Anal. Chem. 2000, 72, 1360-1363
Correspondence
Determination of Biological Toxins Using Capillary Electrokinetic Chromatography with Multiphoton-Excited Fluorescence Jing Wei, Eric Okerberg, Jennifer Dunlap, Cindy Ly, and Jason B. Shear*
Department of Chemistry and Biochemistry and The Institute for Cellular and Molecular Biology, University of Texas, Austin, Texas 78712
We report a highly sensitive and rapid strategy for characterizing biological toxins based on capillary electrokinetic chromatography with multiphoton-excited fluorescence. In this approach, aflatoxins B1, B2, and G1 and the cholera toxin A-subunit are fractionated in ∼80 s in a narrow-bore electrophoretic channel using the negatively charged pseudostationary phase, carboxymethyl-β-cyclodextrin. The aflatoxinsshighly mutagenic multiple-ringed heterocycles produced by Aspergillus fungisare excited at the capillary outlet through the simultaneous absorption of two to three 750-nm photons to yield characteristic blue fluorescence; cholera toxin A-subunit, the catalytic domain of the bacterial protein toxin from Vibrio cholera, is excited through an unidentified multiphoton pathway that apparently includes photochemical transformation of an aromatic residue in the polypeptide. The anionic carboxymethyl-β-cyclodextrin, used to chromatographically resolve the uncharged aflatoxins, enhances emission from these compounds without contributing substantially to the background. Detection limits for these toxins separated in 2.1-µm-i.d. capillaries range from 4.4 zmol (∼2700 molecules) for aflatoxin B2 to 3.4 amol for the cholera toxin A-subunit. Larger (16-µm-i.d.) separation capillaries provide concentration detection limits for aflatoxins in the 0.2-0.4 nM range, severalfold lower than achieved in 2.1-µm capillaries. These results represent an improvement of >104 in mass detectability compared to previously published capillary separations of aflatoxins and demonstrate new possibilities for the analysis of proteins and peptides. Natural molecular toxins are produced by organisms as distinct as bacteria and vertebrates and display a corresponding diversity in both molecular structure and mode of action. In terms of human toxicity, the potency of some biological toxins far exceeds synthetic poisons. Toxins secreted by the bacterium Clostridium botulinum, for example, abolish acetylcholine signaling at the neuromuscular junction by cleaving proteins required for vesicle exocytosis. Because each botulinum toxin molecule can degrade many copies of its target protein, these compounds are lethal at molecular doses ∼100-million-fold lower than sarin, a synthetic 1360 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
cholinesterase inhibitor. Fortunately, human exposure to botulinum toxins is relatively rare, in large part because the Clostridium bacteria that synthesize these toxins are obligate anaerobes. In contrast, a variety of toxin-producing organisms can thrive in the aerobic conditions in which most food supplies are stored. Cholera toxin, synthesized by the bacterium Vibrio cholera, and related enterotoxins produced by Escherichia coli and Salmonella typhimurium, are relatively common food contaminants that cause chronic activation of adenylate cyclase.1 In many instances, consumption of foods contaminated with these bacteria leads to severe dehydration through intestinal fluid loss. The smallmolecule aflatoxins produced by the fungal species Aspergillus flavus and Aspergillus parasiticus also present a continual threat to humans and livestock through ingestion of contaminated corn, peanuts, and other agricultural foods. By potently disrupting normal nucleic acid synthesis and regulation, exposure to aflatoxins can cause a variety of cellular abnormalities, including neoplasms in human liver tissue.2-4 The most lethal isoform, aflatoxin B1 (AFB1), has been shown to form adducts with guanine at the N-7 position of the basesa modification pathway that renders AFB1 among the most genotoxic agents identified.4 To guard against the dangers of aflatoxins, a number of food industries are required to screen products before sale in the United States. Fortuitously, aflatoxins are strongly fluorescent when excited by UV light, and because they often are present in mixtures of four isoforms (B1, B2, G1, G2) with different toxicities, fractionation with a chromatographic or electrophoretic procedure typically is performed before fluorometric analysis.5-7 Corn contaminated with AFB1 can be analyzed using a fluorometricbased HPLC assay with a detection limit of 0.1 ppb.8 Although the capabilities of this method are more than adequate for (1) Pekala, P. H.; Anderson, B. M. in The Pyridine Nucleotide Coenzymes; Everse, J., Anderson, B., You, K.