Engine and exposure

A four-stroke single-cylinder direct-injected diesel engine test bed situated at the University of Rostock in the Chair of Piston Machines and Internal Combustion Engines was used to generate aerosol for the exposure study. Heavy fuel oil HFO 180 was used as a representative fuel for ship operation outside of sulfur emission control areas (SECAs). Distillate diesel fuel (DF) according to DIN EN 590 was used as a reference. The DF fuel represents a modern, sulfur-free distilled fuel as used in inland waterway transportation and SECAs. The engine ran at four different operating points: 100%, 75%, 50%, and 25% load at a nominal speed of 1,500 rev/min. A detailed characterization of the aerosol composition at the ALI exposure system was performed, [27]. The duration of each operation point was set in accordance to their weighting factors as described in ISO 8178-4 E2. The total cycle duration was 2 hours, and this cycle was run twice for a single exposure. To obtain comparable particle deposition doses for the experiments with the two fuel types, clean air dilution ratios of 1:40 and 1:100 were used for the exhaust aerosols of DF and HFO emissions, respectively. The diluted concentrations of PM 2.5 in the aerosols were 340 μg/cm³ and 760 μg/cm³ for DF and HFO, respectively. Based on these values, and a post-experiment gravimetric filter analysis of PM 2.5 and assuming a constant deposition probability of 1.5%, which was determined using previous measurements from ALI exposure systems [11], the particle mass deposited on the lung cell monolayer surface was calculated at 28 ± 1.5 ng/cm² (DF) and 56 ± 0.7 ng/cm² (HFO) for the 4 hour exposure [14].

Physical and chemical profiling of aerosols

Profiling of the particulate matter from the exposure aersosols from both fuel types was performed as in the previous study [14, 15]. To summarize, on-line and off-line analysis techniques, including gas chromatography mass spectrometry, high resolution mass spectrometry (ESI-FTICR-MS), energy-dispersive X-ray spectroscopy (EDX), on-line aerosol mass spectrometry, well as others have been used. Detailed information on the particulate matter analysis has been reported previously [14], and a summary is shown in S1 Fig. On-line analyses were performed in parallel with cellular exposures, and off-line techniques were performed using particulate matter collected on filters during cellular exposures.

Data on gas phase aerosol compounds were derived from on-line photo-ionization mass spectrometry [28]. Aromatic species were measured by resonance enhanced multi-photon ionization (REMPI) mass spcectrometry, which utilizes intense short laser pulses of 266 nm generated by a Nd:YAG laser. Ionization occurs by subsequent ionization of two photons, providing selectivity and enhanced sensitivity for aromatic species. Benzene and butadiene were analyzed by on-line single photon ionization (SPI) mass spectrometry using 126 nm Vacuum-UV radiation generated by an electron beam pumped rare gas excimer lamp [29]. Ionization occurs by absorption of a single photon. Carbonyl compounds were sampled by derivatisation with 2,4-dinitrophenylhydrazine (DNPH) using commercially available cartridges ‘ORBO/555’ (Sigma Aldrich, USA) and subsequently analyzed by GC-MS [27]

Cell culture conditions

The mouse macrophage RAW 264.7 cell line was obtained from ATCC (ATCC^© TIB-71™). Cells were cultured in RPMI-1640 medium supplemented with 10% (v/v) Fetal Bovine Serum and 100 U/ml penicillin, 100 mg/ml streptomycin (Life Technologies, Darmstadt), and cultivated in an incubator at 37°C with 5% CO₂. For stable isotope labeling experiments, cells were seeded 24 hours before the experiment in RPMI-1640 medium as above, with 12.5 mmol/L U-¹³C₆-Glucose (Cambridge Isotope Laboratories, USA) substituted for unlabeled glucose.

