Anal. Chem. 1996, 68, 3922-3927
Detection of Radionuclides in Capillary Electrophoresis Using a Phosphor-Imaging Detector Scott E. Tracht,† LouAnn Cruz,† Carolyn M. Stobba-Wiley,† and Jonathan V. Sweedler*
Department of Chemistry and Beckman Institute, University of Illinois, Urbana, Illinois 61801
A capillary electrophoresis (CE) postcolumn radionuclide detector has been developed that uses a commercial phosphor-imaging detector and has been optimized for low-energy β emitters. Eluant from the separation capillary is deposited on a membrane. Emission from radioactive analytes on the membrane is integrated using the phosphor-imaging system for 10-72 h. Results from the phosphor-imaging system are converted to conventional electropherograms. Modifications to a prior postcolumn CE deposition system have been accomplished by adding a buffer makeup capillary; this increases the electrolyte pH range and improves reproducibility. The limit of detection (LOD) for 35S-labeled analytes is 0.13 amol (8.7 pM or 0.007 Bq), while the LOD for 32P-labeled analytes is 4.9 zmol (0.33 pM or 0.002 Bq), with a linear range for 35S-Met from 1.5 amol to 1.5 fmol. Radioisotopes have been used as probes in a wide variety of applications for a number of years. Several radionuclides are currently used to detect biochemical processes and products. Isotopes, such as 14C and 3H, are used to label organic molecules to monitor in vivo processes; other examples include 35S, used to label cellular synthesis of peptides and proteins, and 32P, used as a marker in DNA synthesis.1,2 Using radiolabeled compounds to monitor biological processes often leads to a large number of analytes,3 so a natural progression is to combine radionuclide probes with a high-resolution separation method. With theoretical plates often in excess of 100 000, capillary electrophoresis (CE) is capable of separating analytes in complex mixtures.4 For example, a number of researchers have applied CE to the separation of analytes in single cells.5 We describe an instrument which combines capillary electrophoresis with radionuclide detection, providing the capability to detect trace levels of several significant biological probes. Previously, we demonstrated low-energy β- particle detection of radionuclides separated by capillary electrophoresis with no degradation in separation efficiency.6 We used postcolumn † Present addresses: S.E.T., Amylin Pharmaceuticals, San Diego, CA; L.C., Department of Chemistry, Oklahoma State University; C. M. S.-W., Eli Lilly & Co., Indianapolis, IN. (1) Ehmann, W. D.; Robertson, J. D.; Yates, S. W. Anal. Chem. 1994, 66, 229R251R. (2) Slater, R. J. In Molecular Biology & Biotechnology; Meyers, R. A., Ed.; VCH Publishers: New York, 1995; pp 779-784. (3) O’Farrell, P. H. J. Biol. Chem. 1975, 450, 4007-4021. (4) McCormick, R. M. In Handbook of Capillary Electrophoresis; Landers, E. J., Ed.; CRC Press: Boca Raton, FL, 1994; pp 287-322. (5) Jankowski, J. A.; Tracht, S. E.; Sweedler, J. V. Trends Anal. Chem. 1995, 14, 170-176.
3922 Analytical Chemistry, Vol. 68, No. 22, November 15, 1996
detection, where eluant from the capillary is deposited onto a membrane previously impregnated with a solid state scintillator. Light emission from the scintillation process on the membrane is imaged onto a charge-coupled device (CCD) using a series of 35 mm camera lenses. While this system provided a large improvement in limits of detection (LODs), and allowed 35S and 3H detection in CE for the first time, further sensitivity enhancements are desired. In addition, the response across the membrane is nonuniform. More light is collected from samples deposited closer to the membrane center. This caused the membrane center to have a high bias relative to response near the edge. Finally, lower pH buffers produced a luminescent background when deposited on the scintillator membrane, causing an interfering signal and degrading LODs. Despite these issues, this system demonstrated that postcolumn detection offers a significant advantage in CE with radiochemical detection. The LOD for a radionuclide is dependent on the decay rate of the nuclide in question. For a given radionuclide, the longer the observation time, the higher the number of decay events available for detection.7 For example, Pentoney and Zare have described an on-column 32P detector with nearly 100% detector efficiency;8 however, our detection system obtained better LODs, even with significantly lower detector efficiency, because we are able to observe the analyte for longer times.6 Our new method of combining CE with radionuclide detection incorporates a number of significant improvements. Instead of a scintillator/CCD, we now use a phosphor-imaging system. As before, radionuclide sample is first eluted from the CE capillary onto a membrane. In this case, the membrane is not part of the scintillation process; instead, it is placed next to a phosphorimaging plate after the CE separation. Radioisotope detection with the phosphor imager is significantly different from our previous scintillator/CCD system. In the new system, β- particle emission excites a BaFBr:Eu2+ complex in the phosphor-imaging plate to a high-energy state. This state is stable for extraordinarily long times, up to several days. To quantitate the radioactive exposure of the plate, the high-energy state is destabilized by a HeNe laser (632.8 nm) that pumps the stable energy state to a higher unstable level, which then emits a photon at ∼390 nm as it decays back to ground state, as shown in Figure 1. The photon is collected by the instrument using a photomultiplier tube (PMT). This system of collecting and storing the energy of β- particle decay is described as photostimulated luminescence (PSL).9-11 (6) Tracht, S. E.; Toma, V.; Sweedler, J. V. Anal. Chem. 1994, 66, 2382-2389. (7) Currie, L. A. Anal. Chem. 1968, 40, 586-593. (8) Pentoney, S. L.; Zare, R. N. Anal. Chem. 1989, 61, 1642-1647. S0003-2700(96)00390-3 CCC: $12.00
© 1996 American Chemical Society
used previously has been changed. While the membrane performed well, it is in short supply. The membrane can also act as an unintented selectivity parameter, as noted by Chiu and co-workers.13 After screening a number of membranes, we selected Hewlett-Packard ink-jet paper as a general purpose binding membrane because it maintains highly efficient separations and is readily available.
