In the present study, we aimed to reveal a molecular signature of the toxicity induced by different organic compound mixtures extracted from diesel exhaust particles produced by diesel/renewable fuels in human lung BEAS-2B cells. A comparative analysis of gene expression changes was used to characterize the common mode of action underlying DEP extract exposure as well as specific effects of individual DEP extract treatments.
Recent studies on the toxicity of different diesel and biodiesel exhaust particles suggest inconsistent data due to different running conditions of engines and experimental approaches [23
]. Here, we used a standardized procedure providing the same conditions for particle collection, extraction and cell treatment thus enabling the accurate comparison of DEP extracts on gene expression changes. Although in vitro exposure to whole particles is more relevant in terms of simulation of real world conditions, many difficulties arising from intact particle collection, the characterization of particles and their behavior in culture medium may consequently complicate the interpretation of results. It has been documented that organic compounds bound to DEP are responsible to a large extent for its genotoxicity and other effects [12
]. To the best of our knowledge, a study investigating effects of organic extracts from various fuels engine emissions on global gene expression changes in a model cell line system has not been published yet, so comparable data are not available. However, several authors reported the results of genotoxicity/mutagenicity of similar organic extracts. Bulky DNA adduct formation in a cellular calf thymus DNA system was consistently increased after treatment of the samples with extracts containing higher levels of PAHs [12
]. In another study that investigated mutagenic properties of organic extracts of diesel and biodiesel fuels, higher PAHs content was associated with increased mutagenicity and bactericidal effects in Salmonella typhimurium
TA98 and YG1041 strains [19
]. These reports are in line with our observation that application of NEXBTL100 extract that contains lower levels of PAHs results in weaker biological response. This suggests that not only genotoxicity/mutagenicity, but also global gene expression changes are strongly affected by the presence of PAHs in the samples.
3.1. Common Cellular Response—4 h Cell Exposure
Despite the variability in chemical composition of individual DEP extracts, we revealed numerous genes and pathways altered in the same manner. Commonly deregulated genes following 4 h exposure were mostly involved in oxidative stress response and consequent events, such as activation of Nrf-2 and AP-1 transcription, antioxidant defense and DNA damage response.
The most significantly deregulated genes were AKR1C2
. A wide substrate specificity of these enzymes determines their implication in the metabolism of various exogenous and endogenous compounds such as steroids, sugars, carbonyls and others. They have a dual role in toxicity: under the control of Nrf-2 transcription factor they exhibit a detoxifying function by conversion of toxic aldehydic products [26
] or catalyze the NADPH-dependent reduction of the o
-quinone products to catechols and thus exacerbating ROS formation [27
]. Induction of AKR1C1
was also observed in BEAS-2B cells exposed to urban particulate matter in the study of Longhin et al., 2016 [28
Polycyclic aromatic hydrocarbons and other organic compounds are capable of producing a substantial amount of ROS, which consequently lead to stabilization and activation of transcription factor Nrf-2 and induction of antioxidants and detoxifying enzymes [27
]. Nrf-2 participates in the regulation of oxidant-stimulated functions, such as autophagy, inflammasome assembly, ER stress/UPR, mitochondrial biogenesis or stem cell regulation as well as protects against toxicity and chronic diseases in normal cells or through pharmacological interventions [29
In our study, we observed elevated expression levels of HMOX1
suggesting anti-oxidant response against ROS production. Surprisingly, we were not able to detect an increase in ROS production by carboxy-H2
DCFDA assay. The same effect was reported in the study of Li et al., 2002 [30
] where DEP extract treatment failed to induce DCF fluorescence in BEAS-2B, while the same treatment and detection method was effective in different cells. However, the authors confirmed the pro-oxidative potential by using a different method reflecting mostly production of superoxide radical. They also demonstrated a decrease of the GSH/GSSG ratio in BEAS-2B cells but not in THP-1 macrophages, as well as increased expression of HMOX1
, pro-inflammatory cytokine IL-8
, activation of JNK and decreased cell viability. In contrast, we observed no effect on reduced glutathione levels, suggesting effective replenishment of depleted GSH stores possibly due to the enhanced expression of GLCM
. Moreover, other recent studies have demonstrated the pro-oxidative and pro-inflammatory effect in BEAS-2B and other cell lines upon exposure to DEP chemicals [31
]. Importantly, the authors underscore the close relation of increasing oxidative potential of DEP extracts and the higher content of PAHs.