Air-liquid interface exposure

An automated ALI exposure system station (VITROCELL Systems, GmbH, Waldkirch, Germany) with 18 exposure positions was used as the interface for cellular exposures of the diesel engine exhaust [30]. ALI exposures were performed in a custom built mobile HICE S2 bio safety laboratory, placed next to the engine hall. Diluted DF or HFO aerosol was led through heated stainless steel lines from the engine test bed into the ALI system. The combustion aerosols coming from the engine’s exhaust pipe were cooled and diluted with sterile air, and then transferred directly to the ALI for cellular exposure. A PM 2.5 impactor removed large particles before the exhaust entered the ALI system. Cells exposed to complete aerosols used the humidified exhaust from the engine, while cells exposed to only the gas phase had particles removed directly above the cell chamber using a Whatman HEPA polydisc filter (GE Healthcare). Exposures were performed as described for ALI exposures of BEAS2B and A549 epithelium cells [14].

Cells were seeded in FCS supplemented RPMI-1640 on transferrable 24 mm Transwell^® inserts with a 0.4 μm pore polyester membrane (Type #3450, Corning, NY, USA) 24h before exposure at a density of 1 x 10⁶ cells/mL/insert (2.1 x 10⁵ cells/cm² growth area) with 1.5 mL cell culture medium provided beneath the insert membrane. For cell exposure, the culture medium on the apical side was completely removed and cells were placed in the ALI exposure system with RPMI-1640 medium without FBS, supplemented with 10 mM HEPES, provided at the basolateral side. Cells were then exposed for 4h to the diluted and conditioned (85% r.h., 37°C, maintained by a software controlled humidifier in the ALI exposure system) aerosols with a controlled flow of 100 mL/min for each insert [13, 30]. Cells were exposed to both the complete aerosol (with both particle phase and gas phase), and the gas phase only (with particles filtered out of the complete aerosol by a high efficiency particle membrane filter). All fuel-specific results represent both the filtered and unfiltered treatments, unless otherwise stated.

LDH release assay

After exposure, medium from the compartment under the membrane was collected and frozen at -80°C for later analysis. An aliquot was used for quantification of released lactate dehydrogenase (LDH), an indicator of plasma membrane integrity. An LDH detection kit was used in accordance with the manufacturers’ instructions (Roche, Mannheim, Germany) with slight modifications: the dye solution was diluted 1:1 (v/v) with PBS to slow down the reaction time caused by elevated LDH values due to the high cell densities used for ALI exposure experiments. After 20 minutes, before saturation was reached, the reaction was stopped and the absorbance of the reaction mix was measured at 490 nm with a microplate reader. Absorbance read from the samples was normalized to the absorbance of blank medium. Statistical analysis was performed using an analysis of variance (ANOVA) test, followed by a post-hoc Tukey test for pairwise statistical analysis. Cells exposed with HEPA-filtered ambient air humidified to 85% r.h. at 37°C used as a control treatment. Cells used for a positive control for cellular toxicity were kept under control conditions and lysed with Triton X-100 (Sigma Aldrich) for 30 minutes prior to the end of the exposure period.

Metabolite extraction and GC-MS processing

For metabolomics experiments, cells kept in an incubator at 37°C with 5% CO₂ were used as a control treatment. Metabolite extraction was performed as previously described [31]. Briefly, after washing cells with 0.9% NaCl, a 1:1:1 extraction using methanol, water, and chloroform was used to quench metabolism and create a three-phase separation. Cells were agitated for 20 minutes and then centrifuged to enforce phase separation. 200 μL of the polar phase was used for gas chromatography coupled to mass spectometry (GC-MS) analysis. The interphase was then used for proteomics analysis.

Compound derivatization was performed with a Gerstel autosampler directly before measurement on the GC-MS. Dried metabolites were dissolved in 15 μL of 2% methoxyamine hydrochloride in pyridine at a temperature of 40°C for 30 minutes. Then, 15 μL of 2,2,2-trifluoro-N-methyl-N-trimethylsilyl-acetamide + 1% chloro-trimethyl-silane was added and incubated at 40°C for 30 minutes. Methoxyamine hydrochloride derivatization breaks apart cyclic forms of some metabolites, reducing confusing of metabolites with different forms (such as glucose) eluting at different times based upon their structure. 2,2,2-trifluoro-N-methyl-N-trimethylsilyl-acetamide + 1% chloro-trimethyl-silane is used to block polar groups and allow for efficient transition of the metabolites into the gas phase.

