Anal. Chem. 2004, 76, 7102-7108
Radionuclide Imaging of Miniaturized Chemical Analysis Systems Martin Lave´n,† Susanne Wallenborg,‡ Irina Velikyan,§ Sara Bergstro 1 m,† Majda Djodjic,† Jenny Ljung,† † † ,† Oskar Berglund, Niklas Edenwall, Karin E. Markides,* and Bengt Långstro 1 m§,|
Department of Analytical Chemistry and Department of Organic Chemistry, Institute of Chemistry, Uppsala University, Box 599, SE-751 24 Uppsala, Sweden, Gyros AB, Uppsala Science Park, SE-751 83 Uppsala, Sweden, and Uppsala Imanet AB, Box 967, SE-751 09 Uppsala, Sweden
We propose radionuclide imaging as a valuable tool for the study of molecular interactions in miniaturized systems for chemical analysis. Sensitive and quantitative imaging can be performed with compounds labeled with short-lived positron-emitting radionuclides, such as 11C and 68Ga, within selected parts of the system. Radionuclide imaging is not restricted to transparent materials since the relatively energetic positrons can penetrate high optical density materials. Experimentally, a radiotracer is introduced into the object of study, which is subsequently placed on a phosphor storage plate. After exposure, the plate is scanned with a laser and a digital, quantitative image can be reconstituted. To demonstrate the concept, three types of microstructures suited for integration in chemical analysis systems were imaged with 11C- and 68Ga-labeled tracers. The influence of factors such as geometry of the object and type of radionuclide on resolution and sensitivity was investigated. The resolution ranged from 0.9 to 2.7 mm (fwhm). Measuring low amounts of radioactivity in the three structures, 2-20 Bq could be detected, which corresponded to 2.3-500 amol or 2.4-110 pM tracer. The imaging approach was applied to study analyte concentration and sample dilution effects on the performance of a capillary extraction column integrated in an automated LC-ESI-MS system. The utility of the technique was further illustrated by imaging of microchannels in a zeonor plastic compact disk and in a poly(dimethylsiloxane) material for the study of nonspecific peptide adsorption. Since the tracer concept was first envisioned in 1913, substantial progress has followed, particularly in the field of medicine.1 The tracer principle refers to the idea of employing a radionuclide, incorporated in a molecule or in a stand-alone ionic state, to trace the chemical and biological behavior of its stable counterpart. In positron emission tomography (PET), molecules labeled with * Corresponding author. Fax: +46 18 471 36 92. E-mail: Karin.Markides@ kemi.uu.se. † Department of Analytical Chemistry, Uppsala University. ‡ Gyros AB. § Department of Organic Chemistry, Uppsala University. | Uppsala Imanet AB. (1) Wagner, H. N., Szabo, Z., Buchanan, J. W., Eds. Principles of Nuclear Medicine, 2 ed.; W. B. Saunders Co.: Philadelphia, PA, 1995.
7102 Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
short-lived positron emitters, such as 11C, 13N, 15O, 18F, and 68Ga, can be used to study biochemistry and pharmacology in humans and research animals in vivo.1-3 Recently, significant improvements in synthetic chemistry, especially 11C-carbon monoxide carbonylation reactions, have permitted labeling of a significant number of new short-lived radiotracers.4,5 This has led to the introduction of the PET microdosing concept, used for tracer and early clinical drug development.6 Tracer technology not only visualizes receptor and drug distribution in a biological system, it also offers unique possibilities for studying the chemical behavior of compounds in a chemical analysis system. Extraction efficiency and system losses can readily be determined by detection of the γ-radiation associated with positron decay.7 Detection of the positrons, on the other hand, provides the opportunity to perform quantitative imaging. Capillary liquid chromatography columns have thus previously been imaged in our laboratory with 11C-labeled tracers.2 The method, using a PhosphorImager system, permits quantitative imaging of the entire column, rather than standard off-column detection, and thus a means to study events occurring within the column. The miniaturization drive in analytical chemistry has been spurred by the search for improved analytical performance and has led to the development of downscaled separation techniques such as microcolumn liquid chromatography (micro LC), capillary electrophoresis (CE), and capillary electrochromatography. With the introduction of the micro total analysis system concept in the early 1990s, miniaturization was encompassed in all steps in the analytical chain, from sampling to detection.8 For the analysis of very small sample volumes, miniaturization is a prerequisite, since minimal dead volumes and sample losses are required. Downscaling also yields numerous additional potential gains such as improved separation speed and efficiency, the possibility to (2) Långstro ¨m, B.; Kihlberg, T.; Bergstro¨m, M.; Antoni, G.; Bjo¨rkman, M.; Forngren, B. H.; Forngren, T.; Hartvig, P.; Markides, K.; Yngve, U.; O ¨ gren, M. Acta Chem. Scand. 