Laser-induced fluorescence detection at 266 nm in capillary electrophoresis Polycyclic aromatic hydrocarbon metabolites in biota

Journal of Chromatography A, 907 (2001) 291–299 www.elsevier.com/locate/chroma Laser-induced fluorescence detection at 266 nm in capillary electrophoresis Polycyclic aromatic hydrocarbon metabolites in biota ab cba ´ J. Kuijt , C. Garcıa-Ruiz , G.J. Stroomberg , M.L. Marina , F. Ariese , aa, * U.A.Th. Brinkman , C. Gooijer a Vrije Universiteit , Department of Analytical Chemistry and Applied Spectroscopy , De Boelelaan 1083, 1081 HV Amsterdam , The Netherlands b ´´´´ Universidad de Alcala , Departamento de Quımica Analıtica , Facultad de Quımica , Ctra . Madrid–Barcelona , Km . 33,600, ´ E28871 Alcala de Henares , Madrid , Spain c Vrije Universiteit , Institute for Environmental Studies , De Boelelaan 1115, 1081 HV Amsterdam , The Netherlands Received 30 June 2000; received in revised form 29 September 2000; accepted 13 October 2000 Abstract The separation of five phenolic polycyclic aromatic hydrocarbon metabolites (hydroxy-PAHs) has been performed by cyclodextrin-modified micellar electrokinetic chromatography (CD-MEKC) using a 30 mMborate buffer (pH 9.0) containing 60 mMsodium dodecyl sulfate and varying concentrations of g-cyclodextrin (g-CD). A concentration of 12.5 mMg-CD was found to provide a baseline separation of the five hydroxy-PAHs. We applied conventional fluorescence and laser-induced fluorescence (LIF) detection, using a new, small-size, quadrupled Nd–YAG laser emitting at 266 nm. The best limits of detection, in the low ng/ml range, were achieved using LIF detection. For all analytes, linearity was observed up to ca. 100 ng/ml. As an application, conjugated pyrene metabolites in hepatopancreas samples from the terrestrial isopods Oniscus asellus and Porcellio scaber were separated and detected. Finally, flatfish bile samples from individuals exposed to polluted sediment or crude oil, which were part of an interlaboratory study, were analyzed by CD-MEKC with conventional fluorescence and LIF detection to determine the 1-hydroxypyrene concentrations. 2001 Elsevier Science B.V. All rights reserved. Keywords : Laser-induced fluorescence detection; Detection, electrophoresis; Fish; Polynuclear aromatic hydrocarbons 1. Introduction electrophoresis (CE), mainly because LIF detection using excitation in the deep UV range is not straightAt present, laser-induced fluorescence (LIF) deforward. Until now, such excitation required the use tection for analytes exhibiting native fluorescence, of large frame argon ion lasers or excimer laserfollowing excitation in the deep UV range, has been pumped dye laser systems [1,2]. Unfortunately, these hardly exploited for detection purposes in capillary laser systems are expensive and impractical in analytical applications because of their large dimensions. *Corresponding author. Fax: 131-20-4447-543. E-mail address : [email protected] (C. Gooijer). In this study, the performance in LIF detection of 0021-9673/01/$ – see front matter 2001 Elsevier Science B.V. All rights reserved. PII: S0021-9673(00)01040-2 292 J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 a new, small-size, quadrupled Nd–YAG laser emitwas used to prepare buffers and analyte standard ting 266-nm radiation at a high pulse frequency was solutions. Buffers were adjusted to the required pH investigated. Such a laser system will be of special using 1 MNaOH. Boric acid and ascorbic acid were interest in bioanalytical studies since the 266-nm obtained from J.T. Baker (Deventer, The Netheroutput is fully compatible with the excitation of lands), absolute ethanol (analytical-reagent grade) ¨ natively-fluorescent amino acids (tyrosine, from Riedel-de Haen (Seelze, Germany) and sodium tryptophan and phenylalanine). In the present study, dodecyl sulfate (SDS) from Fluka (Buchs, Switzerhowever, its appropriateness for detection of polynuland). 