Figure 1. Schematic diagram of the phosphor-imaging system, showing the imaging plate and read-out process. Areas of the plate exposed to radiation emit blue photons upon excitation with the laser.
Figure 2. Diagram of the capillary membrane connection. The electrophoretic separation takes place in the analytical capillary on the left. Ground contact is maintained by a silver electrode painted on the outside of the analytical capillary. Makeup buffer is supplied from the capillary on the right. The membrane is Hewlett-Packard ink-jet paper.
A number of improvements have been made to the CE electrical ground system. The original CE ground used a conductive silver layer painted on the outside of the outlet end of the capillary to complete the CE circuit.6 This allowed the capillary along with the electrode to be moved across the membrane. Silver is still painted on the outside of the capillary to complete the CE circuit; however, we now taper the capillary to the inside diameter, using a technique similar to that of Kriger and co-workers, prior to applying three coats of silver paint.12 This greatly reduces the exposed silica capillary face in contact with the membrane, as shown in the schematic on the left side of Figure 2. The taper improves the membrane deposition process and improves reproducibility. The capillary is now held at a 45° angle to the membrane and the last 1 cm of the capillary is unsupported to provide flexibility to respond to irregularities in the membrane and its support. Finally, selection of the correct membrane used for postcolumn fraction collection is important to maintain high separation efficiency. The quaternary ammonium peptide binding membrane (9) Sonoda, M.; Takano, M.; Miyahara, J.; Kato, H. Radiology 1983, 148, 833838. (10) Takahashi, K.; Miyahara, J.; Shibahara, Y. J. Electrochem. Soc. 1985, 132, 1492-1494. (11) Johnston, R. F.; Pickett, S. C.; Barker, D. L. Electrophoresis 1990, 11, 335360. (12) Kriger, M. S.; Cook, K. D.; Ramsey, R. S. Anal. Chem. 1995, 67, 385-389.
EXPERIMENTAL SECTION CE System. A commercial capillary electrophoresis combination autosampler and power supply (Model 300, Analytical Technology Inc., Boston, MA) is used to analyze radionuclide standards and homogenates of cells. The capillary outlet tip is tapered to the inner diameter using a portable grinder (Model 395, Dremel, Racine, WI) equipped with a rubber polishing tip (Dremel, Model 464). The capillary is slowly rotated and brought into contact with the polishing tip, turning at approximately 10 000 rpm. The taper runs from approximately 0.5 mm from the end to the tip. An electrode is constructed by applying a conductive silver paint (Catalog No. 22-202, GC Electronics, Rockford, IL) to the outside of the final 4 cm of the outlet end of the capillary. A total of three coats are applied. The coating is heated to 120 °C for 2 h between each coat and after the final coat. The coating is stable for several weeks when using a basic buffer (pH 8.4) and several days when using an acidic (pH 3.5) buffer. The extra silver layers extend the life of the ground in acidic media. HPLC tubing (0.020 in. i.d.) approximately 1 cm in length is epoxied in place so the capillary extends approximately 0.8 cm past the tubing end. The tubing provides support for the CE column, allowing the capillary to be mounted at a 45° angle to the membrane. A buffer makeup capillary is constructed by tapering the outlet end using the same method as for the analytical separation capillary. A capillary 60 cm long with a 100 µm i.d. is used for all analyses. For the analysis of standards, the makeup capillary outlet is mounted approximately 1 mm above and 1 mm behind the analytical capillary outlet and is held in place with epoxy glue (1:1 wt ratio) (Catalog No. S-35, Devcon, Wood Dale, IL). The capillary assembly is then mounted in a custom-built black delron holder so both capillaries contact the membrane surface. The delron holder allows both capillaries to pass through and contains the ground connection set-screw to complete the electrical circuit. The HPLC tubing is inserted partly into the delron holder, and held in place with a set-screw so the tubing and capillary combination extend 2 cm past the holder. A 20× dissecting scope (Model BM, Meiji Techno) is used to inspect the capillarymembrane contact. Radiolabeled Cells. The radiolabeling procedure is similar to methods developed by Church and Lloyd.14 Buccal