-S., Eds.; Academic Press: New York, 1982; pp 350-355. (2) Forster, P. L.; Eisenstadt, E.; Miller, J. H. Proc. Natl. Acad. Sci. U.S.A. 1983, 80, 2695-2698. (3) McCann, J.; Spingarn, N. E.; Kobori, J.; Ames, B. N. Proc. Natl. Acad. Sci. U.S.A. 1975, 72, 979-983. (4) Wang, J.-S.; Groopman, J. D. Mutat. Res. 1999, 424, 167-181. (5) Diebold, G. J.; Zare, R. N. Science 1977, 196, 1439-1441. (6) Zeringue, H. J.; Shih, B. Y. J. Agric. Food Chem. 1998, 46, 1071-1075. (7) Cole, R. O.; Holland, R. D.; Sepaniak, M. J. Talanta 1992, 39, 1139-1147. (8) Maragos, C. M.; Greer, J. I. J. Agric. Food Chem. 1997, 45, 4337-4341. 10.1021/ac991077c CCC: $19.00
© 2000 American Chemical Society Published on Web 02/15/2000
screening food samples (the USDA action level for human foods is 20 ppb), alternative procedures must be developed to probe volume-limited samples, including single cells that have been infected with relatively few molecular copies of toxin. Investigation of heterogeneities between individual cells (or cell colonies) with respect to toxin uptake, localization, and chemistry would likely provide valuable information concerning factors that influence aflatoxin genotoxicity.4 Unlike standard HPLC and gel electrophoresis techniques, microcolumn separation procedures readily accommodate samples in the picoliter to nanoliter rangesvolumes appropriate for cellular and subcellular sampling. This reduction in scale often yields large improvements in mass detection limits, making it possible to measure zeptomole to attomole quantities of fluorescent peptides, proteins, and neurotransmitters.9-12 Cole et al.,13 and Maragos and Greer8 used capillary electrokinetic chromatography (CEKC) with laser-induced fluorescence (LIF) detection to improve instrumental mass detection limits for aflatoxins to the 0.05-1.0-fmol range for low-nanoliter sample injection sizes. Concentration detection limits were lower in the Maragos and Greer work, which used 75-µmi.d. separation capillaries to achieve a detection limit of ∼10 nM for AFB1 standards. No extensive steps were taken in either the Cole or Maragos studies to optimize fluorescence detection limits. In previous studies, we demonstrated that multiphoton-excited (MPE) fluorescence can be a highly sensitive and versatile strategy for probing small quantities of neurotransmitters and redox cofactors fractionated by capillary electrophoresis (CE).14,15 These biological species can be excited efficiently through simultaneous absorption of two or three near-IR photons when irradiated with focused output from a femtosecond-pulsed titanium: sapphire (Ti:S) laser. For serotonin and several other hydroxyindoles, photoreactions initiated by multiphoton excitation are found to generate products with enhanced fluorescence properties and better detection limits than the parent compounds.10 The low mass detection limits for MPE fluorescence derive from high spatial confinement of the excitation region (probe volumes of ∼1 fL can be achieved by focusing the laser beam with high numerical aperture optics) and efficient rejection of scattered excitation light. Because the excitation photons are lower in energy than MPE fluorescence photons, absorption-based filters with large discrimination ratios can be used without concern for filter autofluorescence. Moreover, near-IR photons that do strike the bialkali photocathode in our detector are converted to counts with very poor efficiency. As a result of these unique characteristics, scattering background in MPE fluorescence measurements can be virtually eliminated without using lossy spatial and narrowband spectral filters. In the present work, we evaluate MPE fluorescence and photochemistry as strategies for probing small quantities of aflatoxins and the A-subunit of the cholera toxin protein (CTA) (9) Lillard, S. J.; Yeung, E. S.; McCloskey, M. A. Proc. SPIE Int. Soc. Opt. Eng. 1997, 2980, 133-144. (10) Gostkowski, M. L.; Wei, J.; Shear, J. B. Anal. Biochem. 1998, 260, 244250. (11) Timperman, A. T.; Oldenburg, K. E.; Sweedler, J. V. Anal. Chem. 1995, 67, 3421-3426. (12) Lee, T. T.; Yeung, E. S. J. Chromatogr. 1992, 595, 319-325. (13) Cole, R. O.; Holland, R. D.; Sepaniak, M. J. Talanta 1992, 39, 1139-1147 (14) Gostkowski, M. L.; McDoniel, J. B.; Wei, J.; Curey, T.; Shear, J. B. J. Am. Chem. Soc. 1998, 120, 18-22. (15) Gostkowski, M. L.; Shear, J. B. J. Am. Chem. Soc. 1998, 120, 12966-12967.
