Environ. Sci. Technol. 2003, 37, 314-320
Bioaccumulation of LAS in Feral Fish Studied by a Novel LC-MS/MS Method J O H A N N E S T O L L S , * ,†,‡ ROBERTO SAMPERI,‡ AND ANTONIO DI CORCIA‡ Environmental Toxicology and Chemistry, Institute of Risk Assessment Sciences, Utrecht University, Bolognalaan 50, P.O. Box 80058, 3508 TD Utrecht, The Netherlands, and Dipartimento di Chimica, Universita di Roma ‘La Sapienza’, Piazzale Aldo Moro 5, I-00185 Roma, Italy
The extent of bioaccumulation of linear alkylbenzene sulfonate (LAS) in feral organisms is presently unknown. To enable LAS determination in biota samples, LAS and its coproducts (methylbranched LAS, dialkyltetralin sulfonates) are extracted from tissues using matrix solid-phase dispersion, isolated by strong anion exchange chromatography and determined by HPLC-electrospray-tandem mass spectrometry. All analytes were quantified in sediment dwelling Tubifex sp. with the sum of the concentrations exceeding 1 µmol/g. Since a portion of LAS was present adsorbed to solids in the guts, the actual body residue was lower than reported lethal LAS body residues. The concentrations of individual constituents in bream muscle samples from the river Saar and fathead minnows caged in the river Arrone ranged up to 2 nmol/g. The apparent bioaccumulation factors in the caged fish are consistently higher than laboratory data, presumably due to a combination of LAS present in the guts adsorbed to suspended material, coingestion of LAS with bacterial detritus, and depressed metabolic activity due to sublethal effects. Given the small deviation between laboratory and field bioaccumulation data and the high detection frequency (>90%) of C13-2- and C13-iso-LAS, the latter two constituents are suitable markers for LAS contamination in fish.
Introduction Linear alkylbenzene sulfonate (LAS) is the workhorse synthetic surfactant with a global annual production exceeding 3 million tons (1) and is discharged into the sewer after use. Despite effective removal during wastewater treatment LAS occurs in surface waters (2-7). Consequently, aquatic organisms are exposed to LAS. Only recently, parent compound specific LAS bioaccumulation data were reported. The bioaccumulation potential of the LAS constituents, expressed as bioconcentration factor (BCF), ranges between 2 and 1000 L/kg (8). The validity of these findings for the field situation has not yet been demonstrated. One important difference
between the controlled laboratory experiments and the exposure of feral organisms is that natural waters contain suspended solids and dissolved organic chemicals, which are known to reduce the fraction of organic compounds that is available for uptake via the gills (9, 10). Second, the LAS concentrations in the experiments were far above those in natural waters in order to obtain LAS concentrations in fish that are measurable by fluorescence detection, the technique initially developed for that purpose (11). The fluorescence technique is expected to be not sufficiently sensitive for detecting LAS in feral fish considering reported LAS concentrations in surface waters and the observed BCFs. This is also true for the sulfonated coproducts of LAS production, dialkyltetralinsulfonates (DATS) and monomethylbranched LAS (iso-LAS). Therefore, an analytical method for LAS determination in feral organisms has to be more sensitive. Besides, it must be more specific since relatively small concentrations of LAS and its coproducts have to be discriminated against other compounds that may be abundant in the organism extracts. HPLC coupled to electrospray ionization-mass spectrometry (HPLC-ESI-MS) fulfills both requirements for other surfactants and their metabolites, such as alcohol ethoxylates in wastewater (12), anionic surfactants of the alkyl and the alkyl ether sulfate type in natural and wastewater (13), perfluorinated alkylsulfonic and fatty acids in water, biological tissues (14-16) and blood samples (17), and for LAS metabolites in water (18). Recently, a LC-ESI-MS method for quantitating LAS and its sulfonic acid type coproducts such as dialkyltetralinsulfonates (DATS) and monomethylbranched LAS (isoLAS) in aqueous samples