Crude oil is a complex mixture that includes polycyclic aromatic hydrocarbons (PAHs) as one of its major components. The toxicity of some chemically substituted PAHs found in oil, such as the methylated species, are relatively understudied. A combination of chemical fractionation and analysis coupled with a bioassay was used to identify a subset of oil PAHs that activated aryl hydrocarbon receptor (AHR). Silica gel chromatography was used for primary and secondary oil fractionation, and standard and reverse phase high performance liquid chromatography (HPLC) were used for the final fractionation steps. Both gas chromatography (GC-) and HPLC-coupled with mass spectrometry (MS) were used to separate and identify compounds present in the petroleum fractions. Bioactivity of the individual fractions was identified and measured using a recombinant yeast strain that expressed the human aryl hydrocarbon receptor complex (AHRC) transcription factor that is composed of human AHR and the ARNT proteins. AHRC activation by oil components results in expression of β-galactosidase, and readout from this enzymatic activity is proportional to the amount and potency of the compounds that activated the system. Silica gel separations produced 25–29 fractions that were assessed for bioactivity using the AHRC reporter system. Bioactivity peaked with the fractions that contained larger PAHs that included four ring compounds such as the triphenylenes, benzanthracenes, and chrysenes (MW 228 + additional methyl groups). When tested as individual compounds, the triphenylenes and benzanthracenes were less potent than the chrysenes, so the latter constituted more of the AHRC signaling activity in the oil fractions. The chrysenes in bioactive fractions were mixtures of the parent compound along with mono-, di-, tri-, and tetra-methyl derivatives and other PAHs. The six possible mono-methylchrysenes were obtained and tested for AHRC activity and for their concentrations in oil. Chrysene, 1-, 2-, 3-, and 6-methylchrysene were present, but 4- and 5-methylchrysene were not detected in the bioactive fractions of oil that were resolved by HPLC. When tested individually in the AHRC bioassay, 4-methylchrysene was the most potent ligand, and 5-methylchrysene was the least potent. Synthetic mixtures of PAHs were reconstructed based upon the chemical composition of one fraction with the high AHRC activity. Collectively, these data show that: 1) the six methylchrysene isomers are within an order of magnitude of chrysene in their ability to activate the AHRC bioassay; 2) although they are a minor group, the chrysene compounds in oil potently activate AHRC signaling; 3) chrysenes diminish as oil weathers, while triphenylenes of identical molecular weight persist, 4) this methodology can be useful for identification and characterization of the bioactivity of sub-fractions and individual compounds found in oil.
Crude oil is well recognized as a toxic agent to people and organisms in the environment (1 –3 ). Some of the individual compounds in oil have been well studied, but many of the chemical constituents have not been examined for bioactivity or toxicity. Here we have compared fresh and weathered oil from the MC252 spill for the endpoint of aryl hydrocarbon receptor (AHR) activation. Pyrogenic (combustion related) PAHs interact with a cellular protein called the aryl hydrocarbon receptor as one way in which they ultimately cause toxic effects (4, 5 ). It is likely that the same is true for some of the less studied petrogenic PAHs and related compounds in oil since they are structurally similar to the pyrogenic PAHs. Here, we report on coupling of a bioassay with separation technology and analytical chemistry to help identify and characterize some of the important PAHs in oil.
1. Oil samples
Most of the data presented here was obtained using fresh (unweathered) oil from the MC252 well. Additionally, MC252 oil samples were obtained from the National Institute of Standards and Technology (NIST). NIST samples SRM 2779 (fresh oil, https://www-s.nist.gov/srmors/view_detail.cfm?srm=2779) and SRM 2779 (highly weathered, https://www-s.nist.gov/srmors/view_detail.cfm?srm=2777) have been extensively characterized. Oildimethylsulfoxide (DMSO) extracts of both fresh and weathered samples were prepared by mixing 100 ul of the oil sample with 900 ul of DMSO in a microcentrifuge tube. The samples were mixed on the highest setting of a vortex machine and then extensively mixed on a shaker at 300 rpm for 24 hr. The oil was allowed to phase separate without movement for 1 hr and then the DMSO extract portion was carefully isolated. Most of the oil did not dissolve into the DMSO phase, so the DMSO extract only contains a small portion of the total oil.
