Direct Coupling of High-Performance Thin-Layer Chromatography

Apr 13, 2009 - Cyanobacteria are pathogenic prokaryotes and known for producing a high variety of cyclic hepatotoxic peptides in fresh and brackish wa...
0 downloads 7 Views 1MB Size
Download PDF
Anal. Chem. 2009, 81, 3858–3866

Direct Coupling of High-Performance Thin-Layer Chromatography with UV Spectroscopy and IR-MALDI Orthogonal TOF MS for the Analysis of Cyanobacterial Toxins Iris Meisen,*,† Ute Distler,† Johannes Mu¨thing,† Stefan Berkenkamp,‡ Klaus Dreisewerd,§ Werner Mathys,† Helge Karch,† and Michael Mormann§ Institute for Hygiene, University of Mu¨nster, Robert-Koch-Strasse 41, D-48149 Mu¨nster, Germany, Sequenom GmbH, Mendelssohnstrasse 15 d, 22761 Hamburg, Germany, and Institute of Medical Physics and Biophysics, University of Mu¨nster, Robert-Koch-Strasse 31, D-48149 Mu¨nster, Germany Cyanobacteria are pathogenic prokaryotes and known for producing a high variety of cyclic hepatotoxic peptides in fresh and brackish water. Prominent members of these toxins are microcystin LR (MC LR) and nodularin (Nod), which are under suspicion to cause cancer. Various analytical methods have been reported for the detection of these cyclopeptides, and these are mainly based on liquid chromatography combined with mass spectrometric techniques. Here, we introduce a new approach based on the direct coupling of high-performance thin-layer chromatography (HPTLC) with infrared matrix-assisted laser desorption/ionization orthogonal time-of-flight mass spectrometry (IR-MALDI-o-TOF MS) using the liquid matrix glycerol. The analysis of the cyclopeptides involves the application of three complementary methods: (i) HPTLC separation of MC LR and Nod, (ii) their detection and quantification by UV spectroscopy at λ ) 232 nm, and (iii) direct identification of separated analytes on the HPTLC plate by IR-MALDI-o-TOF MS. Calibration curves exhibited a linear relationship of amount of analyte applied for HPTLC and UV absorption (R2 > 0.99). The limits of detection were 5 ng for UV spectroscopy and 3 ng for mass spectrometric analysis of individual peptides. This novel protocol greatly improves the sensitive determination of toxins from pathogenic cyanobacteria in complex water samples. It was successfully applied to the detection and quantification of MC LR and Nod in a spiked, processed environmental water sample. Cyanobacterial blooms are a growing problem in waters around the world. The most frequently found cyanotoxins are cyclic peptide toxins of the microcystin (MC) and nodularin (Nod) * To whom correspondence should be addressed. Phone: +49-251-8353250. Fax: +49-251-8355341. E-mail: [email protected]. † Institute for Hygiene. ‡ Sequenom GmbH. § Institute of Medical Physics and Biophysics.

3858

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

families.1 Their toxicity is due to a marked inhibition of protein phosphatases 1 and 2A2 which causes mainly functional and structural disturbances of the liver.3-5 In addition to being hepatotoxic, Nod is also a direct carcinogen in rat liver.6 Structurally, MCs are monocyclic heptapeptides containing five nonproteinogenic amino acids. The general chemical structure is cyclo(DAla1-X2-D-MeAsp3-Z4-Adda5-D-Glu6-Mdha7-) in which X and Z are variable L-amino acids (Adda, 3-amino-9-methoxy-2,6,8-trimethyl10-phenyldeca-4,6-dienoic acid; Mdha, N-methyldehydroalanine). Due to amino acid substitution in positions 2 and 4, giving rise to the naming system, and structural variations such as methylation and demethylation, more than 70 congeners have been identified to date.7,8 The species with leucine (L) and arginine (R) at positions 2 and 4, respectively, is known as microcystin LR (MC LR). Figure 1A shows the structure of MC LR which is the best characterized among all cyanotoxins since it is the one most commonly found. Moreover, it also belongs to the most potent toxins of the group.9,10 In recent studies, it has been shown that MC LR has tumor-promoting activity in the rat liver.11,12 Since cyanobacteria producing MC LR are frequently found in fresh waters including lakes and water reservoirs that (1) Sivonen, K.; Jones, G. Cyanobacterial toxins. In Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring, and Management; Chorus, I., Bartram, J., Eds.; E & F Spon: London, 1999; pp 41-111. (2) Mackintosh, C.; Beattie, K. A.; Klumpp, S.; Cohen, P.; Codd, G. A. FEBS Lett. 1990, 264, 187–192. (3) Yoshizawa, S.; Matsushima, R.; Watanabe, M. F.; Harada, K.-I.; Ichihara, A.; Carmichael, W. W.; Fujiiki, H. J. Cancer Res. Clin. Oncol. 1990, 116, 609–614. (4) Honkanen, R. E.; Zwiller, J.; Moore, R. E.; Daily, S. L.; Khatra, B. S.; Dukelow, M.; Boynton, A. L. J. Biol. Chem. 1990, 265, 19401–19404. (5) Codd, G. A.; Morrison, L. F.; Metcalf, J. S. Toxicol. Appl. Pharmacol. 2005, 203, 264–272. (6) Ohta, T.; Sueoka, E.; Iida, N.; Komori, A.; Suganuma, R.; Nishiwaki, R.; Tatematsu, M.; Kim, S.-J.; Carmichael, W. W.; Fujiki, H. Cancer Res. 1994, 54, 6402–6406. (7) Welker, M.; Brunke, M.; Preussel, K.; Lippert, I.; von Do¨hren, H. Microbiology 2004, 150, 1785–1796. (8) Svrcek, C.; Smith, D. W. J. Environ. Eng. Sci. 2004, 3, 155–185. (9) Carmichael, W. W. J. Appl. Bacteriol. 1992, 72, 445–459. (10) Fawell, J.; Hart, J.; James, H.; Parr, W. Water Supply 1993, 11, 109–121. (11) MacKintosh, R. W.; Dalby, K. N.; Campbell, D. G.; Cohen, P. T. W.; Cohen, P.; MacKintosh, C. FEBS Lett. 1995, 371, 236–240. (12) Runnegar, M.; Berndt, N.; Kong, S. M.; Lee, E. Y. C.; Zhang, L. Biochem. Biophys. Res. Commun. 1995, 216, 162–169. 10.1021/ac900217q CCC: $40.75  2009 American Chemical Society Published on Web 04/13/2009

