Anal. Chem. 1990, 62, 2051-2053
heights and concentration of nucleotides was examined over a range of 30-500 pM and found to be linear for all bases. This suggests that the present method may be utilized for estimation of absolute RNA amounts in a tissue extract. T o demonstrate this, we recorded the peak heights of nucleotides versus mass of RNA prior to base hydrolysis. Figure 5 shows the resulting plot for yeast RNA. By calculation of the response of the instrument to the quantity of a nucleotide standard injected, the amount of RNA in a sample can be determined. Furthermore, if the ratio of the four bases in a specific RNA source is known, detection of any peak is sufficient to determine the amount of RNA. Therefore, for situations in which speed is of prime importance, measurement of the uracil peak will permit such an estimation in less than 2.5 min. This method is clearly applicable to samples containing a single RNA species as well as those containing mixtures of different RNAs, as would result from bulk RNA extraction from tissue.
DISCUSSION The clearest advantage CZE maintains over other means of nucleotide separation is the required sample quantity. In our experiments, injection of less than 20 pg of each ribonucleotide provided electropherograms with a signal-to-noise ratio of 3:l. In comparison to HPLC and isotachophoresis the amount of sample necessary per run is less by approximately lOO-fold, and in comparison to paper electrophoresis, by many orders of magnitude. The method of microcolumn liquid chromatography, however, allows a comparable detection limit of ribonucleotides, as demonstrated by Banks and Novotny (12). Nucleotide resolution in CZE is as good or better than that for each of the previously demonstrated methods. In the experiments done without CTAB, all product species of RNA base-hydrosylate were resolved. HPLC and ion-exchange chromatography also effect such a separation, but isotachophoresis has been shown to distinguish only the 2’ from the 3‘ isomer for CMP, and paper electrophoresis apparently has been unable to separate any of the isomers. Even for our more rapid and reproducible experiments utilizing CTAB, resolution rivals that seen in many HPLC and ion-exchange separations and is superior to that in isotachophoresis and paper electrophoresis. Short separation time is a third attribute of CZE. Complete separation of ribonucleotides is possible in 5 min, provided a CTAB-conditioned capillary is available, and no capillary regeneration time is needed between separations. Isotachophoretic separation demands more than 20 min. HPLC requires approximately 20 min for a separation and another 20 min between runs to equilibrate the column. Paper electrophoresis and ion-exchange chromatography traditionally have taken several hours or more. The use of electropherogram peak heights in quantitation requires either that the migration rate of the standard is exactly equal to the rates for the samples or, if velocities are different, that diffusion of bands is negligible. In this study, migration rates did show minor variations, but diffusion of nucleotides is low for the time scale of our separations. The
205 1
effective diffusion coefficient for 5’-AMP, for example, has been estimated a t 1.6 X cmz/s (11). In summary, the combined advantages of CZE analysis of ribonucleotides from base-hydrolyzed RNA, namely the amount of sample required, the resolution, and the time of analysis, indicate this method to be highly efficacious. In this study, no exceptional effort was made in optimizing the procedure. Also, enhanced accuracy and precision in quantitation are possible through analysis of corrected peak areas. With the prospects of further improvement and also of automation, we believe that CZE may become the method of choice for quantitating base composition of RNA and for measuring RNA amount in tissue. The feasibility of coupling CZE to a radioisotope detection system (13),either on-line or off (14), suggests the potential to monitor incorporation of radiolabels in purine and pyrimidine metabolism.
