Diaromatic sulphur-containing ‘naphthenic’ acids in process waters

w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 Available online at www.sciencedirect.com ScienceDirect journal homepage: www.elsevier.com/locate/watres Diaromatic sulphur-containing ‘naphthenic’ acids in process waters Charles E. West a, Alan G. Scarlett a, Andrew Tonkin a, Devon O’Carroll-Fitzpatrick a, Jos Pureveen b, Erik Tegelaar b, Rafal Gieleciak c,d,1, Darcy Hager c, Karina Petersen e, Knut-Erik Tollefsen e, Steven J. Rowland a,* a Petroleum and Environmental Geochemistry Group, Biogeochemistry Research Centre, University of Plymouth, Drake Circus, Plymouth PL4 8AA, UK b Shell Global Solutions International B V, Rock and Fluid Science, Kessler Park 1, 2288 GS Rijswijk, The Netherlands c Canmet ENERGY, Natural Resources Canada, Devon, Alberta, Canada T9G 1A8 d Institute of Chemistry, The University of Silesia, 9 Szkolna Street, 40-006 Katowice, Poland e Norwegian Institute for Water Research (NIVA), Gaustadalléen 21, N-0349 OSLO, Norway article info abstract Article history: Polar organic compounds found in industrial process waters, particularly those originating Received 2 July 2013 from biodegraded petroleum residues, include ‘naphthenic acids’ (NA). Some NA have been Received in revised form shown to have acute toxicity to fish and also to produce sub-lethal effects. Whilst some of 22 October 2013 these toxic effects are produced by identifiable carboxylic acids, acids such as sulphur- Accepted 24 October 2013 containing acids, which have been detected, but not yet identified, may produce others. Available online 1 November 2013 Therefore, in the present study, the sulphur-containing acids in oil sands process water were studied. Keywords: A fraction (ca 12% by weight of the total NA containing ca 1.5% weight sulphur) was Naphthenic acids obtained by elution of methylated NA through an argentation solid phase extraction col- GCxGC-MS umn with diethyl ether. This was examined by multidimensional comprehensive gas GCxGC-SCD chromatography-mass spectrometry (GCxGC-MS) in both nominal and high resolution Accurate mass mass accuracy modes and by GCxGC-sulphur chemiluminescence detection (GCxGC-SCD). SO2 acids Interpretation of the mass spectra and retention behaviour of methyl esters of several synthesised sulphur acids and the unknowns allowed delimitation of the structures, but not complete identification. Diaromatic sulphur-containing alkanoic acids were suggested. Computer modelling of the toxicities of some of the possible acids suggested they would have similar toxicities to one another and to dehydroabietic acid. However, the sulphurrich fraction was not toxic or estrogenic to trout hepatocytes, suggesting the concentrations of sulphur acids in this sample were too low to produce any such effects in vitro. Further samples should probably be examined for these compounds. ª 2013 Elsevier Ltd. All rights reserved. * Corresponding author. Tel.: þ44 1752 584557. E-mail addresses: [email protected], [email protected] (S.J. Rowland). 1 On leave from Institute of Chemistry. 0043-1354/$ e see front matter ª 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.watres.2013.10.058 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 1. Introduction Naphthenic acids (NA) are reported to be amongst the toxic polar constituents of produced water from various petroleum production processes, including from conventional and from less conventional petroleum reserves, such as oil sands. Deposits of oil sands have been found as far apart globally as China, Venezuela and Canada and are a major source of fossil fuels (e.g. Bycott et al., 1999; Gosselin et al., 2010). The oil sands of Alberta, Canada exceed most conventional oil reserves in volume. These remained uneconomic for many years due to the costs of removing the oil from the sand, but are now produced from both surficial and sub-surface deposits (Gosselin et al., 2010). Surficial deposits are processed by the Clark process in which treatment with an aqueous solution of hot alkali removes sand, fines and unwanted organic acidic material from the bitumen. A by-product of the process, after much recycling of the alkaline water, is a large volume of process-affected water (OSPW), containing sediment and organic compounds, which is currently stored in lagoons (Gosselin et al., 2010). There are concerns about possible leakage of OSPW into the surrounding environment (Kean, 2009; Schindler, 2010; Jordaan, 2012). The water is alkaline, saline and typically contains up to about 100 mg L1 of a complex mixture of organic compounds including the thousands of carboxylic acids known as NA. The latter term has become used because the infrared spectrum of the acidic extract resembles that of NA refined from petroleum (MacKinnon and Boerger, 1986). However, increasingly detailed chemical and biological analyses of the OSPW NA mixtures have indicated that there are many other compounds in OSPW than in commercial refined NA and that some of the NA differ in structure (Rowland et al., 2011a), proportions (Grewer et al., 2010), biodegradability and toxicity (Scott et al., 2005) from the NA in commercial refined samples. This may be due to the refining and production methods used to obtain the latter. It is possible that the NA in oil sands are more similar to those in biodegraded unrefined petroleum (e.g. Watson et al., 2002), but few detailed analyses of the latter have been made to date. Several