}_{\mathrm{4}}^{\mathrm{2}-}$ ratios for coal combustion (Dai et al., 2020), and soluble K${}^{+}$ and Cl${}^{-}$ for plant burning (Jain et al., 2020). The detailed chemical species of these specific source emission PM${}_{\mathrm{2.5}}$ samples also supported the results. Moreover, low $\mathrm{OC}/\mathrm{EC}$ ratios of high TC content, high NO${}_{\mathrm{3}}^{-}$, F${}^{-}$, Na${}^{+}$, Ca${}^{\mathrm{2}+}$ and Mg${}^{\mathrm{2}+}$, V, and Mn of automobile exhaust; Pb and As, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NH${}_{\mathrm{4}}^{+}$ of coal combustion; soluble K${}^{+}$ and Cl${}^{-}$; and a high $\mathrm{OC}/\mathrm{EC}$ ratio of high TC for plant biomass burning found in the current study (Figs. 2–5) could also be corresponding potential aerosol source markers.

4.2

Common PM${}_{\mathrm{2.5}}$ components related to specific combustion sources

Generally, the automobile exhaust PM${}_{\mathrm{2.5}}$ had a high TC content and a low $\mathrm{OC}/\mathrm{EC}$ value with a considerable EC content (Fig. 2), varying with specific vehicle types (Fig. S5–8). The contents of the carbon fractions from diesel vehicles were 2.39 times more than gasoline exhausts (Fig. S5), and the $\mathrm{OC}/\mathrm{EC}$ ratios of diesel exhausts were 37.3 % of gasoline vehicles, owing to both considerable contents of EC and OC from diesel vehicle emission PM${}_{\mathrm{2.5}}$. Some diesel vehicles showed higher EC emissions with age, so exhaust-cleaning devices for them are suggested. In addition, the amounts of OC and EC in exhausts gradually decreased with the strengthened emission standards they met (Wong et al., 2020). In PM${}_{\mathrm{2.5}}$ samples obtained from coal combustion (Fig. S6), the TC contents of bituminous coals were 3.97, 6.41, and 11.6 times higher than those of honeycomb coals, anthracite coals, and industrial coals, respectively, because bituminous coals contain higher volatile fractions. Emissions of non-methane VOCs increase with the volatile content of the coal (He et al., 2022). The vast majority of organic aerosols from bituminous coal are generated in the ignition and fierce combustion phases, which account for 99.9 % of the entire combustion process, while these two phases of anthracite coal generate only 77 % of the entire process (Zhou et al., 2016). Moreover, as the volatile matter in the coal decreases, the temperatures at which weight loss begins and ends shift to higher values, which may be due to the lower number of aliphatic chains present. It has been reported that bituminous maximum weight loss occurs in the range 490–600 ${}^{\circ }$C, while in the case of anthracite coals it occurs between 750 and 870 ${}^{\circ }$C (De la Puente et al., 1998). Therefore, in addition to the way of combustion and the use of combustion stoves, the coal quality related to different coal types and origins determines the carbonaceous fractions of the PM emitted by coal combustion (Zhang et al., 2022). In the PM${}_{\mathrm{2.5}}$ samples from plant biomass combustion (Fig. S7), OC contents were 2.21 times higher than EC contents, except that pine branches contained higher EC and rapeseed straw had considerable contents of EC and OC. The OC in ambient PM${}_{\mathrm{2.5}}$ dominated the carbonaceous component (Fig. S8), consistent with the North China Plain and Indo-Ganges Plain (Flores et al., 2020; Xu et al., 2019). Combining the TC contents and $\mathrm{OC}/\mathrm{EC}$ ratios, carbonaceous components in ambient PM${}_{\mathrm{2.5}}$ mainly originate from semivolatile organic compounds (SVOCs) (Wang et al., 2018). Previous studies have reported that carbonaceous aerosols mainly originated from fossil fuel combustion in transportation, coal combustion in power plants and industries, and biomass combustion (Kang et al., 2018; Zhang et al., 2015). Thus, to control the ambient carbon aerosol pollution, in addition to reducing the precursor emissions of secondary organic aerosols (SOA), controlling primary aerosols, especially EC from diesel vehicles, might be an effective measure.