ACKNOWLEDGMENT We wish to thank Dean Appling and the members of his lab (University of Texas, Austin) for their advice and intestines. Registry No. 5’-AMP, 61-19-8; 5’-CMP, 63-37-6; 2’-CMP, 85-94-9; 3’-CMP, 84-52-6; 2’-AMP, 130-49-4; 3’-AMP, 84-21-9; 5’-GMP,85-32-5;3’-GMP, 117-68-0; 2’-GMP, 130-50-7;5’-UMP, 58-97-9; 3’-UMP, 84-53-7; 2’-UMP, 131-83-9; CTAB, 57-09-0. LITERATURE CITED (1) Davidson, J. N.; Smellie, R. M. S. Biochem. J . 1952, 52, 594-599. (2) Wade, H. E. Biochem. J . 1980, 77, 534-542. (3) Kreamer, W.; Zorich, N.; Liu, D. S. H.; Richardson, A. Exp. Gerontol. 1979, 14, 27-36. (4) Analytical and Preparative Isotachophoresis; Holloway, C. J., Ed.; Walter de Gruyter: BerlinINew York, 1984. (5) Tsuda, T.; Takagi, K.; Watanabe, T.; Satake, T. HRC CC,J. H@h Resoluf. Chromatogr. Chromatogr. Commun. 1988, 1 1 , 721-723. (6) Dolnik, V.; Liu, J.; Banks, J. F., Jr.; Novotny, M. V.; Bocek, P. J . ChromafOQr. 1989, 480, 321-330. (7) Liu, J.; Banks, J. F., Jr.; Novotny, M. J. Microcolumn Sep. 1989, 1 , 136-141. (8) Terabe, S.; Ishikawa, K.; Utsuka, K.; Tsuchiya, A.; Ando, T. 26th International Liquid Chromatography Symposium; Kyoto, Japan, Jan 25-26, 1983. (9) Tsuda, T. HRC CC,J . High Resolut. Chromatogr. Chromatogr. Commun. 1987, IO, 622-624. (10) Kiso, Y. Topics Surrounding Zone Eiecfrophoresis, Ionic Processes Through Matrices; Nankodo: Tokyo, 1971; pp 81-87. (11) Huang, X.; Coleman, W. F.; Zare, R. N. J . Chromatogr. 1989, 480, 95-110. (12) Banks, J. F., Jr.; Novotny, M. V. J . Chromatogr. 1989, 475, 12-21. (13) Pentoney, S. L., Jr.; Zare, R. N. Anal. Chem. 1989, 61, 1642-1647. (14) Huang, X.; Zare, R. N. Anal. Chem. 1990, 62. 443-446.
’
Present address: Genomyx, 460 Point San Bruno Bhrd., South San Francisco, CA 94080.
Xiaohua Huang’ Jason B. Shear Richard N. Zare* Department of Chemistry Stanford University Stanford, California 94305 RECEIVEDfor review March 5,1990. Accepted May 31,1990. Support for this work by Beckman Instruments, Inc., is gratefully acknowledged.
Labeling Reaction for Time-Resolved Luminescence Detection in Liquid Chromatography Sir: In high-performance liquid chromatography (HPLC), chemical derivatization (or labeling) of analytes to obtain
products with better detection characteristics is a very well-known procedure to improve detection limits. Due to
0003-2700/90/0362-2051$02.50/00 1990 American Chemical Society
2052
ANALYTICAL CHEMISTRY, VOL. 62, NO. 18, SEPTEMBER 15, 1990
COOH
Q-"
COOH
@OH
Figure 1. Derivatization reaction of thiols with 4-MSA
its inherent sensitivity, in fluorescence detection impressive results have been achieved and a wide variety of labeling reactions has been investigated. Unfortunately, in the analysis of real samples the chromatographic performance is not always sufficient to detect the labeled analytes separately from other peaks. Hence, the detection is frequently hindered by fluorescence background originating from fluorescent contaminants, an excess of labeling reagent and/or fluorescent side products. Here we present an approach to enhance the selectivity of detection by making use of labels with a long luminescence decay time (i.e. longer than 0.1 ms) and applying time resolution. With a pulsed Xe lamp and a gated photomultiplier the luminescence of these labels can easily be discriminated from short-lived background fluorescence and scattered light. Furthermore, since long-lived luminescence in aqueous solutions at room temperature is quite exceptional, time-resolved luminescence detection has an inherent selectivity. Besides several organic molecules (the most well-known being biacetyl), a few lanthanide ions show long-lived luminescence in liquid solutions at room temperature ( I ) . In aqueous solutions the decay times of (for instance) Eu(II1) and Tb(II1) are respectively 0.1 and 0.4 ms. Compared to the organic compounds, the lanthanide ions have the advantage that oxygen removal is not required to observe long-lived