Article pubs.acs.org/est
Chemometrics-Assisted Effect-Directed Analysis of Crude and Refined Oil Using Comprehensive Two-Dimensional Gas Chromatography−Time-of-Flight Mass Spectrometry Jagoš R. Radović,† Kevin V. Thomas,‡ Hadi Parastar,§ Sergi Díez,† Romà Tauler,† and Josep M. Bayona*,† †
Department of Environmental Chemistry, IDAEA-CSIC, Jordi Girona 18-26, Barcelona 08034, Spain NIVA, Norwegian Institute for Water Research, Gaustadaléen 21, Oslo 0349, Norway § Department of Chemistry, Sharif University of Technology, Tehran 11155-3516, Iran ‡
S Supporting Information *
ABSTRACT: An effect-directed analysis (EDA) of fresh and artificially weathered (evaporated, photooxidized) samples of North Sea crude oil and residual heavy fuel oil is presented. Aliphatic, aromatic, and polar oil fractions were tested for the presence of aryl hydrocarbon receptor (AhR) agonist and androgen receptor (AR) antagonist, demonstrating for the first time the AR antagonist effects in the aromatic and, to a lesser extent, polar fractions. An extension of the typical EDA strategy to include an N-way partial least-squares (N-PLS) model capable of relating the comprehensive two-dimensional gas chromatography−time-of-flight mass spectrometry (GC × GC−TOFMS) data set to the bioassay data obtained from normal-phase LC fractions is proposed. The predicted AhR binding effects in the fresh and artificially weathered aromatic oil fractions facilitated the identification of alkyl-substituted threeand four-ring aromatic systems in the active fractions through the weighting of their contributions to the observed effects. A N-PLS chemometric model is demonstrated as a potentially useful strategy for future EDA studies that can streamline the compound identification process and provide additional reduction of samples’ complexity. The AhR binding effects of the suspected compounds predicted by N-PLS and identified by GC × GC−TOFMS were confirmed using quantitative structure−activity relationship (QSAR) estimates. endocrine processes by increasing estrogen metabolism;9 however, certain PAHs and other oil components are known to act as androgen receptor (AR) antagonists, blocking the AR and, thus, interfering with endocrine processes, too.10 These specific effects are poorly understood, and the oil components that cause them are generally unknown, especially following weathering. In vitro techniques are available to assess these effects, including reporter gene assays, such as the CALUX assay for AhR agonists and the yeast androgen screen (YAS) for AR agonists and antagonists. They can be used in effectdirected analysis (EDA) in combination with physicochemical fractionation and chemical analysis to identify the compounds exerting toxic effects.11 Oil fractionation usually involves liquid chromatography, and the fractions obtained are normally characterized by gas chromatography coupled to mass spectroscopy (GC−MS).10 However, for the characterization of complex samples, such as oil, comprehensive two-dimensional gas chromatography (GC × GC) has been successfully applied for compound-class oil
1. INTRODUCTION Crude oil and petroleum-derived product spills due to tanker and offshore platform incidents, countless everyday ballast and operational spills, and the illicit discharge of bilge oil from ships remain major sources of marine pollution.1 When released into the environment, the spilled oil undergoes physical, chemical, and biological processes (weathering) and is rapidly transformed. Because of its altered composition and nature (e.g., more polar oxygenated products),2 bioavailability and toxicity of weathered oil can be modified relative to neat oil,3,4 however, its composition and effects require further investigation. The toxic effects of oil on marine organisms range from acute and sublethal nonspecific narcotic effects to specific chronic effects that can affect feeding, growth, and reproduction and can result in irreversible tissue damage.5 Much attention has been given to the effects of the polycyclic aromatic hydrocarbons (PAHs) found in oil and, in particular, to carcinogenic PAHs.6 Recent studies have also identified the importance of alkyl-substituted PAH homologues in the overall toxicity of complex mixtures of aromatic compounds.7,8 The metabolic activation of carcinogenic PAHs is thought to take place through the aryl hydrocarbon receptor (AhR)-mediated induction of the CYP family of P450 monooxygenase. It has also been proposed that CYP1A induction can interfere with © 2014 American Chemical Society
Received: Revised: Accepted: Published: 3074
October 31, 2013 January 15, 2014 February 11, 2014 February 11, 2014 dx.doi.org/10.1021/es404859m | Environ. Sci. Technol. 2014, 48, 3074−3083
Environmental Science & Technology
Article
Figure 1. Flowchart of the applied EDA strategy and data arrangement used for the N-PLS modeling.
fingerprinting,12 the elucidation of oil weathering processes,13 and the characterization of petroleum fractions.14 It enables high-resolution, nontarget compound detection, and its potential for EDA studies is recognized,8 but lacks practical validation. Different chemometric methods have been proposed for the modeling of GC × GC data obtained from the analysis of complex samples.15−17 A potentially powerful multivariate regression approach to analyzing GC × GC data is N-way partial least-squares (N-PLS),18 which is an extension of PLS to higher-order data sets. N-PLS has been applied to data from GC × GC for the quantitative analysis of aromatic and naphthenic content in naphtha,19 the identification of gasoline adulteration,20 the quantification of kerosene in gasoline,21 and the quantification of naphthalenes in jet fuel oil.22 Therefore, we recognized an interest to investigate its potential applicability in the characterization of oil. In this paper, coarse fractions of fresh and artificially weathered (evaporated and photooxidized) oils, obtained by open-column liquid chromatography, were tested for the presence of AhR agonists and AR antagonists. Subsequently, the AhR-active fine fractions (obtained by normal-phase semipreparative HPLC) were analyzed by GC × GC coupled to time-of-flight mass spectrometry (TOFMS). An N-PLS multivariate regression method was developed able to identify the main compounds contributing to the observed effects, which were then confirmed using quantitative structure− activity relationship (QSAR) estimates.
2. MATERIALS AND METHODS 2.1. Chemicals and Reagents. All analytical-grade chemicals (purity >95%) included in this study were purchased from Dr. Ehrenstorfer (Augsburg, Germany). All the solvents were SupraSolv and LiChrosolv grade from Merck (Darmstadt, Germany). Silica gel and aluminum oxide for column chromatography were also obtained from Merck. 2.2. Oils. Oil samples were obtained from local sources. North Sea crude (NSC) is a light, sweet (