toxicological effects have been attributed to NA (reviewed by Scarlett et al., 2012). For instance, when an acid extract of a whole OSPW was fractionated by distillation of the NA fraction esterifiable with diazomethane (Frank et al., 2008) the acute toxicity to bacteria in a screening assay was EC50 w 40e60 mg L1. Furthermore, when OSPW NA were fractionated by argentation solid phase extraction (Agþ SPE), a fraction eluting with 5% diethyl ether: 95% hexane and containing aromatic acids, was at least as toxic as the alicyclic NA (Scarlett et al., 2012). It was also apparent from the latter fish assay that some of the toxicity of the esterifiable NA was not accounted for by the alicyclic and aromatic acids alone (Scarlett et al., 2012). Fractions of the NA from Agþ SPE, including diethyl ether and methanol eluates, have not yet been fully characterised or assayed for toxicity. The diethyl ether fraction, which is the subject of the present report, contained aromatic compounds, as shown by C/H ratios and UV spectrophotometry, but in addition, about 50% of the total sulphur associated with the NA was present in this fraction (Jones et al., 2013). 207 Multidimensional comprehensive gas chromatography (GCxGC) with sulphur chemiluminescence detection (SCD) established that several major GC-resolvable sulphur compounds were present (Jones et al., 2013). Numerous analyses of other OSPW acid extracts by electrospray ionisation Fourier Transform Ion Cyclotron Resonance Mass Spectrometry (FT-ICR-MS) or Orbitrap MS have also indicated that a variety of sulphur species are often present (e.g. Barrow et al., 2010; Bataineh et al., 2006; Headley et al., 2011a,b). More attention has been drawn to these sulphur compounds as more has begun to be understood about the NA in OSPW, and indeed in petroleum generally (e.g. Headley et al., 2011a,b; Panda et al., 2009). In the present study a sulphur-containing diethyl ether eluate Agþ SPE fraction of a methylated (esterifiable) OSPW extract was examined by GCxGC-MS with both nominal and high resolution (HR) mass accuracy modes and by GCxGCSCD/flame ionisation detection (FID). Several acids were synthesised for comparison. The toxicity of the fraction was determined as cytotoxicity and estrogenicity in a rainbow trout in vitro (hepatocyte) assay and acute toxicity (lethality) for Fathead minnow (Pimephales promelas) and the water flea (Daphnia magna) predicted. 2. Materials and methods Authentic acids were purchased from Sigma (Poole, U.K.) or synthesised. Syntheses were based on FriedeleCrafts acylation or alkylation of either dibenzothiophene (Sigma, Poole, UK) or naphtho[2,1-b]thiophene synthesised previously (Kropp et al., 1997), with methylsuccinic anhydride or b-butyrolactone (Sigma, Poole, U.K.) in the presence of aluminium trichloride (Fig. 1). The methods were essentially those of Smith et al. (2008). The resulting keto acids (from acylation) were reduced to the acids by Huang-Minlon modification of the WolffeKishner reaction (cf Smith et al., 2008). The fractionated OSPW extract was obtained from Syncrude West In-Pit as described previously (Reinardy et al., 2013; Scarlett et al., 2012). Smaller amounts of an acidic extract from a different oil sands company (wishing to remain anonymous) was also obtained at a different time and location in order to briefly test the generality of occurrence of the sulphur compounds in OSPW-derived NA (Rowland et al., 2012). Both authentic acids and OSPW NA extracts or fractions were converted to the methyl or trideuteriomethyl esters as stated previously (West et al., 2013). Argentation SPE was conducted essentially as previously (Jones et al., 2012). Multidimensional comprehensive gas chromatographymass spectrometry (GCxGC-MS) and GCxGC-SCD/FID analyses were conducted on four different instruments; two allowed GCxGC-MS with nominal mass resolution, one with higher mass resolution. One of the nominal mass instruments used an identical configuration of GC columns to that installed in a GCxGC-SCD/FID instrument to allow ease of comparison between the SCD/FID responses and the mass spectrometer. Details are provided in the online supplementary information (Table 1S). 208 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 Modelling of toxic effects was conducted using ADMET Predictor software, as described previously (Scarlett et al., 2013). 2.1. Toxicity assay with primary hepatocytes Assays of cytotoxicity and estrogenicity were performed essentially as described previously (Tollefsen et al., 2003, 2008, 2012). Primary hepatocytes were obtained by a two-step perfusion procedure of juvenile rainbow trout (size 200e500 g) livers. Cells (90% viability) were diluted to 500,000 cells mL1 in serum-free L-15 medium supplemented with 0.29 mg mL1 L-glutamin, 4.5 mM NaHCO3, 100 Units mL1 penicillin, 100 mg L1 streptomycin and 0.25 mg mL1 amphotericin (Cambrex, East Rutherford, NJ, USA), and 200 uL cell suspension was plated in 96-well primaria cell culture plates (Falcon, Becton Dickinson Labware, Oxnard, CA, USA). After 24 h of acclimatization, cells were exposed for 96 h to fractions, blanks, reference compounds and positive and negative controls. At the end of the exposure, 100 mL growth medium was transferred from each well to a Maxisorp nunc-immunoplate (Nunc, Roskilde, Denmark), sealed with sealing tape and stored at 80  C for subsequent vitellogenin (Vtg) protein analysis. Cytotoxicity was determined directly in the cell