luminescence (2). The lanthanide ions Eu(II1) and Tb(II1) have been successfully applied as luminophoric labels in fluoroimmunoassay techniques (3, 4 ) . The lanthanide ions have low inherent absorptivities. When complexated with certain ligands, their luminescence intensities can be increased by sensitized excitation ( 5 , 6). To achieve efficient sensitizing, two conditions have to be fulfilled: the ligand must form a complex with the lanthanide(II1) ion and there must be a good match between a donating electronic (triplet) level of the ligand and the accepting level of the lanthanide(II1) ion. If the complexes are too weak, sensitizing is only possible if the solvent is thoroughly deoxygenated since otherwise nonradiative decay of the excited ligand dominates (7). Sensitizing of Tb(II1) or Eu(II1) luminescence as a detection method in batch and in HPLC has been described in the literature (8-13). In the present paper the detection via lanthanide luminescence is described for thiol-containing analytes. To this end the analyte is derivatized to a product able to form a complex with Tb(II1) and to sensitize its luminescence. Use was made of a commercially available derivatization reagent for fluorescence spectroscopy, i.e. 4-maleimidylsalicylic acid (denoted as 4-MSA). The salicylate group has sensitizing properties for Tb(II1) luminescence ( 4 , 9, 13, 14) and the maleimide group reacts with thiols, as schematically depicted in Figure 1;the reaction may be followed by hydrolytic opening of the imide ring of the adduct (15-17). Thus, by using 4MSA, thiol-containing analytes can be converted into sensitizing compounds for Tb(II1) luminescence and, after addition of Tb(III), time-resolved detection of these products will be
possible. It is shown for some thiols that (compared to fluorescence detection) by this method higher selectivity and sensitivity can be obtained.
EXPERIMENTAL SECTION Chemicals. Tb(C1)3.6H20was obtained from Aldrich (Milwaukee, WI) and 4-maleimidylsalicylicacid (4-MSA)from Molecular Probes (Eugene, OR). Glutathione was purchased from Boehringer Mannheim (Mannheim,FRG), @-mercaptoethanolwas from Janssen (Beerse, Belgium), thioglycolic acid from Merck (Darmstadt, FRG), and L-cysteine and Trizma base were from Sigma (St. Louis, MO). HPLC-grade acetonitrile was obtained from Baker (Deventer, The Netherlands). Stock solutions of the thiols were prepared in twice distilled water acidified with hydrochloric acid (pH 3.0). A stock solution of 4-MSA (2 x M) was prepared in ethanol. Instrumentation. Batch experiments were carried out with a Perkin-Elmer (Beaconsfield, U.K.) MPF-44 fluorescence spectrometer, supplied with a continuous XBO 150-Wxenon lamp and Hamamatsu type R777-01-HA photomultipliers. The HPLC system consisted of a Gilson (Villiers-le-Bel,France) 302 HPLC pump equipped with a Gilson 802c manometric module, a Valco six-port injection valve equipped with a 20-wL loop, a stainless-steel column (170 X 3.1 mm id.) packed with RoSil C18HL 5 pm (RSL, Eke, Belgium), a Kratos (Ramsay,NY) URS 051 postcolumn unit (containing two pumps), and a Perkin-Elmer LS-2 filter fluorometer. In addition to a UGll filter, a chemical filter was placed in the excitation beam: a homemade cell (path length 1.5 cm) with quartz windows, filled with an aqueous solution of 1 X M K2Cr04and 1 X M NaOH. This filter combination has a window from 300 to 340 nm. Fluorescence of derivatized thiols was detected at 410 nm, Tb(II1) emission at 545 nm. For time-resolved luminescence detection, a delay time of 0.1 ms and a gating time of 2.0 ms were used. RESULTS AND DISCUSSION Batch Experiments. First the derivatization of thiols with 4-maleimidylsalicylicacid was studied in batch by fluorescence spectroscopy. In this preliminary investigation it was not attempted to reach optimum reaction conditions. A mixture of 1 x M thiol (p-mercaptoethanol, thioglycolic acid, L-cysteine, or glutathione) and 2.5 X M 4-MSA