1,5-Naphthalenedisulfonic acid and g-cycloclear aromatic hydrocarbon (PAH) metabolites is dextrin hydrate were purchased from Aldrich (St. tested. Louis, MO, USA). b-Glucuronidase/aryl sulfatase Because the metabolism of PAHs after PAH (from Helix pomatia, EC 3.2.1.31 and EC 3.1.6.1, 30 exposure is very fast for most organisms, the conand 60 U/ml, respectively) was obtained from centrations of PAH metabolites in biotic matrices are Merck (Darmstadt, Germany). Buffer solutions were generally much higher than those of the parent filtered over 0.45-mm syringe filters (Schleicher & PAHs. It is therefore advantageous to focus attention Schuell, Dassel, Germany) before use. on the determination of PAH metabolites, instead of on their parents. In the last decade the quantification 2.2. Standard solutions and samples of PAH metabolites in excreta (urine, bile and feces) has become an important method for the biomonitorThe hydroxy-PAHs used in this study are 1-hying of PAH exposure in various terrestrial and droxypyrene, 3-hydroxybenzo[a]pyrene, 1-hydroxyaqueous ecosystems [3–7], as well as for monitoring phenanthrene, 1-hydroxychrysene and 2-hydroxyhuman PAH exposure [8–13]. naphthalene. Stock solutions containing these five Specifically, 1-hydroxypyrene has extensively hydroxy-PAHs (10 mg/ml) as well as the fish bile been used as a biomarker for PAH exposure. The samples were distributed as part of a European determination of 1-hydroxypyrene and other hyUnion (EU)-funded interlaboratory study (SMT 4droxy-PAHs is performed after enzymatic hydrolysis CT 98-2250) [17]. The stock solutions had been of phase II conjugates such as 1-hydroxypyrene prepared with ethanol–water (80:20, v/v), containing sulfate and 1-hydroxypyrene glucuronide. Although 5 mg/ml ascorbic acid to avoid oxidation. From the standards are not yet commercially available, the stock solutions, 1 mg/ml solutions were prepared potential of direct monitoring of phase II metabolites using ethanol–water (80:20, v/v) with 5 mg/ml has also been explored [5,9,10,13]. ascorbic acid. In order to prepare calibrant solutions, Phenolic and conjugated PAH metabolites are further dilutions were made by mixing with ethanol– usually analyzed by liquid chromatography (LC) water (80:20, v/v) without ascorbic acid. combined with fluorescence detection [3–13]. Few Hepatopancreas samples from the isopods Porcelstudies report on the separation of hydroxy-PAHs by lio scaber and Oniscus asellus, which contained CE [14–16], but until now CE has not been applied conjugated pyrene metabolites, were obtained as to the detection of these compounds in real (biotic) described by Stroomberg et al. [5]. Briefly, the samples. In the present study the separation and isopods were starved for 3 days and individually detection of five hydroxy-PAHs, and of pyrene and placed in scintillation vials for 48 h, together with its metabolites, in biological matrices (fish bile and 100 mg of ground leaves spiked with pyrene at a isopod hepatopancreas) is studied. level of 10% (w/w) dry mass. After dissection, hepatopancreases of 10 animals were pooled in 100 ml of Tris buffer (pH 9.0) and 20 ml of protease K 2. Materials and methods was added. Proteolysis was performed for 18 h at 378C. Next, an equal volume of ethanol was added 2.1. Chemicals and materials and the remaining debris was precipitated by 3-min centrifugation at 5000 rpm. Individual metabolite Water demineralized and distilled in the laboratory solutions were obtained by fractionation of the J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 293 samples by means of reversed-phase LC, with a Meylan, France), emitting at 266 nm with an average gradient of acetonitrile and ammonium acetate buffer output power of 5.4 mW, a pulse width of 0.34 ns (10 mM, pH 5.0) [5]. For prolonged use, all stanand a repetition rate as high as 7.8 kHz, was focused