cultures at the end of the exposure period with cellular viability probes according to Tollefsen et al. (2008). In essence, metabolic activity and membrane integrity were determined by the probes Alamar blue (AB) and 5-carboxyfluorescein diacetate acetoxymethyl ester (CFDA-AM), respectively. The cells were incubated with Tris buffer (50 mM, Ph 7.5) containing 5% AB and 4 mM CFDA-AM on an orbital shaker (100 rpm, 30 min) and fluorometric readings were performed with a Victor V3 multilabel counter (Perkin Elmer, Waltham, MA, USA) using excitation and emission wavelength pairs of 530e590 nm (AB) and 485e530 nm (CFDA-AM). Determination of the relative expression of Vtg, a biomarker for estrogenicity, was performed by a capture enzyme linked immunosorbent assay (ELISA) as described in Tollefsen et al. (2008). Plates containing cell culture media were thawed and incubated overnight (4  C) to allow binding of proteins to the well surface. Vitellogenin protein detection was achieved by a capture ELISA with monoclonal mouse anti-salmon Vtg (BN-5, Biosense Laboratories, Bergen, Norway) and goat anti-mouse IgG (Bio-Rad, Hercules, CA, USA) as primary and secondary antibody respectively, both diluted 1:6000. A solution with 3,30 ,5,50 e tetramethylbenzidine (TMB) plus (KEMENTEC diagnostics, Taastrup, Denmark) was used as enzyme substrate and the reaction stopped by addition of H2SO4 (1 M). The absorbance was measured at 450 nm using a Thermomax microplate reader (Molecular Devices, USA). All data were normalised to a positive control (30 nM 17bestradiol for Vtg production and 10 mM CuSO4 for cytotoxicity) and a negative (solvent) control (water or DMSO). The cytotoxicity and Vtg production was characterised by non-linear regression using a sigmoidal concentration-response (variable slope) model in GraphPad Prism v. 4.03 (GraphPad Software Inc., La Jolla, CA, USA). 3. Results and discussion The mixtures of acidic and non-acidic organic compounds in OSPW are extremely complex, and like those of petroleum generally (Panda et al., 2009), they can even be termed 4 1' 1' 3 2' 2' (II) (I) AlCl3 or (IV, VI) R= (III, V) R= Fig. 1 e Structures of purchased and some of the synthetic acids. w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 ‘supercomplex’. For this reason, for decades, most NA in petroleum and oil sands have remained unidentified. The application of GCxGC-MS has allowed some individual NA in OSPW and petroleum to be identified (e.g. Rowland et al., 2011a, b) but even so, several of these have only been identified tentatively and await confirmation or rebuttal by synthesis. The task is daunting, but possibly the best chromatographic separations to date have been achieved using a pre-separation of methyl esters of the acids by Agþ SPE, followed by examination by GCxGC-MS in both nominal and high mass resolution modes employing (for aromatic acids) a primary GC column coated with an ionic liquid stationary phase (e.g. West et al., 2013; Reinardy et al., 2013). Previously, examination of the ether eluate from Agþ SPE of a methylated fraction of an OSPW extract by GCxGC-SCD established the presence of several major resolved peaks, but numerous minor peaks in a complex, partially resolved, mixture were also present (Jones et al., 2013). In the present study this relevant retention area of the GCxGC-SCD chromatogram (Fig. 1S) was the focus for examination of samples by GCxGC-MS (Fig. 1S) and GCxGC-HRMS, using similar chromatographic conditions for both MS and SCD methods. We noted the presence of numerous components for which the GCxGC-SCD response indicated the presence of sulphur (Fig. 1S) and for which accurate mass molecular ions indicated the presence of one sulphur and two oxygen atoms. The presence of the 34S isotopomeric molecular ion was also apparent in each case due to the improved spectral quality following SPE fractionation. The accurate masses of the fragment ions (Table 1) indicated the loss of radicals containing the two oxygens with retention of the sulphur in the cations. This, along with the derivatisation behaviour, including formation of both methyl and trideuteriomethyl esters, indicated that the sulphur compounds were carboxylic acids and not sulfoxides or sulfonyl species. This is also consistent with the interpretations from IR spectroscopy (Jones et al., 2013) and is in itself an important advance: previously it was not known whether the sulphur compounds were sulfoxides, or thiols with hydroxy or with keto substituents, for example (e.g. Barrow et al., 2010; Headley et al., 2011a,b; Bataineh et al., 2006). At least three series of sulphur compounds were recognised. We focus here on those with ten double bond Table 1 e Accurate masses of ions observed from unknowns (i. e. A, B, D; Fig. 2) in a sulphur-rich fraction of oil sands process-affected water naphthenic acids (methyl esters) with ten double bond equivalents (DBE). Data obtained with instrumentation configuration 3 (Table 1S). Measured m/z Inferred composition C17H16O2S 284.0884 (Mþ.) 197.0432 (Bþ) C13H9S C18H18O2S 298.1032 (Mþ.) 211.0584 (Bþ) C14H11S C19H20O2S 312.1180 (Mþ.) 225.0742 (Bþ) C15H13S Theoretical m/z Accuracy (ppm) 284.0871 4.6 197.0425 298.1028 3.6 1.5 211.0582 312.1184 1.2 1.3 225.0738 1.8 209 equivalents, but the complexity of the mixture showed that many more minor compounds including isomers of each series, were present (Fig. 1S). 