was prepared in 5 x M aqueous Tris buffer (pH 7.0). After a reaction time of 2 h a t room temperature, the fluorescence spectra were recorded. Maximum excitation of the products was observed at 303-305 nm and maximum emission at 405-410 nm. The spectral maxima of derivatized L-cysteine shifted to longer wavelengths after standing for more than about 3 h; such a shift was not observed for the other thiols under investigation. Like other fluorescence derivatization reagents containing maleimide groups ( 1 5 ) ,4-MSA itself is almost nonfluorescent. Subsequently a Tb(II1) solution was added to the fluorescent reaction mixture to give a final concentration of 2.5 X lo-* M Tb(II1). This caused a decrease in fluorescence intensity and, as shown in Figure 2, the characteristic narrowbanded Tb(II1) emission became clearly visible at a maximum excitation wavelength of 322 nm. To observe sensitized Tb(111) luminescence, a pH as high as 11-12 as described for sensitizing by 5-sulfosalicylate or salicylate (9, 14) is not required; instead a pH of about 7 suffices. It is emphasized that unreacted 4-MSA does not have sensitizing properties. From the absorption spectra of 4-MSA with and without Tb(II1) in Tris buffer at pH 7.0, it can be derived that complexation does occur, but obviously Tb(II1) sensitizing is not effective. Thus, an excess of derivatization reagent does not perturb the detection. HPLC Experiments. The derivatized thiol (L-cysteine or glutathione) was injected into the mobile phase consisting of water/acetonitrile 70/30 (v/v) at a low pH (Le. 2.7, HCl added) to prevent proton dissociation. The column effluent was brought to pH 7.0 by postcolumn addition of 5 X IO-* M Tris
ANALYTICAL CHEMISTRY, VOL. 62, NO. 18, SEPTEMBER 15, 1990
600
350
350
2053
250
Wavelength (nm) F W s 2. Fluorescence (-)
and sensitized Tb( I I I) luminescence (-- -) excitation and emission spectra of the 4MSA adduct of glutathione in Tris buffer pH 7.0 (uncorrected spectra). buffer (pH 7.5) to achieve good complexation with Tb(II1); Tb(II1) was added in a second addition line (2 X M TbC13 in water). Buffer and Tb(II1) solutions were added separately to prevent hydrolysis of Tb(II1) in the postcolumn solution. Flow rates for both postcolumn solutions were 0.25 mL/min, while the flow rate of the HPLC mobile phase was 0.50 mL/min. Due to the broad-banded excitation and the high Tb(II1) concentration, some Tb(II1) background emission .at 545 nm was observed as a result of direct excitation. Obviously, the derivatized thiols can be detected by both timeresolved Tb(II1) luminescence and by normal fluorescence detection: when the latter was applied no Tb(II1) solution was added. Thus a direct comparison could be made. The limit of detection obtained by sensitized Tb(II1) luminescence detection for both derivatized L-cysteine and glutathione was 1.5 X lo-' M (SIN = 3) in standard solutions, corresponding to respectively 0.4 and 0.9 ng on column. These detection limits were a factor of about 5 more favorable than those found by fluorescence detection. For L-cysteine a linear calibration curve was measured ( r = 0.9997, n = lo), varying the concentrations from 2.0 X lo-' to 1.6 X lod M (all reaction mixtures contained 2 X M 4-MSA). The repeatability of the signal of a mixture of 2 X lo4 M L-cysteine and 2 X M 4-MSA was 2% (relative standard deviation, n = 6). As an example of the analysis of a complex matrix, in Figure 3 chromatograms of derivatized L-cysteine in 10 times diluted urine (which was filtered over a millipore filter and a disposable CI8cartridge) obtained by fluorescence and time-resolved Tb(II1) luminescence detection are shown. The difference in the selectivity of the detection methods is clearly demonstrated. Whereas the ~-cysteine-4-MSAadduct can be discriminated very well from coeluting peaks by time-resolved sensitized Tb(II1) luminescence detection, this is impossible in the fluorescence detection mode.