1999, 53, 651-669. (3) Hartvig, P.; Bergstro ¨m, M.; Långstro ¨m, B. Toxicol. Lett. 2001, 120, 243251. (4) Kihlberg, T.; Långstro ¨m, B. J. Org. Chem. 1999, 64, 9201-9205. (5) Karimi, F.; Kihlberg, T.; Långstro¨m, B. J. Chem. Soc., Perkin Trans. 1 2001, 1528-1531. (6) Bergstro ¨m, M.; Grahne´n, A.; Långstro ¨m, B. Eur. J. Clin. Pharmacol. 2003, 59, 357-366. (7) Jacobson, G. B.; Moulder, R.; Lu, L.; Bergstro¨m, M.; Markides, K. E.; Långstro ¨m, B. Anal. Chem. 1997, 69, 275-280. (8) Manz, A.; Graber, N.; Widmer, H. M. Sens. Actuators, B 1990, 1, 244-248. 10.1021/ac040070e CCC: $27.50
© 2004 American Chemical Society Published on Web 11/03/2004
integrate different sample handling steps for analysis in an automated high-throughput manner, and flow rates close to optimal for electrospray ionization-mass spectrometry (ESI-MS), as well as less solvent consumption and the possibility of performing parallel separations on a single microchip device.9,10 However, a number of challenges have to be faced as a result of the reduced dimensions and increased surface-to-volume ratios in microstructures, such as control of liquid flow rate, surface chemistry, and a reliable and reproducible manufacturing process.11,12 Quantitative radionuclide imaging can be of significance in the evaluation and optimization of such miniaturized analysis systems. Selected parts of the intact system can be studied in a lateral fashion, which permits the analysis of within-system processes, that can be determined to position and magnitude. This is not as straightforward using end-point detection, where the generated data are the sum of all events within the system. An additional advantage of positron detection is the ability to study molecular interactions in a range of materials and solutions. The relatively energetic positrons can penetrate high optical density materials, which is not appropriate with optical UV and fluorescence imaging techniques.13-15 The penetration range is dependent upon the density of the material, but independent of parameters such as temperature, pH, and salt content. The only source of background signal is low-intensity cosmic radiation. These features make radionuclide imaging suitable for the study of real systems, where absorbing or quenching materials and background signals, from unwanted light scattering or fluorescence,16 can complicate detection using optical techniques. Tracer technology relies on the ability to produce labeled compounds. Some short-lived positron-emitting radionuclides can only be produced by accelerators, but others can be obtained from simple portable generators, such as a 68Ge/68Ga generator. A significant amount of labeling methods for PET tracers have been published, ranging from small molecules to large biomolecules, and the field is expanding.17 A great advantage of utilizing shortlived positron-emitting radionuclides is that tracers can be produced with a very high specific radioactivity. This feature permits highly mass sensitive analysis and thus imaging of small amounts of analyte. The objective of this study was to demonstrate and discuss possibilities and limitations of radionuclide imaging for the study of molecular interactions in miniaturized systems for chemical analysis. Aspects such as sensitivity and resolution are discussed in conjunction with geometry and material of the studied object as well as type of radionuclide employed. The discussion is based on the experience of radionuclide imaging of three separate microstructures suited for integration in chemical analysis systems. Applications of the concept are also presented. The performance of a capillary extraction column, integrated in an automated LC-ESI-MS system designed for metabolic stability analysis,18 was (9) Sanders, G. H. W.; Manz, A. TrAC, Trends Anal. Chem. 2000, 19, 364378. (10) de Mello, A. J. Lab Chip 2001, 1, 7N-12N. (11) Mitchell, P. Nat. Biotechnol. 2001, 19, 717-721. (12) Ehrnstro ¨m, R. Lab Chip 2002, 2, 26N-30N. (13) Wu, J. Q.; Pawliszyn, J. J. Chromatogr., B 1994, 657, 327-332. (14) Mao, Q. L.; Pawliszyn, J. Analyst 1999, 124, 637-641. (15) Wu, X. Z.; Wu, J. Q.; Pawliszyn, J. Electrophoresis 1995, 16, 1474-1478. (16) Hawkins, K. R.; Yager, P. Lab Chip 2003, 3, 248-252. (17) Welch, M. J., Redvanly, C. S., Eds. Handbook of Radiopharmaceuticals; John Wiley & Sons Ltd.: New York, 2003.