The metabolite extracts were measured on an Agilent 7890 GC with a 30 m DB-35MS capillary column (Agilent Technologies) with an internal diameter of 0.25 mm and a film of 0.25 μm. The GC was connected to an Agilent 5975C MS operating in electron ionization (EI) at 70 eV. The MS source was kept at a constant temperature of 230°C and the quadrupole at 150°C. The detector was operated in scan mode with an m/z range of 70 to 800.

1 μL of derivatized sample was injected at 270°C in splitless mode. Helium was used as the carrier gas at a flow rate of 1 mL/min. The GC temperature program started at 80°C with a hold for 6 minutes, followed by a gradient of 6°C/min to 300°C, a hold for 10 minutes, and an additional gradient of 10°C/min to 325°C. The total splitless GC-MS run time of one sample was 59 minutes. An alkane mix was run with the experimental sequence in order to provide retention index calibration for the experimental samples. The above program was used for the measurement of polar intracellular metabolic extracts from samples. In order to determine secretion of lactic acid, both control and experimental cellular medium was analyzed with the above parameters in a 25 minute method.

All chromatograms were normalized to a standard alkane mix (Sigma 68281) to standardize retention times (RT) to retention indicies (RI) across sample runs. To identify metabolites found in experimental chromatograms, an in-house developed metabolite library was used, and mass spectra as well as normalized RI values were compared. This library was created by running analytical standards run on the same instruments using the same parameters as our experimental samples. All identified metabolites were verified using the NIST 11 spectral database [32]. Metabolic analysis and mass isotopomer distribution (MID) calculation, including correction for naturally occuring stable isotopes, was performed with the MetaboliteDetector software [33, 34].

Metabolic data were factor-normalized based on standardized pool samples run with each GC-MS sequence. The pool samples were a standardized extract of intracellular metabolites from control conditions, to provide a baseline and control for GC-MS drift and buildup on the GC column. Changes in the pool samples were calculated, and this factor was applied to normalize the experimental samples. In order to reduce noise and to focus on metabolites changing between conditions, a further filtering was applied, removing all compounds present in less than 80% of all treatments. All metabolic statistical significance measurements were performed using an unpaired, two-tailed student’s t-test unless otherwise noted.

Stable isotope labeling by amino acids in cell culture (SILAC)

RAW 264.7 mouse macrophages were cultured for six passages as above, with either 48.67 μg/mL H4-lysine (lysine 0, Sigma-Aldrich) or D4-lysine (lysine 4, Sigma-Aldrich) to achieve complete labeling of the proteome [35]. In order to detect unlabeled contaminants for each sample the reverse experiment was performed by exchanging lysine 0 and lysine 4.

Proteome extraction and LC-MS/MS analysis of peptides

Proteome extraction was performed as previously described [31]. In brief, after metabolite extraction as above, the proteins were present in the interphase of the sample. The tertiary and secondary structure of the proteins were broken down with TCEP and chloroacetamide, and LysC protease was added to digest the proteins. After this, the samples were used for liquid chromatography coupled to two-dimensional mass spectrometry (LC-MS/MS).

The protein extracts were digested using an automated sample-preparation workflow [36]. The purified peptides were lyophilized and resuspended in 0.06% FA/3% ACN buffer and separated on an in-house packed reversed-phase chromatography column (20 cm length, 75 μm ID, 3 μm—Dr. Maisch C18). A 155 min gradient (solvent A: 5% acetonitrile, 0.1% formic acid; solvent B: 80% acetonitrile, 0.1% formic acid) was applied for the samples. A volume of 5 μL sample was injected and the peptides eluted with gradients of 4 to 76% ACN and 0.1% formic acid in water at flow rates of 0.25 μL/m. The samples were injected into a Q-Exactive Orbital-trap mass spectrometer (Thermo-Fisher GmbH, Germany) using electrospray ionization at spray voltage of 2.2 kV and measured twice in a data-dependent acquisition mode, selecting the top 10 peaks for HCD fragmentation. MS acquisition was performed at a resolution of 70,000 in the scan range from 300 to 1700 m/z. Dynamic exclusion was set to 30 s and the normalized collision energy to 26eV. The mass window for precursor ion selection was set to 2.0 m/z.