fractionated by CEKC in micrometer-diameter channels. Negatively charged carboxymethyl-β-cyclodextrin is found to be an effective differential complexation agent for the aflatoxins, providing baseline resolution of the B1, B2, and G1 aflatoxins in ∼80 s and yielding a severalfold enhancement in fluorescence intensity versus cyclodextrin-free buffer systems. Concentration detection limits for the aflatoxins are more than 10-fold lower than achieved in the previous CEKC report;8 moreover, because sample volume requirements are extremely low, the use of ultrasmall separation channels with MPE fluorescence detection provides greater than a 104-fold improvement in mass detection limits. Low-attomole analysis of CTA is achieved apparently through an MPE photochemical transformation analogous to the hydroxyindole reactions.10 Although this process has yet to be characterized in detail, photochemical detection of proteins is shown to offer expanded capabilities for probing multiple toxin types in a single sample. EXPERIMENTAL SECTION Chemicals and Materials. Carboxymethyl-β-cyclodextrin (average of 3 carboxymethyl substitution sites per molecule) and sulfated β-cyclodextrin (average of 14 sulfate substitution sites per molecule) were obtained from Supelco (Bellefonte, PA), HPLCgrade acetonitrile (99.9%) was obtained from EM Science (Gibbstown, NJ), and CTA was purchased from both Sigma (St Louis, MO) and Calbiochem (La Jolla, CA); all other chemicals were obtained exclusively from Sigma. Water was purified using a Barnstead UV water system, and all aqueous solutions were filtered with 0.2-µm pore-size cellulose acetate syringe filters (Valuprep, Ann Arbor, MI) to remove particulates. Aflatoxin stock solutions were prepared at 1.0 mg/mL in acetonitrile, and the CTA stock solution was prepared in 20 mM Tris buffer (pH 7.5). Stock solutions were stored at -20 °C for up to 3 weeks and were diluted to the desired sample concentrations with 20 mM Tris buffer (pH 7.5). Uncoated fused-silica separation capillaries were obtained from Polymicro, Inc. (Phoenix, AZ). Cyclodextrin solutions were prepared and used at concentrations between 2 and 10 mM. CEKC Analysis with MPE Fluorescence. The design and use of the MPE fluorescence system is described in greater detail elsewhere.14 A positive high potential is applied at the injection end of the separation capillary using a (30 kV dc power supply (Spellman CZE1000R), and the high-voltage electrode and buffer/ sample reservoirs are enclosed in an interlocked plastic box during operation for user safety. Detection is performed at the capillary outlet, which terminates in a grounded cuvette containing buffer solution. Multiphoton-excited fluorescence detection of analytes is achieved by focusing 750-nm output from a femtosecond-modelocked Ti:S laser (Coherent Mira 900F) to a nearly diffractionlimited spot at the outlet aperture of the capillary using a 1.3 NA oil immersion microscope objective (Zeiss Fluar, 100×, infinity corrected). In this approach, laser light is focused through a cover slip a short distance (∼10-100 µm) into a modified cuvette that serves as the outlet electrolyte reservoir (held at electrical ground). The separation capillary enters the cuvette through a septum on one side and terminates at the beam focus near the opposing side of the cuvette so that analyte molecules intersect the multiphoton excitation volume before diffusing into bulk solution in the outlet reservoir. Fluorescence is collected through the excitation objective and is reflected from the excitation beam path using a long-pass dichroic mirror (625DCXR, Chroma Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1361
Technology). Residual laser scatter is removed from the light collection pathway using several absorption filters (two Schott Glass BG39 filters and 2 cm of 1 M CuSO4 solution), and fluorescence is detected using a bialkali photomultiplier tube (PMT, model HC125-02, Hammamatsu). Signal from the PMT is digitized by a photon counter (model SR400, Stanford Research Systems, Sunnyvale, CA) which transfers data via a GPIB interface to a Macintosh running a LabView-based data acquisition program (National Instruments). Cuvette measurements of toxin MPE fluorescence are performed by focusing 750-nm output from a Spectra Physics Tsunami Ti:S laser to a strongly diffraction limited spot using the 1.3 NA objective. Safety Considerations. The high-voltage inlet for electrophoretic separations should be isolated from users during operation with an interlocked plastic box. In addition, aflatoxins are extremely genotoxic substances that should