dards and samples were stored in the refrigerator at into the liquid core of the capillary by a quartz lens 2208C. Before use, the solutions were sonicated for with a focal distance of 3 cm. A 253, N.A. 0.4 5 min. reflective objective (Ealing, Holliston, MA, USA) was used to collect the fluorescence emission. The 2.3. Instrumental set-up 9558 QA photomultiplier tube (EMI, Middlesex, UK) used for detection was connected to a 456H An LS-50B fluorescence spectrometer (Perkinpower supply (Ortec, Bracknell, UK) operated at Elmer, Beaconsfield, UK) was used to record fluo1000 V. Neutral density filters were used in order to rescence excitation and emission spectra. Elecdecrease the irradiance of the capillary detection tropherograms were obtained using a Prince highwindow. A UG-1 bandpass filter and BG-14 or voltage/injection system (Lauerlabs, Emmen, The GG-13 cut-off filters (Schott, Mainz, Germany) were Netherlands) in combination with a conventional FPused to collect all emission and discriminate against 920 fluorescence detector (Jasco, Tokyo, Japan) or a scattered excitation light. laboratory-built set-up for LIF detection. CE separations using the conventional fluorescence detector were performed at a voltage of 20 kV with a 90 3. Results and discussion cm375 mm I.D. uncoated fused-silica capillary (BGB Analytik, Anwil, Switzerland) with an effecThis section first discusses the CE separation and tive length of 60.5 cm. For the LIF experiments a fluorescence detection of five hydroxy-PAHs which 104 cm capillary with an identical effective length would be the target analytes in the fish bile samples. was used; the separation was run at 23.3 kV to The parent PAHs are usually not observed in such maintain the same field strength (221 V/cm). The samples. Therefore, it is not necessary to involve the injection pressure was 25 mbar and the injection time parent compounds in the separation. Emphasis is on 0.1 min. For data collection a computer equipped the performances of the new 266-nm LIF method with CLASS-VP data-acquisition software was used. compared with conventional fluorescence detection. The LIF set-up is shown in Fig. 1. The light from Subsequently, the determination of pyrene metaboa quadrupled Nd–YAG NanoUV laser (Uniphase, lites in isopods and fish bile samples will be discussed. For such complex matrices the LIF method will, no doubt, suffer from the disadvantage that it offers no selectivity in excitation settings: only one, single, wavelength is available, i.e., 266 nm. Comparison with conventional fluorescence detection is, therefore, of distinct interest. 3.1. CE of hydroxy-PAHs To perform conventional fluorescence detection under optimum conditions, the excitation and emission fluorescence spectra of the hydroxy-PAHs were recorded in the selected buffer system (see below). If more than one maximum was present, as is the case for the excitation spectra, the maximum giving the Fig. 1. Detection system for LIF detection. UVL: 266-nm laser; highest signal-to-noise ratio (S/N) in the elecF1: neutral density filter; L: focusing lens; C: capillary (focusing tropherograms was chosen. The maxima used for point); O: reflecting objective; M: mirror; F2: bandpass or cut-off filter; PMT: photomultiplier tube. detection in CE are given in Table 1. The conven- 294 J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 Table 1 a Limits of detection (ng/ml) for five hydroxy-PAHs using CE with LIF and conventional fluorescence (CF) detection Compound CF LIF, UG-1 LIF, BG-14 3-Hydroxybenzo[a]pyrene 2 (383/435) 25 2 2-Hydroxynaphthalene 40 (275/344) 3 4 1-Hydroxyphenanthrene 7 (306/361) 3 2 1-Hydroxychrysene 35 (307/370) 10 6 1-Hydroxypyrene 1 (347/387) 4 2 a In parentheses (CF): excitation/emission wavelengths used for detection. tional fluorescence detector has a fixed excitation slit jections close to the detection limit. As shown in of 10 nm, but the