thus studied using a 11C-labeled tracer ([11C]flumazenil). Microchannels in a zeonor plastic compact disk (CD) served as a model for microsample preparation systems in the CD format and channels in an elastomeric poly(dimethylsiloxane) (PDMS) material as a model of an interface for coupling LC and CE.19 These latter microstructures were imaged to study nonspecific peptide adsorption, using 68Ga-labeled peptides. EXPERIMENTAL SECTION Materials. Formic acid (p.a.) and 2-propanol were obtained from Merck (Darmstadt, Germany). Acetonitrile (Chromasolv) and potassium dihydrogen phosphate were purchased from Riedel de Hae¨n (Seelze, Germany). Dipotassium hydrogen phosphate (ultrapure bioreagent) was obtained from J. T. Baker (Phillipsburg, NJ). Extraction column packing material Oasis HLB (30-µm particle diameter) was purchased from Waters Corp. (Milford, MA). Angiotensin II (human) and vasoactive intestinal peptide were purchased from Sigma-Aldrich. 1,4,7,10-Tetraazacyclocyclododecane-1,4,7,10-tetraacetic acid (DOTA) was obtained from Macrocyclics (Dallas, TX). Labeling Procedures. [11C]Flumazenil ([11C]Ro 15-1788) was obtained by N-alkylation of N-desmethylflumazenil with [11C]methyl iodide in alkaline solution.20 Peptides (vasoactive intestinal peptide and angiotensin II) were conjugated with DOTA and labeled with 68Ga,21 obtained from a 68Ge/68Ga generator (Cyclotron C., Obnisk, Russia). Microstructures. Three types of microstructures were investigated. Extraction columns (Figure 1A) integrated in an automated capillary LC-ESI-MS system, designed for microsomal metabolic stability analysis,18 were studied using [11C]flumazenil. The columns were prepared by slurry packing fused-silica capillaries (0.5-mm i.d., 0.75-mm o.d., and 50-mm length) with Waters Oasis HLB packing material in 2-propanol. Each sample (200 µL) was injected with a CMA Microdialysis autosampler (CMA Microdialysis AB, Stockholm, Sweden), operated at 4 °C, and loaded onto the extraction column with a Beckman programmable solvent module 126 (Beckman, Coulter, CA) at 1 mL/min using a 5 mM formic acid mobile phase. Microsome samples were prepared from livers of male Sprague-Dawley rats and contained 1 mg/mL microsomal proteins, dissolved in a 100 mM phosphate buffer (pH 7.4). Prior to injection onto the extraction column, the microsome samples were centrifuged at 143000g at 4 °C for 4 min. Before imaging, screens and unions were removed from the extraction column. The second structure investigated was microchannels molded in a PDMS material (Figure 1B). These were imaged using 68Ga-labeled peptides. Three parallel separate channels of 0.18mm diameters were cast with fused-silica capillaries as templates and a Petri dish as a holder, according to a procedure described elsewhere.19 The distance from the channels to the outer surface of the PDMS structure was ∼1 mm. Liquids were introduced and removed using a 50-µL glass syringe. (18) Lave´n, M.; Markides, K.; Långstro¨m, B. J. Chromatogr., B 2004, 806, 119126. (19) Bergstro ¨m, S. K.; Samskog, J.; Markides, K. E. Anal. Chem. 2003, 75, 54615467. (20) Maziere, M.; Hantraye, P.; Prenant, C.; Sastre, J.; Comar, D. Int. J. Appl. Radiat. Isot. 1984, 35, 973-976. (21) Velikyan, I.; Beyer, G.; Långstro ¨m, B. Bioconjugate Chem. 2004, 15, 554560.
Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
7103
Figure 1. Cross-section drawings of three microstructures imaged with compounds labeled with short-lived positron-emitting radionuclides. (A) Extraction column packed with Oasis HLB particles with an internal diameter of 0.5 mm and an external diameter of 0.75 mm. (B) Microchannel in a PDMS material of 0.18-mm diameter. The distance from the channel to the outer surface was ∼1 mm. (C) Microchannel in a zeonor plastic CD, with a width of 200 µm and a depth of 40 µm. The lid thickness was 40 µm.