2. Oil Separation
We developed a methodology to fractionate oil to identify components with more or less AHRC activity as shown in the Figure 1. The first silica gel column chromatography separations used 20 ml of crude oil dissolved in 50 ml of hexane, with hexane and methylene chloride as a mobile phase. A total of 16–20 fractions of equal volume were collected in repeated trials with the first silica gel column. The separation began with pure hexane followed by gradually increasing the solvent polarity by using methylene chloride. Monitoring with a UV lamp allowed for identification of the aromatic fractions. When the non-fluorescent alkanes eluted, the polarity of the hexane mobile phase was gradually increased from 3% to 20% methylene chloride to separate the aromatics. The alkanes eluted in the first five fractions, followed by mono aromatic fractions, then the polyaromatics, and finally the polar compounds were resolved as shown in Figure 2. A dark brown material that did not elute from the silica gel column was probably composed of asphaltenes and other heavier resin type compounds.
The most active compounds in the yeast bioassay eluted in a fraction that contained various chrysenes, pyrenes, fluorenes, and lesser amounts of methylphenanthrenes. These active fractions were then fractionated again into additional sub-fractions using a second silica gel. A total of 25–29 equal fractions were collected using 3% dichloromethane in hexane as the mobile phase, which were concentrated by evaporation to 2 ml and characterized by GC-MS. Once the fractions were chemically characterized, portions were used for further separation by HPLC. Other portions of these fractions (0.1 ml) were fully air dried and then dissolved in an equivalent volume of dimethylsulfoxide (DMSO), such that they could then be used in bioassays. Hexane and dichloromethane were not compatible solvents for bioassays. DMSO is a biocompatible solvent that can was used in concentrations of 1% in cell culture without overt cytotoxicity.
2. Recombinant Yeast Screen for AHR Activating Chemicals
Screening fractions with an AHRC reporter assay makes it possible to identify the principal bioactive AHR ligand compounds within complex mixtures of chemicals such as oil. A genetically engineered yeast strain (S. cerevisiae) named YCM3 was used as the bioassay. The methodology for the yeast AHRC reporter assay has been described (6, 7 ). YCM3 is available as product MYA-3637 from American Type Culture Collection. This yeast expresses the human aryl hydrocarbon receptor (AHR) and the aryl hydrocarbon receptor nuclear translocator (ARNT) proteins. The yeast strain contains a reporter gene (lacZ) which encodes the enzyme β–galactosidase. The promoter region of the lacZ reporter gene was engineered to contain five binding sites for the AHRC transcription factor. Structurally suitable ligands bind AHR and “activate” it. The ligand activated AhR moves into the cell's nucleus and partners with the ARNT protein, becoming a transcription factor called AHRC (aryl hydrocarbon receptor complex). The ligand-bound AHRC then binds the specific DNA sequences in the lacZ reporter gene, triggering its expression. The degree of expression of the reporter gene results in a proportional (graded) expression of the β–galactosidase protein it encodes. The amount of β–galactosidase enzyme is then measured by adding a colorimetric substrate (o-nitrophenol-β-galactopyranoside, ONPG). The o-nitrophenol liberated by the enzymatic action on the substrate quantitatively reflects the potency and efficacy of the AHR ligand(s) that initiated the process. This human AHR reporter assay constructed in yeast has the same transcription factor function as in a vertebrate cell. Activity of the AHRC regulated reporter gene serves as a surrogate for the genes expressed in organisms such as fish or humans when they encounter ligands such as PAHs. Coupling the yeast bioassay with chemical separation and analysis results in a powerful methodology for dissecting oil chemicals toxic response. The process of fractionation and analysis can be repeated multiple times to reach a condition of separation and bioactivity identification. Screening the oil fractions is described in the Figure 1. The bioassay is conducted in 96 well plates with a volume of 200 ul synthetic galactose medium per well (7). Two microliters of DMSO (1% final volume) containing an oil fraction or an individual compound is added. The positive control AHR ligand, β-naphthflavone (β-NF dissolved in DMSO) to a final concentration of 1 uM that provides for optimal expression of the reporter gene. Signaling from this β-NF control is set to 100% and other treatments are expressed as percent AHR signal relative to the positive control. The bioassay is conducted in triplicate and repeated at least three times on separate days. Results are reported as fitted dose response curves, mean effective concentrations that produce a 50% response (EC50) with standard deviations, and relative potency factors (relative to the EC50 for chrysene).