emission slit can be varied (10, 18 Table 1, the LODs were in the low ng/ml range for or 40 nm bandwidth). An emission bandwidth of 40 three of the test analytes, but about 10-fold higher nm gave the best results in terms of S/N. for 2-hydroxynaphthalene and 1-hydroxychrysene. As far as the CE separation is concerned, micellar Using peak areas, linear ranges from 10 to 500 electrokinetic chromatography (MEKC) as such does ng/ml were obtained for 3-hydroxybenzo[a]pyrene 2 not suffice: using a 30 mMborate, 60 mMSDS and 1-hydroxypyrene (with Rvalues of 0.9997 and buffer (pH 9.0), all analytes were found to migrate 0.9993, respectively; five data points in duplicate for simultaneously. This is not surprising because all both). For peak heights, linearity was observed from 2 target compounds are rather hydrophobic and there10 to 100 ng/ml with Rvalues of 0.9992 and fore have similar affinity for the micelles [18]. In 0.9969 (four data points in duplicate) for 3-hydroxyorder to generate selectivity, g-cyclodextrin (g-CD) benzo[a]pyrene and 1-hydroxypyrene, respectively. was added to the buffer, an approach proven to be For 1-hydroxyphenanthrene, linearity was observed valuable in the separation of a different set of from 25 to 200 ng/ml, using either peak areas or 2 hydroxy-PAHs [14]. The selectivity that can be peak heights (R50.9940; four data points in dupli2 achieved with a g-CD–SDS buffer system is based cate, and R50.9952; five data points in duplicate, on the different partitioning of the analytes between respectively). Because the LODs for 2-hydroxythe micellar, aqueous and g-CD phases. The affinity naphthalene and 1-hydroxychrysene were rather of a particular analyte for a CD molecule strongly high, no linear ranges were determined for these depends on its fit in the CD cavity. Although b-CDs compounds. are generally considered to be more efficient comIn general, the linear ranges are shorter when plexing agents than g-CDs, the latter are preferred in using peak heights, which can be readily explained CD-MEKC because the b-CD molecules are not able by the peak broadening occurring at levels above ca. to bind analytes strongly in micellar buffer systems. 100 ng/ml. This peak broadening is presumably It has been suggested that this can be ascribed to the caused by overloading of the micellar and g-CD co-inclusion of individual SDS molecules which phases. The co-inclusion of individual SDS molereduces the actual cavity size, especially in the case cules into CD cavities, as mentioned above, implies of b-CDs [19,20]. that the effective micellar concentrations are lower In this work we used a 30 mMborate, 60 mMthan expected from the total SDS concentration and SDS, pH 9.0 buffer with varying concentrations of the critical micellar concentration. Also, the cong-CD (10, 12.5, 15 and 20 mM) to achieve the centration of g-CD molecules with a cavity large separation of the five hydroxy-PAHs. A concenenough for strong inclusion of hydroxy-PAHs will be tration of 12.5 mMwas found adequate to separate reduced. These assumptions are supported by the all hydroxy-PAHs. observation that the peak of 3-hydroxyThe limits of detection (LODs) of the hydroxybenzo[a]pyrene, which shows the highest affinity for PAHs obtained with conventional fluorescence dethe g-CD molecules, shows tailing at high contection were determined from three replicate incentrations, which indicates overloading of the g-CD J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 295 phase. On the other hand, the fronting observed for filter with a transmission (T) of 6.3%, the capillaries the 1-hydroxypyrene peak at higher concentrations could be used for at least several weeks, while suggests overloading of the micellar phase. maintaining adequate detection sensitivity. A typical With the conventional fluorescence detector opLIF-detected electropherogram of a standard mixture timum detection