3.1. Sulphur-containing acids with ten double bond equivalents The molecular ions in the mass spectra of the first series of compounds (Fig. 3) suggested they each had ten double bond equivalents (DBE). Although these could also reasonably be attributed to polycyclic compounds, the Agþ SPE retention behaviour, C/H ratios and UV spectra (Jones et al., 2013) of components in this fraction, suggested the compounds were instead aromatic, rather than alicyclic. Indeed, a mixture of synthesised and commercial dibenzothiophene and naphthothiophene alkanoic acids (Fig. 1; methyl esters) eluted with diethyl ether in this Agþ SPE fraction and the preceding ether:hexane fraction, whereas alicyclic acids eluted in fractions eluting with hexane or earlier eluting ether:hexane fractions (Jones et al., 2012, 2013). The mass spectrum of the methyl ester of the lowest molecular weight unknown compound (Fig. 3A; Table 1) was characterised by a molecular ion (m/z 284.088; C17H16O2S) and a major fragment ion (m/z 197.043) consistent with loss of C4H7O2. The number of DBE in the m/z 197 base peak ion fragment (9 DBE) suggested an aromatic moiety. Possibilities for the unknown included dibenzothiophenes, naphthothiophenes or dibenzothiopyrans with acid side chains, amongst others. Since dibenzothiophenes in petroleum are well known (reviewed by Kropp and Fedorak, 1998), we purchased 4(dibenzo[b,d]thiophen-20 -yl)butanoic acid (Fig. 1: I), converted the acid to the methyl ester and examined the mass spectrum (Fig. 4A). This contained the expected molecular ion (m/z 284) and S isotope ion (m/z 286) plus a dominant ion m/z 197, as in the spectrum of the unknown (Fig. 3A). However, also present in the spectrum of (I) was a major radical ion m/z 210, indicative of loss of a 74 Da moiety, attributed to the well known McLafferty rearrangement (McLafferty and Turecek, 1993). This rearrangement is possible because (I) contains a g-H at C4 of the butanoate chain; a g-H adjacent to a double bond (e.g. carbonyl group) is a structural requirement for the McLafferty rearrangement to occur. The m/z 210 ion was not present in the mass spectrum of the unknown (Fig. 3A) which suggests that the latter did not contain a g-H. The methyl ester of the authentic acid (I) eluted later than the unknown on the first dimension GC column, suggesting more branching in the unknown, but in about the same position on the second dimension GC column, suggesting a similar structure in the nucleus. Structures that satisfy the chromatographic and spectral features of the unknown include methyl esters of a dibenzothiophene (DBT), a naphthothiophene (NT) or a dibenzopyran (DBP) acid substituted with a methyl branched side chain with no g-H; others may also be possible. Formation of the base peak ion m/ z 197 (Fig. 3A) could then be attributed to the facile benzylic cleavage between C2 and C3 of a C4 chain with a methyl branch at C2. Such benzylic cleavages are common in the methyl esters of thiophenyl carboxylic acids more generally (Charrié-Duhaut et al., 2000). Numerous isomers of such acids are possible, but only one major chromatographic peak with 210 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 Fig. 2 e Extracted ion chromatograms (m/z 284 D 197D298 D 211D312 D 225) from GCxGC-MS of methyl esters of a sulphurrich OSPW fraction isolated by argentation solid phase extraction, illustrating the distributions of sulphur-containing alkanoic acids (methyl esters, A-E) with ten double bond equivalents. Capital letters refer to unknowns, mass spectra of which are shown in Fig. 3AeE. GCxGC-MS conditions were as per configuration 2 (Table 1S). these spectral features was observed in the GCxGC-MS chromatograms (Fig. 2; Fig. 1S). The presence of a methyl branch in an alkanoate side chain would be consistent with the impaired further microbial degradation, as has been observed with other aromatic acids (Smith et al., 2008). In further support of these arguments, synthesis (albeit in low, <6% yield) of a mixture of isomers of 3-dibenzothiophen20 (and 30 ?)-yl-butanoic acids (e.g. II; Fig. 1) which had a methyl branch at the C3 position, by FriedeleCrafts alkylation of dibenzothiophene with b-butyrolactone, produced compounds which, once converted to the methyl esters exhibited mass spectral ions at m/z 284 (50%, Mþ.) and m/z 211 (Bþ, 100%). The base peak ion m/z 211 is indicative of loss of a 73 Da moiety, attributed to fragmentation between C2 and C3 of the branched C4 chain (Fig. 1; II). In order for the spectrum of a dibenzothiopyran alkanoic acid to satisfy the spectral features of the unknown and assuming benzylic cleavage is the dominant fragmentation (Charrié-Duhaut et al., 2000), substitution of a C3 branched alkanoate chain in the thiopyran ring would be required. This cannot be ruled out as a possibility yet, since the only authentic thiopyran carboxylic acid available to us was 3methyl-6H-dibenzo[b,d]thiopyran-2-carboxylic acid. The mass spectrum of the methyl ester was dominated by a molecular ion, unlike the unknowns, but it is not known whether this is typical. In summary, the unknown may be a dibenzothiophene, naphthothiophene or dibenzothiopyran with a methyl branched alkanoic acid side chain. Clearly though, it is a C16 Scontaining aromatic carboxylic acid. The mass spectra (Fig. 3B, C: Table 1) of the next eluting unknowns (two isomers) with ten DBE in this series were characterised by a molecular ion and major fragment ion again consistent with loss of C4H7O2. The number of DBE in the m/z 211 base peak ion fragment