ganglia are dissected from Aplysia californica (a sea slug), washed in gentamicin sulfate (150 µg/mL in artificial sea water), and cultured in 0.7 mL of methionine-free L15 media (Catalog No. 95-0192DJ, Gibco BRL, Gaithersburg, MD), with salts added to make the media isotonic with Aplysia cells.15 After 12 h, 1 mCi of 35S-Met (Catalog No. SJ1015, Amersham, Arlington Heights, IL) is added. The ganglia are cultured for another 24 h, followed by removal of the unincorporated 35S-Met by washing 12 times with 0.7 mL of (13) Chiu, R. W.; Walker, K. L.; Hagen, J. J.; Monnig, C. A.; Wilkins, C. L. Anal. Chem. 1995, 67, 4190-4196. (14) Church, P. J.; Lloyd, P. E. J. Neurosci. 1991, 11, 618-625. (15) Schacher, S.; Proshansky, E. J. Neurosci. 1983, 3, 2403-2413.
Analytical Chemistry, Vol. 68, No. 22, November 15, 1996
3923
methionine-free media, waiting 10 min between each wash. Ganglia are then cultured an additional 12 h in 0.7 mL of nonradiolabeled media. Injection and Separation Parameters. Methionine standards are injected onto a 65 cm × 75 µm i.d. fused silica capillary (Polymicro Technology, Phoenix, AZ) using a pressure of 30 mbar for 6 s (15 nL volume), followed by an injection of buffer at 20 mbar for 6 s. The buffer injection provides that buffer escapes from the inlet side, instead of sample, as the outlet side is switched from an outlet vial to the membrane. The separation buffer is 300 mM acetic acid with 7 M urea. The pH of the urea-buffer combination is ∼4.6, and the viscosity (used in calculating injection volume)16 increases to 1.5 times that of water.17 The makeup capillary buffer is 300 mM acetic acid, and the height difference between the buffer reservoir and the membrane is 10 cm, giving a flow rate of 270 nL/min. The separation potential is 25 kV with a current of ∼10 µA, and the capillary temperature is maintained at 20 °C. ATP standards are separated in 10 mM borate at pH 8.3 using a 65 cm × 75 µm i.d. fused silica capillary. The same buffer is used in the makeup capillary with a height difference of 4 cm, producing a flow rate to the membrane of 108 nL/min. A 20 mbar pressure injection for 6 s gives a volume of 15 nL, followed by an injection of buffer using the same parameters. A separation potential of 20 kV is applied, giving a current of ∼30 µA. The peptide standards separation uses a 75 cm × 75 µm i.d. fused silica capillary. The buffer is 10 mM borate, pH 8.4. A makeup capillary is not required as the high electroosmotic flow rates at this pH prevent reverse migration of the OH-. An injection of 10 mbar for 3 s gives an injected sample volume of 3 nL. (Higher injection volumes caused the current to drop to zero during the run.) An applied separation potential of 20 kV produces a current of ∼20 µA. Cellular Injections. The CE system used to inject neurons, previously labeled with radioisotope, consists of an inverted microscope (Zeiss Axovert 100, Thornwood, NY) equipped with a micromanipulator (Eppendorf Model 5171, Madison, WI). The capillary inlet end is attached to the micromanipulator on the microscope stage. The micromanipulator is used to position the capillary inlet (tapered to a narrow outer diameter using hydrofluoric acid)18 near the cells for injection. For injections, a negative voltage (Model PS/MJ30N0400-11; Glassman High Voltage, Whitehouse Station, NJ) is applied to the outlet end in a Petri dish. Once the injection is made, the capillary inlet is removed from the cell Petri dish and placed in a Petri dish containing running buffer on top of a 1.2 cm thick plate of Plexiglas to insulate the high voltage from the stage, and the outlet end is placed on the membrane, as shown in Figure 2. A potential is then applied to the inlet side of the capillary using a positive high-voltage power supply (Bertan Model 230, Hicksville, NY). While the Plexiglas plate prevents arcing to the microscope, it leaves a very high voltage (20 000 V) exposed. Care must be taken to avoid accidental electrical shocks. For cellular injections, a ganglion is treated with 1% protease (Catalog No. P-8811, Sigma, St. Louis, MO) in L15 media15 for (16) Jandik, P.; Bonn, G. Capillary Electrophoresis of Small Molecules and Ions; VCH: New York, 1993; Chapter 3. (17) Wolf, A. V.; Brown, M. G.; Prentiss, P. H. In Handbook of Chemistry and Physics; Weast, R. C., Ed.; CRC Press: Boca Raton, FL, 1975; p D-264. (18) Kristensen, H. K.; Lau, Y. Y.; Ewing, A. G. J. Neurosci. Methods 1994, 51, 183-188.