sensitivity for all compounds reof the five hydroxy-PAHs is shown in Fig. 2. High quires programmed adjustment of the excitation and efficiency peaks were obtained with plate numbers of emission wavelengths. This proved to be quite 128 000, 199 000, 241 000, 191 000 and 214 000 difficult, because the migration times were found to (peak numbers 2–6, respectively). show slight run-to-run variations and the compounds The fluorescence excitation spectra (not shown) of are migrating closely together. Moreover, the LODs the hydroxy-PAHs suggest that only 2-hydroxyfor 2-hydroxynaphthalene and 1-hydroxychrysene naphthalene and 1-hydroxypyrene can be excited will often be inadequate for the determination of efficiently at 266 nm. For the other metabolites the these compounds in real samples. Therefore, it is extinction coefficients at 266 nm are rather low. certainly worthwhile to explore the possibilities of Interestingly, these differences are not reflected in LIF detection. the LIF results: LODs in the low ng/ml range were The LIF set-up (Fig. 1) was optimized under achieved for all compounds studied (Table 1). flow-driven conditions (pressure5100 mbar), using a Probably, measurements were performed under satu26 solution of 5?10 M1,5-naphthalenedisulfonic acid. ration conditions: despite the low molar absorpThe background level was checked after every tivities at 266 nm, maximum excitation is reached adjustment to avoid collection of stray light from the due to the high peak power. When using the BG-14 capillary core and wall. This was done by adjusting cut-off filter, the LODs for 1-hydroxypyrene and the pressure to zero: then, in the absence of flow, the 3-hydroxybenzo[a]pyrene were better than with the fluorescence signal almost instantaneously collapses UG-1 bandpass filter because the UG-1 filter, while due to photodecomposition. conveniently excluding ambient light, is hardly transFirstly, the capillary holder position was optimized parent above 400 nm. Compared with the results so that the focal point on the capillary was exactly in obtained with conventional fluorescence detection, the center of the reflective objective. Then, the the LODs were significantly better for 2-hydroxyfocusing into the capillary liquid core was optimized naphthalene and 1-hydroxychrysene. by adjusting the position of the focusing lens. Subsequently, the distance of the reflective objective to the illuminated point of the capillary was optimized to achieve optimum fluorescence collection and scattered light rejection. Finally, the position and angles of the mirror projecting the fluorescence image on the photomultiplier tube were varied for optimum fluorescence intensity. Proper alignment of the LIF set-up could easily be maintained over several months, without any re-alignment. Also, excellent long-term output power stability was observed, as was apparent from the stable and flat baselines obtained. It should be realized that the peak power of the laser used here is still as high as 1.8 kW, despite its modest average power of 5.4 mW. It was observed Fig. 2. Electropherogram of a mixture of five hydroxy-PAHs (20 that such a high peak power damages the capillary ng/ml) using 266-nm LIF detection. Emission filter: BG-14. Other within a few hours. Also, air bubbles were formed conditions: see text. EOF: electroosmotic flow; 15ascorbic acid; within the capillary, probably due to heating of the 253-hydroxybenzo[a]pyrene; 352-hydroxynaphthalene; 451-hybuffer. Fortunately, when using a neutral density droxyphenanthrene; 551-hydroxychrysene; 651-hydroxypyrene. 