suggested an aromatic moiety, again consistent with Agþ SPE elution, UV spectrophotometry and C/H ratio (Jones et al., 2013). Possibilities for the unknowns therefore include C17 acids with a structure similar to the previous unknown but with a further methyl group retained in the formation of the base peak ion m/z 211, presumably due to methyl group substitution on the nucleus. In order to investigate this further, some isomeric C17 DBT and NT with branched alkanoate side chains were synthesised and the mass spectra of the methyl esters examined. First, dibenzothiophene was reacted with methylsuccinic anhydride in the presence of aluminium trichloride (viz: FriedeleCrafts acylation; Fig. 1). The products were assigned to the expected keto acids by infrared spectroscopy (two absorptions assigned to carbonyl stretches in acids (1709 cm1) and ketones conjugated with an aromatic ring (1677 cm1)) and GCeMS of ester derivatives. Esterification of the products with either BF3-methanol or bis-silyltrifluoroacetamide yielded a mixture, which produced two major peaks when examined by GCeMS. These were assigned to two isomeric keto acids with methyl butanoate side chains, the methyl substituent either alpha or beta to the carboxyl carbon. The position of the alkanoate substituent on the aromatic ring is unknown. WolffeKishner reduction of the keto acids produced the desired acids, again as two major isomeric acids (III,IV). The products were assigned to the expected acids by infrared spectroscopy (one absorption, assigned to carbonyl stretch in acids (1702 cm1)) and GCeMS of ester derivatives. Esterification of the products with either BF3-methanol or bissilyltrifluoroacetamide yielded a mixture, which produced two major peaks when examined by GCeMS. These were assigned to two isomeric acids with methyl butanoate side chains, the methyl substituent either alpha (IV) or beta (III) to the carboxyl carbon. The mass spectra (Fig. 4B, C) of the methyl esters of these contained the expected molecular ions (m/z 298) and S isotope ions (m/z 300) plus a dominant ion m/z 197, as in the spectrum of the C16 unknown (cf Fig. 3A). Also present in the spectra were major radical ions m/z 224 (III, first eluting) and 210 (IV, second eluting), indicative of losses of a 74 Da moiety or a 88 Da moiety respectively, attributed to the well known McLafferty rearrangements and allowing assignment of the first-eluting as the isomer with the methyl substituent beta to the carboxyl carbon (III) and the second as the alpha isomer (IV). These spectra were dissimilar to that in the 211 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 (A) 120 197 Relative Intensity (%) 100 80 60 40 284 20 179 99 115 152 165 221 253 0 50 100 150 200 250 300 m/z (B) (C) 120 120 211 80 60 40 298 20 105 165 100 267 150 60 298 40 241 0 50 80 20 199 59 211 100 Relative Intensity (%) Relative Intensity (%) 100 200 250 0 300 350 59 50 105 100 165 178 198 150 200 m/z (D) 250 300 350 (E) 120 225 225 100 Relative Intensity (%) 100 Relative Intensity (%) 267 m/z 120 80 60 312 40 20 0 240 80 60 312 40 20 59 50 165 113 192 212 237 253 281 59 165 113 192 209 237 253 281 0 100 150 200 m/z 250 300 350 50 100 150 200 250 300 350 m/z Fig. 3 e Nominal mass electron ionisation spectra of methyl esters of unknowns A-E (cf Fig. 2) in a sulphur-rich OSPW fraction isolated by argentation solid phase extraction, illustrating the distributions of a series of sulphur-containing aromatic alkanoic acids with ten double bond equivalents. GCxGC-MS conditions were as per configuration 2 (Table 1S). 212 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 (A) 120 197 210 Relative Intensity (%) 100 80 284 60 40 99 20 152 165 85 139 253 184 0 50 100 150 200 250 300 m/z (B) (C) 120 120 197 80 60 224 40 298 20 0 80 197 60 298 40 20 59 50 98 100 152 165 184 208 267 0 150 200 250 300 350 59 50 88 100 152 165 184 150 238 200 m/z m/z (D) (E) 120 267 250 300 350 120 197 80 60 224 40 210 100 Relative Intensity (%) 100 Relative Intensity (%) 2 210 100 Relative Intensity (%) Relative Intensity (%) 100 298 20 80 197 19 60 298 40 20 152 165 59 74 0 50 184 267 208 59 0 100 150 200 250 300 350 m/z 50 88 100 152 165 184 150 237 200 250 267 300 350 m/z Fig. 4 e Nominal mass electron ionisation spectra of methyl esters of purchased and synthesised reference compounds. GCxGC-MS conditions were as per configuration 3 (Table 1S). spectrum of the C17 unknowns, which contained a major ion m/z 211 (Fig. 3B, C). Next, naphtho[2,1-b]thiophene was reacted with methylsuccinic anhydride in the presence of aluminium trichloride (viz: FriedeleCrafts acylation; Fig. 1). Keto-acids were reduced to acids by WolffeKishner reaction (vide infra). The keto acids and acids were characterised in the same way as the DBT keto acids (above) and the mass spectra of the methyl esters of the 213 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 acids, (the first-eluting as the isomer with the methyl substituent beta to the carboxyl carbon (V) and the second as the alpha isomer (VI)) were indistinguishable from those of the corresponding DBT acids above and thus again dissimilar to the spectra of the C17 unknowns. The NT acids eluted slightly later than the corresponding DBT acids and thus later still than the corresponding unknowns. Clearly the spectra of these synthetic products do not allow definitive assignment of the C17 unknowns (i.e. those producing the spectra shown in Fig. 3B, C), although the unknowns and synthetic