The graph in Figure 3 reflects the screening of MC252 oil fractions and the prominent peak of AHR ligand activity around fraction 22. Most fractions exhibited activity, but with serial dilutions the PAH concentrations became low enough to resolve the more potent components. The data are shown as a dilution rather than an actual concentration since the samples are complex mixtures of compounds. The more active fractions (18–25) were further sub-fractionated and analyzed to reveal a composition of larger 4-ring compounds, many of which were methylated. Figure 4 shows compounds identified within one of the highly bioactive sub-fractions that contained chrysenes and other alkylated PAHs. Sub-fractions such as this were then be resolved further using HPLC. While we lacked standards for many of these methylated PAHs, we were able to obtain and use all six of the mono-methyl isomers of chrysene as references to identify their relevant proportions in fractions. The results of this quantitative assessment are shown in Figure 6. There were about three times more mono-methylchrysenes than chrysene itself, with the predominant isomers being the 1-, 2-, 3, and 6-methyl forms. Interestingly, the 4- and 5- methylchrysene isomers were barely detectable in this oil sample and were always relatively low when compared to the other methylated chrysenes.
Based upon analysis of various fractions, we attempted to reconstitute the AHRC signaling components found in one of the prominently active sub-fractions. First, we tested pure PAHs that were commercially available and determined EC50 (effective concentration giving a 50% response) in the AHRC bioassay. These results are shown in Table 1. In general, the chrysene family of compounds was more potent than other classes of PAHs. Next, we added together chrysene and mono-methylchrysenes in proportional concentrations shown in Figure 5 and compared this synthetic mix to the parent sub-fraction. Even though the chrysene compounds in the synthetic mix were a minority of the total PAHs in the fraction, they produced about a third of the AHRC signal relative to the fraction (data not shown). This result suggests that small amounts of the more potent PAHs plus larger amounts of the less potent PAHs constituted the whole of the AHRC signal in the sub-fraction.
Figure 6 shows the structural features of three major 4-ring PAHs that were present in the more bioactive fractions of MC252 oil. Additionally, there were related compounds present, usually in more abundance, with 1 to 4 methyl groups (e.g. from mono- to tetra-methylated forms) in the oil. The chrysene structure is more elongated and perhaps it is this feature that makes chrysenes generally better ligands for AHR as indicated in Table 1. We compared DMSO extracts of SRM 2779 (fresh MC252) oil standard to the relevant aged reference SRM 2777 in the AHRC bioassay. In Figure 7, a decline of about an order of magnitude in the signaling activity was observed for the weathered oil relative to the fresh oil. This loss of AHR activity suggests that the oil is losing potency as well as changing in composition as it ages in the environment. Losing the stronger AHR ligands, such as the chrysenes, might account for the drop off in AHR activity in the aged oil sample.
Figure 8 shows chemical analysis of fresh and weathered oil samples over time. As degradation by various processes occurs with time, the methylchrysene component of the oil diminishes and methyltriphenyleneisomers are more persistent. The difference may be due to physical and biochemical differences between these two classes of molecules. Since the two classes of compounds share molecular weights and have similar chromatographic properties, mistaken identification might occur if internal standards are not present. If the more bioactive chrysenes are being eliminated by the environmental degradation, then a loss of AHRC activity would be expected.
We have described a combined approach that paired oil fractionation and analytical chemistry with cellular bioassays to follow a particularly relevant bioactivity, namely, the activation of the aryl hydrocarbon receptor (AHR) to form the aryl hydrocarbon receptor complex (AHRC). AHRC activation by oil compounds is quantitatively indicated by the degree of expression of a reporter gene (β-galactosidase) in this bioassay. Activation of the AHR is one of the key pathways that leads to downstream effects and toxicity of PAHs. Numerous studies have shown that when AHR is not present, normally toxic ligands such as dioxins and benzo[a]pyrene lose their effects (4 ). Thus, activation of the AHR transcription factor is a key event that starts a toxic response that is common to numerous chemicals, including many of the multi-ringed PAHs in oil (5, 8 ).