An excessive amount of ROS can further activate intracellular signaling cascades, including the mitogen-activated protein kinase. We detected significant upregulation of HSPB1
, a protein chaperon involved in stress resistance and regulation of apoptosis. HSPB1
maintains glutathione in its reduced form and decreases the amount of reactive oxygen species (ROS) produced in cells exposed to oxidative stress or tumor necrosis factor TNFα [32
]. This could further support the hypothesis of a vigorous antioxidant response, which effectively neutralizes ROS and restores GSH levels.
ROS generation by DEP extracts arising from enzymatic metabolism of organic compounds as well as formation of reactive intermediates possibly cause DNA damage with consequent cell cycle arrest, senescence and cell death. Our data strongly suggest activation of p53 signaling due to the modulation of genes involved in “DNA damage response (ATM induced) pathway” (BIK
). We also observed the induction of SFN
, a direct p53 effector, suggesting cell cycle arrest in response to DNA double strand breaks. Suppression of CCNB2
by all extract treatments, as confirmed by qRT-PCR, may indicate inhibition of cell cycle. It has been shown that the activity of CCNB2
, a key factor essential for transition from G2 to mitosis, is under the control of p53 which represses transcription of CCNB2
and causes arrest in G2 phase upon DNA damage [33
]. Elevated expression of GLIPR1
, another p53 target, could further contribute to increased ROS production, promote cell cycle arrest and apoptosis, as observed in prostatic cancer cells [34
]. On the other hand, pro-apoptotic BNIP3
was significantly downregulated by all treatments. This suppression of BNIP3
might be mediated by p53 acting in favor of the protection against hypoxia-induced cell death as documented in Feng et al. [35
]. Suppression of CDKN2A
, a stabilizer of p53 protein, BIK
may suggest the suppression of p53 activity and anti-apoptotic response of the cells. FOSL1
induction might be related to AP-1 transcription, induction of plasminogen activator urokinase (PLAU
) and modulation of cell adhesion [36
]. Other genes contributing to DNA damage response indicated cellular senescence and/or autophagy (PLAU
The role of plasminogen activator (PLAU
) and its inhibitor (SERPINB2
) in senescence is not fully understood; however, the study of West et al. [37
], described alterations in plasminogen activator activity during replicative senescence leading to the disruption of extracellular matrix maintenance with possible deleterious consequences on tissue homeostasis. According to our results, in the study of Longhin et al. [28
], the authors also found a strong induction of SERPINB2
in BEAS-2B cells exposed to whole airborne particles. Exposure to air pollution may cause acute exacerbation of idiopathic pulmonary fibrosis [38
]. Interestingly, enhanced expression of PLAU
and simultaneous suppression of CTGF
, a growth factor promoting the fibrosis, observed in our study, may suggest the anti-fibrotic response. Plasminogen activator and plasminogen activator-inhibitor are components of the plasminogen activation system, which has been implicated in fibrosis reduction [39
]. Similarly, a defensive role of Nrf2 target, TXNRD1
, against fibrosis has also been described [40
]. Moreover, activation of peroxisome proliferator-activated receptor α (PPARα) participating in the processes of both physiological and toxicological response to various endogenous or exogenous substances may also protect against lung fibrosis [41
]. The multiple regulatory role of PPARα in various processes related to oxidative stress, lipid metabolism and inflammation has been documented [42
3.2. Differential Response Detected by Analysis of Variance—4 h Cell Exposure
A vast majority of genes detected by ANOVA among all treatments were distinctively modulated in response to NEXBTL100 treatment and exhibited more than 1.5-fold change in expression levels compared to a group median involving other gene sets (B0, B30 and B100). Most of the NEXBTL100-specific genes were involved in DNA replication, cyclin B2 related events and entry into mitosis. Elevated expressions of MCM8, RBL1, RPS27, CDC25C, ENSA and ZWINT may suggest deregulation towards enhanced proliferation compared to other DEP extract treatment. Accordingly, repression of RAE1, a mitotic checkpoint regulator, as well as CDKN2A, a tumor suppressor that inhibits G1/S transition and establishes cell cycle arrest, could further support the hypothesis of enhanced proliferative potential of NEXBTL100 extract.