Airborne redox-active metals are usually linked to the oxidation stress of PM${}_{\mathrm{2.5}}$. Different types of automobiles emitted diverse metal contents (Fig. S9). Metal elements in automobile exhaust are primarily contributed by fuels, lubricants, and engine component abrasion. Because Mn is a common anti-detonator that delays and prevents the oxidation of hydrocarbons and increases the octane number, which not only increases the thermal efficiency of the engine, but also improves the emission performance of the vehicle (Cheung et al., 2010), the Mn content was greater in gasoline vehicle exhausts than in diesel vehicles. Although there are multiple sources of traffic Pb emissions such as fuel combustion and brake wear (Wang et al., 2019; Panko et al., 2019), the automobile exhaust Pb content of gasoline vehicles were greater than diesel vehicles owing to oil combustion. Moreover, for the same vehicle type (LDDVs-1 and LDDVs-2; HDDVs-1 and HDDVs-2; SDGCs-1 and SDGCs-2), the stricter the emission standard required, the lower the exhaust metal contents. The metal contents in the PM${}_{\mathrm{2.5}}$ of trucks was higher than that of passenger cars (Wu et al., 2016). In the combustion PM${}_{\mathrm{2.5}}$ of 10 coal types (Fig. S10), Pb contents were the highest than other heavy metals, similar to available findings (Zhang et al., 2020). The PM${}_{\mathrm{2.5}}$ metals from bituminous coal were significantly lower, as indicated by the coal quality analysis, bituminous coal has a low ash content which is mainly derived from non-combustible minerals in coal. These findings suggested that coal maturity might be an important factor influencing the metal composition of particulates emitted from coal combustion (Shen et al., 2021; Zhang et al., 2021). Heavy-metal contents in biomass-burned PM${}_{\mathrm{2.5}}$ varied widely with raw plant types (Fig. S11) but were dominated by Cr and Ni. Different plant species and even different plant parts differ significantly in their ability to take up and accumulate metals from the soil (Zhao et al., 2020). Moreover, because of the high enrichment factors of some metals for crop straws (Zhang et al., 2016; Sun et al., 2019), they also released more Cr, Ni, and Co during burning than fuel woods. Total metal emissions were highest in corn cob but lowest in peanut straw burning PM${}_{\mathrm{2.5}}$. The heavy metals enriched in urban ambient air PM${}_{\mathrm{2.5}}$ showed a slightly seasonal pattern (Fig. S12), while contents of V, Co, and As were relatively low and less affected by seasonal changes. Accordingly, supported by the metal profiles of anthropogenic combustion sources and ambient aerosols, to control the environmental airborne heavy-metal pollution, key targets might be the Pb, Cu, and As from honeycomb, anthracite, and industrial coal combustion; Cu from vehicle exhausts and especially V from light-duty diesel vans with the CN.III emission standard and Mn from gasoline vehicles, Cr and Ni from biomass, especially crop straw burning.