not come in contact with the user in any manner. Special care should be taken to avoid exposure to aflatoxin dust and solutions. RESULTS AND DISCUSSION Aflatoxins exhibit one-photon excitation maximums at deep(∼260 nm) and near-UV (∼365 nm) wavelengths and fluorescence maximums between ∼425 and 450 nm.1 The Maragos and Greer procedure8 for determination of aflatoxins with CEKC used an ultraviolet (325 nm) helium-cadmium laser as the LIF source, which provided good spectral isolation of the excitation and emission wavelengths but a poor match to the principal (nearUV) excitation band. As an alternative approach, we investigated the use of a wavelength-tunable mode-locked Ti:S laser to promote MPE fluorescence of aflatoxins. An AFB1 solution was subjected to tightly focused Ti:S laser light between 730 and 770 nm, and substantial emission could be produced at all tested wavelengths. Measurement of signal as a function of laser power (Figure 1) revealed that excitation of AFB1 is achieved primarily through absorption of two 750-nm photons at the lowest powers examined and three photons at higher powers. These results indicate that both moderate-energy (“near-UV”) and high-energy (“deep-UV”) electronic states can be accessed through multiphoton excitation of aflatoxins and that the predominant excitation mode changes with excitation power in a manner consistent with the expected power scaling for the two processes.16 Aflatoxins (Figure 2) are uncharged over a wide pH range, making electrophoretic separation impractical unless a pseudostationary phase is established in the separation buffer. Because micelle solutions formed from deoxycholic acid8 or SDS13 generate prohibitive fluorescence background when subjected to tightly focused Ti:S light, we evaluated anionic β-cyclodextrin (β-CD) derivatives in our system as differential complexation agents for aflatoxins.17 Baseline resolution of AFB1, AFB2, and AFG1 was readily achieved using Tris buffers modified with either carboxymethyl-β-CD or sulfated β-CD that can form host-guest inclusion complexes with hydrophobic (or amphipathic) compounds.18 The use of an anionic β-CD as a differential complexation agent in the separation medium yields aflatoxin-β-CD complexes (16) Shear, J. B. Anal. Chem. 1999, 71, 598A-605A. (17) Janini, G. M.; Muschik, G. M.; Issaq, H. J. Electrophoresis 1996, 17, 15751583. (18) Szejtli, J. Cyclodextrins and Their Inclusion Complexes; Akademiai Kiado: Budapest, 1982.
1362 Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
Figure 1. Aflatoxin B1 multiphoton-excited fluorescence (F) as a function of the time-averaged laser power (P). When saturation and photobleaching are low, the slope of a log(F) vs log(P) plot reveals the number of laser photons that must be absorbed to promote analyte fluorescence. By fitting a line to the 10 lowest laser power points (∼427 mW), excitation is shown to take place principally through absorption of two photons (slope 2.0). At somewhat higher powers (∼35-80 mW), the slope of the data increases to ∼2.7, indicating an increasing role of three-photon excitation; the falloff in slope at the highest powers tested suggests that the available pool of fluorophore molecules may be partially depleted through photobleaching reactions.
Figure 2. Structures of aflatoxins B1, B2, and G1.
that have a net negative charge, making it possible to separate components through a chromatographic mechanism. Fortuitously, incorporation of β-CDs in the separation buffer improves the fluorescence quantum yields of aflatoxins19,20 without substantially increasing MPE fluorescence background. We attempted to measure intrinsic MPE UV fluorescence from CTA using a detection channel optimized for light collection in (19) Francis, O. J., Jr.; Kirschenheuter, G. P.; Ware, G. M.; Carman, A. S.; Kuan, S. S. J. Assoc. Off. Anal. Chem. 1988, 71, 725-728. (20) Cepeda, A.; Franco, C. M.; Fente, C. A.; Va´zquez, B. I.; Rodrı´guez, J. L.; Prognon, P.; Mahuzier, G. J. Chromatogr., A 1996, 721, 69-74.
Figure 3. Electropherogram showing separation of a mixture of 97 nM AFB1, 96 nM AFB2, 92 nM AFG1, and 1.6 µM cholera toxin A-subunit in a 2.1-µm capillary. Separation was performed in a pH 7.5 20 mM Tris,10 mM carboxymethyl-β-cyclodextrin buffer system, with a field strength of ∼735 V/cm (capillary length ∼20.4 cm). Sample injection was accomplished by applying a field of ∼145 V/cm for 5 s, yielding injection volumes for the four components of ∼4.5-8.5 pL. Analytes were excited using ∼185 mW of 750-nm Ti:S laser light focused nearly to the diffraction limit using a 1.3 NA oil microscope objective.