3924
Analytical Chemistry, Vol. 68, No. 22, November 15, 1996
30-35 min at 34 °C and then transferred to a dish containing the running buffer of 300 mM acetate pH 3.6 with 7 M urea (the final pH of the urea-buffer combination is ∼4.6). Individual cells are loosened from the main ganglion using a glass microrod with a pulled tip. The dish is placed onto the microscope stage, and single buccal neurons are injected into an 84 cm × 75 µm separation capillary by applying a voltage of -24.9 kV at the outlet end immersed in running buffer. The makeup capillary contains 300 mM acetic acid delivered at a rate of 526 nL/min from a 19.5 cm height. A voltage of 20.7 kV applied at the inlet end produces a current of ∼17 µA. Fraction Collection. Analytes eluted from the CE outlet are collected on a membrane moving underneath the outlet. The membrane fraction collection system consists of a strip chart recorder (Model 0157-0000; Cole-Parmer, Chicago, IL) with a collection membrane attached to the strip chart paper by doublesided tape (Catalog No. 666, 3M, Minnesota, MN). The membrane used for our application is paper originally designed for ink-jet printers (Catalog No. HP 51636H, Hewlett-Packard). The chart speed is 3 cm/min for all separations. The membrane fraction collection system and the ground side of the electrophoresis system are contained in a Plexiglas box with interlocks to interrupt current flow if the box is opened when power is applied.19 Detection. After CE eluant collection, the membrane is cut into sections 30 cm long and placed on a phosphor-imaging plate (Catalog No. BAS-IIIs, Fuji, Stamford, CT). For the detection limits we require, we expose the plates from 10 to 72 h with a 15 h typical exposure. The plates are read with a phosphor-imaging system (Model BAS-1000, Fuji). Data are collected into a twodimensional image of the plate and exported to a Macintosh Quadra 800 (Apple, Cupertino, CA) for further analysis. A routine in MacBAS software (Version 2.0, Fuji) running on the Macintosh is used to extract a row of data 10 pixels wide from the image, which contain the CE eluant results. The result from the extraction is an X vs Y plot of the distance measured on the membrane verses the summed PSL intensity across the 10 pixels. This plot is exported to a text file. Data from the different membrane sections are transferred an IBM-PC clone (MIS, Sunnyvale, CA) and combined using the DOS COPY command. The reconstructed electropherograms are plotted using Sigma Plot (Jandel, San Rafael, CA). The Fuji system accumulates some background radiation-induced single-pixel spikes. Spikes are removed in Sigma Plot using a macro; the spike is removed if it is greater than 3 times the average of the two preceding and two following data points. A calculated average is used to replace the original spike. Savitsky-Golay filtering is done using TableCurve (Jandel). A 1% filter (defined by Jandel as using a moving window consisting of 1% of the total number of data points) is used for all smoothing routines. LODs are calculated using 3 times the standard deviation of the baseline noise as criteria for peak detection. All exposure times for the linear curve calibration are close to 15 h and are corrected to a standard 15 h exposure by multiplying the peak area by the ratio (15/actual hours exposed). Reagents. Radiochemicals are from Amersham Co. (Arlington Heights, IL). The 35S-Met used for the standard curve (Catalog No. SJ 1515) is in an aqueous solution stabilized with 2-mercap(19) Sweedler, J. V.; Shear, J. B.; Fishman, H. A.; Zare, R. N.; Scheller, R. H. Anal. Chem. 1991, 63, 496-502.
toethanol and pyridine-3,4-dicarboxylic acid. The 35S-Cys (Catalog No. SJ 232) contains potassium acetate and dithiothreitol as preservatives. The (R-32P)ATP (Catalog No. 10160) is also from Amersham. All amounts of radionuclide are based on manufacturer specifications adjusted for radioisotopic decay. Cysteine-containing peptides are labeled with 35S-Cys by forming a disulfide linkage. Benzofuroxan is used to form the disulfide bond.20 The reaction is completed in running buffer. Peaks are identified by reacting the peptides individually with 35SCys and comparing the migration times of the individual labeled peptides to peak migration times of the peptide mixture. Peptides are purchased from Sigma Chemical Co. (St. Louis, MO) and American Peptide Co. (Sunnyvale, CA). All other chemicals are from Sigma Chemical Co. RESULTS AND DISCUSSION The method we use to collect analytes on a membrane is an extreme example of reduced volume (