Epidemiological studies have also shown mortality closely related to the WSIs such as sulfate and nitrate in aerosols (Ostro et al., 2009; Liang et al., 2022). Among the WSIs contents of various automobile exhaust PM${}_{\mathrm{2.5}}$ (Fig. S13), NO${}_{\mathrm{3}}^{-}$ and Ca${}^{\mathrm{2}+}$ were the most abundant anion and cation, respectively. The high NO${}_{\mathrm{3}}^{-}$ in the automobile PM${}_{\mathrm{2.5}}$ may be due to NO${}_x$ production during high-temperature combustion, while the high Ca${}^{\mathrm{2}+}$ content should be related to additives in automobile fuels and calcium-based lubricants (Hao et al., 2019; Yang et al., 2019). Moreover, the exhaust WSIs decreased with the strengthened automobile emission standards required. Coal combustion PM${}_{\mathrm{2.5}}$ contained relatively higher SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NH${}_{\mathrm{4}}^{+}$ concentrations followed by Cl${}^{-}$ than other WSIs species (Fig. S14). Among various coal types, industrial coals emitted highest SO${}_{\mathrm{4}}^{\mathrm{2}-}$ followed by honeycomb and industrial coal with also high NH${}_{\mathrm{4}}^{+}$, but bituminous coals emitted low WSIs which were mainly NO${}_{\mathrm{3}}^{-}$, F${}^{-}$ and Na${}^{+}$, Ca${}^{\mathrm{2}+}$. The WSIs emission factors of honeycomb coal were generally higher than those of lump coal (Yan et al., 2020). For biomass combustion emissions (Fig. S15), Cl${}^{-}$ and K${}^{+}$ were dominant WSIs in PM${}_{\mathrm{2.5}}$ from straw-type fuels (Tao et al., 2016; Sillapapiromsuk et al., 2013), but fuelwood-type combustion emitted high NO${}_{\mathrm{3}}^{-}$. Plant species absolutely determine the emissions (Liao et al., 2021). Finally, there were also high levels of NO${}_{\mathrm{3}}^{-}$, SO${}_{\mathrm{4}}^{\mathrm{2}-}$, and NH${}_{\mathrm{4}}^{+}$ in ambient air PM${}_{\mathrm{2.5}}$ (Zhang et al., 2019) (Fig. S16), even higher than the investigated combustion sources, so other sources like the secondary aerosols may also contribute. Consequently, target combustion primary aerosols WSIs might include the NO${}_{\mathrm{3}}^{-}$ from vehicle exhausts and fuelwood burning; SO${}_{\mathrm{4}}^{\mathrm{2}-}$ and NH${}_{\mathrm{4}}^{+}$ from honeycomb, anthracite, and industrial coal combustion; and Cl${}^{-}$ and K${}^{+}$ from biomass (especially crop straw) burning.

4.3

PM${}_{\mathrm{2.5}}$ toxicity related to specific sources by pivotal chemical components

The complexity of the sources and compositions of atmospheric PM${}_{\mathrm{2.5}}$ leads to different toxicological effects (Newman et al., 2020; Kelly, 2021). The toxicological effects of PM${}_{\mathrm{2.5}}$ are not comparable among different studies, owing to distinct exposure concentrations, biological models, endpoints, and PM${}_{\mathrm{2.5}}$-generation methods (Kelly and Fussell, 2020; Park et al., 2018). In this study, we employed same exposure conditions and biological endpoints in order to obtain comparable toxicity data for PM${}_{\mathrm{2.5}}$ from different sources. Our mass-normalized results demonstrated that automobile exhaust PM${}_{\mathrm{2.5}}$ induced the highest lethality and cellular ROS and TNF-$\mathit{\alpha }$ production, that coal combustion PM${}_{\mathrm{2.5}}$ induced the highest cellular IL-6 production, and that plant biomass burning PM${}_{\mathrm{2.5}}$ induced considerable cellular TNF-$\mathit{\alpha }$ and ROS production (Fig. 6). Generally, various toxicities of combustion emission primary PM${}_{\mathrm{2.5}}$ were much greater than the urban ambient air PM${}_{\mathrm{2.5}}$ (Fig. 6), owing to the higher concentrations of specific toxic components in PM${}_{\mathrm{2.5}}$ from these sources. The Supplement included exhaustive cytotoxicity indicators from each individual source (Figs. S17–20). While the survival rate of cells exposed to the CN.III emission standard PM${}_{\mathrm{2.5}}$ was the lowest and the capacity to induce cells to produce ROS was the highest for CN.IV, automobile exhaust had a similar potential to cause cells to produce inflammatory cytokines (Fig. S17). The capability to induce IL-6 production in cells was highest for industrial coal PM${}_{\mathrm{2.5}}$, whereas bituminous coal had the highest survival rate of the cells and TNF-$\mathit{\alpha }$ induction capacity (Fig. S18). From Fig. S19, we can see that the PM${}_{\mathrm{2.5}}$ cytotoxicity of straw and branch burning was analogous, but it should be noted that the cell viability of various straw PM${}_{\mathrm{2.5}}$ differ significantly, which may be related to the raw fuel characteristics.