acids eluted in the same Agþ SPE fractions. In summary, the unknowns are most likely to be methyl dibenzothiophene or methyl naphthothiophenes with methylpropanoate acid side chains or methyldibenzothiopyrans with a methylethanoic acid side chain (C17 acids). Next in this series, a pair of isomeric unknowns was observed (Fig. 2), with the corresponding mass spectral features expected of dimethyl analogues of the previous unknowns with acid side chains (Fig. 3D, E). Again the spectral features (Fig. 3D, E: Table 1) were consistent with the presence of sulphur and formation of the base peak ion by benzylic cleavage. These unknowns are most likely to be dimethyl dibenzothiophenes or dimethyl naphthothiophenes with C2methyl branched propanoate acid side chains (or dimethyldibenzothiopyrans with shorter alkanoic acid side chains): C18 acids. In summary, firmer identification of the unknowns will require synthesis and characterisation of yet more acids. Several such acids were synthesised over 40 years ago for testing as antimalarials (e.g. Das et al., 1973), but unfortunately the mass spectra were not published. The current study has added to the spectral database of such acids (e.g. Fig. 4), but more syntheses and spectral characterisation are needed, especially of alkyl-substituted DBT, NT and DBPs with branched alkanoic acid side chains. 3.2. Environmental relevance It is likely that the occurrence of the S-containing acids in the NA is a result of biodegradation of the corresponding sulphurcontaining hydrocarbons (reviewed by Kropp and Fedorak, 1998). For example, from transformation of alkyl polyaromatic thiophenes, many of which are known to occur in petroleum and in oil sands bitumen (e.g. Strausz et al., 2011). The biotransformation of methyl and 2,8-, 3,4-, and 4,6dimethyl DBT and 1-methylnaphtho[2,1-b]thiophene has been studied (Kropp et al., 1997; Saftic et al., 1993) due to their occurrence in fossil fuels and the biotransformations of DBT via the Kodama pathway have been known for decades (Kodama et al., 1973; Bressler and Fedorak, 2001). All the cultures tested previously were able to degrade the unsubstituted ring of 3, 4-dimethyl-DBT to give 6,7-dimethyl-hydroxy-2formylbenzothiophene and 6,7-dimethylbenzothiophene2,3-dione, among other products. No carboxylic acids were reported in these earlier studies, though presumably formation of hydroxy acids from the hydroxy-2formylbenzothiophenes would be facile. Indeed, the pathway proposed for biotransformation of naphthothiophenes including a methyl isomer, involved bis-hydroxylation of a benzenoid ring, followed by ring opening to form hydroxy benzothiophene carboxylic acids and methyl hydroxy benzothiophene carboxylic acids, which were observed previously (Kropp et al., 1997). Tentative identification of diaromatic sulphur-containing alkanoic acids in the NA herein prompted us to consider the contribution that such compounds might make to the proportion of toxicity unaccounted for by alicyclic and aromatic non-sulphur containing NA (Scarlett et al., 2012). First a commercial computer model, ADMET Predictor (cf Scarlett et al., 2013) was used to predict the toxicities of selected acids to a number of endpoints (Table 2). Predicted toxicities (LC50) of some members of the families of the candidate unknowns including dibenzothiophene, naphtho [3,2b]thiophene, naphtho[2,1b]thiophene, naphthothiolane, napthenobenzothiophene and dibenzopyran acids, for the Fathead minnow P. promelas and the water flea, D. magna were in the approximate range 0.5e7.5 mg L1 (2e28 mM; Table 2), which are fairly comparable to those of aromatic NA (w5e8 mg L1; Scarlett et al., 2012) and known components of NA such as dehydroabietic acid (DHAA; LC50 w 1 mg L1 (w3.3 mM) to Zebrafish larvae and Rainbow trout (Scarlett et al., 2012). Neither positional isomerism, nor structural isomerism influenced the predicted values greatly (Table 2). Thus, it appears that whilst such acids (once confirmed) might indeed contribute to the acute toxicity of NA, if sufficiently abundant, they are not especially toxic and the total ether eluate Agþ SPE (i.e. sulphur-rich) fraction was not especially abundant (w12e20% of material recovered as a proportion of the summed weights of the hexane through ether SPE eluates; Jones et al., 2012, 2013). Well-described assays of cytotoxicity and estrogenicity based on cultures of Rainbow trout hepatocytes (Tollefsen et al., 2003, 2008, 2012) were also used herein to assay a few purchased aromatic S-containing acids and DHAA (Scarlett Table 2 e Toxicities of selected thiophenic alkanoic acids predicted to Fathead minnow, P. promelas and water flea, D. magna predicted by ADMET Predictor modelling package. Values shown are estimated lethal concentrations for a 50% response (LC50) in mg LL1. Examples chosen to illustrate the toxicities of different structural (e.g. naphtho[3,2 b] thiophenes and naphtho [2,1 b]thiophenes) and positional isomers. Acid Dibenzo[b,d]thiophen-20 -yl-2-methylpropanoic Dibenzo[b,d]thiophen-40 -yl-2-methylpropanoic Dibenzo[b,d]thiophen-10 -yl-2-methylpropanoic 2-methyl-3-(naphtho[3,2b]thiophen-30 -yl) propanoic 2-methyl-3-(naphtho[3,2b]thiophen-20 -yl) propanoic 2-methyl-3-(naphtho[2,1b]thiophen-30 -yl) propanoic 2-methyl-3-(naphtho[2,1b]thiophen-20 -yl) propanoic 2,3-dihydronaphtho[4,3,b]thiophen-20 -yl)-2methylpropanoic 2-methyl-3-(6,7,8,9-tetrahydronaphtho[4,3,b] thiophen-20 -yl)propanoic P. D. promelas magna 1.5 2.1 3.0 1.4 