296 J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 To summarize this section on standard solutions, it can be concluded that the 266-nm LIF detection method investigated here is generally applicable for analytes exhibiting native fluorescence, despite the fixed excitation wavelength. The detection limits are slightly better than those obtained with conventional fluorescence detection. A quantitative comparison with other LIF detection methods is difficult to make since differences in detection configuration play a role. In an MEKC separation of hydroxy-PAHs, Smith et al. [14] achieved LODs in the 0.08–3 mM range, values distinctly less favorable than the data in Table 1. However, they used a capillary with an inner diameter of 50 mm and made use of fiber optics. In general it can be stated that continuous Fig. 3. Electropherograms of pyrene metabolites isolated from the wave argon ion lasers perform better in terms of hepatopancreas of (a) Oniscus asellus and (b) Porcellio scaber achievable detection limits. In preliminary experiusing 266-nm LIF detection. Emission filter: BG-14. Buffer: 30 ments on naphthalene sulfonates, using 266-nm LIF mMborate, 60 mMSDS, 10 mMg-CD, pH 9.0. 151-Hydroxypyrene sulfate; 25unknown pyrene metabolite; 351-hydroxydetection, LODs in the low nMrange were obtained pyrene; 45pyrene; 551-hydroxypyrene glucoside. [21], which is about one order of magnitude higher than obtainable using excitation by a continuous wave argon ion laser [1,2]. Such a difference in mentioned above, recorded with LIF detection using performance should be expected in view of the the BG-14 cut-off filter, are shown in Fig. 3. pulsed nature of the present laser (duty cycle: 3?Theoretical plate numbers are typically 241 000, 26 10 ). 93 000, 137 000, 97 000 and 91 000 for peak numbers 1–5. The conjugated PAH metabolites could be 3.2. Pyrene metabolites in isopods well separated by CE using a 30 mMborate, 60 mM SDS, 10 mMg-CD, pH 9.0 buffer. However, pyrene In a previous study it was concluded that the and 1-hydroxypyrene glucoside showed partial overpyrene metabolite concentrations in isopods, spelap (R: 0.40) with this buffer composition. Their s cifically, Oniscus asellus and Porcellio scaber, can resolution required 12.5 mMg-CD in the buffer. be used as a biomarker for PAH exposure in the Unfortunately, under these conditions the 1-hydroxyterrestrial environment [5]. Monitoring the conjupyrene sulfate peak overlapped with a compound gated pyrene metabolites directly, instead of 1-hypresent in the sample matrix. Nevertheless, when droxypyrene after enzymatic hydrolysis, can be using both buffers systems all target analytes can be advantageous because the analyte detectability is monitored. 3–5-times higher in that case [9,10,13]. Moreover, detection of the conjugated PAH metabolites instead 3.3. 1 -Hydroxypyrene in fish bile of the hydroxy-PAHs can give important information regarding phase II metabolic pathways. Here, the As a second application, two fish bile samples applicability of CE–LIF was tested for the detection were analyzed, bile from plaice (Pleuronectes platesof pyrene, 1-hydroxypyrene and the three conjugated sa) exposed to crude oil, and from flounder (Platichpyrene metabolites, 1-hydroxypyrene sulfate, 1-hythys flesus) exposed to contaminated sediment. These droxypyrene glucoside and an unknown phase II samples had been distributed as part of an interpyrene metabolite. Since the conjugates are not laboratory study [17]. The methods used in this study commercially available, quantification was not posby other laboratories were LC with conventional sible. fluorescence detection and gas chromatography– Electropherograms of the two isopod samples mass spectrometry (GC–MS). J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 297 In this application the quantitative aspects of CE with conventional fluorescence and LIF detection, were tested, the target analyte being 1-hydroxypyrene. In view of the data assembled in Table 1, for this particular metabolite conventional fluorescence detection will probably be the method of choice: compared to 266-nm LIF it performs equally in terms of LODs but it provides additional excitation selectivity. The fish