The third structure studied was microchannels in a Zeonor plastic CD, manufactured at Gyros AB (Uppsala, Sweden). The microchannel system consisted of six interlinked channels, each 3.2 cm long, 40 µm deep, and 200 µm wide, and a laminated lid with a thickness of 40 µm (Figure 1C). The CD was treated with oxygen plasma to create a hydrophilic surface within the microstructure. The hydrophilic surface is necessary for using capillary action as a mode of liquid transport. Samples, containing 68Ga- or 68Ga-labeled peptides, were introduced to the inlet of the channel system by a pipet whereby the liquid was drawn into the channels by capillary action. At given positions in the channels (20.5 mm downstream from the inlet hole), a hydrophobic patch was applied. The patch acted as a valve which was “closed”; i.e., no liquid passed the patch, during application of the sample and incubation. A strict channel volume definition could be obtained through this design. The valve was “opened” by spinning the CD at a predetermined speed, thereby removing liquids from the channels. Imaging. The object of study was placed on a PhosphorImager plate (Molecular Dynamics, Amersham Biosciences). The exposure time was ∼1 h for 11C-labeled analytes and 2 h for compounds labeled with 68Ga. The plate was subsequently scanned with a PhosphorImager SI unit (Molecular Dynamics, Amersham Biosciences), with a pixel size of 50 µm and analyzed with ImageQuant 5.1 software. Extraction columns were analyzed by drawing a histogram, or line graph, on-axis to the column, and peak heights were calculated. An integration procedure was used for CD and PDMS microchannels, where pixel intensities over a selected area were summed. The liquid volumes analyzed were 0.1 and 0.2 µL for CD and PDMS microchannels, respectively. 7104
Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
Microstructures were used for one tracer application and were discarded after imaging. Resolution, Sensitivity, and Reproducibility. The resolution was determined by drawing a line graph crossing the studied structure perpendicularly and calculating the full width of the peak at half its maximum height (fwhm). Fourteen nonpacked fusedsilica capillaries (0.5-mm i.d., 0.75-mm o.d.) were imaged to determine the resolution that could be obtained for extraction columns. The resolution of CD and PDMS microstructures was obtained by the analysis of 6 and 11 structures, respectively. The sensitivity of radionuclide imaging of extraction columns was assessed by injecting a microsome solution of 100 nM 11C-labeled flumazenil (20 pmol) onto three extraction columns and employing time to decrease the amount of radioactivity. The sensitivity of imaging of CD and PDMS microchannels was investigated by the construction of calibration curves from channels filled with 68Ga of different concentrations. A crystal scintillation counter was used for radioactivity measurements of applied tracer solutions. The uniformity of a phosphor storage plate can be assessed by exposing the plate to droplets of the tracer of equal volume and amounts of radioactivity.22 In this study, plates were exposed to 24 spots of 68Ga solution, each of a volume of 20 µL, which had been evenly distributed on paper using a pipet. The reproducibility of extraction column imaging was investigated by injecting the same amount of labeled flumazenil (20 pmol) on extraction columns on two separate occasions, using a total number of 11 columns. Peak heights were calculated and the reproducibility was assessed by the relative standard deviation (RSD) of these. The reproducibility of CD and PDMS microstructure imaging was assessed by filling the channels with a solution of 68Ga-labeled peptides of a concentration ranging from 5 to 75 µM. The RSD of each six-channel system was calculated for the CD structure, analyzing a total number of 27 such systems on 18 different days. In the PDMS structure, the RSD was calculated for 3 channels where a 5 µM 68Ga-labeled peptide solution had been introduced, analyzing a total number of 18 channels on 5 different days. Applications. The effect of analyte concentration on the extraction performance of capillary extraction columns was investigated by duplicate injections of 30, 300, and 3000 nM (6, 60, and 600 pmol) [11C]flumazenil in a microsome solution, performed on two occasions. Peak heights were compared and analyzed by the statistical test analysis of variance (ANOVA), using Microsoft Excel, with R ) 0.05. Sample dilution effects on the extraction performance were investigated by the injection of a microsome solution of 1000 nM [11C]flumazenil with a 20-µL injection loop. The microsome solution was thereafter diluted 10 times with an aqueous solution of 5 mM formic acid and used for injection with a 200-µL sample loop. This procedure resulted in injections of the same amount of analyte at different sample concentrations. To obtain a system where day-to-day comparisons could be reliably performed, a nonpacked fused -ilica capillary, filled with the same amount that had been injected on the extraction column, was used as a (22) Sihver, W.; Sihver, S.; Bergstro¨m, M.; Murata, T.; Matsumura, K.; Onoe, H.; Andersson, Y.; Bjurling, P.; Fasth, K. J.; Westerberg, G.; O ¨ gren, M.; Jacobsson, G.; Lundqvist, H.; Oreland, L.; Watanabe, Y.; Långstro¨m, B. Nucl. Med. Biol. 1997, 24, 723-731.