and associated with the toxicity of these molecules.9 Major alteration of the Adda amino acid such as a change of the conjugated diene geometry from 6E to 6Z renders Nod and MCs nontoxic.16,17 Various techniques for the detection, structural elucidation, and quantification of MC LR have been reported in the literature. These include nonchromatographic methods like the mouse bioassay, the protein phosphatase inhibition assay, and the enzyme-linked immunosorbent assay using antibodies raised against specific microcystin congeners and congener-independent against Adda.18-23 Unfortunately, when analyzing identical samples the results often exhibit poor agreement between the methods.24 Taking advantage of UV absorption of the conjugated diene in the Adda residue, various liquid chromatography (LC) separations combined with UV detection at 238 nm have been developed.25,26 Currently, LC coupled to mass spectrometry is frequently used as a sensitive method for the characterization of cyanobacterial toxins. In particular, electrospray ionization (ESI) and matrixassisted laser desorption/ionization mass spectrometry (MALDI MS) allow the screening for cyanotoxins.27 In addition, full structural characterization can be achieved by use of tandem mass spectrometry (MS/MS) experiments.28-34 MALDI MS is a rapid and robust technique with very low sample consumption and gives rise to mainly singly charged ions. Thus, MALDI MS is increasingly utilized for the detection of MCs.35-37 In a recent study a method was presented simplifying congener identification by removing cation adducts and enhancing

Figure 1. Structures of cyanobacteria toxins: (A) microcystin LR; (B) nodularin. D-Ala, D-alanine; L-Leu, L-leucine; D-MeAsp, D-methylaspartic acid; L-Arg, L-arginine; Adda, 3-amino-9-methoxy-2,6,8trimethyl-10-phenyldeca-4,6-dienoic acid; D-Glu, D-glutamic acid; Mdha, N-methyldehydroalanine; Mdhb, 2-(methylamino)-2-dehydrobutyric acid. Ring positions are numbered 1-7.

supply water treatment facilities the World Health Organization (WHO) has proposed a provisional upper limit for MC LR of 1 µg/L in drinking water13,14 to protect consumers from adverse effects. Nodularin is structurally similar to MC LR. It is a cyclic pentapeptide with the chemical structure cyclo(D-MeAsp1-L-Arg2Adda3-D-Glu4-Mdhb5-) where Mdhb corresponds to 2-(methylamino)-2-dehydrobutyric acid (Figure 1B). Only a few naturally occurring variations of Nod, mainly produced by methylation/ demethylation, have been described.15 The key structural element of both Nod and MCs is the β-amino acid Adda which is unique to cyanobacterial hepatotoxins (13) WHO. Guidelines for Drinking-Water Quality. Addendum to Volume 2. Health Criteria and Other Supporting Information; World Health Organization: Geneva, Switzerland, 1998. (14) Falconer, I.; Bartram, J.; Chorus, I.; Kuiper-Goodman, T.; Utkilen, H.; Burch, M.; Codd, G. A. Safe levels and safe practices. In Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring, and Management; Chorus, I., Bartram, J., Eds.; E & F Spon: London, 1999; pp 155-177.