In Figures 1 and 2 we indicate our methods for conducting multiple fractionations and screening to find some of the specific bioactive compounds present in oil. Results presented here demonstrate the plausibility of this approach and that it is possible to go down as far as to resolve and characterize individual compounds. One of the problems encountered was that many of the compounds we detected were methylated species, but we lacked the commercially available standards to go further and map methylation positions on these molecules. For example, only a few di-methylchrysene isomers and no tri- or tetra-methylchrysene iosmers are available as commercial products. This limitation stopped us from going further into analysis of these species, and we still know little about these compounds other than they are present and relatively abundant as chrysene family members in crude oils. We were, however, able to obtain all six mono-methylchrysene isomers and compared them to the parent chrysene response. The tertiary fractions from the screen, shown in Figure 1, revealed that two of the methylchrysenes, the 4-and 5- isomers, were greatly underrepresented in the oil (Figure 4). This is interesting in that the 4-methyl isomer was very potent in the AHRC assay (Table 1) and that 5-methylchrysene, which is a well-documented combustion by-product, is reported to be the most carcinogenic member of the chrysene family (reviewed by IARC, and National Toxicology Program, Report on Carcinogens). Thus, the oil may be particularly lacking in two of the methylchrysenes that might be the most bioactive species. Additionally, this result might be informative regarding how alkylated PAHs form in the conditions found deep in the earth. The preferential genesis of specific alkylated species might reflect the substrates and/or the chemistry that is occurring during alkyl-PAH formation.
We compared AHRC transcriptional activation by chrysene compounds both in the yeast-based AHR signaling bioassay and in human HepG2 cells. The results from the HepG2 cells were previously published (8). The yeast bioassay uses the human AHR and ARNT proteins as the transcription factor, and the HepG2 human cell line has endogenous AHR and ARNT proteins that induce target gene expression, particularly cytochrome P450 1A1. Ideally, we would have observed complementary results from the yeast and HepG2 transcriptional assays, but they were somewhat distinct. The HepG2 cells did not recognize the chrysene isomers as being particularly different, but the yeast assay showed more variability, with 4-methyl being an exceptionally potent AhR ligand. One reason for the difference may have been that the HepG2 cells are metabolically competent and readily degrade chrysenes (9 ). This may explain the need for higher concentrations of chrysenes needed in HepG2 cells for cytotoxicity and expression of cytochrome P450 1A1 protein. This metabolic capacity may have buffered some of the more subtle concentration differences that the yeast assay detected. There are several caveats to consider for these in vitro studies. It is important to note that the concentrations of PAHs reported here and in the study with HepG2 cells are high and may not be environmentally relevant. The reported aqueous solubility limits of the PAHs used here are less than the amounts we used in these assays, raising the potential issue of PAH insolubility in the in vitro assays. The use of cell culture media rather than water may permit higher amounts of PAHs to dissolve, and the addition of DMSO as a solvent vehicle may also help with solubility. If the PAHs and mixtures we are testing are indeed above the solubility limits in the bioassay media then it compromises the validity of the concentrations that we have reported. It is also important to note that the yeast and HepG2 cell assays are simple in vitro model systems that may not fully mimic more complex in vivo models. The in vitro models are more useful for detection and mechanistic studies rather than predicting in vivo biological effects such as toxicity.
The major PAHs detected in sub-fractions as shown in Figure 5 were composed of a mixture of 4-ring PAHs along with many methylated isomers and other compounds. Figure 6 shows the parent molecular structures. The chrysene molecule has a more elongated, rectangular structure that may account for the greater potency in activating AHR signaling, whereas the triphenylene and pyrene molecules are more compact. These structural differences might also explain why chrysenes might be shorter lived in the environment. Being stronger AHR ligands with more extended structure, the chrysenes might be better inducers of their own metabolism, and their carbon bond structures may be more accessible to metabolic enzymes or susceptible to environmental degradation. It would be interesting to compare the rates of degradation of these three compounds under controlled biotic and abiotic conditions to test this idea. If we assume that this hypothesis is correct, then it might explain the results shown in Figure 7 in which the extract of weathered MC252 oil was of lower AHR activity than the extract of fresh MC252 oil. Rather than being a simultaneous loss of all AHR active compounds, there could be selective elimination of some of the more potent ones as oil weathers. In summary, Figure 8 indicates the changing composition of oil over periods of time and reflects the disappearance of methylchrysenes and the persistence of methyltriphenylenes. This could be a possible reason for a loss of AHRC activity in our bioassay.
We thank our oil spill research colleagues for consultation, assistance, and general support. This research was made possible by funding from the Gulf of Mexico Research Initiative to the Coastal Waters Consortium. The financial sources had no role in the design or execution of the study, data analysis, decision to publish, or manuscript preparation. These and other relevant data are publically available through the Gulf of Mexico Research Initiative Information & Data Cooperative (GRIIDC) at https://data.gulfresearchinitiative.org (doi R5.x280.000:0004, doi R5.x280.000:0009, doi R5.x280.000:0010, doi R5.x280.000:0011, doi R5.x280.000:0012, and doi R5.x280.000:0013).