has been presented (7) and applied to river water samples, showing that DATS and iso-LAS to be less biodegradable than LAS (7). In contrast to gas chromatographic analysis of LAS in water (5) and sediments (19), HPLC-ESI-MS is sensitive and selective without derivatization. It has already been employed in investigating LAS bioaccumulation in organisms exposed to elevated concentrations of the dodecane LAS homologue in experimental streams (20). Here we ask whether the presence of LAS and its byproducts in river water results in measurable LAS concentrations in feral organisms. Second it is of interest whether the concentration in fish can be predicted with reasonable accuracy from laboratory bioconcentration data. These questions were addressed in the present study by first developing a novel method for preparation of biological samples and for HPLC-ESI-MS/MS determination of LAS in feral organisms. The method was applied to bream (Abramis brama) muscle samples taken during the last 10 years in the river Saar (Germany) and to tubifex (Tubifex sp.) samples collected from the sediment of the river Arrone close to the outlet of a wastewater treatment plant. In the same location a caging experiment with fathead minnows (Pimephales promelas) was performed. The resulting bioaccumulation data were compared to those obtained in the laboratory (with the same fish species) in order to evaluate the usefulness of the latter for estimating field bioaccumulation of LAS. To the best of our knowledge, this is the first report of the occurrence of LAS and its coproducts in feral organisms.
Materials and Methods * Corresponding author phone: (49) 211 797 7860; fax: (49) 211 798 7505; e-mail:
[email protected]. Present address: Henkel KgaA, D-40191 Du ¨ sseldorf, Germany. † Utrecht University. ‡ Universita di Roma ‘La Sapienza’. 314
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 2, 2003
Chemicals. Methanol, acetonitrile, hexane, ethyl acetate, and dichloromethane were of HPLC-grade quality or better and purchased from different commercial suppliers. They were redistilled in glass prior to use, to reduce LAS blank levels. 10.1021/es020082m CCC: $25.00
2003 American Chemical Society Published on Web 12/10/2002
Distilled water was further purified (Milli-Q Plus, Millipore, Bedford, MA) prior to use. Hydrochloric acid 37% (Merck, Darmstadt) was distilled in glass to obtain a 10.25 M (densitometry) HCl. Formic acid (99%), the sodium salt of 4-n-(sulfophenyl)-octane (C8-1-LAS), and Supelclean LC-SAX were supplied by Sigma-Aldrich-Fluka (Zwijndrecht). Glass Microfiber Filters (GF/C, 15 cm) were from Whatman (Maidstone, England). The graphitized carbon black was Carbograph 4 (Restek, Bellefonte, PA). Octadecylsilica (ODS), glass solid-phase extraction tubes (6 mL), and Teflon frits were obtained from Baker Mallinckrodt (Deventer, The Netherlands). ODS was washed with redistilled MeOH prior to use. The extraction cartridges were prepared by packing 0.5 g of GCB or SAX material into a 6 mL syringe and placing frits above and below the sorbent bed. Standards. C8-1-LAS was added with the redissolution solvent prior to injection and served to control the instrument performance. The 1-isomer of n-(p-sulfophenyl)decane (C101-LAS) was employed as a surrogate standard to correct for losses during the analytical procedure and allowed for quality control. C10-1-LAS and its dodecane homologue (C12-1-LAS) were synthesized by sulfonation of 1-phenyl-decane and 1-phenyl-dodecane (Sigma-Aldrich-Fluka, Zwijndrecht, The Netherlands) according to the method of Gray et al. (21) and purity exceeded 99% (LC-ESI-MS). Pure 2-, 3-, 4-, 5-, and 6-isomers of (p-sulfophenyl)dodecane were synthesized to a purity of more than 98% as described earlier (22). Biota Samples. Bream (Abramis brama) muscle samples from two locations in the river Saar, Gu¨dingen and Rehlingen (Germany), were kindly provided by the German Environmental Specimen Bank. About 30 fishes are collected annually at each site. Their fillets are pooled, homogenized, ground to a fine powder, and stored in 10 g-aliquots. After sampling, the chain of processing steps is carried out under cryogenic conditions, and