(15) Mazur-Marzec, H.; Meriluoto, J.; Plin´ski, M.; Szafranek, J. Rapid Commun. Mass Spectrom. 2006, 20, 2023–2032. (16) Harada, K. I.; Ogawa, K.; Matsuura, K.; Murata, H.; Suzuki, M.; Watanabe, M. F.; Itezono, Y.; Nakayama, N. Chem. Res. Toxicol. 1990, 3, 473–481. (17) Rinehart, K. L.; Namikoshi, M.; Choi, B. W. J. Appl. Phycol. 1994, 6, 159– 176. (18) Chen, T.; Wang, Q.; Cui, J.; Yang, W.; Shi, Q.; Hua, Z.; Ji, J.; Shen, P. Mol. Cell. Proteomics 2005, 4, 958–974. (19) Masango, M.; Myburgh, J.; Botha, C.; Labuschagne, L.; Naicker, D. Water Res. 2008, 42, 3241–3248. (20) Chu, F. S.; Huang, X.; Wei, R. D.; Carmichael, W. W. Appl. Environ. Microbiol. 1989, 55, 1928–1933. (21) Fischer, W. J.; Garthwaite, I.; Miles, C. O.; Ross, K. M.; Aggen, J. B.; Chamberlin, A. R.; Towers, N. R.; Dietrich, D. R. Environ. Sci. Technol. 2001, 35, 4849–4856. (22) Zeck, A.; Weller, M. G.; Bursill, D.; Niessner, R. Analyst 2001, 126, 2002– 2007. (23) An, J.; Carmichael, W. W. Toxicon 1994, 32, 1495–1507. (24) Mountfort, D. O.; Holland, P.; Sprosen, J. Toxicon 2005, 45, 199–206. (25) Harada, K. I.; Matsuura, K.; Suzuki, M.; Oka, H.; Watanabe, M. F.; Oishi, S.; Dahlem, A. M.; Beasley, V. R.; Carmichael, W. W. J. Chromatogr. 1988, 448, 275–283. (26) Rivasseau, C.; Martins, S.; Hennion, M. C. J. Chromatogr., A 1998, 799, 155–169. (27) Maizels, M.; Budde, W. L. Anal. Chem. 2004, 76, 1342–1351. (28) Zhang, L.; Ping, X.; Yang, Z. Talanta 2004, 62, 193–200. (29) Diehnelt, C. W.; Dugan, N. R.; Peterman, S. M.; Budde, W. L. Anal. Chem. 2006, 78, 501–512. (30) Kubwabo, C.; Vais, N.; Benoit, F. M. Rapid Commun. Mass Spectrom. 2005, 19, 597–604. (31) Frias, H. V.; Mendes, M. A.; Cardozo, K. H. M.; Carvalho, V. M.; Tomazela, D.; Colepicolo, P.; Pinto, E. Biochem. Biophys. Res. Commun. 2006, 344, 741–746. (32) Li, C.-M.; Chu, R. Y.-Y.; Hsieh, D. P. H. J. Mass Spectrom. 2006, 41, 169– 174. (33) Spoof, L.; Vesterkvist, P.; Lindholm, T.; Meriluoto, J. J. Chromatogr., A 2003, 1020, 105–119. (34) Allis, O.; Dauphard, J.; Hamilton, B.; Shuilleabhain, A. N.; Lehane, M.; James, K. J.; Furey, A. Anal. Chem. 2007, 79, 3436–3447. (35) Neumann, U.; Campos, V.; Cantarero, S.; Urrutia, H.; Heinze, R.; Weckesser, J.; Erhard, M. Syst. Appl. Microbiol. 2000, 23, 191–197.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

3859

detection of protonated MC molecular species through the addition of zinc sulfate.38 The quantitative analysis of cyanobacterial toxins by MALDI MS using internal standards has also been reported.39 High-performance thin-layer chromatography (HPTLC) is a valuable tool for the separation of various analyte species. It is an inexpensive technique with which many samples can be separated in parallel and stored for further investigation. In the past, TLC in combination with UV spectroscopy was applied to the screening analysis of cyanobacterial bloom samples.40 Recently, postchromatographic derivatization of purified cyanobacterial hepatotoxins to colored or fluorescent products has been investigated.41 However, bacterial toxins can only be identified by comparing UV chromatograms and Rf values with those of available standard toxins. The direct coupling of HPTLC to MS greatly increases the information content by adding the molecular weight. Various methods for combining both techniques have been described for a large variety of analytes, and most reports have involved UVMALDI MS with solid matrixes so far.42-45 The inherent disadvantage of this technique is in the wetting of the HPTLC plate with an extraction solvent in order to dissolve the analyte molecules for their incorporation into matrix crystals. This may also lead to an unwanted diffusion and loss of spatial resolution. In addition, only a thin layer (less than 100 nm) is ablated per single laser pulse in UV-MALDI MS; thus, only small numbers of matrix-incorporated analyte molecules on the surface of the plate are desorbed. As has been shown recently, these drawbacks can be overcome by using infrared MALDI orthogonal time-of-flight mass spectrometry (IR-MALDI-o-TOF MS) with glycerol as the liquid matrix.46 Analyte diffusion in the matrix is negligible, and the penetration depth of the Er:YAG IR laser used is about a few micrometers, thus enabling the ablation of a significantly higher amount of material. Furthermore, decoupling of the desorption/ ionization process from the mass determination provides a high mass resolution and mass accuracy.47,48 Up to now, this direct HPTLC-MS coupling technique has been successfully applied (36) Welker, M.; Fastner, J.; Erhard, M.; von Do ¨hren, H. Environ. Toxicol. 2002, 17, 367–374. (37) Saker, M. L.; Fastner, J.; Dittmann, E.; Christiansen, G.; Vasconcelos, V. M. J. Appl. Microbiol. 2005, 99, 749–757. (38) Howard, K. L.; Boyer, G. L. Rapid Commun. Mass Spectrom. 2007, 21, 699–706. (39) Howard, K. L.; Boyer, G. L. Anal. Chem. 2007, 79, 5980–5986. (40) Pelander, A.; Ojanpera¨, I.; Sivonen, K.; Himberg, K.; Waris, M.; Niinivaara, K.; Vuori, E. Water Res. 1996, 30, 1464–1470. (41) Pelander, A.; Ojanpera¨, I.; Lahti, K.; Niinivaara, K.; Vuori, E. Water Res. 2000, 34, 2643–2652. (42) Fuchs, B.; Su ¨ ss, R.; Nimptsch, A.; Schiller, J. Chromatographia [Online early access]. DOI: 10.1365/s10337-008-0661-z. Published Online June, 17 2008. http://www.springerlink.com/content/635815x3m45k7g71/. (43) Guittard, J.; Hronowski, X. P. L.; Costello, C. E. Rapid Commun. Mass Spectrom. 1999, 13, 1838–1849. (44) Fuchs, B.; Schiller, J.; Su ¨ ss, R.; Schu ¨ renberg, M.; Suckau, D. Anal. Bioanal. Chem. 2007, 389, 827–834. (45) Gusev, A. I. Fresenius’ J. Anal. Chem. 2000, 366, 691–700. (46) Dreisewerd, K.; Mu ¨ thing, J.; Rohlfing, A.; Meisen, I.; Vukelic´, Zˇ.; PeterKatalinic´, J.; Hillenkamp, F.; Berkenkamp, S. Anal. Chem. 2005, 77, 4098– 4107. (47) Rohlfing, A.; Mu ¨ thing, J.; Pohlentz, G.; Distler, U.; Peter-Katalinic´, J.; Berkenkamp, S.; Dreisewerd, K. Anal. Chem. 2007, 79, 5793–5808. (48) Ivleva, V. B.; Sapp, L. M.; O’Connor, P. B.; Costello, C. E. J. Am. Soc. Mass Spectrom. 2005, 16, 1552–1560.