Proteomics Data Analysis and Bioinformatics

The recorded MS-files were analyzed using the MaxQuant software package (version 1.2.2.5) with the IPI-mouse database (version 3.84) complemented with frequently observed contaminants using oxidation and acetylation as variable modifications and carbamidomethylation of cysteins as fixed modification with a maximum of 2 missed cleavages.

Data obtained from the mass spectrometric data was log-transformed for the Aerosol/Gas replicates. Means for the four replicates were calculated and used for the determination of significantly regulated proteins. The cut-off for the experiments was set to 10% for the up- and downregulated proteins. The regulated proteins were used for the Gene Ontology analysis using the DAVID online tool [37].

Chemical and physical properties of the ship engine exposure aerosol

In order to understand the composition of the aerosols that the cells were exposed to, analysis of the chemicals in both the particulate matter (PM) (see ref. [14] and S1 Fig for a summary) and the gas phase (new data presented in this work in Fig 1) of ship engine emission from both fuels was performed. Note that the data presented in S1 Fig and Fig 1 are representing the exposure concentrations as subjected to the cells in the ALI systems (thus including the different dilution ratios that were applied to generally equalize the PM concentrations). The physical characterization (PM size distribution and mass) has been reported in detail previously [14]. This information, along with the new data on the gas phase composition (Fig 1), has been used to obtain as well as estimate and validate the estimated exposure dose for HFO and DF PM (see above). In the following, the previously reported chemical composition of the particulate matter is summarized and the here newly reported gas phase composition results of the exposure aerosol are discussed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Characterization of gas phase from HFO and DF aerosol (cell exposure concentrations including dilution).

The concentrations of each component measured in the aerosol are shown on the right, while the concentration ratios are shown on the left. Exponents refer to method used: (1) Gas monitor, (2) On-line photo ionisation mass spectrometry (REMPI-MS), (3) Gas chromatography mass spectromery, (4) On-line photo ionisation mass spectrometry (SPI-MS).

https://doi.org/10.1371/journal.pone.0157964.g001

The PM of the HFO exposure aerosols exhibits much higher concentrations of organic compounds compared to the DF PM. This includes the carcinogenic polycyclic aromatic hydrocarbons as well as aliphatic compounds, such as alkanes (S1 Fig). Investigations with modern high-resolution analytical methods, such as comprehensive two-dimensional gas chromatography with ultra-high resolution mass spectrometry, depict that HFO PM also exhibits a largely increased chemical complexity, as well as considerably higher amounts of toxic transition metals, such as vanadium, nickel and iron [14]. Conversely, the PM of DF aerosols contains more “carbonaceous soot,” measured either thermo-optically as elemental carbon (EC) or optically as black carbon (BC) [15]. The reduced EC and BC values (“soot”) in HFO exhaust is likely due to the high sulfur content of the HFO fuels, which quenches the formation of the larger aromatics and therefore also of soot in the combustion process [38]. In the previous study, the relatively higher content of rather pure carbonaceous soot in the DF PM was associated with the surprisingly high biological activity of DF PM. The results of the comprehensive gas characterization of the exposure aerosol are depicted in Fig 1. With the applied dilution differences, many gas phase compounds (including the bulk pollutants CO and NO) are 2 to 3 times more abundant in the DF exposure aerosol compared to HFO. One of the main differences is that the sum of the alkylated benzenes are found at higher concentrations in DF aerosol while the alkylated phenanthrenes (three-ring PAHs) are considerably more abundant in the HFO aerosol (Fig 1). These compounds are partially unburnt fuel residues and the differences are due to the different boiling point cuts in DF and HFO generation [15]. Smaller alkylated PAHs are currently suspected to be biologically active and potentially health relevant [39–41]. Carboxylic compounds in sum are about equally abundant in HFO and DF aerosol, with acetaldehyde and acetone being more concentrated in the HFO aerosol, while formaldehyde and other low abundant aldehydes are more abundant in the DF aerosol. Furthermore, the content of SO₂ is very high in the HFO aerosol compared to DF. The SO₂ concentration, however, is below the concentration where adverse effects are expected, which was determined in the previous study to be 2 ppm for transcriptomics, and 3.3 ppm for LDH release [14].