bile samples (50 ml) were weighed and hydrolysed enzymatically for 2 h at 378C after the addition of 20 ml of enzyme (bglucuronidase/aryl sulfatase) and 180 ml of water. After incubation, 1000 ml of absolute ethanol was added and the sample was centrifuged for 5 min at 7000 rpm; next, the supernatant was taken for analysis. As was already indicated above, it was not surprising that analyzing the fish bile samples by CE with 266-nm LIF detection and using the BG-14 cut-off filter, gave rise to a large number of peaks in the electropherograms, especially with the oil-exposed Fig. 4. Electropherograms of fish bile samples after enzymatic hydrolysis using 266-nm LIF detection. Emission filter: GG-13. fish bile. The interfering peaks inhibited reliable Other conditions: see text. (a) Oil-exposed plaice bile, (b) sedirecognition of most of the hydroxy-PAHs discussed ment-exposed flounder bile. The 1-hydroxypyrene peaks are in Section 3.1 by means of spiking. Only 1-hydroxyindicated by arrows. pyrene, which was found to be present at a much higher concentration than the other hydroxy-PAHs, could be unambiguously identified. In order to sented in Table 2. It is interesting to compare the improve the selectivity of LIF detection, a GG-13 results with those obtained in the interlaboratory cut-off filter, which removes all radiation below ca. study [17]. For the sediment-exposed flounder bile 395 nm, was used instead. This caused a substantial the concentrations as determined independently by reduction of the number of peaks for the sedimentseven laboratories were found to be in the 3.061.9 exposed fish bile, but not for the oil-exposed fish bile mg/g range (n57, 95% confidence interval), while (Fig. 4). for the oil-exposed plaice bile the range was 1.260.4 It was expected that the use of conventional (n56). The results presented in Table 2 are fully in fluorescence detection would further improve the line with these data. The small differences between selectivity, since the excitation and emission maxima the LIF and CF data might be systematic, but the of 1-hydroxypyrene are at rather high wavelengths range of 1-hydroxypyrene concentrations determined (cf. Table 1). Indeed, the electropherograms now by the other laboratories is to broad for a definite became much cleaner (Fig. 5), with the elecconclusion. tropherogram of the flounder bile exhibiting only one large peak. Quantification was performed by using the exter4. Conclusions nal standard method with calibration curves in the range from 16 to 100 ng/ml (four data points in Compared to other available deep UV laser systriplicate). Before injection the oiland sedimenttems the quadrupled Nd–YAG laser used in this exposed fish bile samples were diluted with ethanol– study has several advantages. It is compact, inexpenwater (80:20, v/v) to give a total dilution of 503sive and rugged (expected operating lifetime: 10 000 and 1003, respectively. Relevant results are preh). Next, the alignment of the LIF set-up is easily 298 J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 filter is used to reduce the incident power. The last characteristic implies that the 266-nm CE–LIF detector can be generally used for analytes exhibiting native fluorescence; even when the molar absorptivity at 266 nm is poor. This is illustrated by the LODs obtained for the five hydroxy-PAHs studied: similar LODs (2–6 ng/ml) are obtained for all of them although three of them have low absorptivities at 266 nm. Of course, if real samples have to be analyzed, the inherent lack in excitation selectivity can be disadvantageous. This is clearly illustrated by the electropherograms of the oil-exposed fish bile. Nevertheless, the 1-hydroxypyrene concentrations determined in both bile samples using 266-nm LIF detection are comparable to those obtained with conventional fluorescence, and within the range of concentrations found by other