the near-UV range, but standard emission from the sole tryptophan residue was nearly undetectable at the concentrations examined in these studies. Fortunately, multiphoton excitation was found to elicit broad visible emission from CTA solutions, signal that could be used to detect this protein at low-attomole levels. Analysis of a mixture of CTA and aflatoxins using CEKC with MPE fluorescence (Figure 3) highlights one of the key advantages of multiphoton excitation: the capability for analyzing sample components via excitation of distinct photophysical/photochemical processes with the same laser source. Although the specific nature of CTA emission has yet to be determined, experimental observations support the hypothesis that an aromatic residuesmost likely tryptophansis photochemically transformed to a product that emits in the visible region. First, the MPE visible emission spectrum for CTA is centered at ∼530-550 nm, clearly distinct from protein phosphorescence spectra.21 In addition, we have observed previously that hydroxyindoles can undergo photochemical conversion to visible-emitting products when excited with multiple, near-IR photons.10 In recent (unpublished) studies, we have determined that a variety of indolic and phenolic species can produce MPE visible emission, including catecholamines and their trihydroxyindole derivatives,22 aromatic amino acids, peptides, and proteins. The detectability of these species varies considerably, and in general, the best detection limits are achieved for indolic compounds. To minimize the possibility that visible emission from CTA solutions was caused by a small-molecule contaminant, toxin was purchased from two separate sources, and (21) Ross, J. B. A.; Laws, W. R.; Rousslang, K. W.; Wyssbrod, H. R. In Topics in Fluorescence Spectroscopy; Lakowicz, J. R., Ed.; Plenum Press: New York, 1992; Vol. 3, pp 50-52. (22) Yui, Y.; Itokawa, Y.; Kawai, C. Anal. Biochem. 1980, 108, 11-15.
both samples yielded nearly identical electrophoresis peaks in the visible detection channel. In addition, dialysis of CTA solution using a 10 000 molecular weight cutoff membrane did not eliminate this peak, and no visible emission could be detected from CTA solutions analyzed in a conventional steady-state fluorometer. Multiphoton analysis of aflatoxins and CTA separated in 2.1µm-i.d. channels yielded a system reproducibility of ∼5% for analyte peak heights and ∼2% for migration times. Concentration LODs (defined as the signal equivalent of 3 times the rms noise) were 2.4 nM for AFB1, 0.8 nM for AFB2, 2.5 nM for AFG1, and 400 nM for CTA. Electrokinetic injection volumes ranged from ∼4.5 to 8.5 pL for the analytes, yielding mass LODs for these species of 11 zmol, 4.4 zmol, 13 zmol, and 3.4 amol, respectively. These results represent an improvement of >104 in mass detectabilitysand a 10-fold reduction in concentration detection limitsscompared to previous CEKC analysis of aflatoxins using UV-excited fluorescence.8,13 Moreover, the use of larger (16-µmi.d.) capillaries with this MPE system improves concentration LODs for the aflatoxins to 0.2-0.4 nM. No previous microcolumn analyses of CTA could be found for the purpose of comparison; however, the mass LOD for CTA is comparable to that achieved for tryptophan and tryptophan-containing proteins using UVexcited fluorescence after capillary electrophoretic separations.11,12 In summary, we have explored the use of multiphoton-excited fluorescence as an alternative to conventional fluorescence for probing extremely small quantities of toxins fractionated with capillary electrokinetic chromatography. These results demonstrate that MPE fluorescence has the versatility to probe dissimilar toxins through different spectroscopic mechanisms and can offer extremely low mass detection limits for compounds separated in micrometer-diameter channels. Although some of the improvement in mass detectability of aflatoxins can be attributed to the use of extremely small inner diameter capillaries, one generally expects concentration detection limits to suffer as the probe volume is reduced. The fact that concentration detection limits are significantly lower in this work than in previous CEKC reports implies that multiphoton excitation may play an important role in achieving low detection limits. These capabilities should prove valuable in characterizing samples that could have multiple, unknown toxin components and could play an important role in defining the interactions of toxin molecules with cellular materials at new levels of reduction. ACKNOWLEDGMENT This work was supported by grants from the Office of Naval Research (Grant N00014-97-1-0494) and the Searle Scholars Program/Kinship Foundation. We gratefully acknowledge T. Curey for assistance in the acquisition of multiphoton emission spectra.
Received for review September 17, 1999. Accepted January 3, 2000. AC991077C
Analytical Chemistry, Vol. 72, No. 6, March 15, 2000
1363