3.1 5.3 7.6 4.6 0.7 4.0 3.3 3.6 1.3 2.2 0.6 1.0 1.0 2.8 214 w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 et al., 2012; Jones et al., 2012) e the latter assayed as a positive control, since literature values for the toxicity of DHAA are available (vide supra). The EC50 determined for DHAA was w3 mg L1 (w10 mM), fairly close to values from other assays and validating the assay (C; Fig. 2S). Under these conditions, four adventitious purchased aromatic S-containing acids had EC50 values >60 mg L1 (>250 mM; Fig. 3S). No estrogenicity was observed for these acids. Since no accurate method of quantifying the sulphurcontaining acids in the OSPW is available until they are better identified and suitable internal standards can be made, similar toxicity assays were also conducted on the whole sulphur-rich ether eluate. Under these conditions, no EC50 values were reached when up to 10 mg L1 of the ether fraction or of samples from comparable volumes of SPE column eluants were examined (viz: procedural blank fractions; Fig. 2S). Thus, in summary, it seems that the sulphur-rich fraction is not responsible for a major contribution (if any) to the acute toxicity of at least this NA sample when tested in vitro. Neither was any increase in vitellogenin production (viz: estrogenicity) produced by the ether fraction or by column blanks. This is consistent with the non-estrogenicity of the four purchased acids tested herein and with previous findings that non-hetero aromatic acids appear to be sufficient to account for the weak estrogenicity of this OSPW NA. Whether the sulphur-containing acids are responsible for other sub-lethal effects and whether the results differ for other OSPW, will require further study. Future studies will concentrate on the remaining Agþ SPE (methanol eluate) NA fraction and on a wider range of process waters. 4. Conclusions A series of sulphur-containing aromatic carboxylic acids was partially identified in oil sands process water and compared with a number of synthesised analogues. The unknowns are most likely dibenzothiophenes or naphthothiophenes with C2 branched methylpropanoate acid side chains (or possibly dibenzothiopyrans with a C3 acid side chain). The toxicities of a range of similar acids calculated by a computer model for Fathead minnow and water flea were LC50 w 0.5e7.5 mg L1 (2e28 mM). However, a sulphur-rich acids fraction of the water determined by in vitro toxicity assay, showed no appreciable activity. Further process waters should be assayed before general conclusions can be drawn, but these results may give a first indication of the toxicity of such fractions of OSPW. The S-rich water fraction showed no measurable estrogenicity, as reflected in no induction of vitellogenin. Acknowledgements Funding of this research was provided by an Advanced Investigators Grant (no. 228149) awarded to SJR for project OUTREACH, by the European Research Council, to whom we are also extremely grateful. We thank Drs R.A. Frank and L.M. Hewitt of Environment Canada for supplying a bulk OSPW acid extract. Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2013.10.058. references Barrow, M.P., Witt, M., Headley, J.V., Peru, K.M., 2010. Athabasca oil sands process water: characterization by atmospheric pressure photoionization and electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Anal. Chem. 82, 3727e3735. Bataineh, M., Scott, A.C., Fedorak, P.M., Martin, J.W., 2006. Capillary HPLC/QTOF-MS for characterizing complex naphthenic acid mixtures and their microbial transformation. Anal. Chem. 78, 8354e8361. Bressler, D.C., Fedorak, P.M., 2001. Purification, stability, and mineralization of 3-hydroxy-2-formylbenzothiophene, a metabolite of dibenzothiophene. Appl. Environ. Microbiol. 67, 821e826. Bycott, P.W., Taylor, J.M.G., Shi, D., Nicolai, R., Reh, L., Jiayu, N., Jianyi, H., 1999. Formation and distribution of heavy oil and tar sands in China. Mar. Pet. Geol. 16, 85e95. Charrié-Duhaut, A., Lemoine, S., Adam, P., Connan, J., Albrecht, P., 2000. Abiotic oxidation of petroleum bitumens under natural conditions. Org. Geochem. 31, 977e1003. Das, B.P., Cunningham, R.T., Boykin, D.W., 1973. Naphthothiophenes.3. Preparation of 4-naphtho[1,2-b] thiophenemethanols and 5-naphtho[1,2-b] thiophenemethanols and attempts to prepare 5-naphtho[2,1b]thiophenemethanols as antmalarials. J. Med. Chem. 16, 1361e1365. Frank, R.A., Kavanagh, R., Burnison, B.K., Arsenault, G., Headley, J.V., Peru, K.M., Van Der Kraak, G., Solomon, K.R., 2008. Toxicity assessment of collected fractions from an extracted naphthenic acid mixture. Chemosphere 72, 1309e1314. Gosselin, P., Hrudey, S.E., Naeth, M.A., Plourde, A., Therrien, R., Van Der Kraak, G., Xu, Z., 2010. Environmental and Health Impacts of Canada’s Oil Sands Industry. Royal Society of Canada, Ottawa, Ontario, Canada. Available: http://www.rsc. ca/documents/RSCreportcompletesecured9Mb_Mar28_11.pdf (accessed 05.07.11.). Grewer, D.M., Young, R.F., Whittal, R.M., Fedorak, P.M., 2010. Naphthenic acids and other acid-extractables in water samples from Alberta: what is being measured? Sci. Total Environ. 408, 5997e6010. Headley, J.V., Barrow, M.P., Peru, K.M., Fahlman, B., Frank, R.A., Bickerton, G., McMaster, M.E., Parrott, J., Hewitt, L.M., 2011a. Preliminary fingerprinting of Athabasca oil sands polar organics in environmental samples using electrospray ionization Fourier transform ion cyclotron resonance mass spectrometry. Rapid Commun. Mass Spectrom. 