Toxicity of aerosol exposure to macrophage cells

DF and HFO aerosol (1:40 and 1:100 diluted, respectively) exposure both led to a detectable loss of membrane integrity due to cell death after 4 hours, with a slightly higher toxicity for the DF aerosol (Fig 2). When the particulate phase was removed from the aerosol through filtering, resulting in an exposure of the cells to the combustion gases only, the toxicity clearly decreased and no significant difference to either the controls or the cell-free “blank” medium could be observed.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Toxicity of aerosol exposure on macrophages occurs with exposure to both fuel types.

(a) Lactate dehydrogenase (LDH) assays were performed on cellular medium after exposure to complete aerosol and filtered aerosols (without particles). DF dilution was 1:40, while HFO dilution was 1:100 in order to obtain comparable particle levels. Blank represents cell free medium, used as a negative control. Triton represents cells lysed with Triton, used as a positive control. Levels represent the mean of three biological replicates (*** = p < 0.001. Error bars represent s.e.m.) (b) Density plot of Log2 fold change distribution of regulated proteins of RAW 264.7 cells in response to DF and HFO particles. RAW 264.7 cells show broadening of the protein regulation in response to DF particles (blue) in comparison to HFO particles (red), which indicates higher amount of regulated proteins in DF treated samples.

https://doi.org/10.1371/journal.pone.0157964.g002

The clear reduction of released LDH after particle removal indicates a predominantly particle mediated toxicity for DF aerosol exposure. For HFO aerosol exposure, the gas phase (filtered aerosol) shows some toxicity, although a 2.5 times larger dilution was set for the HFO emissions in order to establish comparable particle deposition doses. This suggests that the HFO gas phase shows more acute cytotoxicity than the DF gas phase, which might be attributed to the higher concentration of gaseous toxic compounds, such as gaseous aldehydes and ketones [27] or alkylated polyaromatic species in the HFO exhaust [39–41]. Thus, with the tested dilution ratios, the gases formed by the DF are not, or in the case of HFO combustion to a part, responsible for the observed aerosol toxicity. The main toxicity, however, is governed by the particle exposure, and it is likely that DF particles show even a stronger cytotoxic impact compared to the HFO particles. These results are in line with our previous study on lung epithelial cells, in which DF particle exposure was shown to produce larger biological effects than HFO particle exposure at comparable deposition doses [14]. These findings are surprising, as the concentration of known toxic compounds (such as PAH, oxygenated species, or heavy metals) is much higher in HFO particles than DF particles. Note that LDH release is a rather strong cytotoxicity endpoint, reflecting irreversible cell death by membrane damage. A lower LDH-based toxicity, therefore, does not exclude the presence of strong adverse cellular effects.

Results from the proteomics analysis show that particles from DF ship diesel combustion aerosols cause broader proteomic response in RAW 264.7 cells than particles from HFO ship diesel combustion. Fig 2 shows a larger distribution of proteins that are differentially regulated in response to DF in comparison to HFO, which indicates higher amount of up and down regulated proteins in DF combustion treated samples. This result is very similar to the previous observations in lung epithelial cells [14], and reinforces this finding in a further lung-relevant cell type. The broadening of protein regulation does not itself prove increased toxicity, but instead shows a wide biological reaction to the given aerosol exposure conditions. In conjunction with the cytotoxicity data, the proteomics results support for the conclusion that DF particle exposure leads to a higher cytotoxic effect in macrophages compared to the HFO particle exposure, likely due to the relatively high soot content of the DF aerosol.

Differential metabolic profiles of aerosol-treated macrophages

Non-targeted metabolomics analysis uncovered 230 compounds present in the four different conditions. After normalization and filtering out compounds present in less than 80% of each treatment, 203 compounds were subjected to principal component analysis (PCA) and ANOVA analysis. PCA analysis shows groupings based upon fuel type (70.8% of explained variance), with minimal separation due to particle phase (7% of explained variance) (Fig 3). This result is the opposite of the toxicity profile, suggesting that metabolic effects are due to the gas phase of the aerosol more than the PM.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Metabolic profile of macrophages exposed to combustion aerosols.