3860

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

to HPTLC-separated gangliosides,46 oligosaccharides,49 phospholipids,47 and tumor-associated glycosphingolipids.50 Here, we describe the direct coupling of HPTLC with IRMALDI-o-TOF MS for the analysis of cyclic peptide toxins MC LR and Nod. For detection and quantification of the analytes, three complementary techniques were applied comprising (i) HPTLC separation of MC LR and Nod, (ii) their detection and quantification by UV spectroscopy at λ ) 232 nm, and (iii) the direct identification of the separated analytes on the plate by IR-MALDIo-TOF MS. EXPERIMENTAL SECTION Chemicals. MC LR from Microcystis aeruginosa and Nod from Nodularia spumigena were obtained from Alexis Biochemicals (Axxora Deutschland GmbH, Lo¨rrach, Germany) and used without further purification. Glycerol (p.a.), p-anisaldehyde, and the calibration peptides bradykinin fragment 1-7 and melittin were obtained from Sigma-Aldrich (Deisenhofen, Germany) and used as purchased. Ethylacetate (p.a.) and 1-propanol (p.a.) were from Merck (Darmstadt, Germany), as were methanol and chloroform, which were distilled prior to use. Reference Solutions. Stock solutions of MC LR and Nod were prepared in pure distilled methanol and adjusted to a concentration of 250 µg/mL. Reference solutions were immediately aliquoted, dried under a stream of nitrogen, and stored at -20 °C. Prior to use the samples were freshly dissolved in pure methanol and diluted to final concentrations according to the demand of the experiment. High-PerformanceThin-LayerChromatography.Thesamples were applied to HPTLC plates with an automatic TLC applicator (Linomat IV, CAMAG, Muttenz, Switzerland). MC LR/Nod mixtures were separated on glass-backed silica gel 60 precoated HPTLC plates (size 10 cm × 10 cm, thickness 0.2 mm, no. 1.05633.001; Merck). The plates were developed for 45 min with 1-propanol/ethylacetate/water (30/50/20, v/v/v), containing acetic acid at a final concentration of 5%. For staining of peptides, plates were cut and bands were visualized by spraying a solution of p-anisaldehyde. The spraying solution was prepared by addition of 8 mL of concentrated sulfuric acid and 500 µL of p-anisaldehyde to a mixture of 85 mL of methanol and 10 mL of glacial acetic acid.41 After treatment with the staining reagent the plate was placed onto a heating plate (105 °C) until appearance of purpleblue peptide bands. The unstained part of the plates was further investigated using UV spectroscopy and IR-MALDI-o-TOF MS. If not immediately analyzed developed HPTLC plates were stored at -20 °C. No evidence for oxidation by atmospheric oxygen of the analytes used in this study was observed. UV-Vis Spectroscopy of MC LR/Nod Mixtures Separated by HPTLC. After chromatographic separation of toxin mixtures with various dilutions (100-900, 10-90, and 1-9 ng of each peptide), unstained peptides were detected and quantified with a CD60 scanner (Desaga, Heidelberg, Germany, software ProQuantR, version 1.06.000) prior to mass spectrometric analysis. Chromatographic lanes were scanned in remission mode at (49) Dreisewerd, K.; Ko ¨lbl, S.; Peter-Katalinic´, J.; Berkenkamp, S.; Pohlentz, G. J. Am. Soc. Mass Spectrom. 2006, 17, 139–150. (50) Distler, U.; Hu ¨ lsewig, M.; Souady, J.; Dreisewerd, K.; Haier, J.; Senninger, N.; Friedrich, A. W.; Karch, H.; Hillenkamp, F.; Berkenkamp, S.; PeterKatalinic´, J.; Mu ¨ thing, J. Anal. Chem. 2008, 80, 1835–1846.