CONCLUSION The experiments described above indicate that lanthanide labeling of analytes enabling long-lived time-resolved luminescence detection provides a good sensitivity and an enhancement of selectivity so that complex matrices can be more
Figure 3. Chromatograms of L-cysteine (1 X lod M) derivatlzed with 4-MSA in 10-fold diluted urlne recorded by Ruorescencedetection (right) and time-resolved Tb(1I I) luminescence detection (left).
simply analyzed by HPLC. Extension of this approach to other functional groups would be worthwhile. Of course, in general some sample pretreatment may remain necessary, depending on the problem at hand. Investigations directed on optimization of reaction conditions are currently under way.
LITERATURE CITED (1) Gooijer, C.; Baumann. R. A.; Velthorst, N. H. Prog. Anal. Spectrosc. 1987, IO, 573-599. (2) Kropp, J. L.; Windsor, M. W. J. Chem. phvs. 1965, 42, 1599-1608. (3) Hemmila. I.; Dakubu. S.;Mukkala, V. M.; Siitari, H.; Lhgren, T. Anal. Chlm. Acta 1984, 45, 335-343. (4) Bailey, M. P.; Rocks. B. F.; Riley, C. Analyst 1984, 709, 1449-1450. (5) Dawson, W. R.; Kropp. J. L.; Windsor, M. W. J. Chem. Phm. 1966, 45, 2410-2418. (6) Schreurs, M.; Somsen, G. W.; Gooijer, C.; Velthorst, N. H.; Frei, R. W. J . ChrOmetOgr. 1989. 482, 351-359. (7) Heiier, A.; Wasserman, E. J. Chem. Wys. 1965, 42, 949-955. (8) Hirschy, L. M.; Dose, E. V.; Winefwdner, J. D. Anal. CMm. Acta 1983, 747, 31 1-316. (9) Bailey, M. P.;Rocks, 6. F.; Riley. C. Anal. Chim. Acta 1987, 207,
335-338. (lo) DiBella, E. E.; Weissman, J. B.;Joseph, M. J.; Schultz, J. R.: Wenzel, T. J. J . Chrometogr. 1985, 328. 101-109. (11) Wenzel, T. J.; Coiiette. L. M.; Dahlen, D. T.; Hendrickson. S. M.; YarmalOff, L. W. J. ChfOmatOgr. 1968. 433. 149-158. (12) Wenzel, T. J.; Coiiette, L. M. J. Chrometogr. 1988,436, 299-307. (13) Siepak, J. Anal. Chim. Acta 1989, 278, 143-149. (14)Charles, R. G.; Riedel, E. P. J . Inorg. Nucl. Chem. 1966, 28, 527-536. (15) Wu, C.-W.; Yarbrough, L. R.; Wu, F. Y.H. Blochsmistry 1976, 75, 2863-2868. (16) KAgedai. 6.; Kaliberg, M. J. Chromatogr. 1982, 229. 409-415. (17) Nakashima. K.; Umekawa, C.; Yoshlda, H.; Nakatsuji, S.; Akiyama, S. J . Chromatogr. 1987, 414, 11-17.
Marry Schreurs Cees Gooijer* Ne1 H. Velthorst Department of General and Analytical Chemistry Free University De Boelelaan 1083, 1081 HV Amsterdam The Netherlands
RECEIVED for review April 5, 1990. Accepted June 11, 1990.