laboratories using LC– fluorescence detection. Fig. 5. Electropherograms of fish bile samples after enzymatic Finally, it is expected that 266-nm LIF will have hydrolysis using conventional fluorescence monitoring with l ex perspective for detection in CE, especially when 347 nm and l 387 nm. Other conditions: see text. (a) Oilem attention is focused on molecules like tyrosine, exposed plaice bile, (b) sediment-exposed flounder bile. The tryptophan and phenylalanine, because the 266-nm 1-hydroxypyrene peaks are indicated by arrows. laser line is close to the fluorescence excitation maxima of these amino acids. maintained over several months and the laser shows excellent long-term output power stability providing stable baselines and reliable detection. Finally, its output power is high enough to saturate the absorpAcknowledgements tion transitions even when a 6.3% Tneutral density The authors wish to thank the Dutch Foundation Table 2 for the Advancement of Science (NOW) for financial 1-Hydroxypyrene concentrations in PAH-exposed fish bile and support and equipment (grant No. 344-006). Also, calibration plot characteristics, using CE with LIF and conventiona the technical assistance of Mr. J. Buijs is much al fluorescence (CF) detection appreciated. Concentration 1-hydroxypyrene (mg/g) LIF CF Plaice bile (oil-exposed) 1.8 (0.07) 1.5 (0.04) References 1.7 (n51) 1.3 (n52) [1] S.J. Kok, I.C.K. Isberg, C. Gooijer, U.A.Th. Brinkman, N.H. Flounder bile (sediment-exposed) 4.5 (0.07) 4.0 (0.28) Velthorst, Anal. Chim. Acta 360 (1998) 109. 4.1 (n52) 3.8 (n52) [2] S.J. Kok, G.P. Hoornweg, T. de Ridder, U.A.Th. Brinkman, N.H.Velthorst, C. Gooijer, J. Chromatogr. A 806 (1998) 355. 2b R0.997 0.993 [3] E. Aas, J. Beyer, A. Goksøyr, Mar. Environ. Res. 46 (1998) a For each sample hydrolysis and further sample treatment was 225. ´ carried out in duplicate; unless stated otherwise, subsequent [4] E. Escartın, C. Porte, Environ. Sci. Technol. 33 (1999) 406. quantification was based on the average of three injections.Values [5] G.J. Stroomberg, J.A. De Knecht, F. Ariese, C.A.M. Van in parentheses: standard deviation (repeatability of three injecGestel, N.H. Velthorst, Environ. Toxicol. Chem. 18 (1999) tions). 2217. b Four data points in triplicate. J . Kuijt et al . /J . Chromatogr . A 907 (2001) 291 – 299 299 [6] M.M. Krahn, D.G. Burrows,W.G. MacLeod Jr., D.C. Malins, [14] C.J. Smith, J. Grainger, D.G. Patterson Jr., J. Chromatogr. A Arch. Environ. Contam. Toxicol. 16 (1987) 511. 803 (1998) 241. ¨ [7] F. Ariese, Ph.D. Thesis,Vrije Universiteit, Amsterdam, 1993. [15] U. Krismann, W. Kleibohmer, J. Chromatogr. A 774 (1997) [8] F. Ariese, M. Verkaik, G.P. Hoornweg, R.J. Van de Nesse, 193. S.R. Jukema-Leenstra, J.W. Hofstraat, C. Gooijer, N.H. [16] X. Xu, R.J. Hurtubise, J. Chromatogr. A 829 (1999) 289. Velthorst, J. Anal. Toxicol. 18 (1994) 195. [17] F. Ariese, J. Beyer, D.Wells, Preparation and Certification of ˇ´ [9] R. Singh, M. Tucek, K. Maxa, J. Tenglerova, E.H. Weyand, Fish Bile Reference Materials For Exposure Monitoring of Carcinogenesis 16 (1995) 2909. PAHs in the Aquatic Environment, Vrije Universiteit, Insti- [10] P.T. Strickland, D. Kang, E.D. Bowman, A. Fitzwilliam, T.E. tute for Environmental Studies, Amsterdam, 2000, Report Downing, N. Rothman, J.D. Groopman, A. Weston, CarcinoW-00/07. genesis 15 (1994) 483. [18] Y.F. Yik, C.P. Ong, S.B. Khoo, H.K. Lee, S.F.Y. Li, J. [11] K.S. Boos, J. Lintelmann, A. Kettrup, J. Chromatogr. 600 Chromatogr. 589 (1992) 333. (1992) 189. [19] H. Nishi, T. Fukuyama, S. Terabe, J. Chromatogr. 553 [12] F.J. Jongeneelen, R.B.M. Anzion, P.T. Henderson, J. Chro- (1991) 503. matogr. 413 (1987) 227. [20] S. Terabe, Y. Miyashita, Y. Ishihama, O. Shibata, J. Chroma- [13] C.K. Lee, S.H. Cho, J.W. Kang, S.J. Lee, Y.S. Ju, J. Sung, togr. 636 (1993) 47. P.T. Strickland, D. Kang, Toxicol. Lett. 108 (1999) 209. [21] J. Kuijt, unpublished results.