wavelengths of 238 and 232 nm with a light beam slit width of 0.02 mm × 6 mm and 0.2 mm × 6 mm, respectively. The analysis of p-anisaldehyde-stained bands was performed at a wavelength of 604 nm. For acquisition of a UV-vis absorption spectrum of a single separated unstained band, x- and y-positions were fixed at the center of the band and absorption was measured from λ ) 190 to 900 nm. MALDI Preparation Protocol. Direct HPTLC-MALDI MS analyses were performed from unstained peptide bands. The positions of the analytes, separated chromatographically, were determined using UV chromatography and encircled with a pencil. The HPTLC plate was then cut into pieces with a maximum size of 4 cm × 3 cm to fit into the MALDI sample plate. For matrix application, a series of drops of pure glycerol, 0.2-0.4 µL, were applied next to each other on the marked positions. To avoid or at least minimize interference, care was taken not to wet two neighboring HPTLC bands simultaneously. The glycerol drops were rapidly soaked up by the silica gel and produced spots ∼2-3 mm in diameter. The wetted silica gel areas were readily recognized by their whitish appearance. Parts of the HPTLC plate were mounted by double-sided adhesive pads on a custom-made MALDI sample plate with a milled out central region of 32 mm × 45 mm with a depth of 2 mm. Infrared Matrix-Assisted Laser Desorption/Ionization Orthogonal Time-of-Flight Mass Spectrometry. The modified o-TOF mass spectrometer (MDS Sciex, Concord, ON, Canada), which routinely provides a mass resolution of ∼10 000 and a mass accuracy of ∼20 ppm, has been described in detail previously.46,51 An IR-MALDI port with an Er:YAG infrared laser (Bioscope, BiOptics Laser Systems AG, Berlin, Germany) emitting pulses of ∼100 ns duration at a wavelength of 2.94 µm and a repetition rate of 2 Hz served for desorption/ionization. The focal spot size was ∼250 µm in diameter. Samples were observed with a CCD camera with a resolution of 10 µm. Gas-phase ions are generated in an elevated-pressure ion source filled with nitrogen (p ∼ 1 mbar). No evidence for electric charging effects of the glass-backed HPTLC plate was found due to the high number density of the buffer gas molecules inside the ion source. The low m/z cutoff of the quadrupole was set to 100. All spectra were recorded in the positive ion mode with the TOF-pusher voltage set to 10 kV. The o-TOF instrument was calibrated with a two-point calibration using the protonated species of bradykinin fragment 1-7 and melittin. For acquisition of the IR-MALDI mass spectra, ∼350 single laser pulses were usually applied at several positions adjacent to each other on one band. Mass spectra were processed and evaluated using the MoverZ3 software (version 2002.2.13.0, Genomic Solutions, Ann Arbor, MI). Fortification and Extraction of Water Samples. To 500 mL of an untreated environmental water sample taken from an eutrophic local urban lake (Aasee, Mu¨nster, Germany) 40 µL of a solution containing 0.01 µg/µL MC LR/Nod was added. This results in MC LR and Nod at final concentrations of 0.8 µg/L. The spiked sample and 500 mL of the corresponding nonspiked water sample were passed through a 0.45 µm membrane filter. Water samples were subsequently cleaned and enriched using solid-phase extraction (SPE). SPE columns containing 6 mL of (51) Loboda, A. V.; Ackloo, S.; Chernushevich, I. V. Rapid Commun. Mass Spectrom. 2003, 17, 2508–2516.

octadecyl silica (C18) and an apparatus to support the columns and provide vacuum assist were obtained from J. T. Baker (Mallinckrodt Baker, Griesheim, Germany). SPE columns were preconditioned with 15 mL of methanol followed by 15 mL of water. After water samples were applied with vacuum assistance, the columns were washed with 10 mL of water and bound peptides eluted with 10 mL of methanol. The eluents were evaporated to dryness under a stream of nitrogen, redissolved in 2 mL of methanol/water (2/1, v/v), and extractedtwicewith2mLofchloroform.Theaqueous-methanolic extracts were dried under a stream of nitrogen and finally dissolved in 60 µL of methanol for HPTLC and UV experiments. RESULTS AND DISCUSSION HPTLC of MC LR/Nod Mixtures. An anisaldehyde stain of an HPTLC-separated MC LR/Nod mixture is shown in Figure 2A, lane b. Prior to staining, the part of the HPTLC plate containing lane a was cut for further investigations using UV spectroscopy. The slightly different polarity of the hepatotoxins enabled them to be separated sufficiently. Using the chromatographic conditions applied Rf values for Nod were approximately 0.33 and 0.37 for the less polar MC LR. The successful spatial chromatographic separation of the two peptides is a prerequisite for the examination using UV spectroscopy described below. UV Spectroscopy of MC LR/Nod Mixtures Separated by HPTLC. The exact lateral chromatographic positions of Nod and MC LR were determined by use of UV spectroscopy. A wavelength of λ ) 238 nm was used first since this has been reported as the absorption maximum of MCs.25,52 In agreement with the achieved HPTLC separation, peaks detected for Nod and MC LR are baseline-separated, thus enabling the precise determination of peak areas (data not shown). Subsequently, the wavelengthdependent absorption maxima of Nod and MC LR are inspected by recording UV-vis spectra over the wavelength range from 190 to 900 nm for fixed x- and y-positions corresponding to the center of the bands. The UV-vis spectra obtained for both peptides are shown in Figure 2B together with a reference spectrum recorded from a blank position of the HPTLC plate. As expected, both analytes exhibit strong absorption in the near-UV. Closer examination of the UV-vis spectra (cf. inset of Figure 2B) reveals a slight shift of the absorption maxima to the far-UV for Nod at 231 nm and MC LR at 233 nm when compared to literature values. Consequently, determination of peak areas was repeated by scanning the chromatographic lane at a wavelength of 232 nm. The resulting UV chromatogram is depicted in Figure 2C. Comparison of peak areas determined at 238 versus 232 nm showed increased values for UV absorption measured at λ ) 232 nm (10.5% for Nod and 6.5% for MC LR), thus enabling a more sensitive detection of the peptides. To probe absorption of HPTLC-separated anisaldehyde-stained bands, lane b of Figure 2A was scanned at a wavelength of 604 nm. Photometric analysis yielded considerably reduced peak areas compared to UV absorption of the unstained bands at λ ) 232 nm (data not shown). IR-MALDI-o-TOF Mass Spectrometry of MC LR/Nod Mixtures Separated by HPTLC. The HPTLC-separated, nonstained toxin bands were marked according to the exact positions (52) Lanaras, T.; Cook, C. M. Sci. Total Environ. 1994, 142, 163–169.