reference. The peak height ratio from a packed and the nonpacked reference extraction column was calculated. A total number of nine and eight injections was performed with the 20- and 200-µL injection loops, respectively. The assay was performed on three occasions. The results from the two different groups were analyzed with a statistical t-test, using a confidence interval of 99%. Peptide adsorption to CD microchannels was studied by introducing a 68Ga-labeled peptide solution into the channel system followed by 5-min incubation in an atmosphere of air saturated with water. The peptide solution was thereafter removed from the channels by applying a negative pressure at the injection hole. A wash solution (2 µL) was applied to the channel system, which was subsequently removed by spinning the disk. Adsorption behavior in microchannels situated in the PDMS material was investigated by injecting a solution of 68Ga-labeled peptide into the channel. After 5-min incubation, the peptide solution was washed out with 10 µL of water, followed by an air plug of 20 µL, using a syringe. RESULTS AND DISCUSSION Radionuclide imaging is a tool that permits sensitive analysis of chemical events occurring within selected parts of a chemical analysis system. Three sets of microstructures were investigated to demonstrate the value and possibilities of this imaging technique (Figure 1). By placing a structure containing an appropriate radiotracer on a phosphor storage plate, the emitting positrons, and to some extent the ensuing γ-rays, can produce excitations of BaFBr-Eu2+ complexes bound in the plate, thereby forming a latent image of the radiating source. Subsequent scanning of the plate with a helium-neon laser causes deexcitation of the complexes and a release of photons. By measuring the intensity of the luminescence in connection with the position of the laser beam, a digitalized and quantitative image can simply be reconstituted.23 In Figure 2A, an image is displayed of an extraction column onto which 11C-labeled flumazenil had been injected. It can clearly be elucidated from the image that an extraction column length of 50 mm is more than sufficient. Additionally, a histogram, or a one pixel-width line graph, can be plotted over a selected range of pixels allowing a more chromatographic representation of the extraction column. Adsorption and release behavior of the extraction column can, in this way, be studied in more detail, e.g., as a function of concentration, extraction time, and column length. Resolution. The resolution of the imaging method depends on factors such as the type of radionuclide used and the distance between the radiation source and the phosphor storage plate. A thin object should be used for obtaining optimal resolution, since a short distance between the point of origin of the particle and the plate minimizes dispersion of radiation. To further improve the resolution, the radionuclide with the lowest positron energy should be chosen. By decreasing the positron emission energy, the range of the β+-particle, and thus the dispersion, is minimized. The radionuclide 11C (Eβ+max, 0.96 MeV), with a calculated maximum and mean positron range of 3.4 and 0.75 mm in Plexiglas, yields a higher resolution than 68Ga (Eβ+max, 1.9 MeV), with a maximum and mean range of 7.8 and 2.0 mm, respectively.1 The resolution that was obtained in the imaging of nonpacked capillaries filled with a solution of 11C-labeled tracer (Figure 2B) (23) Johnston, R.; Pickett, S.; Barker, D. Electrophoresis 1990, 11, 355-360.
Figure 2. Image of (A) tracer distribution in an extraction column (length 50 mm) packed with Oasis HLB particles for integration in an automated capillary LC-ESI-MS system, designed for microsomal metabolic stability analysis. A microsome solution of 100 nM 11C-labeled flumazenil was injected onto the column. The entry point of the column is to the left in the figure. (B) An image of a nonpacked fused-silica capillary (length 50 mm) filled with the same amount of tracer as in (A) (20 pmol).