These possible mechanisms were implied by the overall relationships between the measured chemical components with cytotoxicity indicators of PM${}_{\mathrm{2.5}}$ from various specific sources (Fig. 7). In general, both TNF-$\mathit{\alpha }$ and ROS were significantly positively correlated with carbonaceous fractions and redox-active transition metals (V, Cr, Ni), which were the main contributors of automobile exhausts and biomass burning. The IL-6 was significantly positively correlated with some heavy metals (As and Pb, V and Cu), which were the main contributors of coal combustion sources. Potential mechanisms include carbon fractions bound in PM${}_{\mathrm{2.5}}$ possibly being transformed into reactive metabolites and then inducing ROS production in cells (Stevanovic et al., 2019), and the PM${}_{\mathrm{2.5}}$-bound transition metals could also induce ROS production through the Fenton reaction and disrupt the function of enzymes in cells (Verma et al., 2010; Sørensen et al., 2005). Oxidative stress can lead to inflammatory infiltration of neutrophils and stimulate immune cells to produce inflammatory cytokines, among which TNF-$\mathit{\alpha }$ and IL-6 play important roles in inflammation development (Xu et al., 2020). Ultimately, excessive production of ROS leads to dysfunctional endoplasmic reticulum responses, and dysfunctional lipid metabolism in ROS bursts can result in cell membrane damage and even cell death (Piao et al., 2018; Zhao et al., 2004). There have been some related supporting reports. For instance, the OC and EC were significantly associated with biological responses of PM from vehicle emissions collected in tunnels (Niu et al., 2020). The polar or quinone fractions of PAHs in diesel engine exhaust particles significantly contributed to the heightened toxic response (Xia et al., 2004). The PM${}_{\mathrm{2.5}}$ generated from biomass burning contained a substantial concentration of carbonaceous components. In addition, Cr and Ni in PM${}_{\mathrm{10}}$ from straws were highly associated with ROS (Li et al., 2023). In the current study, cellular ROS were also correlated with water-soluble Ca${}^{\mathrm{2}+}$, F${}^{-}$, and Mg${}^{\mathrm{2}+}$, which were the main contributors of automobile exhaust PM${}_{\mathrm{2.5}}$. The Ca${}^{\mathrm{2}+}$ controls the membrane potential and regulates mitochondrial adenosine triphosphate (ATP) production, and excessive Ca${}^{\mathrm{2}+}$ leads to energy loss and more ROS production (Madreiter-Sokolowski et al., 2020). Moreover, the TNF-$\mathit{\alpha }$ was also positively correlated with water-soluble Cl${}^{-}$ and K${}^{+}$, which were the main contributors to plant burning PM${}_{\mathrm{2.5}}$. Therefore, the accumulations of some organic matters with high carbonaceous content (OC, EC) in PM${}_{\mathrm{2.5}}$, typically from automobile exhausts and plant biomass burning, redox-active metals (V, Cr, Ni) and water-soluble anions (Cl${}^{-}$, F${}^{-}$) and cations (Ca${}^{\mathrm{2}+}$, Mg${}^{\mathrm{2}+}$) contributed by various combustions, might induce ROS production in cells, cause cellular damage through oxidative stress and inflammatory response, impair cell viability, and finally harm human health.

Considering the multiple endpoints measured and the PM${}_{\mathrm{2.5}}$ toxicity mechanisms mentioned above, based on the cell viability, the ROS, the inflammatory markers, and the significantly related toxic chemical composition contents (Park et al., 2018), we put forward a general sequence of overall mass-normalized toxicity for these combustion source PM${}_{\mathrm{2.5}}$ to managers. To improve the urban environmental air quality for better public health benefits by controlling aerosol pollution, considering the differential toxicity intensity of each chemical component and its contributions from various sources to ambient aerosols, preferential targets of specific primary PM${}_{\mathrm{2.5}}$ sources and bound pollutants from anthropogenic combustions are suggested as following this sequence: reducing the automobile exhaust PM${}_{\mathrm{2.5}}$ containing high contents of EC, transition metals (V, Cu, Ni, Cr), and ions (Ca${}^{\mathrm{2}+}$, Mg${}^{\mathrm{2}+}$, F${}^{-}$, Na${}^{+}$) from diesel exhausts by strengthening the emission standards and accelerating the phasing out of highly polluting vehicles; lessening coal combustion rich in heavy metals (As, Pb, Cu) by replacement with low-ash clean coals; and decreasing biomass burning containing high OC, Ni, Cr, Cl${}^{-}$, and K${}^{+}$ from rural crop straw emissions and promoting domestic cleaner energy such as natural gas.