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

3861

Figure 3. Direct HPTLC-IR-MALDI-o-TOF mass spectra recorded in the positive ion mode from the separated, unstained Nod/MC LR mix. The areas analyzed are marked with dotted rectangles in the insets (lane a). The corresponding anisaldehyde stain is displayed in lane b. (A) Mass spectrum of 1 µg of MC LR obtained from lane a at 39.5 mm (upper band). (B) Mass spectrum of 1 µg Nod obtained from lane a at 36.1 mm (lower band). For assignment of the numbers to the corresponding ionic species detected and their m/z values refer to Table 1. Peaks derived from analyte-matrix adducts are marked by open squares, and background peaks, which originate either from the glycerol matrix or the silica gel of the HPTLC plate, are assigned by asterisks. For details concerning m/z values and composition of the species detected, cf. Figure S-1 and Table S-1 of the Supporting Information.

Figure 2. HPTLC and UV-vis spectroscopy of a mixture of Nod and MC LR. (A) For chromatographic separation 20 µL containing 1 µg of each hepatotoxin was applied. Lane a: after development of the HPTLC plate the lane remained unstained for further characterization by UV spectroscopy (panels B and C) and mass spectrometry (cf. Figure 3). Lane b: after development this lane was cut and stained with anisaldehyde. (B) UV-vis spectra of HPTLC-separated, unstained MC LR and Nod recorded from 190 to 900 nm. The inset shows an enlarged region of the spectra from 195 to 270 nm. Data were acquired with fixed xand y-positions from the center of the bands. Dashed line: UV-vis spectrum of Nod obtained from lane a. Solid line: UV-vis spectrum of MC LR obtained from lane a. Dotted line: UV-vis spectrum recorded from a blank position of the HPTLC plate. (C) UV chromatogram recorded from the HPTLC plate at 232 nm from the separated, unstained Nod/MC LR mix (see panel A, lane a). For scanning the full HPTLC lane, the x-position was fixed in the middle of lane a, whereas the lateral position was scanned from 10 to 80 mm.

determined by UV spectroscopy, treated with glycerol, and subsequently analyzed directly on the HPTLC plate using IRMALDI MS. The full mass spectra obtained from 1 µg of peptides 3862

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

applied are depicted in Figure 3. Gas-phase ions derived from MC LR (Figure 3A) are detected as protonated and sodiated species of high abundance. The singly sodiated form at m/z 1017.53 ([M + Na]+, peak 2) gives rise to the base peak of the spectrum, which is accompanied by signals corresponding to the protonated ([M + H]+, peak 1), doubly ([M - H + 2Na]+, peak 3), and triply sodiated ([M - 2H + 3Na]+, peak 4) analyte ions similar to the results obtained from standard UV-MALDI preparations.38 The 10-fold magnified region contains peaks assigned with numbers 5-10 representing gas-phase dimers. Protonated species with low abundance [2M + H]+ (peak 5) and dimers of MC LR with 1 to as many as 5 sodium ions attached are present (peaks 6-10). The mass spectrum of Nod displays a similar pattern of ionic species (Figure 3B). Protonated peptides at m/z 825.44 (peak 1) give rise to the base peak of the spectrum which is accompanied by signals corresponding to the singly, doubly, and triply sodiated species (peaks 2-4, respectively). In addition, low abundant dimers are observed as protonated species and as adducts with one to four sodium ions attached (peaks 5-9). Furthermore, ions arising from the formation of analyte-matrix adducts, e.g., [M + Na + glycerol]+, [M - H + 2Na + glycerol]+, [M - 2H + 3Na + glycerol]+, occur with both toxins. Peaks marked with asterisks represent background ions derived from the silica gel of the HPTLC plate

Table 1. Ionic Species and Measured and Calculated m/z Values of the Ions Detected from an HPTLC-Separated Nod/MC LR Mix by Direct IR-MALDI-o-TOF Mass Spectrometrya

a

peak

ionic species

MC LR m/zmeasd

MC LR m/zcalcd

Nod m/zmeasd

Nod m/zcalcd

1 2 3 4 5 6 7 8 9 10

+

995.55 1017.53 1039.51 1061.49 1990.08 2012.07 2034.04 2056.01 2077.99 2099.97

995.56 1017.54 1039.52 1061.50 1990.11 2012.09 2034.07 2056.05 2078.03 2100.02

825.44 847.42 869.40 891.38 1649.86 1671.85 1693.82 1715.80 1737.79 n.d.b

825.45 847.43 869.42 891.40 1649.89 1671.88 1693.86 1715.84 1737.82 1759.80

[M + H] [M + Na]+ [M - H + 2Na]+ [M - 2H + 3Na]+ [2M + H]+ [2M + Na]+ [2M - H + 2Na]+ [2M - 2H + 3Na]+ [2M - 3H + 4Na]+ [2M - 4H + 5Na]+

Amounts of 1 µg of each, cf. Figure 3. b Not detected.