was determined to a fwhm of 1.0 ( 0.05 mm ((SD). Imaging of a 68Ga-labeled tracer in microchannels in a CD (Figure 3) gave a resolution of 0.9 ( 0.03 mm. This exemplifies that imaging of structures with reduced dimensions is beneficial for the resolution. A 68Ga tracer was well suited for use in the CD channel case, since the miniaturized structure, with microchannel dimensions of 40 × 200 µm (depth and width) and a laminated lid thickness of 40 µm, counteracted the inherent lower resolution that can be obtained with 68Ga compared to 11C. The resolution from the study of microchannel structures (0.18-mm i.d.) in a PDMS material using a 68Ga-labeled tracer (Figure 4) was lower, with a fwhm of 2.7 ( 0.3 mm. This can readily be explained by the fact that the distance from the channel to the plate was ∼1 mm, which was the thickness of the PDMS structure design. In this application, the use of a 68Ga tracer yielded a resolution that was sufficient for the study. It can thus be seen that imaging with positron emitters does not yield high-resolution images. However, by considering geometry and type of radionuclide, dispersion effects can be minimized. Sensitivity. The sensitivity is dependent on aspects such as the material and thickness of the object, the type of radionuclide used, and the exposure time. A higher sensitivity can theoretically be obtained using a radionuclide with greater positron emission energy since a larger number of particles will reach the plate. 68Ga would thus yield a better sensitivity than 11C, assuming equal BaFBr-Eu2+ excitation efficiencies. A thin object will yield higher sensitivity, since the positrons will encounter less matter in the path toward the plate, which reduces the probability of electron annihilation. High-density materials will interact with the radiation Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
7105
Figure 3. Image of peptide distribution in a microchannel system in a CD (each channel of length 32 cm, depth 40 µm, and width 200 µm). A solution of 500 µM 68Ga-labeled vasoactive intestinal peptide was introduced to the system at the inlet hole at the top of the image. After 5-min incubation, the channels were washed with 0.1% trifluoroacetic acid in 10% ethanol. The washing liquid was spun out by centrifugation. Note the hydrophobic patches, indicated by the arrow, situated 20.5 mm downstream from the inlet hole.
Figure 4. Image of peptide distribution in PDMS microchannels of 0.18-mm i d. A solution of 300 µM 68Ga-labeled angiotensin II was introduced to the system using a glass syringe. After 5 min of incubation, the channels were washed with 10 µL of water, followed by a 20-µL air plug.
to a greater extent than materials with low densities and will thus yield a lower sensitivity. Since the decay positrons of 11C and 68Ga are relatively energetic, it is also possible to image other materials than these exemplified here, i.e., chromatography particles, fused silica, zeonor plastic, and PDMS. Even such a highdensity material as steel displays theoretical ranges that allow imaging, provided that a sufficiently thin object is studied. The theoretical maximum and mean ranges of 11C-positrons in steel are 0.5 and 0.11 mm, whereas the corresponding ranges of 68Ga-positrons are 1.1 and 0.29 mm, respectively (assuming a density of 7.86 g cm-3). The sensitivity of extraction column imaging was investigated by injecting a solution of 100 nM 11C-labeled flumazenil (20 pmol) onto extraction columns and employing time to decrease the amount of radioactivity. In Figure 5A it can be seen that 4 Bq can 7106 Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
Figure 5. Images of tracer distribution in extraction columns displayed in graph form. Pixel size, 50 µm; exposure time, 1 h. (A) 100 nM 11C-labeled flumazenil microsome sample injected with an activity of 4 Bq. (B) Three concentrations of 30, 300, and 3000 nM analyte injected with an activity of ∼1800 Bq. Each line represents an average of two columns.
be detected with an exposure time of 1 h, using a line graph procedure for peak representation. The exposure time of 1 h is equal to 3 half-lives of the radionuclide, which corresponds to a decay of 87% of the initial amount of radioactivity. Increasing the exposure time would not substantially improve the sensitivity, first because most of the decay has already occurred and second due to an increased background signal with time. In the analysis procedure of CD and PDMS microchannel imaging, pixels were integrated over an area, rather than analysis through line graphs with one-pixel widths. The exposure time was 2 h, which corresponds to a decay of 71% of the initial amount of radioactivity. Approximately 2 Bq could be detected in the CD and 20 Bq in the PDMS structure utilizing 68Ga as the radionuclide. The latter result shows that it is possible to image microstructures that are embedded relatively deeply in a supporting material, as is the case of the PDMS microchannels (Figure 1). Imaging of such an object using optical methods might be problematic due to absorption by the surrounding material. Furthermore, the results demonstrate that there is a potential of increasing the sensitivity by minimizing the thickness of the object. Another aspect of sensitivity to consider is the mass sensitivity of the imaging method. The mass sensitivity is dependent on the specific radioactivity (amount of radioactivity per mass). The maximum specific radioactivity that theoretically can be obtained is inversely proportional to the half-life of the radionuclide. A unique feature of the β+-emitting radionuclides employed in PET technology is their short half-lives. Tracers such as 11C, 15O, 18F, and 68Ga display T1/2 values of 20.3, 2.0, 110, and 68 min. It is thus theoretically feasible to obtain very high specific radioactivities from such short-lived radionuclides and to perform highsensitivity analysis. In our laboratory, 11C-labeled flumazenil is produced with specific radioactivities typically in the range of 30-100 MBq/nmol (measured at the end of synthesis). Since the