Analysis of top-ranked over-represented pathways specific for each DEP extract treatment showed that “Keap-Nrf2 pathway” and “Glutathione biosynthesis pathway” were the most significantly affected by NEXBTL100 4 h exposure. However, NEXBTL100 induced the lowest levels of HMOX1, TXNRD1 and GPX1 suggesting modest anti-oxidative response, possibly caused by the lowest production of ROS. It should be stressed that changes were not significant compared to unexposed controls. However, subtle changes in gene expression levels detected by ANOVA may also contribute to distinctive gene expression patterns of individual DEP extracts treatments. These results indicate that NEXBTL100 induced the weakest oxidative stress response and DNA damage and exhibited elevated expression of genes possibly contributing to increased proliferation compared to other treatments.
On the other hand, the list of the most over-represented pathways modulated by B0, B30 and B100 was similar to each other and involved “Benzo[a]pyrene metabolism” and similar pathways (with AKR1C2 and CYP1B1 being the most contributing deregulated genes) and “Plasminogen activating cascade” (with a significant contribution of SERPINB2, PLAU, PLAT). Additionally, B100 extract, which contained the highest concentrations of carcinogenic PAHs, also affected most significantly the “Cell cycle”, “p38 signaling” and “Senescence and autophagy” pathways (data not presented here).
3.3. Common Cellular Response—24 h Cell Exposure
The major toxic response following 24 h exposure to all DEP extracts was the metabolic activation of PAHs. The key enzymes contributing to modulation of “Benzo[a]pyrene metabolism” and “Metabolism of xenobiotics by cytochrome P450” as well as numerous other pathways were AKR1C2
. AKRs participate in o
-quinone pathway by conversion of PAH-diols into redox active PAH o
-quinones, but also facilitate the redox cycling of the PAH o
-quinones to catechols. Catechols are able to conjugate with a wide range of conjugating enzymes. Conjugation terminates the redox cycling, eliminates formation of electrophilic products and prevents formation of covalent adducts [43
]. We also confirmed significantly elevated expression of other important PAH-metabolic enzymes CYP1A1
by qRT-PCR, although microarray data did not indicate a significant increase, possibly due to the high variability among replicates. CYP enzymes are involved in the formation of trans
-dihydrodiols, the first step of PAH activation and also in the consequent event when dihydrodiols are converted into diol-epoxides, which can covalently bind to DNA and form persistent DNA adducts. The competing role of CYP and AKR enzymes in the metabolic activation of PAH-diols has been observed in human bronchoalveolar cell extracts [44
The induction of CYP enzymes is dependent on the activation of AhR. Our recent data of AhR-mediated activity performed in human AZ-AhR cells (Stable HepG2 Luciferase Reporter Cell Line) by CALUX assay, suggest similar TEQ values for B0, B30 and B100 extracts and lower TEQ value for NEXBTL100 extract (see Figure S2
), which accordingly reflected a lower PAH content. The similar trend could be also expected for BEAS-2B cell line. Our previous findings indicate that concentration of PAHs in a mixture of organic compounds extracted from reference urban dust particulate matter is higher [45
] than in similar organic extracts from standard reference diesel exhaust particle material [25
]. Therefore, also AhR-mediated activity of DEP extracts was lower. These results are in line with our present findings where only partial induction of CYP1A1 and CYP1B1 mRNA confirmed the low activation of the AhR-dependent gene expression. Interestingly, a recent study of Palkova et al., 2015 [25
] analyzing the toxicity of organic fraction extracted from reference material of diesel exhaust particles (SRM 1650b) consistently evidenced its high AhR-inducing activity and potency to induce metabolic enzymes, trigger DNA damage response and disrupt cell cycle progression and proliferation in rat lung epithelial cells RLE-6TN and rat liver epithelial cells WB-F344 in the concentration range of 100–1000 and 50–1000 µg/mL, respectively, suggesting a higher potency to metabolize PAHs compared to other cell lines.