and the matrix itself. Gaseous ions derived from Nod and MC LR and their corresponding m/z values are summarized in Table 1. A priori the here described method should be also applicable under the use of the more frequently employed UV-MALDI conditions. Direct TLC-UV-MALDI MS of peptides has been demonstrated previously. However, the inherent problem of analyte spreading upon matrix application leads to loss of spatial resolution and decrease of local analyte concentration.42 This can impede the analysis especially of species with close Rf values as in the case of MC LR and Nod investigated in the present study. Furthermore, a significantly reduced sensitivity has been reported for UV-MALDI MS of HPTLC-separated oligosaccharides as a result of a high background level over the whole mass range as compared to IR-MALDI MS.49 It is noteworthy that in the case of both analyte species, no fragment ions were observed under the mass spectrometric conditions used, and this contrasts to the results obtained previously using UV-MALDI MS where fragmentation of the Adda side chain has been observed.38 This finding points to the mild desorption/ionization conditions of the IR-MALDI process employed in this study. Furthermore, the formation of dimeric gasphase ions points to the moderate excess of internal energy deposited in the analyte ions.53 In summary, both toxins can be detected by IR-MALDI-o-TOF MS directly from the HPTLC plate as high abundant intact ionized molecules. The characteristic ionization pattern of the cyclic peptide toxins Nod and MC LR are beneficial for their identification in more complex mixtures such as environmental water samples. Quantification of HPTLC-Separated MC LR/Nod Mixtures and HPTLC-IR-MALDI MS of Individual Toxins. To explore the potential of the combined methods for identification and quantification of Nod and MC LR, reference solutions with various concentrations were separated by HPTLC, detected by UV spectroscopy, and subjected to IR-MALDI analysis. Measured peak areas obtained from UV chromatograms were plotted as a function of the amount of peptide. The linear correlations for 100-900 ng of MC LR and Nod are depicted in Figure 4A. All parameters were obtained by linear regression, and intercept and slope values together with the calculated errors, correlation coefficients, standard deviations, and coefficients of determination are listed in Table 2. An excellent linear relationship between amount of sample and peak area was (53) Dreisewerd, K.; Berkenkamp, S.; Leisner, A.; Rohlfing, A.; Menzel, C. Int. J. Mass Spectrom. 2003, 226, 189–209.

obtained for both toxins. The correlation coefficients for linear regression were 0.9988 for MC LR and 0.9982 for Nod, respectively. The HPTLC-separated toxin bands were subsequently analyzed by direct IR-MALDI-o-TOF MS. Both peptides appear as protonated, singly, doubly, and triply sodiated species with a pattern similar to that observed before (see above). The mass spectra reveal high signal-to-noise ratios (S/N) even in the case of those spectra obtained with only 100 ng of MC LR and Nod which are shown as representative examples in Figure 4, parts B and C, respectively. Further inspection of the sensitivity of the combined methods was done using HPTLC separations of 10-fold-diluted sample solutions (10-90 ng). UV absorption could be measured for all analyte dilutions even though the lowest amount of 10 ng appears to be close to the limit of UV detection. Figure 4D shows the plots of the peak areas determined for MC LR and Nod as a function of the amount of sample separated by HPTLC. Analysis of the data points using linear regression showed a perfect relationship between amount of peptide and peak area with correlation coefficients of 0.9982 for MC LR and 0.9981 for Nod, respectively. Additional parameters determined by use of linear regression are listed in Table 2. All mass spectra subsequently acquired from the toxin bands revealed the characteristic ionization pattern already described, and both peptides were detected with high abundance. Even the mass spectra obtained from the lowest amount of sample applied, i.e., just 10 ng of toxins, exhibited a more than sufficient signal-to-noise ratio (S/N ∼ 10) thus enabling the unambiguous detection of both peptides (Figure 4E, MC LR; Figure 4F, Nod). These results clearly indicate that trace quantities of 10 ng which correspond to 1/100 of the proposed provisional upper limit for MC LR in drinking water can be easily detected by UV spectroscopy at 232 nm combined with direct HPTLC-IR-MALDI-o-TOF MS. To determine the limit of detection of the combined methods a further series of diluted toxin solutions (1-9 ng) were investigated. UV spectroscopy revealed absorption for total amounts of sample from 5 to 9 ng. Due to low S/N ratios of UV chromatograms peak areas and peptide amount did not yield a linear correlation (data not shown). However, the exact localization of the chromatographic position of the peptides was achievable, thus facilitating the subsequent mass spectrometric analysis. For evaluation of mass spectra, only those ionic species were taken into account which exhibited the full isotopic pattern and could therefore be unambiguously Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

3863

Figure 4. Linear regression of peak areas determined by UV spectroscopy of HPTLC-separated MC LR and Nod as a function of amount of sample and corresponding HPTLC-IR-MALDI mass spectra. (A and D) Calibration curves obtained by linear regression for 100-900 (A) and 10-90 ng (D) amounts of MC LR and Nod separated by HPTLC. All parameters of the linear fits are listed in Table 2. (B and C) Mass spectra obtained from 100 ng of MC LR (B) and 100 ng of Nod (C). (E and F) Mass spectra obtained from 10 ng of MC LR (E) and 10 ng of Nod (F).