specific radioactivity decreases rapidly with time, it is important to perform the experiments rapidly if true high-sensitivity analysis is required. The detection limit of 4 Bq obtained in extraction column imaging can be converted to a corresponding mass using the specific radioactivity. With an experimental time of 60 min, 0.5 fmol or 2.4 pM [11C]flumazenil can on average be detected. Such a low concentration could however be difficult to handle using the same injection system due to adsorption losses. Utilizing DOTA 68Ga labeling methods, we can currently obtain high specific radioactivities of ∼3 GBq/nmol at the end of synthesis.21 The sample preparation time of 68Ga-labeled peptides for application to microchannels in a CD and PDMS material was longer compared to [11C]flumazenil, however, due to a drying and reconstitution step. With an experimental time of 2 h, ∼2.3 amol, or an analyte concentration of 22 pM, can be detected in the CD microchannels. In the PDMS channels, the corresponding values were 23 amol, or 0.11 nM. These results clearly illustrate the need for highly sensitive methods in submicroliter-volume analysis. The mass sensitivity was considerably higher in the CD microchannel case as compared to extraction column imaging. However, the volume analyzed in the latter case was 2000 times greater, explaining the lower concentration that could be detected on average in the extraction columns. Reproducibility. The phosphorimager method displays a linear dynamic range covering 5 orders of magnitude23 and the phosphor storage plates show good uniformity, as has been shown previously with 11C tracers.22 The RSD of phosphor storage plates, calculated from the exposure of evenly distributed 68Ga spots, was below 7% in this study. The reproducibility of extraction column imaging was investigated by injections of the same amount of labeled flumazenil onto different columns. A precise measurement of the amount of flumazenil retained on the column could be obtained by measuring the activity injected on the column and the fraction collected from the outlet with a crystal scintillation counter. Using the phosphorimager software permitted the summing of pixel intensities from each column and thus another form of measurement of the amount of activity retained on each column. Whereas the RSD of crystal scintillation counter measurements was low (1%), the RSD value obtained with the phosphorimager method was higher (6 and 18%, with an average of 10% for all imaged columns in the study). Although nonuniformity of the phosphor storage plate could contribute to the distribution, more importantly, this variation is a measure of inhomogeneities within the extraction bed. Irreproducibly formed void spaces will cause measurement deviations since the counting efficiency of the phosphor storage plate is dependent on the distance to the source and the medium between the two. By normalizing pixel intensities to the column with the lowest sum intensity, such deviations could be attenuated. This procedure also made possible analysis of columns onto which different amounts of labeled analyte had been injected. The reproducibility assay yielded RSD values of 3 and 4% on two occasions. These values corresponded well with the average RSD of all imaged columns in the study (4 ( 2%). To reduce variations in the analysis of submicroliter volumes, the liquid handling has to be strictly controlled. The CD microchannel structure examplified in this work was designed to minimize variations in the study of peptide adsorption to the
channel surfaces. Hydrophobic patches located downstream from the inlet hole (Figure 3) ensured volume definition and eliminated liquid flow during the experiment (i.e., during the incubation time). Additionally, the interlinked six-channel structure provided six measurements of each sample, increasing the reliability of the data. Imaging of the six-channel system filled with a solution of 68Ga-labeled peptide yielded an average RSD of 7 ( 5%. The RSD ranged from 3 to 21%, with no more than 7% of the measurements exceeding an RSD of 14%. Liquid handling of solutions for introduction into PDMS structures was optimized by setting up a system where the liquid was drawn into the channel, rather than pushed, using a 50-µL glass syringe. The average RSD obtained was 20 ( 6%, ranging from 12 to 28%. Further improvements, e.g., optimizing the design of the microstructure and the production procedure so that the channel can be filled with a more precise volume, could result in a lower variation. Alternatively, imaging variations can be used as a measurement of the precision of the microstructure manufacturing process, if a controlled sample introduction protocol has been established. Applications. A great advantage of radionuclide imaging is the ability to study real systems, rather than models. In this work, the performance of capillary extraction columns in an automated LC-ESI-MS system18 was investigated using authentic microsome samples. Implementing radionuclide imaging on these columns provided the means to study the extraction process as it occurred within