storage is in the gas phase above liquid nitrogen (T < -140 °C) according to obligatory standard operating procedures (23). Tubifex samples were obtained from the sediment of the river Arrone at the location of the caging study. To that end, the top 3 cm of the Arrone river sediment were sampled and transported to the laboratory within 2 h of sampling. The sediment was suspended in tap water, and the suspension was allowed to form a thin film on a glass plate from which the tubifex specimen were harvested with tweezers and transferred to a glass beaker containing clean water. During a period of 3 h the water was exchanged several times in order to remove sediment particles adhering to their exterior and to allow depuration of ingested solids. After washing the tubifex were killed in liquid nitrogen and kept refrigerated (-20 °C) until chemical analysis. Individuals were pooled (0.8 g) for chemical analysis. Caging Study. The caging study was performed in the river Arrone, an effluent of Lake Bracciano (40 km to the Northwest of Rome, It), about 1.5 km southeast of the town Anguillara and 100 m downstream of the discharge of the COBIS sewage treatment plant (STP). The COBIS plant is fed primarily with domestic sewage from a circular pipeline that collects the wastewater of all settlements that are located on the border of the lake. As a result, the travel times of the sewage from the different settlements are different, thereby attenuating the variation of the LAS concentrations caused by the diurnal LAS use pattern. The COBIS plant uses activated sludge treatment, has a capacity of 10 000 m3/day, a hydraulic retention time of 12 h, and aerobic sludge digestion. It achieves a LAS removal percentage of more than 98% (7) without a tertiary treatment step or sludge flocculation. Its effluent (ca. 0.12 m3/s) is discharged into the river (0.1 m3/s just upstream of the caging site) at a perpendicular angle and under highly turbulent conditions. Therefore, mixing is complete upstream of the caging site. Elevated LAS concentrations (7) due to the effluent accounting for a significant
portion of the exposure water and the attenuated variation in diurnal concentration patterns make the COBIS site particularly suitable for the purpose of the present study. During the experiment, we observed brown fluffy flocs of bacterial biomass floating, which originated from the WWTP and disintegrated in the river. The fathead minnow (Pimephales promelas) was used in earlier laboratory LAS bioconcentration experiments (8, 24). The specimen used were reared in the hatchery of Utrecht University, transported to Rome, and kept in Roman tap water until the start of the experiment. On October 24, 1999, 20 fishes were placed into a net, which in turn was placed into a cage made from a steel frame, spanned with meshed wire (1 cm mesh size) that was to protect the net from mechanical damage. The cage was held in place during the experiment in the middle of the river by means of a rope. Grab water samples (2L) were collected from the middle of the stream each day at about 10:30 a.m. ((30 min) in amber glass bottles and preserved by adding 2% formalin and stored in the refrigerator at 4 °C before analysis. pH and the hardness of the water were measured after homogenization of the water samples. On October 29, the fishes were netted out of the cage. All 20 fishes were alive and looked healthy. They were carefully blotted with a paper tissue, killed by cervical dislocation, and put on ice until storage at -20 °C. Chemical Analysis. Water and Suspended Solids. The dissolved LAS fraction and that adsorbed to suspended particles were separated by a standard filtration step (5) employing GF/C microfiber filters. 