4.4 Limitations and perspectives

In the current study, we selected the A549 cell based on previous abundant experimental experiences and also because it has been used popularly in in vitro toxicology studies to elucidate the cellular and molecular mechanisms of PM involved in lungs for many decades (T. Li et al., 2022). However, recently, the human normal bronchial epithelial cell BEAS-2B was preferred over the human lung adenocarcinoma epithelial cell A549. For instance, both cells were used in an aerosol study (Bonetta et al., 2017), the results of which also highlighted the higher sensitivity of BEAS-2B cells with respect to A549 in samples with low levels of pollutants, because the PM${}_{\mathrm{0.5}}$ samples from Italian towns can induce genotoxicity in normal cells, while cancer cells might be resistant to their adverse effects. Therefore, although our results are reasonable under the same exposure conditions, there were still potential limitations of A549 cells since they may be more resistant to exposure to external compounds, and the generally more sensitive BEAS-2B cells are suggested for future studies.

In toxicity assessments, cell vitality reflects the overall health of cells, encompassing factors such as cell membrane integrity, intracellular metabolic activity, and cell proliferation capacity. Decreased cellular vitality may be associated with cell damage, toxic effects, or cellular apoptosis. Inflammation markers are employed to assess the extent and nature of inflammatory reactions, including the production of cytokines and inflammatory mediators, as well as the activation status of inflammatory cells. Inflammation is a complex physiological response typically delineated by the immune and inflammatory reactions of the body to stimuli such as injury or infection. Alterations in inflammation markers can indicate the intensity and nature of the inflammatory response. In this study, multiple biological responses of epithelial cells to various PM${}_{\mathrm{2.5}}$ were evaluated, including that cell viability evaluated the mitochondrial dehydrogenase activity of the living cells, excessive intracellular ROS formation induced by PM${}_{\mathrm{2.5}}$ was responsible for oxidative stress on the cells, and cytokines IL-6 and TNF-$\mathit{\alpha }$ were determined for the effect of PM${}_{\mathrm{2.5}}$ on proinflammatory response in cells. In general, in vitro data can be used to rank various types of particles in terms of the toxic potential, including possible carcinogenicity. Each marker will help one to understand the hazard and toxicity of PM${}_{\mathrm{2.5}}$. However, the toxicity of PM${}_{\mathrm{2.5}}$ may be the result of multiple components acting through disparate physiological mechanisms, with inconsistent relationships among endpoints (Park et al., 2018). For instance, in BEAS-2B cells, oxidative stress generated by H${}_{\mathrm{2}}$O${}_{\mathrm{2}}$ exposure rather than by stimulating cytokine or chemokine responses often results in cytotoxicity, sometimes with no correlation between oxidative damage and cytokine or chemokine responses. Moreover, the TNF-$\mathit{\alpha }$ gene was not detected in BEAS-2B cells exposed to atmospheric PM collected from Benin, but the gene expressions of other inflammatory cytokines (IL-1$\mathit{\beta }$, IL-6, and IL-8) were significantly induced, and decreasing cell viability was highly correlated with high secretions of all the studied cytokines (Cachon et al., 2014). Therefore, in the present study, it was impossible to analyze all the chemicals in PM${}_{\mathrm{2.5}}$ and to determine all the related toxicological endpoints, so unmeasured chemicals and endpoints might also play roles in the incongruous or unexplained results, and we also cannot over-explain the mechanisms just based on statistical relations. To overcome these hurdles, standardization of toxicological studies (experimental methodologies) and reporting guidelines is necessary for tracking and comparing results.