(a) Principal Component Analysis shows separation of fuel types, but no separation by presence of particles. (b) Heatmap represents metabolites with significantly differing abundances between treatments (ANOVA test, p < 0.01). Data was batch normalized, and metabolites found in less than 80% of all treatments were removed. DF: Diesel Fuel, HFO: Heavy Fuel Oil, f: filtered aerosol exposure (n = 4), uf: unfiltered (complete) aerosol exposure (n = 5).

https://doi.org/10.1371/journal.pone.0157964.g003

After ANOVA analysis, 167 compounds were found to be significantly differing between HFO and DF exposure at a cutoff of p < 0.01. Of the compounds discovered, 34 were able to be identified with our in-house metabolite library, and are listed in S1 Table. Clustering of significantly changed metabolite groupings show almost no separation between complete aerosol and gas phase exposure (Fig 3). This suggests that the effect of the particulate matter on the metabolic profile of macrophages is negligible, compared to the gas phase. Most compounds were found at increased levels with HFO exposure, while relatively fewer compounds (including Tyrosine and glycerol) are increased under DF treatment.

Certain metabolites found increased in HFO-exposed macrophages suggest a pro-inflammatory metabolic phenotype, specifically succinic acid and lactic acid (Fig 4). Succinic acid levels have been found to be increased in pro-inflammatory macrophages to aid increased inflammatory cytokine production [26]; this has been shown to lead to an increased glycolytic flux with reduced pyruvate intake into the TCA cycle [42]. Increased glycolysis with decreased pyruvate flux into the TCA cycle would shunt the flux towards lactic acid. This cellular metabolism profile is common in certain cellular conditions, and is often called the Warburg effect [43]. The Warburg effect is mostly associated with tumor cells, however has been recently found to be present in pro-inflammatory macrophages. [44]

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Levels of intracellular metabolite pools in aerosol-exposed macrophages.

Pools of (a) lactic acid, (b) succinic acid, and itaconic acid were measured. Levels represent the mean of 9 biological replicates. *** = p < 0.001. Error bars represent s.e.m.

https://doi.org/10.1371/journal.pone.0157964.g004

Another metabolite, itaconic acid, has been associated with antimicrobial activity of pro-inflammatory macrophages [45]. Itaconic acid has only been found in certain immune cells (including macrophages), and only when these cells were in a pro-inflammatory state. Therefore, it is a good marker for inflammatory macrophages. To identify this metabolite, experimental spectra were compared to an itaconic acid standard measured previously and integrated into our metabolite identification library (S2 Fig). Itaconic acid was found in all macrophages exposed to HFO aerosols, but was not present in DF-treated macrophages (Fig 4), a result independent of the presence of PM therefore an effect of gas phase exposure. Some itaconic acid was also found in the control cells, at higher levels than DF treatment. The DF treated cells have a much more complicated chromatogram, and this would increase the noise baseline for the analysis. The itaconic acid signal might fall below the baseline, and would therefore be not measured even if it is present in low amounts. However, its presence in macrophages stimulated with aerosols could point to a novel role for this metabolite in inflammation-governed environmental health effects.

The metabolic profile of HFO-exposed macrophages also shows increased levels of other compounds compared to DF-exposure. The increase of adenine and uracil, two bases involved in nucleic acid synthesis, suggests increased cellular DNA and RNA production, a process known to be increased in activated macrophages [46]. Adenine is also an important molecule for energy metabolism, present in both ATP and NAD⁺; its increase could be tied to an increased cellular energy production from increased glycolytic flux to lactic acid (which produces ATP, and aids in NAD+/NADH balancing), necessary for macrophages dealing with inflammatory stimuli.

Effects of aerosol treatment on metabolic dynamics: Carbon flux from glucose to intracellular metabolites

Stable-isotope labeling provides a complementary source of information on how glycolytic flux is impacted by combustion aerosol exposure. By looking at the labeling pattern of certain metabolites in cells cultured with uniformly ¹³C-labeled glucose, we were able to gain insight into how the glucose is metabolized in the cell through central carbon metabolism, and the rates of glycolysis and glucose oxidation can be compared between different conditions. We measured the labeling in lactic acid, giving more information into glycolytic flux, as well as two metabolites from the TCA cycle (fumaric acid and glutamic acid), which allows the tracing of how glucose is differentially oxidized through the TCA cycle with exposure to the different combustion aerosols.