2 L of water sample was homogenized by shaking vigorously and filtered over dried (7 h at 105 °C and subsequent equilibration to laboratory air for 24 h) and preweighed filters. Dissolved analytes were extracted from an aliquot of 1 L of the filtrate using the procedure described by Di Corcia et al. (7). LAS adsorbed to the solids retained on the filter were extracted immediately after filtration by washing the suspended material with 40 mL of MeOH and combining the washing with 210 mL of Millipore water. The solids mass was determined by weighing the dried filters. Upon addition of the surrogate spike (C101-LAS) the extract of the suspended solids is prepared like a water sample. Recovery experiments (performed by amending suspended solids retained on the filter with 1.6 nmol C10-1-LAS) showed that the suspended solids were extracted quantitatively (99 ( 6%). Adsorption to the filter material was evaluated by placing two filters on top of each other and corrected for by subtracting the amount of LAS found in a second filter from that in the first filter. Biota Samples. The analytes were extracted from the biota samples by the matrix solid-phase dispersion extraction procedure described earlier (11). In short, the biota sample is weighed and placed into a mortar. 4 g of ODS per gram of sample and the surrogate spike are added and ground to obtain a homogeneous paste, which is allowed to fall dry and powdery. Then the sample is transferred to a glass column fit with a Teflon frit at the bottom. The column thus obtained is eluted with 32 mL of hexane (to remove lipids) and 5 mL of ethyl acetate. Both fractions are discarded. Then 16 mL of a 1:1 mixture of ethyl acetate:methanol (MeOH) elute LAS and its coproducts. This fraction is passed online over a strong anion exchange (SAX) solid-phase extraction (0.5 g of sorbent, held in place by two PTFE frits) cartridge for isolation of the analytes. Weakly acidic matrix constituents are washed off the SAX cartridge with 5 mL of 0.36 M formic acid in MeOH. The analytes were extracted by back flushing the SAX cartridge with 8 mL of a mixture of MeOH:2 M HClaq (4:1). The extract is dried in a water bath at 50 °C under a gentle stream of N2 and resuspended in 300 µL of MeOH containing the internal standard. Instrumental Analysis. Water and suspended matter samples were analyzed using the HPLC-ESI-MS conditions VOL. 37, NO. 2, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
315
reported earlier (19). Biota samples were chromatographed on a C-18 column (250 mm × 3.2 mm i.d., 5-µm particle size, Alltima, Alltech, NL). Acetonitrile (A) and water (B), both containing 1 mM ammonium acetate, were employed in gradient elution. Solvent A was increased from 0 to 31 min from 10 to 60% and during 5 more minutes to 100%, followed by 3 minutes of isocratic elution at 100% A. Dissolved gases in both solvents were removed by an online degasser. The flow rate of the mobile phase was 0.5 mL/min. Detection was by tandem mass spectrometry in the MRM mode. The mass spectrometric conditions were optimized by infusing solutions of pure LAS isomers into the ES-interface of a PE Sciex 365 triple quadrupole mass spectrometer (Foster City, CA) operated in the negative-ion mode detection. These experiments were also employed to obtain CID-spectra of these compounds. For LAS detection, the declustering potential, the focusing potential, and the entrance potential were set at -50 V, -190 V, and 10 V, respectively. The molecular ion was isolated in Q1 and upon fragmentation in the collision cell (collision gas N2, collision energy: 45 V) the daughter ions (183 m/z for LAS, 223 m/z for DATS and 170 m/z for the internal and the surrogate standard) were allowed to pass Q3 in order to be measured. In the HPLC analysis the entire column effluent was introduced into ESIinterface, which was operated at a spraying voltage of 4.2 kV, a spraying gas temperature of 450 °C, and a source gas flow of 6 to 8 L/min. The vacuum in the mass analyzer was below 10-5mbar. Data were collected and handled with Analyst software (PE Sciex, Foster City, CA). Quantitation and Quality Control. Quantitation of each analyte was carried out by relating the area of the analyte with the area of the internal standard (C8-1-LAS). The recovery was calculated by comparing the measured and the added amount of C10-1-LAS (surrogate spike), and the analyte’s concentration was corrected for the recovery of the surrogate spike. One procedural blank was run in each series of five biota samples. Positive detections had to exceed the average of the blank signal by three times the standard deviation of the procedural blank.