25, 1899e1909. Headley, J.V., Peru, K.M., Janfada, A., Fahlman, B., Gu, C., Hassan, S., 2011b. Characterization of oil sands acids in plant tissue using Orbitrap ultra-high resolution mass spectrometry with electrospray ionization. Rapid Commun. Mass Spectrom. 25, 459e462. Jones, D., West, C.E., Scarlett, A.G., Frank, R.A., Rowland, S.J., 2012. Isolation and estimation of the ‘aromatic’ naphthenic acid w a t e r r e s e a r c h 5 1 ( 2 0 1 4 ) 2 0 6 e2 1 5 content of an oil sands process-affected water extract. J. Chromatogr. A 1247, 171e175. Jones, D., Scarlett, A.G., West, C.E., Frank, R.A., Gieleciak, R., Hager, D., Rowland, S.J., 2013. Elemental and spectroscopic characterisation of sub-fractions of an acidic extract of oil sands process water. Chemosphere. 93, 1655e1664. Jordaan, S.M., 2012. Land and water impacts of oil sands production in Alberta. Environ. Sci. Technol. 46, 3611e3617. Kean, S., 2009. Eco-alchemy in Alberta. Science 326, 1052e1055. Kodama, K., Umehara, K., Shimizu, K., Nakatani, S., Minoda, Y., Yamada, K., 1973. Identification of microbial products from dibenzothiophene and its proposed oxidation pathway. Agric. Biol. Chem. 37, 45e50. Kropp, K.G., Fedorak, P.M., 1998. A review of the occurrence, toxicity,and biodegradation of condensed thiophenes found in petroleum. Can. J. Microbiol. 44, 605e622. Kropp, K.G., Andersson, J.T., Fedorak, P.M., 1997. Bacterial transformations of naphthothiophenes. Appl. Environ. Microbiol. 63, 3463e3473. McLafferty, F., Turecek, F., 1993. Interpretation of Mass Spectra, fourth ed. University Science Books, California, USA. MacKinnon, M., Boerger, H., 1986. Description of two treatment methods for detoxifying oil sands tailings pond water. Water Pollut. Res. J. Can. 21, 496e512. Panda, S.K., Andersson, J.T., Schrader, W., 2009. Characterization of supercomplex crude oil mixtures: what is really in there? Angew. Chem. 121, 1820e1823. Reinardy, H.C., Scarlett, A.G., Henry, T.B., West, C.E., Hewitt, L.M., Frank, R.A., Rowland, S.J., 2013. Aromatic naphthenic acids in oil sands process-affected water, resolved by GCxGC-MS, only weakly induce the gene for vitellogenin production in zebrafish (Danio rerio) larvae. Environ. Sci. Technol. 47, 6614e6620. Rowland, S.J., Scarlett, A.G., Jones, D., West, C.E., Frank, R.A., 2011a. Diamonds in the rough: Identification of individual naphthenic acids in oil sands process water. Environ. Sci. Technol. 45, 3154e3159. Rowland, S.J., West, C.E., Scarlett, A.G., Jones, D., 2011b. Identification of individual acids in a commercial sample of naphthenic acids from petroleum by two dimensional comprehensive gas chromatography-mass spectrometry. Rapid Commun. Mass Spectrom. 25, 1741e1751. Rowland, S.J., West, C.E., Scarlett, A.G., Ho, C., Jones, D., 2012. Differentiation of two industrial oil sands process-affected waters by two-dimensional gas chromatography/mass spectrometry of diamondoid acid profiles. Rapid Commun. Mass Spectrom. 26, 572e576. 215 Saftic, S., Fedorak, P.M., Andersson, J.T., 1993. Transformations of methyldibenzothiophenes by three Pseudomonas isolates. Environ. Sci. Technol. 27, 2577e2584. Scarlett, A.G., Reinardy, H.C., Henry, T.B., West, C.E., Frank, R.A., Hewitt, L.M., Rowland, S.J., 2013. Acute toxicity of aromatic and non-aromatic fractions of naphthenic acids extracted from oil sands process-affected water to larval zebrafish. Chemosphere. 93, 415e420. Scarlett, A.G., West, C.E., Jones, D., Galloway, T.S., Rowland, S.J., 2012. Predicted toxicity of naphthenic acids present in oil sands process-affected waters to a range of environmental and human endpoints. Sci. Total Environ. 425, 119e127. Schindler, D., 2010. Tar sands need solid science. Nature 468, 499e501. Scott, A.C., MacKinnon, M.D., Fedorak, P.M., 2005. Naphthenic acids in athabasca oil sands tailings waters are less biodegradable than commercial naphthenic acids. Environ. Sci. Technol. 39, 8388e8394. Smith, B.E., Lewis, C.A., Belt, S.T., Whitby, C., Rowland, S.J., 2008. Effects of alkyl chain branching on the biotransformation of naphthenic acids. Environ. Sci. Technol. 42, 9323e9328. Strausz, P.P., Lown, E.M., Morales-Izquierdo, A., Kazmi, N., Montgomery, D.S., Payzant, J.D., Murgich, J., 2011. Chemical composition of Athabasca bitumen: the distillable aromatic fraction. Energy Fuels 25, 4552e4579. Tollefsen, K.-E., Mathisen, R., Stenersen, J., 2003. Induction of vitellogenin synthesis in an Atlantic salmon (Salmo salar) hepatocyte culture: a sensitive in vitro bioassay for the oestrogenic and anti-oestrogenic activity of chemicals. Biomarkers 8, 394e407. Tollefsen, K.-E., Blikstad, C., Eikva, R.S., Finne, E.F., Gregersen, I.K., 2008. Cytotoxicity of alkylphenols and alkylated non-phenolics in a primary culture of rainbow trout (Oncorhynchus mykiss) hepatocytes. Ecotoxicol. Environ. Saf. 69, 6e73. Tollefsen, K.E., Petersen, K., Rowland, S.J., 2012. Toxicity of synthetic naphthenic acids and mixtures of these to fish liver cells. Environ. Sci. Technol. 46, 5143e5150. Watson, J.S., Jones, D.M., Swannell, R.P.J., van Duin, A.C.T., 2002. Formation of carboxylic acids during aerobic biodegradation of crude oil and evidence of microbial oxidation of hopanes. Org. Geochem. 33, 1153e1169. West, C.E., Scarlett, A.G., Pureveen, J., Tegelaar, E., Rowland, S.J., 2013. Abundant naphthenic acids in oil sands process-affected water: studies by synthesis, derivatisation and GCxGC-high resolution mass spectrometry. Rapid Commun. Mass Spectrom. 27, 357e365.