While both aerosol types lead to a similar amount of lactic acid labeling derived from glucose (Fig 5), labeling of TCA cycle intermediates derived from glucose were significantly reduced in DF exposure compared to HFO exposure (Fig 5). These results suggest that DF-exposed macrophages have a decreased amount of relative glucose oxidation into the TCA cycle compared to macrophages with HFO-exposure. As HFO stimulates a pro-inflammatory phenotype in the macrophages, this decrease in glucose oxidation is presumably tied to an alternative metabolic phenotype. As it is known that macrophages exhibit a wide variety of phenotypic states depending on the stimulant [47], the metabolic changes seen here suggest a macrophage state slightly different from canonical activation states.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Relative oxidation of glucose in the TCA cycle for aerosol-exposed macrophages.

Metabolism of U-¹³C₆-Glucose was measured in macrophages after exposure. (a) M2 fumaric acid isotopologue levels, (b) M2 glutamic acid isotopologue levels, and (c) M3 lactic acid isotopologue levels. Levels represent the mean of 4–6 biological replicates. *** = p < 0.001. Error bars represent s.e.m.

https://doi.org/10.1371/journal.pone.0157964.g005

Effects of combustion aerosol treatment on the proteome in RAW 264.7 macrophages

Analysis of the changes to the proteome of exposed RAW 264.7 cells support the findings of the metabolomics analysis, with differing regulation of proteins by HFO particles compared to DF particles (Fig 2 and S3 Fig). An analysis for gene set enrichment of the regulated proteins using the DAVID online pathway analysis tool [37] showed that different pathways and biological processes are induced in response to DF and HFO particles. Two of the main pathways seen to be affected by PM exposure are endocytosis and the activation of the immune response.

Proteins relating to the GO term endocytosis (GO:0006897) were found to be upregulated in both HFO and DF treated samples, while the different fuel types induced activation of different groups of proteins involved in endocytosis (S8 Fig). Proteins involved in the immune response pathway (GO:0006955) were found to be upregulated only in HFO treated macrophages (p = 0.059), while DF stimulation shows no significant upregulation of the pathway (Fig 6). This data supports the metabolic analysis in so far that enrichment of the immune response pathway GO term indicates an activation of inflammatory pathways on a larger scale, mainly through the NF-kappa-B (NF-kB) signaling pathway (Fig 6).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. HFO particles induce activation of immune response in RAW 264.7 macrophages.

(a) The Gene Ontology term GO:0006955, corresponding to activation of immune response, was found to be significantly up-regulated in HFO-treated samples (p = 0.059) and not regulated in the DF-treated samples. (b) Model of how the regulated proteins found in this study affect the NF-kB immune response pathway in the cell. Stimulation of the toll-like receptor (TLR2) leads to activation of NF-kB. Tumor necrosis factor alpha-induced protein 8-like protein 2 (TNFAIP8L2) acts as a negative regulator of TLR2, preventing hyperresponsiveness of the immune system, and inhibiting NF-kappa-B activation. Peroxiredoxin 2 (Pdrx2) reduces hydrogen peroxide, inhibiting NF-kappa-B activation.

https://doi.org/10.1371/journal.pone.0157964.g006

Stimulation of the toll-like receptor (TLR2) leads to activation of NF-kB, which plays a key role in regulating the immune response. Tumor necrosis factor alpha-induced protein 8-like protein 2 (TNFAIP8L2) acts as a negative regulator of innate and adaptive immunity by maintaining immune homeostasis. TNFAIP8L2 prevents hyperresponsiveness of the immune system by negative regulation of TLR2, and inhibition of NF-kappa-B activation [48]. Peroxiredoxin 2 (Pdrx2) reduces hydrogen peroxide and is involved in redox regulation of the cell. Prdx2 inhibits NF-kB activation, which is induced by H₂O₂, and regulates immune response [49].

These findings in macrophages are in line with the previous study on lung epithelial cells [14]: the endocytosis pathway was found to be upregulated after aerosol exposure from both fuel types, while the immune response was only upregulated after HFO exposure.