Results and Discussion Method Development and Application to Feral Organisms. CID-spectra of the individual isomers of n-(p-sulfophenyl)dodecane demonstrate that m/z 170 is the base peak of the 1-isomer. Other significant product ions were m/z 119 and m/z 80. The other isomers dissociate to form m/z 183 as most important product ion corresponding to the styrenesulfonic acid anion. m/z 119 and m/z 80 are further major product ions, obtained by loss of SO2 from m/z 183 and by cleavage of SO3-, respectively. Closer inspection of the spectra of the 2- to 6-p-(sulfophenyl)-dodecanes reveals that these isomers cannot be differentiated mass spectrometrically for the purpose of quantitative analysis. C10-DATS fragments by losing the alkyl side chains to yield m/z 223, m/z 209, and m/z 207 as the major and m/z 195, m/z 143, and m/z 80 as minor product ions. Therefore, we decided to employ m/z 183, m/z 223, and m/z 170 as diagnostic ions for the LAS, DATS, and the internal and surrogate standards, respectively. The ionization efficiency of LAS increases with increasing content of organic modifier in the HPLC-column effluent and thus with molecular weight. As a result, the mass based response factor relative to C8-1-LAS is similar for all LAS homologues (25). Response differences between the isomers as a result of the ionization were evaluated in flow injection experiments. C12-3, C12-4, and C12-5-LAS do not differ regarding their molar response relative to the internal standard. It amounts to (unitless) 0.96 (C.V: 4.7%). For C12-2 the response is 1.10 (C.V: 2.5%). For these constituents the error made in employing a response factor of 1 is small. C12-6 is more responsive to LC-MS detection with a response factor 316
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 37, NO. 2, 2003
of 1.63 (2.5%). C12-6 coelutes with C12-5 in HPLC-separation (and also in high-resolution GC of the trifluoroethyl derivatives (5)) and is grouped together with C12-3, C12-4 as the inner isomers. In this group, C12-6 is one out of four constituents, and its contribution to this C12-inner is unknown. Therefore, the error made in quantification of the inner isomers is relatively small and cannot be corrected for. Consequently, we chose to employ 1 as a response factor for the inner-isomer groups. For DATS authentic standards are not available; therefore, we assumed the molar response for DATS to be identical to that of LAS. In previous work, complete recovery of LAS from fish tissue was achieved by matrix solid-phase dispersion (MSPD) extraction and subsequent isolation via a liquid-liquid ionpair extraction step (11) employing tetrabutylammonium (TBA). Since high TBA concentrations in the extracts might compromise the electrospray process and thus the high sensitivity of ESI-MS, the latter step was replaced by a silica based solid-phase anion exchange chromatography step, that had been applied for the isolation of LAS from wastewater extracts (26). Our modifications were replacement of acetic with formic acid and backflush elution. The recovery rate of the surrogate standard C10-1-LAS for all analyzed samples was 102 ( 15% (n)37) indicating that no analyte losses occur during sample workup and the procedure’s reproducibility. Blanks and Limit of Quantitation. Owing to its surface activity and ionic character, LAS can be detected with electrospray-mass spectrometry with large sensitivity. By largely eliminating the chemical noise using tandem mass spectrometry in the MRM mode we achieved low limits of quantitation for the surrogate standard (10 pg injected). However, the target analytes were present in reagent blanks of almost all materials employed causing the limit of quantitation of the whole procedure to be higher. Therefore, all solvents were redistilled in glass prior to use. Octadecylsilica was washed with redistilled MeOH to remove traces of LAS. Polyethylene frits of different suppliers were found to be the primary source of LAS in reagent blanks and were replaced by PTFE frits. These measures resulted in a significant reduction of the procedural blank levels but not to a complete elimination. Table 1 shows the blank levels found for the different analytes along with their standard deviation. The limit of quantitation was defined as the average blank concentration plus 3 times the standard deviation. Clearly, the high blank concentrations found for e.g. C11and C12-inner result in high limits of quantitation (ca. 300 pmol/g tissue), while the LOQs of iso-LAS and DATS are relatively low (