the column. This is not possible to perform using conventional system end-point detection, such as MS, where the analyte is registered after column elution. In that case, only the total sum of, for example, peak-broadening contributions from columns, frits, and connections can be observed. Radionuclide imaging, on the other hand, can give the opportunity to focus on different parts of the system, to localize and quantify individual peak-broadening contributions. An important factor to consider in the extraction process is the concentration of the analyte. Increasing the analyte concentration will at some point cause the extraction column to be overloaded, resulting in reduced extraction efficiency. Radionuclide imaging was applied to monitor the within-column peak shape at increased concentrations of [11C]flumazenil (30-3000 nM), in order detect any deterioration in the extraction performance. The results showed that extraction peak heights were not significantly affected by an increase in analyte concentration, tested by ANOVA, as can also be seen in Figure 5B. This indicated a stable extraction pattern over the studied concentration range. The extraction performance can also be affected by the sample concentration. A dilution of the sample could improve the performance, since a greater number of the active sites of the extraction material then should be available for the analytes. This could lead to a more efficient extraction. Radionuclide imaging was used to investigate whether this effect could be induced in the extraction of [11C]flumazenil, using capillary columns. Extraction of samples that were not diluted (using an injection volume of 20 µL) yielded an average peak height ratio of 14.8 ( 1.5 greater than the reference channel. Samples diluted 10 times (using an injection volume of 200 µL) displayed an average peak height ratio of 12.6 ( 0.6. A t-test indicated that the difference between nondiluted and diluted samples was significant. These results Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
7107
showed that the above-discussed positive dilution effects could not be seen under the experimental conditions used in this study. Not only spatial resolution can be obtained utilizing radionuclide imaging but also resolution in the time domain. On-column focusing in capillary columns has been studied in our laboratory, where the radiotracer was monitored for different time periods during the injection process.2 There is a compromise between resolution in time and sensitivity, however, that has to be considered. Studying short time segments also means short exposure times and thus a limited sensitivity. Radionuclide imaging is therefore not well suited for the analysis of very fast chemical processes. The radionuclide imaging approach can be utilized to study peptide adsorption in miniaturized chemical analysis systems. We are currently using 68Ga-labeled vasoactive intestinal peptide and angiotensin II in our laboratory to quantitatively image nonspecific peptide adsorption to microchannels in a zeonor plastic CD and a PDMS material. The microchannels in the former material served as a model for microsample preparation systems in the CD format, whereas the latter was used as a model of a PDMS interface for coupling LC and CE.19 In Figure 3, the tracer distribution reflects the adsorption of 68Ga-labeled vasoactive intestinal peptide to CD microchannels after a washing procedure using 0.1% trifluoroacetic acid in 10% ethanol. Adsorption of 68Ga-labeled angiotensin II to PDMS channels, after a waterbased wash, is shown in Figure 4. Different parameters such as time of peptide incubation, peptide concentration, pH, ionic strength, organic modifiers, buffer additives, washing procedures, and surface modifications can be changed and the effects on adsorption readily be measured with the developed method. Utilizing radionuclide imaging in the investigation of these structures can be of great help in optimizing the surface chemistries in order to reduce undesired interactions.
7108
Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
CONCLUSIONS The possibilities and limitations of employing quantitative radionuclide imaging in the study of microstructures that can be integrated in miniaturized systems for chemical analysis were demonstrated. Labeling a compound with a short-lived positronemitting radionuclide permits sensitive imaging of events occurring within microsystems. High-resolution imaging is restricted by the relatively high energy of the utilized positrons, but this inherent feature of the positrons does, on the other hand, permit imaging of nontransparent materials. Imaging of microstructures is beneficial from a resolution perspective, since dispersion effects are minimized when the distance between the radiation source and the phosphor storage plate is reduced. The ability to study real systems was illustrated by the investigation of the effects of analyte concentration and sample dilution on the extraction performance of a capillary extraction column integrated in an automated capillary LC-ESI-MS system. Currently, we are conducting research on nonspecific peptide adsorption to microstructures with 68Ga-labeled peptides. In the future, we envision radionuclide imaging playing a role in the evaluation and optimization of new miniaturized biotechnological analysis systems. ACKNOWLEDGMENT Financial support from Uppsala Imanet AB and the Swedish Research Council, Contract K5104-706/2001 (K.E.M.) and K3464345/2001 (B.L.), is acknowledged. Received for review April 19, 2004. Accepted September 9, 2004. AC040070E