This study ranked the unequal “toxic effects” based on the same mass concentration of PM${}_{\mathrm{2.5}}$ exposure in body lung fluid systems, while the “health risks” usually related to the inhalation exposure concentration of PM${}_{\mathrm{2.5}}$ in ambient air were not calculated and evaluated quantitatively. Moreover, nonlinear concentration–response functions for various endpoints and different exposure concentrations might also limit the straightforward use of toxicological data to predict health effects (morbidity, mortality) in human populations. Therefore, drawing precise conclusions quantifying or ranking the health risks of PM${}_{\mathrm{2.5}}$ from specific sources or of individual PM${}_{\mathrm{2.5}}$ components is still not an easy task (Kelly and Fussell, 2020). Coupled with source apportionment and exposure levels of ambient aerosol pollution, toxicology and epidemiology studies linking these factors and indicating scientific mechanisms would help one to reach conclusions.

Moreover, the exact effective measures to control these specific key toxic components from the emissions of various combustion sources are indeed a challenge but still need to be explored. The findings of this research provide a specific direction for better air pollution control and public health. In addition to the environmental technological methods of controlling toxic components targeting source materials, combustion processes, and final emissions, the environmental management policies are also beneficial to such aims, like the choice of fuel types, especially for the management of domestic biomass fuel burning. For example, potential solutions include promoting new green energy vehicles and low-ash clean coals and decreasing the diesel exhaust and rural crop straw burning emissions.

5 Conclusions

In the current study, we found that two-thirds of the mass of urban ambient air PM${}_{\mathrm{2.5}}$ in a typical megacity of eastern China originated from primary sources of anthropogenic combustions, including coal, automobiles, and biomass. Because of the significant differences in the chemical compositions, the diverse PM${}_{\mathrm{2.5}}$ from both mixed ambient air and directly from individual combustion sources showed very different mass-normalized in vitro toxicity to human lung epithelial cells, either for the environmental aerosol samples collected from different seasons or for the primary emissions of PM${}_{\mathrm{2.5}}$ from various specific source types. According to the comparative study and correlation analysis, the carbonaceous fractions (OC, EC) and redox-active heavy metals (V, Ni, Cr) assisted by water-soluble ions (Ca${}^{\mathrm{2}+}$, Mg${}^{\mathrm{2}+}$, F${}^{-}$, Cl${}^{-}$) might play important roles in inducing cellular ROS production, causing oxidative stress and inflammation, resulting in cell injury and apoptosis, and damaging human health. These toxic pollutants accumulated in specific-source PM${}_{\mathrm{2.5}}$ varied by the emission types and raw fuel properties. Combined with the chemical composition and general cytotoxicity rank, the preferential controlling targets of specific combustion sources might be automobile exhaust (diesel vehicles with emission standards inferior to CN.IV), coal combustion (high ash and high sulfur coals), and rural plant biomass burning (crop straws). While showing the synthetic effects of mixed compositions and complex sources, in addition to preventing the secondary aerosols from combustions, preferentially targeted reductions of toxic PM${}_{\mathrm{2.5}}$ direct emissions from these primary sources would produce great benefits for public health with improved ambient air quality. Overall, the chemical findings of our toxicological research could help to support precise, oriented, effective, efficient, and economical composition-source-based strategies for urban aerosol pollution control. However, as a prospect, the detailed mechanisms for unequal toxicity of PM with complicated components from various sources and their quantitative contributions to the health effects of an ambient air PM${}_{\mathrm{2.5}}$ mixture still need in-depth study.

Data availability

All the raw data can be provided by the corresponding authors upon request.

The supplement related to this article is available online at: https://doi.org/10.5194/acp-24-1345-2024-supplement.

Author contributions

XSL conceived and supervised the study. WH, YP, MT, HL, and ZZ collected the samples. WH, YP, MT, WL, HL, ZZ, GS, and LX analyzed the chemical compositions. WH, YP, and MT performed the toxicity tests. WH, YP, MT, and XSL analyzed the data. WH and XSL wrote the manuscript draft. XSL, WH, GS, and TM reviewed and edited the manuscript.

Competing interests

The contact author has declared that none of the authors has any competing interests.

Disclaimer

Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors.

Acknowledgements

The authors are very grateful to the anonymous reviewers for their constructive comments.

Financial support

This research has been supported by the National Natural Science Foundation of China (grant nos. 41977349 and 41471418).

Review statement

This paper was edited by Hang Su and reviewed by three anonymous referees.