Interestingly, a higher metabolic rate of organic compounds probably caused the interference with WST-1 assay and gave false positive results of “enhanced” proliferation upon low dose exposures to DEP extracts. Since WST-1 assay is based on conversion of tetrazolium dye by mitochondrial dehydrogenases, the results obtained by WST-1 assay more likely imply enhanced metabolic activity of cells due to the effect of organic compounds and not increased number of cells.
3.4. Differential Response Detected by Analysis of Variance—24 h Cell Exposure
Similar to 4 h incubation, ANOVA also revealed distinct gene expression patterns following 24 h incubation across all treatments. Functional annotation of genes identified by ANOVA revealed that the most distinctive signature was induced by NEXBTL100 extract treatment. Specifically, we found a marked difference in expression levels of genes associated with regulation of chromosome segregation and cytokinesis. Among them, AURKA
is critical for the proper formation of mitotic spindle [46
], while CENPA
controls kinetochore assembly and chromosome segregation [47
] and KIF20A
, a mitotic kinesin, is required for chromosome passenger complex (CPC)-mediated cytokinesis [48
]. It has been demonstrated that p53 acts as a negative regulator of AURKA
activity and reduces its expression level after DNA damage [49
]. Deletion of AURKA
may consequently cause cell cycle arrest. AURKA
requires a number of co-factors for its activation such as microtubule associated protein TPX2
and GTPase Ran. Ran releases TPX2
to bind and activate AURKA
by changing its conformation, stimulating its autophosphorylation and targeting it to spindle microtubules at the pole [50
]. Importantly, overexpression of AURKA
has been linked to tumor development at different levels [51
]. We confirmed the elevated expression of TPX2
upon NEXBTL100 extract treatment by RT-qPCR, while other treatments rather slightly suppressed gene expression level. TPX2
is an essential regulator of spindle function and may indicate the possible negative effect of DEP extracts (excepting NEXBTL100) on mitosis and generally on cell cycle progression [52
3.5. Study Limitations
Although our study represents a comprehensive analysis of toxic effects of organic extracts from various fuels engine emissions, there are several limitations that should be acknowledged.
The first limitation is related to the nature of the samples used for in vitro tests. The results reported here are based on exposing the cells to the extracts from comparable masses of particles. The results therefore represent the “quality” of the particles, expressed as some metric of effect per mg of particulate matter, a metric suitable for attempting to comprehend the mechanisms of the effects of particles on human health. For realistic evaluation of fuels, however, the effects of fuels should be compared based on distance driven, amount of useful work performed by the engine, fuel consumed, or similar metric. Since the total mass of particles emitted varied among the fuels (see Figure 1
B), and the effects are believed to be nonlinear, the results do not necessarily represent the realistic effects of a fuel substitution on human health. In addition, due to the anticipated nonlinearity of effects with respect to dose, expressing the effects per kg of fuel would be of limited use. The real exposure is not to raw, undiluted exhaust, and the effective dilution ratio (reciprocal of the fraction of raw exhaust in inhaled air) considerably varies with conditions (mainly proximity of vehicles to population and atmospheric conditions). Evaluation of “realistic” fuel effects would therefore necessitate an arbitrary and difficult to justify assumption of a certain dilution ratio.
Another limitation is the fact, that we used organic extracts rather than exhaust particles to investigate biological effects of various fuel emissions. Although particles mediate some unique effects, particularly oxidative stress induction that cannot be directly mimicked by extracts, organic fraction contains PAHs, i.e., compounds with highest genotoxic activity and most immediate impact on human health. It should be also noted that the use of organic extracts increase the bioavailability of PAHs compared to whole particles experiments therefore the results may overestimate the effects of DEP-associated PAHs. Oxidative stress induction measured in our study yielded mostly negative results, although this may be caused by technical limitations of the test methods.
Finally, although gene expression profiling is a robust tool that covers simultaneously whole-genome gene expression changes, it still measures a single endpoint (mRNA levels). To get more insight into the mechanism of action of tested toxicant(s), gene expression profiling experiments should be integrated into larger studies examining multiple end-points at the molecular, cellular, tissue, and physiological levels in the context of the whole organism. Therefore, our study does not aim to reveal the precise mechanism of action (e.g., cell cycle regulation or other fundamental cellular processes) but rather provides new testable hypothesis that could be subsequently confirmed by further experiments.