Table 2. Calibration Curve Data Obtained by Linear Regression of Peak Areas Detected by UV Absorption of HPTLC-Separated MC LR and Nod at 232 nm and Corresponding Peptide Amounts Applied on the HPTLC Plate

c

A error Bd error Re SDf R2 g a

MC LR 100-900 nga

Nod 100-900 nga

MC LR 10-90 ngb

Nod 10-90 ngb

29.46389 6.66715 0.64166 0.01185 0.99881 9.17729 0.9976214

32.79083 8.03466 0.65748 0.01428 0.99835 11.05967 0.9967027

0.10472 1.29157 0.99688 0.02295 0.99815 1.77784 0.9963034

1.41389 1.12748 0.86317 0.02004 0.99812 1.55197 0.9962435

cf. Figure 4A. b cf. Figure 4D. c Intercept value. d Slope value. e Correlation coefficient. f Standard deviation of the fit. g Coefficient of determination.

discriminated from the noise and background signals. A detection limit of 3 ng of total sample amount was determined with HPTLC-IR-MALDI MS for MC LR (Figure 5A-C) and Nod (Figure 5D-F). 3864

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

UV Spectroscopy and Direct HPTLC-IR-MALDI MS of an Environmental Water Sample Fortified with MC LR/Nod. Finally, the application of the combined methods for the detection of MC LR and Nod isolated from a spiked environ-

Figure 5. Determination of limit of detection for direct IR-MALDI mass spectrometric analysis of HPTLC-separated MC LR and Nod. (A-C) Mass spectra obtained from 4 (A), 3 (B), and 2 ng (C) of MC LR. (D-F) Mass spectra obtained from 4 (D), 3 (E), and 2 ng (F) of Nod.

mental water sample was investigated. Assuming a 100% recovery from the enrichment procedure the amount of sample applied for HPTLC separation was equivalent to 133 ng in both toxins. As a control, a nonspiked 500 mL water sample was processed the same way. The UV chromatograms obtained from the spiked and nonspiked water sample are shown in Figure 6A together with chromatograms obtained from reference solutions. Despite a high background, the chromatogram of the spiked water sample revealed distinct, well-separated peaks for Nod and MC LR. For quantification of the peptides using the peak areas calculated with the software, calibration curves were determined from reference mixtures with amounts of 10, 30, 50, 100, 300, and 500 ng separated by HPTLC. Using the linear fits calculated with correlation coefficients of 0.997 (Nod) and 0.996 (MC LR), amounts of 99.6 ng (Nod) and 98.3 ng (MC LR) were determined for the spiked water sample. This implies recovery rates of 75% (Nod) and 74% (MC LR).

Subsequent HPTLC-MS examination enabled the unambiguous identification of MC LR in the spiked water sample (Figure 6B). The mass spectrum exhibits the characteristic pattern of molecular ions with a high S/N ratio. No interfering background ions are observed. The mass spectrum obtained from the chromatographic position of Nod is shown in Figure 6C. An ionic species with m/z 845.53 gives rise to the base peak of the spectrum, obviously representing an impurity comigrating with Nod under the HPTLC conditions used. However, Nod can be identified from the presence of highly abundant, isotopically resolved protonated as well as doubly and triply sodiated ions. The peak representing singly sodiated Nod molecules is overlapping with the signal for the isobaric [13C2] isotopomer of the unknown comigrating species, thus resulting in a broadened peak at m/z 847.5. These results indicate the potential of the matched techniques for the Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

3865

detection and quantification of MC LR and Nod in environmental water samples.

Figure 6. Analysis of an environmental water sample spiked with MC LR and Nod by UV spectroscopy and direct IR-MALDI-o-TOF MS. Two environmental water samples (500 mL) were either spiked with both 400 ng of MC LR and Nod, or left untreated, and enriched extracts were separated by HPTLC. (A) UV chromatograms recorded from the spiked and nonspiked water samples after HPTLC separation. For quantification, reference solutions containing 30, 50, and 100 ng of each toxin were analyzed. Extract volumes of 20 µL corresponding to one-third of the processed redissolved water samples were applied on the HPTLC plate. (B and C) Direct HPTLC-IR-MALDI mass spectra of MC LR (B) and Nod (C) obtained from the spiked water sample.

3866

Analytical Chemistry, Vol. 81, No. 10, May 15, 2009

CONCLUSION HPTLC coupled to IR-MALDI-o-TOF MS with a glycerol matrix provides an easy means of analyzing the cyclic peptides MC LR and Nod directly from the silica gel HPTLC plates. The exact positions of the HPTLC-separated peptides are determined by UV absorption. Therefore, an IR laser scan over the full chromatographic distance or alternatively running a sample in parallel for visualization is not required. The linear relationship between UV absorption of MC LR and Nod and amount of sample applied on the HPTLC plate allows detection and also quantification of these toxins. The peptides are finally identified by direct IR-MALDI-o-TOF MS according to their typical ionization pattern as protonated and multiply sodiated species. The main advantage of desorption/ ionization by IR photons is the formation of stable, intact ionic analyte species, without fragmentation of the cyanobacterial toxins. A drawback of these combined methods is the missing option for performing MS/MS experiments, which is not yet available on the instrument used. However, HPTLC in conjunction with the sensitivity and high mass accuracy of the o-TOF instrument allows the detection of as little as 3 ng of MC LR and Nod on the HPTLC plate by IR-MALDI MS. The limit of detection for UV spectroscopy was 5 ng of total sample. The established protocol was successfully applied to the detection of MC LR and Nod isolated from a spiked environmental water sample. In conclusion, it could be shown that combining HPTLC with UV spectroscopy and IR-MALDI-o-TOF mass spectrometry directly on the HPTLC plate is an ideal and sensitive tool for the detection and quantification of the cyclic cyanobacteria toxins MC LR and Nod. The protocol represents a robust and cheap alternative to the application of LC-MS techniques and exhibits a high potential for the detection of cyanobacteria toxins in complex water samples with a sensitivity far below the proposed provisional upper limit for MC LR in drinking water of 1 µg/L. ACKNOWLEDGMENT The authors thank Sequenom GmbH (Hamburg, Germany) for providing use of their o-TOF instrument. Ansgar Schu¨tting is kindly acknowledged for providing the water samples. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review January 29, 2009. Accepted March 22, 2009. AC900217Q